Introduction

Chickpea is well known for its high protein content (24.6%) and is consumed as a substitute of animal protein1, 2. Despite having a narrow genetic base, chickpea shows a wide variation in seed size/weight in the germplasm collections3. Seed size is the most desirable trait in chickpea that determines its market price and economic importance in trade and commerce. Apart from that, chickpea seed size is also considered as an important factor for germination, growth, vigor and yield of its subsequent generation of plants4, 5. Therefore, seed size is considered as an important objective for improving yield and product quality in chickpea breeding programs. Seed weight was also proposed as a reliable measure of chickpea seed size6. Several earlier conventional genetics and breeding efforts reported various complex genetic inheritance pattern of seed size/weight such as, high7, 8 and low9 heritability for seed size, dominance of small seeds over large seeds8, 10,11,12 and vice-versa7. Further, seed size/weight inheritance was reported to be monogenic13, digenic6, 14, oligogenic15, and polygenic7, 8, 10, 12, 16.

In view of complex inheritance pattern and wider phenotypic variation in natural germplasm accessions, multiple genomics-assisted breeding strategies including QTL (quantitative trait loci) mapping and genetic association analysis have been deployed to identify major QTLs and candidate genes governing seed size/weight in chickpea by using intra- and inter-specific chickpea mapping populations17,18,19,20,21,22,23,24. A recent study reported two significant seed weight QTLs at two genomic regions, one spanning from 3.07 Mb to 4.15 Mb on LG1 and the other, from 11.12 Mb to13.82 Mb on LG4 of chickpea genome assembly2. In spite of the considerable progress in QTL and association mapping any potential gene regulating seed size/weight has not been delineated and utilized in marker-assisted selection (MAS) so far to develop cultivars with high seed weight and yield in chickpea.

About two-third of phosphorus in cereal and legume seeds is stored in the form of phytate [salt form of phytic acid (PA), myo-inositol 1,2,3,4,5,6-hexakisphosphate, IP6]25. Myo-inositol is a derivative of glucose. Biosynthesis of myo-insitol-1-phosphate from glucose-6-phosphate is catalyzed by myo-inositol phosphate synthase (MIPS). Myo-inositol monophosphatase (IMP) catalyzes the last step of myo-inositol biosynthesis by dephosphorylation of myo-insitol-1-phosphate. This enzyme is also responsible for recycling of myo-inositol by dephosphorylating other inositol compounds26, 27. Recently, repeat-length variation of an SSR (simple sequence repeat) marker (NCPGR90) in the 5′-untranslated region (5′UTR) of CaIMP gene (LOC101496687) was reported to exhibit 13% association with seed phytate content within a natural population of sixty germplasm accessions of chickpea28. A systematic effort has been made in the current study to infer a three way association of a CaIMP gene-derived SSR marker (NCPGR90) with seed weight (SDW) and seed-phytate content using precisely field-phenotyped germplasm accessions and reciprocal intra-specific mapping populations of chickpea. Further, using a chickpea transient system, we demonstrated that this repeat length variation in CaIMP influences translation efficiency of the transcript resulting in differential CaIMP protein expression.

Results

Repeat length variation of a SSR marker NCPGR90 contributes to chickpea seed weight

A collection of genetically diverse 71 chickpea germplasm accessions including landraces with wide geographical distribution across 20 countries of the world was gathered20. A wide phenotypic variation for 100-seed weight ranging from 4.53 to 70.3 g including a normal frequency distribution of target trait was observed in these accessions (Table S1 and Fig. S1A). A SSR marker (NCPGR90) with CT repeat was amplified, genotyped and sequenced among 71 accessions for assessing its potential for fragment length polymorphism across these chickpea accessions. The marker essentially detected seven different CT-repeat length variations (CT10, CT11, CT27, CT29, CT31, CT32, and CT34) (Fig. S1B). The SSR marker produced two major groups of amplicons with fragment size lengths of about 180 bp and about 220 bp that collectively were named as NCPGR90_180 and NCPGR90_220, respectively (Fig. S2A). Notably, seven wild accessions of C. reticulatum and C. echinospermum, belonging to chickpea primary gene pool, possessed short repeats at the SSR marker locus (Fig. S2B). More than 80% of the traditional landraces with geographical locations outside Indian subcontinent used in this study showed presence of NCPGR90_220 alleles (Table S1).

