Phenylpropanoid Content of Chickpea Seed Coats in Relation to Seed Dormancy

The physical dormancy of seeds is likely to be mediated by the chemical composition and the thickness of the seed coat. Here, we investigate the link between the content of phenylpropanoids (i.e., phenolics and flavonoids) present in the chickpea seed coat and dormancy. The relationship between selected phenolic and flavonoid metabolites of chickpea seed coats and dormancy level was assessed using wild and cultivated chickpea parental genotypes and a derived population of recombinant inbred lines (RILs). The selected phenolic and flavonoid metabolites were analyzed via the LC-MS/MS method. Significant differences in the concentration of certain phenolic acids were found among cultivated (Cicer arietinum, ICC4958) and wild chickpea (Cicer reticulatum, PI489777) parental genotypes. These differences were observed in the contents of gallic, caffeic, vanillic, syringic, p-coumaric, salicylic, and sinapic acids, as well as salicylic acid-2-O-β-d-glucoside and coniferaldehyde. Additionally, significant differences were observed in the flavonoids myricetin, quercetin, luteolin, naringenin, kaempferol, isoorientin, orientin, and isovitexin. When comparing non-dormant and dormant RILs, significant differences were observed in gallic, 3-hydroxybenzoic, syringic, and sinapic acids, as well as the flavonoids quercitrin, quercetin, naringenin, kaempferol, and morin. Phenolic acids were generally more highly concentrated in the wild parental genotype and dormant RILs. We compared the phenylpropanoid content of chickpea seed coats with related legumes, such as pea, lentil, and faba bean. This information could be useful in chickpea breeding programs to reduce dormancy.


Introduction
Chickpea (Cicer arietinum L.) is an annual grain legume crop adapted to dry climates and is the second most important pulse in terms of its production and consumption by humans. Cultivated chickpeas are divided into two types-desi and kabuli. The desi type has small, irregularly shaped seeds with thicker and colored seed coats, in which the seed coat accounts for approximately 14% of the total weight of the seed. The kabuli type produces comparatively larger and smoother round seeds with thinner unpigmented seed coats; the seed coat forms approximately 5% of the total weight of the seed [1,2]. The larger-seeded kabuli type was likely derived from the smaller-seeded desi type [3,4], while seeds of the desi type are very similar to the seeds of the ancestor of cultivated chickpea, C. reticulatum Ladiz. [5]. As in other pulses, chickpea seeds are an excellent In our previous study, we showed that dormancy is not directly related to the thickness of the seed coat in chickpea [4], and it is more likely to be mediated by the chemical composition of the seed coat. Here, we aim to take our investigation of dormancy a step further by analyzing the relative concentration of selected phenolic acids and flavonoids in the seed coats of wild and cultivated genotypes and derived recombinant inbred lines of the chickpea.

Results
The chickpea dormancy status was determined by Sedláková et al. [4], where the imbibition and germination were tested, and several gemination coefficients were calculated. The germination percentage of cultivated ICC4958 was 100% after 24 h, while for wild PI489777 it was 36.1% at 24 h, 60.2% at 72 h, 90.9% at 10 days, and 100% at 30 days. Recombinant inbred lines showed a broader distribution. We have selected the group of 16 non-dormant recombinant inbred lines (RILs) for which germination percentages ranged from 70% to 100% at 24 h, 78.5% to 100% at 72 h, 88% to 100% at 10 days, and 89% to 100% at 30 days. We have also selected the contrasting group of 12 dormant RILs for which the germination percentage ranged from 2% to 46% at 24 h, 4% to 56.9% at 72 h, 22% to 83.9% at 10 days, and 24% to 100% at 30 days.
The average content, standard deviation, and minimal and maximal values of measured data for the non-dormant RIL groups are shown in Table 2 and for the dormant RILs group in Table 3. The Wilcoxon test was applied to investigate the differences between the non-dormant and dormant RILs group. Significant differences between the levels of phenolic acids of seed coats of non-dormant and dormant RILs were found in gallic acid, 3-hydroxybenzoic acid, syringic acid, and sinapic acid (Wilcoxon test, p < 0.05). Higher levels of 3-hydroxybenzoic acid (0.146 ± 0.048 pmol/mg) were in dormant RILs, while higher levels of gallic acid (3.814 ± 3.814 pmol/mg), syringic acid (0.043 ± 0.018 pmol/mg), and sinapic acid (0.369 ± 0.357 pmol/mg) were in non-dormant RILs.  Boxplots of four phenolic compounds with significantly different levels (gallic acid, 3-hydroxybenzoic acid, syringic acid, and sinapic acid) between all non-dormant and dormant RILs are shown in Figure 3. Within the non-dormant and dormant RIL groups, outlying observations are marked with an asterisk. There is variability among groups of non-dormant and dormant RILs in their germination times. The measured values of phenolic acids of selected RILs differed substantially from each other: three non-dormant lines (CRIL2-5, CRIL2-6, and CRIL2-50; 100% final germination percentage (FGP) after 24 h) and three dormant lines (CRIL2-106, CRIL2-60, and CRIL2-131; up to 50% FGP after 30 days) were selected and compared. These lines are considered to be the most representative of the given non-dormant and dormant category in terms of seed responses to water uptake and dormancy levels. The comparison was made for four phenolic acids (gallic acid, 3-hydroxybenzoic acid, syringic acid, and sinapic acid) and their levels showed the most significant changes in the non-dormant and dormant categories. Selected non-dormant and dormant RILs are marked in boxplots ( Figure 3); their average contents are represented by black dots. The average concentrations of gallic acid, syringic acid, and sinapic acid were higher in three selected non-dormant RILs. On the other hand, 3-hydroxybenzoic acid had a higher average concentration in three selected dormant RILs.

