Groundnut or peanut is a major oilseed and grain legume mainly cultivated under rainfed regions of tropical, subtropical and temperate countries worldwide. Unlike the Virginia genotypes, the widely grown Spanish cultivars have lost the dormancy trait during domestication and selective breeding, and resulted in introduction of pre-harvest sprouting trait in cultivated groundnut. Developing commercial Spanish cultivars with 14–21 days of dormancy can prevent yield losses due to pre-harvest sprouting. Utilizing GAB offers a distinct advantage over conventional breeding by facilitating efficient tracking of alleles among segregating lines through the use of trait-linked markers (Varshney et al., 2014). In this context, the present investigation employed multi-model genome wide association analysis on mini-core collection using genotyping data generated from “Axiom_Arachis” 58K SNP array and multi-environment phenotyping data to identify the genomic regions and candidate genes regulating dormancy/germination.
Conventional single-locus methods like Generalized Linear Model (GLM) and Mixed Linear Model (MLM) have been frequently deployed for identifying genetic variants in several crops (He et al., 2019). However, these methods have limitations as they neglect combined effects of multiple loci and face issues with multiple test corrections to determine critical values (Odesola et al., 2023). ML-GWAS methods, however, addresses these challenges (Liu et al., 2016). Comparative studies indicated that ML-GWAS has higher statistical power and lower false-positive errors as compare to SL-GWAS methods (Segura et al., 2012; Hu et al., 2018). Investigators typically integrate the strengths of various ML-GWAS algorithms to identify target loci/QTL associated with complex traits, as each algorithm possesses unique characteristics and QTL detection capabilities (Liu et al., 2020).
A total of 9 significant STAs using ML-GWAS were identified for FSD trait using mini-core collection association panel. Previously, a QTL-Seq study reported two genomic regions on B05 and A09 chromosomes for fresh seed dormancy and also developed a potential marker on chromosome B05, GMFSD1 (Kumar et al., 2020). In this study we have developed three more allele specific markers namely, GMFSD2 (A09), GMFSD3 (B05) and GMFSD4 (B09). High-density genetic mapping for FSD identified two dormancy QTLs on chromosomes A04 and A05 (Wang et al., 2022). Similarly, a major stable QTL associated with fresh seed germination was identified on chromosome A04 (Zhang et al., 2022). Moreover, our previous FSD study used a 5K SNP assay based bi-parental genetic mapping and identified five major QTLs on Ah01, Ah06 Ah11, Ah16 and Ah17 chromosomes and two minor QTLs on Ah04 and Ah15 chromosomes (Bomireddy et al., 2022). Additionally, qFSD_A04-1 (119 Mb) on chromosome A04 was observed to be located in the close proximity of qPD_A04-2 (Wang et al., 2022), while the location of qFSD_B05-1 (112 Mb) on chromosome B05 was physically close to the genomic region identified by Kumar et al., 2020. Thus, the identification of significant MTAs around previously reported genomic regions using multi-locus GWAS underscores the method's reliability. In addition to ML-GWAS, our study has also identified 38 significant STAs on 14 chromosomes of cultivated groundnut by SL-GWAS revealing all the possible genomic regions associated with FSD.
Differences in the mapping results from various studies can be attributed to the factors such as seed development stage, population composition or the pedigree of the parents used in the population development and the prevailing environment during crop growth period (Cheng et al., 2014; Magwa et al., 2016). Comparable results were also documented in association mapping studies on wheat and rice seed dormancy (Lin et al., 2016; Lu et al., 2018). Because of genome similarity between the homeologous chromosomes of diploid progenitor genomes (A and B genome), homeologous associations on both sub-genomes of groundnut were identified. Nested association mapping for seed and pod weight in groundnut identified associations on homeologous chromosomes A05/B05, A06/B06 (Gangurde et al., 2020). In other allopolyploids such as wheat, where seed dormancy QTLs were detected on homeologous 3A/3B (Shao et al., 2018), 4A/4B and 5A/5B (Lin et al., 2015) chromosomes.
