The mature seeds (14% moisture level) of two diverse bread wheat (T. aestivum) varieties, “C 306” (good) and “Sonalika” (poor) differing for chapatti (unleavened flat bread) quality were used for the characterization and quantification of phenolic compounds, while their developing seeds were used for quantitative expression analysis of their biosynthesis pathway genes. The details of the two varieties including pedigree and other information were described elsewhere (Singh et al., 2014). The seeds of above varieties were completely dried at 60°C for at least 6 h before extraction of phenolic compounds.

The phenolic compounds were analyzed in three different types of extracts-free (soluble), bound-alkali, and bound-acidic extracts of wheat grains in three replicates. The free (soluble) and bound phenolic compounds were extracted, following the protocol of Adom et al. (2003), with modification. Adom et al. (2003) used both diethyl ether and ethyl acetate for extraction of bound phenolic compounds whereas we used only ethyl acetate for extracting bound phenolics. The final extraction of PCs was done with ethyl acetate and PCs were reconstituted in 80% methanol and dried in a rotary evaporator. The extracted PCs were filtered through a 0.22 μm nylon syringe filter, and stored at −20°C.

Each sample extract was dissolved in 2 ml of 80% (v/v) methanol, filtered through a 0.22 μm nylon syringe filter, and stored at −20°C till injecting into the UPLC system.

Stock solution for each standard was prepared to final concentration of 1.0 mg/ml, filtered through 0.22 μm syringe filter, and stored at −20°C till use. Each standard was sonicated before injecting into the UPLC for analysis. The solution with the highest concentration was prepared by taking 3 μl of each standard and making the final concentration of 3 μg/ml. Different dilutions ranging from 0.005 to 3 μg/ml of each standard solution were prepared from the stock standard mixture for the preparation of linear standard curve.

Phenolic compounds were separated on an ACQUITY™ UPLC system equipped with a binary solvent manager and sample manager (Waters, Milford, MA, USA). The method was developed on an ACQUITY UPLC® BEH C18 column (2.1 × 50 mm, 1.7 μm) equipped with an ACQUITY UPLC BEH guard column containing 0.2 μm metal frit coupled with a TripleTOF® 5600 mass spectrometer (ABSciex, Massachusetts, USA). The spectrometer is a quadrupole-time-of-flight (Q-TOF) which is used for high-resolution quantitative and qualitative analysis of compounds.

The mobile phase consisted of 0.1% formic acid in ACN (solvent A) and 0.1% formic acid in Milli-Q water (solvent B). The gradient for separation was performed as 0–1 min solvent A, 99%; 1–2 min solvent A, 99%; 2–16 min solvent A, 1%; and 16–18 min column equilibration with solvent A, 99% and 18–20 min solvent A, 99%. The flow rate of 300 μl/min was used for separation. MS/MS analysis was done for the fragmentation and identification of the molecules in the sample. Data processing was performed using Analyst TF software (version 1.5.1) and quantification was carried out using Multiquant software (version 2.0.2) and matching was carried out using Peakview software (version 1.2.0.3). The dwell time and scan cycles was automatically set by the software.

The UPLC method used in this study for the identification and quantification of PCs was validated by carrying out linearity, limit of detection (LOD), limit of quantification (LOQ), and precision following the International Conference on Harmonization [(ICH Q2(R1)] guidelines (ICH, 2005).

The expression data of ninety-one probe sets of forty-one genes, which were relevant to phenolic compound biosynthesis pathway, were extracted from our publicly available gene expression data of Affymetrix® Wheat Genome Arrays on the above two bread wheat varieties, “C 306” and “Sonalika” along with the other two bread wheat varieties, “LOK1”(good chapatti quality) and “WH 291” (poor chapatti quality; Singh et al., 2014). The expression data of the four varieties were analyzed for expression potential (heatmap) in three seed developmental stages (7, 14, and 28 days after anthesis) in the software, GENEVESTIGATOR®. The expression data of the two varieties, “C 306” and “Sonalika,” which were used for phenolic compound characterization, were extracted from the published microarray data for differential expression analysis between the good and poor chapatti quality varieties.

