Phytochemical Analysis of Phenolics, Sterols, and Terpenes in Colored Wheat Grains by Liquid Chromatography with Tandem Mass Spectrometry

The colored grain of wheat (Triticum aestivum L.) contains a large number of polyphenolic compounds that are biologically active ingredients. The purpose of this work was a comparative metabolomic study of extracts from anthocyaninless (control), blue, and deep purple (referred to here as black) grains of seven genetically related wheat lines developed for the grain anthocyanin pigmentation trait. To identify target analytes in ethanol extracts, high-performance liquid chromatography was used in combination with Bruker Daltonics ion trap mass spectrometry. The results showed the presence of 125 biologically active compounds of a phenolic (85) and nonphenolic (40) nature in the grains of T. aestivum (seven lines). Among them, a number of phenolic compounds affiliated with anthocyanins, coumarins, dihydrochalcones, flavan-3-ols, flavanone, flavones, flavonols, hydroxybenzoic acids, hydroxycinnamic acids, isoflavone, lignans, other phenolic acids, stilbenes, and nonphenolic compounds affiliated with alkaloids, carboxylic acids, carotenoids, diterpenoids, essential amino acids, triterpenoids, sterols, nonessential amino acids, phytohormones, purines, and thromboxane receptor antagonists were found in T. aestivum grains for the first time. A comparative analysis of the diversity of the compounds revealed that the lines do not differ from each other in the proportion of phenolic (53.3% to 70.3% of the total number of identified compounds) and nonphenolic compounds (46.7% to 29.7%), but diversity of the compounds was significantly lower in grains of the control line. Even though the lines are genetically closely related and possess similar chemical profiles, some line-specific individual compounds were identified that constitute unique chemical fingerprints and allow to distinguish each line from the six others. Finally, the influence of the genotype on the chemical profiles of the wheat grains is discussed.


Introduction
Among nutritional sources of antioxidant compounds necessary for human health, cereal products, which contain flavonoid pigments (plant compounds of a phenolic na-

Introduction
Among nutritional sources of antioxidant compounds necessary for human health, cereal products, which contain flavonoid pigments (plant compounds of a phenolic nature), are now receiving increasing attention [1]. The biosynthesis of various colored flavonoid compounds in certain grain components of cereal plants gives rise to a distinct color ( Figure 1). As a result of the biosynthesis of anthocyanins, cereal seeds can have a color of various shades from bluish gray and reddish to dark purple and almost black. Other classes of flavonoid compounds give the grain a reddish-brown color (proanthocyanidins) or a dark-brown color (phlobaphenes). Anthocyanins have the highest antioxidant potential among the above-mentioned compounds [2]. These substances can accumulate in vegetative and reproductive parts of the plant, where their main physiological role is to protect the plant from excessive UV radiation. Additionally, the concentration of anthocyanins usually increases during exposure to adverse environmental factors [3]. Anthocyanins have been shown to play an important part in the prevention of neurodegenerative diseases [4], atherosclerosis, diabetes, and obesity and to have vasoprotective and anti-inflammatory properties [5,6]. As a consequence, the food industry is interested in researching colored cereals.
In the most common cereal species, soft wheat (Triticum aestivum L.), the grain can have either an unremarkable color or a reddish-brown, bluish-gray, or purple hue. The differences in color are due to the accumulation of certain flavonoid pigments in various layers of wheat grain envelopes [7][8][9]. The biosynthesis of proanthocyanidins in the seed coat causes a reddish-brown hue (trait: "red grain") and is controlled by R genes localized on chromosomes of homoeologous group 3 [10]. The bluish-gray hue appears to be due to the biosynthesis of anthocyanins in the aleurone layer (trait: "blue aleurone") and is regulated by Ba genes introduced into the common wheat genome from wild relatives Anthocyanins have been shown to play an important part in the prevention of neurodegenerative diseases [4], atherosclerosis, diabetes, and obesity and to have vasoprotective and anti-inflammatory properties [5,6]. As a consequence, the food industry is interested in researching colored cereals.
In the most common cereal species, soft wheat (Triticum aestivum L.), the grain can have either an unremarkable color or a reddish-brown, bluish-gray, or purple hue. The differences in color are due to the accumulation of certain flavonoid pigments in various layers of wheat grain envelopes [7][8][9]. The biosynthesis of proanthocyanidins in the seed coat causes a reddish-brown hue (trait: "red grain") and is controlled by R genes localized on chromosomes of homoeologous group 3 [10]. The bluish-gray hue appears to be due to the biosynthesis of anthocyanins in the aleurone layer (trait: "blue aleurone") and is regulated by Ba genes introduced into the common wheat genome from wild relatives such as wheatgrass Thinopyrum ponticum, Triticum boeoticum, and Thinopyrum bessarabicum owing to translocations in the chromosomes of homoeologous group 4 or via a substitution of one of the chromosomes of homoeologous group 4 [11][12][13]. The purple color is a consequence

