**Qualitative and Nutritional Improvement of Cereal-Based Foods and Beverages**

Editors

**Antonella Pasqualone Carmine Summo**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Antonella Pasqualone Department of Soil, Plant and Food Science (DISSPA), University of Bari Aldo Moro Italy

Carmine Summo Department of Soil, Plant and Food Science (DISSPA), University of Bari Aldo Moro Italy

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Foods* (ISSN 2304-8158) (available at: https://www.mdpi.com/journal/foods/special issues/cereal foods).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-0706-4 (Hbk) ISBN 978-3-0365-0707-1 (PDF)**

Cover image courtesy of Antonella Pasqualone.

c 2021 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**



### **About the Editors**

**Antonella Pasqualone** is an Associate Professor of Food Science and Technology at the University of Bari, Italy. Scientifically responsible for many national and international projects, she published 186 articles in refereed international scientific journals, with an h-index of 33 (February 2021). Her main research interests are in the field of cereal science and technology: (i) improvement of the nutritional quality of cereal-based foods by integration with sustainable crops such as pulses; (ii) formulation and characterization of functional pasta and bakery products, also gluten-free; (iii) studies on the baking systems of flatbreads; (iv) improvement of bread shelf-life; (v) activity of polyphenol oxidase in durum wheat and correlations with pasta browning; (vi) effect of drying or baking on the volatile profile of pasta and bread; (vii) food upcycling by reusing food industry waste and by-products in cereal-based foods; (viii) rediscovery and qualitative standardization of non-alcoholic fermented cereal beverages and traditional cereal-based foods.

**Carmine Summo** is an Associate Professor of Food Science and Technology at the University of Bari, Italy. Scientific coordinator and Research Unit member for several research projects, he published 155 articles in refereed international scientific journals, with h-index = 26 (February 2021). His main research interests fall in the field of cereals and legume-based foods: (i) formulation and characterization of functional pasta and bakery products containing bioactive extracts from food industry by-products and waste; (ii) quality characterization of pasta and bakery products, also gluten-free; (iii) improvement of the nutritional quality of cereal-based foods by adding legume flours; (iv) studies on the effect of drying on the volatile compounds of pasta; (v) formulation and characterization of innovative ready-to-eat legume-based foods; (vi) chemical, nutritional and functional characterization of legume varieties to increase the knowledge on the legume properties and avoid the risk of genetic erosion; (vii) studies on the application of air classification to legume flours and on food application of the derived fractions; (viii) production of meat analogues from cereals and pulses flours.

### *Editorial* **Qualitative and Nutritional Improvement of Cereal-Based Foods and Beverages**

#### **Antonella Pasqualone \* and Carmine Summo**

Department of Soil, Plant and Food Science (DISSPA), University of Bari Aldo Moro, Via Amendola, 165/a, I-70126 Bari, Italy; carmine.summo@uniba.it

**\*** Correspondence: antonella.pasqualone@uniba.it

Received: 28 January 2021; Accepted: 3 February 2021; Published: 5 February 2021

**Abstract:** The main directions of research aimed at nutritional improvement have to face either excesses or deficiencies in the diet. To this end, different strategies may be adopted, such as the reformulation of products, the introduction of functional ingredients, and the application of biotechnology to increase the bioavailability of bioactive compounds. These interventions, however, can alter the physico-chemical and sensory properties of the final products, making it necessary to achieve a balance between nutritional and quality modification. This Special Issue offers readers information on innovative ways to improve the cereal-based foods and beverages, useful for researchers and for industry operators.

**Keywords:** functional foods; upcycling; byproducts; bioactive compounds; dietary fiber; new quality; pulses; insects; bread; pasta

Increased consumer awareness of the effects of food in preventing nutrient-related diseases and maintaining physical and mental well-being, has made nutritional improvement an important goal of the food and beverage industry, including the cereal sector. To this end, different strategies may be adopted, such as the reformulation of products, the introduction of functional ingredients, and the application of biotechnologies to increase the bioavailability of bioactive compounds. These interventions, however, can alter the physico-chemical and sensory properties of the final products, making it necessary to achieve a balance between nutritional and quality modification.

The Special Issue "Qualitative and Nutritional Improvement of Cereal-Based Foods and Beverages" collects 17 original research articles and one review aimed at exploring innovative ways to improve cereal-based foods and beverages, an old—if not ancient—group of products which are still on our table every day.

In these articles, a wide range of very different food products is considered, such as bakery products (including white bread, brown bread, durum wheat bread, tortilla, pizza base, muffins and biscuits), fresh and dry pasta, extruded sticks and instant flours, fortified blended foods and Sunsik, the latter being a traditional Korean beverage [1]. Cereal-based beverages, indeed, hold a long tradition and have become known for their sensory and health-promoting attributes [2].

The main directions of research aimed at nutritional improvement have to face either excess or deficiency in the diet. In the latter case, nutrient-rich foods with long shelf-life are needed to prevent malnutrition, whereas in developed countries it is mostly required to decrease the energy value, sucrose, salt, and increase dietary fiber content of foods to prevent obesity and nutrient-related chronic diseases such as cardiovascular disease, hypertension, and diabetes mellitus. The 2030 Agenda for Sustainable Development, adopted in 2015 by the United Nations, set the goals "*to end hunger, achieve food security and improved nutrition, and promote sustainable agriculture*" (Goal 2) and "*to ensure healthy lives and promote*

*Foods* **2021**, *10*, 338

*well-being for all at all ages*" (Goal 3), recognizing non-communicable diseases (NCDs) as a major challenge for sustainable development [3]. In addition, patients with obesity and other chronic underlying conditions are at particularly high risk of developing severe COVID-19 complications [4].

The most natural way to improve the nutritional profile of cereal-based foods is to use wholemeal flour, thus retaining all fiber and micronutrients of wheat caryopsis. Wholewheat flour is a valuable raw material, irrespective of the milling system (stone milling or roller milling) [5]. Higher contents of bioactives can be reached if purple wheat, whose debranning fractions are particularly rich in anthocyanins, is used [6]. Fermentation further improves the nutritional features of wholemeal flours. Fresh pasta prepared by mixing semolina with wholewheat sourdough shows higher content of free essential amino acids and phenolic compounds, lower phytic acid content, and higher antioxidant activity, than control pasta where non-fermented wholewheat semolina is used [7].

Another strategy consists in adding locally available ingredients to reformulate existing cereal-based foods, in order to increase the nutritional value while diminishing the risk of genetic erosion of the local crops and reduce imports. In this context, the leaf powder of *Moringa oleifera*, a plant originating in India and Africa, which is rich in proteins, minerals, and phytochemicals, has been proposed as an additional ingredient to improve the nutritional profile of white and brown bread [8] or fresh pasta [9]. The fortification with moringa leaf increases protein and iron content, but makes bread darker lowering the consumer acceptability, although with a minor impact on brown bread. The addition of moringa leaf powder to fresh pasta in the range 5–15% significantly increases the content of polyphenols even at the lowest percentage. Similarly, legumes can be used which are rich in proteins and complement the amino acid profile of cereal-based foods. Flour of Apulian black chickpeas, an autochthonous black-coated chickpea cultivated in Southern Italy, rich in anthocyanins, has been added to various bakery products [10], namely bread, "focaccia"—an Italian traditional bakery product similar to pizza [11], and pizza crust by substituting flour in the 10–40% range. The rheological properties of dough worsen, resulting in harder and darker final products. However, the nutritional features improve in terms of higher contents of fiber, proteins, and bioactive compounds, as well as higher antioxidant activity [10]. In pasta, the addition of chickpea and hemp flour, previously fermented and enzymatically treated, improves the nutritional profile and protein digestibility, and reduces the sensory drawbacks and the antinutrients (tannins, phytates and raffinose) [12].

Red quinoa or Taiwan djulis (*Chenopodium formosanum* Koidz.) can be used to develop sourdough bread (20% djulis sourdough and 80% wheat flour) [13], whereas germinated wheat, in combination with an extract of *Achyranthes aspera* and *Acanthopanax*, two plants used in Asian traditional medicine, has been proposed to fortify Sunsik, a traditional ready-to-drink Korean cereal-based beverage made of roasted brown rice, barley, adlay, oat, and black beans [1]. Amaranth (*Amaranthus hypochondriacus* L.) and flaxseed (*Linum usitatissimum* L.), instead, have been added to corn-based instant-extruded products, to meet the needs of nutritionally balanced ready-to-eat foods. The addition effectively increases lysine, polyunsaturated fatty acids, minerals, and fiber of the end-products [14]. Seed flour of *Brosimum alicastrum* Sw., a Mexican tree locally named "ramón," characterized by high protein, dietary fiber, and micronutrient content, has been added to wheat tortillas improving the healthy features while keeping pliability unaltered, but with a browner color [15].

Insects are another valuable source of proteins which could be used to overcome the challenge of a more sustainable food chain while improving the nutritional profile of end-products. Meal of winged termites (*Macrotermes bellicosus*) has been added to biscuits prepared with sorghum and wheat flour [16]. Sorghum, which can be grown in tropical areas, makes this biscuit formulation viable in sub-Saharan areas, where protein-energy malnutrition is a major health concern. A significant increase of proteins, minerals, and amino acids is achieved, but biscuits become darker and less hard [16]. In the same geographic area, fortified blended foods (corn-soy blends or wheat-soy blends) are used to prepare viscous porridges in

*Foods* **2021**, *10*, 338

supplementary feeding programs [17]. These food aids often result in products too viscous for being fed to infants and young children, therefore it is needed to dilute them, lowering the nutritional value and energy density. The addition of cowpea has been proposed, also in combination with extrusion cooking, to obtain a porridge able to deliver the correct amount of nutrients at lower viscosities [17]. Extruded cooked products are rehydrated easily without cooking [18], which is important to save energy sources.

The use of waste and byproducts from the food industry, where upcycling is becoming increasingly important to improve sustainability, represents another mean for enhancing the nutritional and healthy features of cereal-based foods [19]. Coffee silverskin, a byproduct from the coffee industry rich in dietary fiber, proteins and bioactive compounds, has been used to produce extruded sticks based on cornmeal and amaranth flour [20], whereas almond skins, a byproduct of almond-based confectionery industry, rich in fiber and phenolic compounds, have been effectively considered as a functional ingredient in biscuits, which become more friable but darker [19].

Finally, to meet the needs related to western lifestyle, the production of muffins containing agave syrup instead of sucrose [21], and the reduction of the sodium content of bread by using a natural low-sodium sea salt [22], have been proposed.

To sum up, this Special Issue gives an interesting contribution to the field and offers readers information on several ways to innovate and improve the cereal-based foods and beverages, which can be useful both for researchers and for industry operators. In most cases, the reformulation with additional ingredients alters the sensory properties, therefore raising the need of communicating a "new quality" to consumers to explain that the differences with conventional counterparts are largely compensated by improved nutritional features.

**Author Contributions:** A.P. writing—original draft preparation; A.P., C.S. writing—review and editing; A.P., C.S. funding acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by: (1) Agropolis Fondation, Fondazione Cariplo, and Daniel and the Nina Carasso Foundation through the "Investissements d'avenir" program with reference number ANR-10-LABX-0001-01, under the "Thought for Food" Initiative (project "LEGERETE"); (2) the Italian Ministry of Education, University and Research (MIUR), program PRIN 2017 (grant number 2017SFTX3Y) "The Neapolitan pizza: processing, distribution, innovation and environmental aspects."

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Foods* **2021**, *10*, 338


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Characteristics of** *Sunsik***, a Cereal-Based Ready-to-Drink Korean Beverage, with Added Germinated Wheat and Herbal Plant Extract**

#### **Bo Ram Kim 1, Seung Soo Park 1, Geum-Joung Youn 2, Yeon Ju Kwak <sup>2</sup> and Mi Jeong Kim 1,3,\***


Received: 16 October 2020; Accepted: 6 November 2020; Published: 12 November 2020

**Abstract:** The purpose of this study was to develop a formulation of *Sunsik* with improved health benefits by adding germinated wheat (GW) and herbal plant extract (HPE) using a response surface methodology (RSM). The central composite experimental design (CCD) was used to evaluate the effects of *Sunsik* with added HPE (2–4%) and GW (10–20%) on total phenolic content (TPC), total flavonoid content (TFC), Trolox equivalent antioxidant capacity (TEAC), 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging capacity, gamma butyric acid (GABA) content, total color changes (-E), browning index (BI), water absorption index (WAI), and water solubility index (WSI). As a result of the CCD, the independent and dependent variables were fitted by the second-order polynomial equation, and the lack of fit for response surface models was not significant except in relation to WSI. The GABA content, TPC, and TEAC were more adequate for a linear model than for a quadratic model, and they might be affected by GW rather than HPE. Alternatively, the TFC, DPPH radical scavenging capacity, WAI, WSI, -E, and BI were fitted with quadratic models. The optimum formulation that could improve antioxidant and physicochemical properties was *Sunsik* with 3.5% and 20% added HPE and GW, respectively.

**Keywords:** cereal-based ready-to-drink beverage; convenient meal replacement (CMR); germinated wheat; response surface methodology (RSM); gamma-amino butyric acid (GABA); antioxidant properties

#### **1. Introduction**

Recently, the increase in single-person and double-income households has shifted consumers' eating behaviors toward the increased consumption of home meal replacements (HMRs) or convenient meal replacement (CMRs) [1]. As ready-to-eat foods, CMRs are a more convenient and simpler meal replacement than HMRs, and they could reduce meal preparation and eating time. The CMR market quadrupled from \$600 million in 2009 to \$2.3 billion 2019. In Korea, the proportion of single-person households is expected to reach 35% of the total population in 2030, and the CMR market is expected to continue to grow.

The types of CMR products are diversifying, such as to include liquid and powder grains, porridges, and cereal bars. Among them, cereal-based beverages are a representative CMR product consumed worldwide because they provide an efficient means to increase the intake of essential nutrients among busy modern people. A few studies investigated the physicochemical and health-conscious properties of various cereal beverages [2,3]. Bembem and Agrahar-Murugkar [2] reported that

millet-based ready-to-drink beverages improved radical scavenging activity, total phenolic content (TPC), and viscosity in the geriatric population. In another study, multigrain beverages prepared with barley, oats, buckwheat, and red rice were identified as providing additional health benefits, such as phenolic content and soluble fiber, to consumers [3].

*Sunsik* has been consumed for a long time as a cereal-based ready-to-drink beverage in Korea. It is made of partially raw or thermal-processed and dried agricultural and marine products [4]. The most common ingredients of *Sunsik* are roasted brown rice, barley, adlay, oat, and black beans [5]. With the recent increase in the demand for healthy foods, much research has reported that additional ingredients, such as various dried vegetables, nuts, and fruits, could be added to *Sunsik* to offer more health-conscious nutrients [6–8]. For example, Park [8] reported that *Sunsik* with added mealworm was higher in antioxidant capacities and in consumer preference than a control *Sunsik*. Regarding the quality of ready-to-drink of *Sunsik*, it should disperse and dissolve well in water or milk within a few minutes. Koh, Jang, and Surh [6] reported that fermented *Sunsik* had a higher soluble solid content, oxidative stability, and amino acids than unfermented *Sunsik*, resulting in an improved solubility and nutrient content. Although several studies reported enhancements in the quality and nutrient content of *Sunsik*, there is limited information on the health benefits of *Sunsik* with added germinated wheat (GW) and herbal plant extract (HPE).

Germination has been identified as an effective processing method to improve the nutritional quality and health-related compounds of cereal [9]. In numerous studies, gamma amino butyric acids (GABA) and phenolic acid compositions were increased as the germination time of wheat increased, suggesting the possibility of GW as a health-conscious ingredient [10–12]. In addition, Dhillon et al. [13] found that the antioxidant activity of and consumer preference for breads were improved when GW flour at 30 ◦C for 72 h was partially used to make bread. The changes in the physiological and biochemical properties of GW might be due to the activation of endogenous enzymes that break down starch and protein into small molecules [14,15]. The activation of endogenous enzymes may also play a role in increasing the solubility of *Sunsik* with added GW when it mixes with water or milk. In addition, plant herbal medicines, such as *Achyranthes aspera*, safflower seed, and Acanthopanax, have been used for the prevention of various diseases in traditional treatments in Asian countries [16,17]. It is known that safflower seeds are rich in lignin, flavonoid, and serotonin and have excellent effects on bone diseases, such as osteoporosis [18]. As previously published in many studies, the extracts of *A. aspera* and *Acanthopanax* showed a reduced inflammatory effect and antioxidant capacities [19–22]. The above-mentioned herbal plant medicines are used not only for therapeutic purposes, but also by adding them to various foods in the form of extracts to increase the health-related functions in the food matrix, such as noodles, drinks, and cookies [23–26]. The HPEs, including *A. aspera*, safflower seed, and *Acanthopanax*, used in this study confirmed previously the pharmacological effects on osteogenic differentiation in human mesenchymal stem cells [27]. The mixture extracts of herbal plants were freeze-dried and then were used in various food products of Gagopa Healing Food Co., Ltd. (Changwon, Korea).

Currently, *Sunsik* with added GW flour and HPE is not available in the marketplace yet. Thus, if GW and HPE are added to commercial *Sunsik*, which is conveniently used as ready to drink beverage, the new *Sunsik* product might be more beneficial to health. The purpose of this study was to determine the optimum formula amounts of GW flour and HPE powder for new *Sunsik* products as cereal-based ready-to-drink beverages. To determine the optimum formulation of *Sunsik*, the response surface methodology (RSM) was adopted using a central composite experimental design (CCD). The antioxidant capacities, GABA, water absorption index (WAI), water solubility index (WSI), total color changes (ΔE), and browning index (BI) were analyzed to optimize the health-conscious nutrients and quality of *Sunsik*; then, the newly optimized *Sunsik* was compared with control *Sunsik* in terms of various health-conscious and physicochemical properties.

#### **2. Materials and Methods**

#### *2.1. Materials*

The *Sunsik* and HPE were provided from Gagopa Healing Food Co., Ltd. (Changwon, Korea). The main ingredients of *Sunsik* consisted of 30% barley, 30% brown rice, 20% adlay, 10% black bean, and 10% oat. In general, each cereal was steamed and then dry-roasted. The four roasted cereals were pulverized in a batch for a production of the Sunsik. The Sunsik used in this study is being sold on the market. Gagopa Healing Food Co., Ltd. (Changwon, Korea) found effects of HPE on osteogenic differentiation through preliminary studies, and the results already published [27]. The HPE used in this study is composed of safflower seed (85%), *A. aspera* (5%), manyprickle acanthopanax (5%), and *Kalopanax septemlobus* (5%) [27]. In addition, the GW used in this study was prepared according to preliminary experiments. Anzunbaengi wheat, which was cultivated in Jinju, Korea, was germinated at 17.6 ◦C for 46.18 h to enhance GABA. After germination, the GW was freeze-dried and then grounded to powder. To develop a cereal-based ready-to-eat beverage to enhance health-related properties, *Sunsik* was formulated with HPE and GW to maximize GABA and antioxidant capacities. The ranges of HPE and GW used in this study were 2–4% and 10–20%, respectively, and the ranges were determined based on samples of five points or more as a result of consumer acceptability (nine-point hedonic scale) of *Sunsik* with added HPE or GW, respectively.

#### *2.2. Experimental Design and Optimization of the Formulation*

The amounts of HPE and GW were optimized using a CCD of an RSM [28]. The independent values were studied at five different levels (− α, −1, 0, + 1, and + α), and the actual levels are presented in Table 1.


**Table 1.** The coded levels and actual values of 13 experiments formulated with a central composite design (CCD).

Table 1 and they were evaluated to maximize the GABA, total flavonoid content (TFC), TPC, 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging capacity, Trolox equivalent antioxidant capacity (TEAC), and WSI and to minimize the WAI, ΔE, and BI. The effects of the two independent variables on the responses (Y) were modeled using the response surface regression, and they were predicted by the following Equation (1) [28]:

$$\mathbf{Y}\_{\mathbf{k}} = \beta\_0 + \beta\_1 \mathbf{X}\_1 + \beta\_2 \mathbf{X}\_2 + \beta\_{12} \mathbf{X}\_1 \mathbf{X}\_2 + \beta\_{11} \mathbf{X}\_1^2 + \beta\_{22} \mathbf{X}\_2^2 \tag{1}$$

where β<sup>0</sup> is a constant, β<sup>1</sup> and β<sup>2</sup> are the linear coefficients, β<sup>12</sup> is the interaction coefficient, and β<sup>11</sup> and β<sup>22</sup> are the quadratic coefficients. X1 and X2 are the levels of HPE and GW, respectively. Yk is the response variable, and each response variable is as follows; Y1 = GABA (μg/g), Y2 = TFC (μg CE/g), Y3 = TPC (μg GE/100g), Y4 = DPPH (μM TE/100g), Y5 = TEAC (mM TE/100g), Y6 = WAI, Y7 = WSI, and Y8 = ΔE, Y9 = BI. To validate the linear or quadratic model, each experimental data of independent variables was compared with the predicted values using the model developed in this study.

#### *2.3. Extraction Procedure of Sunsik Samples*

In total, 5 g of each *Sunsik* sample was extracted with 80% ethanol at 65 ◦C for 2 h, and the supernatants obtained by centrifugation (5000 rpm for 30 min) were evaporated to dryness at 45 ◦C using a nitrogen evaporator (Eyela MG-2200, Tokyo Rikakikai Co. Ltd., Tokyo, Japan). The dried extract was then re-dissolved with 80% ethanol into a final volume of 5 mL. The extract was used to determine the GABA, TEAC, DPPH, TFC, and TPC.

#### *2.4. Gamma-Amino Butyric Acid (GABA)*

The GABA contents of the *Sunsik* samples were determined according to the method described in Sharma et al. [29]. In brief, 0.1 mL of each extract was mixed with 0.2 mL of 0.2 M borate buffer and 1 mL of 6% phenol reagent. Then, 0.4 mL of 7.5% sodium hypochlorite was added, and the mixture was boiled for 10 min in a water bath. The samples were immediately cooled for 5 min, and the absorbance was measured using a spectrophotometer (EMC-11D-V Spectrophotometer, EMCLAB Instruments, Duisburg, Germany) at 630 nm. The GABA was used as a standard curve and prepared with a range of concentrations from 0 to 50 mg. Results were expressed as mg/g.

#### *2.5. Total Flavonoid Content (TFC)*

TFC was determined using the methods previously described by Dahl [30]. The extract of samples (250 μL) was added to 1.25 mL distilled water, and 70 μL of 5% sodium nitrite was added to the mixture. After 6 min, 150 μL of 10% aluminum chloride was added to the mixture. After 5 min, 0.5 mL of 1 N sodium hydroxide was added to the mixture. The absorbance was measured immediately at 510 nm. Distilled water was used as a blank. Catechin was used as a standard curve and prepared with a range of concentrations from 0 to 2.5 mg. The results were reported as catechin equivalents (CE) μg/g.

#### *2.6. Total Phenolic Content (TPC)*

TPC was determined by the method described by de la Rosa et al. [31] with modifications. TPC was measured using the Folin-Ciocalteu method. In total, 100 μL of each extract was added to 2.5 mL of 10% Folin-Ciocalteu reagent, and the mixture was allowed to stand for 2 min. Then, 2 mL of 6% sodium carbonate was added to the mixture, and it was incubated at 50 ◦C for 15 min in a water bath. The absorbance was measured at 760 nm, and distilled water was used as a blank. Gallic acid was used as a standard curve and prepared with a range of concentrations from 0 to 50 mg. Results were expressed as gallic acid equivalents (GAE) mg/g.

#### *2.7. DPPH Radical Scavenging Capacity*

The determination of the effect scavenging of the DPPH radical was based on a procedure previously described by Wong et al. [32]. A 0.1 mM DPPH solution diluted with 100% methanol was prepared. In addition, 0.1 mL of the sample and 1.9 mL of 0.1 mM DPPH were mixed well. The DPPH solution was allowed to stand for 30 min at room temperature in the dark. Then, the absorbance was measured at 515 nm, and 100% methanol was used as a blank. Furthermore, 10 mM Trolox was used as a standard curve and prepared with a range of concentrations from 0 to 500 μM. Results were expressed as μmol of Trolox equivalents (TE) μmol/100 g.

#### *2.8. Trolox Equivalent Antioxidant Capacity (TEAC)*

TEAC was performed as described by Simsek and El [33], with modifications. Briefly, an ABTS<sup>+</sup> stock solution was prepared with 7.4 mM ABTS and 2.6 mM potassium persulfate and mixed. After, the mixture was allowed to stand for 16 h at room temperature in the dark. The ABTS<sup>+</sup> stock solution was diluted with 100% methanol to an absorbance wavelength of 0.7 at 734 nm. Then, 2960 μL of the ABTS<sup>+</sup> stock solution was added to 20 μL of the sample, and absorbance was measured after 7 min. Trolox was used as a standard curve and prepared with a range of concentrations from 0 to 1000 μg. Results were expressed as mmol of TE mmol/100 g.

#### *2.9. Water Absorption Index (WAI) and Water Solubility Index (WSI)*

The WAI and WSI of the optimized *Sunsik* and control samples were determined using methods previously described by Du et al. [34] with slight modifications. In total, 2.5 g of the sample was added to 30 mL of distilled water and mixed in a shaking water bath at 30 ◦C for 30 min. Then, the mixture was centrifuged at 3000 rpm for 15 min. The supernatant and remaining sediment from the mixture were weighted. The supernatant was decanted into an aluminum dish and dried at 105°C overnight using a dry oven. The WAI and WSI were calculated as in the following equations, respectively.

$$\text{WAI} = \frac{\text{weightofthesediment (g)}}{\text{weightofthesample (g)}} \tag{2}$$

$$\text{WSI}(\%) = \frac{\text{weight of dry solids from the supernovaant} \left(\text{g}\right)}{\text{weight of the sample (g)}} \times 100\tag{3}$$

#### *2.10. Color Properties*

The color values of the optimized Sunsik and control samples were determined with a CIE Lab system using a color meter (CR-400, Konica minolta sensing Inc., Osaka, Japan). It was calibrated with a white ceramic plate before measuring the sample. The total color changes (ΔE) and browning index (BI) were calculated as follows [35,36]:

$$
\Delta E = \sqrt{(L\_0^\* - L^\*)^2 + (a\_0^\* - a^\*)^2 + (b\_0^\* - b^\*)^2} \tag{4}
$$

$$BI = \left[100(X - 0.31)\right] / 0.172 \tag{5}$$

$$X = (a^\* + 1.75L^\*) / (5.645L^\* + a^\* - 3.012b^\*)\tag{6}$$

where *L*∗ <sup>0</sup>, *a*<sup>∗</sup> <sup>0</sup>, and *b*<sup>∗</sup> <sup>0</sup> are color parameters for the control and *L*<sup>∗</sup> , *a*∗ , and *b*∗ are color parameters for each Sunsik sample.

#### *2.11. Apparent viscosity of Sunsik Samples*

The apparent viscosity of the optimized *Sunsik* and control samples was measured using a digital rotary viscometer (WVS-0.1M, DAIHAN Scientific, Gang-Won-Do, Korea). First, 45 g of the sample was placed in a 500-mL beaker, and 300 mL of water or milk was poured in, followed by thorough mixing with a magnetic stirrer (MS-20D, DAIHAN Scientific, Gang-Won-Do, Korea). Finally, the thoroughly mixed sample was poured into a 250-mL beaker (SDS 2400, DONG SUNG science, Gang-Won-Do, Korea) and the viscosity of the sample was measured. When measuring the viscosity, the standard was measured when the torque value was close to 50%.

#### *2.12. Cell Proliferative E*ff*ects of Sunsik Samples on Caco-2 and HepG2 Cells*

In total, 15 g of the *Sunsik* samples was extracted with 80% ethanol, evaporated to dryness at 45 ◦C, and re-dissolved in dimethyl sulfoxide (DMSO) according to a previously described method [37]. The Caco-2 (ATCC®HTB-37TM, Manassas, USA) cell was cultured in MEM (Hyclone Laboratories Inc.,

South Logan, UT, USA) with 10% or 20% fetal bovine serum (FBS, Welgene, Daegu, Korea) at 37 ◦C in a humidified incubator with 5% CO2. The cell proliferation of *Sunsik* extracts was determined by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay. The cells (1 × 104/well) were seeded in 96-well plates and then allowed to attach overnight. After overnight, the media included with *Sunsik* extracts were exchanged and incubated for 72 h. After 72 h of incubation, cell proliferation was determined using the MTT Cell Proliferation Assay kit (Roche Ltd., Mannheim, Germany) at 570–655 nm with a SpectraMax®i3 plate reader (Molecular Devices, Sunnyvale, CA, USA).

#### *2.13. Data Analysis*

The Design Expert software (version 11, State-Ease Inc., Minneapolis, USA) was used to analyze the experimental data for best fit model equations and to obtain response plots for each response variable. The combination of independent variables generating the highest overall desirability was selected as the optimum formulation. To validate the optimization process, the Sunsik was prepared using the optimum levels of independent variables and analyzed for the selected responses. The absolute residual error (%) was calculated using the experimental and predicted data through the following Equation (7):

$$Absolute\ residual\ error(\%) = \frac{Actual\ value - Predicted\ value}{Actual\ value} \times 100\tag{7}$$

All experiments were carried out in triplicate, and ANOVA was performed to determine differences among the samples using the XLSTAT software (Addinsoft, Paris, France). When a difference among the samples was identified, the Student Newan–Keul's (SNK) multiple comparison was performed to separate the means.

#### **3. Results and Discussion**

#### *3.1. Fitting the Model and Statistical Analysis*

The RSM is often used to determine the formulation ratio of a new product in the food industry. In this study, a CCD was applied to determine the optimum formulation of HPE and GW to prepare healthy Sunsik, a cereal-based ready-to-drink Korean beverage. The independent and dependent variables were fitted by linear or quadratic equations, and Table 2 shows the statistical results of the regression coefficients, R2, adjusted R2, lack of fit, and p values of the fitted models on analyzed responses by CCD. As shown in Table 2, the lack of fit for response surface models was not significant without the WSI, implying that the response surface models were adequately explained for predicting the relevant responses [28].

**Table 2.** The regression coefficients, R square, adjusted R square, lack of fit, and *p* values of the fitted models on dependent variables.


\*, \*\*, \*\*\* significantly differ at *p* > 0.05, *p* < 0.01, and *p* < 0.001, respectively. β1: herbal plant extract; β2: germinated wheat.

Among the responses, GABA, TPC, and TEAC were more adequate for a linear model than for a quadratic model. Because the β<sup>2</sup> values of GABA (*p* < 0.01), TPC (*p* < 0.05), and TEAC (*p* < 0.01) differed significantly, the GABA, TPC, and TEAC contents of newly developed Sunsik might be affected by GW rather than HPE. The final equations of GABA, TPC, and TEAC as follows:

GABA = 2.09 + 0.017 × HPE + 0.1031 × GW (8)

$$\text{TPC} = 70.57 - 0.3237 \times \text{HPE} + 2.21 \times \text{GW} \tag{9}$$

$$\text{TEAC} = 120.16 + 1.34 \times \text{HPE} + 3.39 \times \text{GW} \tag{10}$$

As described in Table 2, the TFC, DPPH, WAI, WSI, ΔE, and BI were fitted with quadratic models. The final equations of TFC, DPPH, WAI, WSI, ΔE, and BI were coded as follows:

$$\text{TFC} = 30.99 + 1.11 \times \text{HPE} + 3.03 \times \text{GW} + 1.18 \times \text{HPE} \times \text{GW} - 1.45 \times \text{HPE}^2 - 1.47 \times \text{GW}^2 \tag{11}$$

$$\text{DPPH} = 106.59 + 3.71 \times \text{HPE} + 3.32 \times \text{GW} + 2.39 \times \text{HPE} \times \text{GW} - 3.49 \times \text{HPE}^2 - 2.80 \times \text{GW}^2 \tag{12}$$

WAI = 1.85+0.0068<sup>×</sup> HPE−0.0196×GW−0.0394<sup>×</sup> HPE×GW+0.014<sup>×</sup> HPE2 +0.0263×GW2 (13)

$$\text{WSI} = 48.44 - 0.3332 \times \text{HPE} + 4.52 \times \text{GW} + 0.0882 \times \text{HPE} \times \text{GW} - 4.61 \times \text{HPE}^2 - 2.03 \times \text{GW}^2 \tag{14}$$

$$
\Delta E = 0.2224 - 0.2736 \times \text{HPE} + 0.1071 \times \text{GW} - 0.0362 \times \text{HPE} \times \text{GW} + 0.2708 \times \text{HPE}^2 + 0.1384 \times \text{GW}^2 \tag{15}
$$

$$\text{BI} = 20.02 + 0.259 \times \text{HPE} + 0.015 \times \text{GW} - 0.1831 \times \text{HPE} \times \text{GW} - 0.0531 \times \text{HPE}^2 + 0.0061 \times \text{GW}^2 \tag{16}$$

The higher values of R<sup>2</sup> and adjusted R<sup>2</sup> mean desirability of the model to explain the relationships between variables [28]. In this study, the responses with R2 values of 0.8 or higher were TFC, DPPH, WAI, and ΔE, indicating that the fitted equations adequately describe the effects of adding GW and HPE to Sunsik on each dependent variable.

#### *3.2. E*ff*ects of Independent Values on Health-Conscious Properties*

The GABA, TFC, and TPC contents and antioxidant capacities (DPPH radical scavenging capacity and TEAC) of differently formulated Sunsik samples by CCD are shown in Table 3. Significant differences among the 13 samples were found in the GABA (*p* < 0.01), TFC (*p* < 0.001), TPC (*p* < 0.001), DPPH (*p* < 0.05), and TEAC (*p* < 0.05) contents. The GABA content, TFC, and TPC are some of the major compounds that contribute to the antioxidant capacities, such as DPPH and TEAC [11,30,38]. The GABA content and TPC were in the ranges of 1.81–2.25 μg/g and 67–76 μg GE/100g, respectively. As shown in Table 2, the GABA content and TPC were significant in the β<sup>2</sup> value (*p* < 0.01 for GABA and *p* < 0.05 for TPC) but not significant in the β<sup>1</sup> value, indicating that the GABA content and TPC of Sunsik with added HPE and GW were influenced by increased GW. These results were also confirmed in the three-dimensional response surface plots of Figure 1a,c.

**Experiment No. GABA \*\* (Y1,** μ**g**/**g) TFC \*\*\* (Y2,** μ**g CE**/**g) TPC \*\*\* (Y3,** μ**g GE**/**100 g) DPPH \* (Y4,** μ**M TE**/**100 g) TEAC \* (Y5, mM TE**/**100 g) HPE (g) GW (g)** *Sunsik* **(g)** 1 1 5 44 2.01 <sup>±</sup> 0.09 ab <sup>26</sup> <sup>±</sup> 2.30 bc <sup>71</sup> <sup>±</sup> 2.23 cd <sup>96</sup> <sup>±</sup> 3.7 <sup>b</sup> <sup>113</sup> <sup>±</sup> 3.93 <sup>b</sup> 2 2 5 43 2.00 <sup>±</sup> 0.08 ab <sup>24</sup> <sup>±</sup> 1.21 <sup>c</sup> <sup>67</sup> <sup>±</sup> 0.44 <sup>f</sup> <sup>102</sup> <sup>±</sup> 2.9 ab <sup>114</sup> <sup>±</sup> 3.70 <sup>b</sup> 3 1 10 39 2.19 <sup>±</sup> 0.13 <sup>a</sup> <sup>29</sup> <sup>±</sup> 3.68 abc <sup>76</sup> <sup>±</sup> 0.88 <sup>a</sup> <sup>95</sup> <sup>±</sup> 3.1 ab <sup>122</sup> <sup>±</sup> 5.48 ab 4 2 10 38 2.14 <sup>±</sup> 0.16 <sup>a</sup> <sup>32</sup> <sup>±</sup> 3.34 ab <sup>74</sup> <sup>±</sup> 0.97 <sup>b</sup> <sup>110</sup> <sup>±</sup> 9.2 ab <sup>122</sup> <sup>±</sup> 3.30 ab <sup>5</sup> 0.79 7.5 41.71 1.98 <sup>±</sup> 0.06 ab <sup>26</sup> <sup>±</sup> 0.92 abc <sup>69</sup> <sup>±</sup> 0.70 de <sup>96</sup> <sup>±</sup> 6.9 <sup>b</sup> <sup>116</sup> <sup>±</sup> 4.27 <sup>b</sup> <sup>6</sup> 2.21 7.5 40.29 2.12 <sup>±</sup> 0.04 <sup>a</sup> <sup>31</sup> <sup>±</sup> 2.80 abc <sup>72</sup> <sup>±</sup> 0.09 <sup>c</sup> <sup>103</sup> <sup>±</sup> 9.4 <sup>b</sup> <sup>123</sup> <sup>±</sup> 2.86 ab 7 1.5 3.96 44.54 1.81 <sup>±</sup> 0.15 <sup>b</sup> <sup>22</sup> <sup>±</sup> 2.00 <sup>c</sup> <sup>68</sup> <sup>±</sup> 0.40 ef <sup>94</sup> <sup>±</sup> 6.9 ab <sup>119</sup> <sup>±</sup> 3.60 ab

**Table 3.** The experimental values of the health-conscious variables for each independent variable.


**Table 3.** *Cont*.

All values are means of three replications ± standard deviation. Values with the same letter(s) within a column are not significantly different. \*, \*\*, \*\*\* significantly differ at *p* > 0.05, *p* < 0.01, and *p* < 0.001, respectively.

**Figure 1.** Three-dimensional response surface plots of the GABA content (**a**), TFC (**b**), and TPC (**c**). (GW: germinated wheat; HPE: herbal plant extract; GABA: gamma aminobutyric acid; TFC: total flavonoid content; TPC: total phenolic acid).

Conversely, the addition of HPE and GW had significant quadratic effects (*p* < 0.05 for β<sup>11</sup> and *p* < 0.05 for β22) on TFC (Table 2). Figure 1b shows the three-dimensional response surface plots of TFC, implying the TFC of *Sunsik* is increased by both HPE and GW.

The antioxidant properties of 13 Sunsik samples corresponding to the experiments generated by the CCD were determined by DPPH and TEAC (Table 3). The DPPH and TEAC values of the samples differed significantly (both *p* < 0.05) and were in the ranges of 96–110 μM TE/100g and 113–127 mM TE/100 g, respectively. As presented in Table 2, the DPPH value was fitted with a quadratic model while TEAC value was fitted with a linear model. The comprehensive effects of the dependent variables (HPE and GW) on the antioxidant properties of Sunsik are represented by the response surface plots in Figure 2.

**Figure 2.** Three-dimensional response surface plots of DPPH (**a**) and TEAC (**b**).

The Sunsik samples with higher antioxidant activities contained relatively high GABA content, TPC, and TFC. These results are in agreement with previous studies [11], which reported a higher antioxidant capacity of the samples containing higher GABA content, TPC, and TFC. The increments of TPC and GABA content in Sunsik samples could be explained by the addition of GW. Chen et al. [39] reported that phenolic contents in GW increased by lignin synthesis during germination. In addition, another study explained that the GABA content in GW increased via the decarboxylation of L-glutamate [11]. Safflower seed, a major material of HPE, has protective effects against osteoporosis and a beneficial effect on atherogenic risk through various phenolic compounds, such as lignin and flavonoids [25]. Recently, the antioxidant, anti-cancer, anti-inflammatory effects of safflower seeds have been identified by a few studies [25,40,41].

#### *3.3. E*ff*ects of Independent Values on Physicochemical Properties*

The WAI and WSI are important parameters in powdered cereal-based beverages, such as Sunsik, which is eaten by dissolving in milk or water. The WAI and WSI values of the Sunsik samples tested in this study are presented in Table 4. The WAI values of the Sunsik samples were in the range of 1.82–1.95 and did not differ significantly (Table 4). Although there was no statistically significant difference in the WAI values of Sunsik samples, they tended to increase as the amount of HPE increased (Figure 3a). The WAI value of reconstituted powder, such as Sunsik examined in this study, might play a role in preventing its dissolution in milk or water [42]. As shown in the WAI results of Table 2, the linear coefficients of HPE (β1) and GW (β2) were 0.0018 and −0.0195, respectively, implying that GW in newly formulated Sunsik had a negative effect. The WSI is the amount of soluble components released from the Sunsik samples, and the values ranged from 32% to 59% (Table 4). The WSI values of Sunsik with 1.5 g of added HPE and 11.04 g of added GW were the highest among the samples, suggesting the contribution of GW to the solubility of the newly formulated Sunsik samples (Figure 3b).


**Table 4.** The experimental values of the physicochemical variables for each independent variable.

All values are means of three replications ± standard deviation. Values with the same letter(s) within a column are not significantly different. \*\*, \*\*\* significantly differ at *p* < 0.01 and *p* < 0.001, respectively.