The association of CaIMP gene-derived SSR marker (NCPGR90) with 100 SDW was studied using both GLM and MLM approaches. Using native model (GLM_marker + trait) and considering structured population (GLM_marker + trait + Q) and/or kinship by MLM, the NCPGR90 SSR marker alleles exhibited significant association with 100 SDW (R2 = 0.30, p = 0.00002). NCPGR90_220 and NCPGR90_180 marker alleles explained 71.9% (p = 6.1 × 10−7) and 28.1% (p = 2 × 10−5) of the total seed weight phenotypic variation observed in chickpea germplasm accessions, respectively (Fig. 1 and Table S2).

Figure 1
figure 1

Association analysis of NCPGR90 with seed weight and phytate: Three-way association analysis of NCPGR90, 100 SDW and phytate content with each other. Association coefficients (R) are mentioned for each pair (A). Tables showing data of association analysis of NCPGR90 with 100 SDW in 71 chickpea germplasm accessions, F2 mapping population (ICCX-810800 × ICCV95334) and F5 mapping population (ICCV95334 × ICCX-810800) (B) association analysis of 100 SDW with seed phytate content (mg PA/gm dry weight) in 52 chickpea germplasm accessions and 96 individuals of F5 mapping population (ICCV95334 × ICCX-810800) (C) and association analysis of NCPGR90 with seed phytate content in 52 chickpea germplasm accessions (D).

Full size image

Two intra-specific reciprocal crosses were made using kabuli and desi accessions ICCV95334 (CT31, 100 SDW = 51.3 g) and ICCX-810800 (CT11, 100 SDW = 11.6 g), respectively, as parents with contrasting 100 SDW and SSR repeat-length variations. A representative set of 192 mapping individuals derived from each of F2 (ICCX-810800 × ICCV95334) and F5 (ICCV95334 × ICCX-810800) intra-specific populations exhibiting a broader 100 SDW trait variation and normal frequency distribution along with transgressive segregation of target trait (Tables S3 and S4 and Figs S3 and S4) were used to assess association of NCPGR90 SSR marker with 100 SDW following the aforesaid methods. Based on this analysis, the CaIMP gene-based SSR marker exhibited significant association (R2 = 0.27 to 0.31, p = 1.48 × 10−13 to 3.58 × 10−16) with 100 SDW in both mapping populations developed in chickpea (Fig. 1 and Table S2).

Chickpea seed weight is strongly associated with its phytate content

Association of seed weight with seed phytate content was assessed in 52 genetically diverse chickpea germplasm accessions with 100 SDW ranging from 5.8–71.7 g and seed phytate content ranged from 12.26–27.9 mg/g seed dry weight (Table S5). A significant association (R2 = 0.78, p = 6.2 × 10−12) was observed between seed weight and its phytate content, which was further validated (R2 = 0.71, p = 1.9 × 10−16) in the F5 mapping population (Fig. 1 and Table S6) of (ICCV95334 × ICCX-810800). As reported before28, NCPGR90 showed a similar significant association (R2 = 0.29, p = 2.75 × 10−5) with seed phytate content in these 52 germplasm accessions (Fig. 1 and Table S5).

