*Article* **Chemical, Functional, and Technological Features of Grains, Brans, and Semolina from Purple and Red Durum Wheat Landraces**

**Afef Ladhari 1, Giandomenico Corrado 2, Youssef Rouphael 2, Francesca Carella 3, Giuseppina Rita Nappo 4, Cinzia Di Marino 5, Anna De Marco <sup>6</sup> and Domenico Palatucci 3,\***


**Abstract:** A main reason of the increasing interest in cereal landraces is their potential to offer more diversified and functional staple food. For instance, landraces are an underexploited resource of pigmented varieties, appreciated for the high accumulation of phytochemicals with known health benefits. This study characterized the chemical, functional, and technological features of the bran, semolina, and grains of two durum wheat (*Triticum turgidum* L. subsp. *durum*, Desf.) landraces, named 'Purple' and 'Red' for their grain color, collected in Ethiopia and grown and sold in southern Italy as a niche product. Specifically, we analyzed the protein content, dry gluten, ash, total polyphenols, anthocyanins, proanthocyanidins, and specific phenolic acids. We also evaluated the antioxidant activity using DPPH- and ABTS-based methods. The two landraces had positive nutritional features, such as a high protein content, a rich and composite range of secondary metabolites (which include specific phenolic acids and anthocyanins), and antioxidant activities in all the fractions analyzed. The germplasm under investigation therefore has a well-justified potential to yield functional products and to diversify durum wheat-based foods.

**Keywords:** durum wheat; diversity; pigmented cereals; phytochemicals; anthocyanins; antioxidant activity; protein; gluten

#### **1. Introduction**

Wheat is one of the first domesticated cereal plants and it has been globally cultivated for its grains since the dawn of civilization. In the last decade, its world production has increased, currently reaching 750 million tons [1], while the sowing area has fluctuated around 220 million hectares. China (17%) and India (12%) are the top producers and collectively the European Union produces around 15% of the world's total. In Italy, about 2 million hectares are cultivated, prevalently with durum wheat, for a production of 8 million tons [2]. The yield increase in the last century is the joint result of different factors, with plant breeding having a significant role in shaping the morphological and technological features of contemporary varieties [3]. These are characterized by a reduced height, a more efficient assimilate partitioning, diminished sensitivity to photoperiod,

**Citation:** Ladhari, A.; Corrado, G.; Rouphael, Y.; Carella, F.; Nappo, G.R.; Di Marino, C.; De Marco, A.; Palatucci, D. Chemical, Functional, and Technological Features of Grains, Brans, and Semolina from Purple and Red Durum Wheat Landraces. *Foods* **2022**, *11*, 1545. https://doi.org/ 10.3390/foods11111545

Academic Editors: Donatella Bianca Maria Ficco and Grazia Maria Borrelli

Received: 13 April 2022 Accepted: 20 May 2022 Published: 25 May 2022

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

**Copyright:** © 2022 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 (https:// creativecommons.org/licenses/by/ 4.0/).

adaptability to certain agronomic conditions, and resistance to specific races of fungal pathogens [3], traits that are expected to be absent in the old varieties of wheats [4].

Durum wheat (*Triticum turgidum* L. subsp. *durum*, Desf.) is a tetraploid species (AABB; 2n = 4x = 28) with better tolerance to drought and heat than the hexaploid common wheat (*Triticum aestivum* L; AABBDD; 2n = 6x = 42). Durum wheat, also known as hard, pasta, or macaroni wheat, is mainly cultivated in Mediterranean countries, North America, Argentina, and eastern Europe. In Italy, the production is principally located in southern regions such as Apulia and Sicily. Durum wheat is central to the gastronomy of Mediterranean countries because it is employed to produce pasta and couscous, as well as bulgur, puddings, pastries, freekeh, kishk, and other traditional dishes. An important aesthetic and commercial feature of durum wheat semolina is the color. The typical yellow-amber pigmentation is predominantly due to lipophilic carotenoids within the kernel, mainly lutein [5]. Nonetheless, anthocyanins are also another class of pigmented phytochemicals that can be present in high amount in the grains of some wheat varieties. According to the quantity and type, this class of water-soluble pigments can give rise to wheat grains with colors ranging from red to purple [5].

In recent years, the scientific interest and appreciation of the quality of traditional wheat varieties has increased for a more sustainable low-input production of grain, as germplasm with an enhanced phytochemical profile, and as a source of adaptive traits in the face of climate change [6,7]. For instance, considering that wheat is a staple food in several countries, anthocyanin-rich grains can be used to produce a wide range of foods with enhanced nutraceutical and pharmaceutical properties. Old varieties are also gaining popularity to satisfy consumer demand for regional crop production and food manufacturing, to diversify the dietary basket, and to provide commercially novel products richer in health-promoting ingredients [7,8]. Regrettably, the compositional properties of anthocyanin-rich grains of landraces, as well as old durum wheat varieties, have not yet been fully acknowledged, not only if compared with soft wheat varieties, but also with old species such as einkorn, emmer, and spelt [5,9–12].

The aim of this research was to explore the chemical, functional, and technological features (such as the contents of proteins; dry gluten; gluten index; ash; total polyphenols; antioxidant activity) of two differently pigmented *Triticum turgidum* landraces. Moreover, we quantified major anthocyanins (Delphinidin 3-glucoside, Delphinidin 3-rutinoside, Cyanidin 3-glucoside, Petunidin 3-glucoside, Peonidin 3-glucoside, and Malvidin 3-glucoside), anthocyanidins (i.e., cyanidin, delphinidin, malvidin, peonidin, and petunidin), and specific phenolic acids (ferulic acid, *p*-hydroxybenzoic acid, vanillic acid, and *p*-coumaric acid). These two landraces were named 'Purple' and 'Red' according to the color of the grain and were originally collected from the Oromya region, one of regional states of Ethiopia. Cereal landraces are mainly evaluated as a source of inheritable traits that may favor local adaptation and productivity in sustainable agriculture [13], but they also have desirable characteristics related to food quality and nutritional benefits. Therefore, our detailed characterization contributes to demonstrate the value of durum wheat landraces as a rich source of primary and secondary metabolites of interest for human health and nutrition.

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

#### *2.1. Materials*

We analyzed two durum wheat (*Triticum turgidum* L. subsp. *durum*, Desf.) landraces named 'Purple' and 'Red'. They are maintained and multiplied by Agrismarter (Foggia, Italy), marketed by the company Granomischio (Foggia, Italy), and cultivated in areas near the Daunian Mountains (Apulia region, Italy). Grains (i.e., intact caryopses), bran (i.e., kernel components except the flour fraction at the given extraction rate, ~65–75%), and semolina were provided also by the Granomischio company in September 2021. For their analysis, grains were ground to a fine powder using a blender stored at −20 ◦C. Semolina and bran were processed by a professional milling company with a multi-pass roller system. Briefly, grains were purified (mainly to remove low-density particles), surface cleaned (to

remove impurities and possible abiotic and biotic contamination), pearled/scoured, milled, and purified/sieved to semolina.

Chemicals, analytical grade reagents, and standards of phenolic compounds were obtained from Sigma-Aldrich (Milan, Italy).

#### *2.2. Determination of Dry Gluten*

Gluten was extracted from semolina (*n* = 5) according to previously described procedure [14]. Twenty grams of semolina were suspended in 12.5 mL of a 4% monosodium/disodium phosphate buffer at pH of 6.8, diluted to a ratio of 1:40 with a 2% NaCl solution. After 30 min, the dough was washed using a Glutomatic System 2200 (Perkin Elmer, Turin, Italy) with the NaCl solution to remove soluble proteins and starch. The pure gluten obtained was dried in an oven (Heraeus T6200, Progitec, Sabaudia, LT, Italy). The gluten content was expressed as dry gluten per 100 g of material.

#### *2.3. Determination of Gluten Index*

The determination of the gluten index comprised three steps: (i) the gluten extraction and quantification of the wet gluten; (ii) the centrifugation of the wet gluten; and (iii) the calculation of the gluten index, according to standard procedures [14] using a Gluten Index 2100 centrifuge (Bastak, Ankara, Turkey). The gluten index is the percentage ratio of the wet gluten remaining on the sieve (after centrifugation) to the total wet gluten.