Significant differences were observed in the ΔE (*p* < 0.001) and BI (*p* < 0.01) values among the newly formulated Sunsik samples (Table 4), which were in the ranges of 0.22–1.13 and 19.2–20.3, respectively. In the results of the regression coefficients, the HPE addition negatively affected and the GW addition positively affected the ΔE of the newly formulated Sunsik. The three-dimensional response surface plots also showed a similar trend (Figure 3c), indicating that the color of the newly formulated Sunsik was mostly affected by a higher GW amount than HPE amount. Such a result was expected, as more GW (10–20%) was added to Sunsik than HPE (2–4%). The color affects consumer

perceptions of various foods or beverages, and color changes or a brown color during processing or cooking might negatively affect consumer preferences [43]. As shown in Figure 3d, the brown color changes of Sunsik were the result of adding HPE. In a preliminary experiment to determine the range of the HPE amount, consumers tended not to prefer Sunsik with more than 4% HPE added due to its darkened color.

**Figure 3.** Three-dimensional response surface plots of the WAI (**a**), WSI (**b**), ΔE (**c**), and BI (**d**).

#### *3.4. Optimization and Validation*

Cereal-based products like Sunsik are often developed with the addition of two or more ingredients to provide additional health benefits to consumers. In this study, both GW and HPE had a significant effect on the health-related properties and physicochemical characteristics of Sunsik. The additions of GW and HPE in newly formulated Sunsik were response specific. Thus, optimization is needed to attain a formulation with the desired characteristics concerning all the responses.

*Sunsik*, a cereal-based ready-to-drink beverage, was optimized considering maximized properties, such as GABA, TFC, TPC, DPPH, TEAC, and WSI. By contrast, WAI, ΔE, and BI were minimized in Sunsik products. The optimized formula of Sunsik developed in this study was 10 g of GW, 1.79 g of HPE, and 38.21 g of Sunsik corresponding to the highest desirability of 0.719. In addition, the predicted and actual values for optimized formulations of Sunsik are presented in Table 5. Both the predicted and actual values were compared and were verified using absolute residual error values (Table 5). The errors for the responses were found to be less than 5% without ΔE. This indicated the precision of the developed and optimized regression models for the newly formulated Sunsik products.


**Table 5.** Predicted and actual values of the optimized *Sunsik* formulation.

#### *3.5. Health-Conscious and Physicochemical Properties of Optimized Sunsik*

Because the purpose of this study was to develop a newly formulated Sunsik containing GW and HPE to provide health benefits over the commercially available Sunsik, various properties of commercial and optimized Sunsik were compared. The health-conscious and physicochemical properties of both Sunsik samples are presented in Table 6. The GABA content, TPC, and TFC might be major constituents contributing to the antioxidant capacities and antiproliferative cancer cells [38]. Significant differences between the commercial and optimized Sunsik samples with respect to the GABA content (*p* < 0.001), TFC (*p* < 0.001), and TPC (*p* < 0.001) were observed (Table 6). The optimized Sunsik contained more GABA (2.23 μg/g) content, TFC (33.75 μg CE/ 100g), and TPC (73.75 μg GE/100g) than commercial Sunsik (GABA: 1.7 μg/g; TFC 19.8 μg CE/100 g; TPC: 54.4 μg GE/100g), confirming health benefits of optimized Sunsik compared to commercial Sunsik.

In addition, the DPPH (*p* < 0.001) and TEAC (*p* < 0.001) of optimized Sunsik, to which 10 g of GW and 1.79 g of HPE were added, increased significantly compared to commercial Sunsik. Numerous studies have been developed new product with more antioxidant or antiproliferative activities to contribute health benefits of consumed products [7,8,38]. According to Kim and Kim [38], cereal products containing higher phenolic or flavonoid contents had higher antioxidant capacities. In this study, optimized Sunsik contained higher TPC, TFC, DPPH, and TEAC values than the commercial Sunsik. Similar trends were observed in terms of the proliferative activities of cancer cells. The relative proliferative effects on Caco-2 and HepG2 cells after treatment with an extract of the samples are shown as the median effective dose (EC50) in Table 6. The EC50 values of optimized Sunsik for Caco-2 and HepG2 cells were 45.7 and 35.2 mg/mL, respectively. Commercial Sunsik was relatively high in EC50 values of Caco-2 (97.9 mg/mL) and HepG2 (76.2 mg/mL) cells compared to those of optimized Sunsik (Caco-2: 45.7 mg/mL; HepG2: 35.2 mg/mL), indicating relatively low antiproliferative activities. Many studies have reported that foods or beverages with antioxidant activities have cancer-protective effects [37], suggesting that cereal-based beverages could inhibit cancer cell growth. In this study, optimized Sunsik added with GW and HPE showed higher antioxidant capacity and antiproliferative activity than commercial Sunsik.

The WAI, WSI and viscosity of optimized Sunsik with added GW and HPE were compared to commercial Sunsik, and the results are shown in Table 6. The WAI and viscosity of cereal-based beverages are important quality factors [3,4]. According to the finding of Fernandes, Sonawane, and Arya [3], the high absorbing properties in cereal-based beverages resulted in increased viscosity, and high viscosity negatively affected mouthfeel and overall acceptability in sensory tests. According to the results of the current study, the WAI and viscosity of optimized Sunsik with added GW and HPE were less than that of the commercial Sunsik sample. The low WAI and viscosity might contribute to the solubility of Sunsik, which is eaten by dissolving in milk or water, showing higher WSI values in optimized Sunsik than in commercial Sunsik.


**Table 6.** Health-conscious and physicochemical properties of the optimized Sunsik formulation.

All values are means of three replications ± standard deviation. Values with same letter(s) within a row are not significantly different. \*\*\* significantly differ at *p* < 0.001.

#### **4. Conclusions**

This study showed that the CCD and RSM could be used to optimize the formulation of *Sunsik*, a cereal-based ready-to-eat beverage. RSM predicted that a *Sunsik* formula of 10 g GW, 1.79 g HPE, and 38.21 g *Sunsik* would provide a better quality with more health-conscious and physicochemical characteristics. The optimized *Sunsik* is characterized by higher GABA, TPC, TFC, DPPH, TEAC, and WAI values than commercial *Sunsik*. The EC50 of cancer cells, WAI, and viscosity were low in optimized *Sunsik* compared to commercial *Sunsik*. Overall, *Sunsik* with 10 g of added GW and 1.79 g of added HPE might increase various health-related components and biological activities while maintaining the quality of the cereal-based beverage.

**Author Contributions:** Conceptualization, M.J.K.; methodology, B.R.K., S.S.P., Y.J.K., G.-J.Y.; investigation, B.R.K., S.S.P.; data curation, B.R.K., S.S.P.; writing—original draft preparation, B.R.K.; writing—review and editing, M.J.K.; supervision, M.J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by Changwon National University in 2019~2020.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Current Functionality and Potential Improvements of Non-Alcoholic Fermented Cereal Beverages**

**Maria Valentina Ignat 1, Liana Claudia Salant,ă 2,\*, Oana Lelia Pop 2, Carmen Rodica Pop 2, Maria Tofană 2, Elena Mudura 1, Teodora Emilia Coldea 1, Andrei Bors, a <sup>1</sup> and Antonella Pasqualone <sup>3</sup>**


Received: 30 June 2020; Accepted: 30 July 2020; Published: 1 August 2020

**Abstract:** Fermentation continues to be the most common biotechnological tool to be used in cereal-based beverages, as it is relatively simple and economical. Fermented beverages hold a long tradition and have become known for their sensory and health-promoting attributes. Considering the attractive sensory traits and due to increased consumer awareness of the importance of healthy nutrition, the market for functional, natural, and non-alcoholic beverages is steadily increasing all over the world. This paper outlines the current achievements and technological development employed to enhance the qualitative and nutritional status of non-alcoholic fermented cereal beverages (NFCBs). Following an in-depth review of various scientific publications, current production methods are discussed as having the potential to enhance the functional properties of NFCBs and their safety, as a promising approach to help consumers in their efforts to improve their nutrition and health status. Moreover, key aspects concerning production techniques, fermentation methods, and the nutritional value of NFCBs are highlighted, together with their potential health benefits and current consumption trends. Further research efforts are required in the segment of traditional fermented cereal beverages to identify new potentially probiotic microorganisms and starter cultures, novel ingredients as fermentation substrates, and to finally elucidate the contributions of microorganisms and enzymes in the fermentation process.

**Keywords:** cereal beverage; fermentation; functional; non-alcoholic; health benefits

#### **1. Introduction**

Consumers' lifestyles have changed in recent years and will continue to change by being influenced by globalization, economic growth, rapid advances in food science and technology, and/or lifestyle choices and religious restrictions [1–3]. With regard to consumption behaviour, sensory acceptance is still the main choice criteria for consumers [4–6] and it is strongly dependent on cultural backgrounds, as well as previous sensory exposure to a specific food product [7,8].

Beverages are an optimum vehicle to transport nutrients and bioactive compounds into the body as well as to facilitate their bioavailability. Bioactive compounds, such as phytochemicals (e.g., phytoestrogens, phenolic compounds, flavonoids, carotenoids, etc.), dietary fibre, vitamins, fatty acids, probiotics, and minerals, can be incorporated into beverages. The presence of these compounds offers the prospect of using food as a valuable element in disease prevention strategies, particularly in the early stages of the diseases [9–11]. Worldwide, statistics clearly show the growing trend of functional beverage consumption [12], due to their nutrient contents, convenient packaging, design, ease of transportation and storage, and for their shelf-stable nature [13].

Functional beverages could be classified as dairy based, fruit and vegetable based, legume based, cereal based, coffee, or tea. The functionality traits of these beverages address different needs and lifestyles: to boost energy, to fight the ageing process, fatigue, and stress, or to target diseases [14].

Fermentation is widely used to improve the nutritional value, the digestibility level, shelf life, functional properties, texture, taste, and flavour of the beverages [15–18]. A popular class of fermented beverages are those made from cereals: barley (*Hordeum vulgare* L.), maize (*Zea Mays* L.), millet (*Panicum miliaceum* L.), oats (*Avena sativa* L.), rice (*Oryza glaberrima*/*Oryza sativa*), rye (*Secale cereale*), sorghum (*Sorghum bicolor*), and wheat (*Triticum aestivum* L.) [19]. These cereals are a good fermentation substrate and also pose a potentially functional trait, as they contain nutrients that can be easily assimilated by probiotics [20]. The processing steps could include soaking, sprouting, malting, cooking, grinding, and filtering. The fermentation of cereals is influenced by several factors (e.g., length of fermentation, temperature and pH, moisture content of grain, growth factor requirements, cereal nutrients, etc.) which require control by using technological methods to standardize quality.

Fermented beverages are affordable, and their production involves traditional methods to maintain hygiene conditions, product quality, and security [19,21]. Lactic acid bacteria (LAB) dominate the fermentation process and lead to a low pH, which is incompatible with the development of pathogenic bacteria, thereby increasing the shelf life and product safety [22]. Traditional methods exploit mixed cultures of various potentially beneficial microorganisms, referred to as probiotics [23]. Fermented beverages are considered essential because they serve as vehicles for beneficial microorganisms that play an important role in human health, and also for their nutritional, nutraceutical, and pharmaceutical properties [21–24]. Moreover, it was shown that fermentation improves protein digestibility and the bioavailability of minerals and other micronutrients [25].

The consumption of these beverages has the potential to reduce the adverse health and economic impacts of poor diets. Additionally, among the main benefits of cereal-based beverages, there is the possibility of consumption by vegetarians, vegans, and lactose-intolerant consumers [26]. This paper outlines the current achievements and further development for enhancing the functionality of the non-alcoholic fermented cereal beverage (NFCB) segment.

#### **2. Non-Alcoholic Fermented Cereal Beverage Segment**

#### *2.1. An Overview of NFCBs*

Reviewing the existent literature regarding some of the most studied NFCBs, it is shown that they are usually based on barley, maize, millet, oats, rice, rye, sorghum, and wheat [27,28]. Cereals are known as important sources of dietary proteins [29], energy, carbohydrates, vitamins, minerals, and fibre (arabinoxylan and β–glucan), but, at the same time, they are deficient in some basic components (essential amino acids, e.g., lysine) [19,30]. Whole grains are known to have important bioactive compounds and high nutritional values and their regular consumption has a positive effect on health [31–33]. The presence of soluble fibre lowers the glycaemic index of the beverage by slowing down its digestion and absorption, whereas phenolic compounds have antioxidant potential and scavenge harmful free radicals in the body [9].

Most of the traditional and currently produced NFCBs are considered functional foods and wholesome nutritional products [22,34,35]. Functional foods have been intensively studied in recent years and are continuously looked at as research efforts turn to processing food matrices, such as cereals, vegetables, and fruits, into medicine-like products.

A more detailed definition describes functional foods as industrially processed or natural foods that, when regularly consumed within a diverse diet at efficacious levels, have potentially positive effects on health outside basic nutrition [36]. This means that such products should provide a therapeutic benefit if consumed regularly within a diverse diet and with the condition of having the main nutrients extracted in a standardized manner and dosage [37].

Originally, cereal fermented beverages were mainly produced due to the need for conserving and utilizing various cereals and crops with affordable financial implications. Some of them, which are now commercially available as soft drinks and non-alcoholic beverages, were traditionally prepared as alcoholic beverages, having higher contents of alcohol [38].

Currently, "non-alcoholic" is a regulatory term and the laws regarding it vary across the globe. For example, EU regulation no. 1169/2011 simply states that a beverage must be labelled as an alcoholic drink if it has an alcoholic strength by volume (ABV) of over 1.2%. Moreover, in Great Britain, an alcohol-free drink has a maximum alcohol content of 0.05 % ABV; in Germany, a beverage is considered alcohol-free if the maximum alcohol limit is below 0.5 % ABV; in Spain, a non-alcoholic beer contains a maximum of 1 % ABV; in France, alcohol-free beers contain a maximum of 1.2 %; in US, non-alcoholic beers are of a maximum 0.5 % ABV; in China, non-alcoholic drinks can be of up to 0.5 % ABV [39]; in Japan, non-alcoholic beverages contain up to 1% ABV [40]. Depending on the country's traditional recipe, *boza*, a lactic acid fermented drink, has an alcohol content of less than 1% ABV in Turkey and up to 7% ABV in Egypt [41], probably due to microorganisms involved in the fermentation process [42].

Fermented drinks were and are obtained using a combination of fermentable substrates, like cereal mixes, fruits, plants, spices, legumes, and vegetables. The possibility of combining various fermentable substrates and of adding supplementary bioactive compounds to the final product encourages the development of different versions of the same original recipes, according to specific resources, health concerns, and nutritional needs. Such enhancements can be seen, for example, in vitamin A-fortified *mahewu* [43] or in establishing the suitable temperature for saccharification and oligosaccharide production efficiency in *amazake*, containing up to 0.5 % ABV [44].

Globally, there are numerous similar non-alcoholic cereal fermented beverages with similar names and profiles targeted for thirst quenching properties, on nutrition added value, cultural significance, and on providing alternatives to alcoholic beverages. Fermented beverages based on cereals are somewhat common around the world as staple foods, particularly in developing countries and are all made in a similar manner, generally using spontaneous microbial cultures [38]. Although there are various non-alcoholic cereal fermented beverages with different chemical profiles and sensory traits, all of them present certain bioactive compounds and therapeutic agents, which make them beneficial for human health. The functional compounds available and detected in NFCBs have variable values, therefore, it is difficult to assess their functional impact when consumed regularly [17]. If cereals lack certain nutrients, then additional food matrices can be added to enhance the final NFCB, as seen in a case of *mahewu* enhanced with *Moringa oleifera* leaf powder for elevating Ca and Fe contents [45]. Another example of a functional NFCB is a non-alcoholic beverage made of green tea and barley malt wort for delivering superior amino acid content [46].

In Table 1, there are several NFCBs listed with their place of origin, constituent cereals, microorganisms, nutritional compounds, and potential health benefits. As previously mentioned by other authors, the functional outcome of such products is strongly connected to the selection of the cereals, microorganisms, fermentation temperature and time, and other additional food matrices [47–49].


#### *Foods* **2020**, *9*, 1031

22


**Table 1.** *Cont.*

Some of the NFCBs presented in Table 1 are commercially available, with various scales of production, while others are mainly homemade beverages, produced for individual consumption using traditional methods. *Busa*, *kunun-zaki*, *mahewu*, *munkoyo*, *obushera*, *oshikundu*, *pozol*, and *tobwa* are produced in rural and urban areas and their production is essentially a home-based industry as for now, there is no large-scale factory production. *Amazake* is commercially available in Japan and it is classified as a soft drink [90,91]. *Bor¸s* is also industrially produced and used as a flavour enhancer in Romanian gastronomy or it is consumed as a refreshing drink [55]. *Boza* is one of the most popular Bulgarian beverages and is industrially produced at a large scale in all countries of the Balkan Peninsula [92]. *Kvass* is a traditional fermented Slavic and Baltic beverage and is one of the most popular beverages in Russia, numerous varieties emerged due to its popularity and market demand. Currently, *kvass* production is designed according to traditional processes, with the implementation of modern biotechnological methods [72]. *Shalgam* is a traditional beverage of southern Turkish cities and it is commercialized in various markets of European cities [86]. In Table 2, there are more details presented regarding NFCBs and their sensory properties, pH values, nature of use, fermentation status, and production scale.


**Table 2.** Non-alcoholic fermented cereal beverages and their sensory properties, nature of use, and fermentation status.


**Table 2.** *Cont.*

#### *2.2. Processing Technologies and Their Outcome*

The dietary attributes and sensory traits of cereal products can be at times viewed as inferior or deficient in comparison with those of milk and milk-based foods. Some of the reasons behind this include the smaller protein quantities, deficiency in certain amino acids (lysine), the presence of antinutrients (phytic acid, tannins, and polyphenols), and the coarse nature of grains [19].

The fermentation of starchy sources is more complex compared to that of low molecular sugars (glucose or sucrose) because, in general, the starch must at first be converted into fermentable sugars. To achieve an almost complete starch degradation, two main types of amylolytic enzymes are required (α-amylase and glucoamylase) [98].

Several methods have been engaged with the aim of enhancing the nutritional qualities of cereals. These include genetic improvement and amino acid supplementation with protein concentrates or other protein-rich sources, such as grain legumes or defatted oil seed meals of cereals. Additionally, several processing technologies, which include cooking, sprouting, milling, and fermentation, have been put into practice to improve the nutritional properties of cereals, although the best one is probably fermentation. In general, the spontaneous fermentation of cereals leads to a decrease in the level of carbohydrates, as well as some non-digestible poly- and oligosaccharides [35]. Certain amino acids may be synthesized, and the availability of B group vitamins may be improved [99]. Fermentation also provides optimum pH conditions for the enzymatic degradation of phytate, which is present in cereals in the form of complexes with polyvalent cations, such as iron, zinc, calcium, magnesium, and proteins. Such a reduction in phytate may result in a several fold increase in the amounts of soluble Fe, Zn, and Ca [19,100–102].

The current article advances one general processing technique, by compiling all the traditional recipes assessed and integrating germinated and non-germinated grains, to adapt it to an industrial scale with functional improvement. Several traditional processing steps can be applied in the production. The general outline of the process is essentially the same and that presented in Figure 1: the grains, after conditioning, are either soaked/wet milled or the grains are dry milled and the flour is extracted in water afterwards, the mix is boiled to gelatinize the starch, a source of enzymes to hydrolyse the gelatinized starch into fermentable sugars is added, and finally, spontaneous fermentation occurs [19,103].

**Figure 1.** General flow diagram of traditional NFCB production.

#### 2.2.1. Pre-Treatment of Raw Materials

Cereal processing is an essential component of the brewing production chain and the milling process is the main procedure. Before milling, the cleaning and conditioning of the grains is required. The cleaning process allows for the removal of various impurities, depending on the raw material types. The most common grain impurities are shrivelled grains, other cereals, grains damaged by pests, grains with discoloured germs, and sprouted grains. There are also miscellaneous impurities, including extraneous seeds, damaged grains, extraneous matter, husks, ergots, decayed grains, and insects. The conditioning or tempering of grains is performed using the monitored addition of water, which turns the endosperm softer and the bran harder. Doing this prevents the bran from breaking up, aids gradual separation throughout milling, and enhances sieving efficiency [104].

There are two milling categories, namely dry and wet milling, each having its own characteristics. Dry milling removes the germ and the outer fibrous materials of grains, as these by-products are not used in traditional ways [105]. Malting utilizes the power of natural germination when the grains, after absorbing water, germinate in the presence of oxygen to achieve a moisture content of up to 47% [106]. Another by-product is represented by floating kernels, which are unsuitable for malting. Throughout germination, the grain's embryo expands, and rooting begins. Moreover, the germination and steeping processes frequently overlap. It is recommended to keep the germination time and temperature at low values, given that long and warm germination processes lead to longer roots, resulting in larger malt yield losses [104,106].

During germination, grain enzymes start to break down the endosperm high-molecular-weight material into easily digestible components for the yeasts. Drying the malt stops the germination process [104].

In the case of traditional methods, the grains are usually superficially cleaned. A part of the cleaned grains (in variable percentages) is washed and soaked (10 to 20 h) at room temperature (26–35 ◦C). The soaked grains are drained and left for germination for 48 to 72 h with a frequent spraying of tap water. The germinated grains are sun dried (5 to 20 h), so the success of the process depends on the weather, mostly the intensity of the sunshine. Afterwards, the grains (malted and non-malted) are milled separately or in a mixture to obtain flour using rudimentary equipment [107].

Industrial cleaning processes aim at removing impurities and all other materials except for grains, using specific equipment such as magnetic separators, disc or sieve separators, aspirators, destoners, colour sorters, etc. The conditioning process ensures the complete hydration of grains, holding them in suitable containers for specific time intervals. Usually, depending on the grain varieties and initial moisture levels, the soaking time and temperature may be different [104].

For industrial-scale malting (Figure 2), the cereals are dried, and kilning is used to stop further transformations.

**Figure 2.** General flow diagram of industrial NFCB production (compilation from Encyclopaedia Britannica's brewing process).

During controlled drying, the water content should go under 5%, to stop the enzymatic activity whilst colour and flavour compounds are formed. A lot of by-products can additionally be recovered and further capitalized at an industrial scale. Dry milling can be extended to pearling, an abrasive procedure for gradually removing the testa and pericarp, aleurone and sub aleurone layers, and the germ. This results in polished grains and by-products with enhanced contents of bioactive compounds. Alternatively, wet milling, mostly used for producing starch and gluten, can increase the value of the cereals, as it is a source of by-products, with coproducts such as steep solids (abundant in compounds of pharmaceutical interest), germs (used in the oilseed-crushing industry), and bran [104].

#### 2.2.2. Mashing

The traditional mashing process differs greatly, depending on the local culture, and has low efficiency. In Africa, either sorghum or finger millet malt is commonly used as a source of enzymes [22,108]. In the case of gowé production, a beverage based on maize and sorghum, the grains are milled to flour, which is optionally mixed/kneaded with tap or hot water or with a supernatant of a previous production and left to ferment at room temperature. Additionally, this processing technique can include a saccharification step, where a part of the malted sorghum or maize flour is kneaded with water and left for saccharification for 5 to 10 h [107]. For munkoyo, a maize-based beverage with a variable ABV, the specific feature is the usage of *Rhynchosia* roots or watery root extract as an enzymes source [22,77,78]. Starch gelatinization facilitates the activity of α- and β-amylases from the *Rhynchosia* roots for at least 4 h up to a maximum of 24 h, which hydrolyse the starch into fermentable sugars [22,103,109]. Mahewu is an example of a non-alcoholic sour beverage made from corn meal, consumed in Africa and some Arabian Gulf countries [74]. It is prepared from maize porridge further mixed with water. Sorghum, millet malt, or wheat flour is then added and left to ferment [19]. The production techniques of obtaining kvass, a cereal-based beverage produced from rye and barley malt, rye flour, and stale rye bread, uses as raw material either stale sourdough bread or malt rye malt and rye flour [110]. Boza is a colloid suspension, non-alcoholic beverage consumed in Bulgaria, Albania, Turkey, and Romania, made from wheat, rye, millet, maize, and other cereals mixed with sugar or saccharine [92]. Boza's preparation involves six stages: the preparation of raw materials, boiling, cooling, straining, the addition of sugar, and fermentation. Another option for its production is the use of previously fermented boza as an inoculum [59].

The industrial mashing process resembles the beer production process. Between one and three volumes of water are added per volume of milled grains, and during the cooking process, the mixture turns into a mash. The mixture is cooked at a normal pressure or in an autoclave for about 2 h at 4–5 atmospheres.

The macromolecular profile of cereal-based beverages is generally determined by polymeric compounds (proteins, polysaccharides, and polyphenols) and their progress in depolymerization during processing [111,112]. Given that yeasts or specific strains of lactic acid bacteria (LAB) cannot metabolize high-molar-mass substances [113], the macromolecules are solely depolymerized during the malting and mashing process by the malt's intrinsic enzymes. During malting, the enzymatic degradation of polymers is technologically controlled by the degree of steeping, germination time, and germination temperature [114,115]. Modern brewing barley varieties are bred to be balanced in malting performance and to meet the required brewing specifications. The degradation of starch into fermentable sugars (amylolysis) is the primary objective of mashing (substrate production for fermentation) [116], since during the subsequent fermentation process, only low-molar-mass compounds, fermentable sugars, and low-molar-mass proteinaceous compounds are metabolized by microorganisms (e.g., LAB and yeasts) [117,118].

Lautering is the next step in the large-scale process, through which solid and liquid fractions, respectively, the spent grain, composed of sugar-extracted grist or solids remaining in the mash, and the sweet wort with high contents of fermentable sugars are obtained. The spent grain is the major by-product of the brewing industry and represents a valuable source of bioactive ingredients and a potential ingredient for functional foods [119]. In small-scale production, malt extract can be used instead, thus skipping the use of grain malt (including milling, mashing, and lautering) [120].

#### 2.2.3. Cooling and Addition of Yeast, LAB Cultures, and Other Ingredients

After boiling, the mash is gradually mixed with cold water in a 1:1 ratio. This slurry is often filtered or decanted to remove the grinding waste and insoluble plant material. In many traditional processes, where cereals are soaked in water for a few days, a succession of naturally occurring microorganisms will result in a population dominated by LAB. In such types of fermentations of endogenous grains, amylases generate fermentable sugars, which serve as energy sources for lactic acid bacteria. When no malted cereals are used, sucrose is added to the beverage to mimic the malt's sweet taste.

The bacterial flora formed in each fermented cereal drink is influenced by several factors, such as water activity, pH level, salt concentration, temperature, and the composition of the grain matrix, which must be considered in industrial processes. However, most fermented drinks, including the well-known products commonly met in the Western world, as well as those beverages of other origins which are less well studied and characterised, rely on lactic acid bacteria to mediate the fermentation process [19,121]. Lactic acid fermentation contributes towards the nutritional value, shelf life, safety status, and acceptability of a wide range of cereal-based foods [118]. Fermentation is often just one step in the process of fermented food preparation. Other operations, such as volume reduction, salting, or heating, also affect the final product characteristics [19,122]. Depending on the desired product, further steps can be applied afterwards, such as standardisations and the addition of other ingredients like flavourings, sugar, and stabilizers.

#### 2.2.4. Fermentation Process

Despite the lack of process control, dealing with unstandardised microbial flora composition, delayed fermentation, and imperfect reproducibility of the fermentation process, spontaneous fermentation offers complex microbial diversity, providing higher levels of intrinsic stability to the microbial community [123]. This is achievable due to stabilizing interactions between species that prevent and inhibit the proliferation of unwanted microorganisms, including pathogenic ones. Recently, the existence of stability criteria for complex microbial communities was proven [124]. In terms of spontaneous fermentation, this can be explained by the resilience to small perturbations when there is a balance between the availability of resources, namely nutrients, and consumers (e.g., lactic acid bacteria). Commonly, the production of artisanal fermented beverages is conducted in successive batches [55], by using a natural starter from the previous fermentation. At all traditional sites, spontaneous fermentation proceeds after cooling. A 24 h fermentation period is sufficient for

some traditional beverages to develop their characteristic sensory attributes, although in practice, fermentation time can go on for up to three days. Some brews, obtained through a 6 to 15 h fermentation interval of non-malted grain flour, are enhanced with commercial sugar (sucrose) to obtain the sweet taste. This process is a distortion of the original germination technique for gowé production [22]. A modern industrial process is different from the traditional process regarding the introduction and control of thermophilic LAB cultures, which only produce lactic acid, the extension of the product's shelf life by pasteurization and/or chemical preservation, and the inclusion of sugar and/or artificial sweeteners [125,126]. Controlled fermentation also leads to a general improvement in the shelf life, texture, taste, and aroma of the final product. During cereal fermentation, several volatile compounds are formed, which contribute to a complex blend of flavours [127]. Moreover, there is a good opportunity to apply colloidal dispersions in the form of nanoemulsions to deliver food grade nanoparticles, which contain water-insoluble molecules that were formerly unsuitable due to their poor soluble characteristics. Thus, there is a wide range of healthy foods which can be further designed, such as cereal-based fermented beverages enhanced with nanomolecules possessing beneficial health attributes [27].

#### *2.3. Fermentation Microbiota and Safety of NFCBs*

Originally, fermented beverages were only consumed in their native regions, however, due to increasing demand and interest, some of those traditional beverages are available to international markets. The attributes of traditional fermented beverages are influenced by several factors, such as the use of different raw materials, manufacturing methods, natural microbiota, and fermentation conditions. The microbiology of many traditional fermented drinks prepared from the most common types of cereals is quite complex.

There is a lot of diversity in the traditional processing techniques used for cereal-based fermented beverages all over the world, integrating single or multigrain cereals, germinated and/or non-germinated grains. Many types of cereal-based fermented beverages are produced in Africa, such as *togwa* in Tanzania, *mahewu* in Zimbabwe and South Africa, *maw*è in Benin, and *munkoyo* in Zambia and the Democratic Republic of the Congo [9,19,128,129]. Generally, the preparation of these products is a traditional family activity with an uncontrolled fermentation process by diverse microbial communities. The composition of the microbial community in a fermented food product largely determines the key product properties [130,131]. In other words, variations in microbial communities may result in differences in product quality, taste, acceptability, and microbial stability.

In one of our previous papers [55], we proved the impact of processing parameters, namely temperature and batch fermentation cycles, on the chemical composition of bors, . Interestingly, the final composition of cereal-based beverages might not only be influenced by the physical process parameters. Processing practice variation affects the microbial composition of the fermenting microbial community. Despite the decreased pH caused by the lactic acid bacteria activity, the low pH of munkoyo also permitted the development of acidifying bacteria [22]. An important role is played by the initial microbial composition of the raw materials used in the process. The same study also referred to the fact that further investigations might be needed into the soil composition of the harvested raw materials. Along with the prevention of growth of most pathogenic strains [132], a low pH (of 2.5 to 3.5) improves food safety and expands the shelf life of this type of beverage [22]. The pathogenic microorganisms that are aerobes and facultative anaerobes and ferment simple sugars have an optimum pH for growth of 6.0 to 8.0. However, growth can occur at a pH as low as 4.3 and as high as 9, but with a combination of factors (pH, water activity), the control of foodborne pathogen growth can be done [133].

The fermentation of most cereals is natural and involves mixed cultures of yeasts, bacteria, and fungi [134]. Some microorganisms may participate in parallel, while others act in a sequential manner with a changing dominant flora during fermentation. The challenge, though, is the generally uncontrolled nature of the fermentation, which raises safety concerns, as well as the lack of standardization in the methods used, thus further research and development are needed to

improve the traditional fermentation processes [80]. In this regard, the introduction of starter culture technology has led to greater consistency and safety and to better product quality [135].

Furthermore, the main difficulties encountered in the production of cereal-based beverages using traditional processes are linked to the high variability of unit operations and the unhygienic conditions of the processing environment. The soaking and germination parameters (temperature, duration, moisture) vary within and between processors. Moreover, during soaking and germination, the grains can be infested by fungi with the potential development of mycotoxins (aflatoxins) [136].

The continuous study of the fungi, yeasts, and bacteria strongly involved in ensuring a certain quality of the NFCB allows for the optimization of same-product delivery [85,137,138]. Such upgraded fermented beverages are sometimes the outcome of general efforts to enhance original recipes [139]. *Kunun-zaki* is an example of a traditional wheat and sorghum fermented beverage now also commercially available in the form of a powder [68], and there are several strategies proposed to upgrade and re-engineer the process of *gowé* production, a beverage obtained from sorghum and maize [66]. As seen so far, it is possible to prolong shelf life through the co-incubation of probiotic cultures, as seen in the development of some cereal-based fermented beverages [37,140] or to market to a group of consumers looking for healthy and functional foods by using oats, for example, as a main ingredient [141]. In-depth studies are still being conducted to ensure food safety in the processing technology of fermented foods through the keen selection of starter cultures and thorough examination of the specific microorganisms [142,143]. For example, it was recently shown that *Enterococcus faecium* YT52 isolated from *boza* is susceptible to clinically relevant antibiotics and contains low numbers of virulence factors and antibiotic resistance genes. Therefore, the enterocin-producing *E. faecium* YT52 strain poses a low risk to consumer health, and this strain may be used as a starter or a co-starter culture for improving the food safety of fermented products by acting against foodborne pathogens, such as *Listeria monocytogenes* and *Bacillus cereus* [144]. Moreover, rethinking the technological processes for obtaining various cereal-based fermented beverages helps to increase their functionality and overall therapeutic and nutritional properties. Such an example is *boza*, enhanced by fermenting cereals with *Lactobacillus acidophilus*, *Bifidobacterium bifidum*, and *Saccharomyces boulardii*, and by adding chickpea flour to the fermentable substrate for an elevated protein content [145]. Along with these interventions, narrowing the risks of product spoilage through the inoculation of specific strains of LAB and yeasts and processing the cereals in a certain manner repetitively allows the producers to deliver same-quality functional NFCBs.

Fermentation is one of the oldest known biotechnologies and the most economical procedure for producing new foods and ensuring product conservation. The yeasts responsible for fermentation in fermented drinks include species of *Saccharomyces*, *Saccharomycopsis*, *Schizosaccharomyces*, *Pichia*, *Candida*, *Torulopsis*, and *Zygosaccharomyces*. Brewer's yeast, *Saccharomyces cerevisiae,* metabolizes various sugars, mainly into alcohol, but also into other flavour-active substances. The most used microorganisms in fermentation processes belong to *Lactobacillus* species, which synthesize many flavour-active substances and lactic acid and are considered probiotic microorganisms known to support intestinal microbiota [146]. Fermented beverages are known to be rich in bioactive compounds, such as immune globulin peptides and the bioactive hormone cytokinin [19,71,82,147,148]. The use of a wide variety of microorganisms and yeasts is implied in the production of NFCBs by both traditional and modern means.

Largely responsible for the fermentation process are the indigenous microbiota present on the substrate or which can be added artificially as a culture. Basically, there are four main fermentation processes, which include alcohol production, lactic acid development, acetic acid production, and alkaline fermentation [19]. The modern industry for fermentative foods and beverages is innovative, given that it currently employs thermophilic fermentation, DNA technologies, molecular devices, designed starter cultures, genetic engineering, etc. Recombinant DNA technology has provided new insights for enhancing the product quality by designing tailor-made starter cultures that perform better than those found naturally [60]. For example, Basinskiene et al. used *Lactobacillus sakei* KTU05-6 and *Pediococcus pentosaceus* KTU05-10 to ferment extruded rye for obtaining *kvass*, a traditional Lithuanian NFCB. They showed that the pH of the beverages fermented by LAB reached lower values compared to yeast fermentation and a they had a higher amount of organic acids. Innovative technology was applied, such as xylanolytic enzymes and antimicrobial active LAB, to improve the product's functional properties [71].

Functional cereal fermented beverages are becoming more attractive as they represent healthy alternatives for lactose intolerant consumers and for those who avoid certain allergens. For example, a study describes the production of kefir-like riboflavin-enriched beverages based on oat, maize, and barley flours. To obtain this beverage, riboflavin-producing Andean LAB were used, consisting of five *Lactobacillus plantarum* strains and two *Leuconostoc mesenteroides* strains [149].

As previously discussed, ensuring product safety, although difficult at times, is a crucial step in the production of traditional beverages where the fermentation process is spontaneous. Thus, microorganisms found on the brewers' skin, hair, and clothes can alter the product safety, therefore, high standards of hygiene are mandatory. Nevertheless, through the identification of bacteria strains and yeasts, the safety and quality status of the fermented beverages are guaranteed. A safety case study was conducted concerning "*obushera*", a Ugandan traditional fermented cereal beverage where important steps, such as pasteurization and ensuring water quality, are in the loop to ensure product safety and quality, as pathogens can also change the product's sensory characteristics [80].

Superior results can be obtained in the production of NFCBs by better understanding the interaction of microorganisms in the fermented substrates. The beneficial outcomes of controlling microorganisms are mirrored in product safety, increased shelf life, improved nutrient contents and availability, palatability, and enhanced sensory traits. For example, it was shown that *S. cerevisiae* improves LAB growth by transmitting essential metabolites, such as pyruvate, amino acids, and vitamins, while it uses some metabolites produced by LAB as carbon sources [150]. Moreover, Salari et al. concluded in their study that malt and *L. delbrueckii* were the best substrate and lactic strain for producing a functional beverage with the highest cell viability (1.2 <sup>×</sup> 106 cfu/mL after 4 weeks) [151].

#### Probiotics

Non-alcoholic fermented cereal-based beverages contain a wide range of diverse probiotics, depending on their cereal substrate and overall production methods. The processing method should assure the stability of the bacterial composition in order for the final product to possess probiotic functionality [152]. Still, the threshold to declare a beverage a probiotic one must be higher than 10<sup>7</sup> CFU/mL. Moreover, not all lactic acid bacteria possess probiotic activity. *L. rhamnosus*, also present in NFCBs (Table 1), was efficient in treatment and prevention of gastrointestinal disease [152]. The traditional Romanian NFCB, *bors,* , has been consumed since ancient times as a gastric remedy. Several potentially probiotic bacteria—*L. casei*, *L. plantarum*, *L. brevis*, and *L. fermentum*—were isolated from bors, , explaining the traditional consumption of this beverage for curative purposes [56].

The predominant microorganisms in the spontaneous fermentation of the African *mahewu*, a non-alcoholic sour beverage made from corn meal and consumed in Africa and some Arabian Gulf countries, belong to *Lactococcus lactis* subsp. *lactis*. On the other hand, the microbiota identification of Bulgarian boza shows that it mainly consists of lactic acid bacteria and yeasts, such as *Lactobacillus plantarum*, *L. acidophilus*, *L. fermentum*, *L. coprophilus*, *Leuconostoc ra*ffi*nolactis*, *Ln. mesenteroides*, and *Ln. brevis* and *Saccharomyces cerevisiae*, *Candida tropicalis*, *C. glabrata*, *Geotrichum penicillatum*, and *G. candidum*, respectively [19].

In the case of Turkish *boza*, the use of LAB and yeast isolates as starter cultures allows for controlled fermentation studies to be carried out. The selection of proper strains with probiotic and antimicrobial properties enhances the functional properties of *boza* [59].

As seen in the case of African countries, the primary challenge for the development and use of fermented cereal-based probiotic beverages is the common lack of knowledge regarding the health and nutritional benefits of such foods and beverages. Insufficient common knowledge on probiotics and their benefits creates a sense of scepticism among consumers. Moreover, there is also an imperative need to ensure proper facilities for probiotic starter cultures, given that through spontaneous fermentation, the organoleptic and functional qualities of the resulting products are variable [153].

#### *2.4. The Nutritional and Bioactive Composition of Commonly Consumed NFCBs*

Consumers are aware of the importance of maintaining a strong immune system to prevent illnesses and they are actively looking for products which can help maintain their health status and alleviate health problems. It has been scientifically proven that probiotics isolated from functional beverages boost the immune system [154], and that, along with prebiotics, they are able to improve the intestinal homeostasis, immunomodulating ability, and general health of the host [27].

Looking at past publications, it is shown how fermented beverages have transitioned from traditional natural fermented products to beverages formulated with functional ingredients meant to offer cardiovascular health benefits, and then to functional fermented drinks which improve gastrointestinal health, which could then evolve into fermented products containing specific bioactive nanoparticles [134].

NFCBs are receiving increased attention from researchers and consumers more recently due to their proven probiotic characteristics and disease prevention perspectives [17]. The perceived health outcomes of fermented beverages are strongly related to the microbial content and implicit improvement of gastrointestinal health [38]. Moreover, non-alcoholic fermented beverages offer a sense of wellbeing, as they stimulate the metabolic system [126].

Fermented cereal-based foods, including NFCBs, are a potential source of new functional lactic acid bacteria species besides various nutrients and bioactive compounds, with beneficial effects on human health [56]. Furthermore, as briefly mentioned previously, NFCBs are healthy alternatives to the traditionally consumed food probiotics of dairy origin (e.g., yogurt, kefir, etc.), especially for people with lactose intolerance and milk protein allergies [27].

Given the functional components of fermented beverages and their bioactive compounds released through fermentation by cultures, NFCBs have been linked with many potential and some proven health benefits and actions on digestive, endocrine, cardiovascular, immune, and nervous levels [155]. They present beneficial actions for vital body functions and contribute to the prevention and reduction of risk factors for various diseases [19,23]. NFCBs are rich sources of minerals, vitamins, fibre, flavonoids, phenolic compounds, antioxidants, omega-3 fatty acids, plant extracts, sterols/stanols, amino acids, and biopeptides, among others, which could also protect from oxidative stress and inflammation diseases [17,134]. The presence of numerous valuable compounds in NFCBs grants several health benefits upon consumption, as presented in Figure 3.

