The variation of rice mesocotyl length
The analysis of CV (Table 1 & Table S1) showed that the rice mesocotyl length variability in the three different treatments with 0 cm, 4 cm and 6 cm sand culture are 114.43%, 146.65% and 152.55%, respectively. While most mesocotyl lengths in the three groups are in 0.00 ~ 0.50 cm; some are above 1.00 cm. There are 10 (the least) and 20 (the most) rice accessions whose mesocotyl lengths are in the range of 1.00 ~ 3.00 cm when cultured with a sowing depth of 4 cm and 6 cm, respectively (Fig. 1). For each rice accession, the difference in mesocotyl lengths at different sowing depths is within 1 cm. The analysis of variance (ANOVA) showed that there is significant difference in the mesocotyl length of 290 rice accessions in the three groups (Table 2) Although there was difference in the mesocotyl length for the same accession when cultured with different sowing depths, the trend of change in mesocotyl length for each accession is highly correlated to the sowing depths.
The experiments results show that there is extensive variation but similar variation pattern for the mesocotyl length of rice seedling in the 290 rice accessions. According to the experimental data, the coefficient of variation (CV) of the mesocotyl lengths in experimental rice population is large and the averages of dispersion is high. This means that the experimental rice population which we used in this study have huge genetic difference and broad representation. The mesocotyl length in the most of the experimental accessions is short. And the variation of mesocotyl length is relatively slight when the treatment condition is 4 cm. It may be because that the depth of sand was shallow, and the seedling growth may be affected by light, thus the mesocotyl elongation was limited(Feng et al. 2017; Gao et al. 2011; Kondo et al. 2010; Yalan 2016). The mesocotyl length of the three groups are positively correlated with each other, indicating that the variation trend is consistent despite the differences in experimental treatment conditions, which led to the different performances of the mesocotyl lengths of the same accession.
Population structure analysis and whole genome resequencing of 290 rice accessions
There are 115 indica rice accessions, 133 japonica rice accessions and 42 Indica japonica intermediate types in the whole 290 experimental rice accessions from all over the world (Table S2). This germplasm resource population can represent two subspecies of rice and have rich genetic background, so this population is suitable for GWAS of mesocotyl length traits in rice seedlings. The average sequencing depth in the whole genome resequencing was 10 × for the 290 rice accessions and developed 1095430 SNPs genotypes. The MSU database(Xiong et al. 2017) http://rice.plantbiology.msu.edu was used to annotate the identified variations. It is the molecular basic for GWAS.
Detection of SNPs associated with mesocotyl length in rice seedling
The EMMAX software was used to analyze the correlation between the rice seedling mesocotyl length data and the SNP data developed by resequencing in 290 rice accessions, and 922 SNPs significantly was found to be correlated with the mesocotyl length were identified (Table 3). 85, 48 and 186 SNPs (-log (P value) ≥ 8.0) were identified when cultured at dark with 0 cm sand, 4 cm sand and 6 cm sand, respectively. 6 QTL regions across chromosomes 1, 4, 8 and 10 under 0 cm sand dark culture condition; 2 QTL regions across chromosomes 7 and 8 under 4 cm sand culture condition and 4 QTL regions across chromosomes 1, 6, 7 and 9 were the distribution of associated SNP makers. Because Bonferroni correction can be conservative, it may result in the probability of producing false negatives. Therefore, we identified the SNP makers associated with rice mesocotyl length again with the -log(P value) ≥ 7.0. And under 4 cm and 6 cm sand culture conditions, we newly identified 172 and 428 SNP makers, respectively. 172 SNP makers were across 2 QTL regions distributed in chromosomes 6 and 9. 428 SNP makers were in across 5 QTL regions distributed in chromosomes 1, 6, 7 and 12. The Manhattan plots of p-values analyzed the SNP makers associated with rice seedling mesocotyl length showed the visible loci in chromosomes (Fig. 2&3).
The comparative analysis of associated SNP makers in three different culture conditions
There were total 8 chromosomes were identified the SNP makers associated with rice seedling mesocotyl length in three different culture conditions. And there were identified 85, 220 and 614 SNP makers (-log (P value) ≥ 7.0) associations under 0 cm, 4 cm and 6 cm sand culture conditions, respectively. And the comparative analysis of the QTL regions in three different culture conditions showed that there was no overlap among 0 cm,4 cm or 6 cm sand culture conditions, but there were three overlap regions between 4 cm and 6 cm sand culture conditions. The overlap regions were distributed in chromosomes 6, 7 and 9. The length of regions was 297.472Kb、253.012Kb、312.368Kb, respectively. The repetitive SNP maker number in three overlap regions was 112, 9 and 5, respectively.
