Genome-Wide Association Study (GWAS) For Cold Tolerance at The Bud Burst Stage in Rice Using SNP Markers

Background: Rice is a crop that is very sensitive to low temperature, and its morphological development and production are greatly affected by low temperature. Therefore, understanding the genetic basis of cold tolerance in rice is of great signi�cance for mining favorable genes and cultivating excellent rice varieties. However, there were limited studies focusing on cold tolerance at the bud burst stage, therefore, considerable attention should be paid to the genetic basis of cold tolerance at the bud burst stage (CTBB). Results: In this study, a natural population consisting of 211 rice landraces collected from 15 provinces of China and other countries were �rstly used to evaluate the cold tolerance at the bud burst stage. Population structure analysis showed that this population divided into three groups and was rich in genetic diversity. Our evaluation results confered that the japonica rice was more tolerance to cold at the bud burst stage than indica rice. Genome-wide association study (GWAS) were performed through the phenotypic data of 211 rice landraces and 36,727 SNPs dataset under a mixed linear model, and 12 QTLs (P < 0.0001) were identi�ed according to the seedling survival rate (SSR) treated at 4 ℃ , in which there are �ve QTLs (qSSR2-2, qSSR3-1, qSSR3-2, qSSR3-3 and qSSR9) which were co-located with previous studies, and seven QTLs (qSSR2-1, qSSR3-4, qSSR3-5, qSSR3-6, qSSR3-7, qSSR4 and qSSR7) which were reported for the �rst time. Among these QTLs, qSSR9, harboring the highest-peak SNP, explained biggest phenotypic variation. Through bioinformatics analysis, �ve genes (LOC_Os09g12440, LOC_Os09g12470, LOC_Os09g12520, LOC_Os09g12580 and LOC_Os09g12720) were nominated as candidates for qSSR9. Conclusion: This natural population consisting of 211 rice landraces with high density SNPs will serve as


Background
Originated in the tropical and subtropical, rice is one of the main staple foods in the world.Low temperature has a huge impact on rice.More than 15 million hectares of rice cultivation area in the world have been persecuted by low temperature.Severe cold damage exists in many countries, mainly in Japan, South Korea, the United States and China [1].Rice is cultivated in a wide area in China, ranging from 53 °27 'to 18 ° 90' north latitude, especially in the provinces of the Yangtze river basin in China, from 1951 to 1980, japonica rice and indica rice in the Yangtze river suffered from cold injury severely, the disaster areas lose 5 to 10 billion kg of rice every year [2].Early rice in the Yangtze river basin in China is often affected by cold injury at the bud burst stage, resulting in low germination rate and failure to emerge.
Therefore, it is very necessary to study the cold tolerance in rice at the bud burst stage.
Cold tolerance is a complex quantitative trait that is often controlled by multiple genes and environment, and researchers often use bi-parental populations to look for QTLs associated with cold tolerance.The researchers excavated more than 250 QTLs in the bi-parental population using traditional QTL mapping methods during various stages of rice growth [3].Among more than 250 QTLs, many genes throughout the rice whole growth stages have been isolated.At the germination stage, qLTG3-1 was the rst gene to be linked to germination at low temperatures [4].During the seedling stage of rice, many QTLs/genes related to cold tolerance of rice were isolated, including qCTS12 [5], qCTS4 [6], qCtss11 [7], qSCT1 and qSCT11 [8], qLOP2 and qPSR2-1 [9], COLD1 [10] and qCTS-9 [11].qCTS12 was the rst identi ed coldtolerance gene at seedling stage.COLD1 is another important gene related to cold tolerance in rice seedling stage, and it is also the rst cold-tolerance gene involved in signal transduction.At the booting stage of rice, several genes have been isolated, including Ctb1 [12], qCT8 [13], qCTB7 [14], qCTB3 [15] and qCT-3-2 [16], Ctb1 is the rst gene to be linked to cold tolerance at booting stage in rice.Although biparental mapping populations have played a great role in traditional gene mapping, the construction of bi-parental populations entails a major investment in time, and has therefore limited the number of genes excavate to date [17].
In order to excavate more new genes associated with target traits, the exploration of QTLs/genes through the natural population with weak genetic relationship using GWAS has become one of the most popular methods.This method eliminates the need to construct a mapping population, and can simultaneously analyze multiple alleles, using the recombination information from the long-term evolution of natural populations.In recent years, some studies have applied this method to study rice cold tolerance.17 QTLs were detected related to the rice germination at low temperatures using 63 Japanese varieties [18].51 QTLs were mapped by GWAS with the population of 174 rice accessions from China [19].132 QTLs were identi ed from 527 rice varieties for both rice natural chilling and cold shock stresses [20].67 QTLs were mapped for cold stress at the seedling stage of rice, 56 QTLs were newly discovered [21].42 QTLs were found associated with cold tolerance in rice seedling stage, 20 of which have not been mentioned in previous reports [22].31 QTLs were detected related to low temperature germination of rice seeds at rice germination stage using 200 japonica rice varieties [23].47 QTLs were identi ed for cold tolerance at the bud burst stage using 249 indica rice accessions [24].26 QTLs were found related to cold tolerance in rice seedling stage by using a core collection of landraces of rice from 2,262 accessions of Ting's collection [25].In addition, 31 distinct QTLs regions were identi ed in a panel of 257 rice accessions from all of the world for low temperature germination [26].However, there are still few studies using GWAS to explore cold tolerance in rice at the bud burst stage.In order to understand the genetic mechanism of cold tolerance in rice at early stage, it is necessary to search for QTLs related to cold tolerance in rice at the bud burst stage.
In this study, we selected 211 rice landraces from different regions to form a natural population and performed high-throughput sequencing using microarray.The 211 rice landraces were mainly composed of indica and japonica rice, which provided abundant genetic diversity for studying cold tolerance of rice.We treated the natural population with low temperature at 4 ℃, and then recovered it at room temperature.A total of 12 QTLs and 5 candidate genes for qSSR9 related to cold tolerance were identi ed by genome-wide association study of seedling survival rate (SSR), which provided valuable gene resources for cold tolerance research in rice and laid a solid foundation for breeding cold tolerant rice varieties.