The detail protocols of RNA extraction, cDNA synthesis, and quantitative gene expression analysis are described elsewhere (Singh et al., 2014). The primer pairs for quantitative gene expression analysis were designed using Primer Express Software Tool ver. 3.0 (Applied Biosystems, Forster City, CA, USA). The nucleotide sequences of the pathway genes for primer designing were extracted from the GeneChip® Wheat Genome Array (Affymetrix, Santa Clara, USA). ADP- ribosylation factor (ARF) was used as a house-keeping gene for normalization of gene expression. The gene expression analysis was calculated using the method described in Livak and Schmittgen (2001).

Analysis of the molecular ions (MS) at individual peak and the fragments of their main ion of PCs generated by UPLC-QTOF-MS/MS were done by matching the RT, MS, MS/MS fragmentation patterns of their standards. The chemical formulae and the molecular masses of 87 phenolic compounds (Supplementary Table 1) reported in plants were submitted in the Peakview software (version 1.2.0.3). PCs were validated and quantified using their standards. Quantification was done in triplicates for the three extracts of each variety using Multiquant software (version 2.0.2).

Both free (soluble) and bound forms of PCs from six samples of the two diverse wheat varieties for chapatti quality were fractionated, and the bound PCs were further fractionated into alkali-labile and acid-labile components by successive alkaline and acid treatments. Sixty-nine of eighty-seven plant PCs (~79%) were tentatively identified in these extracts by comparing their mass determined on UPLC-QTOF-MS with that of the previously known mass data of the plant PCs (Table 1 and Supplementary Table 1). The UPLC chromatograms of the six extracts were provided in Figure 2 and Supplementary Figure 1. About 50% of plant PCs were detected in all six samples and the remaining PCs were detected in at least one sample. The 69 PCs were grouped into 23 phenolic acids and their derivatives including lignans, 38 flavonoids including anthocyanidins, 4 stilbenoids, 2 chalcones, and 2 coumarins (Table 1).

Out of 69 PCs, the standards of 17 PCs were randomly selected for their validation on UPLC-QTOF-MS/MS. Fourteen PCs (14/17~82%) were validated by matching their RT, MS, and MS/MS (Figure 3 and Supplementary Table 2), while the remaining three PCs such as t-cinnamic acid, benzoic acid, and astringin were not validated. The precursor ion and MS/MS fragmentation patterns of the fourteen compounds (Table 2) were as: vanillic acid (167; 152, 108, 123), vanillin (151; 136, 108, 92), ferulic acid (193; 134, 178), p-coumaric acid (163; 119, 93), caffeic acid (179; 135, 89), quercetin (301; 273, 178, 151), rutin (609; 301, 271), salicylic acid (137; 93), chlorogenic acid (353; 263, 190, 146, 102), hesperidin (611; 303, 465), 2,4-dihydroxybenzoic acid (153; 108), nobiletin (401; 341, 189), sinapic acid (223; 208, 193, 149, 121), and 4-hydroxybenzoic acid (139; 121, 99, 75). Their structural formula was represented in Figure 4.

The identification and characterization of PCs was accomplished using ICH guidelines (ICH, 2005). The correlation coefficient (R²) between different concentration of each standard and their respective peak areas was >0.99 except for ferulic acid (0.93), quercetin (0.98), and vanillic acid (0.96) (Table 2). This also shows that the method developed has good linearity (Table 2). LOD and LOQ values of standards were in the range of 0.1–0.6 ng/ml (lowest for salicylic acid) and 0.3–1.0 ng/ml (lowest for salicylic acid and 2,4-dihydroxybenzoic acid), respectively. Except hesperidin, sinapic acid, and p-hydroxybenzoic acid, the fragmentation pattern, LOD and LOQ of eleven phenolic compounds was carried out in the −ve mode because their respective peak was sharper and their intensity was high in −ve mode as compared to the +ve mode. LOD and LOQ of PCs were very low (within 1 ng/ml) and better than the earlier reported in wheat (>50 ng/ml; Nicoletti et al., 2013). Intra-day variation (Relative standard deviation, RSD) in retention time (RT) for 14 PCs ranged from 0.09 to 2.18% and that for Inter-day from 0.32 to 3.27% (Table 3). For peak area, Intra-day variation (%RSD) for 14 PCs ranged from 0.22 to 2.2% and that for Inter-day from 1.16 to 4.13% (Table 3).