These results showed that there were wide differences in SNP makers associated with rice seedling mesocotyl length among three culture conditions. For each culture conditions there were many specific SNP makers, although there were some overlapped SNP makers between 4 cm and 6 cm sand culture conditions. It indicates that there are multiple genes controlling rice seedling mesocotyl length, and rice seedling mesocotyl length is sensitive to environmental influences.
Identification of candidate genes for seedling mesocotyl length
There are 922 SNP makers associated with rice seedling mesocotyl length across 20 QTL regions and covered 1246 genes. There are little genes known the functions among these genes, and most of these genes are annotated as the unknown expressed protein or putative protein in http://rice.plantbiology.msu.edu (Xiong et al. 2017). And some of these genes are annotated as the regulated genes of growth and development or transcription factors.
For 6 cm sand culture condition, there were 7 QTL regions associated with seedling mesocotyl length. And total 11 SNP makers was located in qML6cm6-1 covered 1687.342 kb(6610901–8298243) region in Chr6. The qML6cm6-1 includes 2 GDSL-like lipase/acylhydrolase genes(LOC_Os06g12410, LOC_Os06g14630)and 1 growth regulator related protein gene(LOC_Os06g13215). Moreover, there were 173 SNP makers located in qML6cm6-2(Chr6:27582320–28071845) covered 489.525 kb region including a MYB gene(LOC_Os12g39640) and 2 Auxin-responsive Aux/IAA gene family member genes OsIAA30(LOC_Os12g40890), OsIAA3༈LOC_Os12g40900༉.
For 0 cm sand dark culture condition, 6 QTL regions were identified associated with seedling mesocotyl length. In qML0cm8, there are 28 SNP makers located in 363.334 kb region (Chr8:25706134–26069468) covered 66 candidate genes including an auxin response factor OsARF21(LOC_Os08g40900). And there were 2 OsWAK receptor-like protein kinase genes (OsWAK34, OsWAK33) found in qML0cm4 and 3(OsWAK95, OsWAK96, OsWAK97) in qML0cm10-1. Furthermore in qML0cm10-1, an OsWAK receptor-like cytoplasmic kinase gene(OsWAK98, LOC_Os10g02360) and an oxidoreductase, aldo/keto reductase family protein gene(LOC_Os10g02380) were identified.
GDSL-like lipase/acylhydrolase genes are mainly expressed in the root of rice seedlings and possibly regulate the cell elongation as well as extracellular matrix. EGR1 with a GDSL-motif domain was as a negative regulator of coleoptile elongation in the context of recent findings on the impact of JA on light signaling(GER1, a GDSL Motif-Encoding Gene from Rice is a Novel Early Light- and Jasmonate-Induced Gene). The main function of MYB family transcription factors was the regulator of secondary metabolism and Cellular morphogenesis, and MYB family transcription factors were expressed in different tissues and organs at different development stages of rice(Hande et al. 2017; Slabaugh et al. 2011). OsSAUR25 and OsSAUR26 are the member of the SUAR gene family which is as part of the primary auxin response in plants and involved in plant growth by regulating auxin synthesis and transport(Xihua et al. 2017; Chen et al. 2014; Kant and Rothstein 2009; Niek et al. 2018; Spartz et al. 2012). OsIAA30 and OsIAA3 are expressed in different tissues and organs at different development stages of rice especially highly expressed in the root of seedling, as reported that these two genes take part in the signal transduction of auxin(Jie et al. 2018; Nakamura et al. 2010). The genes in the wall-associated kinase (WAK) gene family have important function in plant growth and development such as signal transduction, Resistance of pathogens, response to mineral element and cell elongation(Shibo Zhang 2005). In rice, the genes of OsWAK receptor-like protein kinase are expressed in different tissues and organs at different development stages, OsWAK1 is part in the defense of plant disease(Li et al. 2009); and OsWAK11 is regulated by Aluminum (Al), Copper (Cu) and Sodium (Na)(Wei et al. 2014). But the function of six OsWAK genes in our associated QTL regions is not clear. And aldo/keto oxidoreductase gene is expressed the leaf and radicle, part in the regulation of cell elongation(Hur et al. 2009).
Taken together, based on our results and previous publications, we detected 23 candidate genes which maybe have effect to the mesocotyl length of rice seedling at different sowing depths(Table 4)(Kutschera and Wang 2016; Huizhen et al. 2018; Qi et al. 2013; Wang et al. 2007).