**A Novel and Simpler Alkaline Hydrolysis Methodology for Extraction of Ferulic Acid from Brewer's Spent Grain and its (Partial) Purification through Adsorption in a Synthetic Resin**

## **Pedro Ideia 1, Ivo Sousa-Ferreira <sup>2</sup> and Paula C. Castilho 1,\***


Received: 25 March 2020; Accepted: 18 April 2020; Published: 8 May 2020

**Abstract:** This work aims to develop simpler methodologies of extracting ferulic acid (FA) from brewer's spent grain (BSG). BSG is produced by brewing companies at high amounts all over the year and does not possess a direct application. Thus, its use as raw material for extraction of bioactive compounds has gained attention in the last years. FA has different interesting applications in cosmetics, food industry, and pharmaceutics. Several studies aim for its extraction from BSG by various methods, namely alkaline hydrolysis. In the present work, we suggest the use of autoclave to process higher amounts of BSG in a lab scale. A simplification of the regular post-hydrolysis procedures is also proposed to decrease the number of experimental steps and energy costs and to simultaneously increase the extraction yield (up to 470 mg of FA per 100 g of BSG). The adsorption of extracted FA in a synthetic resin is suggested as a partial purification method.

**Keywords:** ferulic acid; brewer's spent grain; alkaline hydrolysis; adsorption; synthetic resin

## **1. Introduction**

Due to the great political and social pressure in reducing pollution arising from industrial activities, large companies no longer consider residues as a waste but as valuable raw materials for other processes [1].

Brewer's spent grain (BSG) is the main solid by-product of brewing industry, produced during the wort elaboration step of beer production [1]. BSG is produced in a ratio of 20 kg per 100 L of beer [2], and the worldwide production is around 38.6 <sup>×</sup> 106 tons/year [3]. Despite being commonly used for animal and even human feed [4–7], excessive BSG availability is gaining attention for other applications, among which are the production or the extraction of high value added compounds, namely oligosaccharides [8], xylitol [9,10], and ferulic acid (FA) [11–13], a phenolic compound belonging to the hydroxycinnamic acids family.

Due to its physiological functions—anti-oxidant, anti-inflammatory, anti-thrombosis anti-microbial, and anti-cancer—as well as its protective effect against coronary disease, FA is considered one of the most important phenolic compounds [14]. Several applications are described, such as vanillin production [15], as preservatives [16,17], and as an ingredient for dermatologic lotions [14], among others. Because FA is covalently linked to the structure of lignocellulosic biomass by ether and ester bonds, conventional extraction techniques (e.g., solid–liquid extraction) are not effective in its separation from the matrix. Other techniques, such as alkaline hydrolysis, are needed in order to cleave these bonds and release FA. In fact, alkaline hydrolysis is able to cleave the lignin/phenolic-carbohydrate complexes structure, resulting

in a phenolic portion, soluble sugars, insoluble lignin, and carbohydrates [18]. Other methods such as enzymatic hydrolysis were developed to recover ferulic acid from lignocellulosic biomass, namely wheat bran [19,20]. The main disadvantages of enzymatic hydrolysis are the cost of enzymes and/or the reaction time. Additionally, for the process to be efficient, control of reaction temperature and pH is required.

Different procedures for alkaline hydrolysis reactions are described in literature, namely those reported by Mussatto et al. (2007) [11] and McCarthy et al. (2013) [6]. The most common is to perform the reaction into auto-pressurized tubes or cylindrical stainless steel reactors at high temperatures and high pressures. Moreia and co-workers [13,21] suggested using microwave assisted extraction (MAE) to promote FA release from lignocellulosic materials, such as BSG. Despite the short time required for extraction, MAE's main limitation is the small amount of BSG the system is able to process in each batch. The present work aimed to optimize FA extraction from BSG by the simplification of the methods described in literature, which resulted in reducing the time and the resources required for extraction. Additionally, extraction in an autoclave allows one to scale-up the FA extraction, bridging the limitations of other processes already described.

Currently, adsorption technology is widely used for the removal of organic compounds from aqueous solutions and heterogeneous mixtures. The main disadvantage associated with the most used adsorbents is the high regeneration cost. This has stimulated the research on new adsorbents such as macrobead synthetic resins, which may provide a cheap and effective chemical regeneration process [22].

Several authors refer to the adsorption of ferulic acid in resins such as Amberlite XAD-16 [23] or Lewatit-type resins [24], aiming at its purification. These last authors studied three different polystirene-based macroporous resins, Lewatit S6328 A (an anionic, strongly alkaline exchange resin), Lewatit S2328 (food grade cationic exchange, strongly acidic), and Lewatit S7968, a resin without functional groups, which gave the best performance on the adsorption of chlorogenic acids from artichoke residues with little sugar co-adsorption. A similar resin Lewatit VPOC1064 was chosen for purification of the extracted FA in the present work.

Our goal in this work was to use the principles of methods already reported and simplify the processes for cleaning as well as partially purify the FA extracted from BSG.

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

### *2.1. Chemicals and Reagents*

All reagents and standards were of analytical reagent (AR) grade. Folin–Ciocalteu reagent, gallic acid (99%), and sodium hydroxide (98%) were purchased from Panreac (Madrid, Spain). Ferulic acid (99%) and sodium carbonate (99.8%) were from Sigma-Aldrich (St. Louis, MO, USA). Absolute ethanol was obtained from Riedel-de Haën, acetonitrile (99.9%) from Fisher Scientific, and formic acid (≥98%) from Sigma-Aldrich. Synthetic resin (Lewatit VPOC1064 MD PH®) was purchased from LANXESS (Köln, Germany).

Dimethylsulfoxid-d6 (DMSO-d6) (≥99%) was purchased from Merck (Darmstadt, Germany).

## *2.2. Raw Material*

Brewer's spent grain (BSG) was provided by a local brewing company (ECM, Empresa de Cervejas da Madeira). The samples were freeze-dried immediately after delivery to our laboratory (NatLab—CQM). Dry BSG was stored at −20 ◦C until use. These samples were used throughout the work with the simple designation of "BSG". A sample of fresh BSG was used for moisture and ash content determinations.

### *2.3. Physicochemical Characterization*

The moisture content determination was carried out on a KERN DBS 60-3 moisture balance running a semi-automatic program, which heats the sample at 120 ◦C until the moisture content is stable for 30 s. Ash content (%) was determined after incineration of BSG samples in a muffle furnace at (500 ◦C for 24 h). Particle size distribution was studied by passing three freeze-dried BSG portions though a set of sieves with decreasing mesh size (1.0, 0.5, 0.25 and 0.125 mm).

## *2.4. Alkaline Hydrolysis—Basic Procedure*

Alkaline hydrolysis was performed according to the procedure outlined in Figure 1A and described in literature. Reactions were performed in Ace pressure tubes using BSG and NaOH solution in a solid:liquid ratio of 1:20 (*w*/*v*). Experimental conditions were optimized and set as 120 ◦C for 1.5 h with 20 mL of NaOH (2%) for 1 g of BSG. Optimization was performed by individual variation of each of the extraction parameters.

**Figure 1.** Proceeding to the alkaline hydrolysis reaction. (**A**) classic procedure; (**B**) simplified methodology.

The tubes were placed into an oil bath and heated. After alkaline hydrolysis reaction, the mixture was cooled to room temperature, and the solid residue, containing mainly cellulose and lignin, was separated by centrifugation. Precipitation of the hemicellulose fraction was triggered by addition of ethanol to a final concentration of 30% (*v*/*v*), and its separation was performed by centrifuging. The supernatant, which contained the ferulic acid, was neutralized with HCl (6M) and concentrated on a rotary evaporator. The aqueous phase was freeze-dried and stored at −20 ◦C until analysis.

## *2.5. Pretreatment by Solid–Liquid Extraction with Acetone*

A portion of BSG was extracted with acetone (60%) in a solid/liquid ratio of 50 mL/g in an ultrasound bath for 1 h. After solid–liquid extraction, the mixture was allowed to cool to room temperature, and the solid residue was separated from the supernatant by centrifugation followed by filtration. The solid residue was washed with distilled water and freeze-dried. Alkaline hydrolysis was carried out in Ace tubes at 120 ◦C for 1.5 h with 20 mL of NaOH (2%) for 1 g of BSG. For comparison, a portion of BSG not subjected to pretreatment was extracted by alkaline hydrolysis in the same conditions.

## *2.6. Extraction by Alkaline Hydrolysis in Autoclave*

An alkaline hydrolysis assay was carried under the same conditions (120 ◦C for 1.5 h with 20 mL of NaOH (2%) for 1 g of BSG) in autoclave. The subsequent procedure was similar to that described for Ace tubes. The resultant extracts were stored at −20 ◦C until analysis.

## *2.7. Simplification of the Procedure*

A modification of the process described in the literature was introduced in order to diminish the number of separation steps and thus improve the yield in the desired product; after alkaline hydrolysis, the mixture was cooled, and ethanol was added (Figure 1B). Precipitated hemicellulose was separated by centrifugation together with lignin and cellulose fraction. The supernatant was neutralized, concentrated, and the aqueous phase was freeze-dried and stored at −20 ◦C until analysis.

#### *2.8. Analysis of Extracts*

The extracts obtained after the procedures described in the previous sections were analyzed in terms of their total soluble solids (TSS), total phenolic content (TPC), and quantification of ferulic acid by HPLC-diode-array detector (DAD).

#### 2.8.1. Total Soluble Solids (TSS) Determination

For TSS determinations [25], extracts were resuspended in water (10 mg/mL) and filtered through membrane filters (0.45 μm). Using an ATAGO RX-1000 refractometer, the TSS was measured based on a calibration curve of sucrose (5–50 mg/L). The results are expressed in milligrams of sucrose equivalent (SE) per 100 g of dry BSG.

#### 2.8.2. Total Phenolic Content (TPC) Determination

TPC was determined by Folin–Ciocalteu method [26]; fifty microliters of the sample in methanol (5 mg/mL) was mixed with 1.25 mL of Folin–Ciocalteu solution (1:10) and 1 mL of Na2CO3 (7.5%). The absorbance at 765 nm was measured after 30 min, and results were expressed as milligrams of gallic acid equivalent (GAE) per 100 g of BSG (dry weight).