Cold tolerance of the 211 rice varieties
In our study, SSR were used for evaluate the CTBB (Table S2).Due to the abundant landrace germplasm resources, the phenotypes of rice varieties within the natural population varied greatly after 4℃ treatment (Table 1;Fig.1).We classify cold tolerance into ve levels of SSR: Extremely sensitive (0≤X≤20), Sensitive (20<X≤40) , Light sensitive (40<X≤60) , Tolerance (60<X≤80) , Extremely tolerance (80<X≤100).Of the 101 rice varieties that were extremely sensitive to low temperature, 99 were indica and 2 were japonica.Correspondingly, Of the 69 rice varieties that were extremely tolerance to low temperature, 5 were indica and 64 were japonica.According to SPSS software 26.0 analysis results, the SSR was signi cantly correlated with indica and japonica, and the correlation coe cient was 0.851 (Table 2), suggesting that japonica was more cold tolerance than indica at the bud burst stage.On the other hand, we found that the cold tolerance of rice was signi cantly related to its geographical distribution.and the correlation coe cient was 0.714 (Table 2).The higher the latitude, the stronger the cold tolerance of rice, which may be due to the signi cant correlation between indica or japonica types with latitude (Table 2).

Population structure and relative kinship
Based on the 36,727 SNPs, the STRUCTURE, Neighbor-joining (NJ) tree method, Principal Component Analysis (PCA) and Kinship were used to analyze the population structure of the natural population (Fig. 2).According to the STRUCTURE analysis, the log likelihood increased gradually from K = 1 to K = 10.The maximum ad hoc measure 1K was observed for K = 3, which indicated that the entire population could be divided into three subgroups (Fig. 2A).The 211 rice varieties could be divided into 3 subgroups by NJ tree (Fig. 2B) and three principal components from this panel (Fig. 2C).In addition, in the relationship analysis, we found that there were two major groups of subgroups and a middle subgroup in the 211 rice varieties (Fig. 2D), suggesting that the landraces population germplasm resources were abundant, which was bene cial for performing GWAS.

Candidate gene analysis
Among these QTLs, we conducted further candidate genetic analysis of the qSSR9.According to the LD decay analysis, a total 244-kb region was identi ed as the candidate region (Fig. 5).There are 39 genes in this region, including 3 hypothetical proteins, 4 transposon proteins, 7 retrotransposon proteins and 16 functionally annotated genes (Table S3).In order to nd possible candidate genes, we analyzed the homology between these 39 genes and 20 characterized cold-tolerance genes, LOC_Os09g12440, LOC_Os09g12470, LOC_Os09g12520, LOC_Os09g12580 and LOC_Os09g12720 were found having a high degree of homology with COLD1, Ctb1, LTG1, OsWRKY71 and OsbZIP73(Fig.6 Table 4).

Discussion
Population structure and phenotypic assessment of a natural population For this study, GWAS was used as a method to reveal complicated genetic variations of cold tolerance.
However, population structure is an important factor affecting the results of GWAS and increases the false positive rate.In this study, a natural population consisting of 211 rice landraces (130 indica rice and 81 japonica rice) was used to assess cold tolerance in rice at the bud burst stage.Most of the rice varieties come from 15 provinces in China, with three from Japan and one from Philippines.The geographical regions spans from the north latitude 15 • to 48 • including temperate zone, tropics and subtropics.This natural population is newly constructed and has rich genetic diversity.Population structure analysis divided the natural population into three groups.Subsequently, the results of PCA and NJ trees support this result that low relatedness were showed from the relative kinship analysis (Fig. 2), which makes it suitable for GWAS.
We used the SSR as the indicator to evaluate the cold tolerance of natural populations.The results show that the SSR ranges from 0% to 100% (Table S2), indica rice is extremely sensitive to temperature, and its SSR was low after cold treatment.Some indica rice varieties even die in the recovery time, while japonica rice is with the characteristic of cold tolerance, and the SSR of most japonica rice is above 90%.This result showed that japonica rice was more tolerance to cold than indica rice at the bud burst stage.