#### *2.4. Determination of Total Protein Content*

Total proteins were quantified using the Kjeldahl method. Briefly, samples (1 g) were digested in 15 mL of 98% H2SO4 in the presence of a catalyst (K2SO4:CUSO4, 9:1 *w/w*). Then, 50 mL of 40% NaOH (*w*/*v*) was added to covert the released ammonium into ammonia, which was distilled and collected in a flask containing a known amount of excess acid (0.1 M HCl). The excess acid was back-titrated with 0.1 M NaOH. The protein content refers to 100 g of substance and is calculated using the following formula:

$$\text{Proteiners (\%)} = \left[ 100 \times (\text{V}\_{\text{NaOH}} \times \text{C}\_{\text{NaOH}} - \text{V}\_{\text{HCl}} \times \text{C}\_{\text{HCl}}) \times 14.0067 \times 5.70 \right] / \text{g}$$

where VNaOH are the liters of NaOH; CNaOH is the molar concentration of NaOH; VHCl are the liters of HCl; CHCl is the molar concentration of HCl; 14.0067 is the atomic weight of nitrogen; 5.70 is the conversion factor for proteins; and g is the weight of the sample in grams [15].

#### *2.5. Determination of Ash*

The quantification of ash was performed essentially as described [16]. Briefly, a weighed sample of semolina (5–10 g) was placed in a platinum capsule and heated at 550–590 ◦C in the muffle furnace Srefo R-1905 (Zhuhai Refine Zhizao, Guangdong, China) until light gray ash was obtained (5 h). The weight of the ash refers to 100 g of dry matter. The measurement was carried out on five replicates.

#### *2.6. Phenolic Extraction*

The phenolic extracts were prepared as reported [17]. The steps of the approach are summarized in Figure 1. About 1.5 g of wheat material was pulverized and suspended in 30 mL of a methanol/hydrochloric acid solution (99:1, *v*/*v*). The mixture was stirred for 30 min at room temperature. The suspension was then centrifuged (PK121R Multispeed, ALC International, Milan, Italy) at 10,000 rpm for 5 min at room temperature. The sediment was extracted five times.

**Figure 1.** Flowchart of the process used for the extraction and analysis of the wheat material.

#### *2.7. Soluble Phenolic Fraction*

Two milliliters of the acidified methanolic extract were dried using a rotavapor (Rotavapor RE 111, Buchi, Switzerland) at 30 ◦C. The dried material was then suspended in 20 mL of 3M KOH and stirred for about three hours at room temperature. The solution was acidified with 1M HCl to pH 3 and extracted three times with a 1:1 (*v*/*v*) mixture of petroleum ether/ethyl acetate. Extracts were treated with anhydrous sodium sulfate, filtered on a Whatman paper Grade 1, dried under a light stream of nitrogen, and resuspended in methanol.

#### *2.8. Insoluble Bound Phenolic Fraction*

The sediment (Figure 1) was suspended in 20 mL of 3M KOH at room temperature for about three hours, under continuous stirring, acidified with 1M HCl to pH 3, and extracted three times with a 1:1 (*v*/*v*) petroleum ether/methylene chloride solution. Extracts were treated as in paragraph 2.2 and resuspended in methanol. All extractions were performed avoiding direct light to minimize photo-oxidation.

#### *2.9. Total Polyphenols*

The total phenol content in the soluble and insoluble bound extracts and in the different organic extracts (Figure 1) was quantified with the Folin–Ciocalteu method using gallic acid as a standard as previously reported [18]. The extracts were solubilized in 5 mL of methanol, with the aid of a sonicator. One milliliter was taken and filtered on a Phenex filter (0.45 μm) and diluted to a 10 mL final volume. Samples (*n* = 5) and solutions of the standard were tested with the colorimetric method and absorbance was read at 765 nm [19]. Specifically, 0.2 mL of the solution of the sample to be analyzed (or of the standard, or Milli Q water, in the case of the blank) was added with 0.8 mL of Milli Q water and 0.2 mL of the Folin–Ciocalteu reagent. The solution was incubated for 5 min and then another 0.8 mL of Milli Q water and 2 mL of an aqueous 8% Na2CO3 solution were added. A standard curve was built with gallic acid at the 4.8, 9.6, 48, 96, 240, and 480 μg/mL concentrations. The minimum threshold for accepting the calibration curve was r = 0.97. Results were expressed as milligrams of gallic acid equivalents.

#### *2.10. Phenolic Acid Quantification by HPLC*

Both phenolic extracts (Figure 1) were analyzed by HPLC in accordance with the already published protocols [20,21]. A Shimadzu LC-8A HPLC instrument (Shimadzu, Milan, Italy) was used with a 2.6 mm 100 Å (100 × 4.6 mm) Kinetex reverse phase column. The eluent phase consisted of a mixture of A (2% AcOH in water, *v*/*v*) and B (methanol), with a constant flow of 1.2 mL/min and a wavelength of the UV detector set at 280 nm. The gradient, in terms of eluent B, was 10% at time 0, 20% at 10 min, 25% at 15 min, and 30% at 30 min. The presence of ferulic acids, *p*-hydroxybenzoic acid, and vanillic and coumaric acids was measured through the corresponding calibration lines obtained from the corresponding standard samples commercially available from Sigma-Aldrich. The calibration curves were linear in the concentration intervals considered. In particular, the detection limits were equal to 12, 0.08, 0.1, and 1.11 μg/mL, respectively, for the four aforementioned phenolic acids. The analyses were performed five times; the results were expressed as micrograms per kilogram of wheat on a dry matter (DM) basis.

#### *2.11. Determination of Anthocyanins Content*

Total anthocyanin extract was prepared essentially as reported [22] and quantification was carried out with a spectrophotometric method [23], using catechin as standard with concentration ranging from 2 to 200 μg/mL. Results were expressed as μg catechin equivalents/g dry weight material. The equation obtained from the standard curve is y = 0.0018x + 0.0146, where y is absorbance at 535 nm and x is concentration of catechin standard.

#### *2.12. Determination of Proanthocyanidin Content*

Proanthocyanidin extract was prepared using previously published procedures [23] and the proanthocyanidin content was determined also as already described [24], using catechin as standard with concentrations ranging from 100 to 1000 μg/mL. Results were expressed as μg catechin equivalents/g dry weight material. The equation obtained from the standard curve is y = 0.0023x + 0.0187, where y is absorbance at 510 nm and x is concentration of catechin standard.

#### *2.13. Radical DPPH Scavenging Capacity*

The DPPH. antioxidant activity of the material under investigation (Figure 1) was evaluated using already published procedures [25] with minor modifications. One milliliter of extraction solvent with different extract dilutions was added to two mL of DPPH in methanol (5 × <sup>10</sup>−<sup>5</sup> M). The reaction was carried out at 25 ◦C for 30 min. After half an hour, the absorbance value reached a constant value, which was used to calculate the percentage of residual DPPH. Radical reduction by antioxidants was monitored by measuring the absorbance at 517 nm using a Perkin Elmer Lambda 7 spectrophotometer (Beckman, Brea, CA, USA). Five extracts were analyzed for each sample, each with four different dilutions. A regression line was also calculated for the reference antioxidant Trolox, with concentrations ranging from 3 to 50 μM. The antioxidant activity was expressed as the ratio between the I50 of Trolox and the I50 of the sample, that is, micromoles of Trolox equivalent (TE) per gram of DM.

#### *2.14. Radical ABTS Scavenging Capacity*

The ABTS antioxidant activity (Figure 1) was evaluated as already published [26], quantifying the ability of natural extracts to convert the radical cation ABTS+, generated from the corresponding acid using as oxidizing agent sodium or potassium persulfate (K2S2O8 or Na2S2O8) in its neutral form. For the ABTS assay, the antioxidant capacity was also expressed as the Trolox equivalent antioxidant capacity (TEAC), a unit of measurement defined as the quantity of Trolox needed to obtain the same antioxidant activity as the sample (micromoles of TE per g of sample).

#### *2.15. Statistical Analysis*

Data are reported as mean value ± standard deviation (SD). The normality of the data distribution was assessed by the Shapiro-Wilk test. The independent Student's *t*-test was employed for mean separation considering as the threshold of statistical significance a *p*-value lower than 0.05. Calculations were performed with the SigmaPlot 12.2 software (Systat Software, San Jose, CA, USA).