**Figure 3.** Health benefits associated with NFCB consumption.

Considering group B vitamins, these are unequally distributed in grain tissues. What counts the most in terms of their potential functionality in cereal-based food and beverages is their biological availability. During thermal food processing, these vitamins are destroyed almost completely, however, lactic acid fermentation represents a great tool for food industrialists interested in developing novel vitamin-fortified products. For example, Capozzi et al. (2012) pointed out the ability of many lactic acid bacteria strains, such as *Lactobacillus delbrueckii, Lactobacillus plantarum, Lactobacillus rhamnosus, Lactobacillus fermentum*, and *Pediococcus lactis*, reported in the composition of several NFCBs (Table 1), to biosynthesize riboflavin [156].

A new concept attributed to non-alcoholic cereal-based beverages is that they are healthy drinks with important impacts on human health. It has been shown that consuming NFCBs leads to liver function improvement, increased levels of lactobacilli and bifidobacteria in the intestinal microbiota [157], balanced gut microbiota, and the prevention of bacterial translocation with reduced incidences of nosocomial infections [27]. Furthermore, because non-alcoholic fermented beverages contain several LAB metabolites, their consumption confers bactericidal, bacteriolytic, and bacteriostatic properties, resulting in therapeutic effects at a digestive level. Antimicrobial compounds identified in functional beverages exhibited activities against several Gram-positive and Gram-negative bacteria and against yeasts and moulds [17]. The functional compounds and potential or proven health benefits of NFCBs are presented in Table 1.

As previously discussed, the potential benefits of NFCBs can be ensured in a safe and non-contaminated production environment to prevent the risk of product spoilage. Moreover, there are diverse nutritional and bioactive components of NFCBs, depending on the cereal substrate, microorganisms, and fermentation parameters, and on the technological process followed.

Currently, novel NFCBs are designed based on traditional recipes for delivering improved functional fermented beverages. Such an example is "*amahewu*" or "*mahewu*", a southern African LAB-fermented non-alcoholic maize-based beverage, which is deficient in vitamin A. In a study published in 2015, the traditional drink was redesigned using provitamin A-biofortified maize in exchange for the traditionally used white maize and it resulted in a functional beverage addressing vitamin A deficiency, a concerning health problem in sub-Saharan Africa [43].

#### 2.4.1. Phenolic Compounds and Antioxidant Activity

Cereal grains and germs contain various major phytochemicals, such as phenolic acids, flavones, phytic acid, flavonoids, coumarins, and terpenes. The bran layers of cereal grains are relatively rich in water soluble and liposoluble antioxidants. Oats include phenolic acids, flavonoids, tocopherols, tocotrienols, and avenanthramides. On the other hand, insoluble fibre of wheat bran contains about 0.5%–1% phenolics and most phenolic compounds, like ferulic acid, are abundant in whole grains [158].

Depending on the cereal substrate, fermentation type, and other added ingredients, each NFCB can have quite different compositions. For example, the phenolic profile of brown beer vinegar indicates that such a fermented beverage is a rich source of phenolic compounds (661 mg GAE/L) and antioxidants, with 30% increased antioxidant activity after acetic fermentation. The same trend was observed regarding some of the drink's individual phenolic compounds. With cohumulone I (4.44 mg CE/L), cohumulone II (6.58 mg/L), 8-prenylnaringenin (2.33 mg CE/L), 6- prenylnaringenin (1.86 CE/L), humulone (5.62 CE/L), and isohumulone (4.14 CE/L) [159], this non-alcoholic fermented beverage can be considered an alternative source of valuable compounds, which could also be part of specific diets.

Fermentation parameters are of high importance in what concerns an NFCB's antioxidant profile. For example, a fermentation temperature of 24◦ C was considered a positive factor influencing the antioxidant activity of *bors,* , a Romanian traditional wheat bran-based fermented beverage [55]. The antioxidant activity ranged from 3.05 to 8.52 mmol/L Trolox.

Some of the ancient traditional fermented beverages are currently produced at industrial scale. Still, the original recipes and the nutritional components are being altered because of thermal treatments

applied to obtain product safety and stability. The addition of natural starters allowed for the increase in phenolic contents and the enhancement of final antioxidant activity. The same trend considering phenolic content between fermentation stages was found in the case of *bors,* [56]. Depending on the processing operations applied, the phenolic compounds ranged considerably (4-hydroxybenzoic acid: 0.9–5.9 mg/L; vanillic acid: 0.6–3.2 mg/L; syringic acid: 0.5–2.5 mg/L; p–coumaric acid: 0.5–1.5 mg/L; sinapic acid: 0.6–2.7 mg/L; ferulic acid: 13.7–47.8 mg/L).

#### 2.4.2. Amino Acids

Although cereals pose few challenges from a nutritional standpoint, especially with regard to the increased starch content upon cooking, the limited amino acid contents of their protein fractions, or the reduced bioavailability of their zinc and iron contents, there are already some solutions meant to correct and further enhance the nutritional status of cereal-based foods and beverages [160].

For example, in a study conducted on *bors,* , an increase in amino acid content was found from one fermentation stage to another. The same study stated that the LAB fermentation of cereals improves the protein quality as well as the level of certain free amino acids by enhanced endogenous proteolysis and/or microbial action. Among the identified amino acids, there were isoleucine and threonine [56].

In general, the natural fermentation of cereals allows for amino acids to be synthesised, while it also helps to enhance the availability of group B vitamins. Fermentation processes also offer suitable pH conditions for phytate enzymatic degradation. Such a reduction in phytate may increase the amount of soluble iron, zinc, and calcium several fold, correcting the nutritional status of cereal-based beverages and foods [19].

#### **3. Future Perspectives to Enhance the Functional Properties of NFCBs**

#### *3.1. Consumers' Preferences and Requirements*

The way we perceive healthiness in foods varies between cultures and is reflected in the consumers' familiarity with health-related information [161]. Moreover, modern-day buyers might show increased interest in foods which can improve well-being [162], reduce the risk of developing illnesses, and satisfy nutrition-related conditions such as food intolerances and allergies. Currently, consumers approach the concept of wellness with a holistic view and are becoming more health conscious, particularly due to the growing incidence of diseases such as type 2 diabetes, coronary heart disease, cancer, and obesity [163]. There is a heightened demand for food products which are nutritious, functional, attractive, with clean labels, and which are ready to eat. Considering such attributes, the beverage sector seems to be one of the most suitable for addressing customers' demands and for increasing the rate of functional product consumption. Market data indicate that functional food products are among people's choices even when economic issues arise. The global functional beverage market has increased rapidly, and it is expected to grow further, due to their sensory appeal and health-promoting attributes [38].

Studies showed that providing information on labels regarding health-related claims associated with functional beverages determines consumers' preferences for a specific product [164]. Not only health-related claims but also the sensory properties determine consumers' preferences for a specific functional beverage [34]. It was suggested that the importance of specific testing conditions for functional beverages will help product developers to reformulate the product with the proper design of experiments, focusing on the consumers' needs [165], as they will determine the product acceptance [166]. The urgent need for educational campaigns designed with accessible and easily understood wording about the nutritional components of foods has already been pointed out [167]. Although the previously mentioned study was conducted in Brazil, insufficient knowledge about nutritional facts for food products is a worldwide issue strongly related to the level of education [168].

#### *3.2. Possibilities of Improving the Appeal and Functionality of NFCBs*

The functional beverage market is competitive and driven by product innovation and health awareness trends concerning optimum nutritional diets. The biological activities and the sensory properties of a beverage arise from individual components, along with chemical and physical interactions within the food matrix during processing, storage, ingestion, and digestion [169]. Beverages deliver nutrients, bioactive components, antioxidants, vitamins, minerals, fatty acids, plant extracts, probiotics, prebiotics, and micronutrients [13]. Cereal-based beverages include a complex mix of different polymers, such as proteins, polyphenols, and polysaccharides. These polymers affect the sensory perception of beverages in terms of mouthfeel, depending on their substance properties [112].

Several strategies have been set to intensify the production, availability, accessibility, and consumption of beverages rich in bioactive compounds. These include combining cereals with pseudocereals (e.g., quinoa, amaranth, etc.), legumes, vegetables extracts, fruits, aromatic plants, and herbs to improve the quality of the final food product [19,56,170,171]. Furthermore, mixing cereals with legumes could improve protein quality [172]. Fruit juices (e.g., apple, pineapple, mango, orange, lemon, peach, lychee, and strawberry) are considered as health-promoting foods and are an important basis of enrichment on which to append an extra functional constituent that can significantly augment the appeal to customers. Cereal beverages are based on grain suspensions. The viscosity, mouthfeel, and sweetness of the drink can be adjusted to the consumers' tastes using enzyme compounds. Wort can be mixed with various plant- and fruit-based juices to obtain a beverage rich in dietary fibre suitable even for athletes and sport amateurs. Moreover, innovative flavours can be obtained by fermenting the extract with specific microorganisms. As previously mentioned, fruits were also employed in the overall production scheme of fermented cereal beverages, as these are the added sugar necessary for initiating the fermentation process. Chemical analyses of ancient organics absorbed into pottery jars from the early Neolithic village of Jiahu, Henan Province, China have revealed that a mixed fermented beverage of rice, honey, and fruit (hawthorn fruit and or grape) was being produced as early as the seventh millennium B.C. [173]. Such findings inspire the development of new functional products, such as yogurt-like beverages made of a mixture of rice, barley, emmer wheat, oats, soy, and grape must [174] or of other milk-based functional drinks using fermented plant juices [175,176].

Plants are valued for their nutrients such as vitamins, dietary fibre, antioxidants, and flavonoids, which have shown nutritious and health-promoting properties [134]. Bioactive compounds derived from fruits and vegetables can be good vehicles for probiotics, prebiotics, and synbiotics (e.g., fermented cereal beverages) [57,58]. Spices such as tarhana herb, mint, and thyme are mixed with wheat flour, yogurt, vegetables, and herbs to produce a traditional Turkish fermented food called tarhana [177,178] and improve taste, aroma, and other profile characteristics. Moreover, it is suggested that adding herbs with proven beneficial compounds to otherwise traditional fermented soft drinks can augment the nutritional profile of fermented cereal beverages and add therapeutic potential [179]. Among the suggested herbs, we can find echinacea (*Echinacea angustifolia*) for antibiotic action and immune system support, ginkgo (*Ginkgo biloba*) for enhancing memory and alertness, guarana (*Paulina cupana*) for improved cognitive performance, kava (*Piper methysticum*) for stress reduction and mental balance, and St John's Wort (*Hypericum perforatum*) for anxiety reduction [180]. Vegetables such as beets, tomatoes, carrots, and cabbage can also be included in the production of some functional NFCBs, as they provide supplementary fermentable substrates and nutrients, and act as prebiotics in the final product [181]. *Shalgam* is a traditional Turkish NFCB made of turnip bulb, purple carrot, salt, sourdough, bulgur, or bulgur flour [182], and is thought to regulate the digestive system's pH and to act as an antiseptic agent. The results of a research study conducted on the co-culture probiotic fermentation of a protein-enriched cereal medium suggests that plant protein may be exploited for achieving protein supplementation of NFCBs. Legumes like chickpeas, individually or in combination with cereals, can provide a good substrate for probiotic microorganism production [145]. Additionally, the composition of legumes, vegetables, and fruits is known to be rich in protein, phytochemicals, dietary fibre, vitamins, and other micronutrients beneficial for human health [183].

Currently, concepts like "green consumerism" and "minimally processed foods" are on the rise, as consumers prefer food lacking synthetic additives. Natural compounds that can be added to functional beverages, such as essential oils and plant extracts including rosemary, peppermint, bay, basil, tea tree, celery seed, and fennel with antimicrobial activity, might represent an alternative to synthetic preservatives. The challenge is to define the optimal dosage of such compounds to have a positive impact on the product's final nutritional status, sensory properties, and consumer acceptance [12].

#### *3.3. Perspectives for Future NFCBs*

Due to the increasing prevalence of lactose intolerance and milk protein allergies, as well as the general aim of controlling cholesterol intake, and along with the growing interest in vegetarianism, research efforts have been focused on the feasibility of using cereals as fermentation substrates for the development of probiotic, prebiotic, and synbiotic beverages [27,113,184]. In the production of various non-alcoholic drinks, probiotic lactic acid bacteria are used to boost the product's functional value [143]. Although ensuring nutritive compounds within NFCB is a crucial step for product development, flavour improvement remains one of the main challenges for the development of LAB-fermented beverages obtained from cereal-based substrates [185].

It is a fact that improved and more appealing NFCB sensory properties would help consumers shift their interest intensively towards such healthy products. The efficacy of probiotics in the treatment of bowel disorders, the prevention of antibiotic-associated diarrhoea, and the improvement of lactose metabolism has been proven [186]. It has been also concluded that both fermentation and acidification with lactic acid have the potential to improve the nutritional quality of cereal-based foods as a method to combat protein malnutrition and iron and zinc deficiencies [187].

Biotechnological processes, such as malting and lactic acid fermentation, are recommended for producing functional beverages with increased contents of functional bioactive components [188]. There are numerous important benefits of enhancing NFCBs through biotechnological processes and it is worth mentioning the reduction of phytates through LAB-fermentation, which in turn leads to the increased absorption of Fe and other minerals [189]. Due to cereal fermentation, quantitative and qualitative changes take place in small molecules, ensuring the high bioavailability of macro- and micronutrients [190,191]. As mentioned earlier, prebiotics and dietary fibre can be added to fermented drinks to heighten their nutritive and functional contents. For example, it was shown that the addition of soy fibre not only improves *Lactococcus lactis* counts but also enhances the beverage characteristics regarding acidity, viscosity, and syneresis [192].

Currently, emerging techniques, such as nanotechnology, are in discussion to enhance the nutrient composition and functionality of NFCBs. However, continuous research and technological improvements are required to better design NFCBs and ensure product safety. It is also crucial to improve the quality of the main ingredients and it is imperative to integrate food safety management systems for industrial scale production. The proposed solutions include the development of new production technologies for obtaining functional NFCBs by extending the spectrum of raw materials used and applying new biotechnological resources (enzymes and lactic acid bacteria) [71].

#### **4. Conclusions**

Science and technology have the potential to produce superior functional NFCBs and to deliver consistent product quality, to improve shelf life, and to enhance nutritional values to finally meet the consumers' demands. Past research studies have proven that lactic [55] and acetic [159] fermentation enhance the nutrient content and their bioaccessibility in the case of cereal-based fermented beverages. Adding probiotics and prebiotics to a beverage is perhaps one of the most convenient ways of turning it into a functional beverage.

The research efforts on the enrichment of non-alcoholic fermented cereal beverages are still in their early stages but appear to be more promising than ever. Upcoming studies should focus on traditional non-alcoholic fermented beverages around the world to identify new potential probiotic microorganisms and starter cultures, new ingredients as potential substrates, and to elucidate the contributions of microorganisms and enzymes to the fermentation process.

Knowledge about processing applications and bacterial strains is essential to control the production methods and to design proper mixes of microorganisms, ensuring product safety. Afterwards, starter cultures with expected outcomes can be used for the industrial production of standard-quality fermented beverages with functional attributes.

The effect of advanced technologies on NFCB functional properties during processing requires additional studies to ensure that these technologies can prevent the loss of product quality and nutritive compounds. Moreover, for both scientific and industrial actors, the main challenge is to manage the large-scale production of fermented beverages without losing the unique flavours and other properties associated with the original products. Given this, it is highly recommended to explore the sensory properties of NFCBs when obtained from a combination of cereals, legumes, fruits, and plants.

The current review investigates some of the most scientifically documented traditional non-alcoholic beverages from all over the globe, highlighting their functional compounds and associated health benefits upon consumption. Moreover, processing technologies and their outcome were highlighted and discussed from a large-scale production standpoint. NFCBs were also reviewed, concerning fermentation microbiota and product safety, highlighting the need to apply starter cultures to ensure food safety and standard quality. The nutritional and bioactive compositions of commonly consumed NFCBs were reviewed, showing the functional compounds of NFCBs and their associated health benefits.

NFCBs are associated with functional benefits to one's health status, as discussed. However, new studies should be carried out to produce new NFCBs, combining probiotic fermented beverages and products such as fruit juices, vegetables, and cereals, meant to address specific health concerns. Future cereal-based fermented beverages need to be balanced regarding sensory properties, nutritional composition, alcohol content, and resource investments to be even more attractive, healthy, and affordable.

**Author Contributions:** Conceptualization, L.C.S. and M.V.I.; Writing—original draft preparation, M.V.I., L.C.S., C.R.P., O.L.P., T.E.C. and A.B.; Writing—review and editing, L.C.S., M.V.I., A.P. and T.E.C.; Supervision, L.C.S., M.T., A.P. and E.M.; Funding acquisition, L.C.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by two grants, as follows: from the Ministry of Research and Innovation, CNCS-UEFISCDI, project number PN-III-P2-2.1-CI-2018-1462, within PNCDI III; and from the European Foundation for Alcohol Research (ERAB), Belgium, project number Ref EA 15 45.

**Acknowledgments:** The publication was supported by funds from the National Research Development Projects to finance excellence (PFE)-37/2018–2020 granted by the Romanian Ministry of Research and Innovation.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Nutritional Features and Bread-Making Performance of Wholewheat: Does the Milling System Matter?**

### **Maria Ambrogina Pagani 1, Debora Giordano 2, Gaetano Cardone 1, Antonella Pasqualone 3, Maria Cristina Casiraghi 1, Daniela Erba 1, Massimo Blandino 2,\* and Alessandra Marti 1,\***


Received: 27 June 2020; Accepted: 29 July 2020; Published: 1 August 2020

**Abstract:** Despite the interest in stone-milling, there is no information on the potential advantages of using the resultant wholegrain flour (WF) in bread-making. Consequently, nutritional and technological properties of WFs obtained by both stone- (SWF) and roller-milling (RWF) were assessed on four wheat samples, differing in grain hardness and pigment richness. Regardless of the type of wheat, stone-milling led to WFs with a high number of particles ranging in size from 315 to 710 μm), whereas RWFs showed a bimodal distribution with large (>1000 μm) and fine (<250 μm) particles. On average, the milling system did not affect the proximate composition and the bioactive features of WFs. The gluten aggregation kinetics resulted in similar trends for all SWFs, with indices higher than for RWFs. The effect of milling on dough properties (i.e., mixing and leavening) was sample dependent. Overall, SWFs produced more gas, resulting in bread with higher specific volume. Bread crumb from SWF had higher lutein content in the wheat *cv* rich in xanthophylls, while bread from RWF of the blue-grained *cv* had a moderate but significantly higher content in esterified phenolic acids and total anthocyanins. In conclusion, there was no relevant advantage in using stone- as opposed to roller-milling (and vice versa).

**Keywords:** wholewheat flour; stone milling; roller milling; bioactive compounds; bread; pigmented wheat; dough rheology

#### **1. Introduction**

Whole grains have been (and still are in several countries) the most important energy source of mankind. They constitute a valuable and inexpensive source of numerous nutrients and phytochemicals, including fiber, phenolic compounds, minerals, and vitamins, mainly located in the germ and bran regions [1,2]. Although a universally accepted definition of whole grain has yet to be formulated [3], it is widely recognized that a whole grain product must contain bran, germ and endosperm in the same proportions as in the original, native grain [3]. Whatever the official definition is, the relationship between the consumption of whole grain foods and a lower incidence of diseases in the Occident has been suggested by numerous epidemiological studies [4–8].

Thanks to public institutions (governments and health organizations) and extensive promotional campaigns, consumers nowadays are fully aware of the numerous health benefits associated with the consumption of products made from whole grain flour (WF). Consequently, the demand for wholegrain products—in particular for staple foods such as bread and pasta—has been constantly growing [9]. This trend is recognizable even in countries where products made from refined flour have always been preferred for their undisputed superior sensory characteristics.

At present, two milling processes are mainly used to produce WF from wheat: (i) single–stream milling, and (ii) multiple-stream milling with recombination [10]. The former—where stone mills are used—is the world's oldest system for flour production. The stone mill is quite a simple machine, formed by two horizontal and overlapping grinding stones where the upper revolves while the other (bedstone) is stationary [11]. Grains are fed in the gap between the two stones and undergo shear, compression and abrasion forces. The ground fractions stay together during the entire milling process, and they are collected (with an extraction rate of 100%) at the bottom of the bedstone without separation according to particle size or composition [12].

Multiple-stream milling became established in the mid-1800s [11]. According to this process, wheat grains are progressively ground by means of a sequence of cast iron roller mills followed by sieving and sifting of the ground material [13]. Due to differences in composition, bran, germ and endosperm exhibit different friability and breakage patterns, thus facilitating the separation of bran and germ from flour, which derives from the endosperm. The separation of the three regions is enhanced by conditioning the kernels before milling [11,13]. The flour extraction rate is usually about 70% [12]. Even this milling process may lead to a WF if the different "fraction streams"—originating with the repeated grinding and sieving steps—are gathered at the end of the process. The extraction rate of the recombined WF should be about 100%, and the endosperm, germ and bran should be present in the same ratio as in the native whole grain [10]. Roller milling guarantees higher productivity, flexibility and more constant results over time; consequently, this process is highly preferred for industrial applications [13,14]. Moreover, the separation of bran and germ fractions during the process allows further treatments before recombination, such as mild heating and/or bran grinding, improving both technological performance and storage of the WF [13,15].

On the other hand, stone milling has been rediscovered in recent years by small farmers and bread-making artisans as it requires relatively low capital inputs [16]. Certainly, the simplicity and cheapness of the process (only one operation; grain tempering is not mandatory) also makes it suitable for household milling, favoring whole grain consumption for some segments of population [14]. Nevertheless, despite these advantages, stone milling not only is characterized by a lower yield but may also worsen the rheological properties of WF due to varying degrees of heat development [16,17] according to the type of small-scale mills used [14]. The negative effects of stone milling can be reduced by choosing suitable wheat *cultivars (cvs)* and/or farming procedures, as suggested by Gélinas et al. [18].

Despite the great interest in stone milling and increasing demands for WF-produced food, few studies have compared the effects of stone and roller milling on the same wheat varieties to identify and evaluate analogies and differences between the two kinds of WF [15,19]. Some research has found similarities in proximate chemical composition and phenolic profile of WF from stone milling (SWF) and roller milling (RWF) [17,20,21]. On the contrary, by promoting high heat development due to friction, compression and shear phenomena, stone milling leads to a significant loss in aminoacids and unsaturated fatty acids [17]. This worsening can affect not only the nutritional aspects of flour but also its technological properties, due to a partial denaturation of gluten proteins, relevant starch damage and/or differences in particle size distribution [15,17,20]. Specifically, SWFs exhibited lower pasting properties, higher water absorption, and lower stability compared to RWFs [14,21].

Regarding the relationship between milling process and bread quality, the data are not in agreement. Liu et al. [21] emphasized that RWFs exhibited the best steam-bread making performance, while according to Kihlberg et al. [22], bread produced using RWF was characterized by regular shape but higher compactness. Nevertheless, differences in bread volume in WFs obtained by the two types of milling can be resolved by adopting sourdough fermentation [14].

The few studies published so far on the effects of the milling process on WF and bread properties obtained by the same type of wheat presented no univocal findings and prompted our study to aim at giving a complete overview of behavior–from milling to baking–of WFs from four types of common wheat (*T. aestivum* L.), all belonging to the bread-making class and characterized by high protein content. Beyond these common traits, the four wheat samples differed in protein strength and physical structure of kernels–hard, semi-hard and soft endosperm–a property that highly impacts milling behavior and performance [23]. Moreover, two pigmented varieties were distinguished by their richness in bioactive compounds, differently located in the kernel (polyphenols and anthocyanins in the external layers and xanthophylls in the endosperm), whose retention has to be carefully ensured along the whole transformation chain.

#### **2. Materials and Methods**

#### *2.1. Wheat Samples*

The present study analyzed the SWF and RWF derived from four common wheat (*Triticum aestivum* L.) samples.

Table 1 compares the main information regarding the wheat samples. The two pigmented wheat *cvs* were chosen for their interesting bioactive compound content [24].


**Table 1.** Main information of common wheat samples.

TW, test weight; TKW, thousand kernel weight.

#### *2.2. Milling Procedures*

An aliquot (70 kg) of each sample was stone-milled (Molino Tomatis; Niella Tanaro, Italy), without kernel conditioning through a single-stream process, without any sifting, to produce SWF. The stone mill used was made of two discs of natural French burrstones, from the district of La Ferté-sous-Jouarre (France). The stones, 1.3 m in diameter, are arranged on a vertical axis, with the upper stone rotating at 70 rpm. The distance between the two stones at their centers was about 1 mm. The opposing surfaces of the stones were subdivided into 10 harps and grooved from the center to the circumference. After the proper cleaning operation, 40 kg of each sample was milled and the flour discarded, in order to reach stable operative conditions (speed, temperature) of the stone mill before sampling. The remaining 30 kg aliquot was then milled, carefully mixed to favor fraction blending and subsampled for rheological and chemical analysis. Another aliquot (10 kg) of grains was conditioned (16 h at 20 ◦C) till reaching 16% moisture, and then submitted to multiple-stream milling by using a roller mill (Bona lab-scale mill, Labormill 4RB, Monza, Italy). This lab scale mill, equipped with 4 rollers, 0.07 m in diameter and 0.20 m wide were horizontally arranged can simulate the industrial milling process e by separating the different parts of the wheat kernel: the external coarse bran, the intermediate layers (middlings) and the inner endosperm (refined flour). The kernels are milled in three steps: a first break phase passing through fluted rolls and two reducing phases with smooth rolls. Gap settings of the break and reducing rolls were adjusted to 0.4 mm, 0.2 mm and 0.05 mm, respectively. The feed rate was adjusted to about 8 kg/h. The three milling fractions obtained (refined flour, middlings and coarse bran) were recombined to produce the RWF. WF yield was about 100% for both milling processes. Refined flours obtained from roller milling were produced for bread production and functional characterization (see Section 2.3.2).

Samples were stored in a polypropylene bag at 4 ◦C and under vacuum until used. Samples were used after two weeks of resting.

#### *2.3. Methods*

#### 2.3.1. Particle Size Distribution

Particle size distribution of SWFs and RWFs was assessed in single by means of an automatic mechanical sifter (AS 200, Retsch GmbH, Düsseldorf, Germany) equipped with 8 sieves: 1000 μm, 800 μm, 710 μm, 500 μm, 425 μm, 315 μm, 250 μm, 160 μm. The test was carried out on 100 g of WF, setting 1.5 mm of oscillation for 5 min.

#### 2.3.2. Chemical Analysis

The moisture content of WFs was determined by means of thermo-balance (MA 210.R, Radwag Wagi Elektroniczne, Radom, Poland), by drying the sample at 130 ◦C until its weight did not change by 1 mg in 10 s. Ash (AACC 08-01.01) and damaged starch (AACC 76-31.01) contents were evaluated according to AACC standard methods [25]. The amounts of protein (AOAC 34.01.05 No. 925.31), fat (AOAC 31.04.02 No. 963.15), total (AOAC 31.04.02 No. 985.29), soluble and insoluble dietary fiber contents were measured according to AOAC standard methods (AOAC 31.04.02 No. 991.43) [26]. Total and water soluble arabinoxylans were evaluated as previously reported by Manini et al. [27]. Total starch content was calculated as what remained after moisture, protein, ash, fat and total fiber determinations had been accounted for. α-amylase activity was determined according to the AACC method 22-02.01 [25].

The SWF and RWF from each *cv* were analysed for soluble phenolic acids (SPAs) and cell wall-bound phenolic acids (CWBPAs) and total antioxidant capacity. Flours from Bona Vita and Skorpion *cvs* were further analysed for xanthophylls and total anthocyanin content (TAC), respectively. Extraction of phenolic acids and xanthophylls and their quantification by means of RP-HPLC was performed as reported by Giordano et al. [28]. The antioxidant capacity was determined by means of FRAP (Ferric Reducing Antioxidant Power) and the ABTS [2,2 -Azino-bis(3-ethylbenzthiazoline-6-sulfonic acid)] assays adapted into QUENCHER method as described by Serpen et al. [29]. The TAC was determined on samples (1 g) extracted using 8 mL of ethanol acidified with 1 N HCl (85:15, *v*/*v*) for 30 min. The absorbance was measured after centrifugation at 20,800× *g* for 2 min at 540 nm as reported by Siebenhandl et al. [30]. TAC was expressed as mg cyanidin-3-*O*-glucoside (Cy-3-glc) equivalents/kg of sample (db).

Moisture, ash, protein, fat, soluble and insoluble dietary fiber, total and water soluble arabinoxylans, and amylase activity were evaluated in duplicate, whereas damaged starch content and bioactive compounds in triplicate.

#### 2.3.3. Rheological Properties

Gluten aggregation properties were measured in triplicate by means of the GlutoPeak (Brabender GmbH and Co KG, Duisburg, Germany) device, by using distilled water as solvent; 9:10 flour:solvent ratio, 2750 rpm as paddle speed rate and 35 ◦C as temperature of water circulating bath. The main indices considered were: (i) maximum torque (the highest consistency value reached, evaluated in GlutoPeak Units, GPU); (ii) peak maximum time (time required to achieve the maximum torque, evaluated in seconds, s); and (iii) aggregation energy (the area under the curve between 15 s before and 15 s after the maximum torque; evaluated in GlutoPeak Equivalents, GPE).

Mixing properties were evaluated in duplicate by using the Farinograph-E (Brabender GmbH and Co KG, Duisburg, Germany) device equipped with a 50 g mixing bowl, according to the standard method ICC 115/1 [31].

Dough leavening properties were determined in single by means of the Rheofermentometer F4 (Chopin Technologies, Villeneuve La Garenne CEDEX, France) device on 315 g of dough at 30 ◦C for 4.5 h. Dough samples were prepared by mixing flour, in a spiral mixer (Artisan, KitchenAid®, Whirlpool, Benton Harbor, USA) with fresh yeast (2 g/100 g of flour; Carrefour, Annecy, France) and

salt (NaCl; 1 g/100 g of flour; Candor®, Com-Sal s.r.l., Pesaro, Italy). The amount of tap water added, and the kneading time used were previously determined by means of the farinographic test (Table 2).


**Table 2.** Amount of water, kneading and leavening times used in bread-making for each sample.

#### 2.3.4. Bread Preparation and Characterization

Dough samples were prepared from either WFs or refined flours (obtained by roller milling) in the same conditions as those reported for the rheofermentographic test. After kneading, the samples were left to rest for 15 min, then divided into portions of 250 g, modelled in cylindrical shapes, put into aluminum pans (length: 12.5 cm; width: 6 cm; height: 5 cm) and leavened at 30 ◦C and 70% relative humidity in a combined proofer oven (Self-Cooking Center®, Rational International AG, Heerbrugg, Switzerland), until the dough exceeded the top of the baking pans by about 2.5 cm. Then, the leavened dough samples were baked (Self-Cooking Center®, Rational International AG, Heerbrugg, Switzerland) at 200 ◦C for 30 min (85% relative humidity). One baking test was performed, yielding two loaves for each sample.

Two hours after baking, the loaves were characterized in terms of specific volume, through the ratio between the bread volume—evaluated according to the standard method AACC 10-05.01 [25]– and its weight. Bread height was determined by measuring the maximum height of the slice by means of image analysis (Image ProPlus, v6; Media Cybernetics, Inc., Rockville, MD, USA). Loaf specific volume was determined on two loaves, while bread height was evaluated on three central slices of each bread, for a total of six replicates.

SPAs, CWBPAs, total antioxidant capacity, and xanthophylls (only for flour from Bona Vita *cv*) and TAC (only for flour from Skorpion *cv*) were carried out as described in Section 2.3.2, on bread samples obtained from SWF, RWF and refined flours of pigmented *cvs*, which provide a higher AC and the possibility to investigate the impact of milling method also on the bioactive compounds responsible for flour color. Before analysis, bread samples were ground to a fine powder (particle size <300 μm) with a Cyclotec 1093 sample mill (Foss, Padova, Italy). The same grinding procedure was carried out for bread crust (about 3.5 mm thick), and crumb after freeze-drying (−80 ◦C for 72 h; Alpha 1–2 LD plus; Deltek s.r.l., Naples, Italy). All samples were stored at −25 ◦C. Bioactive compounds in bread were evaluated in triplicate.

#### *2.4. Statistical Analysis*

Statistical analysis (t-test) was carried out in order to identify significant differences between SWF and RWF from the same *cv* by using Statgraphics Plus 5.1 (StatPoint Inc., Warrenton, CT, USA) at the 1% (\* *p* < 0.01) significance level. Data obtained from the functional characterization of flours and breads were analyzed separately for each *cv* using analysis of variance (ANOVA), by comparing raw material (flour), bread crust and bread crumb obtained from refined flour, SWF and RWF. A 0.01 threshold was used to reject the null hypothesis. The REGW-F test was performed for multiple comparisons, by using SPSS for Windows statistical package, Version 24.0 (SPSS Inc., Chicago, IL, USA).

#### **3. Results**

#### *3.1. Particle Size Distribution*

Regardless of the type of wheat, stone milling led to WF with a higher number of particles from 315 to 710 μm (from 32.2% for CWRS to 45.9% for Bona Vita *cv*) (Figure 1). On the contrary, in WF obtained by roller milling (i.e., RWFs) such "intermediate" fractions extended from 7% for CWRS to 22% for Bolero *cv*. In addition, recombination of roller milling fractions led to a peculiar, bimodal distribution with large (>1000 μm, mainly composed by bran) and fine (<250 μm) particles. Indeed, the large particles extended from 15% to 25% (except for CWRS) in RWFs, whereas they did not exceed 8% in SWFs. On the other hand, fine particles represented more than 50% (*w*/*w*) in RWFs, reaching 75% for samples with the highest kernel hardness (i.e., CWRS); whereas this fraction was only 30% of the mass in SWFs.

**Figure 1.** Effect of stone milling (black bars) and roller milling (dash bars) on the particle size distribution of whole grain flours from Bolero *cv* (**a**), CWRS (**b**), Bona Vita *cv* (**c**) and Skorpion *cv* (**d**). CWRS: commercial Canada Western Red Spring Wheat.

#### *3.2. Chemical Composition and Bioactive Compounds in WFs*

The milling system did not significantly affect the chemical composition of WFs, except for the moisture content that decreased by 17% and 11% when Bona Vita and Skorpion *cvs* were milled using the stone milling system (Table 3). Although the fiber content did not change, roller milling caused a slight but significant decrease in the total arabinoxylan content of both CWRS and Bona Vita *cv* (about 19% and 12%, respectively). In contrast, a significant increase in this parameter was found in RWF from Skorpion *cv* (about 70%). As regards the water soluble arabinoxylan fraction, the roller milling system resulted in an increase of about 63% for only Bona Vita *cv* (Table 3). The effect of milling on the bioactive compound concentration was sample and compound dependent. RWFs resulted in a significantly higher content of CWBPAs, CWB-ferulic acid only in Bona Vita and Skorpion *cvs*. A similar trend was observed for CWB-sinapic acid, with a significant difference observed in the CWRS. RWF also showed a significant higher TAC in the blue-grained Skorpion *cv* and a higher ACFRAP in the yellow-grained Bona Vita *cv*. Conversely, no difference was observed for SPAs in any of the samples and for xanthophyll in Bona Vita *cv*.


*Foods* **2020** , *9*, 1035

deviation. The asterisks indicate significant differences between the means of stone- and roller-milled samples of each *cv* (\* *p* < 0.01). The absence of asterisk indicates a not significant

difference. n.d.: not determined.

#### *3.3. Gluten Aggregation Properties*

The gluten aggregation profiles of the WFs were consistent in showing a slower aggregation when the samples were milled by stone milling (Figure 2). Specifically, the peak maximum time was significantly higher in SWFs than RWFs (86 vs. 71 s, 86 vs. 64 s, 101 vs. 79 s, and 72 vs. 58 s for Bolero, CWRS, Bona Vita, and Skorpion *cvs*, respectively).

**Figure 2.** Effect of stone milling (solid line) and roller milling (dash line) on gluten aggregation properties, assessed by GlutoPeak®, of whole grain flours from Bolero *cv* (**a**), CWRS (**b**), Bona Vita *cv* (**c**) and Skorpion *cv* (**d**). CWRS: commercial Canada Western Red Spring Wheat; GPU: GlutoPeak Units.

In the case of maximum torque and aggregation energy, the effect of the milling system depended on the type of wheat. Specifically, SWFs showed a significantly higher maximum torque than RWFs (55.1 vs. 43.7 GPU and 46.2 vs. 37.3 GPU for Bolero and Bona Vita *cvs*, respectively), as well as significantly higher aggregation energy (1220 vs. 1001 GPE, 1099 vs. 864 GPE, and 1080 vs. 908 GPE, for Bolero, Bona Vita and Skorpion *cvs*, respectively) (Figure 2a,c,d). On the contrary, significantly lower values for both indices were found in the case of CWRS (52.2 vs. 59.7 GPU and 1216 vs. 1387 GPE for maximum torque and aggregation energy, respectively) (Figure 2b).

#### *3.4. Mixing Properties*

As regards the mixing properties (Figure 3), stone milling resulted in a significantly higher water absorption value for Bolero *cv* (61.7 vs. 64.2%, RWF vs. SWF) and Bona Vita *cv* (64.7 vs. 68.1% RWF vs. SWF). On the contrary, the milling system did not significantly affect the dough development time.

**Figure 3.** Effect of stone milling (solid line) and roller-milling (dash line) on the mixing properties, assessed by Farinograph®, of whole grain flours from Bolero *cv* (**a**), CWRS (**b**), Bona Vita *cv* (**c**) and Skorpion *cv* (**d**). CWRS: commercial Canada Western Red Spring Wheat; FU: Farinographic Units.

In addition, stone milling led to a significant increase in dough stability only in the case of Bolero *cv* (from 4.6 to 6.2 min for RWF and SWF). Finally, stone milling significantly influenced mixing resistance (evaluated as the degree of softening) only for CWRS and Bona Vita *cv*. Specifically, a higher degree of softening was observed in SWF (88 FU) compared to RWF (65 FU) in the case of CWRS, while this parameter decreased (from 138 to 119 FU, for RWF and SWF, respectively) in the case of Bona Vita *cv*.

#### *3.5. Leavening Properties*

The greatest impact of stone milling on dough leavening properties was observed for CWRS and Bolero cv, but each sample exhibited a different trend (Figure 4). Specifically, in the case of CWRS, stone milling caused a decrease in maximum dough development (from 40 to 33 mm), instead an opposite trend was found for Skorpion cv (from 37 to 42 mm). Moreover, stone milling resulted in increased dough height at the end of the test for CWRS (from 24 to 33 mm), as opposed to decreased height for the Bona Vita cv (from 34 to 20 mm). Moreover, the time required to reach maximum dough development was lower in stone milling for Bolero cv (2.36 and 1.38 h, for SWF and RWF respectively), as opposed to an increase in this index for CWRS (1.36 and 4.49 h, for SWF and RWF respectively), whereas no differences were observed for Bona Vita and Skorpion cvs (Figure 4).

**Figure 4.** Effect of stone milling (solid line) and roller milling (dash line) on dough development, assessed by Rheofermentometer®, during leavening of whole grain flours from Bolero cv (**a**), CWRS (**b**), Bona Vita cv (**c**) and Skorpion cv (**d**). CWRS: commercial Canada Western Red Spring Wheat.

As regards the volume of CO2 developed (Figure 5), stone milling resulted in an increase in this index for Bolero (from 1150 to 1398 mL) and Skorpion (from 1532 to 1659 mL) cvs. In addition, regardless of the type of wheat, stone milling led to an increase in the amount of CO2 released (from 124 to 197 mL, from 195 to 211 mL, from 191 to 205 mL, and from 260 to 332 mL for Bolero, CWRS, Bona Vita and Skorpion cvs, respectively). Finally, no difference was observed in terms of the retention coefficient between the two milling approaches for the four samples.

**Figure 5.** Effect of stone milling (solid line) and roller milling (dash line) on the total gas production (grey line) and on the retained gas (black line) in the dough, assessed by Rheofermentometer®, of whole grain flours from Bolero *cv* (**a**), CWRS (**b**), Bona Vita *cv* (**c**) and Skorpion *cv* (**d**).

#### *3.6. Bread Properties*

As for bread properties, stone milling resulted in higher bread height for Bolero *cv* and CWRS, instead no significant differences were observed for Bona Vita and Skorpion *cvs* (Figure 6). As regards volume and specific volume, bread from CWRS, Bona Vita and Skorpion *cvs* showed significant higher values when the stone milling system was used. However, bread from Bona Vita and Skorpion *cvs* exhibited a large bubble in the crumb layer under the upper surface of the bread, due to a collapse of the crumb structure.

**Figure 6.** Effect of stone and roller milling on bread height (H; cm), volume (V; mL) specific volume (SV; mL/g). Data are presented as mean ± standard deviation. The asterisks indicate significant differences between the mean of the bread from stone and roller milled flours of each *cvs* (\* *p* < 0.01). The absence of asterisk indicates a not significant difference. CWRS: commercial Canada Western Red Spring Wheat.