Identi cation of QTLs/candidate genes controlling CTBB
In this study, we found 12 QTLs using SSR as indicator.Among these QTLs, seven of them (qSSR2-1, qSSR3-4, qSSR3-5, qSSR3-6, qSSR3-7, qSSR4 and qSSR7) were reported for the rst time, and the other ve QTLs (qSSR2-2, qSSR3-1, qSSR3-2, qSSR3-3 and qSSR9) were co-located with previous studies.The physical distance of the peak SNP for qSSR2-2 was located at 4.4Mb on chromosome 2.It overlapped with OsWRKY71 that is a transcriptional suppressor that encodes GA signaling in aleurone cells and coldtolerant [28]).qSSR3-1, qSSR3-2 and qSSR3-3 were co-located with qLVG3 [29], a QTL for low-temperature vigor of germination.In qLVG3 region,, there were two characterized cold-tolerance genes (OsMYB2 and OsCIPK03).OsMYB2 is a MYB transcription factor and plays a regulatory role in tolerance of rice to salt, cold injury and dehydration stress [30], while OsCIPK03 is a calcineurin B-like protein-interacting protein kinases, the overexpression of OsCIPK03 transgenic plants signi cantly improved the tolerance to cold stress [31].The QTL qSSR9 explained the largest phenotype variation in our study overlapping with clr9 that is a QTL associated with culm length growth rate under cold stress [32], however, the knowlwdge of candidate genes underlying qSSR9 is still gaping.
The abundant SNPs dataset of our nature population through chip strategy makes it feasible to locate qSSR9 on a small genomic region.The analysis of candidate genes shows that there are 39 candidate genes undelying qSSR9, among these candidate genes, ve genes (LOC_Os09g12440, LOC_Os09g12470, LOC_Os09g12520, LOC_Os09g12580 and LOC_Os09g12720) might be the target genes, because these ve candidated genes share the same branch of the characterized cold tolerance genes COLD1, Ctb1, LTG1, OsWRKY71 and OsbZIP73, respectively (Fig. 6).For theses characterized cold tolerance genes, COLD1 encodes a G protein signal regulator and it can interact with RGA1, the α subunit of G protein, to sense low temperature, activate Ca 2+ channel, and enhance the activity of G protein GTP-enzyme to enhance cold tolerance of rice [10]; Ctb1 encodes F-box protein, which interacts with an E3 ubiquitin ligase subunit SKP1 and is involved in cold tolerance at booting stage [12]; LTG1 encodes casein kinase and regulates cold response in rice and affects auxin transport, synthesis and signal transduction, and positively regulates low temperature tolerance of rice during vegetative growth period [33]; OsbZIP73 Jap is up-regulated by low temperature and the plant hormone abscisic acid (ABA), suggesting that OsbZIP73 is involved in ABA-dependent low temperature signaling pathways [34].However, that other genes underlying qSSR9 cannot be ruled out, such as LOC_Os09g12360, LOC_Os09g12390, LOC_Os09g12450, LOC_Os09g12615, LOC_Os09g12640 and LOC_Os09g12650.Although these genes do not share the same branches with the characterized cold tolerance genes, they are very homologous with some the characterized cold tolerance genes (Fig. 6).Further studies, such as, gene complementation analysis, are necessary to elucidate which allele is more favourable.

Conclusions
In this study, a natural population consisting of 211 rice landraces were used to assess CTBB by using GWAS under a mixed linear model, 12 QTLs were detected on chromosomes 2, 3, 4, 7, and 9, ve genes (LOC_Os09g12440, LOC_Os09g12470, LOC_Os09g12520, LOC_Os09g12580 and LOC_Os09g12720) might be the target genes for qSSR9 after candidate gene analysis, these QTLs/genes will be conducive to further mining favorable gene resources and breeding of rice varieties.

Materials And Methods
Plant material rice gene SNP microarray 'OsSNPNKs' was used for genotyping.The microarrays were evenly distributed throughout the genome, with an average distance of < 1 Kb between each other.Genotyping based on Affymetrix AXIOM ®2.0.The Target Prep Protocol QRC (P/N 702990) kit manual was used for DNA ampli cation, DNA fragmentation, microarray hybridization, DNA binding single base extension, and signal ampli cation.Staining and scanning were performed using a GeneTitan® multi-channel instrument [27].

Declarations
LD heatmap around peak on chromosome 9

Figure 2 Population
Figure 2

Table 1
SSR (seedling survival rate) of cold tolerance at the bud burst stage (CTBB)

Table 3
Summary of the signi cant SNPs detected by GWAS and the overlapped QTLs/genes reported previously

Table 4
Candidate genes in the qSSR9 region