Quantification of the 14 phenolic compounds in the six wheat samples (free, alkaline, and acidic extracts) of two diverse bread wheat varieties, “C306” and “Sonalika” differing for chapatti quality was done on UPLC-QTOF-MS (Table 4). These PCs were highly concentrated in acidic extracts as compared to the free and alkaline extracts. In the poor chapatti quality bread wheat variety, Sonalika, the concentration of the fourteen PCs was higher as compared to that of the good chapatti quality variety, “C306.” Among 14 PCs, ferulic acid was found to be highly concentrated in acidic extract in grains of the variety, “Sonalika” (232.64 ± 2.24 μg/100 g; Table 4). In this variety six PCs such as ferulic acid, salicylic acid, vanillic acid, quercetin, rutin, and vanillin were present in at least 1 ppm (>100 μg/100 g) in grain. Quercetin was not detected in the good chapatti quality variety, “C306.”

The microarray expression data (Affymetrix's GeneChip® Wheat Genome Arrays, Singh et al., 2014) of 91 probe sets for 41 genes related to phenolic compound biosynthesis pathway revealed the temporal distribution pattern of the majority of the probe sets during seed development in the wheat varieties (Figure 5). The expression of about 51% of the probe sets was very low and/or not detected and that of the remaining probe sets were either in early stage or late stage of seed development. The expression level of the probe set, Ta.9320.1.S1_at of caffeoyl CoA O-methyl transferase was high during seed development in all the varieties. The probe set, TaAffx_80118.1.S1_at of 2-hydroxy isoflavanone synthase highly expressed in the poor chapatti quality varieties (“Sonalika” and “WH 291”) during seed development. The five probe sets, namely Ta.8548.1.A1_at, Ta.26595.1.A1_at, Ta.254.1.S1_s_at, Ta.11954.1.A1_at, and Ta.2001.1.S1_at of cinnamoyl-CoA reductase, cinnamoyl alcohol dehydrogenase, benzoate 4-monoxygenase, O-methyl transferase, and catechol O-methyl transferase, respectively, showed strong expression potential in early stage of seed development in the varieties (Figure 5). On the basis of expression potential, the ninety-one probe sets were divided into two groups-I comprising of 37 probe sets and II of 54 probe sets. The expression of the majority of the probe sets in group-IA and IIB showed low expression level (Figure 5).

The fold change analysis showed that out of 91, 44 probe sets representing 23 genes showed ≥1.5-fold change difference in expression between the good (“C 306”) and poor (“Sonalika”) chapatti varieties (Table 5). About 45, 64, and 73% of the probe sets showed the negative expression in the good chapatti variety, “C 306” at early (7 DAA), mid (14 DAA), and late (28 DAA) stage of seed development, respectively. The probe set, TaAffx_80118.1.S1_at of 2-hydroxy isoflavanone synthase showed highly negative expression throughout seed development in the good chapatti quality variety, “C 306.” The four probe sets, Ta.9643.2.S1_x_at, Ta.13798.1.S1_at, Ta.13990.1.S1_at, and TaAffx_80118.1.S1_at, of cinnamate 4-hydroxylase, cinnamyl alcohol dehydrogenase, cinnamoyl CoA reductase, and 2-hydroxy isoflavanone synthase, respectively showed about five-fold negative expression at late stage of seed development in the good chapatti quality variety, “C 306.” The two probe sets, Ta.9172.1.S1_at and Ta.2001.1.S1_at, of chalcone synthase and catechol O-methyl transferase, respectively showed at least four-fold positive expression at late stage of seed development in the good chapatti variety, “C 306.” The chromosome location of the 44 probe sets was determined by homology search with the IWGSC's wheat whole genome sequence databases (Table 5). The chromosome location revealed that the majority of 23 genes were coded by multiple loci.