## 2.8.3. Quantification of Ferulic Acid

Ferulic acid concentrations were determined by high performance liquid chromatography (HPLC) using a UV detector (at 320 nm) and a Phenomenex Gemini C18 (5 μm, 250 × 0.3 mm i.d.) column. The HPLC analysis, adapted from Gouveia and Castilho (2011) [26], was performed on a Dionex ultimate 3000 series instrument (Dionex, Sunnyvale, CA, USA) coupled to a binary pump, a diode-array detector (DAD), an autosampler and a column compartment. Samples (5 mg/mL) were prepared in the mobile phase and filtered through 0.45 μm membranes (Millipore, Burlington, MA, USA). Then, samples were injected into the equipment under the following conditions: column at 30 ◦C, acetonitrile/0.1% formic acid as mobile phase (isocratic elution, 75:25), a flow rate of 400 μL/min, and injection volume of 10 μL. Results are expressed as mg of FA per 100 g of BSG (dry weight).

Identification was performed comparing retention times with those obtained from commercial ferulic acid standard. Quantification was based on the UV signal response at 320 nm, and the resultant peak areas in the chromatograms were plotted against concentrations obtained from standard. Calibration curve (5–100 mg/L) was prepared by diluting the stock solutions (1000 mg/L in methanol) with the initial mobile phase. Quantification was carried out by plotting peak area versus concentration (*R*<sup>2</sup> = 0.9994).

#### 2.8.4. Statistical Analysis

All samples were assayed in triplicate (*n* = 3), and results are given as means ± standard deviations. Differences between means were tested by ANOVA using SPSS Statistics 22 software.

#### *2.9. Purification by Adsorption on a Synthetic Resin*

Synthetic resin (Lewatit VPOC1064 MD PH®) was used to promote adsorption of FA from the extract obtained after alkaline hydrolysis using the simplified method previously described (Figure 1B).

#### 2.9.1. Kinetic Studies with FA Standard

Before purification of alkaline hydrolysis extract, adsorption kinetic studies with FA standard were performed. During these studies, several conditions were tested in three different assays. In the first assay, a proportional variation of FA concentration and the amount of adsorbent were carried up within three different tests. A second assay was achieved in two tests, where different initial concentrations of FA were studied. Finally, in the third assay, the effect of temperature in the adsorption was evaluated. Two tests were performed at room temperature (22–25 ◦C) and 6 ◦C (controlled ice bath), respectively. Table 1 resumes the conditions of the different assays.


**Table 1.** Conditions of the adsorption mixtures during in adsorption kinetic studies. The volume of ferulic acid (FA) solution used in each teste was constant (50 mL).

A supernatant sample was collected every 2 min in the first 10 min, every 10 min during the next 90 min, and every 20 min until 180 min of adsorption. After filtration of supernatants, FA concentration was determined. Adsorption percentage, the amount of FA adsorbed per gram of resin, and *C*/*C*<sup>0</sup> were used to evaluate the adsorption process.

The equation below was applied to the experimental data, and the parameters α, β, γ, δ and θ were determined using the Solver Microsoft Excel add-in program.

$$\frac{\mathbf{C}}{\mathbf{C}\_0} = a + \beta \cdot \mathbf{e}^{-\frac{\mathbf{f}}{\overline{\mathbf{y}}}} + \delta \cdot \mathbf{e}^{-\frac{\mathbf{f}}{\overline{\mathbf{g}}}}$$

where *C*<sup>0</sup> is the initial concentration of the absorbate (g/L), and *C* is the concentration at the time *t* (g/L).

2.9.2. Purification of an Alkaline Hydrolysis Extract

As outlined in Figure 2, a portion of 2 g of alkaline hydrolysis freeze dried extract was dissolved in 40 mL of water at 40 ◦C with vigorous stirring. After cooling the mixture to room temperature, a portion of 10 g of pre-activated resin (according to the supplier, treatment with 6% HCl and 4% NaOH and washing with distilled water) was added to the flask. The mixture was magnetically stirred for 2 h and filtered under reduced pressure. The filtrate was used to determine the FA adsorption yield, and the loaded resin was further stirred with 50 mL of ethanol:water (70%) in order to promote desorption. A new filtration allowed us to separate the resin from the liquid phase containing the FA. The determination of desorbed FA was possible through analysis of the filtrate.

**Figure 2.** Procedure for the partial purification of FA by adsorption on a synthetic resin.

The extracts were analyzed in terms of FA quantification by HPLC-DAD according to the conditions mentioned before. 1H NMR was performed to verify the partial purification of the extract obtained by alkaline hydrolysis of BSG and treated with the Lewatit resin. For NMR analysis, 10 mg of each extract were dissolved in 1 mL of DMSO-d6 and transferred to 5 mm NMR tubes. 1H spectra were recorded on a Bruker UltraShield 400 Plus NMR (Bruker, Billerica, MA, USA) at 10,061 MHz and 400 MHz. Acquisition parameters for 1H were: size of fit 65 k; spectral width 4401 Hz; acquisition time 64 k; relaxation delay 1 s; number of scans 512.

## **3. Results and Discussion**

## *3.1. Physicochemical Characterization of BSG*

The moisture content of supplied BSG was 68.44 ± 0.93%, and the ash content was 4.18 ± 0.03%. Sifting of freeze-dried material revealed that BSG was provided as a fine powder with particle size between 1 and 0.25 mm (Table 2), and it was further used without any separation.


**Table 2.** Particle size distribution.

## *3.2. Optimization of Alkaline Hydrolysis Conditions*

Table 3 summarizes the results for optimization of alkaline hydrolysis reaction in Ace pressure tubes. Different reaction temperatures (60, 80, 100, and 120 ◦C) were tested for 1 h with NaOH (2%). Reaction time (1 to 3 h) was tested at 100 ◦C using NaOH (2%). Optimization of alkali solution concentration was performed in reactions at 100 ◦C for 1 h. Optimal conditions were set as 120 ◦C for 1.5 h with 20 mL of NaOH (2%) for 1 g of BSG.


**Table 3.** Conditions and results obtained for optimization of the parameters for alkaline hydrolysis reaction in Ace pressure tubes. FA yield is expressed as mg of FA per 100 g of brewer's spent grain (BSG).

## *3.3. Pretreatment by Solid–Liquid Extraction with Acetone*

The results compiled in Table 4 reveal that extraction with acetone is effective in removing free sugars from the matrix, resulting in a significant decrease in TSS for pretreated BSG compared with the untreated portion. FA concentration in the extracts obtained by alkaline hydrolysis of both untreated and pretreated BSG portions did not show a statistically significant difference. Results obtained in previous works [27] showed that solid–liquid extraction with 60% acetone is efficient in the extraction of free form compounds from BSG. The present data show that solid–liquid extraction is not efficient in the extraction of FA from lignocellulosic materials, since it is covalently bonded to their structure, but soluble solids such as mono and disaccharides are partially removed.

**Table 4.** Comparison of the results obtained for alkaline hydrolysis of not pretreated and treated BSG (1 h). Total soluble solids (TSS) is expressed as mg of sucrose equivalent per 100 g of BSG, total phenolic content (TPC) is expressed as gram of gallic acid equivalent per 100 g of BSG, and FA yield is expressed as mg of FA per 100 g of BSG.


<sup>a</sup> indicates not significant differences and <sup>b</sup> indicates significant differences. SE: sucrose equivalent; GAE: gallic acid equivalent.

## *3.4. Extraction by Alkaline Hydrolysis in Autoclave*

The characterization of the extracts (Table 5) showed that differences on TSS were not significant. However, there was a statistically significant increase of TPC and FA yield for those obtained by alkaline hydrolysis in autoclave compared to those obtained in Ace pressure tubes. This might have been due to a higher contact exchange between BSG and alkali solution in autoclave, resulting from the greater volume of the reaction vessels in which hydrolysis was performed. Taking into account the aspects mentioned above, alkaline hydrolysis of BSG in an autoclave might be an interesting process

for a possible scale-up of the extraction process, since even everyday laboratory equipment is capable of processing large amounts of BSG.


**Table 5.** Comparison of the results obtained for alkaline hydrolysis on Ace pressure tubes and in an autoclave (1.5 h). TSS is expressed as mg of sucrose equivalent per 100 g of BSG, TPC is expressed as gram of gallic acid equivalent per 100 g of BSG, and FA yield is expressed as mg of FA per 100 g of BSG.

<sup>a</sup> indicates not significant differences and <sup>b</sup> indicates significant differences.

## *3.5. Simplification of the Procedure*

Table 6 shows the results for the simplification of the procedure after alkaline hydrolysis in comparison with the normal procedure (schematized in Figure 1). Both TSS and TPC showed a large increase (of about 30.9% for TSS and 122.97% for TPC). FA yield was increased in about 80% with the change of procedure, up to 476.99 ± 25.94 mg (FA)/100 g (BSG, dry weight). The increase was probably due to the solvent washing during the addition of the ethanol to the alkaline liquor, which resulted in reduction of losses associated with the process. Thus, in addition to reducing the experimental steps and the energetic resources required for extraction, simplification of the procedure also permits obtaining a greater amount of FA. TSS increase may be an indication that larger carbohydrates are degraded into smaller, soluble sugar molecules.

**Table 6.** Comparison of the results obtained for normal and simplified procedures applied after alkaline hydrolysis. TSS is expressed as mg of sucrose equivalent per 100 g of BSG, TPC is expressed as gram of gallic acid equivalent per 100 g of BSG, and FA yield is expressed as mg of FA per 100 g of BSG.


<sup>a</sup> indicates not significant differences and <sup>b</sup> indicates significant differences.

## *3.6. Purification by Adsorption on a Synthetic Resin*

## 3.6.1. Kinetic Studies with FA Standard

Three assays were performed to study the adsorption kinetics under different experimental conditions, according to Table 1.

After determination of FA concentration in the various supernatant samples, *C*/*C*<sup>0</sup> was calculated for each time *t*, and parameters α, β, γ, δ and θ were determined.