Similarities and Differences in Metabolites among the Lines
According to Table 1 and Figure 3, the largest number of biologically active compounds (55) was found in lines S29 BLUE (4Th-4B) and S29 BLACK (4Th-4D), and the smallest (18) in the control line (differences of S29 from all the other lines are significant according to a two-sided test for proportions [Spearman's rank correlation analysis], p = 0.00001-0.0177). Similar data were obtained for the polyphenol family: the largest (33 and 36) and smallest (12) numbers of such compounds were detected in the same lines as mentioned above. Phenolic compounds were found more often than nonphenolic compounds (p = 0.00001-0.0016) in all studied lines except for S29 BLUE (4Th-4D). In this line, the two classes of compounds showed almost equal numbers of members (p = 0.3428). Overall, in terms of the numbers of substances of a phenolic nature (53.3-70.3% of the total number of identified compounds) and a nonphenolic nature (46.7-29.7%) the studied lines were similar.
(p = 0.00001-0.0016) in all studied lines except for S29 BLUE (4Th-4D). In this line, the classes of compounds showed almost equal numbers of members (p = 0.3428). Overal terms of the numbers of substances of a phenolic nature (53.3-70.3% of the total num of identified compounds) and a nonphenolic nature (46.7-29.7%) the studied lines w similar. The results of cluster analysis of all the compounds (Figure 4) showed that two c ters can be distinguished in the dendrogram. The first cluster is formed by lines BLACK (4Th-4B) and S29 BLUE (4Th-4D) and the adjacent S29 BLUE (4Th-4B) line. Plotting of dendrograms separately for phenolic and nonphenolic families of substances indicated that nonphenolic compounds differentiate lines by grain color ( Figure  S1). Even clearer separation by grain color was noted when the lignin family of compounds was utilized for the tree construction. Similar data were obtained on anthocyanins, flavones, and terpenoids. Unambiguous separation by substituted chromosomes was not achieved by means of any one family of substances. In some cases (e.g., for sterols and flavonols), one cluster was distinguished on the basis of the seed color, and the other cluster on the basis of chromosome substitution ( Figure S1).  Plotting of dendrograms separately for phenolic and nonphenolic families of substances indicated that nonphenolic compounds differentiate lines by grain color ( Figure S1). Even clearer separation by grain color was noted when the lignin family of compounds was utilized for the tree construction. Similar data were obtained on anthocyanins, flavones, and terpenoids. Unambiguous separation by substituted chromosomes was not achieved by means of any one family of substances. In some cases (e.g., for sterols and flavonols), one cluster was distinguished on the basis of the seed color, and the other cluster on the basis of chromosome substitution ( Figure S1).
Examination of the chemical composition of wheat grains by the families of compounds within the phenolic and nonphenolic classes revealed that the lower number of biologically active substances detected in the control line can be explained by the absence of seven families of phenolic substances: coumarins, flavan-3-ols, flavanones, flavonols, phenolic acids, dihydrochalcone, and stilbenes. Flavonols were found in all the lines except for the control (S29). Furthermore, in S29, the number of substances belonging to the most numerous (in this study) "flavones" was 1.4-3.2-fold lower as compared to the other lines. The lower number of nonphenolic substances detected in the control line can be explained by the absence of the following families: alkaloids, anabolic steroids, carboxylic acids, carotenoids, cycloartanols, di-and triterpenoids, propionic acids, purines, sesquiterpenoid plant hormones, thromboxane receptor antagonists, and unsaturated fatty acids. Accordingly, the colored-grain lines showed a 3-6-fold greater number of substances in the carboxylic acid family, 2-3-fold in the sterol family, 2-4-fold in the anthocyanin family, and 1.5-2.7-fold in the flavone family as compared to the unpigmented-grain control (S29). It should also be noted that among the phenolic compounds, selgin (from