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

#### *3.1. Analysis of Dry Gluten, Gluten Index, Protein Content, and Ash*

The two wheat varieties analyzed had a similar percentage of dry gluten, around 12.5–12.8% (Table 1). Nonetheless, *T. durum* 'Purple' had a higher gluten index (GI of 37 against 31 of the 'Red'), values that can be considered low in durum wheat [27–29]. The GI is a widely accepted parameter to express gluten strength, and it is considered highly inheritable [30]. Hence, the observed difference reveals distinct technological features of the two landraces. Nonetheless, the relationship of the GI with the protein content should not be thought as linear [29,31]. For both varieties, the latter was not far from the upper limit for durum wheat, usually from 7% to 18%, with an average of 12%. The protein content and the gluten strength are the main features of the starting material that determine the quality of pasta [28]. Even so, the protein content is often considered more important than the strength of the gluten, being more closely correlated with the positive features of dried pasta, although this relationship can be affected by the processing method and the gluten composition [32]. The semolina of 'Purple' and 'Red' had ash contents of 1.25 and 1.35%, respectively. The ash content in the grain is under genetic and environmental control. It is typically influenced by mineral fertilization and positively correlated with the protein content [33]. A low ash content in milled durum grains is considered a quality feature, although it does not significantly affect the industrial performance of the semolina. In Italy, as in other countries, the ash content in common and durum wheat is regulated by law (D.P.R. 9 February 2001, n. 187). The observed values exceed the threshold for "*semola*" and are within limits for "*semolato*". It should be added that the ash content lowers with sequential debrannings. For instance, in a group of 11 Italian varieties encompassing traditional and contemporary durum wheat cultivars, the ash content reached an acceptable level for "*semola*" after five successive debranning treatments [34].



#### *3.2. Polyphenol Content*

The grain, semolina, and bran of both 'Purple' and 'Red' were subjected to an evaluation of the total polyphenol content because these are a wide group of chemically diverse secondary metabolites that are generally known for their positive effects on human health [35,36]. Cereals, as many other plant species, contain different phenolic compounds, and the most important are the derivatives of the cinnamic acid (i.e., coumaric, caffeic, and ferulic acids), flavonoids, and lignans [37]. It is believed that their function is predominantly non-nutritive [37].

The quantification of total polyphenols indicated that the bran of the 'Purple' landrace had a significantly higher amount (almost double) of both phenolic fractions, while differences were not significant for the insoluble fractions of the semolina and grains (Table 2).


**Table 2.** Soluble and insoluble polyphenolic fraction in 'Purple' and 'Red' landraces. Values are reported as mean ± SD (*n* = 5). Significant differences between the two varieties for bran, semolina, or grain are indicated by lowercase letters (*t*-test, *p* < 0.05).

Specifically, 'Purple' bran and grains had an almost double phenolic content (112 and 103 μg/g DM of the equivalents of gallic acid, respectively) compared to those of the 'Red' (64 and 65 μg/g DM of gallic acid equivalents, respectively), while the semolina content of 'Red' (49 μg/g DM of gallic acid equivalents) was significantly higher than that of the semolina of 'Purple' (31 μg/g DM of gallic acid equivalents). The same was true for the insoluble bound phenolic fractions. In 'Purple', the phenolic content in the bran and grain was 40% and 23% higher compared to 'Red'. This can be explained considering that the phenolic content in wheat grains is mainly concentrated in the bran with a limited contribution of the organs and tissue originating the milled semolina. Nonetheless, the two varieties differed in the variation of the ratio insoluble/soluble polyphenols across the analyzed material. This parameter little varied in 'Purple' (coefficient of variation, CV: 7.9%), while the CV was 24.3% for 'Red', whose semolina had the lowest insoluble/soluble polyphenol ratio.

#### *3.3. Determination of Phenolic Acids*

HPLC analysis of the semolina revealed several phenolic derivatives, with four compounds (i.e., ferulic, *p*-hydroxybenzoic, vanillic, and *p*-coumaric acids) being the most abundant. The analysis of the two landraces indicated a similar profile in terms of quantity and rank of the chemical compounds, with the ferulic acid always present in a predominant amount, followed, for both varieties, by the vanillic acid or the *p*-coumaric acid in the soluble or insoluble fraction, respectively (Table 3).

**Table 3.** Identification and quantification of the acids in the soluble and insoluble bound phenolic fraction in semolina of 'Purple' and 'Red' based on retention times and calibration lines for comparison with the corresponding commercial products. Values are reported as mean ± SD (*n* = 5). Mean values between the two varieties were not statistically different (*t*-test, *p* ≥ 0.05).


In the 'Red' variety, the ferulic acid was 639 mg/kg DM and 18.3 mg/kg DM in the insoluble fraction and in the soluble fraction, respectively (Table 3). The ferulic acid is considered the predominant free- and bound-form of polyphenols in cereals, especially (brown) rice and corn [38]. In durum wheat, the ferulic acid contents are slightly influenced by the environment in normal agronomic conditions, and it is mostly influenced by the genotype and altered by abiotic stress [39]. In the soluble fraction, of the three other quantified phenolic acids, the most abundant was vanillic acid, with concentrations in the range of 7.2–7.4 μg/kg DM, followed by *p*-hydroxybenzoic acid, with concentrations in the range of 3.1–3.3 μg/kg DM. The *p*-coumaric acid was the less abundant constituent, with concentrations in the range of 2.2–2.5 μg/kg DM. In the insoluble fraction of the three other quantified phenolic acids, the most abundant was *p*-coumaric acid, with concentrations in the range of 18.5–19.2 μg/kg DM, and with much lower concentrations of vanillic acid (3.1–3.4 μg/kg DM) and *p*-hydroxybenzoic acid (2.1–2.3 mg/kg DM). Overall, the *p*-hydroxybenzoic and the vanillic acids were present in higher quantities in the soluble fractions of the semolina, while the amount of ferulic and *p*-coumaric acids was larger than insoluble fraction. Finally, we did not observe differences between the two varieties in the acids in the soluble and insoluble bound phenolic fraction of the semolina.

#### *3.4. Anthocyanins and Proanthocyanidins Content*

Anthocyanins and proanthocyanidins (expressed as the micrograms of catechin equivalents per gram of DM) were also quantified in the grain, semolina, and bran of 'Purple' and 'Red' (Table 4).

**Table 4.** Anthocyanins and proanthocyanidins in the 'Purple' and 'Red' landraces. Values are reported as mean ± SD (*n* = 5). Significant differences between the two varieties for bran, semolina, or grain are indicated by lowercase letters (*t*-test, *p* < 0.05).


The content of anthocyanins in the grain of 'Purple' was higher than in 'Red', almost double in the bran (72.9 vs. 36.3 μg/g DM) and triple in the grain (116.6 vs. 39.2 μg/g DM) (Table 4). In contrast, the semolina of 'Red' contained 25% more anthocyanin than 'Purple' (16.8 vs. 12.5 μg/g DM, respectively). A similar trend was observed for proanthocyanidins (polymers or oligomers of anthocyanidin). Their content was 33% higher in the bran of the 'Purple' compared to that of the 'Red' (1530 vs. 1031 μg/g DM of catechin equivalents, respectively) and 47% more in the grain of the 'Purple' than that of the 'Red' (3437 vs. 1807 μg/g DM of catechin equivalents, respectively). On the other hand, the amount of proanthocyanidins in the semolina was almost double (+47%) in the 'Red' compared to the 'Purple' (466 vs. 244 μg/g DM of catechin equivalents, respectively).

Anthocyanins are a class of water-soluble pigments belonging to the flavonoid family. They show a range of pharmacological activities because of their antioxidant and anti-inflammatory properties, with potential therapeutic benefits [40]. In durum wheat, anthocyanins mainly accumulate in the pericarp and aleurone [5]. This may explain why the less colored 'Red' variety yielded semolina with a significantly higher content of anthocyanins and proanthocyanidins. Anthocyanin-rich grains can be then used to produce functional foods and considering that these compounds are in the less noble coat of the grain, it has been also proposed that grains can be also exploited as a natural source to extract these pigments [5].