Figure 3 resumes the evolution of *C*/*C*<sup>0</sup> along the 180 min of adsorption in the different assays. Figure 3a is related to assay 1, where dispersion of both FA and resin increased in proportion within the tests *AdsA*, *AdsB*, and *AdsC*. Adsorption isotherms show that the variation in the concentration of adsorbed FA during the equilibrium was proportional to the dispersion of both adsorbate and adsorbent

in the mixture. Assay 2 (Figure 3b) was performed to study the effect of the initial concentration of FA in the adsorption process. The FA concentration in test *AdsHC* was four times higher than that in test *AdsLC*. Because the amount of resin was similar in both tests, isotherms indicated a more efficient adsorption process for *AdsLC*. Regarding the study of the effect of temperature in the adsorption process, Figure 3c resumes the results obtained under room temperature (*AdsRT*) and 6 ◦C (*AdsT6*). Results showed that, at room temperature, the equilibrium was achieved faster than at 6 ◦C. In all tests,

it was possible to establish that 100–120 min is enough to reach equilibrium.

**Figure 3.** Adsorption kinetic isotherms for the assays performed to study the effect of (**a**) dispersion of FA and resin, (**b**) the initial concentration of FA, and (**c**) the temperature in the adsorption mixture.

Parameters for adsorption isotherms equations were determined and compiled in Table 7. Based on kinetic studies results, the best conditions for adsorption of FA standard may be an intermediate between the conditions of assays *AdsA*, *AdsLC*, and *AdsRT*. For real samples, where the goal is to purify FA from a variety of compounds of different chemical nature, the ideal conditions might be different, namely because of the competition for adsorption in the resin.


**Table 7.** Determination of parameter estimates and sum of squared differences (SSD) by MS Solver for adjustment of adsorption kinetic curves.

## 3.6.2. Purification of an Alkaline Hydrolysis Extract

Adsorption of FA from the alkaline hydrolysis extract showed to be effective (90.83%), where around 4.6 milligrams of compound were adsorbed into the resin. The desorption was achieved by adding two portions of 25 mL of 70% ethanol followed by continuous stirring for 30 min and filtration. The filtrates were combined, concentrated, and the concentration of FA was determined by HPLC-DAD. Results showed that 68.70% of the adsorbed FA was desorbed under these conditions, indicating that about 1.4 milligrams of FA remained adsorbed.

1H NMR spectra (Figure 4) suggests a partial purification of the extract obtained by alkaline hydrolysis of BSG in an autoclave and a simplified procedure. Figure 4A corresponds to FA standard spectrum. FA peaks were identified in both Figures 4B and 4C for initial and partially purified extracts, respectively. Because FA concentration increased after purification, the peaks between 6.3 and 7.5 ppm and the peak at 3.8 ppm slightly increased in the final extract when compared to the original extract. On the contrary, peaks at 8.4 and 5.3 ppm as well as regions 3.5 to 3.7, 3.0 to 3.4, 2.6 to 2.8, 1.5 to 2.3, and 1.0 to 1.3 ppm decreased in the final extract compared with the initial extract. This suggests a decrease in the concentration of these compounds after partial purification with synthetic resin.

113

## **4. Conclusions**

Because of its high availability and potential, several works are being developed around BSG. One of the interesting compounds present in the lignocellulosic structure of BSG is ferulic acid. Since FA is not extracted from the BSG matrix by conventional solid–liquid extraction methods, alternative techniques such as alkaline hydrolysis are often applied. The methods described in literature usually comprise a pretreatment step to remove compounds resulting from the brewing process. Alkaline hydrolysis is commonly followed by centrifugation, precipitation of hemicellulose fraction in the alkaline liquor, and neutralization. In the present work, the use of autoclave to perform the alkaline hydrolysis and a simplification of the post-extraction process were suggested and were shown to increase FA yield. The extraction in an autoclave resulted in an increase of FA extraction yield of around 38% when compared with the extraction in pressure tubes. An increase of about 84% in the extraction yield was achieved in a small scale extraction in pressure tubes when a simplification of the post-extraction process was applied. A partial purification by adsorption on a synthetic resin was also suggested, constituting a potential approach to obtain ferulic acid in a higher degree of purity.

During the studies, *p-*coumaric acid was co-extracted in molar proportion (*p-*CA:FA) between 1:4 and 1:8, depending on the extraction conditions.

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

**Funding:** This research was partially supported by FCT—Fundação para a Ciência e a Tecnologia with funds from the Portuguese Government, under de projects UIDB/00674/2020 (Centro de Química da Madeira) and UIDB/00006/2020 (Centro de Estatística e Aplicações). PI grant through the project M1420-01-0145-FEDER-000005—Centro de Química da Madeira—CQM+ (Madeira 14–20) is also acknowledged.

**Acknowledgments:** Authors acknowledge Empresa de Cervejas da Madeira for the supply of BSG samples.

**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* **Radical Scavenging and Antimicrobial Properties of Polyphenol Rich Waste Wood Extracts**

## **Anita Smailagi´c 1, Petar Ristivojevi´c 2, Ivica Dimki´c 3, Tamara Pavlovi´c 3, Dragana Dabi´c Zagorac 1, Sonja Veljovi´c 4, Milica Fotiri´c Akši´c 5, Mekjell Meland <sup>6</sup> and Maja Nati´c 2,\***


Received: 12 February 2020; Accepted: 2 March 2020; Published: 10 March 2020

**Abstract:** The main focus of this study is to assess radical scavenging and antimicrobial activities of the 11 wood extracts: oak (*Quercus petraea* (Matt.) Liebl*., Q*. *robur* L., and *Q. cerris* L.), mulberry (*Morus alba* L.), myrobalan plum (*Prunus cerasifera* Ehrh.), black locust (*Robinia pseudoacacia* L.), and wild cherry (*Prunus avium* L.). High-performance thin-layer chromatography (HPTLC) provided initial phenolic screening and revealed different chemical patterns among investigated wood extracts. To identify individual compounds with radical scavenging activity DPPH-HPTLC, assay was applied. Gallic acid, ferulic and/or caffeic acids were identified as the compounds with the highest contribution of total radical scavenging activity. Principal component analysis was applied on the data set obtained from HPTLC chromatogram to classify samples based on chemical fingerprints: *Quercus* spp. formed separate clusters from the other wood samples. The wood extracts were evaluated for their antimicrobial activity against eight representative human and opportunistic pathogens. The lowest minimum inhibitory concentration (MIC) was recorded against *Staphylococcus aureus* for black locust, cherry and mulberry wood extracts. This work provided simple, low-cost and high-throughput screening of phenolic compounds and assessments of the radical scavenging properties of selected individual metabolites from natural matrix that contributed to scavenge free radicals.

**Keywords:** wood waste; phenolic profile; planar chromatography; DPPH-HPTLC assay; antimicrobial activity

## **1. Introduction**

Ageing processes of some alcoholic beverages are one of the most important practices during their production. This contributes to improved sensory characteristics such as aroma, color, taste and astringency. The most commonly used material in cooperage is oak heartwood barrels. Alternative wood species such as chestnut, cherry and mulberry are also used in Balkan cooperages, in different forms such as wood chips and staves [1]. Nowadays, notable studies have showed that agri-food wastes and by-products, including waste from barrel production, represent an inexhaustible source of valuable biologically active compounds. Additionally, this waste represents a low-cost material, which can be used as material for the production of the extracts. Recently, various extraction techniques were reviewed and compared with classical extraction procedures used for recovery of the antioxidant compounds from wastes [2]. Using simple, fast and inexpensive eco-friendly extraction methods for phenolic compounds represents an efficient method and advantage for further implementation in the food, pharmaceutical and cosmetic industries [3–5].

From the production of wood barrels, it is estimated that more than 200 tons of wood waste is available annually in Serbia [6]. Forests and other wooded land occupy ~2.5 million hectares, which is about one third of the territory of the Republic of Serbia. These natural populations of Serbia contain a large number of economically important forest tree species (oak, beech, black locust, spruce, pine and fir) together with autochthonous and introduced wild fruit trees species (wild cherry, cherry plum, mulberry, wild pear, wild apple, cornelian cherry, hazel and walnut) which are used for timber production, afforestation and erosion prevention, for grafting, in human diet, in medicine, in industrial processing and in landscape architecture [7,8].

In Serbia, barrels are mostly made of oak (pedunculate oak, *Quercus robur,* or sessile oak, *Quercus petraea* (Matt.) Liebl. L.) and Turkey oak (*Quercus cerris* L.) but sometimes black locust (*Robinia pseudoacacia* L.), myrobalan plum (*Prunus cerasifera* Ehrh.), mulberry (*Morus alba* L.) and even wild cherry (*Prunus avium* L.) are used as a cheaper substitute. Oak is the most widespread deciduous tree in Serbia, a national tree with strong historical and religious importance. Myrobalan plum is a native tree in Southeast Europe, has great genetic importance for horticultural breeding, and has spread throughout the whole country in all kinds of micro-climatic and pedologic conditions. Mulberries are very common, since ex-Yugoslavia used to be the fifth largest silk producer in the world with more than 2.5 million white mulberry trees [9]. Wild cherry, a noble tree, is widely distributed by birds; its seeds are used for generative rootstock production and its fruits are suitable for table consumption and as a local medicine [10,11]. Black locust is mostly used in construction works, as technical or ornamental wood, and is most commonly used as firewood. It has special value as a honey species for beekeeping [12].