Relationship of Non-Dormant and Dormant RILs to Parental Genotypes
The measured contents of phenolic acids in the group of non-dormant RILs corresponded with a parental genotype of the cultivated chickpea (ICC4958). For both cultivated ICC4985 and non-dormant RILs, the most abundant phenolics are 4-hydroxybenzoic and gallic acid with high concentrations of salicylic acid-2-O-β-D-glucoside and salicylic acid. The correspondence between the non-dormant RILs and the cultivated parental genotype was also demonstrated by syringic, caffeic, and 3-hydroxybenzoic acids being the least abundant phenolic compounds (Tables 1 and 2).
The amount of phenolic acids in the group of dormant RILs corresponded to a parental genotype of the wild chickpea (PI489777). For both wild PI489777 and dormant RILs, the most abundant phenolic acid was 4-hydroxybenzoic acid, with very high concentrations of salicylic acid-2-O-β-D-glucoside, gallic, salicylic, and vanillic acids. On the other hand, caffeic, syringic, and 3-hydroxybenzoic acids were the least abundant phenolics in both wild PI489777 and dormant RILs (Tables 1 and 3). The concentrations were similar in the cultivated vs. wild and non-dormant vs. dormant categories.

Contents of Flavonoids and Their Glycosides
In total, 26 flavonoids were analyzed. Of these, 13 were detected in the seed coats of both parental genotypes, except for catechin, which was not detected in wild PI489777, while its level in ICC4958 was 0.273 ± 0.041 pmol/mg.
Boxplots of the five flavonoids (quercitrin, quercetin, naringenin, kaempferol, and morin) with significantly different levels compared between all non-dormant and dormant RILs are shown in Figure 5. Within the non-dormant and dormant RIL groups, outlying observations are marked with an asterisk.   There was variability among the groups of non-dormant and dormant RILs in their germination times. The measured values of flavonoids of selected RILs differed substan-tially from each other; therefore, three non-dormant RILs (CRIL2-5, CRIL2-6, and CRIL2-50; 100% FGP after 24 h) and three dormant RILs (CRIL2-106, CRIL2-60, and CRIL2-131; up to 50% FGP after 30 days) were selected and compared. These lines are considered to be the most representative lines of the given non-dormant and dormant category in terms of seed responses to water uptake and dormancy levels. The comparison was performed for five flavonoids (quercitrin, quercetin, naringenin, kaempferol, and morin); their levels showed the most significant changes in the non-dormant and dormant categories. Selected non-dormant and dormant RILs are marked in boxplots ( Figure 5); their average contents are represented by black dots. All five flavonoids (quercitrin, quercetin, naringenin, kaempferol, and morin) had higher average concentrations in the three selected non-dormant RILs.

Relationship of Non-Dormant and Dormant RILs to Parental Genotypes
The measured contents of flavonoids in the group of non-dormant RILs corresponded to parental genotypes of the cultivated chickpea (ICC4958). For both cultivated ICC4958 and non-dormant RILs, the most abundant flavonoids were myricetin and gallocatechin, and the least abundant were isovitexin and isoorientin (Tables 4 and 5).
Similarly, the amounts of flavonoids in the group of dormant RILs corresponded to the parental genotype of the wild chickpea (PI489777). For both wild PI489777 and dormant RILs, the most abundant flavonoids were gallocatechin and myricetin, with considerably higher concentrations in dormant RILs than in the wild PI489777. A very high concentration of naringenin was detected in both the wild PI489777 and in the dormant RILs. On the other hand, quercetin, isovitexin, isoorientin, and kaempferol are the least abundant flavonoids in both wild PI489777 and dormant RILs (Tables 4 and 6). There were similarities between the concentrations in the cultivated vs. wild and non-dormant vs. dormant categories.