The analyses of bioactive compounds and AC were carried out on bread samples obtained from SWF, RWF and refined flours of pigmented *cvs*, chosen to provide more interesting phytochemical comparisons (Table 4). Compared to WFs, the removal of wheat bran led to a significant decrease in several antioxidant compounds both in refined flours and derived breads. On average, bread obtained from refined flours showed a lower content of CWBPA (−94%), SPA (−89%), ACABTS (−27%), ACFRAP (−90%), zeaxanthin (−64%, Bona Vita *cv*) and TAC (−122 times, Skorpion *cv*). Conversely, lutein was higher (+29%) in bread from refined flour compared to WF samples. As highlighted in Table 4, bread samples obtained from the blue-grained *cv* showed levels of CWBPA (crumb and crust), SPA (crust), ACFRAP (crumb) and TAC (crumb and crust) significantly higher for RWF as compared to SWF. RWF bread (crumb and crust) obtained from Bona Vita *cv* showed a significantly higher ACFRAP, but a significantly lower content of lutein, than SWF. No difference in the relative abundance of CWBPAs was observed by comparing refined flour, SWF and RWF: ferulic acid was in both flour and bread the most representative (88%), followed by sinapic acid (5%) and *p*-cumaric acid (4%). Conversely, as far as the SPAs were concerned, a different behavior was observed in all compared flours: bread-making increased the relative abundance of soluble ferulic acid (on average from 27% to 46%), while reducing the percentage of soluble sinapic acid (from 41% to 32%).

**Table 4.** Cell wall-bound phenolic acids (CWBPAs), soluble phenolic acids (SPAs), antioxidant capacity (AC), xanthophylls (lutein and zeaxanthin) and total anthocyanin content (TAC) in raw material (flour), bread crust and bread crumb obtained from refined white flour, stone (SWF) and roller-milled (RWF) whole-grain flour of Bona Vita and Skorpion *cvs*.



**Table 4.** *Cont.*

Cell wall-bound phenolic acids (CWBPAs) and soluble (free and conjugated forms) phenolic acids (SPAs) are the sum of the single phenolic acids determined by means of RP-HPLC/DAD and are expressed as mg/kg dry basis. Xanthophylls (lutein and zeaxanthin) are expressed as mg/kg dry basis. Total anthocyanin content (TAC) is expressed as mg Cy-3-glc eq/kg dry basis. Antioxidant capacity (AC) measured by means of the ABTS and FRAP assays is expressed as mmol TE/kg dry basis. For each *cv*, value followed with different letters are significantly different (one-way ANOVA, *p* < 0.01), according to the REGW-F test.

#### **4. Discussion**

Despite the numerous websites and skilled marketing operations that declare the uniqueness and authenticity of SWF, the effects of stone milling—in comparison with those promoted by roller milling—on the chemical, rheological, and bread-making properties of the related flours have not yet been investigated systematically. The present study seeks to fill this gap. Moreover, since differences in kernel characteristics affect the milling process, four types of wheat were chosen for their variations in hardness, gluten strength, and richness in biocomponents. Specifically, the two wheat samples frequently used in the bread sector were Bolero (soft white winter wheat *cv*) and a commercial Canada Western Red Spring (CWRS). The other two wheat *cvs* were Bona Vita (medium-hard winter wheat with yellow endosperm for the high xanthophyll content) and Skorpion (medium-hard winter wheat with a blue external layer, rich in anthocyanins).

There is a widespread belief among consumers that, from a nutritional point of view, SWF are better than RWF, thus the label "made with stone-ground flour" is a powerful marketing tool for both producers and retailers [32]. In accordance with the AACC International definition of whole grain [33], neither milling process selected the anatomical regions, and endosperm, bran and germ have to be present in the same proportions as in the intact caryopsis. In agreement with other authors [17,20,22], no significant changes in the proximate composition were found regardless of the milling process used (Table 3). The few differences between SWFs and RWFs concerned moisture, which was significantly higher in RWFs from Bona Vita and Skorpion *cvs*. This result might firstly be related to the lack of conditioning before stone milling; moreover, a drop of moisture might be associated with heat development during this milling process, as mentioned by several authors [14,15,17].

The total dietary fiber content of WFs was included in the range 9.3%–14.3% (Table 3), similar to that observed for WFs examined within the European HEALTHGRAIN Project [10]. Regarding the potential effects of milling process on fiber fraction content, the SWF and RWF of each wheat sample exhibited no differences in total, insoluble and soluble fractions related to the milling process. Regarding the arabinoxylan fraction, the differences in the amounts of total and water-extractable arabinoxylans did not show a common trend regardless of the milling process used (Table 3). Nevertheless, in the two pigmented *cvs*, both of these parameters were considerably higher than in Bolero *cv* and CWRS, highlighting that Skorpion and Bona Vita *cvs* contain other interesting nutritional traits in addition to polyphenols or xanthophylls suitable for their exploitation. Although the occurrence of several macronutrients was unchanged in the compared WFs, RWFs of pigmented *cvs* resulted in a higher content of antioxidant compounds than SWF. Carcea et al. [20] reported no compositional difference regarding the total polyphenol and alkylresorcinol contents between stone milled or roller milled flours. In our study, in particular, a significantly higher content of CWBPAs and TAC were present in RWFs. Results could be related to the higher heat generated during stone milling due to friction [15]. Prabhasankar and Rao [17] observed that the higher temperature detected in SWF (85 ◦C), resulted in protein degradation, a reduction of amino acid content and a loss of some essential fatty acid compared to RWF (32 ◦C). Similarly, by comparing a stone milling process (60 ◦C) to a watermill process able to generate lower temperatures (30 ◦C), Di Silvestro et al. [34] observed a decrease in bound phenolic fraction, while no effect was detected for arabinoxylans and β-glucans.

In conventional roller milling, the importance of kernel tempering (or conditioning) to guarantee high yield and high quality of flour is widely recognized [11–13]. Indeed, particle size distribution after the first break and, consequently, the behavior of the "broken material" in all the remaining milling passages, is strictly influenced by kernel moisture [35]. On the contrary, no mention is made about the need to modify the native moisture of kernels before stone milling [11,16]. Such differences, and specifically, the increased moisture of the pericarp assured by tempering before roller milling could be the main reason for the differences in particle size distribution between SWFs and RWFs (Figure 1). Indeed, by lowering the native friability of the bran layers, moistening facilitates their separation from the starchy endosperm in large flakes during roller milling. At the same time, the increased endosperm moisture—in respect to the native kernel—induces the efficacious breakage of this region, yielding a high percentage of fine particles (Figure 1). This pattern is congruent with the results of Kihlberg et al. [22] and Ross and Kongraksawech [14], the latter investigating eight different small-scale mills. Although a common trend in all RWFs is recognizable, the moisture distribution inside the kernels could not be optimal irrespective of their hardness, as tempering conditions were the same for all wheat samples (moisture before milling equal to 16%). Bearing in mind the results obtained by Doblado-Maldonado et al. [36], CWRS could present, particularly in the external layers, lower moisture than required for optimal roller milling; this physical condition might account for the low percentage of large bran particles (Figure 1). Roller milling of medium-hard (i.e., Bona Vita and Skorpion *cvs*) and soft (Bolero *cv*) wheat, likely optimally moistened, yielded WFs with similar particle size distribution, characterized by a high percentage of both coarse and fine particles. On the contrary, when stone milling was applied, due to its different breakage system and lack of conditioning before milling, bran and endosperm regions exhibited a similar behavior during the breakage actions. Consequently, particles in SWFs were more homogeneously distributed in classes of different size, particularly in medium-sized classes (from about 300 to 700 μm) (Figure 1): the pattern was similar for all varieties, regardless of their hardness.

Evaluation of particle size distribution is important for understanding the rheological properties of dough: indeed, the particle size of bran and/or flour influences several features, including water absorption and gluten aggregation kinetics. Nevertheless, the literature has yet to indicate the 'optimal' particle size distribution for bread-making [37,38].

Although the milling system did not affect the protein content of the WFs (Table 3), some changes in protein properties—that are important for bread-making performance—were highlighted by the rheological tests. The analysis of gluten aggregation properties by means of a rapid shear-based method (i.e., GlutoPeak test) indicated that gluten proteins were able to aggregate and show a peak (Figure 2), which represents the maximum extent of gluten formation before its breaking due to the intense shear-stress [39]. Overall, RWFs exhibited faster gluten aggregation (lower peak maximum time), required less energy to aggregate and resulted in lower maximum consistency (except for CWRS) than SWFs, suggesting gluten weakening. A similar trend has been observed while comparing refined and whole flours due to the interference of fiber in network formation [40,41]. In the case of RWFs, the weakening of the gluten network could be due to depolymerization phenomena, favored by the presence of free-SH groups, particularly abundant in WFs with coarse particles (average particle size: 830 μm) [42]. Certainly, the presence of high amounts (more than 15% *w*/*w*) of large particles (>1000 μm size) in RWFs from Bolero, Bona Vita and Skorpion *cvs* might have negatively affected protein-protein interactions via physical mechanisms [42]. Similarly, the low percentage (only 7%) of the same size class in the RWF from CWRS might account for its opposite performance: both maximum torque and aggregation energy exhibited higher values than those determined in SWF.

Moving to the mixing properties evaluated by the farinographic test (Figure 3), the milling process did not seem to have a conclusive effect on such properties, that are greatly affected by the type of wheat. As emphasized by Ross and Kongraksawech [14], the farinographic indices were primarily influenced by *cv* and less by the milling process. In contrast to the gluten aggregation kinetics (evaluated by the GlutoPeak test on a slurry), the mixing properties were evaluated on a dough applying lower stress to the system (63 rpm vs. 2750 rpm). Thus, the apparent different findings could be attributed to the differences in the test conditions.

In general, the water absorption index was higher in SWFs (in Bolero and Bona Vita *cvs*), probably as the consequence of their higher (although not significant) amounts of damaged starch (Table 3), as the role of bran particle size, proposed as a valid explanation by Kihlberg et al. [22], was not highlighted. Stability was significantly affected only for Bolero *cv* (6.2 and 4.6 min for SWF and RWF, respectively). An important role might be played by the distribution of large/coarse bran particles (>1000 μm size) (Figure 1) which were three times higher in the RWF of Bolero *cv*. They were probably responsible for the weakness in its gluten network and, consequently, the significant decrease in dough stability. Moreover, the high percentage of large bran particles could impair not only dough properties during bread making but also the bioavailability of minerals, as indicated by Miller Jones et al. [10]. Nevertheless, the role of bran particle size on dough and bread characteristics needs to be further investigated as the results of works on this subject are still contradictory [38].

Regardless of the milling process, Skorpion and Bona Vita *cvs* showed similar leavening profiles (Figure 4), in agreement with their similar proximate composition (Table 3) and trends observed through the GlutoPeak test (Figure 2) and the farinographic test (Figure 3). Anyway, Skorpion WFs (both SWF and RWF) resulted in good dough development (Figure 4d) and gas production (Figure 5g,h), likely due to the high dough stability as shown by the farinographic test.

Differences in rheological properties associated with the milling systems were evident only for Bolero *cv* and CWRS. As considering the indications of the other rheological tests, Bolero *cv* and CWRS showed an opposite behavior according to the milling process; moreover, despite a similar protein content (Table 3), these samples were characterized by relevant differences in protein quality (Figure 2). Specifically, as for Bolero *cv*, SWF reached similar dough heights than RWF but faster, probably due to its higher—although not significant—damaged starch content as a quick source of simple sugars for yeast growth. In addition, RWF produced the least gas (Figure 5b) and the lowest bread volume (Figure 6). Both leavening properties and bread-making performance might be due to the high percentage of large particle size in RWF in Bolero *cv* (Figure 1). As expected, RWF of CWRS performed best during leavening in terms of dough development and time to reach it, indicative of the ability of this wheat type to withstand leavening stresses. Other reasons which might account for this result include good gluten aggregation (Figure 2) and mixing (Figure 3) properties, associated with the high damaged starch content and the low fiber percentage (Table 3) among the samples considered in this study.

Among the rheological tests used to predict the bread-making performance of samples, only the GlutoPeak tests (Figure 2) agreed for all samples as volume, specific volume and height of bread (Figure 6). Indeed, for all the wheat types, both the loaf height and the specific volume of bread samples produced with SWFs were higher than those obtained from RWFs, in agreement with the observations by Kihlberg et al. [22]. Our findings were also congruous with those of Gélinas et al. [18] which showed that dough mixing properties of WFs—in terms of farinographic absorption and stability—did not always relate to bread properties and, therefore, did not explain why some varieties performed better than others.

As the characteristics of bread are related not only to dough properties—generally evaluated by tests carried out at temperatures below 30 ◦C—but also to phenomena occurring during baking, we can hypothesize that, during baking, proteins in dough from SWFs might have retained extensibility for a longer time, assuring a higher bread volume. According to the literature [14,22,38,43,44], particle size distribution of WF might represent another trait able to influence bread volume. As previously discussed, stone milling produced a large amount of medium-coarse particles (from 300 to 700 μm) (Figure 1) that, according to Doblado-Maldonado et al. [15], could be considered the most advantageous for bread production. The particle size distribution observed in RWF (especially fine and large particles, simultaneously) (Figure 1) accounted for the low bread development. Indeed, small particles (<250 μm) could have a negative effect on bread characteristics as they promptly interfere in protein-protein interactions due to their high contact surface [43]. Also large particles might exert an undesirable action towards gluten development and gas cell stabilization [42] and bread appearance and texture [45].

Despite their high volume, in the case of Bona Vita and Skorpion *cvs*, stone milling resulted in a huge bubble just under the crust, together with the collapse of the underlying crumb (Figure 6). These behaviors cannot be attributed to differences in α-amylase activity (data not shown). The pattern of the corresponding rheofermentograph traces (Figure 4) allows us to hypothesize that, at the beginning of baking, the gas produced by yeasts in large amounts was not efficaciously held inside the gluten network and gathered in the upper part of loaf, causing the formation of a big bubble and a partial collapse of the underlying region.

The slight but significantly higher content of bioactive compounds in RWF flour compared to SWF (Table 3) was confirmed after bread-making. In particular, a higher CWBPA and TAC content found in both RWF bread crumb and crust for the blue-grained *cv*, resulting in a higher AC (FRAP assay, Table 4). As observed in previous studies [46,47], during bread-making significant changes occur in both bioactive compounds and AC. In the present study, bread-making caused a significant loss of the antioxidants responsible for the grain and flour pigmentation (xanthophylls and anthocyanins). Nevertheless, an increase in the AC was observed in the bread crust (Table 4). This could be due to the neo-formation of Maillard reaction products [46,48].

#### **5. Conclusions**

Most consumers believe that only the stone-milling process is able to preserve all the nutrients and bioactive compounds of wheat grains as, in this process, all kernel regions form a single stream. Indeed, the roller-milling process (a multiple-stream approach where the fractions must be recombined to obtain WF) is wrongly but commonly associated with a partial depletion of the native nutrients of the kernel. Our results proved that SWFs have neither a better proximate composition, nor a better bioactive compound concentration than RWFs. Only for blue-grained *cv* (Skorpion) RWFs resulted in a slight but significant higher content of CWBPAs and TAC compared to SWFs and this feature was observed also in bread.

The comparison of SWF and RWF properties highlighted a different particle size distribution. Indeed, during the grinding of caryopsis through the stone or the roller-milling, compression, shear, and cutting stresses exhibited different intensities and degrees (due to intrinsic kernel factors and process conditions), promoting the formation of large bran particles and very small flour particles in RWFs, while a more homogeneous particle size distribution was observed in SWFs. Although it can be assumed that these physical features could greatly affect the surface properties and the hydration properties of flour, only the GlutoPeak test, a quite recent rheological approach proposed for evaluating the protein-protein aggregation kinetics in wheat, highlighted significant differences in the gluten properties of WFs according to the milling process which were congruent with their bread-making performances.

The rheological differences between the WFs obtained from stone- or roller-milling, although significant, do not make one process clearly preferable to the other one. However, further information on the sensory profile of bread is worthy of interest. Nevertheless, the lower productivity of the former is acceptable for artisan or home-made processes, while the higher flexibility and versatility of multiple-stream roller-milling, and its fully automated management, can better satisfy industrial purposes. On the other hand, the effect of heat treatments (for stabilizing bran and germ) on the nutritional features of RWFs should be considered, as well as the effect of re-milling large bran particles on the technological performances.

**Author Contributions:** Conceptualization, M.A.P., A.M. and M.B.; methodology, G.C., M.C.C., D.E., D.G. and A.P.; validation, G.C., M.C.C. and D.G.; formal analysis, G.C., M.C.C., D.E. and D.G.; investigation, G.C., M.C.C. and D.G.; resources, M.A.P., A.M., M.C.C., A.P. and M.B.; data curation, G.C., M.C.C., D.E., D.G. and M.B.; writing—original draft preparation, M.A.P., G.C., M.C.C. and D.G.; writing—review and editing, A.M., M.B. and A.P., M.C.C., and D.E.; visualization, G.C. and D.G.; supervision, A.M. and M.B.; project administration, A.M., M.C.C. and M.B., funding acquisition, M.A.P., M.B, M.C.C., and D.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors acknowledge the financial support of the Italian Ministry of Education, University and Research (MIUR), program PRIN 2015 (Grant Number 2015SSEKFL) "Processing for healthy cereal foods".

**Acknowledgments:** The authors thank Simone Tomatis (Molino Tomatis S.N.C, Niella Tanaro, Italy), Francesca Vanara (Università degli Studi di Torino), Daniele Noé and Franca Criscuoli (Università degli Studi di Milano) for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Bio-Functional and Structural Properties of Pasta Enriched with a Debranning Fraction from Purple Wheat**

### **Parisa Abbasi Parizad 1, Mauro Marengo 1, Francesco Bonomi 1, Alessio Scarafoni 1, Cristina Cecchini 2, Maria Ambrogina Pagani 1, Alessandra Marti 1,\* and Stefania Iametti 1,\***


Received: 20 January 2020; Accepted: 6 February 2020; Published: 8 February 2020

**Abstract:** A colored and fiber-rich fraction from the debranning of purple wheat was incorporated at 25% into semolina- and flour-based pasta produced on a pilot-plant scale, with the aim of increasing anthocyanin and total phenolic content with respect to pasta obtained from whole pigmented grains. The debranning fraction impaired the formation of disulfide-stabilized protein networks in semolina-based systems. Recovery of phenolics was impaired by the pasta making process, and cooking decreased the phenolic content in both enriched samples. Cooking-related losses in anthocyanins and total phenolics were similar, but anthocyanins in the cooked semolina-based pasta were around 20% of what was expected from the formulation. HPLC (High Performance Liquid Chromatography) profiling of phenolics was carried out on extracts from either type of enriched pasta both before and after cooking and indicate possible preferential retention of specific compounds in each type of enriched pasta. Extracts from cooked samples of either enriched pasta were tested as inhibitors of enzymes involved in glucose metabolism and uptake, as well as for their capacity of suppressing the response to inflammatory stimuli. Results of both biological tests indicate that the phenolics in extracts from both cooked pasta samples had inhibitory capacities higher than extracts of the original debranning fraction at identical concentrations of total bioactives.

**Keywords:** pigmented wheat; anthocyanins; polyphenols; alpha-amylase inhibition; anti-inflammatory activity

#### **1. Introduction**

Many studies have shown that anthocyanins and other polyphenolic compounds have anti-inflammatory and antioxidant properties that may play a positive effect in preventing chronic diseases ranging from cardiovascular diseases to metabolic syndrome. The relationship between specific anthocyanins and their biological activities, such as anti-inflammatory, anti-obesity, anti-diabetes, has been the subject of a number of recent reviews [1,2]. Anthocyanins and polyphenols have also been reported to control intracellular signaling cascades as the process of inflammation progresses within the cells [3–5]. Numerous studies have demonstrated that anthocyanins can exert the beneficial effects in diabetes by acting on various molecular targets and regulate different signaling pathways in multiple organs and tissues such as liver, pancreas, kidney, adipose, skeletal muscle, and brain [6].

A critical aspect of most of the reported bioactivities for this class of chemical species relates to the poor absorption of several phenolics that impairs their presence at high enough concentration in biological fluids [7,8]. However, many of the health benefits associated with anthocyanins bioactivity relate to the effects these molecules reportedly exert on proteins in the intestine. Among the significant targets that do not require a transit of phenolics across the gastrointestinal epithelia are enzymes involved in glucose metabolism, such as pancreatic amylases involved in the starch enzymatic breakdown and the brush-border alpha-glucosidase relevant to glucose uptake [9–14].

One possible strategy to ensure adequate uptake of anthocyanins and phenolics is their incorporation into staple foods [8,15]. In this frame, pigmented grains have received particular attention as they may represent the starting ingredients to produce staple foods such as pasta or bread. High concentrations of phenolic compounds are present in the outer layers of a number of varieties of common grains such as wheat, corn, and rice [16]. In the case of colored grains, anthocyanins are responsible for their purple, blue, or red color [17]. These compounds are present in various amounts in bran layers as either the glycosylated or aglycone form [18].

The growing interest for new products and the visual appeal of naturally colorful products has led to the introduction of "colored" bread and pasta to the market. Most of the available products are prepared by using whole grains as the source of anthocyanins [16,19]. Recently, an innovative milling process has been developed in order to separate bran components to be used in specific processes [20,21]. In the case of colored grains, the usually discarded outermost part of the kernels is a suitable ingredient for the production of enriched food. According to Zanoletti et al., [20], it was possible to produce pasta with a high amount (up to 15% (*w*/*w*)) of bran fraction, with fiber and polyphenol content sensibly higher than in products prepared from the corresponding whole grains. The enriched pasta samples were characterized mainly in terms of chemical composition and cooking behavior, and only limited data are available for the retention of the bioactive properties of the incorporated materials.

This study attempts to fill evident gaps in the studies mentioned above by studying the effects of processing (and cooking) on retention of the bioactives' functionality in pasta samples prepared by incorporating at least 25% of anthocyanin-rich bran fractions from purple wheat debranning into pasta made from semolina or common wheat flour. This amount was chosen to provide nutritionally relevant amounts of both dietary fibers and phenolics in a product with attractive color features, avoiding in the meantime excessive changes in the pasta-making process and allowing the retention of the structural integrity of the cooked product. The focus in this study was on the in vitro inhibitory capacity of phenolics toward specific enzymes involved in the carbohydrate metabolism, as well as on their anti-inflammatory properties.

#### **2. Materials and Methods**

#### *2.1. Chemicals and Enzymes*

Unless otherwise specified, all chemicals and enzymes (namely, rat intestinal acetone powders of alpha-glucosidase (EC 3.2.1.20) and porcine pancreatic alpha-amylase (EC 3.2.1.1)) were from Sigma-Aldrich (Milan, Italy).

#### *2.2. Raw Materials*

An anthocyanin-rich bran fraction (henceforth, DF) was obtained by the debranning of commercial purple wheat, essentially as reported in [20]. The DF used in the studies reported below corresponds to an abrasion level of 3.7%, with respect to the whole grain and had a particle size in the 500–700 μm range. Refined common wheat flour (12.5% protein, dw; ashes <0.5%, dw) and durum wheat semolina (14.4% protein, dw; ashes 0.85%, dw) were provided by Molino Quaglia (Vighizzolo d'Este, Padua, Italy) and by F.lli De Cecco (Fara San Martino, Chieti, Italy), respectively.

#### *2.3. Pasta*

Pasta was produced from either commercial flour or semolina that was enriched by dry mixing with 25% (*w*/*w*) of the bran fraction obtained by the debranning of purple wheat (DF, see above). Macaroni-shaped pasta samples were produced using the DeFENS pilot plant, essentially as described in [22]. In particular, moisture in the dough was adjusted to a final level of 31.8%, and mixing and extrusion were completed in 20 min. Drying was carried out at 60 ◦C for 12 h. Each sample of pasta was cooked in tap water (1 L water per 100 g pasta) at the optimum cooking time, according to the AACC method 16–50 [23]. The sensory acceptability of the cooked pasta was assessed by ten untrained panelists. Prior to further characterization, samples of dried uncooked pasta were ground to <250 μm in a laboratory mill, whereas samples of cooked pasta were frozen in a deep freezer at −80 ◦C and lyophilized (Alpha 2-4 LD freeze dryer, Martin Christ, Osterhode am Harz, Germany) prior to further characterization.

#### *2.4. Protein Solubility and Thiol Accessibility*

The solubility of proteins in pasta samples was determined in triplicate using buffers of increasing dissociating ability, as described elsewhere [24,25]. Results are expressed as (mg soluble protein)/(g total proteins) to account for the protein content of individual samples, assessed through a dye-binding method [26]. Accessible thiol groups were determined (in triplicate) by using the spectrophotometric thiol reagent 5,5 -dithiobis-(2-nitrobenzoate) (DTNB) as also described in [24,25]. Results are expressed as μmol thiols/g total protein, to account for the protein content of individual samples.

#### *2.5. Phenolics Extraction*

Each sample (2 g) was defatted with petroleum ether and extracted twice with 15 mL of an ethanol/HCl mixture (15 mL of a mixture made up of 65 volumes of 95% ethanol and 35 volumes of aqueous 0.3 M HCl). The pooled ethanol/HCl extracts were used for analytical measurements and HPLC profiling. Extracts to be used with Caco-2 cells were by using water in place of dilute HCl, in order to avoid interference with the cellular assays and cell viability issues [13]. Assays were carried out in triplicate for each sample.

#### *2.6. Determination of the Total Polyphenols and the Total Anthocyanins Content*

Total polyphenols (as gallic acid equivalents) and total anthocyanins (as cyanidin-3-*O*-glucoside) were measured as in [27] on individual extracts from both cooked and uncooked pasta samples.

#### *2.7. Anthocyanin and Phenolics Profiling by RP-HPLC*

RP-HPLC profiling was performed on a C18 column (5 μm, 4.6 mm × 250 mm, Waters, Milan, IT) fitted to a Waters 600 E HPLC, equipped with a 996 PDA (Photo Diode Array) detector, by adapting procedures reported elsewhere [9]. Extracts (0.1–0.2 mL) were loaded on the column, and components eluted at a solvent flow of 0.8 mL min−1, using a gradient from 100% A (0.1% trifluoroacetic acid in water) to 100% B (0.1% trifluoracetic acid in acetonitrile): 0% to 5% B in 5 min; 5% to 40% B from 5 to 40 min; 40% to 70% B from 40 to 48 min. The eluate was monitored at 520 nm (for anthocyanins), 350 nm (for rutin and quercetin), 320 nm (for ferulic acid), and 280 nm (for catechin and epicatechin). Calibration was carried out by using suitable standards, and results are expressed as content of individual species in the original sample, on a dry matter basis.

#### *2.8. Enzyme Inhibition Studies*

Alpha-Glucosidase: Rat intestinal acetone powder was used in these assays, following established procedures [10–14] and testing inhibition by ethanol/HCl extracts from individual samples, diluted as appropriate in ethanol/HCl. Blanks were prepared in the absence of enzyme and of the ethanol/HCl mixture, whereas controls were complete reaction mixtures containing the appropriate volumes of the ethanol/HCl mixture used for extraction, but no bioactives. Acarbose (from a stock solution in ethanol/HCl) was used as the reference inhibitor. Tests were carried out in triplicate.

α-Amylase: Enzymatic activity was measured according to established procedures [10–14], and inhibition was tested by using ethanol/HCl extracts from individual samples, diluted as appropriate in ethanol/HCl. Blanks were prepared in the absence of enzymes, whereas controls were complete reaction mixtures, containing the appropriate volumes of the ethanol/HCl mixture used for extraction, but no bioactives. Acarbose (from a stock solution in ethanol/HCl) was used as the reference inhibitor. Tests were carried out in triplicate.

#### *2.9. Immunomodulatory Properties of Extracts*

Human intestinal epithelial Caco-2 cells were provided by Maria Rosa Lovati (Dipartimento di Scienze Farmacologiche e BioMolecolari, University of Milan, Milan, Italy). All experiments involving cultured Caco-2 cells have been carried out in the DeFENS Cell Culture Laboratory, a core facility of the Department of Food, Environmental and Nutritional Sciences (University of Milan, Milan, Italy). The experimental setup allows us to monitor the effects of extracts for inhibiting IL1β-stimulated synthesis of NF-κB in the transfected Caco-2 cells. In short, the test measures inhibition of luciferase expression in Caco-2 cells transiently transfected with the pNiFty2-Luc plasmid (InvivoGen, Rho, Italy), that combines five NF-κB binding sites with the luciferase reporter gene *luc*, allowing expression of the luciferase gene in the presence of NF-κB upon stimulation with interleukin 1β (IL1β) at a final concentration of 20 ng mL−<sup>1</sup> in the presence/absence of extracts from individual samples. It should be noted that the extracts used in this study were prepared with aqueous ethanol in the absence of HCl, as noted above (Section 2.5). The experimental setup for these experiments has been reported in detail elsewhere [13]. The activity of the expressed luciferase was measured in cellular extracts by adding ATP and D-luciferin, followed by monitoring bioluminescence in a VICTOR3 1420 Multilabel Counter (PerkinElmer, Waltham, MA, USA) Each assay was carried out at least in triplicate.

#### *2.10. Statistical Analysis*

Analysis of variance (one-way ANOVA) was carried out by using Statgraphic Plus v. 5.1 (StatPoint Inc., Warrenton, VA, USA). The addition of DF to each pasta samples was considered as a factor for ANOVA. Results are reported as averages ± SD. The number of replicates for individual measurements is given under the Materials and Methods section for individual measurements, and in the legend to individual tables.

#### **3. Results and Discussion**

#### *3.1. Molecular Organization of Proteins in Enriched Pasta*

The possible effects of incorporating high levels of a bran-derived fraction into pasta on the overall structure of the protein network in the product were investigated by taking into account the conditional solubility of proteins and the accessibility of cysteine thiols to water-soluble reagents in the uncooked products [25,28]. The conditional solubility approach offers a simple way for estimating the nature of interprotein bonds, due to their different sensitivity to the action of chaotropes in the presence/absence of disulfide-breaking agents. Measuring the accessibility of residual thiols in the presence/absence of chaotropes provides information on the overall compactness of the protein network and complements protein solubility data.

Results are presented in Table 1, which also include values of these parameters for reference semolina-only pasta made in the same pilot plant under very similar processing conditions. The protein solubility data in Table 1 indicate the prevalence of urea-sensitive hydrophobic interactions in protein networks formed from wheat flour, as well as their lower compactness with respect to those formed by proteins in semolina. However, a comparison with the reference semolina pasta highlights the well-known "destructive" effect of the addition of a bran-derived fraction on the formation of a disulfide-stabilized protein network in semolina-based systems [29]. This is evident when considering that the DTT-dependent increase in soluble proteins is about 25% in the enriched pasta, whereas, in the reference semolina-based pasta, the breakdown of disulfide bonds by DTT results in a 4-fold increase of the amount of solubilized proteins.


**Table 1.** Properties of the protein network in uncooked pasta.

Addition of anthocyanin-rich bran fraction (DF) was considered as a factor for ANOVA. Different letters in a row indicate a significant difference at *p* < 0.05 (Tukey test; *n* = 4). DTT

A comparison of data on thiol accessibility is also informative, as the number of accessible thiols in either enriched pasta sample is by far lower than what expected from the properties of the starting material or—in the case of semolina-based enriched pasta—from the figures obtained for non-enriched samples. Freely accessible thiols (i.e., those accessible in the absence of chaotropes) in wheat flour or semolina proteins are in the 3–4 micromolar range when expressed on a protein basis [24,25]. The addition of chaotropes results in a 3-fold increase of accessible thiols for wheat flour-based pasta, and almost doubles the amount of accessible thiols in semolina.

A detailed molecular-level investigation of these observations is beyond the scope of this report. However, it seems reasonable to hypothesize that the observed decrease in accessible thiols—regardless of the presence/absence of chaotropes—may relate to some physical features of the enriched pasta that do not necessarily involve the properties of the protein network itself and may involve protein/polysaccharide interactions, as observed, for instance, in a rice-based pasta [30].

#### *3.2. Total Anthocyanins and Total Polyphenol Incorporation in Enriched Pasta*

The total anthocyanins content (TAC) and the total polyphenol content (TPC) in both cooked and uncooked pasta samples—prepared from either common wheat flour or semolina and 25% (*w*/*w*) of the polyphenol-rich fraction obtained from the debranning of purple wheat grains (DF)—are reported in Table 2. The values are compared with those expected from the content of each subclass of phenolics in the original bran fraction used to enrich both pasta samples (TAC, 690 mg Cya-3-O-glycoside eq/kg; TPC, 47,400 mg GAE/kg [13,20]). Separate analyses indicated that the contribution of phenolics from either the wheat flour or the semolina used in this study was negligible.

The data in Table 2 indicate an apparent decrease in both TAC and TPC in uncooked pasta samples with respect to what expected from the mixing formula. Losses in either family of phenolics may have occurred in the pasta-making process (most likely as a consequence of the drying steps in pasta production [7,31,32]), but the observed decrease may also be attributed to difficulties in recovering either species from the highly structured protein matrix formed upon kneading and drying.

The fact that apparent recovery figures in uncooked pasta are lower for semolina-based samples than for the flour-based ones offers circumstantial support for the latter hypothesis. In other words, the stiff protein network likely present in pasta could make the ethanol/HCl extraction procedure less effective, an effect most evident in semolina-based pasta because of the high protein content of semolina and the relevance of covalent disulfide bonds in the stabilization of interprotein networks formed by *T. durum* (*Triticum durum*) proteins in semolina-based dough in contrast with the prevalence of non-covalent hydrophobic interactions in the stabilization of interprotein networks formed by *T. aestivum* proteins in wheat flour dough [33,34].


**Table 2.** Efficiency and stability of phenolics incorporation.

Addition of DF was considered as a factor for ANOVA. Different letters in a row indicate a significant difference at *p* < 0.05 (Tukey test; *n* = 3). \* As calculated from formulation and previous or current analytical data [20].

As also evident from Table 2, cooking induced a further decrease of the phenolics content in both pasta samples. At this stage, we are unable to discriminate between this decrease being due to the release of the bioactives in the cooking water or to their sensitivity to temperature. In general, it seems safe to assume that there should be no issues for the yield of the ethanol/HCl extraction procedure when applied to cooked and lyophilized pasta samples, given the highly porous structure of the lyophilized samples.

It may be noted that the losses observed upon cooking flour-based enriched pasta are similar when considering anthocyanins alone or total phenolics. Conversely, cooking semolina-based enriched pasta gave loss in total phenolics slightly lower than those observed in common wheat-flour enriched pasta, as expected from the higher tenacity of the protein network in semolina-based pasta according to plentiful literature reports and to the protein network stability data reported in the previous section of this report. In spite of this, losses in anthocyanins upon cooking semolina-based enriched pasta were the highest, resulting in the final content of these species not exceeding 22% of what was calculated for the untreated formulation, based on proportion among individual ingredients and on their content in the various bioactives.

HPLC profiling was used to address whether any particular chemical species was involved in the compositional changes reported in Table 2 before and after cooking. To avoid possible sources of confusion [9], the HPLC profiling in this study takes into account only the aglycones of the most abundant species present in ethanol/HCl extracts of DF, namely, cyanidin and delphinidin (as representative of anthocyanins), and ferulic acid, quercetin and rutin (as representative of not-colored phenolics). In this general frame, it is worth mentioning that glycosylated and non-glycosylated anthocyanins are present in almost equivalent amounts in purple wheat [13], whereas most of the non-pigmented phenolics are present in their glycosylated forms. Also, the amount of ferulic acid derivatives in the debranning fractions from pigmented wheat varieties is reportedly much lower than in bran fractions from non-pigmented wheat varieties, where ferulic acid and its glycosylated forms may account for more than 80% of the total phenolics [35].

The results of HPLC profiling carried out on ethanol/HCl extracts from the various samples are reported in Table 3, which also provides—as a reference—the expected content in individual aglycones, calculated from quantitative HPLC profiling of ethanol/HCl extracts from DF. As already observed for the data presented in Table 2, the content of individual anthocyanin aglycones is lower than expected even in the uncooked samples, suggesting that the yield of the extraction procedure may be impaired by processing. This hypothesis is circumstantially supported by the apparent (although not statistically significant) increase in the content of each of the two anthocyanin aglycones in the cooked and lyophilized pasta samples.

A comparison among the data in Table 2; Table 3 also makes it evident that recovery figures for representative anthocyanin aglycones in uncooked pasta (30–50% of the expected, Table 3) are much lower than recovery figures for total anthocyanins (90–65%, Table 2), suggesting that the glycosylated forms of anthocyanins are less sensitive to matrix-related recovery issues. Extending the above comparison to the cooked samples, it appears evident that the abundant glycosylated anthocyanins are much more prone to being released upon cooking than their aglycones, and that anthocyanin aglycones are retained by the matrix in cooked pasta independently of whether the enriched pasta was based on semolina or wheat-flour (see Table 3), at difference with what observed with total anthocyanins (see Table 2).


**Table 3.** Content of representative aglycones in enriched pasta.

Different letters in a row indicate a significant difference at *p* < 0.05 (Tukey test; *n* = 4). \* As calculated from formulation and previous analytical data on DF [20].

Figures for the aglycone forms of representative non-pigmented phenolics indicate relevant differences in the efficiency of their incorporation in non-cooked pasta. Whereas ferulic acid was incorporated almost completely regardless of the use of flour or semolina in the formulation, rutin levels in pasta were roughly 30% of what expected, and quercetin fared even worse, showing a 10% incorporation, again regardless of the use of flour or semolina. This is in keeping with the data in Table 2, indicating a 30% retention of total phenolics.

Conversely, bioactive losses upon cooking were highest for the aglycone form of ferulic acid (30–50%) and statistically negligible for the rutin aglycone. In contrast, about 50% of the quercetin aglycone was lost upon cooking flour-based pasta, but no significant losses were observed in the case of semolina-based pasta. In this frame, it should be noted that losses in total phenolics upon cooking were ranging from 50 to 60 percent (see Table 2), again with the lowest losses being recorded for semolina-based pasta.

#### *3.3. Inhibition of Enzymes Relevant to Glucose Metabolism and Uptake*

Ethanol/HCl extracts from cooked pasta were tested for their capability of inhibiting pancreatic α-amylase and brush border alpha-glucosidase. Cooked pasta was used for these studies, in order to evaluate the possible effects of the bioactives in both types of enriched pasta after undergoing all the required steps prior to their consumption.

The data in the upper panel of Figure 1 indicate that extracts from either type of pasta had an inhibitory activity towards pancreatic alpha-amylase remarkably higher than the reference drug acarbose or extracts from the debranning fraction (DF) used in the formulation of either type of pasta (on a weight basis, as anthocyanin equivalents). Extracts recovered from semolina-based pasta were found to be less effective inhibitors than those from flour-based pasta, in particular at low anthocyanin concentrations.

A comparison with the profile of individual classes of phenolics in the extracts (see Table 3) does not offer valuable hints for a straightforward interpretation of these results. Further investigation is required to assess whether the difference between extracts from the two pasta samples could depend on the absence of synergistic effects among individual classes of phenolics (and individual chemical species, as reported in previous studies on alpha-amylase inhibition with various classes of phenolics of different origin [11,13,14], possibly as a consequence of losses or alterations of specific individual molecules during the processing and cooking steps.

**Figure 1.** Inhibition of pancreatic alpha-amylase (upper panel) and brush-border alpha-glucosidase (lower panel) by HCl-ethanol extracts from cooked pasta, the original debranning fraction (DF), and by the reference drug, acarbose. Concentration of bioactives in extracts from either type of cooked pasta and from the debranning fraction (DF) is given as cyanidin equivalents. Results are reported as averages ± SD (*n* = 3).

The data on glucosidase inhibition by extracts from either type of cooked pasta are reported in the lower panel of Figure 1 and indicate that brush border alpha glucosidase is less sensitive to phenolics in the HCl-ethanol extracts (and to the reference drug, acarbose) than pancreatic alpha-amylase, confirming a number of previous reports [13,14]. Also, in the case of alpha-glucosidase, extracts from flour-based pasta appear more active than those from semolina-based pasta at identical concentrations of anthocyanins. As already pointed out for alpha-amylase inhibition, extracts from either type of pasta were much more efficacious in alpha-glucosidase inhibition than extracts from the debranning fraction (DF) used in the formulation of either type of pasta (on a weight basis, as anthocyanin equivalents).

As in the case of alpha-amylase, the differences in alpha-glycosidase inhibitory activity between extracts from the two pasta samples are most evident at low concentrations (<15 mg/L) of bioactives. Again, comparison with data reported in previous work [13] indicates that the inhibitory effects of anthocyanins extracted from cooked pasta samples towards alpha-glycosidase were higher (about 50% inhibition at 25 mg/L total anthocyanin, almost regardless of the type of pasta, see Figure 1) than those for extracts from the same debranning fraction used in this study, which gave approximately 30% inhibition at the same concentrations.