The determination of the anthocyanin compounds by HPLC revealed the presence of significant differences between the material and the variety (Table 5). Overall, the cyanidin 3-glucoside was the most abundant compound, followed by peonidin 3-glucoside. Quantitative differences between the varieties were most pronounced for the bran, with the 'Purple' having on average a threefold higher amount than 'Red' of the detected molecules. Comparing the material, as expected, anthocyanins were in much lower quantities in the semolina, with the two major anthocyanins detected only in the 'Purple' variety. Interestingly, the analysis of the grain revealed both qualitative and quantitative differences. Specifically, the differences in relative terms between the two varieties were more limited

compared to the bran. Moreover, only the grain of the 'Red' variety contained detectable amounts of delphinidin 3-glucoside and delphinidin 3-rutinoside.

**Table 5.** Anthocyanin composition in the 'Purple' and 'Red' landraces (mean ± SD; *n* = 5) determined by HPLC. Values are reported as mean ± SD (*n* = 5). Significant differences between the two varieties for bran, semolina, or grain are indicated by lowercase letters (*t*-test, *p* < 0.05).


Cyanidin 3-glucoside is often the most abundant anthocyanins in colored cereals (e.g., rice and corn), as well as in most of the plants [40]. For purple common wheat, cyanidin 3-glucoside, peonidin 3-glucoside, and cyanidin 3-galactoside have been described as the most abundant compounds [41,42]. In blue common wheat, cyanidin 3-glucoside is predominant, with pelargonidin 3-glucoside and cyanidin 3-galactoside present in lower amounts [43]. Although intra-varietal differences in colors are sufficiently explained by variation in anthocyanins [44], it is not straightforward to correlate the color of a variety with the type, number, and quantity of pigments, also considering that the influence of the anthocyanins on the plant tissue hue (and tint) is determined by various factors besides their total amount and ratio [45,46]. It is therefore interesting that the grain of the 'Red' variety contained delphinidins in a low amount compared to other pigments, while these were not detected in the 'Purple'. These anthocyanins are typically associated with dark grains [5]. For instance, in a blue common wheat, delphinidin 3-glucoside and delphinidin 3-rutinoside accounted for 69.3% of the detected anthocyanins [42]. Nonetheless, they were also not found in a purple common wheat [42]. In addition to wheats also in purple rye grains cyanidin-3-glucoside is the predominant anthocyanin, followed by peonidin-3 glucoside [47,48] as in our 'Purple'.

We also quantified major anthocyanidins (i.e., cyanidin, delphinidin, malvidin, peonidin, and petunidin) in the 'Purple' and 'Red' bran, semolina, and grain (Table 6) because these are typical of colored wheat grains [10]. In grains, cyanidin was the most abundant aglycones for all the material, followed by delphinidins. Other works indicated that cyanidin also the main aglycone in purple common and durum wheat varieties, but it was followed by peonidin [49]. Petunidin was detected in smaller quantities only in the two brans, while peonidin and malvidin only in the grains of the 'Purple'.

**Table 6.** Anthocyanidin composition in the 'Purple' and 'Red' (mean ± SD; *n* = 5) determined by HPLC. Values are reported as mean ± SD (*n* = 5). Significant differences between the two varieties for bran, semolina, or grain are indicated by lowercase letters (*t*-test, *p* < 0.05).


13

The total anthocyanin content was higher in the 'Purple' than in the 'Red'; precisely, it was just over double of that contained in the bran (25 vs. 12.01 μg/kg DM), more than double of that contained in the grain (12.29 vs. 5.09 μg/kg DM), and almost 10 times that contained in the semolina (5.36 vs. 0.56 μg/kg DM).

#### *3.5. Antioxidant Activity*

Tests for the evaluation of antioxidant activity show the highest values for the grain in both varieties compared to bran and semolina. Moreover, in both assays the highest values were observed for the 'Purple' variety compared to that 'Red' (Table 7). Mean values were not significantly different between the semolina of the two varieties. In the brans, were higher values in the DPPH test (respectively, ABTS test) were recorded for the 'Purple' (resp. 'Red'). As previously reported [50], the highest antioxidant capacity was observed in whole grains, which contain more bioactive compounds of health interest, such as insoluble fiber, phenolic acids, and alkylresorcinols [51].

**Table 7.** Antioxidant activity (μmol Trolox equivalent/g DM) in 'Purple' and 'Red' landraces. Values are reported as mean ± SD (*n* = 5). Significant differences between the two varieties for bran, semolina, or grain are indicated by lowercase letters (*t*-test, *p* < 0.05).


#### **4. Conclusions**

The characterization of *Triticum durum* 'Purple' and 'Red', two varieties imported from Ethiopia and grown in southern Italy, highlighted their distinctive features such as an above-average protein content, which should positively influence the pasta-making quality. The ferulic acid was particularly abundant among phenolic acids in both the soluble and insoluble phenolic fractions of the grain, semolina, and bran. Even if related to the analysis of two landraces, our work also revealed the qualitative and quantitative diversity in anthocyanin content in durum wheat. It should be added that the material under investigation derives from a non-mass-production system in which grains are processed with a multi-pass methodology using a rolling mill. These systems generate less heat during grinding also because multiple passes allow the achievement of size reduction more gradually [52]. In the future, it may be worth investigating to what extent the high antioxidant activities in the semolina may be affected by the milling method. The lower gluten index, high protein level, rich and composite range of secondary metabolites, along with the antioxidant activities, indicate that the germplasm under investigation has interesting features for the niche market of functional durum wheat products in specific geographical areas as an alternative to the mass-produced Italian goods required for international markets.

**Author Contributions:** Conceptualization, D.P. and A.D.M.; methodology, D.P., C.D.M. and A.L.; formal analysis, G.R.N., F.C., C.D.M. and D.P.; investigation, F.C. and D.P.; data curation, Y.R., G.C., D.P. and A.D.M.; designed the research study, supervision, writing—original draft preparation and the last version of the manuscript, Y.R., G.C., A.D.M. and D.P. All authors have read and agreed to the published version of the manuscript.

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

**Data Availability Statement:** The data are contained within the article.

**Acknowledgments:** We thank the non-profit organization "Associazione Italiana per la Promozione delle Ricerche sull'Ambiente e la Salute umana" (AIPRAS) for its support.

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

#### **References**


**Alison Lovegrove 1, Jack Dunn 1, Till K. Pellny 1, Jessica Hood 1, Amanda J. Burridge 2, Antoine H. P. America 3, Luud Gilissen 4, Ruud Timmer 5, Zsuzsan A. M. Proos-Huijsmans 6, Jan Philip van Straaten 6, Daisy Jonkers 7, Jane L. Ward 1, Fred Brouns <sup>8</sup> and Peter R. Shewry 1,\***


**Abstract:** Five cultivars of bread wheat and spelt and three of emmer were grown in replicate randomised field trials on two sites for two years with 100 and 200 kg nitrogen fertiliser per hectare, reflecting low input and intensive farming systems. Wholemeal flours were analysed for components that are suggested to contribute to a healthy diet. The ranges of all components overlapped between the three cereal types, reflecting the effects of both genotype and environment. Nevertheless, statistically significant differences in the contents of some components were observed. Notably, emmer and spelt had higher contents of protein, iron, zinc, magnesium, choline and glycine betaine, but also of asparagine (the precursor of acrylamide) and raffinose. By contrast, bread wheat had higher contents of the two major types of fibre, arabinoxylan (AX) and β-glucan, than emmer and a higher AX content than spelt. Although such differences in composition may be suggested to result in effects on metabolic parameters and health when studied in isolation, the final effects will depend on the quantity consumed and the composition of the overall diet.

**Keywords:** bread wheat; emmer; spelt; fibre; metabolites; minerals; phenolics; fertilisation; health benefits

#### **1. Introduction**

Wheat is the major staple crop in temperate countries, with annual global yields exceeding 700 million tonnes. About 95% of the total production is hexaploid bread wheat (*Triticum aestivum* L. subsp. *aestivum*, genome constitution AABBDD) which originated about 10,000 years ago, with most of the remaining 5% production being tetraploid pasta wheat (*Triticum turgidum* L. subsp. *durum*) (AABB genomes). Bread and durum wheats are "free threshing" (the hulls being readily separated from the grain during harvest), which is regarded as an advanced trait. However, both bread and pasta wheats have been subjected to intensive breeding, focusing on improving their agronomic performance and increasing their yield and quality (for making bread and pasta, respectively). Hence, although there is wide genetic variation in both species, modern cultivars (those developed by scientific breeding during the last few decades) tend to be less genetically diverse than older cultivars and traditional types of wheat dating from before the application of breeding (called land races) [1].