Wood waste has a potential to be reused in the food and pharmaceutical industry due to its richness in potentially bioactive phenolic compounds with high antioxidant and antimicrobial activity. In our previous research [6], ellagic acid was abundant in sessile and pedunculate oak wood. It was also found in Turkey oak, black locust and myrobalan plum, but in much lower quantities. Mulberry contained the largest concentration of *p*-hydroxybenzoic acid and stilbenoids in comparison with other wood species, while myrobalan plum showed the highest content of protocatechuic acid and 5-*O*-caffeoylquinic acid. Wild cherry was characterized by richness in flavonols, flavanones, flavones, isoflavones and flavanonols [6,13,14], with taxifolin as the most abundant phenolic compound [6]. Extracts from sessile and pedunculate oak, black locust, myrobalan plum, wild cherry and mulberry showed notable antioxidant capacity, with the highest radical scavenging activity in the latter extract. Turkey oak showed the lowest radical scavenging activity [6]. According to the literature, phenolic acids were identified as the major contributors to the antioxidant capacity in wood samples, including gallic, protocatechuic, *p*-coumaric and ellagic acid and all the ellagitannins, due to their characteristic structure [15]. The following phenolic acids: ferulic acid, caffeic acid, protocatechuic acid, gallic acid, *p*-coumaric acid and chlorogenic acid, also present in some wood species, exhibit strong free radical scavenging properties on silica plates [16].

It is proposed that phenolic compounds can damage the bacteria cell membrane by interacting with the proteins of the cell membrane, or can be involved in interaction with cellular enzymes [17], which may directly or indirectly cause metabolic dysfunction and finally bacterial death [18]. Phenolic compounds are able to inhibit bacterial quorum sensing signal receptors, enzymes and secretion of toxins [19]. The type, structure and concentration of phenolic compounds, as well as the microorganism used, will influence the bacterial growth. Large doses of phenolic compounds may be toxic for bacteria, but lower doses can be used as substrates [17].

Some phenolic compounds present in several wood species showed antimicrobial activity. Taxifolin exhibited antibacterial activity against six known clinical pathogens: *Escherichia coli*, *Listeria*

sp., *Pseudomonas aeruginosa*, *Bacillus* sp., and *S. aureus* [20]. Oxyresveratrol, the most abundant stilbene in mulberry, was active against the methicillin-resistant *S. aureus* [21].

Among flavonoids present in wild cherry wood, flavonols were distinguished by effective antimicrobial activity against resistant bacteria [22]. Methanolic extract (80%, *v*/*v*) from oak bark (*Q. robur* L.) showed moderate bactericidal, fungicidal, bacteriostatic and fungistatic activity on *S. aureus*, *Enterobacter aerogenes* (today known as *Klebsiella aerogenes*) and *C. albicans* [23]. *Q. robur* bark showed strong antibacterial activities against *Pseudomonas aeruginosa, M. flavus* and *E. coli,* and moderate effects against other bacterial species [24]. Heartwood and resin of cherry wood exhibited cytochrome inhibition and antifungal activity [25], while cherry wood extracts possessed noticeable antimicrobial activity against 9 out of the 11 wine organisms tested [17]. Oak wood has abundant ellagitannins, which are toxic to microorganisms, and provides good resistance to fungal degradations [26].

Antimicrobial resistance presents a global problem since resistant pathogens can cause life-threatening conditions that become incurable with one or more known drugs. The mechanisms of the antibacterial activities of many plant-derived flavonoids are different than those of conventional drugs, which open new possibilities in enhancement of antibacterial therapy [27]. In addition, many synthetized drugs have side-effects, which are small in the case of plant-derived compounds [27]. Due to all these reasons, the development of alternative drugs derived from natural resources is an attractive option.

Radical scavenging activity using DPPH-HPTLC (high performance still layer chromatography) assay and antimicrobial activity on wood waste extracts are not investigated so far. Thus, the main aim of this research was to assess radical scavenging and antimicrobial activities of the wood waste extracts from mulberry (*M. alba* L.), myrobalan plum (*P. cerasifera* Ehrh.), black locust (*R. pseudoacacia* L.), wild cherry (*P. avium* L.), and different species of oaks (*Q. petraea* (Matt.) Liebl*., Q*. *robur* L. and *Q. cerris* L.) and consider their usage in the pharmaceutical and food industries. Phenolic compounds were separated by using HPTLC, while radical scavenging activity was determined using DPPH-HPTLC.

## **2. Materials and Methods**

## *2.1. Chemicals*

Ethyl acetate was purchased from Merck (KGaA, Darmstadt, Germany); formic acid, hexan 2,2-diphenyl-1-picrylhydrazyl (DPPH) and phenolic standards from Sigma-Aldrich (Steinheim, Germany); and 2-aminoethyl diphenylborinate (NTS) from Fluka (Steinheim, Germany). Gallic acid, ferulic acid and caffeic acid were supplied by Sigma Aldrich (Steinheim, Germany).

## *2.2. Samples and Preparation of Wood Extracts*

Eleven different wood staves of different geographical origins were analyzed (Table 1). In total three samples of Pedunculate oaks (*Quercus robur* L.), three of sessile oaks (*Quercus petraea* (Matt.) Liebl), and one sample of Turkey oak (*Quercus cerris* L.), black locust (*Robinia pseudoacacia* L.), myrobalan plum (*Prunus cerasifera* Ehrh.), wild cherry (*Prunus avium* L.), and mulberry (*Morus alba* L.) were included. Nine staves were stored for the whole year in the open air at cooperage industry VBX-SRL. D.O.O. in Kraljevo, Central Serbia., while two samples (sessile oak from Kuršumlija and Turkey oak) were not seasoned [6]. The wood age of the oak wood staves was over 60 years, while the wood age of non-oak wood staves was more than 40 years.

Firstly, the staves were grinded in a mill for wood and sieved until granulation of 0.5–1.5 mm was obtained. The sawdust (2.5 g) was extracted with 25 mL of ethanol (60%, *v*/*v*), in Erlenmeyer flasks, with constant stirring in a magnetic stirrer for seven days in darkness and room temperature (20 ± 2 ◦C) [6]. The extracts were centrifuged twice (5 min at 8000 rpm). For investigation of antimicrobial activity, the extracts were evaporated with a rotary evaporator and diluted in methanol until the concentration of 50 mg mL−<sup>1</sup> was reached. The extraction yield of each extract was calculated from the weight of the extract residue obtained after solvent removal and the weight of waste wood employed in the extraction procedure.


**Table 1.** Selected wood waste extracts of different forest trees for DPPH-HPTLC (high-performance thin-layer chromatography) and antimicrobial testing assay.

#### *2.3. High-Performance Thin-Layer Chromatography and Image Analysis*

HPTLC Silica gel 60F254 plates were used for both HPTLC fingerprint and DPPH-HPTLC assay (Merck, Germany). The oak and wild cherry samples (5 μL), black locust, myrobalan plum and mulberry (2 μL), and four standard compounds: gallic acid, ferulic acid, caffeic acid and *p*-coumaric acid (2 μL, c = 1000 ppm), were applied as bands (8 mm) using Linomat 5 system (Camag, Muttenz, Switzerland).

The mobile phase consisted of a mixture of ethyl acetate:hexan:formic acid:water (11:2:1:0.5 *v*/*v*/*v*/*v*). The plates were developed at room temperature (20 ◦C) in a twin-trough-chamber (CAMAG) saturated with the vapors of mobile phase for 15 min, at a developing distance of 70 mm. The obtained HPTLC chromatograms were derivatized with 2-aminoethyldiphenylborate solution (NTS - 0.2% in ethanol) in order to intensify the fluorescence of compounds.

For DPPH-HPTLC assay, a developed HPTLC chromatogram was immersed manually for 3 seconds (s) in DPPH·methanol solution (0.2%) and then photographed every 30 s for 15 min. Images of the plates were captured with mobile phone (Huawei P Smart) equipped with a 13-pixels camera. All developed plates were photographed both before and after derivatization and saved as TIF files.

Images of the HPTLC chromatograms were analyzed using free available Image J software. The obtained results for each sample were cropped and denoised by using median filter with three pixels width filter. Further, images were transformed and the tracks were outlined with a rectangular selection tool. The line profile plots were generated with Plot Profile option for each sample. Profile plot displays a 2-D graph of the intensities of pixels along a line.

#### *2.4. Principal Component Analysis*

The line profiles were obtained using ImageJ software [28]. Principal Component Analysis (PCA) was applied using PLS ToolBox, v.6.2.1 (Eigenvector Research, Inc. 196 Hyacinth Road Manson, WA 98831, USA), for MATLAB (7.12.0(R2011a) (http://www.eigenvector.com/software/pls\_toolbox.htm, Eigenvector Research, Inc., Wenatchee, WA). The data were pre-processed using correlated optimized warping (COW), standard normal variate (SNV) and mean centering to improve multivariate models.

## *2.5. Bacterial Strains and Growth Conditions*

Antibacterial activity was tested using eight indicator strains in line with their growth requirements (Table 2). Suspensions were adjusted to McFarland standard turbidity (0.5) (BioMérieux, Marcy-l'Étoile, France), which corresponds approximately to 1 <sup>×</sup> 108 CFU mL<sup>−</sup>1.


**Table 2.** Indicator strains used in testing antimicrobial activity of selected extracts from forest trees.

\* Strains isolated from the human oral cavity [29]. All reference strains belong to Department of Microbiology, Faculty of Biology, University of Belgrade.

#### *2.6. Well-Di*ff*usion Method*

A modified well-diffusion method [30] was performed for initial screening of the antimicrobial potential of the selected 11 wood waste extracts. Wells were made of sterile bottom parts of pipette tips (200 μL) and placed on the LA/BHA/TSA solid medium (Table 2). According to growth requirements of used strains, 6 mL of LA/BHA/TSA soft agar was inoculated with 60 μL of the appropriate strain and poured into Petri dishes over the solid medium. Molds (5 mm in diameter) were removed after soft agars solidification and 20 μL of each extract (1 mg/well) was added. Vancomycin and nystatin were used as a positive control (antibiotic/mycotic, viz., 0.2 mg/well), for bacterial strains and *C. albicans*, respectively. As a negative control, 20 μL of methanol was used. The Petri dishes were incubated at 37 ◦C, for 24 h. After the incubation, bacterial susceptibility and zones of inhibition were measured and expressed in mm.