As discussed earlier, identification of candidate genes underlying the QTLs/MTAs provide insights for better understanding of the trait. As ABA, GA and ethylene have been demonstrated to be associated with seed dormancy and germination regulation in many crops, identifying genes involved in the regulation of their metabolic pathways is of major interest. The involvement of ABA signaling, along its interaction with GAs/ethylene tends to modulate seed dormancy and germination initiation. Therefore, the genes retrieved in this study from both ML-GWAS and SL-GWAS models were thoroughly reviewed in previous literature for assessing their functional role in ABA and GA signaling pathways. Fascinatingly, cytochrome P450 705A (Aradu.FN562) identified as an important participant in ABA catabolism. A cytochrome P450 superfamily protein (CYP707A) in Arabidopsis encodes ABA 8'-hydroxylases, an enzyme involved in ABA 8'-hydroxylation pre-dominant for ABA catabolism. Expression profiling indicated that cyp707a2 mutant displayed six times higher ABA levels, resulting in hyper seed dormancy compared to wild types (Kushiro et al., 2004). It indicated that CYP707A2 negatively regulates seed dormancy by declining ABA levels during seed imbibition. Supporting this, cytochrome P450 superfamily protein gene copies displayed high transcript abundance in ICGV 91114 (non-dormant) gene expression atlas, indicating their positive role in regulating germination (Bomireddy et al., 2022).
WRKY family transcription factor family protein (Aradu.76148) identified in this study is known to be involved in ABA signaling. Lack of WRKY transcription factor 41 (WRKY41) in imbibed seeds of Arabidopsis resulted in decreased ABI3 (play crucial role in seed dormancy) expression while overexpressing transgenic WRKY41 lines had increased ABI3 expression (Ding et al., 2014). Examination of the double mutant wrky41 aba2 revealed that the regulation of ABI3 expression and seed dormancy is a combined effort between WRKY41 and ABA. Therefore, WRKY41 acts as a key regulator of ABI3 expression, thereby influencing seed dormancy. The identified protein kinase superfamily protein (Aradu.VA4EQ) in this study is known for its positive role in ABA signaling. In Arabidopsis, SNF1-RELATED PROTEIN KINASE 2.2 (SnRK2.2) and SnRK2.3, were reported to regulate ABA responses in seed dormancy and germination by mediating ABA signaling (Fujii et al., 2017). Double mutants of snrk2.2 and snrk2.3 exhibited decreased expression of several ABA-induced genes, demonstrating their positive role in ABA signaling. Similarly, in Arabidopsis, redundant ABA-activated SnRK2s were identified as prominent regulators of seed maturation and dormancy (Nakashima et al., 2009). Dormancy/auxin associated family protein (Aradu.2Q7VA) or Auxin responsive factors (ARF) identified in this study were reported to stimulate ABA signaling to induce seed dormancy (Liu et al., 2013). Defects in auxin signaling of MIR160-overexpressing plants and auxin receptor mutants significantly reduce dormancy, while increase in auxin biosynthesis prolong dormancy.
Serine/threonine-protein phosphatase 7 long form homolog (Aradu.25KN6) and myb transcription factor (Aradu.GFS4B) identified in this study are known for their role in ABA signaling. In Arabidopsis, loss of ABSCISIC ACID-INSENSITIVE1 (ABI1) which encodes 2C class of serine/threonine phosphatases (PP2C), leads to enhanced ABA responsiveness, indicating its negative role in ABA signaling (Gosti et al., 1999). Further, overexpressing HON (Protein Phosphatase 2C family group) lines revealed that it suppresses dormancy by impeding ABA signaling (Kim et al., 2013). However, some PP2Cs like HIGHLY ABA-INDUCED PP2C GENE1 (HAI1) were reported to promote ABA signaling (Saez et al., 2004). MYB transcription factor 96 (MYB96) was presumed to fine tune seed dormancy as it enhances ABA biosynthetic NCED genes and down regulate GA biosynthetic GA20ox1, GA3ox1 genes in Arabidopsis (Lee et al., 2015). myb96-1 mutant seeds exhibited germination earlier than the wild MYB96-1, while activation-tagging of myb96-1D seeds delayed the germination process. Differential gene expression analysis identified MYB60 as the promising candidate regulating dormancy with higher transcript abundance observed in the dormant Tifrunner gene expression atlas (Bomireddy et al., 2022).