**Citation:** Lovegrove, A.; Dunn, J.; Pellny, T.K.; Hood, J.; Burridge, A.J.; America, A.H.P.; Gilissen, L.; Timmer, R.; Proos-Huijsmans, Z.A.M.; van Straaten, J.P.; et al. Comparative Compositions of Grain of Bread Wheat, Emmer and Spelt Grown with Different Levels of Nitrogen Fertilisation. *Foods* **2023**, *12*, 843. https://doi.org/10.3390/foods12040843

Academic Editors: Donatella Bianca Maria Ficco, Grazia Maria Borrelli and Grant Campbell

Received: 16 January 2023 Revised: 7 February 2023 Accepted: 13 February 2023 Published: 16 February 2023

**Copyright:** © 2023 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 (https:// creativecommons.org/licenses/by/ 4.0/).

Ancient wheats are diploid einkorn (*T. monococcum*, L., AA genome), tetraploid emmer (*T. turgidum* L. subsp. *dicoccum* Thell., AABB genomes) and hexaploid spelt (*T. aestivum* L. subsp. *spelta* Thell, AABBDD genomes) and are generally hulled as opposed to free threshing. Although the term "ancient" is taken to imply that the genotypes grown today are similar to those grown in antiquity, this is certainly not the case, at least for spelt and emmer. Einkorn is a distinct species, which includes cultivated and wild forms, while emmer and spelt are subspecies of *T. turgidum* and *T. aestivum*, respectively. Furthermore, modern commercial cultivars of spelt may have introgressions (transfer of genetic information) from bread wheat due to cross-breeding, while all types of wheat cultivated today have been grown and hence selected (either unconsciously or deliberately) over thousands of years [2].

The composition of wheat grain is determined by the genotype, the environment, the farming system and the interactions between the genotype and these factors. Environmental factors are particularly important when comparing ancient and modern wheats, as the former are often grown in organic or low input systems with low nitrogen application to avoid lodging (bending of the plant at or near ground level, making harvest difficult and often leading to premature germination of grains) while modern semi-dwarf wheats are more usually grown in intensive high input systems with high nitrogen fertilisation [3].

The breeding and selection of modern bread wheats have focused on increasing yield and improving breadmaking quality, which is largely determined by the content and composition of gluten proteins. It has therefore been suggested that this has led to modern wheats having lower contents of micronutrients (minerals, vitamins) and bioactive components (phytochemicals) and higher contents of proteins that may lead to adverse reactions and diseases such as coeliac disease, wheat allergy and non-coeliac wheat sensitivity [4,5]. Hence, there has been increased interest in ancient wheats, which are assumed to have more favourable compositions for health [2].

We have therefore carried out detailed analyses of grain samples of three commercial cultivars of emmer and five cultivars each of spelt and bread wheat. All of the cultivars are adapted to Northern Europe and corresponded, with one substitution due to unavailability, to those selected to compare the effects of breads on health as part of the "Well-on-Wheat?" research consortium programme (https://www.wellonwheat.org, accessed on 1 February 2023). These samples were grown in replicate field trials in two Northern European countries (the UK and the Netherlands) for two years with low (100 kg/Ha) and high (200 kg/ha) applications of nitrogen fertilisation to reflect the different inputs used for ancient and modern wheats. Wholemeal samples were analysed for the major types of dietary fibre in white flour (arabinoxylan and β-glucan), polar metabolites, protein (as nitrogen), phenolics and mineral micronutrients to identify differences in composition.

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

#### *2.1. Grain Samples*

Commercial samples of five cultivars each of bread wheat (RAGT Reform, Capo, Bernstein, Kometus, Akteur) and spelt (Comburger, Zollernspelz, Attergauer, Bauländer Spelz, Franckenkorn) and three cultivars of emmer (Ramses, Roter Heidfelder, Späths Albjuwel) were grown in two years (2017–2018 and 2018–2019) at two nitrogen levels (100 and 200 kg N/Ha) in Flevoland (WUR Field crops, Lelystad, Flevoland, Netherlands, 52◦53 94.69 N, 5◦56 56.77 E) in three randomised replicate 6 × 1.5 m plots and at Rothamsted Research (Harpenden, Hertfordshire, UK, 51◦48 19.79 N 0◦21 11.39 W) in three randomised replicate plots of 1 × 1 m. All trials were autumn sown, but sowing, fertiliser application and harvest dates varied between sites and years, according to local conditions. Standard agronomic practices for the two sites were used. Grain samples of emmer and spelt were mechanically dehulled. Whole grain samples at about 14% water content were milled in two stages: firstly, a Retsch ZM 200 Model Ultra-Centrifugal Mill (Retsch Gmbh, Dusselgorf, Germany) using a 0.5 mm ring sieve and then a Glen Creston Ball Mill Retsch

Gmbh, Dusseldorf, Germany) using 5 ball bearings in a 5 cm diameter canister for 4 min for each sample.

#### *2.2. Genotyping*

The samples were genotyped using the Axiom 35k Wheat Breeders Genotyping Array (Thermo Fisher Scientific, Inc., Waltham, MA, USA) using the Affymetrix GeneTitan (Thermo Fisher Scientific, Inc.) [6]. Alleles were identified using the Affymetrix proprietary software package Axiom Analysis Suite V4.0.3.3 (Thermo Fisher Scientific, Inc.) and prior model 'Axiom\_WhtBrd-1.r3'. A Dish QC threshold of 0.8 and call rate cut-off of 90% were used to adjust for hybridisation rates of spelt and emmer to the array. A distance matrix was generated from the scores using R package SNPRelate (Bioconductor Open Source, Harvard, MA, USA) [7]. The first two Principal Components accounting for over 25% of the variance (PC1:19.76%; PC2:5.53%) were plotted as a PCA plot.

#### *2.3. Enzyme Fingerprinting of Arabinoxylan and β-Glucan*

Three technical replicates of flour were digested with endoxylanase and lichenase (β-glucanase) to release arabinoxylan oligosaccharides (AXOS) from arabinoxylan (AX) and gluco-oligosaccharides (GOS) from β-glucan, respectively [8]. The oligosaccharides were separated using a 2 mm × 250 mm Carbopac PA-1 (Dionex) column [8] (dx.doi. org/10.17504/protocols.io.babriam6, accessed on 1 February 2023). The areas under the oligosaccharide peaks were combined to give total AX and total β-glucan (expressed in arbitrary units), respectively.

#### *2.4. NMR Spectroscopy of Polar Metabolites*

Sample preparation for 1H-NMR was carried out as described by [9]. Signal intensities for characteristic spectral regions for 29 major metabolites were compared with a library of spectra of standards analysed under the same conditions.

#### *2.5. Mineral Analysis*

Nitrogen was determined on each biological replicate by Dumas combustion, using a Leco combustion analyzer (Leco Corp., St. Paul, MN, USA). Iron and zinc were determined by Optima 7300 DV Inductively Coupled Plasma–Optical Emission Spectrometer (ICP– OES) (Perkin Elmer, Waltham, MA, USA) after digestion with nitric and perchloric acids. Certified external standards and in-house standards were used to monitor performance using Shewhart Control Charts.

#### *2.6. Total Phenolics*

Total phenolics were determined based on [10]. Triplicate 75 mg samples of each biological replicate were vortexed with 1.5 mL acidified methanol and then mixed at 850 rpm on an Eppendorf Thermomixer (Eppendorf Ltd., Stevenage, UK) for 2 h at 23 ◦C. After centrifugation (Eppendorf Ltd., Stevenage, UK) at 5000× *g* for 10 min, 1 mL of supernatant was removed into a fresh Eppendorf tube and 200 μL aliquots mixed with 1.5 mL of ×10 diluted Folin–Ciocalteau reagent (Sigma-Aldrich, St. Louis, MO, USA) and left to stand for 5 min. 1.5 mL of 6% (*w*/*v*) aqueous sodium carbonate solution was added, mixed and stood at room temperature for 90 min. The absorbance at 725 nm was then measured (Jenway 6715 UV/Vis spectrophotometer, Cole-Parmer, St Neots, UK), and the concentration of phenolics was calculated using ferulic acid as a standard (Sigma-Aldrich) and a standard curve from 20, 40, 100, 150 and 200 μg/mL with three technical replicates of each point.