Phenylpropanoid Contents in Relation to Dormancy Status
The PLS-DA biplot ( Figure 6) presents a closer look at the relationship between dormancy status and the amount of phenolic acids and flavonoids present. The first two PLS components explained 59.51% of the variability in phenolic acids and flavonoids. The coefficient of determination for the corresponding model was 0.67. Gallic acid was identified as the metabolite that distinguished most clearly between the non-dormant/cultivated parental genotype and dormant/wild parental genotype groups. Dormant CRIL2-27 deviated the most in this regard, with the amount of gallic acid most similar to non-dormant RILs, i.e., relatively high. There was greater overlap between the groups in the other metabolites with major discriminating effects, such as morin, kaempferol, and quercetin. In particular, these flavonoids are very abundant in dormant CRIL2-48 and CRIL2-21, and are rare in non-dormant CRIL2-65, CRIL2-81, CRIL2-80, CRIL2-45, and ICC4958, which is not in accordance with the majority of observations in the respective groups.

Discussion
Flavonoids and phenolic substances have been extensively analyzed in seeds and seed coats [10,41]. Contrasting pigmentations of seeds are found in many crops as a result of selection for less pigmented seeds during domestication [11,14,[42][43][44]. Pigmentation has a protective role against biotic and abiotic stress; it adversely impacts palatability; and consequently, less pigmented seeds are desirable for cultivation and processing. However, with the increased interest in nutritional aspects, there has been increased interest in pigmentation, as many of these metabolites that contribute to increased pigmentation have health-promoting and antioxidant activities. While the content of secondary metabolites has been studied in seeds of legumes, these studies used only cultivated crops [45][46][47][48][49]. Comparisons of wild progenitors and the respective crops are much rarer [42]. The seed coats of chickpeas (and legumes in general) contain many phytochemicals, especially flavonoids and phenolic acids [8].
Connections between the expression of genes that determine proanthocyanidin and flavonoid biosynthetic pathways and seed dormancy, caused by a water-impermeable seed coat, were shown in Arabidopsis [40], Medicago truncatula Gaertn [50][51][52], and soybean [15]. According to Galussi et al. [53], several water-repelling substances that also join cell walls are responsible for physical dormancy. These substances include polyphenols, lignins, condensed tannins, and pectins, as well as some celluloses and hemicelluloses. Higher concentrations of polyphenolic substances were observed in colored seed coats of soybeans [54,55], lentils [56], peas [45][46][47]49], and chickpeas [48,49]. A relationship between the accumulation of proanthocyanidins, lignins, and the alteration of physical properties of the seed coat leading to cracking was found in soybean [55]. We found that the presence of epicatechin was positively related to the dormancy of seeds, as Zhou et al. [15] and Li et al. [38] demonstrated in soybeans. In addition to genetic factors, the contents of phenolics and flavonoids are influenced by environmental factors such as storage conditions and the access of seeds to oxygen. Moreover, Zhou et al. [15] directly tested the effect of particular phenolic compounds on the germination of soybeans and found a negative effect of epicatechin at higher concentrations.
Wild chickpea, C. reticulatum (PI489777), has medium to dark brown seeds and contained higher concentrations of 10 out of the 13 phenolic compounds that were detected than the genotype of the cultivated chickpea (ICC4958). Corresponding dormant RILs (considered as more pigmented) contained higher concentrations of phenolic compounds. In the case of flavonoids, only 4 out of the 13 detected substances (31%) had a higher concentration in the wild seeds of PI489777. This corresponded with non-dormant RILs (considered as less pigmented) having higher flavonoid contents than dormant RILs, similarly to cultivated ICC4958 having higher flavonoid contents than wild PI489777. However, the differences are not that distinct, and these relatively minor differences between wild and cultivated genotypes could be due to the cultivated parental genotype being of the desi type, which is more pigmented compared with the kabuli type, and thus contained higher contents of polyphenolic substances (13-fold more phenolic acids and 10-11-fold more flavonoids) [48]. Higher phenolic contents in wild and cultivated desi chickpeas in comparison to cultivated kabuli chickpeas were also found by Kaur et al. [57]. Moreover, there was also a difference in the rate of imbibing between kabuli and desi chickpea types. Non-pigmented kabuli seeds imbibed more quickly (within 4-8 h), whereas pigmented desi seeds imbibe more slowly [58] (within 24 h); this is partially related to differences in the thickness of their respective seed coats [59].