the flavonol family) and abscisic acid [dormin; abscisin II; (S)−(+)-abscisic acid] from the class of nonphenolic compounds (sesquiterpenoid plant hormone family) were found only in lines with a substitution of chromosome 4B. A number of compounds (peonidin-3-O-glucoside, caffeic acid derivative, apigenin, isorhamnetin, kaempferol, rhamnetin II, taxifolin-O-pentoside, salvianolic acid G, undecanedioic acid, cyclopassifloic acid glucoside, sespendole, berberine, and β-sitostenone) were found only in some lines with a substituted 4D chromosome (Table S1,  Among the 125 compounds identified in this study in wheat grains, 58 substances turned out to be unique, that is, each was detectable in only one of the seven analyzed lines. The lowest number of unique compounds (three and four) was found in lines S29 BLACK (4Th-4B) and S29, respectively, and the highest number (17 and 15) in S29 BLACK (4Th-4D) and BW BLACK (4Th-4D). The rest of the lines were somewhere in between. These data are in good agreement with the contribution of the unique compounds to the total pool of detected substances ( Figure 5). It is worth mentioning that the difference between the proportions estimated by ratios-(1) the number of unique substances in a line to the sum of unique substances for all wheat lines under study and (2) the number of unique substances in a line to the total number of biologically active substances in this line-was 3.2-fold in the control line S29: the largest difference among the seven lines ( Figure 5). In all the studied lines, the contribution of phenolic compounds to the pool of unique substances was predominant (60.0-88.2%).
to the sum of unique substances for all wheat lines under study and (2) the number of unique substances in a line to the total number of biologically active substances in this line-was 3.2-fold in the control line S29: the largest difference among the seven lines ( Figure 5). In all the studied lines, the contribution of phenolic compounds to the pool of unique substances was predominant (60.0-88.2%). Effects of various factors on the compounds' diversity in the seven lines were analyzed by one-way ANOVA on ranks (Table 2). It was found that factors "Chromosome Effects of various factors on the compounds' diversity in the seven lines were analyzed by one-way ANOVA on ranks (Table 2). It was found that factors "Chromosome Substitution", "Grain Color", and "Genotype of Line" affect the diversity of the chemical compounds, but "Genotype of Parental Line/Cultivar" does not. A multiple pairwise comparison of proportions of compounds (in the total number of compounds) in the groups of the lines having substituted chromosomes 4B and 4D did not uncover any differences between the lines (p = 0.688, Duncan test). No such differences were revealed in the groups of lines with blue and black colors of grains (p = 0.229, Duncan test). Nonetheless, an effect of an interaction of two factors "Chromosome Substitution × Grain Color" on the diversity of compounds was detected ( Figure 6A). In the group of the lines with substituted chromosome 4D, there were no differences between the blue-and black-grained lines (p = 0.807, Duncan test), whereas in the group of the lines with substituted chromosome 4B, such differences between the blue-and black-grained lines were found (p = 0.0023, Duncan test), with significantly lower diversity of the compounds in the latter group. In the group of blue-grained lines, lower diversity of the compounds was observed in the lines with substituted chromosome 4D, while in the black-grained lines, the effect of chromosome substitutions was opposite: higher diversity ( Figure 6B). mosome 4D, there were no differences between the blue-and black-grained lines (p = 0.807, Duncan test), whereas in the group of the lines with substituted chromosome 4B, such differences between the blue-and black-grained lines were found (p = 0.0023, Duncan test), with significantly lower diversity of the compounds in the latter group. In the group of blue-grained lines, lower diversity of the compounds was observed in the lines with substituted chromosome 4D, while in the black-grained lines, the effect of chromosome substitutions was opposite: higher diversity ( Figure 6B).