Chickpea accessions with longer NCPGR90 repeat expressed higher CaIMP protein

The CT repeat of NCPGR90 is located at the 5′UTR of chickpea IMP gene (CaIMP) and initiates after 68 bases from the transcription start site29. CaIMP (LOC101496687) is located between 6,150,779 bp and 6,155,608 bp on LG 1 of chickpea genome assembly and is close to one significant seed weight QTL mentioned above30. A representative of 10 accessions with various NCPGR90 SSR repeat length and different seed weight were randomly picked for qRT-PCR to compare transcript level of CaIMP gene at normal growth condition. A maximum of 1.6 fold difference in the steady-state transcript abundance of CaIMP gene was observed in these accessions. Similarly, eight random mapping individuals and two parents of F5 intra-specific population of (ICCV95334 × ICCX-810800) with homozygous and heterozygous NCPGR90 alleles and different seed weights did not show any significant variation in transcript abundance of CaIMP gene (Fig. 2). Several lines of evidences documented the importance of sequence and structure of 5′UTR in determining level of protein expression31,32,33. Therefore, CaIMP protein level in the same chickpea germplasm accessions and mapping individuals/parents was assessed by western blotting. Remarkably huge variation in the CaIMP protein expression level was observed in these accessions irrespective of their corresponding transcript levels, with higher CaIMP protein levels were observed in the accessions with longer CT repeats and higher seed weight. Same trend of differential CaIMP protein level was observed in the F5 mapping individuals and parents with different seed weight. Specific activity of CaIMP enzyme was determined in the same germplasm accessions and mapping individuals. CaIMP possesses broad substrate specificity however; it showed maximum activity with myo-inositol 1-phosphate in in vitro system29. Myo-inositol 1-phosphate was used as a substrate to measure activity of CaIMP in the germplasm accessions and mapping individuals mentioned above. CaIMP activity in these accessions/individuals ranged from 29.6–13.9 μmol of inorganic phosphate released/mg/min. Although, the fold of catalytic activity of CaIMP did not accurately follow the expression levels of CaIMP protein in these accessions/individuals, however, those expressing higher CaIMP protein exhibited higher CaIMP enzymatic activities than those having lower CaIMP protein level. CaIMP catalyzes the last step of biosynthesis of myo-inositol, which is a precursor of phytic acid. Therefore, phytic acid content was measured in the same accessions and F5 mapping individuals. Phytic acid content ranged from 27.9–17.8 mg/g of seed dry weight in these accessions/individuals. The accessions/individuals with higher CaIMP protein level and activity exhibited higher seed phytic acid content (Fig. 2).

Figure 2
figure 2

Expression analysis of CaIMP in accessions and mapping individuals: qRT-PCR analysis (top) and western blot analysis (bottom) of CaIMP gene and protein expression in 10 different chickpea germplasm accessions (A) and in 8 individuals of F5 mapping population of ICCV95334 × ICCX-810800 (B). CT repeat number, 100 Seed weight (100 SDW in gram), CaIMP specific activity (μmol Pi mg−1 protein min−1), and seed phytate content (mg PA/gm) of 10 chickpea germplasm accessions and 8 individuals of mapping population are shown. Chickpea Elongation factor 1-α (CaEF-1α) was used as internal control for normalization in qRT-PCR and RuBisCo was used as loading control for western blot. Standard deviation was determined from three biological replicates.

Full size image

Repeat length variation in CaIMP gene influenced its protein level

Variation in CaIMP protein level in different chickpea germplasms could have also resulted from the variation in the CaIMP transcript level and/or due to their different genetic background. Therefore, a transient expression system was used to investigate any role of length variation in CT repeat within the 5′UTR of CaIMP gene in its expression under the same genetic background. A 1.5 kb DNA fragment upstream to the translation start site of CaIMP having CT31 or CT11 was inserted in to a binary plasmid to drive expression of GUS reporter gene. These constructs were used to transiently transform young chickpea leaves of one mapping parent ICCV95334 along with another plasmid expressing GFP reporter gene for normalization of transformation efficiency. Normalized transcript abundance of GUS gene from two constructs varying only in CT repeat length did not show any significant difference. However, normalized GUS activity, a measure of GUS protein expression level, in the samples transformed with the construct having CT31 was about 2.7-fold higher than that in the samples transformed with the construct having CT11 in three biological replicates (Fig. 3A) suggesting that the CT repeat length variation in CaIMP 5′UTR affected translation instead of transcription.

Figure 3
figure 3

Transient expression analysis of CaIMP: (A) Beta-glucuronidase (GUS) was transiently expressed by agroinfiltration in young leaves of chickpea ICCV95334 under the control of 1.5 kb CaIMP promoter and 5′UTR containing CT11 or CT31. GUS mRNA level was assessed by qRT-PCR. CaEF-1α gene was used as internal control (a). Specific activity of GUS enzyme was determined for GUS protein expression (b). Standard deviation was determined from three biological replicates. Fold expression of a Green fluorescence protein (GFP) gene from a plasmid agroinfiltrated in two leaf samples simultaneously with the GUS constructs was determined by qRT-PCR and was used to normalize transformation efficiency. (B) Secondary structure and folding energy (dG) of first 200 base CaIMP mRNA from (a) ICCX-810800 (CT11) (b) ICCV95334 (CT31) was predicted using mfold tool. The CT repeat and the translation start site are shown by arc and arrow, respectively.