The most commonly detected phenolic acids in chickpeas were dihydroxybenzoic acid, p-coumaric acid, gallic acid, chlorogenic acid, ferulic acid, 4-hydroxybenzoic acid, and syringic acid. Among the most commonly detected flavonoids observed in chickpeas are pinocembrin, quercetin, catechin, luteolin, and myricetin [8,[60][61][62]. Elessawy et al. [63] reported that chickpea and pea seed coats had the most similar compositions of polyphenolic substances among the legumes they compared. The most important phenolic acids and flavonoids in peas were protocatechin, vanillic acid, syringic acid, caffeic acid, ferulic acid, and p-coumaric acid [8,12,45]. Troszynska and Ciska [45] also observed differences in phenols between colored and white seed coats of peas, where protocatechuic, gentisic, and vanillic acids were dominant in colored peas, while hydroxycinnamic, ferulic, and coumaric acids were the most abundant in white-seeded peas. These findings correspond with our results, where vanillic acid was not detected in cultivated ICC4958 and several non-dormant, less pigmented seeds, while wild PI489777 and dormant RILs with more pigmented seeds had generally higher concentrations. A comparison of parental genotypes showed a higher concentration of coumaric acid in cultivated ICC4958, which corresponded to Troszynska and Ciska's [45] findings of more coumaric acid in less pigmented genotypes. In contrast to Troszynska and Ciska [45], however, ferulic acid had higher levels in dormant RILs with more pigmented seed coats. Jha et al. [47] reported that gallic and caffeic acids are present only in seeds of a purple-flowered variety of pea and the corresponding RILs with pigmented seed coats, and are absent in varieties with white flowers and non-pigmented seeds and their corresponding RILs. They also mentioned the higher concentrations of epigallocatechin, vanillic, and 3,4-dihydroxybenzoic acids in the pigmented seed coats of pea. Our findings do not completely correspond with the results of Jha et al. [47]. However, these researchers worked only with domesticated, e.g., non-dormant genotypes, differing in pigmentation. Here, we detected p-coumaric acid, gallic acid, ferulic acid, 3-hydroxybenzoic and 4-hydroxybenzoic acid, syringic acid, catechin, quercetin, luteolin, and myricetin, corresponding to the results of Aguilera et al. [60], Fratianni et al. [61], and Magalhaes et al. [62]. According to Amarowicz and Pegg [64], legume seed coats primarily contain 4-hydroxybenzoic acid, protocatechin, gallic acid, vanillic acid, and syringic acid; however, vanillic acid was not detected in cultivated ICC4958 and nine RILs (regardless of the dormancy state). Gallic acid, which was mentioned above, was also not detected in several dormant RILs. In lentil genotypes, vanillic acid glucoside and gallic acid were identified as the most abundant phenolic acids in seed coats [65]; in terms of significance, both vanillic acid and gallic acid are ranked first in bean seed coats [8]. Lentil genotypes with colored seed coats are also abundant with flavonoids tricetin and luteolin [56]. On the other hand, syringic acid, which was previously identified as the primary substance found in legume seeds [64], was present in both parental chickpea genotypes and all derived RILs. In contrast to this finding, Elessawy et al. [63] did not identify syringic acid in the seed coats of the analyzed legume genotypes. This was further supported by Jha et al. [47], where none of the pea genotypes, regardless of seed coat pigmentation, contained syringic acid. In our study, we did not observe a significant difference in ferulic acid concentration between parental genotypes and RILs, which does not correspond with the finding of Jha et al. [47] where ferulic acid content was significantly higher in white flowered peas with non-pigmented seed coat. According to the study by Quintero-Soto et al. [66], the most abundant flavonoid in cultivated desi chickpea is catechin, which is assumed to be present mainly in colored chickpea seed coats. In our study, the catechin levels in cultivated ICC4958 were not very significant, but catechin was absent in wild PI489777 and several dormant RILs. This finding suggests that in the case of catechin, pigmentation probably does not play a crucial role; instead, catechin has been introduced to cultivated chickpeas by domestication. In contrast, a higher abundance of catechin was observed in wild soybeans [38]. Low-tannin species of lentil are also very low in levels of catechin [47,67].
All these metabolites are largely soluble, but there are also more complex polymerized insoluble phenolics in the seed coats of peas [11], faba beans [68], and lentils [69]. Although their direct involvement in seed-coat-imposed dormancy was not confirmed in peas [11], they still might contribute to this by providing substrates for lignin or suberin pathways for impregnating the seed coat.
Notable is the high content of salicylic acid and its glucoside in the seed coats of wild chickpea. Salicylic acid is known to be involved in plant responses to certain pathogens; it regulates diverse aspects of the plant's responses to abiotic stresses through extensive interactions with other growth hormones [70]. Salicylic acid also plays a role in germination under stressful conditions, although its precise role and the underlying molecular mechanisms have not been fully elucidated.
We observed differences in the contents of selected phenolic acids and flavonoids between the seed coats of contrasting wild and cultivated chickpeas. We found that the differences were mainly in the presence of gallic acid, syringic acid, and sinapic acid for phenolic compounds, and quercetin, kaempferol, and naringenin for flavonoids. We noted particularly that the wild parental genotype and corresponding dormant RILs contained generally higher levels of phenolic acids and lower levels of flavonoids. The biosynthesis of phenolic acids is apparently more pronounced in dormant chickpea genotypes, and there is a possible relationship between the phenolic content and dormancy of the seeds. Metabolic and chemical compositions of the seed coats of chickpeas in relation to their germination, together with information about the anatomical structure, could be employed in further chickpea-breeding programs. This may prove especially useful in relation to dormancy-breaking mechanisms, which are essential for the successful establishment of crops. The presence of polyphenolic substances in the seed coats of chickpeas could also be part of plant breeding programs, due to their nutritional possibilities.