The identified Transcriptional regulator of STERILE APETALA-like (Araip.14WNT) is notably known to be involved in GA catabolism. APETALA 2 (AP2)-domain-containing transcription factors (ATFs), including OsAP2-39 in rice and ABI4 in Arabidopsis, were reported to play a prominent role in ABA and GA antagonistic crosstalk (Yaish et al., 2010; Shu et al., 2016). The AP2/ethylene-responsive element binding factor (AP2/ERF) family constitutes a substantial group of plant transcription factors, playing diverse roles at various plant developmental stages. In rice, transcriptome analysis of OsAP2-39 overexpression lines unveiled upregulation of OsNCED-I (ABA biosynthesis gene), increasing endogenous ABA levels, and enhanced GA-inactivating gene OsEUI, resulting in reduced GA content (Yaish et al., 2010). OsAP2-39 directly governs the expression of OsNCED-I and EUI, elucidating a novel mechanism regulating ABA/GA balance and ultimately influencing rice growth.
Ethylene-responsive transcription factors (Araip.LL89K) and F-box protein interaction domain protein (Aradu.XB89Z) identified in this study are the significant contributors for GA signaling. Ethylene (ET) was reported to break seed dormancy by antagonizing ABA biosynthesis and signaling, thereby promoting seed germination (Corbineau et al., 2014). In Arabidopsis, Delay of Germination1 (DOG1) interacts with ethylene responsive factor 12 (ERF12) and co-regulate seed dormancy (Li et al., 2019). High transcript levels of ethylene-responsive transcription factors in seed and embryo of ICGV 91114 (Arachis hypogaea sub spp. fastigiata) gene expression atlas revealed its functional relevance in seed germination (Sinha et al., 2020; Bomireddy et al., 2022). In plants, F-box proteins regulate various physiological processes in various ways. Song et al. (2012) had isolated OsFbx352 gene (encoding for F-box domain protein) from rice to characterize its role in germination. Overexpression of OsFbx352, (a F-box protein in rice) demonstrated lower ABA contents by reduced expression of ABA synthesis genes (OsNced2, OsNced3) and increased ABA catabolism gene expression (OsAba-ox2, OsAba-ox3) leading to seed germination. Whereas, knockdown of OsFbx352 led to higher ABA contents and suppressed seed germination, revealing its regulatory in modulating ABA metabolism. In groundnut, high gene expression value of F-box/RNI-like superfamily protein (Arahy.LZ56CD) in all the six selected tissues of seed and pod of non-dormant ICGV 91114 also indicated its positive regulatory role in germination (Bomireddy et al., 2022). In Arabidopsis, transgenic plants overexpressing for AtTLP9 (Tubby-F-box like protein) also demonstrated hypersensitiveness to ABA (Lai et al., 2004). As all these genes were explored for their involvement in ABA/GA/Ethylene signaling pathways, these were identified as promising candidates in the hormonal regulation of dormancy and germination dynamics.
The allele calls of the representative panel of mini-core accessions for the identified significant MTAs revealed the presence of all favorable dormant alleles in an accession result in increased dormancy duration. Conversely, an increase in the number of unfavorable non-dormant alleles corresponded to a decreased dormancy duration. These MTAs show potential for developing KASP assays, and accessions with all the favorable dormant alleles can be used as donors in MABC programs aiming for dormancy of ≥ 30 days. However, if the goal is to achieve fresh seed dormancy of 14–21 days, accessions with a combination of dormant and non-dormant alleles for these markers can be utilized. Thus, these assays could play a crucial role in future molecular breeding programs, offering targeted and efficient means to select lines with 2–3 weeks of dormancy.