## *2.7. MIC Assay*

A broth microdilution method was used to determine the minimum inhibitory concentration (MIC) and minimum bactericidal concentrations (MBC) of the selected 11 wood waste extracts. Extracts were tested in the concentration range from 0.02 to 2 mg mL−<sup>1</sup> by performing two-fold serial dilutions with the appropriate medium in 96-well microtiter plates. Negative control (control of bacterial and yeast growth) and sterility control (blank, only appropriate medium) were also tested. The final concentration of the solvent control (methanol) in the first wells was 10%. Vancomycin, streptomycin and nystatin were tested as positive controls in concentration range from 0.001 to 0.4 mg mL<sup>−</sup>1. Beside negative and sterility controls, each well was inoculated with 20 μL of bacterial/yeast culture (approx. 1 <sup>×</sup> 106 CFU mL−1), reaching a final volume of 200 <sup>μ</sup>L. In addition, 22 <sup>μ</sup>L of resazurin indicator was added to each well. Microtiter plates were incubated for 24 h at 37 ◦C. In the presence of living bacterial cells, blue colored resazurin was being irreversibly reduced to pink colored and highly red fluorescent resorufin [31]. The lowest concentration of each extract which showed no change in color of resazurin was defined as MIC value. MBC/MFC values were determined by sub-culturing the dilutions from wells without color changes on agar plates. Plates were incubated 24 h at 37 ◦C and bacterial/yeast growth was monitored. The lowest concentration without growth was defined as MBC/MFC value. The results were expressed in mg mL<sup>−</sup>1.

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

#### *3.1. Line Profiles of Investigated Extracts*

Investigated wood samples contained several characteristic phenolic compounds at R*<sup>F</sup>* values of: 0.28, 0.35, 0.43, 0.74, 0.86 (gallic acid), 0.91 (ferulic acid), 0.88 (caffeic acid) and 0.91 (*p*-coumaric acid) (Figure 1). Based on HPTLC profiles, wood extracts contained bands with R*<sup>F</sup>* values from 0.28 to 0.91. There are five different patterns in the investigated samples: *Quercus* samples showed one band of weak intensity with R*<sup>F</sup>* value at 0.85, while sample 1 (Pedunculate oak—*Q. robur* L.) had the highest intensity peak of this compound. Further, black locust (sample 8) showed greenish bands with R*<sup>F</sup>* at 0.75, 0.82 and 0.87, clearly different from standard phenolic acids (Figure 2a).

**Figure 1.** Line profiles of investigated wood extracts based on HPTLC analysis: (**A**) oak samples (no. 1–7); (**B**) non-oak samples (no. 8–11): black locust (*Robinia pseudoacacia* L.) (no. 8), myrobalan plum (*Prunus cerasifera* Ehrh.) (no. 9), wild cherry (*Prunus avium* L.) (no. 10) and mulberry (*Morus alba* L.) (no 11).

**Figure 2.** HPTLC chromatograms of samples: *Q. robur* (no. 1–3), *Q. petraea* (no. 4–6), *Q. cerris* (no. 7), *Robinia pseudoacacia* (no. 8), *Prunus cerasifera* (no. 9), *Prunus avium* (no. 10), mulberry (no. 11) and four standard compounds (gallic acid (no. 12), ferulic acid (no. 13), caffeic acid (no. 14) and *p*-coumaric acid(no. 15)); (**a**) under UV light at 366 nm; (**b**) under UV light at 254 nm; (**c**) DPPH-HPTLC chromatogram.

Wild cherry contained one characteristic peak at 0.36, whereas myrobalan plum had a different pattern from other wood samples with three characteristic peaks at R*<sup>F</sup>* values of 0.28, 0.35 and caffeic acid (Figure 1). A different profile of myrobalan plum in comparison with other wood samples could be seen also by HPLC [6], where it contained significantly larger amounts of protocatechuic acid and 5-O-caffeoylquinic acid than other wood samples. The peak profiles for mulberry showed it contained peaks at 0.43, 0.84, gallic, ferulic and/or *p*-coumaric acids.

Mulberry sample showed hardly visible blue band with R*<sup>F</sup>* at 0.86, recognized as gallic acid, while black locust and oak wood samples contained gallic acid in higher amounts. In addition, wild cherry contained *p*-coumaric and ferulic acid in greater quantities than mulberry, and caffeic acid in greater quantities than myrobalan plum, which was observed neither on HPTLC plates nor line profiles.

## *3.2. DPPH-HPTLC Assay*

It was previously shown that the total antioxidant activity of each extract through the DPPH assay, mulberry and myrobalan plum wood extracts had significantly higher DPPH values in comparison to the other samples [6]. The single compounds with radical scavenging activity and their contribution to the total radical scavenging activity were investigated by DPPH. -HPTLC assay. Substances exhibiting radical scavenging properties (yellow bands against a purple background) were located between R*<sup>F</sup>* values at 75 and 92 (Figure 2c). The most dominant zones in the HPTLC-DPPH. fingerprints were compounds with R*<sup>F</sup>* values at 0.87, 0.91 and 0.92, which could be recognized as gallic, ferulic and/or caffeic acids (zone 4). Extracts no. 8–11 showed strong radical scavenging activities, mainly due to the previously detected phenolic compounds, while *Quercus* samples revealed one weak band with R*<sup>F</sup>* at 75 against the purple background. *P. cerasifera* contained two bands at 0.84 and caffeic acid, and were recognized as radical scavengers. The, *M. alba* sample showed radical scavengers with R*<sup>F</sup>* values at 0.84, gallic, ferulic and/or *p*-coumaric acids. These compounds have been recognized before as strong radical scavengers on silica plates [16].

## *3.3. Principal Component Analysis*

Visual inspection of HPTLC chromatograms is a subjective method and mainly depends on the analyst's perception. On the other hand, multivariate chemometrics analysis applied on the HPTLC chromatogram provides an objective classification of the investigated samples for identification of phenols most responsible for classification, as well as identification of outliers. The HPTLC system was optimized to separate and identify all phenols from different wood extracts.

Principal component analysis (PCA) is a commonly used multivariate technique. It accounts for most of the variation of total variability, visualizes the structure of data by grouping objects into two or three dimensions, and identifies important variables responsible for discrimination between wood samples. PCA as an initial multivariate technique was applied on the data matrix (11 samples × 389 variables) obtained from HPTLC chromatograms, where variables represent the intensities of pixels along the 389 length lines. The first two Principal Components (PCs) accounted for 33.35% and 20.09% of the total variability, respectively. The first five principal components describe 87.85% of total variability. From the PC score plot (Figure 3a), six *Quercus* samples were positioned on left side of PC score plot, while other four wood samples were misclassified and positioned on right side of PCs score plot. The loading plots (Figure 3b,c) demonstrated the significant contribution of polyphenolic compounds to the total variability. The most influential phenolic compounds discriminating between *Quercus* and the other wood samples were compounds with R*<sup>F</sup>* values at 0.35, 0.43, 0.86 and 0.91. In contrast to other types of wood samples *Quercus* samples contained low amounts of phenolic compounds with R*<sup>F</sup>* values at 0.35, 0.43, 0.86 and 0.91. Polar compounds with low R*<sup>F</sup>* values could be some phenolic acids and/or glycosides. These phenolic compounds may be identified as characteristic taxonomical markers between wood species.

**Figure 3.** Principal component analysis (PCA) of HPTLC chromatogram: (**A**) The PC score plot; (**B**) and (**C**) The loading plots. 1–3 Pedunculate oaks (*Quercus robur* L.), 4–6—sessile oaks (*Quercus petraea* (Matt.) Liebl), 7—Turkey oak (*Quercus cerris* L.).

## *3.4. Well-Di*ff*usion Method*

Antimicrobial potential of the extracts was tested against eight representative human and opportunistic pathogens. Besides clear zones of inhibition, bacteriostatic/fungistatic effect of tested extracts was also observed. Wood waste extracts in general showed the highest antimicrobial potential against *S. mutans, S. pyogenes* and *L. monocytogenes* strains in tested concentration of 1 mg/well (Figure 4). The wild cherry extract (10) inhibited the growth of *S. mutans* and *S. aureus* yielding the largest zones of inhibition (21.7 and 19.8, respectively), compared to other extracts, towards to the mentioned pathogens. Additionally, only wild cherry and mulberry extracts (10,11) showed moderate bactericidal effect against *E. faecalis*. This indicator strain due to its higher resistance was excluded for further MIC testing. Mentioned extracts also showed high bacteriostatic effect against MRSA. Additionally, the wild cherry extract showed clear bactericidal effect only against *C. albicans* and *L. monocytogenes*, while other extracts acted more bacteriostatically. On the other hand, other wood extracts showed overwhelmingly bacteriostatic/fungistatic effect against almost all pathogens, including *E. coli*. All pathogens were susceptible to tested vancomycin and nystatin mycotic.

**Figure 4.** Antimicrobial potential of wood waste extracts in well-diffusion method. \*V/N—Vancomycin/ Nystatin. Values within columns represent a mean of inhibition zones and expressed in mm. *Q. robur* (no. 1–3), *Q. petraea* (no. 4–6), *Q. cerris* (no. 7), *Robinia pseudoacacia* (no. 8), *Prunus cerasifera* (no. 9), *Prunus avium* (no. 10), mulberry (no. 11).

#### *3.5. MIC Assay*

For evaluation of new antimicrobials, the assessment of minimum inhibitory concentration (MIC) is usually the first step [27]. The MIC is the minimum concentration that causes visible inhibition of bacterial growth. Plant extracts with MIC < 100 μg mL−<sup>1</sup> and purified compounds with MIC < 10 μg mL−<sup>1</sup> are considered promising [27]. However, bactericidal activity, determined by MBC value in time-kill assays, is also an important parameter in assessing the antimicrobial activity. MBC and MIC parameters complement each other, and MBC below four times MIC value suggests the bactericidal action of a tested compound [27].

The obtained MIC values were in range from 0.02 mg mL−<sup>1</sup> of extract 11, to MRSA to 2 mg mL−<sup>1</sup> in the case of activity of extract 9 (myrobalan plum) against *C. albicans* (Table 3).


**Table3.**Minimuminhibitory,minimumbactericidalandminimumfungicidalconcentrations(MIC/MBC/MFC)of11woodextractstowardsselectedpathogens

(red). *Q. robur* (no. 1–3), *Q. petraea* (no. 4–6), *Q. cerris* (no. 7), *Robinia pseudoacacia* (no. 8), *Prunus cerasifera* (no. 9), *Prunus avium* (no. 10), mulberry (no. 11).