Seed Material
Seeds were obtained from Dr C. Coyne, USDA Pullman, USA, and originated from field-grown plants from the 2021 season on Central Ferry Farm, Pullman, WA, USA. Seed coats were dissected from dry seeds (approx. 7 seeds, depending on their size) of parental genotypes-cultivated non-dormant C. arietinum (ICC4958) and wild dormant C. reticulatum (PI489777). In addition, 16

Analysis of Phenolic Metabolites
Homogenized seed coats (≈20 mg) were mixed with 1 mL of solvent (acetone:water:acetic acid, 80:19:1) and treated for 10 min in an ultrasonic bath. After spinning in a centrifuge at 14,500× g, the supernatant was transferred into the new vial and kept at −20 • C until analyzed. Analysis of free phenolic acids and flavonoids was performed according to the protocol described inĆavar Zeljković et al. [71,72]. Briefly, LC-MS/MS measurements were carried out using an Ultra Performance LCMS 8050 system (Shimadzu, Kyoto, Japan) with a triple quadrupole mass spectrometer equipped with an electrospray ionization (ESI) source operating in negative mode. The samples were injected (5 µL) into a reversed-phase column (Acquity UPLC BEH C18, 1.7 µm, 2.1 × 100 mm, Waters, Milford, MA, USA) with a corresponding pre-column (Acquity UPLC BEH C18 VanGuard pre-column, 1.7 µm, 2.1 mm × 5 mm). The mobile phase consisted of a mixture of 15 mM formic acid (pH 3, adjusted with NH 4 OH) (solvent A) and ACN (solvent B) at a flow rate of 0.4 mL/min. The linear gradient consisted of 10% B for 1 min, 10-13% B for 2 min, isocratic 13% B for 4 min, 13-25% B for 3 min, 25-70% B for 1.2 min, isocratic 75% B for 0.8 min, back to 10% B within 0.5 min, and equilibration for 3.5 min. The effluent was introduced into an electrospray source (interface temperature of 300 • C, heat block temperature of 400 • C, and capillary voltage of 3.0 kV). To achieve high specificity in addition to high sensitivity, the analysis was performed in multiple reaction monitoring (MRM) mode. All standards and reagents were of the highest available purity and purchased from Sigma Aldrich Company (Prague, Czech Republic).

Statistical Analyses
Statistical analyses were performed in RStudio (R Software ver. 4.1.0) using the packages corrplot, ggplot2, and pls. Values for phenolic acids and flavonoids that were not detected were replaced by two-thirds of the minimal detected value in the respective variable in order to carry out the subsequent statistical analysis. The relative differences (in log-scale) in the content of phenolic acids and flavonoids between two parental genotypes were computed and shown with barplots. In addition to the differences in the mean values of the three replicates, the largest and smallest differences in the individual replicates were displayed; thus, the significant difference could be calculated (i.e., points below zero (and points above zero) indicate that the corresponding Wilcoxon test yields a p-value of 0.1, the smallest achievable p-value with this test when comparing two groups with a sample size of 3). The Wilcoxon test was applied to determine the differences between the non-dormant and dormant RILs groups. Boxplots for the phenolic acids and flavonoids showing significant differences (at a statistical significance level α = 0.05) between the groups were derived. The data were log-transformed and a PLS-DA biplot was constructed (taking dormancy as the response).