Real-time expression analysis of seventeen genes related to phenolic compound biosynthesis pathway revealed differential expression during seed developmental stages between the two diverse bread wheat varieties, “C 306” and “Sonalika” differing for chapatti quality (Table 6; Supplementary Table 3; Supplementary Material 1). Eleven genes showed low expression in the good chapatti variety, “C 306” in comparison to the poor chapatti variety, “Sonalika” at mid (14 DAA) and late stage (28 DAA) of seed development (Table 6). One gene, phenylalanine ammonia lyase-5 (PAL-5), showed high expression in the good chapatti variety during seed development. Three genes, 4-coumarate CoA ligase (4-CL), chalcone synthase (CHS), and caffeic acid-O-methyl transferase (COMT) showed low expression at mid stage and high expression in the late stage (28 DAA) of seed development in the good chapatti variety. Two genes, ferulate-5-hydroxylase-1 (F5H-1) and caffeoyl CoA O-methyl transferase (CCoMT-1), showed high expression at mid stage and low expression at the late stage of seed development.

Phenolic compounds are ubiquitous in plant and they represent a large group of biomolecules. PCs have both structural and functional properties. The role of PCs in biotic and abiotic stresses has been extensively studied in plants. Majority of PCs possess health benefits due to their unique structure features. However, their role in processing quality is limited in cereal crops including wheat. The effect of PCs on bread quality has been studied and they play largely negative role. It is presumed that they can also affect other type of end-use food products such as chapatti which is unleavened flat bread. Chapatti is now globally recognized an end-use wheat food product. The characterization of PCs using modern analytical tools such as UPLC-QTOF-MS or MS/MS and functional genomics analysis of their biosynthesis genes in diverse bread wheat varieties differing for chapatti quality have not been studied. In this study, identification and quantifcation of PCs and functional genomics of their biosynthesis genes and their comparative analysis have been studied in a good chapatti variety and a gold standard variety, “C 306” and a poor chapatti variety, “Sonalika.” In this study, a very large number (69) of PCs were identified in their seeds. Dinelli et al. (2009, 2011) had identified ≥70 PCs using a small set of wheat germplasm. Most of the PCs were detected in bound forms (acidic and alkaline extract), which were in agreement with the previous studies (Dinelli et al., 2009, 2011; Vaher et al., 2010 and Nicoletti et al., 2013). In the good chapatti variety, “C 306,” nine PCs (hinokinin, coutaric acid, fertaric acid, p-coumaroylqunic acid, kaempferide, isorhamnetin, epigallocatechin gallate, methyl isoorientin-2′-O-rhamnoside, and cyanidin-3-rutinoside) were identified and these were not identified in the poor chapatti variety, “Sonalika” in this study. Similarly, four PCs (tricin, apigenindin, quercetin-3-O-glucuronide, and myricetin-3-glucoside) were only identified in the poor chapatti variety, “Sonalika.” Therefore, about 20% (9+4 = 13 PCs) of the identified PCs are unique to each other and may be “variety or genotype” specific PCs. They may be informative after further validation.