The results discussed above may be interpreted by taking into account the different phenolics profile in the extracts used in this study (see Table 3). These figures suggest that some of the phenolics originally present in DF are better retained in the cooked products than others or, conversely, that the retained ones are more powerful inhibitors than those lost in any of the steps leading to the final cooked pasta. As for the differences observed at low concentrations of anthocyanins in the inhibition assays, a working hypothesis could ascribe them to a different interplay among the individual species that are retained in each system and are responsible for specific inhibition of either enzyme. Indeed, a number of studies have pointed out synergistic inhibitory effects among various molecules in this general class of compounds [10,13,14].

#### *3.4. In Vitro Study of Anti-Inflammatory Activity of Cooked Pasta Extracts on Caco-2 Cells*

The anti-inflammatory activity of extracts from either type of enriched pasta was tested on the same model used in a number of previous studies [3–5]. In short, the model allows us to estimate the effects of extracts for inhibiting IL1β-stimulated synthesis of NF-κB in suitably transfected Caco-2 cells. Anthocyanidins are among the phenolics reportedly able to suppress the expression of inflammatory mediators such as cyclooxygenase (COX-2) by attenuating various forms of cellular signaling, including pathways involving NF-κB and MAPK [1,4–7,36].

From a methodological standpoint, it has to be underscored that the extracts from cooked pasta samples used in the experiments involving cells were prepared in aqueous ethanol to avoid cell viability issues related to residual traces of HCl [13]. This experimental detail prevents any immediate comparison with the ethanol/HCl extracts discussed above but allows straightforward comparison with the acid-free extracts from the same debranning fraction used in the formulation of the pasta samples used here and already tested for their anti-inflammatory activity in previous studies [13].

The data in Figure 2 highlight the ability of extracts from both types of cooked pasta to suppress NF-κB expression in the cellular model at concentrations much lower than those required for inhibiting enzymes relevant to glucose metabolism. As observed for enzyme inhibitory activities, the higher efficacy of extracts from flour-based pasta in repressing response to IL-1β with respect to the semolina-based one seems to level off at a high concentration of phenolics (as anthocyanin equivalents).

**Figure 2.** Immunosuppressive effects of the aqueous ethanol extracts from both types of cooked pasta and the original debranning fraction (DF) Data are presented as percent inhibition of IL1β-stimulated expression of NF-κB. Acid-free extracts were used in all these assays. Results are reported as averages ± SD (*n* = 3).

Figure 2 makes also evident that acid-free extracts from DF were sensibly less efficacious in suppressing NF-κB expression in the cellular model used here than that of similar extracts from either type of cooked pasta all over the concentration of anthocyanins tested here. As discussed above, these differences likely stem from the different phenolics profile in the various extracts (see Table 3). Again, as in the case of enzyme inhibition, some of the phenolics in DF could be better retained in the cooked products than others, and the retained ones may include species that are particularly efficient in suppressing inflammatory response either as individual compounds or through synergistic effects [10,13,14]. Work currently in progress will hopefully contribute to elucidating in sufficient detail the molecular determinants of the observed differences and offer some useful clues as to the suspected synergies among individual components in the extracts.

#### **4. Conclusions**

Pilot-plant scale incorporation of a phenolics- and fiber-rich fraction (from the debranning of purple wheat) into both semolina- and flour-based pasta was tested at 25% content of the debranning fraction, an amount suitable for avoiding excessive changes in the pasta-making process itself, as well as (1) preserving structural integrity of the product, (2) ensuring some attractive color characters in pasta, and (3) providing added value by incorporating nutritionally relevant amounts of both dietary fibers and phenolics.

Protein solubility and thiol accessibility approaches indicated that protein networks in flour-based enriched pasta were mostly stabilized by hydrophobic interactions and had lower compactness than those in the semolina-based enriched sample. However, a comparison with non-enriched semolina-based pasta indicated that the addition of the bran-derived fraction in the enriched pasta resulted in a sensible destabilization of disulfide-stabilized protein networks in semolina-based systems.

Either type of enriched pasta had a content in anthocyanins and total phenolics much higher than what reported in literature for products obtained from whole pigmented grains, but lower than what was expected from what was calculated from their formulation, at least on the basis of previous analytical data on the debranning fraction of purple wheat used in this study [13,20,21]. Losses in either anthocyanins or phenolics may have occurred as a consequence of thermal treatments in the pasta-making process (most likely as a consequence of the drying steps in pasta production) [7,8,10,15,31]. However, the decrease observed for both anthocyanins and phenolics before cooking could also stem by analytical recovery issues, as suggested by the observation that retention figures in uncooked pasta were lower in semolina-based samples than in flour-based ones.

Cooking induced a further decrease of the phenolics content in both pasta samples. In the case of flour-based enriched pasta, similar cooking losses were observed for anthocyanins alone and total phenolics. Conversely, cooking semolina-based enriched pasta gave a loss in total phenolics slightly lower than those observed in flour-based enriched pasta, but losses in anthocyanins upon cooking semolina-based enriched pasta were the highest, resulting in a final content of these species never exceeding 20% of what was expected on a formulation basis.

HPLC profiling of phenolics was carried out on extracts from either type of enriched pasta both before and after cooking. Although the characterization reported here may hardly be seen as complete, results are suggesting the possible "preferential retention" of specific compounds in each type of enriched pasta, both before and after cooking. The relevance of specific/preferential retention was evident from measurements of biological activities, which were carried out using suitable extracts from cooked samples of either type of enriched pasta.

Extracts from cooked pasta samples were tested for inhibitory capacity towards enzymes involved in glucose metabolism and uptake, and for the ability to suppress the cellular response to inflammatory stimuli. Results of both types of test (i.e., "in vitro" inhibition of enzymes and suppression of response to inflammatory stimuli in a widely used cellular model [1,3,13,36]) indicated that the phenolics retained in both samples of cooked, enriched pasta may be regarded as the "good" ones, at least in terms of biological activity. Indeed, extracts from either type of enriched pasta had inhibitory capacities

higher than extracts of the original debranning fraction used for formulating these products at identical concentrations of bioactives.

A coarse estimate from data in Table 2 indicates that a 60 g serving (dry weight) of pasta enriched with suitable fractions from purple wheat debranning could provide between 1.8 and 2.4 mg of total anthocyanins in the cooked product. This should result in an estimated duodenal total concentration of anthocyanins in the range of 4 to 6 mg/L. In this concentration span, the activity of enzymes involved in glucose metabolism is decreased (decrease is in the 20% to 25% range, see Figure 1), as is the response of epithelial cells to inflammatory stimuli (decrease is in the 75% to 60% range, see Figure 2).

The findings reported here may be seen as exciting from the viewpoint of obtaining products with possible physiological impact and of general appeal to the consumer (with the bonus of added nutritional value as a consequence of their content in dietary fibers) by using established production processes. On the other hand, from the standpoint of the food chemist or biochemist, these results bring forward a number of daunting challenges. Indeed, providing a molecular-based rationale for the observations reported above will require substantial efforts, including a more thorough characterization of components in the various samples and, possibly, the use of specific mixtures of individual components to verify, in vitro at least, the observed effects and the occurrence of possible synergies among specific compounds.

**Author Contributions:** Conceptualization, M.A.P., A.M., S.I.; Methodology, P.A.P., M.M., A.S.; C.C.; Formal analysis A.M., S.I., F.B.; Writing—original draft preparation, P.A.P., M.M., S.I.; Writing—review and editing, F.B., M.A.P., A.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors acknowledge the financial support of the Italian Ministry of Education, University and Research (MIUR), program PRIN 2015 (Grant Number 2015SSEKFL) "Processing for healthy cereal foods". M.M. is the grateful recipient of a post-doctoral fellowship from the University of Milan.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Fresh Pasta Manufactured with Fermented Whole Wheat Semolina: Physicochemical, Sensorial, and Nutritional Properties**

#### **Simonetta Fois, Marco Campus, Piero Pasqualino Piu, Silvia Siliani, Manuela Sanna, Tonina Roggio and Pasquale Catzeddu \***

Porto Conte Ricerche Srl, Località Tramariglio, 07041 Alghero (SS), Italy; fois@portocontericerche.it (S.F.); campus@portocontericerche.it (M.C.); piu@portocontericerche.it (P.P.P.); siliani@portocontericerche.it (S.S.); sanna@portocontericerche.it (M.S.); roggio@portocontericerche.it (T.R.)

**\*** Correspondence: catzeddu@portocontericerche.it; Tel.: +39-(0)79-998448

Received: 30 August 2019; Accepted: 16 September 2019; Published: 18 September 2019

**Abstract:** Fresh pasta (SP) was prepared by mixing semolina with liquid sourdough, whole wheat semolina based, and the effects of sourdough inclusion were evaluated against a control sample (CP) prepared using semolina and whole wheat semolina. Physicochemical, nutritional, and sensorial analyses were performed on pasteurized fresh pasta, before and after cooking. The optimum cooking time was not affected by whole wheat sourdough, whereas differences were found in color, firmness, and cooking loss. Changes of in vitro digested starch fractions in SP pasta were affected by a higher cooking loss. Overall, SP samples were characterized by improved nutraceutical features, namely higher content of free essential amino acids and phenolic compounds, lower phytic acid content, and higher antioxidant activity. Sensory analyses (acceptability and check-all-that-apply (CATA) tests) showed significantly higher scores for the SP, and the differences were enhanced when the consumers were informed about the product composition and how it was manufactured. Consumers checked for more positive sensory parameters for the SP than the CP.

**Keywords:** sourdough; fiber; amino acids; phenolic compounds; phytic acid

#### **1. Introduction**

Pasta is a staple food in the Mediterranean area, and it is produced and consumed worldwide. It is a good source of carbohydrates and proteins, with interesting nutritional properties, i.e., a low glycemic index. In recent years there has been a trend towards the production of whole wheat pasta, which represents a good source of fiber. The intake of dietary fiber exerts beneficial effects on human health. In fact, bran, a by-product of wheat milling, obtained from the outer layers of wheat kernel, contains fibers, vitamins, and minerals. Many advantages have been associated with the consumption of bran fiber, in terms of risk reduction of hypertension, breast cancer, and type 2 diabetes, and in terms of prevention of colon disease and gastric cancer [1,2]. Bran fiber is resistant to digestion and absorption at the small intestine level, and thus reaches the colon where bacterial fermentation occurs and produces short chain fatty acids [3]. On the other hand, wheat bran contains phytic acid, recognized as an anti-nutritional compound [4], which reduces the nutritional value by chelating ions (such as Ca2<sup>+</sup>, Fe2<sup>+</sup>, Mg2<sup>+</sup>, and Zn2<sup>+</sup>). Mineral deficiency could lead to decreased function of the immune system and reduced body growth and development [5]. With regards to bread making technology, the use of bran has some drawbacks which negatively affects volume, texture, and sensory acceptance [6]. Pasta prepared with the addition of bran has an inferior technological quality as compared with pasta prepared with semolina [7] or wheat flour [8], because bran interferes with gluten development, especially when bran presents inappropriately sized particles [9].

The general recognition of the positive nutritional effects of fiber consumption increase the demand for technological solutions to overcome the negative effects of bran supplementation. In bread making technology, fermentation of bran by microbial strains has been suggested as a method to reduce the negative effects of phytic acid [10] and to improve the volume and sensory properties of bread containing bran [11].

Few scientific papers have reported on the use of fermentation technology in pasta making. A gluten-free pasta was produced by [12] using buckwheat flour and 24 h fermented semolina, a vitamin B2-enriched pasta was produced by [13] using a 16 h prefermented semolina, and fresh pasta was produced by [14] using semolina and semolina-based liquid sourdough. In this study, fermented whole wheat semolina was used as a functional ingredient in pasta making, in order to ameliorate the detrimental effects of bran fraction over the structure and sensory features, while retaining the advantageous effects of bran on human health. Physical, chemical, sensorial, and nutritional characteristics were evaluated.

#### **2. Materials and Methods**

#### *2.1. Raw Materials*

Commercial whole semolina (Integrale, Selezione Casillo S.r.l., Corato, Bari, Italy), and commercial semolina (Extra Arancio, Selezione Casillo S.r.l., Corato, Bari, Italy) were used in this study. The whole semolina had the following percent composition, as is or on the basis of dry matter (DM): moisture 14.1%, ash 1.6% DM, protein 12.5% DM, fiber 7.8% DM, dry gluten 8.5% DM, gluten index 60, alveographic W 199 (J <sup>×</sup> 10<sup>−</sup>4) and P to L ratio 5.12. The composition of semolina was the following: moisture 14.0%, ash 0.75% DM, protein 13% DM, fiber 2.7% DM, dry gluten 11% DM, gluten index 88, alveographic W 176 (J <sup>×</sup> <sup>10</sup><sup>−</sup>4) and P to L ratio 1.31.

#### *2.2. Preparation and Maintenance of Liquid Whole Wheat Semolina-Based Sourdough (LWS)*

A liquid whole wheat semolina-based sourdough (LWS) was prepared starting from the semolina-based liquid sourdough used by Fois et al. [14], which was refreshed for several days by back-slopping using whole wheat semolina and water, at a ratio of 1:1.5:1.5, in order to obtain a dough yield of 200. Back-slopping was done in the bioreactor GL MINI 25 (Esmach S.p.A., Grisignano di Zocco, Italy). The fermentation process was conducted at 26 ◦C for 5 h, and then LWS was kept at 5 ◦C until the subsequent daily back-slopping. The product was monitored daily in order to achieve stable values of pH and total titratable acidity (TTA).

The ripe sourdough had a pH value of 4.3, a TTA of 12.4 mL NaOH (0.1 mol/L) in 10 g, and a viable cell number of approximately 107 cfu/g for yeast and 10<sup>5</sup> cfu/g for lactobacilli. Yeast cells were enumerated on Rose Bengal Chloramphenicol agar, and lactobacilli on de Man, Rogosa and Sharpe (MRS) modified agar medium [14]. Media and supplements were purchased from Oxoid (Basingstoke, England).

#### *2.3. Physicochemical Analyses of Pasta and LWS*

A thermogravimetric analyzer Thermostep (Eltra GmbH, Haan, Germany) was used for moisture and ashes content determination, at 105 ◦C and at 600 ◦C respectively. Both TTA and pH were determined with an automatic titrator (Crison, Hach Lange, Barcelona, Spain), after homogenization of 10 g of sample in 90 mL of distilled water. After 30 min of gentle stirring for sourdough, and 60 min for pasta, the pH was determined, and the samples were titrated to a pH of 8.5 with NaOH 0.1 mol/L. The TTA was reported as mL of NaOH per 10 g of sample. The total dietary fiber (TDF) content of pasta was measured using the Total Dietary Fiber assay kit (Megazyme, Wicklow, Ireland). The color was determined on raw spaghetti placed side by side without any gap, using a CM-700d spectrophotometer (Konica Minolta, Osaka, Japan), using Standard Illuminant D65/10◦. Prior to measurements, the instrument was calibrated against the white tile. CIE L\*a\*b\* color space coordinates,

lightness (L\*), color in the red/green field (a\*) and color in the blue/yellow field (b\*), were computed. The Euclidean distance between colors, calculated as ΔE76, was used to estimate the range of perceived difference between samples of close chroma [15]:

$$
\Delta E\_{76} = \sqrt{(L\_2^\* - L\_1^\*)^2 + (a\_2^\* - a\_1^\*)^2 + (b\_2^\* + b\_1^\*)^2} \tag{1}
$$

0 < Δ76 < 1 the difference is unnoticeable; 1 < Δ76 < 2 the difference is only noticed by an experienced observer; 2 < Δ76 < 3.5 the difference is also noticed by an unexperienced observer; 3.5 < Δ76 < 5 the difference is clearly noticeable; 5 < Δ76 gives the impression that these are two different colors

Protein content (*N* × 5.27) was determined in 200 mg of pasta by the crude protein AACC (American Association of Cereal Chemists) combustion method 46–30 [16] using a Rapid N Cube analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany).

#### *2.4. Pasta Making*

The pasta was manufactured using the pasta maker "La Monferrina Dolly" (Moncalieri, Italy) equipped with a bronze die. The control pasta (CP) was prepared by mixing semolina (700 g) and whole wheat semolina (280 g) with 318 mL of water. Pasta with LWS, hereafter called sourdough pasta (SP), was prepared by mixing 38 mL of water, 700 g of semolina, and 560 g of LWS (consisting of 280 g of whole meal semolina and 280 mL of water). The dough was mixed for 20 min before extrusion into "spaghetti" of 3 mm diameter. After production, the CP and the SP were immediately pasteurized (at 97 ◦C for 3 min) and packaged under modified atmosphere (CO2:N2 = 30:70) [14]. For the analysis of total starch, protein digestibility, free amino acids, total phenolic compounds, antioxidant activity, and phytic acid, the samples were homogenized using a cryogenic mill (SpexSamplePrep, Stanmore, UK) and stored at −80 ◦C. All the analyses were replicated three times (*n* = 3).

#### *2.5. Cooking Quality and Texture Analysis*

The optimum cooking time (OCT), cooking loss (grams of solids in cooking water per 100 g of pasta as is), and swelling index (SI), i.e., grams of absorbed water per gram of pasta DM were measured for the CP and the SP according to the AACC Approved method 66–50 [16].

The texture of pasta, cooked to OCT, was evaluated according to the AACC Approved method 66–50 [16], using a TA.XTPlus Texture Analyzer (Stable Microsystems, Godalming, UK). The spaghetti strands were rinsed in cool water (4 ◦C) for 30 s to avoid overcooking. Tests were performed on 5 spaghetti strands, cut crosswise by the plexiglass blade probe A/LKB-F, at a test speed of 0.17 mm/s and a distance of 4.5 mm. The maximum force (*N*) of the curve, referred to as "firmness", was computed. The software Texture Expert Exceed (v1.21) (Stable Microsystems, Godalming, UK) was used for texture data processing.

#### *2.6. In Vitro Starch Digestibility*

In vitro digestion of starch was performed on the CP and the SP, in order to quantify rapidly digestible starch (RDS), slowly digestible starch (SDS), and inaccessible digestible starch (IDS). RDS is the glucose released after 20 min of in vitro digestion. SDS and IDS are defined as the glucose released in the time frame between 20 and 120 min and between 120 and 180 min, respectively. IDS is defined as "inaccessible digestible starch" since it is not actually digestion-resistant starch, but just physically inaccessible to the digestive enzymes, and it was made accessible by homogenization of the sample after 120 min of in vitro digestion (Sanna et al., 2019).

Samples were cut 2 cm long and cooked to OCT, and then processed as in Sanna et al., [17].

#### *2.7. Protein Digestibility and Free Amino Acid Analysis*

The raw and cooked CP and SP samples were analyzed for protein digestibility [18,19] and free amino acid content. Free amino acid extract was prepared mixing 10 mL of 0.1 M hydrochloric acid solution and 1.5 g of homogenized sample, then, the mixture was vortexed for 10 s and centrifuged at 18,000× *g* at 4 ◦C for 15 min. Finally, the supernatant was filtered through a 0.22 μm PTFE syringe filter (Phenomenex, Macclesfield Cheshire, UK). The amino acid analysis was performed using an Agilent 1200 series HPLC system (Santa Clara, California, CA, USA), equipped with a binary pump with integrated vacuum degasser, an autosampler, a thermostated column compartment, and a diode array detector (DAD). The system was controlled by the Agilent Chemstation chromatography manager. The pre-column derivatization, gradient eluent method, and injection program were performed according to Agilent Application Note [20], using an Agilent Poroshell HPH-C18 column (4.6 × 100 mm, 2.7 μm pore size; Agilent Technologies, Santa Clara, California, USA) with HPH-C18 fast guard (Agilent Technologies, Santa Clara, California, USA), at a temperature of 40 ◦C. The pre-column derivatization was performed using *o*-phthalaldehyde reagent (OPA), 9-fluorenylmethyl chloroformate reagent (FMOC), and borate buffer supplied by Agilent Technologies. Standard solutions were purchased from Agilent Technologies, whereas GABA (γ-aminobutyric acid) was purchased from Sigma-Aldrich S.r.l. (Milano, Italy). Results were expressed as mg/100 g DM.

#### *2.8. Total Phenolic Content and Antioxidant Activity*

The total phenolic content (TPC) and 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activity were determined on raw and cooked CP and SP, by suspending 1 g of sample in 10 mL of an 80% aqueous methanol solution (20:80, *v*/*v*). The mixture was shaken for 2 h at 750 rpm in a thermomixer (Thermomixer Comfort, Eppendorf), then centrifuged at 800× *g* for 10 min. Supernatant was filtered through a 0.22 μm PTFE (Polytetrafluoroethylene) syringe filter (Phenomenex, Macclesfield Cheshire, UK) and stored at −20 ◦C until analyses.

The TPC of sample extracts was determined using the Folin–Ciocalteau method [21] with the following modifications: 0.2 mL of the sample extract (or 80% methanol for the blank) was mixed with 1.5 mL of Folin–Ciocalteau reagent, previously diluted with water (1:10 *v*/*v*), and 1.5 mL of saturated sodium carbonate solution (7.5% *w*/*v*). The mixture was allowed to stand in the dark at room temperature for 1 h, then the absorbance was read at 735 nm, against the blank. The gallic acid calibration curve was built in the range 25–600 mg L−<sup>1</sup> (y = 0.0051x + 0.071 *R*<sup>2</sup> = 0.999) and results were expressed in terms of Gallic acid equivalents (GAE mg mL<sup>−</sup>1).

The antioxidant activity of pasta was determined through the evaluation of free radical scavenging effect on the DPPH radical, according to [22] with some modifications: 1.4 mL of DPPH solution (0.10 mM in an 80% aqueous methanol) was mixed with 0.1 mL of sample extract, in a 1.5 mL centrifuge tube, vortexed, and then allowed to stand for 30 min in the dark. The discoloration of DPPH against the 80% aqueous methanol was monitored after 30 min, measuring the absorbance at 517 nm. The antioxidant activity of the sample was expressed as the percentage discoloration of DPPH solution, by the following equation:

$$\% \text{sickoration} = [(\text{AbsDPPH} - \text{AbsSample}) \text{(AbsDPPH}] \times 100\tag{2}$$

where, AbsDPPH is the absorbance of the DPPH solution without extract, and AbsSample is the absorbance of the sample solution after 30 min of reaction.

All reagents were analytical grade and purchased from Sigma-Aldrich S.r.l. (Milano, Italy).

#### *2.9. Phytic Acid Determination*

Phytic acid was determined on the raw and cooked CP and SP using the Phytic Acid (Phytate)/Total Phosphorus assay kit (Megazyme, Wicklow, Ireland).

#### *2.10. Consumer Testing*

An acceptability test and a check-all-that-apply (CATA) method [23] were performed on the CP and SP samples. The test was carried out by 54 consumers, 26 women and 28 men, most of them recruited on the basis of interest and willingness. They were regular pasta consumers, aged between 32 and 60 years, not trained on sensory analysis of pasta products. Approximately, 40 g of packed sample were supplied to each consumer and the test was performed at their own home, under real conditions of use and consumption [24], so that consumers could season the pasta with the preferred sauce. The optimal cooking time, and the avoidance of spice and chili were suggested.

The evaluation of the SP and CP samples was carried out in two separate conditions, at two different sessions. The first condition was a blind test and the second condition was an informed test, in which consumers, before evaluating pasta, were informed on how the SP and CP pasta were produced.

Consumers were asked to give a judgment for acceptability by scoring for the following attributes: flavor, taste, texture, and overall acceptability. A nine-point structured hedonic scale ranging from 1 (extremely disliked it) to 9 (extremely liked it) was used, and a sample was considered acceptable when it scored above 5 (neither like nor dislike). Then, consumers answered the CATA questionnaire, containing 12 phrases related to sensory, hedonic, and functional properties of the samples, which were the following: Do you believe that it is a wholesome food? Do you believe that it is ideal for a balanced diet? Do you feel it satiating? Do you sense a distinctive odor? Does it remind you of home-made pasta? Does it feel hard to chew? Do you feel it is tasty? Do you feel any unusual taste? Do you feel it "al dente" (cooked to OCT)? Does it absorb the sauce well? Do you feel it is sticky? Is it pleasantly sour?

The consumers were forced to answer "yes" or "no", checking all phrases that applied as suitable to describe the product.

#### *2.11. Statistical Analysis*

Standard ANOVA procedure (randomized complete design with three replicates and two treatments) was applied on the dataset. The means were separated by the LSD test at *p* = 0.05 significance level, using the Statgraphics Centurion software package (version16.1.11, Statpont Technologies Inc, Warrenton, VA, USA).

Hedonic scores collected in the acceptability test were analyzed by analysis of variance (ANOVA). The Cochran's *Q* test at *p* = 0.05 was used to analyze the CATA data.

#### **3. Results**

#### *3.1. Physicochemical Characteristics and Cooking Properties*

Physicochemical characteristics of samples are reported in Table 1. The fermentation process did not affect ashes, moisture, and protein content, which were similar in the SP and CP samples, nor the total dietary fiber content, which was measured prior to cooking. Moisture increased after cooking, similarly in the CP and SP, and this was obviously due to the water absorbed during cooking. Ashes decreased after cooking, probably because mineral salts were lost in the boiling water. After cooking, protein percentage increased in the CP, whereas it did not vary in the SP and probably this was an effect of cooking loss, as discussed later and displayed in Table 2. Cooking caused an increase in the pH of the SP and a decrease of the TTA values in the CP and SP, indicating that organic acids diffused into the cooking water. The lightness (L\*) and the yellow color (b\*) were significantly higher in the SP than in the CP, as showed in Table 1, suggesting a greater retention of the pigments which could contribute to a higher antioxidant activity [14,25]. The Euclidean color distance, ΔE76, was 4.26, depicting a clearly noticeable difference at human sight [15].

Table 2 shows the results of cooking quality analysis and texture of cooked pasta. The optimal cooking time was seven minutes for both the SP and the CP. The CP and SP showed the same swelling index, indicating the same amount of water absorbed, whereas differences were found in the cooking loss, which was significantly higher for the SP samples. Analysis of solids leached into the cooking water, indicate that the SP released more proteinaceous substances than the CP. Moreover, the SP showed a lower firmness than the CP.



significant difference at *p* ≤ 0.05. 1 DM, dry matter; TTA, total titratable acidity; L, lightness; a, color in the red/green field; b, color in the blue/yellow field; CP, control pasta; SP, sourdoughpasta. a,b,c Means with different letters for each parameter indicate significant differences (*<sup>p</sup>* < 0.05).

**Table 2.** Cooking quality parameters, texture properties, and protein digestibility and availability of cooked pasta. CP, control pasta; SP, pasta with sourdough. Mean values are reported.


Different superscript letters for the same treatment denote a statistically significant difference at *p* ≤ 0.05. \*, grams of solids in cooking water per 100 g of pasta (dry matter); \*\*, grams of absorbed water per gram of pasta (dry matter); \*\*\*, g of proteins lost in water after cooking per 100 g of pasta (dry matter). a,b Means with different letters for each parameter indicate significant differences (*<sup>p</sup>* < 0.05).

#### *Foods* **2019** , *8*, 422

During the cooking of pasta, starch is normally leached into the water, together with soluble proteinaceous material. Kordonowy and Youngs [26] reported that the cooking loss was higher in bran-containing food, as a consequence of water-soluble components of bran and gluten dilution. In the SP the protein loss in the cooking water was higher than in the CP, probably because proteins have been partially hydrolyzed to peptides and amino acids by microbial proteases and, to a greater extent, by endogenous proteases and peptidases, which are active at a low pH. The proteolytic activity on the gluten proteins also explains the lower firmness of cooked pasta.

#### *3.2. In Vitro Starch Digestibility*

The results of glucose release after in vitro starch digestion and total starch values are reported in Figure 1. No significant differences were found between samples for the rapidly digestible starch (RDS), slowly digestible starch (SDS), and total digestible starch (TDS), whereas a significant difference was found for the inaccessible digestible starch (IDS), which resulted in being significantly lower in the SP sample. The differences observed for the TDS (44 g/100 g DM for the SP against 48 g/100 g DM for the CP), nevertheless not significant at *p* = 0.05, are fairly high. The acidic conditions of the SP before the heat treatment (pasteurization) should have been responsible for a stricter interaction between starch and gluten [27], and thus an increase of IDS, the starch fraction indigestible because of the food structure, was expected in the SP. Moreover, several studies have investigated the effects of dietary fiber in food and pasta on in vitro digestibility [28,29] showing that dietary fiber might have been responsible for the formation of a polysaccharide network that could encapsulate the starch granules during processing. The entrapment of starch reduces accessibility to enzymatic degradation, and therefore reduces the sugars released in the blood [30]. Sourdough technology applied to whole wheat bread already has been proved to retard postprandial glucose and insulin response of bread, with respect to yeast leavened whole wheat bread [31]. The decrease of IDS in the SP could be explained taking into account the higher cooking loss observed in the SP samples, which was a consequence of a greater disruption of food structure during cooking, thus altering the relative percentages of starch fractions. Sourdough determines the partial hydrolysis of proteins, causing a higher loss of peptides and amino acids in the cooking water and, at the same time, a weakening of the gluten matrix. A weaker gluten matrix is less able to entrap swollen starch granules during cooking, resulting in higher cooking loss of starch, which explains the lower content of some starch fractions in the SP.

**Figure 1.** Results of in vitro digestion of pasta samples. RDS, rapidly digestible starch; SDS, slowly digestible starch; IDS, inaccessible digestible starch; TDS, total digestible starch. Black bars, CP and grey bars, SP. Different letters (a, b) for the same starch fraction denote a statistically significant difference between samples at *p* ≤ 0.05.

#### *3.3. Protein Digestibility and Amino Acid Content*

The digestibility of protein and the total protein availability (i.e., protein content\*digestibility) are reported in Table 2**,** expressed as a percentage of the average of triplicate runs. The digestibility is significantly lower in the SP (83%) as compared with the CP (86%), and consequently the value of protein availability is lower in the SP. The method used to estimate protein digestibility [18] measures the pH drop, after 10 min of hydrolysis, of an aqueous protein suspension that has been adjusted to a pH of 8.0 using NaOH. During enzymatic hydrolysis, carboxyl (-COO−) and amino (-NH3+) groups are released from proteins. Protons, (H+), released into the surrounding reaction medium give rise to a decrease of the pH. The lower protein digestibility found in the SP is, therefore, due to the fact that part of the accessible peptidic bonds, which are the source of protons measured during the procedure, had already been broken by protease of the liquid sourdough. Moreover, as observed by Desai et al. [19] phenolic compounds (which are present at higher amount in the SP sample, as discussed later on) can contribute to a decrease in the protein digestibility. Therefore, pH drop method results must be carefully interpreted when they come out from the analysis of fermented products.

Table 3 shows the quantity of the 20 individual free amino acids (FAAs) measured in raw and cooked CP and SP. Aspartic acid, glutamic acid, asparagine, arginine, alanine, GABA and tryptophan had the highest concentrations in all samples, whereas the least abundant FAAs were glutamine, histidine, methionine and isoleucine. In the SP, the mean value of total FAAs was 197.41 mg/100 g and 171.91 mg/100 g, raw and cooked, respectively. For the CP it was 159.65 and 153.39 mg/100 g, raw and cooked, respectively. The SP had the highest total FAA content, likely due to the proteolytic activity of the sourdough, and significant (*p* < 0.05) differences could be found between the SP and CP samples for all amino acids, except for glycine, alanine, tryptophan, lysine, and proline. Commonly, the amount of each FAA was higher in the SP than the CP, except for GABA, asparagine, and glutamine which were higher in CP. SP also had a significantly higher content of total free essential amino acids (EAAs).


**Table 3.** Free amino acid content in pasta samples (mg/100 g of pasta dry matter).

Significance of the *F*-test after ANOVA: ns, not significant; \*, significant for *p* ≤ 0.05; \*\*, significant for *p* ≤ 0.01; \*\*\*, significant for *p* ≤ 0.001. <sup>1</sup> FAA, free amino acids and <sup>2</sup> EAA, essential amino acids.

These results show the important role of the microbial fermentation in order to obtain a higher content of FAAs and EAAs, since the acidic conditions of sourdough activated the cereal proteinases. Furthermore, microbial peptidases released small peptides and FAAs into the food matrix [32]. Such a phenomenon already has been observed in bread. Lappi et al., [31] found more solubilized and smaller molecular weight proteins and peptides in sourdough than in yeast bread.

The high level of glutamic acid in the SP was an effect of microbial deamidation of glutamine [33]. This metabolic pathway plays an important role in pH homeostasis and acid resistance of microorganisms. As reported by Zhao et al. [34], a high level of glutamic acid can also result in a more pronounced food taste. Glutamic acid is a metabolic precursor of GABA, a non-protein amino acid naturally present in cereals in small quantities and having different health benefits. Numerous studies describe an increase of GABA in sourdough produced with selected lactic acid bacteria strains [35]. In this study a lower amount of GABA was found in the SP with respect to the CP, likely due to a consumption by the yeast [36]. Moreover, the decrease of asparagine observed in the SP as compared to the CP, also can be due to the yeast metabolism. These results confirm that microbial activity can affect nutritional properties and the taste of fresh pasta.

#### *3.4. Total Phenolic Content, Antioxidant Activity, and Phytic Acid Content*

The total phenolic content (TPC) and the antioxidant activity, measured as a percentage of discoloration of DPPH radical with respect to a blank sample, are reported in Table 4. In raw pasta, both the TPC and the antioxidant activity values were higher in the SP samples (37.0 mg/100 g pasta DM and 6.3%) than the CP samples (23.5 mg/100g pasta DM and 3.6%). It is known that fermentation improves the bioavailability of phenolic compounds and induces enhancement of antioxidant activity, through a mechanism reviewed by Hur et al. [37]. In addition, cooking had a positive effect on the TPC and antioxidant activity, which significantly increased in both the CP and SP. The increase of the TPC in pasta after cooking has already been reported in pasta enriched with bran fractions [38], and it was mainly ascribable to the increase of ferulic acid [39,40]. The increase of the antioxidant activity is consistent with the observed increase in the TPC, in agreement with Fares et al. [39], who stated that ferulic acid esters that are linked to cell walls are released by the pasta matrix during cooking and that they do not lose their antioxidant capacity, even after the hydrothermal treatment. Note that the combined effect of fermentation and cooking made the level of the TPC in the SP to be about three times higher than that of the CP sample, but the concomitant increase in antioxidant activity (from 6.33% to 7.83%) did not have such a high extent. This is probably because ferulic acid shows a relatively weak antiradical effect [41].


**Table 4.** Phytic acid and antioxidant activity in raw and cooked pasta. CP, control pasta and SP, pasta with sourdough.

Significance of the F-test after ANOVA: ns, not significant; \*\*, significant for *p* ≤ 0.01; \*\*\*, significant for *p* ≤ 0.001. Different superscript letters for the same treatment denote a statistically significant difference at *p* ≤ 0.05. <sup>1</sup> mg of gallic acid per 100 g of pasta (dry matter); <sup>2</sup> percentage of discoloration referred to blank sample. a,b,c Means with different letters for each parameter indicate significant differences (*p* < 0.05).

The amount of phytic acid was different between samples (Table 4), being lower in the SP as compared with the CP, both in raw and cooked samples. The phytic acid increased significantly after cooking, in both SP and in CP.

Bioavailability of essential nutrients, such as minerals and amino acids, is strongly reduced by the chelating properties of phytic acid in food stuff containing bran. Kordonowy and Youngs [26] found that the addition of bran in pasta increased the phytic acid content, and also observed the loss of phytic acid into the cooking water. The chelating properties of phytic acid can be inactivated at acidic conditions by the endogenous phytase of wheat [42]. In our work the significant reduction of phytic acid in the SP with respect to the CP was due to the acidic conditions generated by fermentation, and this is an important consequence from a nutritional point of view.

#### *3.5. Consumer Testing*

The main purpose of this study was to analyze how the use of fermented whole wheat semolina affected sensory properties of fresh pasta, taking into account that other authors reported lower sensory scores for high fiber pasta than for semolina pasta [7,26]. Figure 2 shows the results of the acceptability test performed on the CP and the SP, which was planned in two subsequent steps, the first one aiming to know the responses of consumers not informed on the SP properties (blind test), and the second one to know the responses of consumers informed on the preparation of the sample (informed test). As indicated in Figure 2, the scores of the CP sample were significantly lower for most of the sensory parameters as compared with those of the SP samples, except for the flavor which was similar to the value of the SP sample in the blind test but not in the informed test. Significant differences between the blind and informed tests, for SP samples, were found for flavor and overall acceptability, indicating the significant influence of information on food acceptability. Previous researches demonstrated that information on food stuff (brand, manufacturing, nutritional properties, etc.) might affect its hedonic rating [43,44].

**Figure 2.** Results of the acceptability sensory test performed on the CP sample (white bars) and on the SP samples, the latter as blind (grey bars) and informed tests (black bars). Different letters (a, b, c) for the same sensory parameter denote a statistically significant difference between samples at *p* ≤ 0.05.

Table 5 reports the relative frequencies of positive answers of consumers subjected to the CATA test. For the most part, the SP samples collect a greater number of positive responses than CP sample. Only the questions "Do you feel it hard to chew" and "Do you feel it sticky" collected a lower score and CP sample was found to be harder and more sticky than SP samples. To note that in the informed

test the SP sample collected a greater number of positive responses than in the blind test, confirming that the information on the food preparation (i.e., the fermentation of semolina) had a positive impact on consumers' expected food quality. The information provided on pasta influenced positively the "satiating" sensation, the "al dente" property and the "absorption of sauce", which obtained an higher score than the SP blind sample, whereas the "unusual taste", observed in the blind sample, disappeared in the informed one.

**Table 5.** Check-all-that-apply (CATA) test results. The reported values indicate the relative frequency of positive answers. In the first column the main attributes representing the questions addressed to the consumers, as follows: Do you believe that it is a wholesome food? Do you believe that it is ideal for a balanced diet? Do you feel it is satiating? Do you sense a distinctive odor? Does it remind you of home-made pasta? Does it feel hard to chew? Do you feel it is tasty? Do you feel any unusual taste? Do you feel it "al dente" (cooked to OCT)? Does it absorb the sauce well? Do you feel it is sticky? Is it pleasantly sour?


Significance of the Cochran *Q*-test \*\*\*, significant for *p* ≤ 0.001. Different superscript letters for the same treatment denote a statistically significant difference at *p* ≤ 0.05. a,b,c Means with different letters for each parameter indicate significant differences (*p* < 0.05).

#### **4. Conclusions**

The addition of fermented whole wheat semolina affected many quality features of fresh pasta. Differences were found in color, firmness and cooking loss, while the optimal cooking time was the same for both samples. Notably, the SP samples were characterized by improved nutraceutical characteristics, showing a higher content in total and essential free amino acids, phenolic compounds, and resulting DPPH scavenging activity, and a decreased content of phytic acid. The results of sensorial analysis indicate an increase in the overall quality of pasta obtained using fermented whole wheat semolina, suggesting a new way for the sensorial improvement of high fiber pasta. The results of the acceptability test highlighted the differences between the CP and SP, with the latter having higher scores for all sensory parameters. The highest overall acceptability score was obtained from the SP sample after consumers were informed that the SP contained sourdough, indicating consumer interest in the addition of the functional ingredient. This study demonstrated that whole wheat sourdough is a valuable functional ingredient in fresh pasta making. Studies are in progress with in vivo trials, to investigate the nutraceutical properties of this innovative fresh pasta.

**Author Contributions:** Conceptualization, P.C; formal analysis, S.F, M.C, P.P.P, S.S, and M.S; funding acquisition, T.R; investigation, S.F, M.C, P.P.P, S.S, and M.S; resources, T.R; supervision, P.C; writing—original draft, S.F and M.C; writing—review and editing, P.C.

**Funding:** This work was supported by Sardegna Ricerche, Science and Technology Park of Sardinia, and Regione Autonoma della Sardegna, under grant program art.9 LR 20/2016 (2017) to Porto Conte Ricerche.

**Acknowledgments:** We gratefully acknowledge Paola Conte, PhD, for technical assistance in fiber analysis.

**Conflicts of Interest:** The authors declare that there are no conflicts of interest in this research article.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **The E**ff**ect of** *Moringa oleifera* **Leaf Powder on the Physical Quality, Nutritional Composition and Consumer Acceptability of White and Brown Breads**

#### **Laurencia Govender and Muthulisi Siwela \***

Discipline of Dietetics and Human Nutrition, School of Agricultural, Earth and Environmental Sciences, University of KwaZulu-Natal, Private Bag X01, Scottsville 3209, Pietermaritzburg 3201, South Africa; GovenderL3@ukzn.ac.za

**\*** Correspondence: Siwelam@ukzn.ac.za; Tel.: +27-33-260-5459

Received: 30 October 2020; Accepted: 3 December 2020; Published: 21 December 2020

**Abstract:** Fortifying popular, relatively affordable, but nutrient-limited staple foods, such as bread, with *Moringa oleifera* leaf powder (MOLP), could contribute significantly to addressing under nutrition, especially protein and mineral deficiencies, which are particularly prevalent among a large proportion of populations in sub-Saharan African countries. The current study aimed to determine the effect of MOLP on the physical quality, nutritional composition and consumer acceptability of white and brown breads. The texture, colour and nutritional composition of white and brown bread samples substituted with 5% and 10% (*w*/*w*) MOLPs were analysed using standard methods and compared with the control (0% MOLP). A consumer panel evaluated the acceptability of the bread samples using a nine-point hedonic scale. Bread samples became darker as the concentration of MOLP was increased, whilst nutrient levels increased. The overall consumer acceptability of the bread samples decreased with increasing concentrations of MOLP. However, brown bread samples were significantly more acceptable compared with corresponding white bread samples (*p* < 0.05). Under the experiment conditions of the current study, it seems that the bread containing 5% MOLP can be used to contribute significantly to addressing malnutrition, with respect to protein deficiency.