The lowest MIC values (viz., 0.03 mg mL−1) were recorded against MRSA (extracts 5, 7 and 8), *S. aureus* (extracts 4–7) and *S. pyogenes* (extracts 1, 3, 5, 9–11). *S. mutans* also showed high sensitivity to some of the tested extracts with MICs below 0.2 mg mL−1. MIC values for *L. monocytogenes* were in range from 0.03–0.75 mg mL−1, while extracts 9–11 significantly inhibited the growth rate of this pathogen. Compared to Gram-positive isolates, *E. coli* was less sensitive to the tested extracts. *Candida albicans* showed poor sensitivity to the action of all extracts, with the exception of extract 10 with obtained MIC value of 0.25 mg mL−1. Alañón et al. [17] also concluded that yeasts had a stronger resistance to wood extracts than bacteria, since only toasted American oak wood and wild cherry wood extracts inhibited their growth. MICs for vancomycin, streptomycin and nystatin were lower compared to the tested extracts (0.001–0.4 mg mL−1). Additionally, MRSA showed resistance to all antibiotics on the highest concentration tested (0.4 mg mL<sup>−</sup>1). Interestingly, non-seasoned sessile oak (sample 5) showed lower MIC against MRSA and *L. monocytogenes* than seasoned oaks (samples 1, 2, 3, 4, 6).Comparing the results for *Q. robur* with the results for oak bark (*Q. robur*) [24], higher values for MIC were found against *L. monocytogenes* and *E. coli*, but lower values against *S. aureus*. In addition, the values of MIC for streptomycin were significantly lower than Elansary et al. [24] obtained. MBC and MFC values of tested extracts varied from 0.03–2 mg mL<sup>−</sup>1. The lowest MBC was recorded against *S. aureus* for extracts 2, 5, 8, 10 and 11.

There was a strong simultaneous activity against all pathogens tested of extract 10 from the wild cherry wood. This could be explained by its richness of phenolic compounds which was observed in previous research [6]. For example, kaempferol is a potential candidate against different pathogenic microbes, and effective against fluconazole-resistant *Candida albicans* and Methicillin-resistant *S. aureus* (MRSA) [22]. In addition, galangin exhibited selective anti-cytochrome and antifungal activity [25], and showed antimicrobial activity against *S. aureus* [25,32], and methicillin-sensitive and methicillin-resistant *S. aureus*, *Enterococcus* spp., and *P. aeruginosa* [33]. Flavone apigenin showed strong activity against Gram-negative bacteria [34], while quercetin and apigenin derivatives showed strong antibacterial properties against Gram-negative and Gram-positive bacteria [35]. Some phenolic acids (gallic, caffeic and ferulic acids) showed antibacterial activity against Gram-positive (*S. aureus* and *Listeria monocytogenes*) and Gram-negative bacteria (*E. coli* and *P. aeruginosa*) with a greater efficiency than conventional antibiotics such as gentamicin and streptomycin [36]. Contrarily, chlorogenic acid, which was not abundant in wood species, showed no activity against Gram-positive bacteria [36]. Interestingly, noticeable antimicrobial activity of cherry wood against wine organisms was observed before [17], but, to our knowledge, its antimicrobial activity against human and opportunistic pathogens has not been investigated so far.

However, MIC values for extracts 1–7 against *S. aureus*, *L. monocytogenes* and *E. coli* were similar to MICs obtained from some other *Quercus* spp. bark extracts (Table 4), but MICs recorded towards *C. albicans* were lower compared to the results of this study.


**Table 4.** Summarized MICs values for other waste extracts obtained from literature data.

*Sa*—*S. aureus*; Mr—MRSA; *Lm*—*L. monocytogenes*; *Sm*—*S. mutans*; *Sp*—*S. pyogenes*; *Ec*—*E. coli*; *Ca*—*C. albicans*; (-)—not tested.

On the other hand, *P. avium* stem bark extracts from Nigeria showed lower antimicrobial activity against *S. aureus* and *E coli*, with MICs 6.25 mg mL−<sup>1</sup> and 12.5 mg mL<sup>−</sup>1, respectively [43]. Compared to extracts 9 and 10, *Prunus cerasoides* showed similar antibacterial activity towards MRSA [44]. Unlike extract 11, originating from *M. alba*, bark extracts originating from *Morus mesozygia* showed significant antimicrobial activity against *C. albicans*, with obtained MIC of 0.16 mg mL−<sup>1</sup> (Table 4). Interestingly, higher susceptibility of *C. albicans* was also observed for *Picea abies* and *Larix decidua* bark extracts [37]. In the literature, no significant correlation was found between antimicrobial activity and total phenolic content [17,46], as well as between antimicrobial activities and antioxidant capacity. However, structure-function of the phenolic extracts have more influence on the antimicrobial activity than the total phenol content [17]. Finally, according to Cowan [47], a wide variety of specialized metabolites show antimicrobial activity *in vitro*, such as tannins, terpenoids and alkaloids, also found in wood.

## **4. Conclusions**

Wood waste from forest trees is a source of different bioactive metabolites which could find application in the food and pharmaceutical industries. Radical scavenging and antimicrobial activities of the wood waste extracts appeared to be a valuable bio-functional source. In general, HPTLC fingerprint identify the main phenolic acids present in investigated samples and revealed chemical patterns among investigated wood extracts. DPPH-HPTLC assay identified gallic, ferulic and/or caffeic acids as compounds with the highest contribution to total radical scavenging activity. Based on PCA plot, six *Quercus* samples were separated from other extracts showing strong radical scavenging activities.

Wood samples were the most active against MRSA, *S. aureus* and *S. pyogenes*. The lowest MIC and MBC values were detected in mulberry extract against MRSA. Activities were also distinguished against MRSA (extracts of non-seasoned sessile oak (5), Turkey oak, black locust and mulberry) and S*. aureus* (Turkey oak and all sessile oak extracts). The largest zones of inhibition of the growth of *S. mutans* and *S. aureus* were observed for wild cherry extract. Among sessile and pedunculate oak extracts, non-seasoned sessile oak extract (5) was distinguished by lower MIC against MRSA and *L. monocytogenes*. Extracts of myrobalan plum, wild cherry and mulberry significantly inhibited the growth rate of *L. monocytogenes*. *E. coli* was less sensitive to the tested extracts. *C. albicans* showed poor sensitivity to the action of all extracts, with the exception of the wild cherry extract.

Wild cherry wood extract can be commercially important due to good simultaneous activity against all pathogens, and is a valuable source for various formulations: Wild cherry and mulberry wood extracts with given antimicrobial activities can be especially useful in preserving perishable foods with short shelf life.

**Author Contributions:** Conceptualization, M.N. and P.R.; methodology, P.R.; software, D.D.Z.; investigation, P.R., A.S., I.D. and T.P.; resources, S.V.; writing—original draft preparation, A.S., P.R. and T.P.; writing—review and editing, M.N., M.F.A. and M.M.; supervision, M.F.A. and M.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ministry of Education, Science and Technological Development, Republic of Serbia, grant number 172017, and Research Council of Norway (project No 11060-"NORWEGIAN FRUIT GENETIC RESOURCES—HEALTHY, TASTE & NO WASTE").

**Acknowledgments:** Authors would like to thank Ivanka Ciri´ ´ c for technical assistance during the course of analysis.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

## **References**

1. Fernández de Simón, B.; Sanz, M.; Cadahía, E.; Martínez, J.; Esteruelas, E.; Muñoz, A.M. Polyphenolic compounds as chemical markers of wine ageing in contact with cherry, chestnut, false acacia, ash and oak wood. *Food Chem.* **2014**, *143*, 66–76. [CrossRef]


© 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* **Extract from Broccoli Byproducts to Increase Fresh Filled Pasta Shelf Life**

## **Luisa Angiolillo, Sara Spinelli, Amalia Conte \* and Matteo Alessandro Del Nobile**

Department of Agricultural Sciences, Food and Environment, University of Foggia, Via Napoli, 2571121 Foggia, Italy; luisa.angiolillo@unifg.it (L.A.); sara.spinelli@unifg.it (S.S.); matteo.delnobile@unifg.it (M.A.D.N.)

**\*** Correspondence: amalia.conte@unifg.it; Tel.: +39-088-158-9240

Received: 23 October 2019; Accepted: 22 November 2019; Published: 27 November 2019

**Abstract:** The aim of the study was to evaluate the efficacy of extract from broccoli byproducts, as a green alternative to chemical preservation strategies for fresh filled pasta. In order to prove its effectiveness, three different percentages (10%, 15%, and 20% *v*/*w*) of extract were added to the filling of pasta. A shelf life test was carried out by monitoring microbiological and sensory quality. The content of phenolic compounds before and after in vitro digestion of pasta samples was also recorded. Results underlined that the addition of the natural extract helped to record a final shelf life of about 24 days, that was 18 days longer in respect to the control sample. Furthermore, results highlighted that the addition of byproducts extract to pasta also increased its phenolic content after in vitro digestion. Therefore, broccoli byproducts could be valorized for recording extracts that are able to prolong shelf life and increase the nutritional content of fresh filled pasta.