The concentration of PCs in wheat determined in this study is in the range of (0.03 ± 0.01 μg/100 g) for chlorogenic and p-coumaric acid to (232.64 ± 2.24 μg/100 g) for ferulic acid and is very low in comparison to medicinal herbs as in our study they are present in microgram per 100 g samples where as in medicinal plants they are present in milligram per 100 gram of samples as reported by Chen et al. (2012). However, the content of PCs in wheat food products reported by Vaher et al. (2010) and Nicoletti et al. (2013) was several folds higher than our study except few PCs. This may be due to environmental effect on the wheat plants taken for study and also due to the genotype of the plant. Both these factors affect the phenolic content (Mpofu et al., 2006). Also phenolic content is affected by method of extraction (Verma et al., 2009). Of these 69 PCs, 14 PCs were validated using their standards (Table 2). Retention time (RT) and fragmentation pattern of the PCs were matched with that of their standards. The fragmentation pattern of ferulic acid (193, 178) and caffeic acid (179, 135) were similar to that reported by Wang et al. (2012) and that of vanillic acid (167, 152, 123, 108), sinapic acid (223, 208, 193, 149, 121), chlorogenic acid (353, 191) were similar to that reported by Theerasin and Baker (2009). The concentration of all 14 PCs was found to be higher in acidic extract of “Sonalika.” The concentration of hesperidin was found to be lowest among 14 PCs in all the three extracts of “C 306” and “Sonalika.” Quercetin was present only in three extracts of “Sonalika” while it was absent in “C 306.” Similarly, ferulic acid was absent in alkaline extract of “C 306” while present in remaining extracts of “C 306” and “Sonalika.” It is present in high concentration (232.64 ± 2.24 μg/100 g) in acidic extract of Sonalika which is in agreement with previous studies which shows ferulic acid is most abundant phenolic compound in wheat grain (Moore et al., 2006 and Zilic et al., 2012b). Among plant metabolites, phenolic compounds are most profuse, and ubiquitous having antioxidant activities which positively affect human health (Kim et al., 2006 and Li et al., 2008). Ferulic acid in alkylated form is a potent anti-carcinogen (Imaida et al., 1990) and also exhibit anti-apoptotic effect on peripheral blood mononuclear cells (PBMCS) exposed to H₂O₂ induced oxidative stress (Khanduja et al., 2006). Similarly, caffeic acid and its derivatives act as inhibitor of HIV-1 integrase (Bailly and Cotelle, 2005; Bailly et al., 2005), asthma and allergic reactions (Koshihara et al., 1984). Quercetin is a versatile flavonol with several pharmacological functions such as antioxidant, neurological, antiviral, anticancer, cardiovascular, antimicrobial, anti-inflammatory, hepatoprotective, protective of the reproductive system, and anti-obesity agent (Jan et al., 2010; Dajas, 2012; Joseph and Muralidhara, 2013). Similarly other flavonoids and phenolic acids identified in this study- hesperidin, nobiletin, rutin, vanillic acid, vanillin, sinapic acid, etc. have several health benefits. In this study, all 14 PCs showed high variation among different extracts and between the varieties. Therefore, PCs can be extended to a larger wheat germplasm set for determination of their genetics factors or DNA-based molecular markers using QTL and/or association mapping approaches (Roy et al., 1999, 2010).