Full size image

Folding of mRNA at the beginning of genes is one of the major factors that affect translation initiation and, thereby, protein level. Lower structural stability (higher folding energy) of mRNA near the translation start codon was suggested to favor translation initiation34,35,36. The secondary structure and folding energy of CaIMP mRNA was predicted using mfold tool37. Predicted folding energy (dG) of the first 200 nt of CaIMP mRNA with CT31was much higher (−19.22 kcal/mol) than that of the mRNA with CT11 (−26.28 kcal/mol) (Fig. 3B) suggesting that CaIMP mRNA with CT31 would be more favored for translation initiation than that with CT11.

Discussion

In the present study, we used a set of diverse chickpea germplasm accessions and mapping populations derived from two intra-specific reciprocal crosses of two accessions with contrasting seed weight trait to propose that length variation of an SSR in the 5′UTR of CaIMP gene is strongly associated with seed size/weight. Previously, the same repeat element was shown to be moderately associated with phytate content and drought tolerance using a limited set of germplasm accessions of chickpea. In the previous report, repeat length variation of the SSR marker was correlated with differential transcript level of CaIMP in a genetically diverse set of chickpea accessions. However, transcript level of a gene in different accessions may vary due to various reasons including differential activities of the proteins regulating transcription of that gene and different genetic backgrounds of the accessions. In this study, we have demonstrated using a transient chickpea system in a single genetic background that the SSR repeat length variation affected rather translation than transcription of CaIMP. We have also shown that CaIMP protein level and its enzymatic activity corresponded well with the CT-repeat length, seed phytate level and seed weight in natural population of chickpea.

A SNP-based genome-wide genotyping approach to identify major QTLs governing chickpea seed weight has delineated a genomic region spanning from 3.07 Mb to 4.15 Mb on LG1. However, none of the five SNPs identified at this QTL region was in predicted genes probably due to narrow candidate region2. The CaIMP gene that resides within 2 Mb of this region (LOC101496687, CaLG01: 6,150,779–6,155,608) could be a potential candidate for this QTL responsible for seed weight variation in chickpea.

Phytic acid is a ubiquitous storage form of phosphate in eukaryotes. It regulates many physiological functions such as stress response, development, phosphate sensing and homeostasis38. Phosphates released from phytates are important for signal transduction, DNA synthesis and regeneration of adenosine triphosphate (ATP) and thus making it an essential energy source for seeds39. Previously, in cereals, mutations in various other genes were shown to reduce phytate content and seed weight40,41,42. Here we propose that one of the mechanisms that natural chickpea population use to regulate seed size and phytate content is through variation in CT repeat length at the 5′UTR of the CaIMP gene and thus, by variation in CaIMP protein level. Although adequate dietary phosphate is essential for health, a reduction in phytic acid content in cereal and legume grains is desired because besides sequestering inorganic phosphate, phytic acid chelates essential minerals causing health problem43, 44. Several attempts were made before to reduce phytic acid content in seeds by isolating low phytic acid mutants and raising transgenic plants. However, these efforts were usually accompanied with compromised agronomic performance of the seeds and plants45, 46. Varietal difference of IMP protein expression due to repeat length variation in the 5′UTR and associated seed size variation have not been reported in other plants. 5′UTR of the corresponding IMP gene (Medtr2g026060) in Medicago truncatula possesses CT5 whereas, its orthologs in soybean (Glyma15g115500) and common bean (Phvul03g084500) showed presence of CT2CA3 and CTCACT2, respectively, in the similar positions47,48,49. Therefore, it appears that presence of CT/CA repeats in the 5′UTR of IMP gene is an evolutionary feature in legumes, particularly, in chickpea.

To summarize, we have shown a three-way association between a polymorphic SSR at the 5′UTR of CaIMP gene, seed weight and seed phytate content in chickpea using a set of genetically diverse germplasm accessions and two mapping populations derived from reciprocal crosses. Chickpea germplasm accessions and mapping individual with longer CT repeat exhibited higher CaIMP protein level and enzymatic activity, and seed weight and phytate content. Using chickpea transient system we have shown that CT repeat length variation influenced translation of CaIMP gene, and mRNA fold energy predicted a preferred translation initiation for CaIMP mRNA with longer CT repeat. Phytate content is reported to regulate many physiological properties including seed viability, and increased seed phytate level has negative effects on health. High chickpea seed weight is a desirable market trait. Therefore, CT repeat length variation at the 5′UTR of CaIMP can be used as a valuable marker for selection of chickpea lines with desirable seed trait without compromising the plant performance.