#### *2.7. Statistical Analysis*

All data were analysed using analysis of variance (ANOVA) in Genstat 21 (VSN International, Hemel Hempstead, UK). The block structure was Trial/Block/Subblock, where Trial captures the location and year of each trial. There are 3 blocks within each trial and

2 sub-blocks within each block to which the Nlevel treatment was applied. Lines were considered to be applied to plots within each sub-block. The treatment structure was Nlevel\*(Grain/(cultivarBreadwheat + cultivarEmmer + cultivarSpelt)) where the factor Grain tests for differences between the three grain types. The nested factors cultivarBreadwheat, cultivarEmmer and cultivarSpelt test for differences between the lines within each grain type. The Nlevel factor tests for differences between the two Nlevels and their interactions with grain type and lines are also included. Some variables were transformed in order to meet the normality and homoscedasticity assumptions of the analysis. Means and 95% confidence intervals are given in Tables 1 and 3 while the transformations used are indicated in Supplementary Tables of means and *p* values.

Principal component analysis (PCA) and orthogonal partial least squares-discrimination analysis (OPLS-DA) were carried out using SIMCA-P software (version 13, MKS Umetrics) (Sartorius UK Ltd., Epsom, UK).

#### **3. Results**

Five cultivars each of bread wheat (RAGT Reform, Capo, Bernstein, Kometus, Akteur) and spelt (Comburger, Zollernspelz, Attergauer, Bauländer Spelz, Franckenkorn) and three cultivars of emmer (Ramses, Roter Heidfelder, Späths Albjuwel) were selected, all of which have been grown commercially in Northern Europe. The genomic relationships between the cultivars were determined using the Axiom 35k Wheat Breeders Genotyping Array, comprising 35,143 single nucleotide polymorphism (SNP) markers. Principal component analysis (PCA) showed a clear separation of the three cereal types, confirming that the spelt lines used did not have recent introgressions from bread wheat (Figure 1).

#### *3.1. Grain Composition*

In order to determine the variation in composition within and between bread wheat, spelt and emmer, the four environments (site × year combinations) were treated as blocks in the statistical analyses. In addition, to compare the effects of nitrogen fertilisation on grain composition, the data for the 100 kg/Ha and 200 kg/Ha treatments were analysed separately and compared formally by including nitrogen as a factor in the statistical analyses. The ranges of contents are illustrated in Figures 2, 3 and 5, while Tables 1–4 and Supplementary Tables S1–S8 present means, 95% confidence intervals, SEMs and observed statistical significance determined by ANOVA.

The groups of components are discussed below.

#### *3.2. Protein and Minerals*

Grain protein content (determined as N × 5.7) (Figure 2A, Table 1) overlapped in range between bread wheat, spelt and emmer but was highest in spelt and lowest in bread wheat. It was also higher at 200 kg/Ha than at 100 kg/Ha. ANOVA (Table 2) showed statistically significant differences in protein content between bread wheat, emmer and spelt and statistically significant effects of nitrogen on their protein contents. By contrast, the three cereal types did not differ significantly in their response to nitrogen, and there was little variation in the effects of nitrogen between the cultivars within a single cereal type.

**Figure 2.** Contents of protein, minerals and total phenolics in grains of the three cereal types grown in four environments. Colour code: red, emmer; green, spelt; blue, bread wheat. The bar shows the range of the whole dataset. The box shows the middle two quartiles, separated by the horizontal line, which is the median, and the vertical lines are the upper and lower quartiles, respectively. Outliers are shown as circles. The x is the mean average. All analyses are expressed on a dry weight basis.


**Table 1.** Means and 95% confidence intervals (in parentheses) of contents of selected minerals, metabolites and groups of metabolites in grain of the three types of wheat grown with 100 and 200 kg N/Ha.

The box plots show that the contents of calcium (Figure 2B) and magnesium (Figure 2C) were not affected by nitrogen, but the content of magnesium was lower in bread wheat grain than in grains of spelt or emmer. The contents of iron and zinc were lowest in bread wheat grain (Figure 2D,E), while the content of iron was lower in emmer than in spelt grain. The contents of both minerals were also higher at 200 kg N/Ha than at 100 kg N/Ha. However, ANOVA showed significant differences between the contents of all minerals in the grain types and significant effects of nitrogen fertilisation on the contents of all minerals except magnesium, which showed a significant interaction due to a difference between nitrogen fertilisation in emmer only (Table 2). There were also significant differences between the contents of all minerals (except zinc in spelt and iron in all grain types) between the cultivars within each type, but no differences in the effects of nitrogen between the cultivars within the types (except for iron in spelt) (Table 2 and Table S1).

#### *3.3. Total Phenolics*

Phenolics are the most abundant phytochemicals in wheat grain [11]. The contents of total phenolics varied widely (Figure 2F), with significant differences between cereal types and cultivars within types (Table 2). There was an interaction between cereal type and nitrogen, with total phenolics being lower at 100 kg N/Ha for all cereal types apart from spelt where total phenolics were higher at 100 kg N/Ha than at 200 kg N/Ha.

#### *3.4. Polar Metabolites*

The contents of polar metabolites in the samples were determined by 1H NMR spectroscopy. This allowed the quantification of monosaccharide (glucose, fructose, galactose), disaccharide (maltose, sucrose) and trisaccharide (raffinose) sugars, organic acids (malic, acetic, fumaric), the sugar alcohols inositol and galactinol, the "methyl donors" choline and betaine and thirteen amino acids (alanine, aspartic acid, asparagine, glycine, glutamic acid, glutamine, γ-amino butyric acid (GABA), isoleucine, leucine, phenylalanine, tyrosine, tryptophan and valine). Data for all components are given in Supplementary Tables S2–S4, while selected components and groups of components are shown in Tables 1 and 2 and Figure 3.

The contents of all metabolites and groups of metabolites overlapped between the cereal types, but differences between the ranges in the types are observed (Figure 3). Asparagine is of particular interest because it is the limiting factor for the formation of acrylamide during processing [12,13]. The contents of total amino acids (Figure 3A) and of asparagine (Figure 3B) were significantly lower in bread wheat and higher in spelt, with a significant effect of nitrogen. However, there were no differences in the effects of nitrogen between and within the three cereal types.

The contents of total sugars (mono-, di- and trisaccharides) and total organic acids varied widely (Figure 3C,D), but both were significantly higher in emmer. Total organic acids were also significantly lower in bread wheat (Table 2). The contents of sugars were significantly affected by nitrogen, with no differences between the effects of nitrogen on the different cereal types or cultivars within the types (Table 2). The concentration of glycine betaine was about 10-fold greater than that of choline, which is typical for wheat [14]. Although the ranges overlapped (Figure 3E,F), they were significantly higher in spelt and lower in bread wheat, with little effect of nitrogen or variation within the types (Table 2).

Raffinose (the trisaccharide galactose, glucose, fructose) is a non-digestible and fermentable carbohydrate (being part of the FODMAP (fermentable oligosaccharides, disaccharides, monosaccharides and polyols) fraction) while galactinol (1-alpha-D-Galactosylmyo-inositol) and inositol ((1*R*,2*S*,3*r*,4*R*,5*S*,6*s*)-Cyclohexane-1,2,3,4,5,6-hexol) are precursors in raffinose synthesis [15]. Raffinose (Figure 3I) accounted for about a quarter of the total sugars (Figure 3C) and was significantly lower in bread wheat than in emmer or spelt. The contents of inositol were about half of those of raffinose and were higher in spelt (Figure 3G). Galactinol was present at much lower concentrations and was significantly higher in emmer and lower in bread wheat than in spelt (Figure 3H). The contents of raffinose and inositol were affected by nitrogen level and, with the exception of galactinol in spelt, varied significantly between cultivars of the three cereal types (Table 2).