Phenolic compounds identified in this study are biosynthesized via phenylpropanoid biosynthesis pathway. In this pathway, PAL is the key enzyme as it converts the amino acid, phenylalanine, into t-cinnamic acid (Nair et al., 2004). t-cinnamic acid is converted into p-coumaric acid by cinnamate-4-hydroxylase (C4H) (Boerjan et al., 2003) and p-coumaric acid is further converted to produce several phenolic acids and flavonoids by a series of enzymes (Figure 1). The genes of phenolic compound biosynthesis pathway enzymes were analyzed for their expression level in the two diverse bread wheat varieties differing for chapatti (unleavened flat bread) using microarray and real-time quantitative PCR (qRT-PCR). In order to study the expression pattern of seventeen phenylpropanoid pathway biosynthesis genes in the two varieties, qRT-PCR was done at two seed developmental stages i.e., 14 DAA and 28 DAA (Table 7). Out of 17, 12 genes were phenolic acid biosynthesizing genes including PAL, vanillin dehydrogenase, vanillin synthase, caffeic acid-O-methyl transferase, caffeoyl CoA O-methyl transferase, hydroxy cinnamoyl-CoA shikimate transferase, 4-hydroxycinnamoyl CoA ligase, ferulate 5 hydroxylase 1, cinnamoyl-CoA reductase, cinnamyl alcohol dehydrogenase, cinnamate-4-hydroxylase 1, p-coumaroyl quinate/shikimate-3-hydroxylase, benzoate-4-monoxygenate and the remaining five genes were flavonoid biosynthesis pathway genes including chalcone synthase, flavone synthase, flavonol synthase, quercetin- 3-O-glucoside L–rhamnosyl transferase, and 1,2-rhamnosyl transferase. At mid stage (14 DAA) of seed development, the expression of 15 genes (83%) was higher whereas at 28 DAA stage, expression of 14 genes (77%) was higher in the poor chapatti variety, “Sonalika,” in comparison to the good quality variety “C 306.” However, about 50% (8 out of 17 genes) of gene expression data of microarrays and qRT-PCR are in agreement i.e., validated at least at the late stage (28 DAA) of seed development. Among them, only chalcone synthase showed high expression in the good chapatti variety and the remaining seven genes showed low expression in the good chapatti variety. Chalcone synthase is a key enzyme in flavonoid/isoflavonoids biosynthesis pathway and catalyses the conversion of p-coumaroyl CoA into naringenin chalcone (Dao et al., 2011; Ma et al., 2016). The seven genes are flavonol synthase, cinnamyl alcohol dehydrogenase, cinnamate 4-hydroxylase, benzoate 4-monoxygenase, caffeoyl- CoA O-methyl transferase, quercetin-3-O-glucoside L-rhamnosyl transferase, and flavone synthase. These genes are responsible for biosynthesis of key enzymes of hydroxy-cinnamic acid, hydoxy-benzoic acid, and flavonoid pathways. Hence the majority of phenolic compounds showed low concentration in good chapatti variety may be collaborated with that of low expression of the seven genes which are identified both through the microarray and qRT-PCR data. These eight genes are candidate genes for developing markers for phenolic compounds-based chapatti quality determination through biparental mapping population. The variation in the expression of the remaining nine genes in qRT-PCR and microarray data (Tables 5, 7) may be due to hypersensitivity of qRT-PCR technique as compared to microarray which sometimes differ the results of the two methods (Etienne et al., 2004). Ma et al. (2016) have studied the expression pattern of PAL, C4H, COMT, C3H, and 4CL which were higher in early stages of seed development and low in later stages of seed development. In our study also gene expression follows the similar trend except for few genes (Table 7). Both data revealed that the majority of genes showed down expression in the good chapatti variety in comparison to the poor chapatti variety. The gene expression level in the good chapatti variety was largely in agreement with that of phenolic compounds. The level of variation of 12 genes between good and poor chapatti quality wheat varieties is high and has potential in development of markers though biparental QTL or association mappings (Roy et al., 1999, 2010).

The relationship was drawn by considering the quantitative gene expression data of the late stage of seed development (i.e., 28 DAA) and quantitative measurement of phenolic compounds. The statistical significance of the relationship was not calculated as we used only two diverse bread wheat varieties for chapatti quality. However, it is found that the expression data of majority of the genes on qRT-PCR corroborated with that of PCs. The up-expression data of seven enzymes genes (7/17 = ~41%) such as B4MO, VDH, VS, C4H1, C3H, FLS, and Q3-RhaT in the poor chapatti quality variety was corroborated with the high content of their immediate products i.e., 4-hydroxybenzoic acid, vanillic acid, vanillin, p-coumaric acid, caffeic acid, quercetin, and rutin, respectively (Table 7). In this study, we corroborated the expression data of only ~41% of genes with the content of their immediate products whereas Ma et al. (2016) found 100% of gene expression data with their immediate products in wheat. Similar relationship was drawn for the expression data retrieved from the microarrays of the late stage of seed development (i.e., 28 DAA) and quantitative measurement of phenolic compounds. It was found that the high expression of five enzyme genes such as B4MO, COMT, C4H1, FLS, and Q3-RhaT was corroborated with the high content of their immediate products such as 4-hydroxy benzoic acid, ferulic acid, sinapic acid, p-coumaric acid, quercetin, and rutin, respectively in the poor chapatti quality. Therefore, in this study the expression level of only four enzyme genes (4/17 = ~24%), B4MO, C4H1, FLS, and Q3-RhaT, was corroborated on both qRT-PCR and microarrays as well as with that of their immediate products (PCs).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.