**Keywords:** shelf life; byproducts; fresh pasta; vegetable extracts; antimicrobial activity

## **1. Introduction**

Pasta is one of the main constituents of the Mediterranean diet as it contains significant amounts of complex carbohydrates, proteins, B-vitamins, and iron [1]. Pasta can be made with different kinds of flours (semolina, farina, wheat flour, etc.) mixed with water. Fresh pasta has more than 24% moisture and its water activity ranges from 0.92 to 0.99, thus it requires refrigeration [2]. It can be prepared with eggs in the dough or by filling a sheeted dough with a spiced mixture of ground meat, cheese or vegetables as for tortellini and ravioli. In the last decade, fresh filled pasta gained great national and international popularity, even though distribution beyond the Italian borders still represents a real problem due to the rapid microbial proliferation. In fact, this product is very susceptible to spoilage microorganisms and therefore, addition of preservatives or reduced oxygen packaging are necessary to prolong the shelf life that, even under refrigerated temperatures, lasts only two or three days [3]. Specifically, the pH of pasta without any preservatives may drop, thus indicating spoilage and increased coliforms [4]. Italian law [5] prescribes a pasteurization treatment before final packaging in order to reduce the growth of vegetative microbial forms and also to improve cooking behaviors. Generally, the thermal treatment is carried out in an injected steam belt pasteurizer and in addition to reducing water activity, it helps to increase starch gelatinization with consequently less water absorption during cooking [6]. For shelf life prolongation different methods have been applied to fresh pasta; the most common approach is based on chemical preservatives, as organic acids, and modified atmospheres (MAP) with low O2 concentrations (below atmospheric levels) and high CO2 concentrations (20% or higher), proper coupled with N2 as an inert gas filler [7–10]. Castelvetri et al. [11] demonstrated that packaging atmospheres with more than 30% CO2 were capable of extending fresh filled pasta shelf life, even if low residual O2 in the package headspace may cause mold growth. The search for more natural technologies of food preservation greatly promoted the exploration of antimicrobial

compounds of vegetable or animal origin, as essential oils, enzymes or chitosan [12]. Among natural antimicrobial compounds, substances recordable from leaves, flowers, seeds, and peels are becoming very interesting [13]. Many components with useful properties can be found in food byproducts [14]. The possibility to recycle food byproducts may also represent a mean to face their environmental and economic impact. Among plants, *Brassica oleracea,* that belongs to the Brassicaceae family, represents one of the most abundant byproducts producers. It comprises 3500 species such as cauliflower, broccoli, kale, cabbage, and Brussels sprouts. Byproducts from *Brassica oleracea* are rich in phenols, as flavonoids, phenolic acids, and tannins, usually extracted by solvent extraction [15]. This extraction method has low selectivity and utilizes high energy cost, elevated solvents, and high temperatures [16]. For this reason, alternative methods of extraction have been investigated. Among them, the supercritical fluid extraction (SFE) with improved selectivity, automation, and environmental safety, represents a valid alternative [17].

According to the above-reported considerations, the aim of the study was to evaluate the efficacy of broccoli byproducts extract, obtained by SFE, to improve fresh filled pasta shelf life. To this aim, bioactive substances were first extracted from broccoli byproducts and subsequently added to fresh filled pasta to verify their effects on microbial and sensory quality. The evaluation of the consequent polyphenols content in pasta samples was also assessed.

## **2. Material and Methods**

## *2.1. Raw Materials*

Broccoli stems and leaves (*Brassica oleracea*) were provided by a local company in Foggia, Southern Italy. The samples were dried at 30–35 ◦C in a dryer (SG600, Namad, Rome, Italy) for 48 h. The dried samples were reduced to fine powder (≤250 μm) by a hammer mill (16/BV-Beccaria s.r.l., Cuneo, Italy) and then stored at 4 ◦C until further utilization.

## *2.2. SFE*

Supercritical fluid extraction was carried out to collect active compounds from byproducts, above all polyphenolic compounds. It was performed using process conditions (150 bars, 35 ◦C, 20% ethanol and 10 min of dynamic extraction time) previously described by Arnáiz et al. [18], by the supercritical fluid extractor Speed SFE-2 (Applied Separation, Allentown, USA). The extract was placed overnight in vacuum oven at 30 ◦C to remove ethanol. The solid residue was collected in 25 mL of water.

## *2.3. Fresh Filled Pasta Production*

Fresh pasta samples were produced with durum semolina (provided by Agostini mill Montefiore dell'Aso, Ascoli Piceno, Italy). Semolina and distilled water (30% *v*⁄*w*) were mixed for about 20 min to prepare the pasta dough. The samples were prepared using a pilot scale extruder (60VR; Namad, Rome, Italy) equipped with a roller sheeter (Raff, Minipan, Massa Lombarda, Italy) and a compressor (mod. Rondostar, Rondo Doge, Burgdorf, Switzerland) in order to obtain a 6 mm sheeted dough. The filling was prepared by mixing 65% (*w*/*w*) ricotta cheese, 19% (*w*/*w*) grated cheese, 16% (*w*/*w*) fresh spinach, 0.30% (*w*/*w*) potato flour, and 0.01% (*w*/*w*) salt. The sheeted dough and filling were combined in a modified double sheet ravioli machine (mod. PRP 300, Genoa, Italy) to prepare 12 cm diameter fresh pasta samples in the form of ravioli. Each sample consisted of two square dough sheets containing the filling. Four different formulations were prepared: fresh filled pasta without any addition (CNT), and the other three samples with increasing concentrations of broccoli extract in the filling: 10% (*v*/*w*) (10-BE); 15% (*v*/*w*) (15-BE), and 20% (*v*/*w*) (20-BE). The product was conveyed through a 3-m chamber equipped with a perforated steel conveyor belt (Custom, Italgi, Genoa, Italy). By steam injection at 91 ± 1 ◦C for 9 min, the pasteurization was carried out. After pasteurization the product passed through two fans to eliminate the condensed vapor on the surface and then was cooled to 4 ◦C and packaged in bags with anti-fog high-barrier multilayer film made up of polyethylene

terephthalate, ethylene-vinyl alcohol, and polyethylene. The film oxygen transmission rate (OTR) was 6.19 cc/m2/day, the water vapor transmission rate (WVTR) was 1.208 g/m2/day, and the thickness was 50 μm (Di Mauro Officine Grafiche spa, Salerno, Italy). All the samples were stored for about 2 months at 4 ◦C, without light.

## *2.4. Microbiological Analyses*

For microbiological analyses, about 10 g of sample were aseptically removed from each package, placed in a stomacher bag, diluted with 90 mL of sterile NaCl solution, and homogenized with a stomacher LAB Blender 400 (Pbi International, Milan, Italy). Serial dilutions in sterile saline solution were plated onto appropriate media. The media and conditions were the following: plate count agar (PCA) incubated at 30 ◦C for 48 h for aerobic mesophilic bacteria and at 7 ◦C for 10 days for psychrotrophic bacteria; Violet Red Bile Glucose Agar (VRBGA) incubated at 37 ◦C for 24 h for *Enterobacteriaceae*; Baird-Parker Agar, supplemented with egg yolk tellurite emulsion, incubated at 37 ◦C for 48 h for *Staphylococcus* spp.; Sabouraud Dextrose Agar, added with 0.1 g/L chloramphenicol (C. Erba, Milan, Italy), incubated at 25 ◦C for 48 h for yeasts and 25 ◦C for 5 days for molds. Reinforced Clostridial Medium (Oxoid, Milan, Italy) was used for the sulfite-reducing clostridia; after heat treatment of samples at 80 ◦C for 10 min to destroy the vegetative cells, the plates were incubated at 37 ◦C for 48 h in anaerobic conditions, thus avoiding contact with air. The count was carried out with the most probable number (MPN) method. Aerobic spore-forming bacteria were detected and counted on Nutrient Agar (Oxoid, Milan, Italy) after 48 h at 30 ◦C; all vegetative forms were previously destroyed by heat treatment of samples at 80 ◦C for 10 min. All media and supplements were from Oxoid (Milan, Italy). All microbiological analyses were performed twice on two different samples (one sample from two different trays). In order to quantitatively determine the microbial acceptability limit (MAL), a modified version of the Gompertz equation was fitted to the experimental data, as reported in previous studies [19,20]. The Italian law [5] fixes the threshold for total microbial count (TMC), staphylococci, and clostridia at maximum values of 106, 5 <sup>×</sup> <sup>10</sup>3, and 103 Colony Forming Unit (CFU)/g, respectively.

#### *2.5. Sensory Analysis*

During the entire storage period, at selected times, both uncooked and cooked fresh pasta samples were subjected to a time intensity evaluation. Towards the aim, eight trained tasters were involved in the panel test. The panelists were asked to evaluate color, odor, and overall quality of uncooked samples and color, odor, taste, consistency, and overall quality of pasta cooked in food grade tap water at 100 ◦C. A nine-point rating scale, where 1 corresponded to 'extremely unpleasant' and 9 to 'extremely pleasant', was used to perform the panel test. [21]. The panelists were selected on the basis of their sensory skills (ability to accurately determine and communicate the sensory attributes, the appearance, odor, flavor, and texture). Prior to testing pasta, the panelists were trained in the sensory vocabulary and identification of particular attributes, by using commercial pasta. The analyses were performed in isolated booths, located in a standard taste panel kitchen. In order to determine the sensory acceptability limit (SAL), intended as the storage time to reach the sensory threshold, a modified version of the Gompertz equation was fitted to the sensory data [19,20]. The sensory threshold was set equal to 5.

## *2.6. Chemical Analyses*

#### 2.6.1. Extraction of Polyphenols from Cooked Pasta Samples

The extraction of polyphenols from both control and enriched pasta samples was based on the method also described by Rashidinejad et al. [22]. Briefly, 1 g of each cooked sample was homogenized and extracted in a water bath with 50 mL of 95% methanol containing 1% HCl at 50 ◦C and 200 rpm. The mixture was cooled, filtered, and washed with 2 mL of the same solvent.

## 2.6.2. In Vitro Digestion of Cooked Pasta Samples

Simulated gastric and intestinal digestions were carried out on both control and enriched pasta samples using the method of Rashidinejad et al. [22]. In brief, 1 g of each cooked sample was added with 10 mL of simulated filtered gastric fluid (SGF) at 37 ◦C and incubated in an orbital shaker at 37 ◦C at 235 rpm for 10 min. After adjusting the pH of the solution to 2.0, the treatment continued for a further 2 h at 95 rpm. Then, 36 mL at 37 ◦C of simulated intestinal fluid (SIF) were added to each gastric digestion sample and stirred at 37 ◦C at 95 rpm for 4 h. After 30 min from the beginning, the pH was adjusted to 6.8. In order to prepare the SGF sample, 2 g of NaCl, 7 mL of HCl (36%), and 3.2 g of purified porcine pepsin (in 1 L of deionized water, pH 1.2) were used, while to prepare the SIF sample a monobasic potassium phosphate solution (6.8 g in 250 mL of deionized water) with 77 mL of sodium hydroxide (0.2 M) and 500 mL of deionized water was mixed. Finally, 10 g of pancreatin and 0.05 g of porcine bile extract were added to the mixture and the pH was adjusted again to 6.8. For each sample, the digestion was carried out in triplicate.