Methods

Seventy one germplasm accessions representing various chickpea agro-ecological growing areas of the world, collected from International Crop Research Institute for Semi-Arid Tropics (ICRISAT), India were used for genetic association analysis. Additionally, 192 mapping individuals represented from each of a F2 and a F5 intra-specific mapping population (desi accession ICCX-810800 × kabuli accession ICCV95334) and (ICCV95334 × ICCX-810800), respectively with contrasting 100-seed weight trait were included. These germplasm accessions and mapping individuals along with mapping parental accessions were phenotyped for 100-seed weight (g) in field at two geographical locations (ICRISAT and Delhi) in India as described before20. DNA was extracted from fresh leaf tissues of one month-old plants grown in the field using the cetyltrimethyl ammonium bromide (CTAB) method50. All the germplasm accessions and mapping individuals were genotyped with the SSR marker (NCPGR90) using the primer pair Ca14825.1 F: 5′-TATAGAGAGAGAAAGAGAGAGG-3′; Ca14825.1R: 5′-CTAAGAGCACATACGGTTTTGT-3′. Quantitative real-time PCR (qRT-PCR) was carried out according to the procedure described before51. The expression of CaIMP mRNA was calculated according to delta-delta Ct method of the system. Chickpea Elongation factor 1-α (EF-1α) was used as internal control for normalization. All the primers used are mentioned in Text S1. The protein extraction, partial purification and IMP assay was performed as described before29. IMP enzyme activity assay was performed by colorimetric estimation of released inorganic phosphate (Pi)52. Polyclonal antibody corresponding to IMP protein was raised in rabbit using bacterially expressed protein. Rabbit polyclonal antibody against IMP (1:1,000 dilution) was used as primary antibody and the western blot was probed with horseradish peroxidase-conjugated donkey anti-rabbit-IgG secondary antibody (1:10,000) (Amersham, Little Chalfont, UK). Bands were detected using ECL reagent (Promega, Madison, USA) following manufacturer’s instructions. Phytic acid (PA) content was estimated according to previously described method53. To determine the association potential of IMP gene-derived SSR marker with seed weight trait, the marker genotyping data across accessions were subjected to the non-parametric Kruskal-Wallis one-way ANOVA using PAST version 3.0 software54. The detailed association analysis was performed using the marker genotyping and 100-seed weight field phenotyping data of 71 chickpea germplasm accessions and 192 representative mapping individuals from each of F2 and F5 intra-specific populations. An association test between SSR marker and 100-seed weight (g) was performed using a general linear model (GLM) and subsequently by a mixed linear model (MLM) that relied on both population structure (Q) as well as kinship (K) statistics55. The degree of association of IMP gene-based SSR marker with seed-weight trait was estimated by the R2 using a model with the SSR and adjusted P-value following a false discovery rate (FDR-controlling method. The SSR marker exhibiting significant association with seed weight trait at the lower FDR adjusted P-value (threshold P < 10−5) and higher R2 (degree of SSR marker-trait association) was considered for further analysis in chickpea.

For transient gene expression analysis, the 1.5 kb upstream sequences of IMP gene upto the translation start codon were amplified from two low and high seed weight chickpea mapping parental accessions ICCV95334 and ICCX-810800 and cloned in pCAMBIA1301 vector to direct expression of beta-glucuronidase (GUS) reporter protein. The GUS gene in this plasmid possesses an intron for detecting GUS expression from plant cells only. Agrobacterium tumefaciens strain EHA105 cultures (OD600 0.6–0.8) carrying this recombinant plasmid and pCAMBIA1302 plasmid having GFP were mixed equally, resuspended in infiltration medium (10 mM Mgcl2, 10 mM MES, 200 µM acetosyringone) and infiltrated in to young leaves of two-month-old chickpea ICCV95334. The leaves were kept in dark for twelve hours after infiltration and expression of the reporter genes was assessed after three days. GUS activity was estimated by monitoring cleavage of the substrate 4-methylumbelliferyl β-D-glucuronide (MUG). Transformation efficiency was normalized by fold change in GFP transcript expression measured by qRT-PCR.