PCA analysis of the metabolite dataset showed partial separation of the three cereal types, based on 48.9% of the total variance (Figure 4A). To improve the discrimination between cereal types, we repeated the analysis using supervised multivariate analysis (orthogonal partial least squares discrimination analysis, OPLS-DA), selecting for differences between the cereal types (Figure 4C). This gave clear separation between the three types with the loadings plot (Figure 4F) showing that emmer was characterised by high contents of maltose, galactinol and glucose and spelt by higher levels of amino acids (glycine, asparagine, leucine, isoleucine and valine) compared with bread wheat. OPLS-DA was also used to separate the samples based on nitrogen level (Figure 4B), the loadings plot (Figure 4E) showing higher contents of amino acids at 200 kg N/Ha. These differences are also illustrated by the difference plots in Supplementary Figure S1.



**Figure 3.** Contents of selected polar metabolites and groups of metabolites in grains of the three cereal types grown in four environments. Colour code: red, emmer; green, spelt; blue, bread wheat. The bar shows the range of the whole data set. The box shows the middle two quartiles, separated by the horizontal line, which is the median, and the vertical lines are the upper and lower quartiles, respectively. Outliers are shown as circles. The x is the mean average. All analyses are expressed on a dry weight basis.

**Figure 4.** Multivariate analysis of the contents of polar metabolites in grains of the three cereal types grown in four environments. Principal component analysis PCA (**A**) and orthogonal partial least squares discrimination analysis (OPLS-DA) (**B**,**C**) selecting for differences between nitrogen treatments (**B**) and cereal types (**C**). (**D**–**F**) loading plots for (**A**–**C**), respectively. Colour code: red, emmer; green, spelt; blue, bread wheat.

#### *3.5. Dietary Fibre*

The contents of AX and β-glucan in the grain types are shown in Figure 5A,B and in Table 3. The contents of both AX and β-glucan were lower in emmer grain, with AX being highest in bread wheat grain and β-glucan highest in spelt grain. Hence, the ratio of AX: β-glucan was lower in spelt grain than in the other cereals (Figure 5D). The combined contents of these two components were also highest in bread wheat grain (Figure 5C), reflecting the fact that the content of AX was about three- to four-fold greater than that of β-glucan. There was significant variation in the contents of AXOS between the cultivars of the three types of wheat and of β-glucan between the cultivars of bread wheat and emmer, but no significant effects of nitrogen fertilisation (Table 4).

AX and β-glucan were determined by enzyme fingerprinting, which uses enzymes (endoxylanase and lichenase, respectively) to digest the polymers to release oligosaccharides separated and quantified by HP-AEC. The oligosaccharides released have defined structures, and their proportions, therefore, provide information on the structures of the polymers. In the case of AX, the oligosaccharides (AXOS) comprise chains of one to five xylose residues, one or more of which may be substituted with either one or two arabinose residues. The ratio of monosubstituted to disubstituted AX may affect the properties of the molecules and is generally higher in spelt and bread wheat than in emmer (Figure 6F). β-glucan comprises linear chains of glucose molecules linked predominantly by β(1-4) bonds. However, these β(1-4) bonds are interspersed with β(1-3) bonds that generally occur every three to four glucose residues, although some longer stretches of β(1-4) linked glucose residues (up to 14) have been reported. The distribution of β(1-3) bonds results in conformational changes in the linear glucan molecules, which affect their solubility and viscosity. Lichenase is a type of β-glucanase that releases mainly gluco-oligosaccharides (GOS) of three or four glucose residues (called G3 and G4), reflecting the relative abundances of β(1-3) and β(1-4) bonds. The ratio of G3:G4 GOS is higher in emmer and lower in spelt than in bread wheat (Figure 5E, Table 3).

**Table 3.** Means and 95% confidence intervals (in parentheses) of contents and compositions of arabinoxylan and β-glucan in grain of the three types of wheat grown with 100 and 200 kg N/Ha.


**Figure 5.** Contents, ratios and structures of arabinoxylan and β-glucan in grains of the three cereal types grown in four environments. Colour code: red, emmer; green, spelt; blue, bread wheat. The box shows the middle two quartiles, separated by the horizontal line, which is the median, and the vertical lines are the upper and lower quartiles, respectively. Outliers are shown as circles. The x is the mean average.

Differences in the structures of AX and β-glucan in the cereal types are illustrated by the multivariate analysis in Figure 6. OPLS-DA confirmed that there were no effects of nitrogen on AX and β-glucan structure (Figure 6B) but gave clear separation of the cereal types (Figure 6C), with the loadings plot (Figure 6F) showing that spelt differed from emmer and bread wheat in having higher proportions of G3 and G4 GOS and bread wheat higher proportions of substituted AXOS These differences are also illustrated by the difference plots in Supplementary Figure S2

**Table 4.** *p*-values from ANOVA of the proportions of AXOS and GOS in the three cereal types and cultivars.


Statistically significant values (*p* < 0.05) are given in bold. Some variables required transformation, square root (Sqrt) or loge, to meet the assumptions of the analysis.

**Figure 6.** Multivariate analysis of the contents of AXOS and GOS in grains of the three cereal types grown in four environments. Principal component analysis PCA (**A**) and orthogonal partial least squares discrimination analysis (OPLS-DA) (**B**,**C**) selecting for differences between nitrogen treatments (**B**) and cereal types (**C**). (**D**–**F**) loading plots for (**A**–**C**), respectively. Colour code: red, emmer; green, spelt; blue, bread wheat.

#### **4. Discussion**

We have carried out comparative analyses of the grain of five cultivars each of bread wheat and spelt and three cultivars of emmer, focusing on components that may contribute to health effects. The cultivars selected were all commercially available in Germany at the time of the study, and analyses of flours blended from commercial grain samples and of doughs and breads produced from the blended flours using sourdough and yeastbased systems have been reported elsewhere [16]. The cultivars were grown in replicated randomised field trials at sites in the UK and Netherlands for two years, giving four environments (sites and years). Furthermore, two levels of nitrogen fertiliser were applied to reflect the commercial use of low and high input systems for ancient wheats (emmer, spelt) and modern bread wheat, respectively.

Because only small numbers of cultivars were compared, the results cannot be taken to represent the full range of diversity within the three types of wheat. Nevertheless, wide variation within each type was observed, resulting from the effects of genotype, environment and nitrogen fertilisation. However, because only two sites and years were compared, it was not possible to calculate the separate contributions of genotype, environment and G x E interactions, and the year/site combinations were, therefore, treated as "environments".

The variation in composition within emmer, spelt and bread wheat resulted in overlapping ranges in the contents of all components determined: protein, minerals, polar metabolites and AX and β-glucan fibre. Nevertheless, some statistically significant differences between the three cereal types were observed.

The lower protein content of modern bread wheats observed here is well-established and considered to result from "yield dilution"; the higher yields of modern wheats result from an increased accumulation of starch that dilutes other components [17]. Similarly, modern semi-dwarf bread wheats are known to have lower contents of Fe, Zn and Mg. This may result from the effects of the semi-dwarf phenotype, perhaps combined with some yield dilution [18,19]. The modern bread wheats were also significantly lower in asparagine, the precursor of acrylamide, which formed during processing [12] but also had lower contents of glycine betaine and choline, which have benefits for cardiovascular health by reducing the concentration of homocysteine in blood [20,21]. Hence, it is not possible to conclude that any of the three types of cereal is consistently "better" in terms of its content of polar metabolites.

Published values for the contents of total dietary fibre in whole grains of bread wheat range between 11.5–15.5% dry weight, of which about half (5.5–7.4% dry weight) is AX, with a lower content of β-glucan (0.51–0.96%) [22]. White flour contains significantly less fibre (due to the removal of the bran), about 4% dry weight, with AX and β-glucan accounting for about 70% and 20% of the total, respectively [23].

In the present study, bread wheat had higher contents of AX and β-glucan than emmer and higher AX content than spelt. Although spelt was higher in β-glucan, this component was present in lower concentrations than AX, and hence, the sum of the two types of fibre was highest in bread wheat. A meta-analysis of dietary fibre components in whole grain also showed slightly lower contents of AX in spelt than in bread wheat, with a wider range [24]. Some differences in fibre structure were also observed, with spelt having a lower ratio of G3:G4 GOS released from β-glucan while emmer had a lower ratio of monosubstituted:disubstituted AXOS released from AX. The significance of these differences for the behaviour of the AX and β-glucan fractions in foods and in the gastrointestinal tract is not known.