#### 2.6.3. Total Phenolic Content

To measure the total phenolic content (TPC) in all the undigested and digested samples, the Folin–Ciocalteu assay was used. TPC was determined as described by da Silva et al. [23] with slight modifications. Briefly, 0.5 mL of sample (that obtained in Section 2.6.1 and that recorded in Section 2.6.2) and 2.5 mL of Folin–Ciocalteu reagent diluted in water (1:10 ratio) were left to rest for 5 min. An amount of 2 mL of Na2CO3 (4 g/100 mL) was then added. The mixture was allowed to rest again for 2 h in darkness. The absorbance was read by a spectrophotometer (UV1800, Shimadzu Italia s.r.l.) at 740 nm. Total phenols content was expressed as mg of gallic acid equivalents (GAEs) per g of pasta, according to a previously recorded calibration curve. For each sample, the analyses were carried out in triplicate.

#### *2.7. Statistical Analysis*

Experimental data were compared by one-way ANOVA analysis. A Duncan's multiple range test, with the option of homogeneous groups (*p* < 0.05), was used to determine significance among differences. To this aim, Statistica 7.1 for Windows 152 (StatSoft Inc., Tulsa, OK, USA) was used.

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

#### *3.1. Total Phenolic Content*

TPC of both undigested and digested pasta samples is shown in Table 1. As can be seen, the TPC in the indigested control samples (0.63 mg GAEs/g) was significantly (*p* < 0.05) lower than that found in pasta enriched with broccoli extract (1.84–1.86 mg GAEs/g), even if it did not increase linearly with the quantity of added extract. This finding may have been due to the interaction of polyphenols with ricotta proteins and the concomitant formation of less active complexes [24]. Gallo et al. [25] also stated that milk protein fractions caused a decrease of the in vitro antioxidant activity of polyphenols, as a consequence of the weaker non-covalent bonds between proteins and polyphenols.

The other information that can be deduced looking at Table 1 is that the phenolic content of the digested samples appear to be higher in respect to the indigested ones, with the highest value found in the 20-BE sample (2.61 mg GAEs/g).

The explanation for this trend could be linked to the hydrolyzation exerted by digestive enzymes towards chemical bonds in the phenolic-protein complexes, promoting in this way a greater release and extractability of phenolics at the in vitro level [26,27]. In particular, Gumienna et al. [28] claimed that the action of gastric-digestive enzymes may lead to the development of aglycones phenolic compounds, more reactive than the corresponding glycoside forms. These findings positively encourage the consumption of pasta enriched with broccoli byproducts extract, as a way to promote intake of valuable food for human health, since it is widely demonstrated that phenolic compounds provide extraordinary anticancer, antiviral, antibacterial, cardio-protective, and anti-mutagenic activities [29].


**Table 1.** Total phenolic content (TPC) of undigested and digested cooked ravioli samples.

a–c Data in columns with different letters are significantly different (*p* < 0.05). GAEs: gallic acid equivalents; CNT = fresh filled pasta without any addition; 10-BE = fresh filled pasta with 10% broccoli extract; 15-BE = fresh filled pasta with 15% broccoli extract; 20-BE = fresh filled pasta with 20% of broccoli extract.

#### *3.2. Quality of Fresh Filled Pasta*

Figure 1 describes the growth of total mesophilic bacteria in all the experimental samples. Table 2 reports values of fitting parameters. As can be inferred from the data, in the CNT pasta an immediate growth with an ascendant trend was found, with values from 4.88 log CFU/g to 8.77 log CFU/g until the 22nd day.

**Figure 1.** Evolution of mesophilic bacteria in pasta samples during storage at 4 ◦C. CFU: Colony Forming Unit.

In the control pasta the microbial acceptability limit was reached after six days. On the contrary, the three samples with increasing concentration of broccoli extract, revealed a positive microbial quality during the entire observation period, without significant difference among them. These data were similar to the microbial trend of psychrotrophic bacteria.

Regarding *Staphylococcus* spp., counts were found below the microbial limit for the entire 25 days of observation in all the active pasta samples, while in the CNT microbial growth started from the 15th day and reached the limit after 20 days of storage (Figure 2). Data of MAL are also reported in Table 2.

**Figure 2.** Evolution of *Staphilococcus* spp. in pasta samples during storage at 4 ◦C.

The addition of broccoli extract appeared to be also effective against *Enterobacteriacae*, molds, and yeasts as these microbial groups in the enriched pasta did not grow until the 25th day (data not shown). On the contrary, the CNT sample revealed molds growth, both visible and on plates, at the 13th day (Table 2), *Enterobacteriacae* and yeasts around 5.75 log CFU/g and 5.59 log CFU/g after 20 days. The trend of control samples is not surprising because it is in accordance with other studies on fresh filled pasta [10–12]. Clostridia and aerobic spore forming bacteria were never found in any samples (data not shown).

The results of the microbial quality confirm findings of other authors about the antimicrobial properties of vegetal extracts [30–32], even though the studies were all carried out under in vitro conditions. The application carried out in the current study, not only assessed the potential effects of extract from broccoli byproducts, but also demonstrated that it is possible to significantly extend microbial stability (Table 2). These data appeared to be of particular importance since the literature highlighted that fresh filled pasta packaged in ordinary atmosphere and stored at 4 ◦C generally reached very short shelf life values, accounting for hours to one week, depending on the hygienic production conditions [12]. Another import consideration is that the antimicrobial effect is not determined by the quantity of extract added to the experimental samples. As a fact, looking at Table 1, polyphenols appeared to be similar among samples with broccoli extract. It has been suggested by different authors that the bioactivity of polyphenols is rather related to their structure than to their quantity, with differences from one polyphenol to another [33–35]. Their structure is also dependent on temperature and, in particular, heat treatment [36].

According to the above-mentioned results, the microbiological acceptance of pasta was limited to about six days of storage for the CNT sample and lasted about 24 days for the active samples (Table 2).

Sensory evaluation was carried out for more than one month to know when the product became unacceptable for undesired sensory changes. Specifically, color evaluation allowed recording different results among samples, depending on the extract concentration added to the filling, because extract addition modified the filling color from a whitish green of the CNT to an intense green of the 10-BE and 15-BE ravioli samples, to a too dark green of the 20-BE pasta sample. Therefore, while the first two active samples were accepted for a long period, the 20-BE sample was refused within about two weeks.

Regarding pasta consistency, a gradual hardening of all the samples was found, an inevitable consequence of both pasteurization and storage. As described by [11] pasteurization may influence the texture of fresh filled pasta. In fact, during the thermal treatment hardening takes place due to a different distribution of water within the matrix. The increase in water–starch bonding results in a decrease of available water for the other components of the matrix, thus influencing protein denaturation and starch gelatinization. The addition of the extract helps to obtain a more hydrated structure, in fact, the CNT sample revealed a harder consistence in respect to the other three samples. The consistence of the 10-BE sample was found better than that of 15-BE and 20-BE samples because these two types of ravioli revealed an excessive liquid consistence of the filling. The taste of enriched pasta was perceived acceptable by the panelists, even if they underlined increasing the quantity of broccoli extract, the bitterness of the product increased.

**Table 2.** Shelf life (day) of pasta samples during storage at 4 ◦C, calculated as the lowest value between microbial acceptability limit (MAL) and sensory acceptability limit (SAL) (mean ± SD).


a–d Data in columns with different letters are significantly different (*p* < 0.05). VMT = visible molds time (day); CNT = fresh filled pasta without any addition; 10-BE = fresh filled pasta with 10% broccoli extract; 15-BE = fresh filled pasta with 15% broccoli extract; 20-BE = fresh filled pasta with 20% of broccoli extract.

The overall quality of both uncooked and cooked samples reflected the above discussed trends of specific sensory parameters without great differences between uncooked and cooked products. As an example, Figure 3 reports the trend of overall quality of uncooked pasta. All SAL values are reported in Table 2. It is evident looking at Figure 3 and at the sensory data of Table 2 that the CNT sample appeared to be the less appreciated, being rejected after about 13 days, due to undesired changes in color and consistence. The 20-BE samples were refused within 15 days of storage above all for the undesired color, 15-BE samples remained acceptable for less than one month when uncooked and more than one month when considered after cooking, whereas the 10-BE samples recorded the highest score before and after cooking, with sensory acceptability accounting for more than 40 days.

**Figure 3.** Sensory quality of uncooked fresh filled pasta during storage at 4 ◦C.

Taking into account both MAL and SAL values of Table 2, the shelf life was reported as the lowest value among them. From the results it is possible to highlight that excessive proliferation of mesophilic bacteria provoked in most cases the end of the product shelf life, whereas in the case of the pasta with the highest concentration of extract (20-BE) the main problem was the presence of unacceptable sensory defects. The effects of the extract were very tangible because while the CNT sample was refused after a few days, the two active samples 10-Be and 15-BE remained acceptable for more than 20 days because the extract controlled microbial proliferation and delayed undesired sensory changes.

## **4. Conclusions**

Reuse of food byproducts in a sustainable way may be a possible way to reduce environmental impact and face costs related to their disposal. In this study extract from broccoli byproducts was added for the first time to the filling of fresh pasta to improve quality and prolong shelf life. Results underlined that the addition of broccoli extract helped to record a final shelf life of about 24 days, that was 18 days longer in respect to the control sample. In addition, pasta with broccoli extract showed a higher phenolic content in respect to the free samples, particularly after digestion. The most appropriate extract amount was 10%, with the pasta samples appreciated for color, taste, and consistency.

**Author Contributions:** Data collection: L.A. and S.S.; planning of the work and supervision: A.C. and M.A.D.N.; writing—review and editing: A.C.

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

**Conflicts of Interest:** The authors declare no conflicts 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*