Discussion
Successful extraction of polyphenolic compounds depends on two sequential actions: dissolution of each polyphenolic compound at the cellular level in the matrix of plant material and its diffusion into the external medium (the solvent). This is why it is difficult to develop an extraction procedure suitable for all phenolic compounds. For the extraction of phenolic compounds, various organic solvents are commonly used, such as methanol, ethanol, acetone, ethyl acetate, or combinations thereof, often with different proportions of water. Additionally, an important factor directly affecting the solubility and extraction of these compounds is pH of the extraction medium, which determines the solubility of the soluble compounds and affects the possible solubilization of the hydrolyzable fraction.
Liquid chromatography is a versatile and well-established separation technique often employed for a variety of analytical tasks and allowing the separation of fairly complex mixtures of low-and high-molecular-weight compounds. This method is also suitable for different polarities and acid-base properties of various matrices.

Discussion
Successful extraction of polyphenolic compounds depends on two sequential actions: dissolution of each polyphenolic compound at the cellular level in the matrix of plant material and its diffusion into the external medium (the solvent). This is why it is difficult to develop an extraction procedure suitable for all phenolic compounds. For the extraction of phenolic compounds, various organic solvents are commonly used, such as methanol, ethanol, acetone, ethyl acetate, or combinations thereof, often with different proportions of water. Additionally, an important factor directly affecting the solubility and extraction of these compounds is pH of the extraction medium, which determines the solubility of the soluble compounds and affects the possible solubilization of the hydrolyzable fraction.
Liquid chromatography is a versatile and well-established separation technique often employed for a variety of analytical tasks and allowing the separation of fairly complex mixtures of low-and high-molecular-weight compounds. This method is also suitable for different polarities and acid-base properties of various matrices.
The diversity of phytochemicals may underlie diverse biological activities of the raw material. For instance, under the common name anthocyanins, there are up to 600 individual chemicals [39]. Biological activity of some individual anthocyanins has been tested, and distinct effects on physiological processes in animals and humans (or a lack of any) have been described. Antioxidant activity of anthocyanins is reported to be dependent on structural features of the molecules such as the number of hydroxyl and methyl groups and patterns of glycosylation [40]. Among anthocyanins, the highest antioxidant activity is featured by derivatives of delphinidin and cyanidin, followed by derivatives of malvidin, peonidin, pelargonidin, and petunidin [41]. In addition, a glycoside and rutinoside of cyanidin accelerate the regeneration of rhodopsin, while the derivatives of delphinidin have no effect [42]. Anthocyanidins have been demonstrated to be better inhibitors of cell proliferation than anthocyanins [43], with delphinidin and cyanidin having the best growth-inhibitory property and pelargonidin and malvidin devoid of such effects [44,45].
From these observations, we may conclude that the more compounds are present in plant material, the wider is the expected spectrum of biological activities. Investigation of such diversity is a promising field for the development of functional food programs and for pharmacological research.
Here, we compared the diversity of compounds among colored-grain wheat lines and observed that the anthocyaninless line S29 is characterized by the lowest diversity of all the identified compounds, phenolic compounds in particular (Figure 3). The lower diversity of biologically active compounds in S29 is explained by the absence of seven families of phenolics (coumarins, flavan-3-ols, flavanones, flavonols, phenolic acids, dihydrochalcones, and stilbenes) and 12 families of nonphenolic compounds (alkaloids, anabolic steroids, carboxylic acids, carotenoids, cycloartanols, di-and triterpenoids, propionic acids, purines, sesquiterpenoid plant hormones, thromboxane receptor antagonists, and unsaturated fatty acids). These data imply that the genes of wheatgrass chromosome 4Th and chromosome fragments introgressed into 2A and 7D (including the genes regulating anthocyanin biosynthesis) are responsible for the presence of the above compounds in the grain and thus affect the diversity of biologically active substances in the wheat grain.
Although the black-grained lines contain Pp genes in addition to Ba and one may expect an increased number of biologically active compounds in these lines, there were no significant differences in the number of identified compounds between blue-and black-grained lines having chromosome 4D substituted by 4Th; moreover, a statistically significant decrease in the diversity of compounds was observed in the black-grained lines in comparison with the blue-grained lines having a chromosome 4B substitution. According to the results of our one-way ANOVA on ranks, the diversity of the chemicals is affected by such genetic factors as "Chromosome Substitution," "Grain Color," and "Genotype of Line," but not "Genotype of Parental Line/Cultivar." (Table 2) In support of these data, some differences in the chemical profile were noted among the lines with distinct substitutions of chromosomes and among lines with different colors of grains (Table S1). For example, two compounds belonging to the classes of phenolic and nonphenolic substances-selgin and a sesquiterpenoid plant hormone, respectively-were identified only in the lines with substituted chromosome 4B [S29 BLUE (4Th-4B) and S29 BLACK (4Th-4B)]. This observation suggests that this chromosome carries regulatory factors suppressing the synthesis of these compounds. Removing them by substitution of the chromosomes carrying these repressors activates the synthesis of the compounds in the substitution lines. Some common features can be found among the chemical profiles of the lines with similar chromosomes composition. Even though the sister lines of S29 are genetically related (and there is a line based on E22 that has S29 in its pedigree [46]; Figure S2), some line-specific (unique) compounds were identified (Table 1, Figure 5). They constitute unique chemical fingerprints of each line, allowing to distinguish each line from the six others. The unique compounds of each line are hardly explained by the genetic relationships among the lines but can be considered the main reason for the separation of the analyzed lines into two subclusters observed in the dendrogram and the separation of lines S29 BLACK (4Th-4D) and BW BLACK (4Th-4D), which are characterized by the highest percentage of unique compounds (Figures 4 and 5).