**Keywords:** *Moringa oleifera* leaf powder (MOLP); bread; fortification; nutritional composition; consumer acceptability

#### **1. Introduction**

Food insecurity and malnutrition are significant problems globally, specifically undernutrition (wasting, stunting, underweight and micronutrient deficiencies) and hunger [1]. Moderate and severe hunger affects approximately two billion individuals worldwide [1], which contributes significantly to the high rates of malnutrition seen in the world. Approximately 21.3%, 6.9% and 5.6% of children globally are stunted, wasted and overweight, respectively. Additionally, the 2020 nutrition global report indicates that one in nine and one in three individuals are hungry or malnourished and overweight or obese, respectively [2]. The COVID-19 pandemic has contributed to the increase in undernutrition, especially in countries where people are facing financial difficulties [1]. Monotonous diets consisting of largely starchy staple foods are mostly consumed in low- to middle-income countries, such as South Africa (SA). Further, the majority of communities from these countries consume limited amounts of fruits and vegetables and animal source foods [1]. Animal source foods are high in quality protein, but are less affordable to many impoverished households in SA and other developing countries, compared to plant-based protein sources [3]. This type of diet also lacks dietary diversity and could also lead to micronutrient deficiencies. Micronutrient deficiencies are a public health concern, especially in developing countries such as SA. Vitamin A, iron and zinc deficiencies are particularly problematic [4]. One of the main reasons for the trends of an increase in poor diets is the rising food costs. A basic 28-item food basket in SA in 2020 costs approximately USD 53.69 [3], which can be unaffordable to many, especially those that are hard hit by the current economic situation and global pandemic. The poorest are the most affected and are at significant risk for food insecurity [5]. Over the years, the consumption of indigenous crops such as *Moringa oleifera* has decreased due to more Westernised cultures being adopted [6]. This has resulted in less dietary diversity as foods from the formal markets (supermarkets, etc.) are expensive. Indigenous crops, such as *Moringa oleifera,* are known to be nutrient-rich as well as have many health beneficial properties. *Moringa oleifera* is one of 13 species of the *Moringaceae* family of plants and is widely researched. This plant originated in India and Africa but is now widely grown in other parts of the world [7]. Not only can *Moringa oleifera* thrive under different climatic conditions—i.e., in tropical and subtropical countries—it also has nutritional, antioxidant and phytochemical benefits [8].

Furthermore, Moringa is a good source of iron, which is generally deficient in most leading staple plant-based diets such as the starchy staples [9]. Iron is an important micronutrient, especially during pregnancy as it contributes to foetal growth. A pregnant woman who has iron-deficiency anaemia is at great risk for perinatal and maternal mortality, premature delivery and having a low birth weight infant [10,11]. *Moringa oliefera* is also rich in vitamins such as the provitamin A beta-carotene, folic acid, pyridoxine and nicotinic acids and vitamins C, D and E [12]. Vitamin A deficiency is also common in SA and most other developing countries, especially in the sub-Saharan African region. The human body needs vitamin A, and its deficiency affects vision, growth, development, protein synthesis and could result in a child not being able to reach their full potential, both physically and mentally [13]. *Moringa oleifera* is also a good source of protein and contains 16–19 amino acids. Ten of these amino acids are essential [14]. Wasting can present in the form of protein-energy malnutrition (PEM) and is caused by a deficiency of good quality protein in the diet [15]. When an individual is malnourished, their body goes into a state of starvation and negatively affects the immune system, kidneys, cardiac muscle, liver and gastrointestinal tract [13,15]. Micronutrient deficiencies and PEM are commonly seen in vulnerable population groups, such as of a woman of childbearing age and children under five years. These groups are thus targeted for nutritional interventions.

Staple foods such as bread are commonly consumed food items in SA [16], and since October 2003, wheat flour fortification was made mandatory [17]. However, access to fortified foods still remains challenging to many impoverished individuals, as many of these individuals rely on social grants to purchase food [3,18,19]. Baked bread contains high amounts of energy, carbohydrates and fat, but is limited in other nutrients such as protein, minerals and vitamins [20]. To increase the nutritional composition of bread, *Moringa oleifera* leaf powder (MOLP) could be used as it is rich in proteins and several micronutrients that are deficient in bread. Bread is an affordable source of energy (in the form of starch) and, therefore, would be a suitable candidate for supplementation with MOLP [21]. Several studies have investigated the nutritional composition and consumer acceptability of bread fortified with MOLP (Table 1). Consumers living in different geographical locations (e.g., countries) may show different preference and acceptability levels for the same innovative food product developed from the same conventional product. Thus, there may be differences in consumer acceptability of a MOLP fortified bread across different countries. While studies conducted in other countries found that the dark colour and bitter taste of MOLP impacted negatively on consumer acceptability of bread [22,23], different results may be obtained with consumers living in SA. To the best of the researchers' knowledge, this study is first to investigate consumer acceptability of a MOLP fortified bread in SA. Brown bread is dark in colour, so it may be a suitable food item for supplementing with MOLP as its dark colour might not significantly change due to the addition of MOLP, which is also dark in colour. It seems none of the previous studies listed in Table 1 compared consumer acceptable of corresponding samples of white and brown breads supplemented with MOLP. Furthermore, the other authors from earlier studies [23–28] (Table 1) did not compare the nutritional composition obtained

from their respective study to the estimated average intake values. Consequently, the authors did not determine the exact amount of nutrients that would be obtained from the consumption of usual portions of MOLP fortified bread. Therefore, this study determined the effect of MOLP on the physical quality, nutritional composition and consumer acceptability of white and brown breads in SA.


**1.**Studiesconductedontheadditionof*Moringa*tobread.

**Table**


**Table 1.** *Cont.*

97

#### **2. Methodology**

#### *2.1. Study Design*

This study was a cross-sectional experimental design. Figure 1 presents a conceptual framework of the methodology.

**Figure 1.** Conceptual framework of the methodology.

#### *2.2. Preparation of Bread*

*Moringa oleifera* leaf powder was purchased from a local pharmacy in Pietermaritzburg, SA. The bread (both white and brown) was prepared using a standardised bread making recipe [29]. *Moringa oleifera* leaf powder partially replaced wheat flour at 5% and 10% substitution levels of MOLP in both white and brown breads, respectively. Standard white and brown bread types served as corresponding controls (0% MOLP) for the white and brown bread samples, respectively. Fresh bread samples were baked on the day of data collection.

#### 2.2.1. Ingredients

In this experiment, the ingredients listed below were used.

180–250 mL lukewarm water; 300 g brown/white bread flour; 10 g dry yeast; 3.8 g salt; 15 mL melted butter; 15 g or 30 g MOLP for substitution (This will be added after the flour and salt is sifted).

#### 2.2.2. Method

The method for making bread involved the sequential steps described next. Sift flour and salt into a large bowl. Make a well in the centre of the flour in the large bowl. In a small bowl, mix the dry yeast with 60 mL water and stir until dissolved. Pour yeast mixture into the well of flour and add warm melted butter and 180 mL of water. Mix the ingredients with your fingers, then beat with your hands, adding a little more water if necessary, to make a firm dough. Using your hands, fold and slap the dough against the side of the bowl until it starts to feel elastic and leaves the sides of the bowl. Place the dough onto a floured work surface and knead by folding the far edge towards you, then pushing it firmly away with the heel of the hand. Turn the dough a little and repeat. Continue kneading until the dough is smooth and elastic and springs back when you make a dent with your finger. Place the dough in a clean, warm, oiled bowl, turn it over so that the dough is slightly oiled all over, then cover it with oiled plastic wrap and a cloth. Leave to rise in a warm place for 1 h. After the allocated time, test by pushing a finger into the dough—if the indent remains, it is ready to knock back the dough by punching with your fist several times, squeezing out any large bubbles. Place onto a lightly floured surface and knead three to four times. Pat the dough until round, then fold sides under to make a neat oblong. Press together and seal, then place in a lightly oiled 23 × 12 cm bread tin. Cover loosely with oiled plastic wrap and a cloth, leave in a warm place until it rises to the top of the tin. Preheat the Defy Thermofan Stove (Model 731 MF) oven to 230 ◦C. Bake the bread at 230 ◦C for 15 min, then turn the tins around, reduce to moderate heat, i.e., 180 ◦C, and bake for a further 20–25 min. The bread is cooked if it sounds hollow when the underside is knocked with the knuckles. Take the bread out of tins and leave to cool on a wire rack.

#### *2.3. Physical Quality*

A Hunter Lab colourimeter was used to measure the colour in all six bread samples. The L measured lightness, a measured redness/greenness and b measured yellowness/blueness. A TA-XT2 Plus texture analyser was used to determine the texture of the bread samples [30]. The texture probe analyser (TPA) applied a force to the bread sample at a speed of 2.0 mm/s, and the force compression was recorded.

#### *2.4. Nutritional Composition*

The nutrition analysis was conducted on all six bread samples in duplicate for moisture, protein, fat, fibre (NDF), total mineral (ash), calcium, iron and zinc using standard methods [31].

#### 2.4.1. Moisture

The moisture content of all six bread samples was determined using the Association of Official Analytical Chemists (AOAC) Official Method 934.01 [31]. The bread samples were dried in an air circulated oven at 90 ◦C for 72 h. The moisture content was then calculated using the weight loss content of the samples.

#### 2.4.2. Protein

The protein content was measured with a LECO Truspec Nitrogen Analyser, using the AOAC Official Method 990.03 [31]. The six bread samples were individually placed in a combustion chamber with an autoloader at 950 ◦C. The percentage protein content was calculated.

#### 2.4.3. Fat

The fat content of the six bread samples was determined according to the AOAC Official Method 920.39 [31], following the Soxhlet procedure. A Büchi 810 Soxhlett Fat Extractor was used to extract the fat in petroleum ether which was then used to calculate the percentage fat.

#### 2.4.4. Fibre (NDF)

The fibre content of the bread samples was measured as Neutral Detergent Fibre (NDF). The NDF was determined according to the AOAC Official Method 978.10, using the Dosi-fibre system [31].

#### 2.4.5. Total Mineral Content (Ash)

The ash content, otherwise known as the total mineral content, was determined using the AOAC Official Method 942.05 [31]. The bread samples were placed in a furnace and heated at 550 ◦C for 72 h.

#### 2.4.6. Calcium, Zinc and Iron

The calcium (Ca), zinc (Zn) and iron (Fe) contents were determined according to the AOAC Official Method. The bread samples were dried at 105 ◦C for 2 h, then ashed in a furnace at 550 ◦C for 4 h. Deionized water and hydrochloric acid (HCl) were added to the ash and boiled in a water bath until dry; this was repeated twice to allow minerals to be drawn into the solution. After 24 h, an atomic absorption spectrophotometer was used to measure the Ca, Zn and Fe contents.

#### *2.5. Sensory Evaluation*

Fifty-four students and staff members from the agricultural campus of the University of KwaZulu-Natal (UKZN) were recruited to participate in the study. A pilot study was conducted before the main study using 10 participants to test the acceptability of the recipes and adjust the methods accordingly. In the pilot study, 0%, 10% and 20% (*w*/*w*) substitutions were used; however, the 20% substitution was not well accepted. Therefore, the MOLP substitution levels were adjusted to 0%, 5% and 10% (*w*/*w*) for the main study. The pilot study participants were not allowed to participate in the main study. To prevent participants from communicating with one another, they were placed in separate cubicles. Each of the six samples was assigned a unique three-digit code obtained by a table of random numbers, and the tables of random permutations of nine were used to determine the serving order [32]. Each participant was given about a quarter of a slice of each of the six bread samples on a polystyrene plate. Each panellist was given a cup of water so that they could rinse their palate between tasting each bread sample and the sensory evaluation questionnaire. The questionnaire provided was in English, as this is the language of instruction at UKZN. The questionnaire made use of the nine-point hedonic scale (1 = dislike extremely to 9 = like extremely) together with the sensory attributes (taste, colour, aroma, texture, appearance and overall acceptability), which was explained to the panellist before the commencement of the sensory evaluation. Participants were given limited information on MOLP so that bias was prevented. Research assistants helped panellists, if necessary, during the sensory evaluation.

#### *2.6. Ethical Consideration*

Ethical approval was obtained from the UKZN Humanities and Social Science Ethics Committee (HSS/1244/015D). The gatekeeper's permission was obtained from the registrar of UKZN. All panellists were required to sign a written consent form before participating in the study.

#### *2.7. Statistical Analysis*

Data from the physical quality, nutritional composition and sensory evaluation questionnaires were entered into an Excel spreadsheet and cross-checked for accuracy. Thereafter, data were transferred to the Statistical Package for Social Science® (SPSS) version 25 (IBM Corp., Armonk, NY, USA) for analysis. Appropriate statistical techniques, including the Bonferroni and Tukey tests, were used to analyse the data. A *p*-value of <0.05 was considered to be statistically significant.

#### **3. Results**

*3.1. The E*ff*ect of Moringa oleifera Leaf Powder on the Physical Quality of White and Brown Breads*

Figure 2 shows the six bread samples, and Table 2 presents the effect of MOLP on the colour and texture of the bread.

**Figure 2.** Depicts the different bread samples ((**A**): Brown bread (control, 0%); (**B**): Brown bread (5%) (**C**) *Moringa oleifera* leaf powder [MOLP]); Brown bread (10% MOLP); (**D**): White bread (Control, 0%); (**E**): White bread (5% MOLP); (**F**): White bread (10% MOLP)).

**Table 2.** The effect of *Moringa oleifera* leaf powder fortification on colour and texture of white and brown breads.


Different letters in columns show significant difference according to the Bonferroni test (*p* < 0.05).

The Bonferroni test indicated that there was no significant difference between both control bread and bread containing 5% and 10% MOLPs. However, there was a significant decrease in the lightness of bread with the addition of MOLP. White bread containing 10% MOLP had a darker colour in comparison to brown bread containing 10% MOLP (Figure 2 and Table 2). There was no effect on the texture of the bread samples with the addition of MOLP at different substitution levels.

#### *3.2. The E*ff*ect of MOLP on the Nutritional Composition of White and Brown Breads*

Table 3 presents the nutritional composition of white and brown breads containing MOLP.

The Tukey test indicated that the total mineral content in white bread significantly increased when MOLP was added. There was a slightly significant increase in the total mineral content when 5% MOLP was added to brown bread, and there was a significant increase in brown bread with 10% MOLP (*p* < 0.05). A significant increase in protein content was seen in white bread containing 5% and 10% MOLPs. Further, as the MOLP concentration increased in white bread, so too did the protein content (*p* < 0.05). There was a slightly significant increase in the protein content when 5% MOLP was added to brown bread, and there was a significant increase in brown bread with 10% MOLP (*p* < 0.05). White bread containing MOLP has a higher protein content than brown bread containing

MOLP. Addition of MOLP to a concentration of 10% in white bread and 5% in brown bread would result in a significant increase in the iron contents of the two bread types, respectively.

#### Nutritional Composition of White and Brown Breads Compared to the Estimated Average Requirement

The nutritional composition of the bread was compared to the estimated average requirement (EAR) for protein and iron for vulnerable population groups (children under five years and women of childbearing age). The EAR value is one of the four dietary reference intake values and presents the daily average nutrient intake for particular nutrients for specific gender and age groups [33]. In the current study, an estimated weight for a child aged 1–3 years was 13 kg, 4–5 years was 24 kg, 14–18 years was 55 kg and 19–50 years was 62 kg. A standard size of bread is 30 g. A child aged 1–3 years consumes <sup>1</sup> <sup>2</sup> a slice of bread three times a day, a child 4–6 years consumes approximately one slice of bread three times a day, a female aged 14–18 years consumes two slices of bread three times a day and a female adult aged 19–50 years consumes three slices of bread three times a day. Tables 4 and 5 present the percentage of the EAR for protein and iron, respectively, that would be met from the consumption of estimated bread portions by the respective vulnerable population groups.

#### *3.3. The E*ff*ect of MOLP on the Sensory Acceptability of White and Brown Breads*

Fifty-four students and staff members from the UKZN agricultural campus were recruited to participate in the study. Table 6 indicates the effect of MOLP on the overall acceptability of white and brown breads, as indicated by the percentage distribution of the sensory evaluation scores.

Table 6 shows that as MOLP was increased in either white or brown bread at different substitution levels (5% and 10%), the overall acceptability decreased. However, in terms of the addition of MOLP to bread, there was a less negative effect when MOLP was added to brown bread at different substitution levels in comparison to white bread. Table 7 indicates the effect of MOLP on taste, colour, aroma, texture, appearance and overall acceptability of both white and brown breads.

The Tukey test indicated that there was a significant decrease in taste acceptability when MOLP was added to both white and brown breads, respectively (*p* < 0.05). However, unlike white bread which showed no significant difference in taste acceptability with the addition of either 5% or 10% MOLP, there was a slightly significant decrease in taste acceptability when MOLP was increased from 5% to 10% in brown bread. In terms of taste acceptability, brown bread containing 5% MOLP has significantly higher taste acceptability in comparison to white bread containing 5% MOLP (*p* < 0.05). The colour acceptability significantly decreased when MOLP was added to white bread. There was no significant effect when 5% MOLP was added to brown bread (*p* > 0.05). However, there was a significant decrease in colour acceptability when 10% MOLP was added to brown bread. Brown bread containing 5% MOLP had a significantly higher colour acceptability compared to white bread containing either 5% or 10% MOLP and brown bread containing 10% MOLP.

As MOLP was added to both white and brown breads, respectively, there was a significant decrease in aroma acceptability (*p* < 0.05) and there was a slightly significant decrease when MOLP was increased from 5% to 10% in both white and brown breads. The texture acceptability significantly decreased when MOLP was added to white bread; however, there was no significant difference between white bread containing 5% and 10% MOLPs. The appearance acceptability decreased when MOLP was added to white and brown breads. Additionally, when MOLP was increased to 10%, there was a further significant decrease in appearance acceptability seen for white bread. The overall acceptability significantly decreased when MOLP was increased (*p* < 0.05)—this was especially seen for brown bread. With this being said, brown bread containing 5% MOLP had a significantly higher overall acceptability in comparison to the other MOLP-containing bread.



a *Moringa oleifera* leaf powder;

 Estimated average requirement (EAR) [33].

*Foods* **2020** , *9*, 1910

a

a

b

b

c

d

**Table 3.**

Nutritional composition

 of white and brown breads containing different

concentrations

 of *Moringa oleifera* leaf powder (MOLP) on a dry weight basis (DW)].


*Foods* **2020** , *9*, 1910

**White** 

0

5

10

**Brown bread**

0 (Control)

5

10

 3 (5.6)

 7 (13.0)

 0 (0)

 1 (1.9)

 2 (3.7)

 0 (0)

 2 (3.7)

 4 (7.4)

a number of panellists who gave the score; b

 0 (0)

 1 (1.9)

 2 (3.7)

 0 (0)

 3 (5.6)

 7 (13.0)

 10 (18.5)

 14 (25.9)

percentage of the consumer panel (N = 54) that gave the score.

 3 (5.6)

 9 (16.7)

 7 (13.0)

 9 (16.7)

 10 (18.5)

 13 (24.1)

 10 (18.5)

 16 (29.6)

 8 (14.8)

 3 (5.6)

 16 (29.6)

 9 (16.7)

 3 (5.6)

 25 (46.3)

 15 (27.8)

 37 (68.5)

 7 (13.0)

 51 (94.4)

 0 (0)

 4 (7.4)

 6 (11.1)

 4 (7.4)

 8 (14.8)

 8 (14.8)

 12 (22.2)

 10 (18.5)

 7 (13.0)

 9 (16.7)

 5 (9.3)

 9 (16.7)

 4 (7.4)

 5 (9.3)

 2 (3.7)

 2 (3.7)

 3 (5.6)

 14 (25.9)

 33 (61.1)

 25 (46.3)

 19 (35.2)



#### **4. Discussion**

The bread samples became a darker colour when MOLP was included in the dough; this was particularly prominent in white bread samples (Table 2, Figure 2). This was an expected result as *Moringa oleifera* leaves are naturally a dark green colour due to the high chlorophyll content [22] and are thus responsible for the undesirable colour change. The brown bread prepared in this study became a darker colour but was not as noticeable as white bread, and this could be due to the fact that brown bread is a darker colour to start with due to the chocolate colour of the bran [28], masking the undesirable darker colour seen in white bread at the same concentration of MOLP. These results were consistent with a study conducted by Bourekoua et al. (2018), which found that, as the MOLP concentration increased, the lightness of bread crumb and crust decreased [25]. This dark colour may negatively affect consumer acceptability of bread fortified with MOLP as consumers are more accustomed to bread being a golden-brown colour. With this being said, more individuals are will to try a product that they are familiar with if it was beneficial to their health, thus bread containing MOLP may be acceptable despite being a darker colour. White bread is more commonly consumed than brown bread, therefore another solution to this could be to use a lightening agent to mask the dark colour [23], thus making the bread containing MOLP more acceptable. The use of lightening agents was not investigated in this study. Brown bread containing 5% MOLP had a similar colour to the control thus implies that fortifying with a concentration lower than 5% may result in lighter bread with all the nutritional benefits.

Although MOLP has an undesirable physical attribute, as mentioned earlier, it has many nutritional benefits. Table 3 indicates that the protein concentration increased when MOLP was added to white and brown breads. White bread containing 10% MOLP had the highest protein content (15.5 g/100 g). This was expected as Moringa leaves have a high protein content [34]. The study results were similar to another study which found unfortified bread had the lowest protein content (8.5%) and bread fortified with MOLP contained the highest protein content (13.5%) [24]. Similarly, other authors found that there was a gradual increase in protein content as MOLP was added at different concentration levels [27]. Furthermore, said study had a higher protein content than the current study when 5% MOLP was added to the bread (17.72%).

Table 4 shows results of assessing the potential contribution of MOLP-supplemented bread to the EAR for protein for vulnerable population groups if the amount equivalent to the usual portion size of standard was consumed. It was found that all bread samples containing MOLP would contribute to meeting more than 50% of the EAR for protein for each of the vulnerable groups. Moreover, white bread containing MOLP would contribute more to meeting the EAR for protein for all vulnerable population groups. To fully meet the EAR for protein, a 1–3-year-old would need to consume three slices of the MOLP-containing bread/day. A 4–5-year-old would need to consume 4.5 slices of white bread containing MOLP/day or five slices of brown bread containing MOLP/day. The 14–18 year female group would have to consume 9.5 slices of white bread containing 5% MOLP/day and nine slices of white bread containing 10% MOLP/day and ten slices of brown bread containing MOLP/day. Lastly, the 19–50 year female group would need to consume ten slices of bread containing 5% MOLP/daily and 9.5 slices of white bread containing 10% MOLP. In contrast, 10.5 slices of brown bread containing MOLP would need to be consumed daily. The fact that MOLP increases the protein content is encouraging as animal sources of protein are good but expensive and not affordable to many. The high protein content found in bread containing MOLP could assist in reducing PEM. However, in order not to promote a monotonous diet focused on MOLP bread, MOLP should be incorporated in other commonly consumed food items that are also deficient in protein.

The iron concentration increased in white bread when 10% MOLP was added, and when 5% MOLP was added to brown bread (Table 3). This was an expected result as MOLP generally contains a high iron content. The iron values obtained in the current study agree with the results obtained from previous studies, which showed an increase in iron content as MOLP was added [23,24,27].

Analysis of the percentage contribution of MOLP-containing bread samples to the EAR value for iron for each of the vulnerable population groups is presented in Table 5. It was found that consumption of MOLP-containing bread, up to an amount equivalent of the usual portion size of standard bread, would result in the achievement of more than 100% of the EAR value for iron for each of the vulnerable groups. The iron intake, of each age group, that would result from consumption of MOLP-containing bread, up to an amount equivalent to the usual portion size of standard bread, was also compared with the Tolerable Upper Intake Level (UL). UL refers to the highest amount of a nutrient that should be consumed from food without any adverse effects [33]. The results show that the iron intake of 1–5-year-old children would be way below the UL (40 mg/d) if they consumed MOLP-containing bread up to an amount equivalent to the usual portion size of standard bread. For the 14–18 year and 19–50 year female groups, consumption of all MOLP-containing bread types (except for the brown bread containing 10% MOLP), up to an amount equivalent to the usual portion size of standard bread, would result in an iron intake below the UL value. It is noted that, even for the MOLP-containing bread types that would result in an iron intake above the UL value, it is unlikely to be of health concern because of the limited bioavailability of divalent metal minerals such as iron in plant-based foods. Overall, the results suggest that bread containing MOLP could be a cheaper alternative to improve the iron intake in vulnerable population groups.

Overall, consumer acceptability was higher in the control white and brown breads (0%) compared with their respective bread samples containing MOLP (Tables 6 and 7). This could be lower due to the bitter taste [7] and the fact that it causes the colour of the product to become darker. These results concurred with other studies which found a decrease in overall acceptability with an increase in MOLP [25,27]. There was no significant difference in the instrument texture values of the bread samples. However, the texture acceptability of the bread samples decreased significantly in white bread with the addition of MOLP (*p* < 0.05), whilst there was a marginal difference in the texture acceptability of the control brown bread and bread samples containing MOLP. Thus, it appears that MOLP had a negative effect on the texture of white bread, as perceived by the consumers. MOLP increases the hardness of bread, which is likely due to its high fibre content, but the increase in hardness was perceived in white bread and not in brown bread, probably due to the fact that brown bread already has a high concentration of fibre compared to white bread.

The overall acceptability of brown bread containing 5% MOLP was higher than that of the white bread containing 5% MOLP, indicating that brown bread would be more suitable for fortification using MOLP compared to white bread. For consumers who prefer white bread to brown bread, the darkness imparted to the white bread by MOLP, which was disliked by the consumers, could be resolved by adding a lightening agent together with the MOLP to the white bread dough.

#### **5. Conclusions**

Protein-energy malnutrition and iron deficiency anaemia are public health concerns. A food-based intervention such as fortification of bread with MOLP could improve the protein and iron contents of bread. The results of the current study indicate that bread supplemented with about 5% MOLP could be used to complement existing strategies for addressing malnutrition, especially PEM. However, further research needs to be conducted in order to improve the physical attributes of bread containing MOLP. Further research involving incorporating MOLP in other popular, but nutrient-deficient foods, needs to be conducted to determine the most suitable foods for fortifying with MOLP.

**Author Contributions:** Conceptualization, L.G. and M.S.; methodology, L.G. and M.S.; investigation, L.G. and M.S.; resources, M.S.; data curation, L.G.; writing—original draft preparation, L.G.; writing—review and editing, L.G. and M.S.; visualization, L.G. and M.S.; supervision, L.G. and M.S.; project administration, L.G.; funding acquisition, M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** The Halley Stott Foundation for financial contribution for payment of the statistician.

**Acknowledgments:** The authors would like to thank the volunteers for participating herein. The authors would also like to thank Kaylé Higgs, Masana Makhubele, Nelisiwe Hlope and Raine Booy for assisting with data collection.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **<sup>E</sup>**ff**ect of** *Moringa oleifera* **L. Leaf Powder Addition on the Phenolic Bioaccessibility and on In Vitro Starch Digestibility of Durum Wheat Fresh Pasta**

### **Gabriele Rocchetti 1, Corrado Rizzi 2, Gabriella Pasini 3, Luigi Lucini 1, Gianluca Giuberti 1,\* and Barbara Simonato <sup>2</sup>**


Received: 31 March 2020; Accepted: 13 May 2020; Published: 14 May 2020

**Abstract:** Fresh pasta was formulated by replacing wheat semolina with 0, 5, 10, and 15 g/100 g (*w*/*w*) of *Moringa oleifera* L. leaf powder (MOLP). The samples (i.e., M0, M5, M10, and M15 as a function of the substitution level) were cooked by boiling. The changes in the phenolic bioaccessibility and the in vitro starch digestibility were considered. On the cooked-to-optimum samples, by means of ultra-high-performance liquid chromatography-quadrupole time-of-flight (UHPLC-QTOF) mass spectrometry, 152 polyphenols were putatively annotated with the greatest content recorded for M15 pasta, being 2.19 mg/g dry matter (*p* < 0.05). Multivariate statistics showed that stigmastanol ferulate (VIP score = 1.22) followed by isomeric forms of kaempferol (VIP scores = 1.19) and other phenolic acids (i.e., schottenol/sitosterol ferulate and 24-methylcholestanol ferulate) were the most affected compounds through the in vitro static digestion process. The inclusion of different levels of MOLP in the recipe increased the slowly digestible starch fractions and decreased the rapidly digestible starch fractions and the starch hydrolysis index of the cooked-to-optimum samples. The present results showed that MOLP could be considered a promising ingredient in fresh pasta formulation.

**Keywords:** *Moringa oleifera*; phenolic bioaccessibility; starch digestion; slowly digestible starch; resistant starch

#### **1. Introduction**

Nowadays, one of the most applied strategies to increase the nutritional properties of a certain food and provide consumers with physiological functions is the incorporation of different functional ingredients during formulation [1]. This strategy would be useful to extend health benefits to the maximum number of consumers, contributing to the reduction of nutrient deficiencies, without impairing the eating habits of the population [2]. In this context, durum wheat semolina pasta, a widely consumed product, can be an excellent staple food for the addition of different bioactive compounds [3]. Indeed, pasta formulated with different sources of dietary fiber, proteins, omega-3 fatty acids, and/or bioactive compounds has been produced [4,5]. In this framework, the use of *Moringa oleifera* L. leaf powder (MOLP) in durum wheat semolina pasta formulation could be considered a promising strategy aiming to improve the overall nutritional quality of this food product.

The *Moringa oleifera* L. plant is native to India and is cultivated worldwide for its characteristic nutritional properties and for its variety of end-uses. Every part of the *Moringa oleifera* plant contains important nutrients and phytochemicals, such as vitamins, minerals, essential amino acids, bioactive compounds, and dietary fiber [6]. The leaves of Moringa are considered a valuable source of distinctive classes of polyphenols, including flavonoids, phenolic acids, and lignans [6,7]. Polyphenols have been studied for their potential health-promoting properties, including their antioxidant capacity [8,9]. However, these benefits are not only related to the content of polyphenols in a certain food, but also to their bioaccessibility, bioavailability, and bioefficacy in humans [10,11]. Therefore, MOLP polyphenols' bioaccessibility studies seem to be essential for a first-step investigation on the potential health benefits of this plant ingredient. However, no information is available on the changes in the phenolic profiles following in vitro digestion (i.e., bioaccessibility) for MOLP-enriched cooked fresh pasta. Besides, although the inclusion of MOLP has been reported to substantially improve the nutritional value of cereal-based foods, by increasing both the protein and dietary fiber contents, none of the studies have determined if the incorporation of MOLP could also contribute to modifying the starch digestibility, at least in vitro, in real food systems (i.e., after cooking) [6].

Considering the growing interest in MOLP in food formulation [6], due to its nutrient composition and the bioactive compound profile [7,12], in this work we produced durum wheat semolina fresh pasta with different substitution levels of MOLP, being 0, 5, 10, and 15 g/100 g (*w*/*w*), respectively. The MOLP substitution level up to 15 g/100 g (*w*/*w*) was selected considering that greater levels of MOLP in the recipe could impair the food sensory as well as the technological properties [6].

To better explore the nutritional role of MOLP in fresh pasta production, the present study aimed to evaluate the effect of increasing levels of MOLP in durum wheat semolina fresh pasta by focusing on (i) the phenolic bioaccessibility and (ii) the in vitro starch digestion of cooked pasta.

#### **2. Materials and Methods**

#### *2.1. Materials and Fresh Pasta Sample Preparation*

Durum wheat semolina and dried MOLP were acquired in a local market. As reported on the label, durum wheat semolina's nutritional composition was as follows (g/100 g product): total starch: 70.8 g; total protein: 11.0 g; total fat: 1.8 g; total dietary fiber: 3.0 g. For the dried MOLP (g/100 g product): total starch: 15.1 g; total sugars: 3.1 g; total protein: 29.9 g; total fat: 8.2 g; total dietary fiber: 30.7 g. The MOLP and durum wheat semolina had a particle size smaller than 0.2 mm.

Fresh durum wheat semolina pasta samples with 100% durum wheat semolina (control: M0) and by replacing semolina with 5, 10, and 15 g/100 g MOLP (*w*/*w*), obtaining the M5, M10, and M15 pasta samples were produced, respectively. The dough was made with the addition of 35% *v*/*w* of tap water (37 ◦C) to the pure semolina or the blend semolina–MOLP by using a pasta machine (Mod. Lillodue, Bottene, Italy). The mixing time was 15 min. The resulting dough was extruded through a bronze die for a spaghetti shape (0.22 cm diameter, approximately 25.0 cm length). For each recipe, three pasta production batches were produced on the same day.

#### *2.2. Moisture Content, Water Activity and Pasta Cooking Properties*

The moisture content of the fresh pasta samples was measured with the method 44-15A [13]. Water activity (aw) was measured using a Hygropalm HC2-AW-meter (Rotronic Italia, Milano) at 23 ◦C. The AOAC approved method 66-50 was applied for the optimum cooking time (OCT) determination [13]. In particular, samples were cooked in distilled boiling water (ratio of 1:10, *w*/*v*). At 30 seconds intervals, spaghetti strands were picked from the boiling water and squeezed between 2 glass slides. The OCT for each pasta sample, by definition, is the time for disappearing the white central core of the spaghetti after being squeezed between 2 glass plates.

#### *2.3. Cooking Process and Experimental Details*

Prior to in vitro investigations, the spaghetti (5.0 g) were cooked in boiling water (1:10 *w*/*v*) according to the individual OCT, drained up for 1 min, chopped with a manual meat mincer to

#### *Foods* **2020**, *9*, 628

simulate mastication, and analyzed "as eaten". Three separate in vitro evaluations were conducted, as detailed below.

#### 2.3.1. In Vitro Static Digestion of Cooked Samples for the Evaluation of the Fate of Polyphenols

The protocol involved an oral, a gastric, and an intestinal stage as reported by Minekus et al. [14]. The cooked-to-optimum pasta samples (i.e., 5.0 g) were sequentially hydrolyzed at 37 ◦C through (i) an oral phase, (5 mL of salivary fluid at pH = 7.0 plus human salivary α-amylase (A1031; Sigma-Aldrich; Milan, Italy; 75 U/mL) for 2 min; (ii) a gastric phase (10 mL of a simulated gastric fluid at pH 3.0 plus pepsin (P7012; Sigma-Aldrich; 2.000 U/mL) for 120 min; and (iii) an intestinal phase (20 mL of simulated intestinal fluid at pH = 7.0 plus pancreatin (P7545; Sigma-Aldrich; Milan, Italy; 100 U/mL) and bile salts (B8631; Sigma-Aldrich; Milan, Italy; 10 mM) for a further 120 min. Appropriate amounts of HCl (1 M) and NaOH (1 M) were added for the pH adjustment. Liquid aliquots were carefully removed from each hydrolyzed sample after each hydrolysis phase and stored at −20 ◦C.

#### 2.3.2. Nutritional Starch Fractions Determination

The rapidly digestible starch (RDS) and slowly digestible starch (SDS) were measured with the method of Englyst et al. [15], with minor modifications as detailed by Simonato et al. [5]. The RDS and SDS contents were calculated considering the glucose released after 20 min and 120 min of incubation [15] by measuring the amount of glucose spectrophotometrically using a D-Glucose assay kit (GOPOD, Megazyme, Wicklow, Ireland). The resistant starch (RS) was quantified by a K-RSTAR assay kit (Megazyme, Wicklow, Ireland). The total starch content was calculated as the sum of non-resistant starch and RS following the K-RSTAR assay kit's instructions.

#### 2.3.3. Starch Hydrolysis Index

The cooked-to-optimum spaghetti samples (100 mg) were dispersed in 4 mL of maleic buffer (pH 6), containing an enzyme mixture composed of amyloglucosidase (AMG; 4 μL; 300 U/mL; Megazyme, Wicklow, Ireland) and pancreatic α-amylase (40 mg; 3000 U/mg; Megazyme, Wicklow, Ireland). Samples were incubated in a shaking water bath at 37 ◦C. At selected time intervals (i.e., 0, 30, 60, 120, and 180 min) the reaction was stopped by adding absolute ethanol. Samples were then centrifuged at 2500× *g* for 10 min. The amount of glucose was quantified as previously detailed, after the correction for glucose present in the AMG solution. Values were plotted on a graph vs. time, and the area under the hydrolysis curve (AUHC; 0–180 min) was measured by using the trapezoid rule. A starch hydrolysis index (HI) value was calculated as the AUHC with the product as a percentage of the corresponding area with white wheat bread [16].

#### *2.4. Extraction and Characterization of Untargeted Phenolic Profile by UHPLC-ESI*/*QTOF Mass Spectrometry*

Three replicates (1.0 g) for each cooked-to-optimum pasta batch were extracted in 10 mL of a methanol/water 80:20 (*v*/*v*) solution, by using a homogenizer-assisted extraction with an Ultra-Turrax (Ika T25, Staufen, Germany; 5000× *g*; 3 min) [7]. The extracts were centrifuged (10,000× *g*; 10 min; 4 ◦C), filtered (0.22 μm cellulose syringe filters), and collected [7]. The bound phenolic fraction was extracted from the remaining solid residue [17]. After the alkaline hydrolysis (3 mL of 2 M sodium hydroxide; 1 h; room temperature), the pH was adjusted to 3 with 3 M citric acid and the bound phenolics were extracted with 8 mL of ethyl acetate. After 15 min at 6500 rpm centrifugation, 4 mL of the supernatant was dried under a nitrogen flow at 55 ◦C and the residue was dissolved in 1 mL of 1% formic acid in 80% methanol, vortexed, and centrifuged (10,000× *g* for 10 min). The resulting solution was filtered (0.22 μm cellulose syringe filters) and 200 μL aliquot was transferred to amber vials for analysis.

The modifications in the polyphenol profile after subjecting the cooked samples through the in vitro static digestion method (i.e., Section 2.3.1) were evaluated by ultra-high-performance liquid chromatography-quadrupole time-of-flight (UHPLC-ESI/QTOF) mass spectrometry [7]. Liquid aliquots collected after the oral, the gastric, and the pancreatic in vitro digestion phases were centrifuged at 7000× *g* for 10 min and then filtered (0.22 μm cellulose syringe filters). A mixture of water and acetonitrile (VWR, Milan, Italy; both acidified with 0.1% formic acid) as a mobile phase and an Agilent Zorbax Eclipse-plus C18 column (100 mm × 2.1 mm, 1.8 μm) were used. The gradient was from 6% acetonitrile to 94% acetonitrile in 30 min and the flow rate was 0.220 mL/min. The mass spectrometer worked in the positive scan mode (100–1200 *m*/*z*), injecting 6 μL and source conditions were: sheath gas nitrogen 10 L min−<sup>1</sup> at 350 ◦C; drying gas 10 L min−<sup>1</sup> at 280 ◦C; nebulizer pressure 60 psig, nozzle voltage 300 V, capillary voltage 3.5 kV. Three technical replicates were done for each pasta batch

The software Agilent Profinder B.06 was used to elaborate the raw features. Features were aligned, and the monoisotopic accurate mass was combined with the isotopic profile for the compounds' annotation, thus reaching a level 2 of confidence in annotation (i.e., putatively annotated compounds) [18]. The database Phenol-Explorer 3.6 was used. The mass accuracy tolerance was set to 5 ppm. Phenolic compounds passing the frequency of the detection thresholds (100% of replications within at least one condition) were classified and then quantitative information was produced using calibration curves (in the range 0.05–500 mg/L) from standard solutions of the single phenolic compounds (purchased from Extrasynthese; Genay; France, purity >98%). Selected representative compounds were as follows: cyanidin, quercetin, luteolin, catechin, tyrosol, and ferulic acid. Results were expressed as mg phenolic equivalents/g dry matter (DM). The polyphenols' bioaccessibility was calculated [19]:

Bioaccessibility = (PCA/PCB) × 100

where PCA is the total phenolic subclass content in the samples (mg/g DM) collected after each individual in vitro digestion incubation phase, and PCB is the total phenolic subclass (free plus bound polyphenols) content in the cooked samples before the in vitro digestion.