The bread wheats also had lower contents of raffinose, which is not absorbed in the human small intestine but rapidly fermented in the colon, forming part of the FODMAP fraction, which may contribute to discomfort due to gas production in individuals suffering from non-celiac wheat sensitivity and irritable bowel syndrome (IBS) [25]. However, the relevance of this to symptom control is limited as the major FODMAP fraction in wheat is fructans, which were not measured in the present work.

The effects of nitrogen fertilisation were also determined as it is usual to grown ancient and modern wheats under low input and intensive production systems, respectively. High nitrogen resulted in higher contents of minerals (iron, zinc and magnesium), as reported in a number of studies [26]. Similarly, a significant positive relationship between free asparagine content and total grain protein content has been reported [27,28]. Although only small effects of nitrogen on other components were observed, ANOVA showed that these varied between cereal types. ANOVA also showed interactions between N level and the proportions of AXOS and GOS released in all three types of wheat but no interactions with cultivars within the types.

A number of other studies have compared genotypes of modern and ancient wheats. For example, a series of studies compared the agronomic performance, yield, grain quality traits and contents of a range of "bioactive" components in 15 cultivars each of einkorn, emmer, spelt, bread wheat and durum wheat grown on four sites, with nitrogen fertilisation levels varying between wheat types to reflect commercial practice [29–31]. However, the relevance of such reported differences in composition to human health remains unclear.

This is, at least in part, due to the fact that the relevance of the parameters measured to human health has not been established. For example, although differences between the in vitro antioxidant capacity of cereal flours have been reported [32–34], these cannot be generalized to imply health benefits in humans in vivo [35]. Similarly, glucose released during the digestion of flour in vitro cannot be used to predict glycaemic response in vivo, which is determined by the competing effects of the appearance of glucose and the disappearance (cellular uptake) of glucose in blood [36–38]. As a result, although the glycaemic index (GI) calculated based on in vitro digestion may correlate with that determined in vivo, the absolute values may differ [39]. Such calculations based on in vitro digestion have been reported to result in over-estimation of the in vivo GI by 22% to 50% [40]. Differences in food structure resulting from processing will also affect oral mastication and gastrointestinal digestion. For example, the high density of pasta results in a significantly lower glycaemic response than those of flour and bread [41].

Finally, statistically significant differences in the compositions of cereal flours may not represent significant biological differences when considered in the context of processed food consumed as part of a mixed meal and with other foods consumed over a 24 h intake period.

Comparisons are also often reported using analytical data from different studies and/or data from published food composition tables (for example, [42,43]). This is clearly not valid as the cultivars, growth environments, agronomic practices and methods used for sample preparation and analysis will affect the results obtained.

Taking these factors into account, it is not surprising that different conclusions have been drawn on the health benefits of ancient compared with modern wheats. Thus, whereas Shewry and Hey [24] concluded that based on a comparison of grain samples grown and analysed under the same conditions, there is little evidence that ancient wheats are more "healthy" than modern wheats, Serban et al. [43] suggested that ancient wheat species have health benefits in relation to their nutraceutical composition.

In the context of the data presented here, the small differences in mean compositions of bread wheat, emmer or spelt and their overlapping quantitative ranges are unlikely to result in significant differences in health outcomes, with a possible exception being mineral micronutrients (zinc and iron, which are subject to low intakes in many population groups, with wheat being a significant dietary source [44].

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12040843/s1, Figure S1. Difference plots showing the metabolites contributing to the separations in Figure 4B (A) and Figure 4C (B, C, D). Figure S2. Difference plots showing the AXOS and GOS contributing to the separations in Figure 6A (A) and Figure 6B (B, C). Table S1. Means and SEMs of the contents of minerals and metabolites in grains of individual cultivars of the three cereal types grown in four environments. Table S2. Means and SEMs of minerals and metabolites in cultivars of the three cereal types grown at 100 and 200 kg N/Ha. Table S3. Means and SEMs of contents of polar metabolites in grains of the three cereal types grown in four environments at two Nlevels. Table S4. *p* values from ANOVA of the contents of polar metabolites in grains of the three cereal types grown in four environments at two Nlevels. Table S5. Means and SEMs for AXOS and GOS in grains of the three cereal types grown at 100 and 200 kg N/Ha. Table S6. *p* values from ANOVA of AXOS and GOS in the three types and cultivars of wheat. Statistically significant values (*p* < 0.05) are highlighted. Table S7. Means and SEMs of AXOS in the cultivars of

the three cereals grown in four environments. Some variables required transformation, square root (Sqrt) or loge (log), to meet the assumptions of the analysis. Table S8. Means and SEMs of AXOS in the cultivars of the three cereals grown in four environments with 100 and 200 kg N/Ha.

**Author Contributions:** Conceptualization, F.B., P.R.S., J.P.v.S., L.G., A.L. and D.J.; methodology, A.L., J.D. and J.L.W.; formal analysis, J.D., J.L.W., A.J.B. and J.H.; investigation, J.P.v.S., J.L.W., T.K.P., J.L.W., A.H.P.A., R.T. and Z.A.M.P.-H.; resources, P.R.S. and J.P.v.S.; data curation, J.D., J.H., J.L.W. and Z.A.M.P.-H.; writing—original draft preparation, F.B. and P.R.S.; writing—review and dditing, T.K.P., J.P.v.S., J.L.W., A.L., L.G., A.H.P.A. and D.J.; visualization, supervision, P.R.S., A.L., J.L.W. and D.J.; project administration, F.B. and D.J.; funding acquisition, F.B., P.R.S. and J.P.v.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** "Well on Wheat?" is financed by a grant of the Dutch Government—"TKI- Top Knowledge Institute" and a wide range of additional partners\* from the Agri-Food chain who donated unrestricted research grants. \*These partners are AB-Mauri bakery Ingredients, Made, Netherlands; Borgesius Holding BV-Albert Heijn, Stadskanaal Netherlands; CSM innovation Bakery Center, Bingen, Germany; CYMMIT, Texcoco, Mexico; DSM Food Specialties, Delft, Netherlands; Fazer Bakeries Oy, Helsinki, Finland; Health Grain Forum, Vienna, Austria; ICC- Intl. Vienna, Austria; IWGA, Kansas 66210, USA; Lantmännen EK, Stockholm, Sweden; Mondelez, Saclay, France; Nederlands Bakkerij Centrum, Wageningen, Netherlands; Baking Industry Research Trust, Wellington, New Zealand; Nutrition et Sante, Revel, France; Puratos BV, Groot Bijgaarden, Belgium; Rademaker BV-Bakery equipments, Culemborg, Netherlands; Sonneveld Group BV, Papendrecht, Netherlands; Zeelandia-HJ Doeleman BV, Zierikzee, Netherlands. The project is coordinated and executed by an academic research consortium team (ARCT). Research specialists from funding partners, represented in a Research Steering Team (RST), are entitled to give scientific inputs to ARCT, which on its sole decision, may or may not take inputs into consideration. Decisions on studies set-up, execution and data interpretation of data are exclusively taken by ARCT. Scientific output communications are exclusively organized by ARCT without the involvement of funding partners. Rothamsted Research receives strategic funding from the Biotechnology and Biological Sciences Research Council (BBSRC) and the work forms part of the Designing Future Wheat strategic programme (BB/P016855/1).

**Data Availability Statement:** Full datasets are available from the corresponding author on request.

**Acknowledgments:** We thank Friedrich Longin (University of Hohenheim, Germany) for helping with the provision of grain samples and for dehulling emmer and spelt samples prior to analysis.

**Conflicts of Interest:** Part of the work of DJ outside the submitted work has been financed by publicprivate partnership grants of Top Knowledge Institute (TKI) Agri&Food and Health Holland, the NWO Carbohydrate Competence Center, by Organic A2BV/Mothersfinest BV, EU/FP7 Sysmed-IBD/305564, BIOM/305479 and Character/305676 and H2020 DISCOvERIE/848228. The authors declare no other conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