Chemicals and Reagents
HPLC grade acetonitrile was purchased from Fisher Scientific (Southborough, UK), and MS grade formic acid from Sigma-Aldrich (Steinheim, Germany). Ultra-pure water was prepared by means of a SIEMENS ULTRA clear (SIEMENS Water Technologies, Munich, Germany), and all other chemicals were of analytical grade.

Fractional Maceration
To obtain highly concentrated extracts, fractional maceration was employed. In this technique, the total amount of an extractant (reagent grade ethyl alcohol) is divided into three parts and is sequentially applied to grains (first, the first part, then with the second and third). The infusion time for each part of the extractant was 14 days.

Liquid Chromatography
HPLC was performed on a Shimadzu LC-20 Prominence HPLC system (Shimadzu, Tokyo, Japan) equipped with a UV sensor and a Shodex ODP-40 4E reverse-phase column for the separation of multicomponent mixtures. The gradient elution program was as follows: from time point 0.01 min to 4.00 min, 100% A; from 4 to 60 min, 100-25% A; from 60 to 75 min, 25-0% A; then, a control wash from 75 to 120 min at 0% A. The entire HPLC analysis was carried out with an ESI detector at wavelengths of 230 and 330 nm; the temperature was set to 17 • C, and the injection volume was 1 mL.

MS
This analysis was performed on an ion trap amaZon SL instrument (Bruker Daltonics, Bremen, Germany) equipped with an electrospray ionization source, in negative ion mode. The following optimal parameters were found and applied: ionization source temperature 70 • C, gas flow 4 L/min, nebulizer gas (atomizer) 7.3 psi, capillary voltage 4500 V, end plate bend voltage 1500 V, fragmentation voltage 280 V, and collision energy 60 eV. The ion trap was used in the scan range m/z 100-1700 for MS and MS/MS. The capture rate was 1 spectrum/s for MS and 3 spectra/s for MS/MS. Data collection was controlled by Hystar Data Analysis 4.1 software (Bruker Daltonics, Bremen, Germany). All the measurements were performed in triplicate. The combination of both ionization modes (positive and negative) in MS full scan mode provided extra confidence of the molecular mass determination. A comprehensive table of molecular masses of the target analytes isolated from the EtOH extracts of T. aestivum grains was compiled by comparing the m/z values, retention times, and the fragmentation patterns with the MS/MS spectral data from the literature [28,29,31,34,[49][50][51][52][53][54][55][56][57], and other sources or from searches of databases (MS2T, MassBank, and HMDB).

Data Analysis
A nonparametric test (Spearman's rank correlation analysis) was performed to compare the wheat lines having different grain colors; for estimation of differences between two groups, we used the two-sided version of the test. We also carried out the Kruskal-Wallis H test (one-way ANOVA on ranks), the Fisher F test (two-way ANOVA), and multiple pairwise analysis (Duncan test) in the STATISTICA 10.0 software [58]. To visualize the obtained data, a dendrogram based on Euclidean distances was drawn by the UPGMA.

Conclusions
As shown by a number of pharmacological studies, single-component drugs cannot be sufficiently effective in the treatment of multifactorial diseases. The mixtures of biologically active compounds that possess an ability to interact with each other often turn out to be more effective against a disease as compared to individual components of the mixture. Bioactive natural products containing a wide variety of compounds are considered more attractive for the production of functional foods and pharmacological research than formulations containing only a few components. Currently, the search for raw materials with a wide variety of biologically active compounds is an urgent task. In the present study, diversity of such compounds was investigated in anthocyanin-rich wheat grains by HPLC-MS/MS. Aside from anthocyanin, the study was focused on identifying other families of compounds of a phenolic and nonphenolic nature. A total of 125 biologically active compounds were identified, and among them, 87 were found in wheat grains for the first time. Statistically significantly higher diversity of the compounds was noted in colored grains of wheat in comparison with a control line, whereas between blue-and black-grained groups of lines, no differences were found. The unique chemical profiles with line-specific compounds were determined for each anthocyanin-rich line. The results make these lines promising sources of functional-food ingredients with a wide spectrum of biological activities.
Supplementary Materials: The following are available online, Figure S1: Dendrograms for seven T. aestivum lines. The trees were built using the UPGMA and Euclidean distance from data on different groups of chemicals, Figure S2: The breeding scheme for the development of the blue-and black-grained wheat lines used in this study, Table S1: The presence of biologically active compounds in the wheat lines grouped by chromosome substitution, grain color, or both.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.

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