#### *2.5. Statistical Analysis*

Analyses were run in triplicate on each batch and data were expressed as mean values ± standard deviation. The data were evaluated using a one-way analysis of variance (ANOVA). Differences among means were evaluated by Tukey's HSD tests (*p* < 0.05). The statistical software was R project (version 3.2.3, December 2015). Metabolomic data were pre-processed using the software Agilent Mass Profiler Professional B.12.06 (Agilent Technologies, Santa Clara, CA, USA). Compounds were aligned and filtered by abundance (peak area >5000 counts), normalized at the 75th percentile, and baselined against the median [7]. The metabolomics-based dataset was then exported into SIMCA 13 (Umetrics, Malmo, Sweden) to produce a supervised orthogonal projection to latent structures discriminant analysis (OPLS-DA) model. Hotelling's T2 together with CV-ANOVA (*p* < 0.01) and permutation testing were checked to cross-validate the model. Model parameters (i.e., R2Y and Q2Y) were recorded. The variable selection method variable importance in projection (VIP) was used to point out the phenolic compounds with the highest discrimination ability (VIP score >1) during the in vitro digestion.

#### **3. Results and Discussion**

#### *3.1. Moisture Content, Water Activity and Optimal Cooking Time of Samples*

The moisture content and the aw values of the fresh pasta samples were on average 30.8 g water/100 g of fresh pasta and 0.96, respectively, and were not influenced by the inclusion of MOLP (*p* > 0.05; data not reported). The gradual substitution of semolina flour with MOLP induced a progressive reduction in the OCT, ranging from 5 min for the M0 to 2.5 min for the M15 pasta samples (i.e., 4 min and 3.5 min for M5 and M10, respectively). The progressive decrease in the OCT as a function of the MOLP inclusion level could be due to the great presence of fiber (30.7 g/100 g product) in the MOLP along with the reduction in the total starch content of the samples. For instance, fiber inclusion in wheat pasta can affect the starch–gluten structure, allowing a faster cooking water entrance in the core of the pasta and a resultant faster starch granule gelatinization, thus reducing the OCT [5,20].

#### *3.2. Free and Bound Phenolic Profiles of Cooked-To-Optimum Samples*

On the cooked pasta samples, 152 phenolic compounds were putatively annotated, being 38 flavone equivalents (mainly flavones and flavanones), 30 flavonols, 4 flavan-3-ols, 27 anthocyanins, 36 phenolic acids, and 17 remaining compounds. A comprehensive list of each phenolic compound annotated is provided in File S1, considering both the mass abundances and composite spectra. The most abundant compounds detected were tetramethylscutellarein (a flavone), glycosidic and isomeric forms of quercetin and kaempferol (belonging to flavonols), pyrogallol (a low molecular weight phenolic, characterizing mainly the bound phenolic fraction), and malvidin 3-*O*-galactoside (an anthocyanin).

Thereafter, the main phenolic classes characterizing the different cooked samples were targeted. The results are reported in Figure 1, considering both the free (A) and the bound (B) phenolic contents.

**Figure 1.** Cumulative phenolic composition (as mg phenolic equivalents/g dry matter) in the cooked pasta samples, considering the free (**A**) and bound (**B**) phenolic fractions. M0: 100% durum wheat semolina fresh pasta. M5, M10, and M15 = Fresh pasta produced with 5, 10, and 15 g/100 g *w*/*w Moringa oleifera* L. leaf powder, respectively.

Overall, the greatest (*p* < 0.05) total phenolic content (i.e., sum of the different phenolic classes) was found in the M15 sample, being 2.19 mg/g DM, followed by the M10 (1.58 mg/g DM), M5 (1.24 mg/g) and M0 pasta samples (0.78 mg/g DM), as a function of the increasing inclusion level of MOLP in the formulation. Interestingly, the highest inclusion level of MOLP (i.e., 15% *w*/*w*) was also found to impact the bound phenolic content of the cooked pasta samples (Figure 1B). When considering the specific phenolic composition of the different cooked samples, it was evident that the most represented phenolic classes (in terms of semi-quantitative contents) were low-molecular weight phenolic compounds grouped as tyrosol equivalents (according to the Phenol-Explorer Database), followed by flavonoids

(mainly flavonol and flavone equivalents) and phenolic acids. In addition, the highest increase of polyphenols was observed in the M15 sample when considering the total flavonol equivalents; in fact, this class of compounds moved from 0.19 mg/100 g DM for M0 to 35.73 mg/100 g DM for the M15 sample.

Another phenolic class affecting the phenolic profile of the different cooked pasta samples was tyrosol equivalents. These low-molecular weight phenolics were greater in MOLP-substituted spaghetti when compared to M0. These differences may be related to the inherent phenolic profile of MOLP, along with the specific inclusion level in each recipe. Moringa leaves have been reported as a great source of polyphenols, such as flavonoids [7]. Likewise, an abundance of glycosidic forms of quercetin and kaempferol equivalents (i.e., flavonols), followed by hydroxycinnamic/hydroxybenzoic acids and low-molecular-weight phenolics (i.e., phlorin and protocatechuic aldehyde) has been previously indicated [7]. In addition, previous studies reported that cooking by boiling can cause substantial water-losses and/or oxidative degradation of several antioxidant components [21]. According to the literature, whole cereal grains (such as wheat) are reported to be abundant in bound phenolic compounds, such as phenolic acids (i.e., ferulic acid) and lignans. In fact, these compounds are particularly concentrated in the external bran tissues [22]. However, in this work, we found that the M0-cooked sample was characterized by a lower phenolic content, also when considering the bound phenolic composition (Figure 1). Overall, the trends observed for the M0 sample could be explained by considering the different variables such as (a) the milling process conditions, widely reported as one of the major factors able to affect the phenolic profile of durum wheat semolina [23]; (b) the cooking-by-boiling process used; and (c) the rupture of the plant cell structures as promoted by the extraction method, based on a homogenizer-assisted extraction [7].

#### *3.3. Changes of Phenolic Profiles during In Vitro Static Digestion*

The cooked-to-optimum samples were hydrolyzed through a standardized static digestion method, with the aim to describe the changes in the phenolic profile. In particular, 102 phenolic compounds were found. Flavonoids were the most represented (56 compounds), followed by hydroxycinnamic acids (24 compounds) and tyrosol equivalents. Overall, the phenolic compounds exhibited different bioaccessibility behaviors, mainly imposed by the presence of different food components (e.g., dietary fiber) in each pasta sample, in line with previous findings [8,10,21]. As can be observed in Table 1, lower percentage bioaccessibility values were detected during the entire gastrointestinal process for specific subclasses of compounds, namely anthocyanins, flavanols, and flavonols.

On the other hand, flavones, tyrosols, and phenolic acids had a moderate bioaccessibility during the in vitro digestion process (mainly after 120 min of the pancreatic step). In fact, the higher percentage bioaccessibility values were measured for the tyrosol equivalents in the M0 sample (i.e., 29.21%), followed by flavones characterizing the M5 sample (i.e., 28.26%). However, a linear trend between the polyphenols' bioaccessibility and MOLP increasing levels in the recipe was not observed. The similarities in percentage of bioaccessibility are expected because the polyphenol content was increased but the in vitro digestive conditions were the same.

Recent studies showed that several bioactivities including antioxidant, antiproliferative, immuneregulatory, hormonal regulation abilities, and neuro-/hepato-/cardioprotective effects can be related to the consumption of phenolic-rich foods [21]. However, these health benefits are greatly dependent on the bioaccessibility potential within the human digestive tract. Present findings corroborated the fact that food components–polyphenols interactions should be considered when studying the changes in the bioaccessibility values during the in vitro digestion in a real food system (i.e., cooked matrix) [8]. Another important factor is the impact of the cooking process, that can modify the bioaccessibility of several phenolic compounds [24,25]. Therefore, the relatively high-percentage bioaccessibility values observed in the pancreatic phase for some phenolic classes (such as phenolic acids, flavones, and tyrosol equivalents) are not surprising and could promote an antioxidant environment in the digestive tract [26]. According to the literature [21,27], the detected bioaccessibility trends may be

related to the simulated gastrointestinal digestion conditions used. These latter are not only responsible to break down food matrices and thus release bound phenolic compounds but may also modify the phenolic hydroxyl group (major functional group) of the released phenolics, thus leading to a decrease or increase in the phenolic content in the digestion fluids. In addition, according to the phenolic profile reported for wheat flour, we found a greater bioaccessibility of alkylphenols (quantified as tyrosol equivalents) for the M0 sample during the intestinal step (File S1). In particular, greater percentage bioaccessibility values were measured for three wheat compounds, namely 5-heneicosenylresorcinol, 5-heneicosylresorcinol, and 5-tricosenylresorcinol, which are widely reported as the most abundant in wheat flour [28]. Present findings are difficult to compare with the literature, due to the lack of similar works. To the best of our knowledge, only the work by Caicedo-Lopez and co-authors [12] investigated the changes in the bioaccessibility, intestinal permeability, and antioxidant capacity of the free-phenolic fraction of MOLP after an in vitro gastrointestinal digestion. The authors showed that the greatest content of bioactive compounds was retained in the non-digestible fraction, with higher bioaccessibility values recorded for some phenolics acids, morin, and kaempferol, in line with our findings.

**Table 1.** Bioaccessibility values (expressed as % of phenolic equivalents) for the different phenolic subclasses during the in vitro static digestion of the cooked-to-optimum pasta samples formulated with different substitution levels of *Moringa oleifera* L. leaf powder (MOLP), considering the oral, gastric, and pancreatic phases.


M0: wheat semolina fresh pasta. M5, M10, and M15: fresh pasta produced with 5, 10, and 15 g/100 g *w*/*w* MOLP, respectively. nd = not detected. Within each subclass, means within a column with different superscript letters for the total phenolic content (TPC) of the cooked samples differed at p < 0.05.

Multivariate statistics based on OPLS-DA-supervised modelling was then applied to the metabolomics-based dataset. The OPLS-DA score plot built considering each hydrolyzed cooked sample is reported in Figure 2.

**Figure 2.** Orthogonal projections to latent structures discriminant analysis (OPLS-DA) score plot on the cooked pasta samples' phenolic profile after oral, gastric, and pancreatic phases of in vitro static digestion. M0: wheat semolina fresh pasta. M5, M10, and M15: fresh pasta produced with 5, 10, and 15 g/100 g *w*/*w* MOLP, respectively.

Most of the sample variability characterized the oral and the pancreatic phases of the in vitro static digestion, whilst the gastric phase clustered together. The oral-digested M15 sample clustered differently when compared with the others, likely due to the greater content of anthocyanins and flavonol equivalents (i.e., 0.02 and 0.06 mg/100 g DM, respectively). Interestingly, the OPLS-DA score plot confirmed a characteristic phenolic profile for the M0 spaghetti after the pancreatic phase, likely due to its inherent phytochemical composition when compared with the MOLP-substituted counterparts. The OPLS-DA score plot was also inspected for model accuracy parameters recording the acceptable goodness of fit/prediction values (i.e., R2X = 0.92; R2Y = 0.66; Q2cum = 0.50). Finally, the VIP method was used to rank those phenolic compounds most affected during the in vitro digestion. A list containing these VIP markers is reported in File S1, together with the corresponding VIP score (i.e., their discrimination potential) and standard error. Overall, 41 compounds were detected, including flavonoids (such as glycosidic forms of flavonols and flavones, followed by anthocyanins) and phenolic acids (i.e., hydroxycinnamic acids). The highest VIP scores (i.e., representing those compounds most affected by the in vitro static digestion process) were recorded for stigmastanol ferulate (1.23), isomeric and glycosidic forms of kaempferol (1.19), and other flavones (such as luteolin 7-*O*-rutinoside, apigenin 6,8-di-*C*-glucoside, and chrysoeriol 7-*O*-apiosyl-glucoside). Interestingly, the VIP selection method revealed the presence of several anthocyanins, including acetyl-glycosidic forms of peonidin, malvidin, and petunidin. Regarding the other VIP markers discriminating the in vitro digestion process, we found a good distribution of flavones. In this regard, luteolin (VIP score = 1.16) was clearly related to the presence of MOLP in the recipe, being only detected in the M5, M10, and M15 samples (File S1). Similar findings were obtained for quercetin (presenting a VIP score = 1.08), which was found to be abundant in the MOLP-substituted samples during the pancreatic phase.

#### *3.4. In Vitro Starch Digestion of Cooked Samples*

The nutritional starch fraction contents, along with the HI of the cooked samples, are reported in Table 2.


**Table 2.** Starch fractions (g/100 g dry matter), total starch (g/100 g dry matter), and in vitro hydrolysis index (HI) of the cooked-to-optimum pasta samples formulated with different substitution levels of *Moringa oleifera* L. leaf powder (MOLP). Results are expressed as mean ± standard deviation (*n* = 3).

Values in the same row with different superscripts are significantly different (*p* < 0.05). M0: wheat semolina fresh pasta. M5, M10, and M15: fresh pasta produced with 5, 10, and 15 g/100 g *w*/*w* MOLP, respectively. <sup>1</sup> Calculated using commercial white wheat bread as a reference.

An increase in the SDS and a decrease in the RDS fractions were reported considering the gradual substitution of durum wheat semolina with MOLP. The M10 and M15 cooked pasta samples exhibited the lowest RDS value (i.e., 38.1 and 34.1 g/100 g DM; *p* < 0.05) along with the greatest SDS content (i.e., 18.1 and 20.8 g/100 g DM; *p* < 0.05), when compared with the other samples. From a nutritional standpoint, the RDS fraction was found to be responsible for a rapid increment in the blood glucose levels in humans, while the SDS fraction, being characterized by slow digestion properties, can provide a prolonged release of glucose over time [29]. The mechanism by which the MOLP addition affected the nutritional starch fraction contents of the samples may depend on the interactions among the major constituents (i.e., protein, starch, and fibre polysaccharides) and other minor compounds (i.e., certain classes of dietary polyphenols) [5,8,11,20]. It has been reported that the inclusion of fiber-rich ingredients could modulate the in vitro starch digestion to a certain extent, by changing both the physicochemical properties of the food system [5,30]. For instance, a reduction in the RDS fraction, along with an increment in the SDS fraction exerted by the addition of olive pomace has been reported in wheat-based spaghetti [5]. Lastly, the RS represents, by definition, a certain fraction of starch not degraded in the human small intestine but fermented in the large intestine, with a series of physiological benefits [31]. The RS content of the M0 pasta (i.e., 2.1 g/100 g DM) appeared in line with previous findings for similar food products [32]. However, the RS content of the samples decreased with the increasing inclusion level of MOLP in the recipe, with the lowest value recorded for M15 (i.e., 1.1 g/100 g DM; *p* < 0.05) (Table 2). A possible explanation is that the added MOLP could have contributed to undermine the compact microstructure of wheat pasta, by allowing water and heat to more easily penetrate the pasta during cooking, thus contributing to gelatinize the resistant starch granules present in the core region of the pasta strand to a greater extent [20,33]. In addition, the RS fraction in durum wheat pasta is mainly formed during the pasta extrusion at a high temperature and during the drying process, which in our case was not made [34]. Lastly, the gradual substitution of wheat semolina with MOLP decreased the total starch content of the samples (*p* < 0.05), probably due to a dilution effect of starch exerted by the addition of MOLP, as a consequence of the individual chemical compositions of the selected ingredients.

The starch HI can be used as a predictor of the in vivo glycemic response for a certain starch-based food product [35]. In addition, from the HI values, it is possible to calculate the glycemic index of starch-based foods by applying predictive equations [16]. As reported in Table 2, using white wheat bread as a reference, the HI of the M0 pasta was 47.4. The substitution of a part of the durum wheat semolina with increasing levels of MOLP decreased (*p* < 0.05) the HI of the cooked pasta, the lowest values being recorded for M15 (i.e., 41.8; *p* < 0.05). The decrease in the HI values as a function of the substitution level of MOLP in the recipe could be related to the decrease in the starch content for the replacement of semolina with different quantities of MOLP, as also described in the literature [5], or to the interplay of several factors related to the inherent chemical compositions of the samples. In particular, MOLP contains greater amounts of dietary fiber (about 30.7 g/100 g) and protein

(about 29.9 g/100 g) than durum wheat semolina. This may have contributed to entrap starch granules into a non-starchy network with a limited enzyme accessibility [35,36]. Accordingly, cookies enriched with increasing amounts of alfalfa seed (*Medicago sativa* L.) flour showed a reduction in the in vitro starch digestibility compared with the control [37]. Furthermore, the specific phenolic composition characterizing the MOLP-substituted cooked pasta samples (Figure 1; File S1) could have contributed to modulate, at least in part, the in vitro starch digestion of the samples. In particular, certain classes of phenolic compounds may have a role in modulating the in vitro starch digestion, via the inhibition of the starch digestive enzymes (i.e., α-amylase and α-glucosidase enzymes) and/or through the non-covalent interactions with starch on cooking, thus contributing to the formation of inclusion and non-inclusion starch-complexes with a limited enzyme accessibility [11,38,39]. For instance, secondary metabolites characterizing MOLP-substituted pasta (e.g., flavones, flavonols, and hydroxycinnamic acids; File S1) have already been reported to inhibit both α-glucosidase and α-amylase during in vitro activities [40–42].

#### **4. Conclusions**

Fresh pasta was formulated by replacing durum wheat semolina with 0, 5, 10, and 15 g/100 g *w*/*w* of MOLP. After cooking and following an in vitro digestion process, the phenolic compounds exhibited different bioaccessibility behaviors, with an increase in bioaccessibility observed for flavonols characterizing the digested pasta sample formulated with the greatest inclusion level of MOLP in the recipe. Multivariate statistics highlighted a general abundance of flavonoids and phenolic acids among the discriminant markers. Additionally, the inclusion of MOLP in the pasta influenced the rate of in vitro starch digestion in the cooked samples, showing an increase in the SDS fraction, and a decrease in the RDS fraction and HI values. Taken together, the present findings support the fact that MOLP may represent a valuable ingredient to produce a functional pasta. Future investigations considering technological and sensorial aspects are needed to expand the knowledge on the effect of an MOLP addition in fresh pasta formulation.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2304-8158/9/5/628/s1, File S1: Metabolomics dataset containing polyphenols identified by UHPLC-QTOF mass spectrometry, together with semi-quantitative values for each phenolic class and VIP markers following OPLS-DA multivariate model.

**Author Contributions:** Conceptualization, C.R. and B.S.; methodology, G.R., L.L., G.P. and B.S.; software, L.L. and B.S.; validation, C.R., G.G., and B.S.; formal analysis, G.R. and G.G..; investigation, C.R. and B.S.; resources, G.G., B.S.; data curation, L.L., G.G. and B.S.; writing—original draft preparation, G.R., C.R., G.G., B.S.; writing—review and editing, G.R., C.R., L.L., G.G. and B.S.; supervision, B.S.; project administration, B.S.; funding acquisition, G.G. and B.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Cremona FoodLAB (Fondazione Cariplo e Regione Lombardia, Italy).

**Acknowledgments:** The authors wish to thank the "Enrica e Romeo Invernizzi" Foundation for kindly supporting the metabolomic platform.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **The E**ff**ect of the Addition of** *Apulian black* **Chickpea Flour on the Nutritional and Qualitative Properties of Durum Wheat-Based Bakery Products**

#### **Antonella Pasqualone \*, Davide De Angelis, Giacomo Squeo, Graziana Difonzo, Francesco Caponio and Carmine Summo**

Department of Soil, Plant and Food Science (DISSPA), University of Bari Aldo Moro, Via Amendola, 165/a, I-70126 Bari, Italy; davide.deangelis@uniba.it (D.D.A.); giacomo.squeo@uniba.it (G.S.); graziana.difonzo@uniba.it (G.D.); francesco.caponio@uniba.it (F.C.); carmine.summo@uniba.it (C.S.) **\*** Correspondence: antonella.pasqualone@uniba.it

Received: 2 October 2019; Accepted: 14 October 2019; Published: 16 October 2019

**Abstract:** Historically cultivated in Apulia (Southern Italy), *Apulian black* chickpeas are rich in bioactive compounds such as anthocyanins. This type of chickpea is being replaced by modern cultivars and is at risk of genetic erosion; therefore, it is important to explore its potential for new food applications. The aim of this work was to assess the effect of the addition of *Apulian black* chickpea wholemeal flour on the nutritional and qualitative properties of durum wheat-based bakery products; namely bread, "focaccia" (an Italian traditional bakery product similar to pizza), and pizza crust. Composite meals were prepared by mixing *Apulian black* chickpea wholemeal flour with re-milled semolina at 10:90, 20:80, 30:70, and 40:60. The rheological properties, evaluated by farinograph, alveograph, and rheofermentograph, showed a progressive worsening of the bread-making attitude when increasing amounts of chickpea flour were added. The end-products expanded less during baking, and were harder and darker than the corresponding conventional products, as assessed both instrumentally and by sensory analysis. However, these negative features were balanced by higher contents of fibre, proteins, and bioactive compounds, as well as higher antioxidant activity.

**Keywords:** pulses; re-milled semolina; bread; pizza; focaccia; rheological properties; reofermentograph; bioactive compounds; texture; sensory profile

#### **1. Introduction**

The chickpea (*Cicer arietinum* L.) plays an important role in a healthy and environmentally sustainable diet. In fact, as with other pulses, the chickpea fixes atmospheric nitrogen, thus increasing soil fertility [1]. This species is rich in proteins, and particularly if consumed in combination with cereals to compensate for amino acid deficiencies, can help decrease the dietary intake of meat.

Chickpea is commonly classified in two main types: *kabuli*, characterized by large seeds with beige coats, and *desi*, characterized by small and rough seeds with a black or brown coat [2]. However, there is another type of chickpea, historically cultivated in Apulia (South of Italy), apparently similar to *desi* because of its black coat, but different from the genetic point of view [3]. This *Apulian black* type, which is being replaced by modern cultivars, and is therefore, at risk of genetic erosion [3], has an interesting potential for further commercial development due to its high content of antioxidant compounds, such as anthocyanins and carotenoids [4,5].

Both traditional and new food uses of pulses have been proposed in recent years [6]. Chickpea flour has been used as an ingredient in the production of vegetable beverages [7], extruded snacks [8], canned purée [9], pasta [10], and gluten-free bread [11]. Several attempts have been made to also use

chickpea flour in conventional bread-making, rediscovering an ancient tradition of Albania and Turkey, where chickpea bread was commonly prepared in the past [12]. The addition of 10–30% chickpea wholemeal flour to common wheat flour was proposed, but a negative effect on bread quality in terms of volume, internal structure, texture, and sensory acceptability was observed [13–15]. Blends of different pulses (chickpeas, lentils, and beans, with chickpeas accounting for only 5% of the total flour) was then proposed [16] for sourdough, the latter being helpful to reduce antinutrients such as phytates. The use of dried chickpea sourdough in bread-making was studied, pointing out that 10% was the optimal amount [17]. Together with the addition of 5–30% fermented chickpea flour, the use of 0.5–3% xanthan gum was found to improve the viscoelastic properties of dough [18]. Alternatively, some improvers, such as ascorbic acid and sodium stearoyl-2-lactylate, were helpful when using 25–35% chickpea flour in combination with wheat wholemeal flour [19]. In this frame, however, only one study proposed to enrich durum wheat re-milled semolina with flour of pulses, namely yellow pea, for preparing bread [20], and no research was made to employ chickpea flour in durum wheat bread-making. Moreover, no study considered the use of pigmented chickpeas.

Durum wheat (*Triticum turgidum* var. *durum*) is largely used in the production of pasta and cous cous, and is characterized by tenacious gluten and by the presence of carotenoid pigments. The latter are important from the nutritional point of view due to provitamin A activity, and confer the typical yellowish colour to the end-products, much appreciated by consumers. In the Mediterranean area, part of durum wheat production is used to prepare traditional bread and other bakery products, with a typically dense crumb [21]. In particular, durum wheat re-milled semolina has to be used, which has a particle size similar to that of bread wheat flour (i.e., smaller than that of semolina used for pasta-making) [22], ensuring high hydration rate.

Pizza, originated in Italy, has become a popular food worldwide. The demand of pizza has continued to grow in recent decades, so that food industrial companies have shown growing interest in its production [23]. Pizza is usually prepared with bread wheat flour, but other types of flour have been proposed in the recent years, including durum wheat re-milled semolina [24]. Pizza crust is a convenient food that can be seasoned at home.

"Focaccia" is another Italian traditional bakery product, widely consumed as street food, similar to pizza but containing higher amounts of oil [25]. It may be defined as a leavened greasy flat bread varyingly seasoned, the most typical topping being cherry tomatoes and olives, accompanied by the rosemary and the potato and onion variants. Additionally, in the preparation of focaccia, the use of durum wheat re-milled semolina is quite common [25].

To the best of our knowledge, no papers have considered the enrichment of durum wheat bread, pizza, and focaccia with black chickpea flour. The aim of this work was, therefore, to assess the effect of the addition of *Apulian black* chickpea flour on the nutritional and qualitative properties of durum wheat based bakery products; namely, bread, focaccia, and pizza crust. In particular, a wholemeal flour of chickpeas was used, for accomplishing the current dietary guidelines that highlight the need of increasing fibre intake.

#### **2. Materials and Methods**

#### *2.1. Materials*

*Apulian black* chickpeas (*C. arietinum* L.) were supplied by CerealPuglia s.r.l. (Altamura, Italy). Durum wheat (*T. turgidum* var. *durum*) re-milled semolina was supplied by the milling company Industria Molitoria Mininni s.r.l. (Altamura, Italy). Extra virgin olive oil was supplied by Agridè (Bitonto, Italy). Kastalia stabilized liquid yeast, composed of *Saccharomyces cerevisiae*, salt, and xanthan gum as its stabilizer, and having fermentative power >120 mL CO2 at 20 ◦C and 1 atm, was provided by Lesaffre Italia (Trecasali, Parma, Italy).

#### *2.2. Preparation of the Composite Flours*

*Apulian black* chickpeas were ground at the Food Science laboratory of the University of Bari using a laboratory-scale mill (ETA, Vercella Giuseppe, Mercenasco, Italy) equipped with a sieve of 0.6 mm, to obtain wholemeal flour. Durum wheat re-milled semolina was then used to prepare composite meals containing 10, 20, 30, and 40/100 g of *Apulian black* chickpea wholemeal flour.

#### *2.3. Formulation of the Bakery Products*

Three types of bakery products, namely bread, focaccia, and pizza crust were produced using a composite meal made of re-milled semolina (60/100 g) and *Apulian black* chickpea wholemeal flour (40/100 g). For each product, a control made of pure re-milled semolina was prepared. The formulation, reported in Table 1, was the same for the three types of products, except for the amount of oil, which was not used for preparing bread according to the traditional formulation of Italian durum wheat bread [21]. The quantity of water was added to flour in quantities sufficient to reach a dough consistency of 500 BU, assessed by preliminary farinograph analyses (Brabender instrument, Duisburg, Germany). The preparation of the experimental bakery products was carried out at the bakery laboratory "Buéne" of the Industria Molitoria Mininni s.r.l. (Altamura, Italy).

**Table 1.** Formulation of bread, focaccia, and pizza crust, given for 100 g of flour. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.


<sup>1</sup> This amount is the sum of 10 g in the dough and 10 g used to oil the pans.

#### *2.4. Preparation of Bread*

Flour, water, and yeast were kneaded for 6 min by a spiral mixer (Mecnosud, Flumeri, Italy). Then, salt was added and kneading was continued for other 6 min. The formulation was as in Table 1. The homogeneous dough obtained was left to rise for 1.5 h at 35 ◦C, RH = 20% (Electric oven-leavening combo EKL 1264 TCR, Tecnoeka S.r.l., Borgoricco, Italy), then was divided into 110 g portions manually shaped at 14 cm length, 6 cm width, and 0.5 cm thickness ("ciabatta" bread type), and left to rise again for 30 min at 35 ◦C, RH = 20% (EKL 1264 proofer, Tecnoeka S.r.l., Borgoricco, Italy). Bread was finally baked at 220 ◦C for 20 min in an electric oven (Smeg SI 850 RA-5 oven, Smeg S.p.A., Guastalla, Italy).

#### *2.5. Preparation of Focaccia*

Flour, water, and yeast were kneaded for 6 min by a spiral mixer (Mecnosud, Flumeri, Italy). Then, salt and oil were added and kneading was continued for other 6 min. The formulation was as in Table 1. The homogeneous dough obtained was left to rise for 1 h and 15 min at 35 ◦C, RH = 20% (EKL 1264 proofer, Tecnoeka S.r.l., Borgoricco, Italy), then was divided into 110 g portions of spherical shape, which were manually flattened at 1.5 cm thickness and about 13 cm diameter, put in disposable aluminum pans, previously oiled with other 10 g of extra virgin olive oil, and left to rise again for 1 h and 15 min at 35 ◦C, RH = 20% (EKL 1264 proofer, Tecnoeka S.r.l., Borgoricco, Italy). Focaccia

was finally baked at 220 ◦C for 15 min in an electric oven (Smeg SI 850 RA-5 oven, Smeg S.p.A., Guastalla, Italy).

#### *2.6. Preparation of Pizza Crust*

Flour, water, and yeast were kneaded for 6 min by a spiral mixer (Mecnosud, Flumeri, Italy). Then, salt and oil were added and kneading was continued for other 6 min. The formulation was as in Table 1. The homogeneous dough obtained was divided into 220 g portions of spherical shape which were left to rise for 1 h and 45 min at 35 ◦C, RH = 20% (EKL 1264 proofer, Tecnoeka S.r.l., Borgoricco, Italy). Subsequently, the dough portions were manually flattened to the thickness of 0.5 cm and diameter of about 28 cm and were immediately baked at 380 ◦C for 3 min in an electric oven for pizza, equipped with a refractory cooking stone (G3 pizza oven, Ferrari, Rimini, Italy).

#### *2.7. Chemical Analyses*

Protein (N × 5.7), ash, and moisture contents were determined according to the American Association of Cereal Chemists (AACC) methods 46-11.02, 44–19, and 08–01, respectively [26]. Fat was extracted and determined by Soxhlet apparatus using diethyl ether as solvent. Total dietary fibre was determined by the enzymatic-gravimetric procedure [27]. Carbohydrates were calculated by difference: 100 − (moisture + proteins + lipids + fibre + ash). Energy value (kJ) was calculated by Atwater general conversion factors, by considering the contribution of 8 kJ/g from total dietary fibre also, in accordance with the Annex XIV of the Regulation (EC) number 1169/2011 [28]. Total anthocyanins, total phenolic compounds and antioxidant activity by DPPH method were determined as in [29]. The antioxidant activity by ABTS method was assessed as in [30]. Total carotenoid pigments were determined according to AACC method 14–50.01 [26]. All analyses were carried out in triplicate.

#### *2.8. Determination of the Rheological Properties and Fermentative Attitude of Flours and Composite Meals*

The farinograph indices were determined according to the AACC 54–21 method [26] by a farinograph (Brabender instrument, Duisburg, Germany), equipped with the software Farinograph (Brabender instrument, Duisburg, Germany). Alveograph trials were performed according to the AACC method 54–30A [24] using an alveoconsistograph, equipped with the software Alveolink NG (Tripette et Renaud, Villeneuve-la-Garenne, France). The α-amylase activity was determined by using the Falling Number 1500 apparatus (Perten Instruments AB, Huddinge, Sweden), according to the ISO 3093:2009 method [31]. Rheofermentometer analysis (F3 rheofermentometer, Tripette et Renaud, Chopin Technologies, Villeneuve-la-Garenne, France) was carried out according to the AACC 89-01 method [26] at 28.5 ◦C for 3 h, with a 2000 g weight. All analyses were carried out in triplicate.

#### *2.9. Physical Determinations of Bakery Products*

Texture profile analysis (TPA) was performed on bread and focaccia. Pizza crust was not analyzed due to its very low thickness (0.7–1.0 cm). A Z1.0 TN texture analyzer (Zwick Roell, Ulm, Germany) was used, equipped with a stainless steel square probe (4 cm side) and a 50 N load cell. Data were acquired by means of the TestXPertII version 3.41 software (Zwick Roell, Ulm, Germany). Two centimeter thick slices (3.5 cm × 3.5 cm) were prepared and analyzed. The TPA conditions in the cyclic compression test were: 1 mm/s probe compression rate; 40% sample deformation in both the compressions; and a 5 s pause before second compression. The analyses were carried out in triplicate.

The color indices *L\**, *a\**, and *b\** were measured by using a Chromameter CM-600d (Konica Minolta, Tokyo, Japan). The brown index was calculated as 100 − *L\**. Five replicated analyses were made.

The respective diameter (D), length (L), width (W), and thickness (T) values of bread, focaccia, and pizza crust before and after baking were determined by a caliper and used to calculate the percentage variation due to baking as follows:

% of variation of D (or L, W, T) = [D (or L, W, T) after baking − D (or L, W, T) before baking]/D (or L, W, T) before baking × 100. The analyses were carried out in triplicate.

#### *2.10. Sensory Analysis*

Quantitative descriptive analysis (QDA) of bread, focaccia, and pizza crust respectively, was performed by a sensory panel consisting of 8 trained members, as described in [32]. The sensory panelists (4 males; 4 females; age range 35 to 52) were recruited based on their previous experience in the sensory evaluation of cereal-based foods among technicians and researchers of the laboratory of the Food Science and Technology unit of the Department of Plant, Soil, and Food Sciences of the University of Bari, Italy. All panel members had neither food allergies nor intolerances and were regular consumers of bread, baked goods, and chickpeas. Pre-test sessions were carried out: (i) to define the list of descriptors to be evaluated in the samples object of the study; (ii) to define the intensity range of each descriptor; (iii) to fix the scale anchors of each descriptor; (iv) to verify reliability, consistency, and discriminating ability of panelists when testing bread, focaccia, and pizza crust. The study protocol followed the ethical guidelines of the laboratory. Panelists were given information about study aims, and individually written informed consent was obtained from each participant. All tested samples were food-grade. A total number of 11 sensory descriptors of appearance, smell, texture, and taste were considered. Seven of them, i.e. external color, chickpea odor, crumb elasticity, crumb consistency, crumb moisture, saltiness, and sweetness were evaluated for all three products; namely bread, focaccia, and pizza crust. Another two descriptors, inner color and crumb porosity, were evaluated only in bread and focaccia due to difficulties in separating the crumb from the surface related to the reduced thickness of pizza crust. Greasiness was evaluated only in focaccia, which contained more oil than the other products, whereas the presence of surface bubbles was evaluated only in pizza crust, where it represents a typical feature. The descriptors were rated on an anchored line scale that provided a 0–9 score range (0 = minimum and 9 = maximum intensity). The analyses were carried out in triplicate.

#### *2.11. Statistical Analysis*

Statistical analysis was carried out using XLSTAT software (Addinsoft SARL, New York, NY, USA). Significant differences were determined at *p* < 0.05 by one-way analysis of variance (ANOVA) followed by Tukey's HSD test.

#### **3. Results and Discussion**

#### *3.1. Nutritional Characteristics of the Starting Flours*

*Apulian black* chickpea wholemeal flour showed a significantly higher (*p* < 0.05) content of proteins, lipids, and fibre than re-milled semolina (Table 2). The protein content of chickpea flour was slightly higher than those observed in previous works [4,5], whereas the value observed in re-milled semolina agreed with previous quality surveys [22,33]. The fibre content of chickpea flour was within the range observed in other studies [4,5].

The two types of flour exhibited a similar content of phenolic compounds. However, black chickpea flour was characterized by a significantly (*p* < 0.05) higher content of anthocyanins and carotenoids than re-milled semolina. Consequently, the antioxidant activity was also stronger.

These positive features of *Apulian black* chickpea flour confirmed the results of previous studies [4,5].


**Table 2.** Nutritional characteristics, bioactive compounds, and antioxidant activities of flours used in the production of experimental bread, pizza crust, and focaccia (values are expressed on fresh weight bases).

<sup>1</sup> n.d. = not detected. Different letters in a row indicate significant differences (*p* < 0.05).

#### *3.2. Rheological and Fermentative Characteristics of Flours and Composite Meals*

Composite meals were prepared by mixing *Apulian black* chickpea wholemeal flour with re-milled semolina at 10:90, 20:80, 30:70, and 40:60. The rheological properties of the obtained blends were then evaluated by farinograph, alveograph, and rheofermentograph, in order to assess the suitability to the production of fermented bakery products.

The farinograph parameters measured in pure re-milled semolina agreed with previous works [22]. The addition of black chickpea flour significantly (*p* < 0.05) influenced all farinograph parameters (Table 3). In particular, water absorption, dough development time, and loss of consistency progressively increased, whereas the dough stability decreased. The increase of water absorption was due to the presence of fibre in chickpea wholemeal flour, able to absorb high amounts of water. The dilution of gluten, chickpea flour being gluten free, and the presence of fibre able to interfer with a gluten network, were responsible for the decrease of dough stability, the increase of time needed to develop gluten, and the increase of consistency loss. These results highlight that the addition of chickpea flour worsened the bread-making attitude of wheat flour, with a more evident effect at higher doses. Similar negative effects on farinograph were reported when re-milled semolina was added of other fibre-rich ingredients, such as powdered almond skins [34], or yellow pea wholemeal flour [20]. Moreover, the same effects have been reported for blends of chickpea wholemeal flour with bread wheat flour [14,15] or with semolina for pasta-making [10], evidencing that the bread-making attitude of any gluten-containing flour is depressed by the addition of chickpea flour.

As for the alveograph strength (W), it progressively decreased by the addition of increasing amounts of chickpea flour (Table 3). Moreover, that result was imputable to the increasing content of fibre, contributed by chickpea wholemeal flour, and to gluten's dilution. The values of the alveograph tenacity/extensibility ratio (P/L) ratio, instead, remained almost constant after the addition of chickpea flour. The value of P/L observed for pure re-milled semolina was in the range observed in previous works, whereas W was particularly high, indicating a very good quality level [22].

The fermentative attitude of flour blends was evaluated by measuring the falling number, related to the amylase activity, and by performing the rheofermentograph analysis. The values of falling number showed a significant difference (*p* < 0.05) only when comparing pure re-milled semolina (with the highest amylase activity) to the composite flour containing the highest chickpea amount (with the lowest amylase activity) (Table 4). The decrease of amylase activity with the addition of 40/100 g of chickpea flour reflected the presence of α-amylase inhibitors, reported in chickpeas and in several other pulses [35]. These inhibitors are slightly more active in *desi* than in *kabuli* chickpea cultivars [36], but can be inactivated by heat treatment [35].


**Table 3.** Farinograph and alveograph data of pure durum wheat re-milled semolina and of blends with increasing amounts of *Apulian black* chickpea wholemeal flour.

<sup>1</sup> B.U. = Brabender units. Different letters in a row indicate significant differences (*p* < 0.05).

**Table 4.** The fermentative attitude of pure durum wheat re-milled semolina and of blends with increasing amounts of *Apulian black* chickpea wholemeal flour.


Different letters in a row indicate significant differences (*p* < 0.05).

Rheofermentograph data of pure re-milled semolina agreed with other works [37]. Increasing amounts of chickpea flour significantly reduced (*p* < 0.05) the amount of gas produced (*VT*) during the rheofermentograph analysis, in agreement with the decrease of amylase activity. Moreover, a greater amount of gas was lost by the dough (*VL*) when composite meals were analyzed. Consequently, a progressively lower quantity of gas was retained (*VR*) as the addition of chickpea flour increased, reflecting in a significantly lower coefficient of gas retention. Besides, the loss of gas appeared sooner (*Tx*) in chickpea-added dough than in case of pure re-milled semolina.

The volumetric increase of leavened bakery products depends on both the amount of CO2 developed and the rheological properties of dough, in terms of quality and strength of the gluten network, which allows one to effectively retain the gas developed during fermentation. Therefore, these results were due to the diminished fermentative attitude of chickpea-added dough, as shown by the lower amylase activity, and to its weaker gluten network, evidenced by the alveograph and farinograph analyses.

As for the dough development curve, its maximum height (*Hm*), as well as the height after 3h(*h*), decreased significantly with the increase of chickpea wholemeal flour added, indicating a lower inflation of the dough.

Overall, the bread-making attitude of re-milled semolina worsened by increasing the level of enrichment with black chickpea wholemeal flour. However, studies showed that consumer behavior is changing, driven by awareness of the relationship between nutrition and health. Consequently, whole and fibre-enriched bakery products are accepted better than in the past, despite the lower quality characteristics, particularly if information on the high fibre content is shown on the label [38]. Therefore, the search for the best rheological parameters should no longer be the exclusive criterion for selecting the optimal level of enrichment.

By calculating the theoretical fibre content of the end-products, it was found that the addition of 40/100 g black chickpea wholemeal flour would make it possible to claim the "fibre source" on the label, which requires a content of at least 3 g of fibre per 100 g of product [39]. Lower levels of addition would not reach the conditions for the inclusion of this statement in the label. Considering that information about the presence of fibre influences the acceptance of the modern consumer, leading to a positive perception of unconventional bakery products, the latter were prepared using composite flours of durum wheat re-milled semolina added with 40/100 g of black chickpea wholemeal flour.

#### *3.3. Nutritional and Qualitative Properties of the Bakery Products*

Table 5 reports the nutritional features of the experimental bakery products (analytically determined data, not calculated). Bread, focaccia, and pizza crust enriched with black chickpea flour showed significantly (*p* < 0.05) higher contents of proteins, lipids, ash, and fibre than conventional products, whereas carbohydrates were significantly lower (Table 5). The slight increase in lipids was coupled to higher fibre content; therefore, the energy value of the enriched products was similar to that of the conventional products.

**Table 5.** Nutritional features (values per 100 g, expressed on fresh weight basis) of conventional durum wheat bread, focaccia, and pizza crust and of their black chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of Apulian black chickpea wholemeal flour.


Different letters in a row, for the same bakery product, indicate significant differences (*p* < 0.05).

The addition of black chickpea flour determined a significant increase (*p* < 0.05) of the content of anthocyanins and phenolic compounds in all three baked goods considered (Table 6). However, a marked reduction was observed with respect to the starting flour (reported in Table 2). The decrease of bioactives during thermal processing of *Apulian black* chickpeas was also reported during the preparation of canned sterilized puré [9]. The antioxidant activity, particularly when measured with the ABTS method, was significantly higher in chickpea-enriched bakery products than in durum wheat ones, reflecting the contents of bioactive compounds, specifically anthocyanins and carotenoids.

The variations in the dimensional parameters of bakery products induced by cooking, and their weight loss, are shown in Table 7. Usually, baking involves an increase in volume determined by the thermal expansion of the gases (air trapped during mixing and kneading, water vapor evaporated by the dough, and carbon dioxide originated by leavening). An increase in thickness, indeed, was observed, but without significant differences between conventional durum wheat and chickpea-enriched products. On the other hand, the diameter of circular products, i.e. focaccia and pizza crust, as well as length and width of bread, which had a rectangular shape, decreased with baking. The decrease was significantly more marked (*p* < 0.05) in chickpea-enriched products than in conventional durum wheat products. In bread, the conventional product had an opposite behavior showing an increase of width and maintaining its length almost constant.

**Table 6.** The bioactive compounds and antioxidant activity of conventional durum wheat bread, focaccia, and pizza crust and of their black, chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.


<sup>1</sup> Expressed as mg/kg cyanidin 3-*O*-glucoside (d.m.); <sup>2</sup> expressed as mg/kg β-carotene (d.m.); <sup>3</sup> expressed as mg/g ferulic acid (d.m.); <sup>4</sup> expressed as μmol/g Trolox equivalent (d.m.); <sup>5</sup> n.d. = not detected. Different letters in a row, for the same bakery product, indicate significant differences (*p* < 0.05).

Overall, these results agreed with the predictive analyses on dough rheology which, in turn, were due to the interference by fibre and the dilution of gluten. The high alveograph P/L ratio observed in the chickpea-added dough, in particolar, explains the limited expansion during both leavening and the first phases of baking (oven-spring). Thickness was less affected by this limitation because baking essentially induces an upward push [40].

Weight loss though, was not significantly influenced.

**Table 7.** Baking-induced variations of the dimensional parameters of conventional durum wheat bread, focaccia, and pizza crust and of their black, chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.


Different letters in a row, for the same bakery product, indicate significant differences (*p* < 0.05).

Table 8 shows the colorimetric indices of the external and internal surface of the various bakery products prepared. Pizza was inspected only externally, due to its limited thickness (0.7–1.0 cm). The addition of black chickpea flour caused a significant (*p* < 0.05) decrease of *b\** and an increase of brown index (100 – *L\**) of all the products, which appeared grayish.

Statistically significant differences of *a\** were observed only in the internal part of bread and focaccia, being higher in the chickpea-added products. All products from pure re-milled semolina, instead, were bright yellow, with values of *b\** almost double those of chickpea-added products. Therefore, together with the reduced expansion degree of products during baking, color alteration is another negative effect of adding black chickpea wholemeal flour, which would require an adequate explanation to the final consumer to highlight the nutritional reasons behind it, in order not to appear too unconventional or even unpleasant (Figure 1).

**Figure 1.** The internal structure of bread prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour (**left**) and bread prepared by using only durum wheat re-milled semolina (**right**).

**Table 8.** Color parameters of conventional durum wheat bread, focaccia, and pizza crust and of their black, chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.


<sup>1</sup> External color. <sup>2</sup> Internal color. Different letters in a column for the same bakery product and portion inspected, indicate significant differences (*p* < 0.05).

Table 9 shows the textural parameters of experimental bread and focaccia. Again, pizza crust was not analyzed due to its very limited thickness (0.7–1.0 cm). The chickpea-added bread and focaccia were significantly harder (*p* < 0.05) and more chewy than their counterparts made only of re-milled semolina. The springiness was very similar in all products, whereas the cohesiveness of chickpea-added bread was lower than in conventional durum wheat bread. These results, in agreement with the reduced expansion in volume, can be explained by a worse gluten formation, as already indicated by the alveograph and farinograph parameters of the starting meals, due to the richness in fibre and absence of gluten in chickpea flour.

As for the sensory profile (Table 10), all the chickpea-enriched products were significantly more consistent (*p* < 0.05) than the corresponding conventional products. The addition of chickpea flour determined the significant emergence of an odorous note of chickpeas, a darker color (both external and internal, in agreement with colorimetric data), and an increase in moisture. Saltiness and sweetness, on the other hand, were similar in all the products, as well as crumb porosity.


**Table 9.** Textural parameters of conventional durum wheat bread and focaccia and of their black, chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.

Different letters in a row indicate significant differences (*p* < 0.05).

Greasiness, typical of focaccia but unpleasant if eccessive, was significantly more evident in the conventional focaccia than in the chickpea-added one. The inclusion of oil in the formulation, more abundant in focaccia than in pizza crust and bread, probably influenced the consistency of the focaccia as well, which was slightly softer than bread, in agreement with the results of textural analysis. Other studies reported the positive effect of oil on the consistencies and volumes of bakery products [41].

As for the pizza crust, the presence of surface bubbles was significantly greater in the conventional product than in the chickpea-added one, whose dough was less extensible and less able to retain gas, as shown by rheological analyses.

**Table 10.** The sensory profile of conventional durum wheat bread, focaccia, and pizza crust, and of their black chickpea-enriched versions. DW = product prepared by using durum wheat re-milled semolina; BC = product prepared by using a composite meal containing 60/100 g of durum wheat re-milled semolina and 40/100 g of *Apulian black* chickpea wholemeal flour.


<sup>1</sup> Evaluated on the whole product. Different letters in a row, for the same bakery product, indicate significant differences (*p* < 0.05).

#### **4. Conclusions**

On the basis of the results obtained during the characterization of the composite meals, it emerged that the addition of the wholemeal flour of *Apulian black* chickpeas to durum wheat re-milled semolina caused a decrease in the bread-making attitude, which, however, was countered by a nutritional improvement in terms of higher contents of fibre and proteins. The enriched end-products showed also higher contents of bioactive compounds and an improved antioxidant activity. The positive features should be adequately communicated to the consumer to compensate the significant negative effects of addition of chickpea flour, such as alterations of consistency and color with respect to analogous baked goods made of pure re-milled semolina.

Among the three products evaluated, the best product for consumer would be bread because of its lower content of lipids, and consequently, lower energy value. However, with enrichment levels of chickpea flour as high as 40/100 g, all three products evaluated were able to help fulfil the recent dietary guidelines, which suggest one to consume at least three legume servings per week. Adding chickpea flour to baked goods, therefore, represents a nutritionally effective strategy and a significant step forward to increase the consumption of legumes.

**Author Contributions:** Conceptualization, A.P. and C.S.; formal analysis, A.P., D.D.A., F.C., and C.S.; funding acquisition, A.P. and C.S.; methodology, D.D.A., G.S., and G.D.; project administration, A.P., F.C., and C.S.; software, D.D.A., G.S. and G.D.; writing—original draft, A.P. and C.S.; writing—review and editing, A.P., D.D.A., G.S., G.D., F.C., and C.S.

**Funding:** This work was funded by: (1) Agropolis Fondation, Fondazione Cariplo, and Daniel and the Nina Carasso Foundation through the "Investissements d'avenir" programme with reference number ANR-10-LABX-0001-01, under the "Thought for Food" Initiative (project "LEGERETE"); (2) the Italian Ministry of Education, University and Research (MIUR), program PRIN 2017 (grant number 2017SFTX3Y) "The Neapolitan pizza: processing, distribution, innovation and environmental aspects."

**Acknowledgments:** The Authors acknowledge Industria Molitoria Mininni (Altamura, Italy), particularly Giuseppe Galetta, Vincenzo Tricarico, and Isabella Mininni, for preparing accurately and providing flours, composite meals, and bakery products.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Use of Fermented Hemp, Chickpea and Milling By-Products to Improve the Nutritional Value of Semolina Pasta**

#### **Rosa Schettino, Erica Pontonio \* and Carlo Giuseppe Rizzello**

Department of Soil, Plant and Food Science, University of Bari Aldo Moro, 70126 Bari, Italy; rosa.schettino@libero.it (R.S.); carlogiuseppe.rizzello@uniba.it (C.G.R.) **\*** Correspondence: erica.pontonio@uniba.it

Received: 8 October 2019; Accepted: 20 November 2019; Published: 22 November 2019

**Abstract:** A biotechnological approach including enzymatic treatment (protease and xylanase) and lactic acid bacteria fermentation has been evaluated to enhance the nutritional value of semolina pasta enriched with hemp, chickpea and milling by-products. The intense (up to circa, (ca.) 70%) decrease in the peptide profile area and (up to two-fold) increase in total free amino acids, compared to the untreated raw materials, highlighted the potential of lactic acid bacteria to positively affect their in vitro protein digestibility. Fermented and unfermented ingredients have been characterized and used to fortify pasta made under pilot-plant scale. Due to the high contents of protein (ca. 13%) and fiber (ca. 6%) and according to the Regulation of the European Community (EC) No. 1924/2006 fortified pasta can be labelled as a "source of fiber" and a "source of protein". The use of non-wheat flours increased the content of anti-nutritional factors as compared to the control pasta. Nevertheless, fermentation with lactic acid bacteria led to significant decreases in condensed tannins (ca. 50%), phytic acid and raffinose (ca. ten-fold) contents as compared to the unfermented pasta. Moreover, total free amino acids and in vitro protein digestibility values were 60% and 70%, respectively, higher than pasta made only with semolina. Sensory analysis highlighted a strong effect of the fortification on the sensory profile of pasta.

**Keywords:** hemp; chickpea; milling by-products; fortified pasta; lactic acid bacteria; nutritional value

#### **1. Introduction**

The trend in world population growth (up to 9.6 billion people by 2050) and the necessity to provide a nutritionally balanced diet and to reduce greenhouse gas emissions require relevant production increases in vegetables, as well as a transition to a diet higher in plant- rather than animal-derived proteins [1,2]. Aiming at addressing environmental concerns and meeting nutritional deficiencies and recommendations, the fortification of staple foods (e.g., bread and pasta) has been identified as an effective, sustainable and promising intervention [3,4]. To date, several studies investigated the nutritional value of additional ingredients to be used as wheat alternatives in cereal-based products [4–8]. Due to its popularity [9], pasta has been proposed as a suitable carrier of nutrients, mainly dietary fiber and proteins [4,7,10–13].

Legumes are excellent sources of proteins with high biological value and dietary fibers and they supply high levels of vitamins, minerals, oligosaccharides and phenolic compounds [14]. Moreover, thanks to their functionality (e.g., solubility and water-binding capacity), legume flours have successfully been proposed to enhance gluten-free food formulation and processing [5].

Grain germ and bran (milling by-products) comprise important sources of dietary fiber and α-tocopherol, vitamins of group B, polyunsaturated fats, minerals and different bioactive compounds with health-promoting effects [15]. Pearling by-products, thanks to the high content of dietary fiber and β-glucan, have been suggested as suitable ingredients to produce functional pasta [6]. Hemp has recently raised much interest as a sustainable food ingredient due to the high content (ca. 30%) and biological value of protein, dietary fiber content (ca. 50%) as well as the considerable content of functional compounds (e.g., phenols) with antioxidant and anti-hypertensive properties [16].

Nevertheless, the poor stability to oxidation [17], the high content of fibers and the absence of gluten may impair the wheat alternative's high nutritional value, worsening the technological and sensory profiles of the products [18]. Moreover, the presence of anti-nutritional factors (ANFs), i.e., phytic acid, condensed tannins, raffinose and trypsin inhibitors, further limit the use of such ingredients by the food industry [19]. Although different biotechnological options were suggested to overcome the drawbacks related to the use of non-wheat flours in cereal-derived foods, fermentation with lactic acid bacteria (LAB) seemed to be the best option to both decrease the ANF and improve their nutritional, technological and sensory profile [19,20].

Based on the above consideration, here, hemp and chickpea flours, and wheat milling by-products were proposed as additional ingredients to improve the nutritional quality of semolina pasta. Enzymatic pre-treatments and fermentation with LAB were evaluated as bioprocessing options to enhance protein digestibility and decrease ANF. Bioprocessed ingredients were used to manufacture fortified pasta, and the effects on the nutritional, technological, and sensory properties were investigated in comparison to samples produced with the untreated native ingredients.

#### **2. Materials and Methods**

#### *2.1. Raw Materials, Bacterial Strains and Enzymes*

Commercial hemp (*Cannabis sativa L*.) flour (Sottolestelle, San Giovanni Rotondo, Italy), chickpea grains (*Cicer arietinum L.* var. Pascià, Caporal Grani S.a.s.,Gravina di Puglia, Italy), wheat germ (Molino Rieper, Vandoies di Sotto, Italy) and bran (Molino Careccia, Stigliano, Italy) were used in this study. Chickpea, germ and bran flours were milled using a laboratory mill Ika-Werke M20 (GMBH, and Co. KG, Staufen, Germany) before use. After milling, all the flours were sieved (mesh size 500 μm) to remove the coarse fraction.

*Lactobacillus plantarum* LB1 and *Lactobacillus rossiae* LB5 [21] were used in this study. Strains were routinely cultivated on modified de Man Rogosa and Sharp (Oxoid, Basingstoke, Hampshire, UK) (mMRS) as reported by Rizzello et al. [21]. A commercial xylanase (880 *xylanase* u/g; Depol 761P, Biocatalysts Limited, Chicago, USA) from *Bacillus subtilis* and proteases of *Aspergillus oryzae* (500,000 hemoglobin units on the tyrosine basis/g; enzyme 1 [E1]) and *Aspergillus niger* (3000 spectrophotometric acid protease units/g; enzyme 2 [E2]), routinely used for bakery applications, supplied by BIO-CAT Inc. (Troy, VA, USA), were also used.

#### *2.2. Proximate Chemical Composition of Raw Materials*

Moisture, protein, lipids, total dietary fiber and ash of raw material were determined according to Approved Methods 44-15A, 46-11A, 30-10.01, 32-05.01, and 08-01.01 of the American Association of Cereal Chemists (AACC) [22]. Total nitrogen was corrected by 6.25, 4.99, 6.31 and 5.30 for chickpea, wheat germ, wheat bran and hemp, respectively [23,24]. Available carbohydrates were calculated as the difference [100 − (proteins + lipids + ash + total dietary fiber)]. Proteins, lipids, carbohydrates, total dietary fiber and ash were expressed as % of dry matter (d.m.).

#### *2.3. Bioprocessing*

Wheat germ and bran were mixed (1:4, *wt*/*wt*) prior to dough preparation. Doughs (50 g) were prepared by mixing hemp flour (62.5% *wt*/*wt*), or chickpea flour (62.5% *wt*/*wt*) or milling by-products (40.0% *w*/*w*) with tap water. Dough yield (DY, dough weight × 100/flour weight) was 160 for hemp (H) and chickpea (C), and 250 for milling by-products (WGB). To be used as a mixed starter for sourdough

fermentation, LAB cells were harvested by centrifugation (10,000× *g*, 10 min, 4 ◦C), washed twice in 50 mM phosphate buffer, pH 7.0, and re-suspended in tap water used for the dough making (final cell density in the dough was ca. 7.0 log10 cfu/g). Fermentation was carried out at 30 ◦C for 24 h (HF, CF, and WGBF).

For the enzymatic treatments, doughs with the same DY were prepared. Before mixing, xylanase was added at 50 ppm based on dough weight (HX, CX, and WGBX) and proteases E1 and E2 (HP, CP, and WGBP) were used at 50 ppm (25 ppm for each enzyme) on dough weight. Doughs were incubated at 30 ◦C for 8 h. All the doughs were prepared and incubated in triplicate.

#### *2.4. Determination of the Protein Degradation*

Aiming at selecting the optimal bioprocessing option, total free amino acid (TFAA) concentration and peptide profiles were considered as indexes of the proteolytic degradation and screening parameters. Doughs prior bioprocessing were used as the controls (H, C, and WGB).

Water/salt-soluble extracts (WSEs) from doughs were prepared as reported by Weiss et al. [25] and used for TFAA and peptides analyses. TFAA were analyzed by a Biochrom 30 series Amino Acid Analyzer (Biochrom Ltd., Cambridge Science Park, United Kingdom) with a Na-cation-exchange column (20 by 0.46 cm internal diameter), as described by Rizzello et al. [17]. For the analysis of peptides, WSE were treated with trifluoroacetic acid (TFA, 0.05% *wt*/*v*) and subjected to dialysis (cut-off 500 Da) to remove proteins and FAA, respectively. Then, the peptide concentration was determined by the o-phtaldialdehyde (OPA) method [26]. Peptide profiles were obtained by Reversed-Phase Fast Performance Liquid Chromatography (RP-FPLC) [26,27], using an ÄKTA FPLC equipped with a Resource RPC column and a UV detector (214 nm) (GE Healthcare Bio-Sciences AB, Uppsala, Sweden). Peptide profiles and total peak area were elaborated with Unicorn 4.0 software (GE Life Sciences).

#### *2.5. Microbiological and Biochemical Characterization of Fermented Doughs*

pH values of H, C and WGB and HF, CF and WGBF were determined online by a pH meter (Model 507, Crison, Milan, Italy) with a food penetration probe. The AACC method 02-31.01 was used for the determination of total titratable acidity (TTA) of samples. Presumptive LAB was enumerated using de Man, Rogosa and Sharpe (MRS, Oxoid) agar medium, supplemented with cycloheximide (0.1 g/L). Plates were incubated, under anaerobiosis (AnaeroGen and AnaeroJar, Oxoid), at 30 ◦C for 48 h. WSEs from unfermented and fermented doughs were used for the determination of organic acids, peptides and TFAA concentrations. High-Performance Liquid Chromatography (HPLC) ÄKTA Purifier system (GE Healthcare, Buckinghamshire, United Kingdom) equipped with an Aminex HPX-87H column (ion exclusion, Bio-Rad, Richmond, CA, United States) and a UV detector operating at 210 nm [17] was used to quantify organic acids. The fermentation quotient (FQ) was determined as the molar ratio between lactic and acetic acids. Peptides and TFAA concentrations were determined as described above.

#### *2.6. Pasta Making*

Pasta was manufactured using a pilot plant La Parmigiana SG30 (Manufacture, Fidenza, Italy). Table 1 summarizes the ingredients and protocol used for pasta making. All the doughs had a final DY of 130, corresponding to 23% (*wt*/*wt*) water and 77% (*wt*/*wt*) dry matter (semolina, hemp, chickpea flours and milling by-products).

Unfermented and fermented doughs were obtained as described before and used as ingredients for pasta making. Due to the difference in terms of the DY (160 vs. 250) of the doughs, H, HF, C and CF were used at 11% (*wt*/*wt*), while the level of fortification of WGB and WGBF was 17% (*wt*/*wt*). Unfermented and fermented doughs were mixed with durum wheat semolina and water to obtain pasta samples (*p*H, *p*HF, *p*C and *p*CF and *p*WGB and *p*WGBF, respectively). A control pasta was made without fortification (*p*CT). The process is composed of four stages: i) three-steps mixing (1 min mixing/6 min hydration); ii) extrusion of final dough at 45–50 ◦C through a n. 76 (150 mm diameter) bronze die; iii) cutting the extruded to obtain grooved "macaroni"; and iv) drying using low temperature (55 ◦C)

cycle (Supplementary Table S1). The proximate composition of wheat semolina was moisture, 10.2%; protein, 12.1% of d.m.; fat, 1.8% of d.m.; ash, 0.6% of. d.m.; and total carbohydrates, 75.5% of d.m.

**Table 1.** Formulas for pasta fortified with hemp, chickpea, and wheat germ/bran unfermented and fermented doughs: *p*H, pasta containing 11% of unfermented hemp dough (*wt*/*wt*); *p*HF, pasta containing 11% of fermented hemp dough (*wt*/*wt*); *p*C, pasta containing 11% of unfermented chickpea dough; *p*CF (*wt*/*wt*), pasta containing 11% of fermented chickpea dough (*wt*/*wt*); *p*WGB, pasta containing 17% of unfermented wheat germ and bran (1:4) dough (*wt*/*wt*); *p*WGBF, pasta containing 17% of fermented wheat germ and bran (1:4) dough (*wt*/*wt*); *p*CT, pasta made with durum wheat semolina.


<sup>1</sup> Having DY of 160, 11% of hemp (H and HF) and chickpea (C and CF) doughs contained ca. 6.8% of solids and 4.2% of water. While, 17% of wheat germ/bran doughs (WGB and WGBF) having dough yield (DY) of 250 contained ca. 6.8% dry matter and 10.2% of water. Consequently, all the fortified pasta samples contained ca. 7% of non-wheat flours/milling by-products.

#### *2.7. In vitro Protein Digestibility (IVPD)*

The IVPD of pasta samples was determined according to Akeson and Stahmann [28,29]. In order to mimic the in vivo digestion in the gastrointestinal tract, pasta samples were subjected to a sequential enzymatic treatment. The IVPD is the percentage of the total protein solubilized after enzymatic hydrolysis. The protein quantification was made according to the Bradford method [30].

#### *2.8. Pasta Characterization*

#### 2.8.1. Hydration Test, Cooking Time, Cooking Loss and Water Absorption

The method of Marti et al. [31] (ratio pasta:water of 1:20, 180 min of incubation) was used to determine the hydration at 25 ◦C, while the method of Schoenlecher et al. [32] was used to determine the cooking time. The optimal cooking time (OCT) corresponded to the disappearance of the white core. Cooking loss (expressed as grams of matter loss/100 g of pasta) was evaluated by determining the number of solids lost into the cooking water as proposed by D'egidio, et al. [33]. The increase in pasta weight during cooking (water absorption) was evaluated by weighing pasta before and after cooking. The results were expressed as [(W1 − W0/W0] × 100, where W1 is the weight of cooked pasta and W0 is the weight of the uncooked samples.

#### 2.8.2. Chemical and Nutritional Profile

Chemical characteristics (determined on pasta dough prior to extrusion) and the proximal composition of pasta were determined as reported above.

The protein solubility of pasta (grinded) was evaluated under native and denaturing conditions as reported by Iametti et al. [34]. The concentration of protein and peptides was determined as reported above [7,30].

Raffinose and phytic acid concentrations were determined by using the Megazyme kit Raffinose/D-Galactose Assay Kit K-RAFGA and K-PHYT 05/07 (Megazyme International Ireland Limited, Bray, Ireland), respectively, following the manufacturer's instructions. Condensed tannins were determined using the acid butanol assay, as described by Hagerman [35].

IVDP and starch hydrolysis were determined on pasta samples at the OCT. IVPD was determined as described before. The evaluation of the starch hydrolysis rate was performed using a procedure mimicking the in vivo digestion of starch [36]. Wheat flour bread (WB) was used as the control to estimate the hydrolysis index (HI = 100). The predicted GI of all pasta samples was calculated using the following equation: *p*GI = 0.549 × HI + 39.71 [37].

#### 2.8.3. Texture and Color Analysis

Instrumental Texture Profile Analysis (TPA) was carried out with a TVT-300XP Texture Analyzer (TexVol Instruments, Viken, Sweden), equipped with a cylinder probe (diameter 95 mm). For the analysis, pasta samples were cooked until the OCT, left to cool at room temperature and placed in a beaker (diameter, 100 mm; height 90 mm), filled to about half volume. The selected settings were the following: test speed 1 mm/s, 30% deformation of the sample and two compression cycles. The chromaticity coordinates of the samples (obtained by a Minolta CR-10 camera) were reported in the form of a color difference, ΔE × ab [7].

#### 2.8.4. Sensory Analysis

A trained sensory panel (*n* = 13, aged 21–45 years) assessed the sensory profile of pasta samples. The lexicon consisted of twelve attributes as reported in Supplementary Table S2. A line scale from "not at all" (0) to "very" (10) for each attribute was used for the evaluation. Each pasta sample was cooked according to its own OCT and presented randomized in duplicate. Tap water was used to rinse the mouth between the samples. The study protocol followed the ethical guidelines of the sensory laboratory. A written informed consent was obtained from each participant.

#### *2.9. Statistical Analysis*

All analysis as well as the fermentation and enzymatic treatments were carried out in triplicate. The one-way ANOVA, using Tukey's procedure at *p* < 0.05, was performed for the data elaboration (Statistica 12.5, StatSoft Inc., Tulsa, USA). Principal component analysis (PCA) with varimax rotation was performed to visualize the sensory characteristics of the samples with Unscrambler X10.3 (Camo SA, Trondheim, Norway).

#### **3. Results**

#### *3.1. Proximate Composition of the Raw Materials*

The proximate composition of the flours is reported in Table 2. Hemp flour was characterized by the highest concentration of protein (ca. 37% of d.m.) and total dietary fiber (ca. 39.7% of d.m.), while chickpea flour was characterized by the lowest concentration of fat (ca. 4% of d.m.) (Table 2).


**Table 2.** Proximate composition and microbiological characterization of hemp and chickpea flours and wheat germ/bran mixture.

Data are expressed % of dry matter. a–c Values in the same row with different superscript letters differ significantly (*p* < 0.05).

#### *3.2. Proteolysis and Set-Up of the Bioprocessing*

TFAA and peptide profiles were used as screening criteria for bioprocessing parameters, since they correspond to the organic nitrogen compounds released during the process from native proteins. According to the peptide profiles (Figure 1), H and WGB were characterized by a total peak area significantly lower than C (3479 ± 34 and 5194 ± 25 mAU × mL vs. 15011 ± 53 mAU × mL). As

the consequence of the bioprocessing, changes were observed mainly in the range 20% to 40% of the acetonitrile gradient, while the hydrophilic zone had undergone minimal alterations (Figure 1). Enzymatic treatments led to slight increases in the total peak area of peptides. Values from 6.8% (CP) to 9.4% (WGBP) higher were found when proteases were used. Similarly, increases up to ca. 10% were found in CX and WGBX. HX and H were characterized by a similar peptide area. On the contrary, LAB fermentation caused a significant decrease in the peptide profile area, with HF and CF characterized by a relevant lower area (21 and 71%, respectively) than H and C.

**Figure 1.** Reversed-Phase Fast Performance Liquid Chromatography (RP-FPLC) peptide profiles of hemp, chickpea and wheat germ/bran doughs. (**A**): H, untreated hemp dough; HF, fermented hemp dough; HP, hemp dough treated with proteases (E1/E2); HX, hemp dough treated with xylanase (Depol 761P). (**B**): C, untreated chickpea dough; CF, fermented chickpea dough; CP, chickpea dough treated with proteases (E1/E2); CX, chickpea dough treated with xylanase (Depol 761P). (**C**): WGB, untreated wheat germ/bran (1:4) dough; WGBF, fermented wheat germ/bran (1:4) dough; WGBP, wheat germ/bran (1:4) dough treated with proteases (E1/E2); WGBX, wheat germ/bran (1:4) dough treated with xylanase (Depol 761P).

The concentration of TFAA in H, C and WGB prior to bioprocessing ranged from 802 ± 5 to 1972 ± 12 mg/Kg (Table 3). LAB and enzymes both increased TFAA concentration. Among the enzymatic treatments, proteases led to a concentration from ca. 3% (CP) to 70% (HP) higher than the corresponding untreated controls, while increases in the range 2 (CX) to 40% (WGBX) were found when xylanase was used. LAB led to the highest increases in the TFAA concentration (up to two-fold higher) (Table 3). In detail, the highest TFAA concentration was found in CF (2457 ± 16 mg/Kg).

**Table 3.** Chemical characterization of unfermented/fermented hemp, chickpea and wheat germ/bran doughs. H, dough made with hemp flour; HF, fermented dough made with hemp flour; C, dough made with chickpea flour; CF, fermented dough made with chickpea flour; WGB, dough made with a mixture of wheat germ and bran (1:4); WGBF, fermented dough made with a mixture of wheat germ and bran (1:4).


Hemp and chickpea doughs had a DY of 160; wheat germ/bran dough had a DY of 250. The data are the means of three independent experiments ± standard deviations (*<sup>n</sup>* <sup>=</sup> 3). a–e Values in the same row with different superscript letters differ significantly (*p* < 0.05).

Although both enzymatic treatments caused a moderate increase in the peptides in treated samples, the most extensive protein degradation that occurred during fermentation suggested a more intense potential effect of the LAB on protein digestibility. According to these considerations, fermented samples were subjected to further analysis.

#### *3.3. Chemical Characterization of Doughs*

A decrease in pH of ca. 2 units was achieved in all fermented doughs (Table 3). Significant higher values of TTA were found in HF, CF and WGBF compared to H, C and WGB, respectively. These changes are in accordance to the increases in lactic and acetic acids in fermented doughs. The highest concentration of lactic acid was found in WGBF, while HF contained the highest amount of acetic acid (Table 3). Decreases in the peptide concentrations (up to ca. 80%) were found after fermentation. The highest decrease was found in CF, while, according to the TFAA concentration, the lowest value was found in HF (Table 3).

#### *3.4. Chemical, Technological and Structural Properties of Pasta*

The inclusion of both unfermented and fermented ingredients affected the chemical characteristics of pasta. However, the pH and TTA values differ from *p*CT only when fermented ingredients were used (Table 4). TFAA concentration was higher in all fortified pasta, as compared to the pCT with higher extent when fermented ingredients were used. *p*WGBF contained the highest amount (ca. ten-fold higher than *p*CT) (Table 4).

The experimental OCT of *p*CT was ca. 10 min. Decreases (from 25 to 66%) in OCT were found for fortified pasta, especially when fermented ingredients were used (Table 4). A similar trend was found in terms of water absorption; lower values (13 to 32%) were found in *p*HF, *p*CF and *p*WGBF as compared to *p*CT. On the contrary, the cooking loss increased when pasta was fortified, being higher (up to 66%) when fermented ingredients were used. Hydration was also affected by the fortification; indeed, significantly higher values were found in fortified pasta especially when fermented doughs were used as ingredients. The highest value was reached in *p*WGBF.


Protein solubility in phosphate buffer was very low for all samples (< 2.91 ± 0.4 mg/g), while the addition of denaturing urea corresponded to a higher protein extraction (up to three-fold). The protein solubility in a buffer containing the disulfide reducing DTT was ca. two-fold lower compared to the phosphate buffer (data not shown), nevertheless, when both urea and DTT were used, the highest protein extraction was achieved. Overall, the protein solubility of fortified pasta was higher than the pasta containing only semolina. Moreover, the values of protein solubility were higher in fermented than the corresponding unfermented samples, probably due to the more intense proteolysis.

The hardness of fortified pasta samples was higher than *p*CT. However, when fermented doughs were used, lower values were found as compared to *p*H, *p*C and *p*WGB. Chewiness decreased with the fortification. However, when fermented doughs were used, it was higher in *p*CF and *p*WGBF as compared to *p*C and *p*WGB, respectively. The inclusion of wheat substitutes led to a decrease in the cohesiveness only when unfermented doughs were used. With the only exception of *p*HF, which showed lower value of chewiness, similar values were found between pasta containing fermented doughs and *p*CT.

Compared to *p*CT, lower lightness (L) and higher dE × ab values were found in all fortified pasta samples (Table 4). The highest "a" value, index for greenness (−)/redness (+) was observed for *p*WGB and *p*WGBF (Table 4).

#### *3.5. Nutritional Properties of Pasta*

As expected, the fortification with H, C and WGB improved the content of protein and total dietary fibers (Table 5). A fiber concentration higher than 6% was obtained with the fortification. Nevertheless, fortification also increased the content of the ANF, although the use of fermented doughs corresponded to lower concentration than corresponding unfermented controls. Overall, ca. ten-fold decreases in phytic acid and raffinose were observed in pasta with fermented doughs (Table 5). *p*CF and *p*HF contained the lowest amount of phytic acid and raffinose, respectively. Condensed tannins were from 23% to 59% lower in *p*HF, *p*CF and *p*WGBF as compared to the corresponding pasta with unfermented doughs.

Pasta samples containing fermented doughs were characterized by lower values of HI (up to 79%) as compared to the corresponding unfermented ones, except for *p*HF. The use of unfermented and fermented milling by-products led to the lowest values of HI (60.62 and 42.4, respectively). The *p*GI of unfermented and fermented pasta ranged from 72.99 to 81.27 and from 62.98 to 79.74, respectively. The lowest value was found in *p*WGBF. A similar trend was found in terms of the IVPD. Increases from 22% to 45% were found in fortified pasta as compared to *p*CT. Values 43–64% higher than *p*CT were found in *p*HF, *p*CF and *p*WGBF (Table 5). The fermentation led to pasta having an IVPD from 10% to 22% higher than the corresponding pH, *p*C and *p*WGB.



by-products. The data are the means of three independent experiments ± standard deviations (*<sup>n</sup>* = 3). a–f Values in the same0.05).n.d.:notdetected.

 row with different superscript letters differ significantly (*<sup>p</sup>* <

#### *3.6. Sensory Analysis*

Pasta was subjected to sensory analysis and the results are summarized in Figure 2. The PCA, representing 79.49% of the total variance of the data, showed that pasta samples are scattered in different parts of the plane according to the raw materials used for the production. All fortified samples were in a different part of the plane as compared to the control *p*CT, thus confirming the strong influence of the fortification on the sensory profile of pasta. Moreover, among fortified samples, the fermentation seemed to strongly affect the sensory profile of pasta only when milling by-products were used. Indeed, *p*WBGF and *p*WBG were scattered in different part of the plane. The former was characterized by a greater intensity of pungent odor and flavor and note of whole grains as compared to the corresponding *p*WBG. Only slight differences were found between *p*C and *p*CF and *p*H and *p*HF, respectively. *p*CF differentiated from the former due to the most intense legume note.

**Figure 2.** Loading plot (**A**) and score plot (**B**) of first and second principal components after principal component analysis (PCA) based on based on sensory analysis of pasta: *p*H, pasta containing 11% (*wt*/*wt*; d.m.) unfermented hemp dough; *p*HF, pasta containing 11% (*wt*/*wt*; d.m.) fermented hemp dough; *p*C, pasta containing 11% (*wt*/*wt*; d.m.) unfermented chickpea dough; *p*CF, pasta containing 11% (*wt*/*wt*; d.m.) fermented chickpea dough; *p*WGB, pasta containing 17% (*wt*/*wt*; d.m.) unfermented wheat germ and bran (1:4) dough; *p*WGBF, pasta containing 17% (*wt*/*wt*; d.m.) fermented wheat germ and bran (1:4) dough; *p*CT, pasta made only with semolina and water. The data are the means of three independent experiments ± standard deviations (*n* = 3). The attributes used in sensory analysis were: GA, general acceptability; OI, Odor Intensity; OP, Odor pungent; CH, Color Heterogeneity; S, Sapidity; PF, Pungent Flavor; DF, Delicate Flavor; TE, Texture; CHE, Chewability; ST, Stacking; L, Legume; W, Whole; TO, Toasted.

#### **4. Discussion**

The need for a diversified, balanced and healthy diet and the continued emphasis on the importance of dietary proteins and fibers are pressuring food companies and researchers to develop new products. Pasta is an important staple food; compared to other wheat-based foods, it is characterized by a lower glycemic index (GI), and it has been identified as a suitable carrier of bioactive compounds in daily diet [4,5,20]. Novel pasta recipes including the replacement of wheat flour with alternative flours, as well as the inclusion of pre-fermented ingredients have been recently proposed [20].

Here, hemp, chickpea and milling by-products were used to fortify semolina pasta. Aiming at improving the protein bio-accessibility and digestibility of the non-wheat flours and milling by-products before pasta making, treatments with food–grade enzymes and fermentation were investigated. Proteases from *A. niger*, commercial xylanase (Depol 761P) and LAB (*L. plantarum* LB1 and *L. rossiae* LB5) have been used as pre-treatment of the raw materials. The first selection of the more suitable bioprocess option was carried out by the evaluation of the peptide profiles and the TFAA concentration.

When hemp, chickpea and milling by-products were subjected to enzymatic treatments, a moderate increase in the peptides was observed as the consequence of the proteolysis of the native proteins (proteases) [38] and the release of soluble compounds from the fibrous cellular compartments (xylanase) [39]. On the contrary, fermentation led to a decrease in the peptides (up to ca. 70% lower) and a relevant increase in the TFAA (up to ca. 80% higher), suggesting an intense proteolysis operated by both endogenous and bacterial proteases and peptidases on proteins and their derivatives. It has largely been reported that the biological acidification operated by LAB lead to the activation of endogenous proteases which start the primary proteolysis where medium-sized polypeptides are released and subjected to LAB peptidase activities [40]. Based on these results, fermentation was chosen as the optimal bioprocessing option and its effects on hemp, chickpea, and milling by-products further investigated.

Aiming at investigating the suitability of hemp, chickpea and milling by-products as food ingredients, unfermented and fermented doughs were included in pasta formulation. In order to limit the weakening of the gluten network, the level of fortification was kept below 30% [7,41]. Nevertheless, the cooking performances and textural properties of fortified products were affected by the inclusion of the additional ingredients. A decrease in OCT and the increase in the cooking loss observed in fortified pasta might be due to the lower quality of the gluten network [20]. Overall, fortified pasta was characterized by values of hardness higher than control. However, the magnitude was lower when the fermentation was used as pre-treatment, regardless of the raw material. Data from panel test highlighted that the fortification affected the flavor of pasta. Indeed, *p*CT was characterized by the more intense delicate flavor, while legume, toasted and whole flavors were identified in fortified pasta, according to the raw materials used. The high intensity of pungent flavor and odor, due to the fermentation, also contributed to the differentiation among fortified pasta and *p*CT.

The fortification led to a pasta rich in fiber and protein, regardless of the fermentation process. Indeed, more than 13% of protein as well as ca. 6% (d.m) of fiber were achieved. According to EC Regulation [42] on nutrition and health claims on food products, experimental fortified pasta can be labelled as a "source of fiber" and a "source of protein". Nevertheless, increases in the ANF, as compared to control pasta were found as a result of the fortification. The fermentation with selected LAB led to significant degradation (to traces) of the phytic acid, raffinose and condensed tannins as compared to the corresponding unfermented samples. Phytic acid (Myo-inositol 1,2,3,4,5,6 hexakis [dihydrogen phosphate]) is considered ANF due to the binding capacity towards essential dietary minerals, proteins and starch, thus reducing their bioavailability. The degradation of phytic acid during fermentation is achieved mainly through plant phytases [19] activated by LAB acidification. Moreover, a specific role of the organic acids on phytase activity has recently been proposed [43]. Cation chelation from organic acids may inhibit the aggregation of minerals and other molecules by phytic acid, thereby increasing their digestibility [43]. When present at high concentration, i.e., in legumes, raffinose is

considered an ANF. However, LAB contribute to its enzymatic hydrolysis during fermentation [44], thus increasing product digestibility and reducing digestive discomfort [45]. The degradation of condensed tannins through LAB has already been proposed. It involves several enzymatic activities such as tannase, polyphenol oxidase and decarboxylase [46].

Beside the improvements in terms of ANF, an ca. 20% higher IVPD was found in fermented pasta samples as compared to the corresponding unfermented ones. Control pasta was characterized by an IVPD value 45% and 64% lower than the fortified samples (unfermented and fermented, respectively). Pasta containing non-wheat flours and milling by-products had a lower value of *p*GI compared to control, probably due to the higher concentration of dietary fibers and resistant starch, and a further decrease was found when the fermented flours were used. This effect could be attributed to biological acidification, which is among the main factors that decreases the starch hydrolysis rate and HI [36].

#### **5. Conclusions**

The study highlights the suitability of the fortification as a tool to improve the nutritional quality of pasta. Nevertheless, the pre-treatment of the non-wheat flours seems to be necessary to overcome the nutritional, structural and sensory drawback related to the use of such ingredients. Lactic acid bacteria fermentation has successfully been used to include hemp and chickpea flours and milling by-products in pasta making. LAB contributed to the increase in free amino acid content and decrease in phytic acid, raffinose and condensed tannins as compared to the corresponding unfermented doughs containing pasta. Moreover, fermentation improved protein digestibility and decreased the starch hydrolysis rate. Structural properties, cooking quality and sensory profiles were strongly affected by the fortification. Aiming at limiting the loss of rheological properties and cooking quality caused by the incorporation of non-wheat ingredients, further optimization of the technological processes may be needed.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2304-8158/8/12/604/s1. Table S1: Drying cycle used for making pasta; Table S2: List and definition of the attributes used for the sensory analysis made on pasta samples.

**Author Contributions:** R.S. carried out the laboratory work; E.P. contributed to the draft and critically revised the final manuscript; C.G.R. was the scientific advisor, responsible for the research funding and designed the experimental work. All authors read and approved the final manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** Biocatalysts (Biocatalysts Limited, Chicago, USA) is gratefully thanked for having provided the Depol 761P.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
