This article is protected by copyright. All rights reserved.
New insights into the genetic basis of natural chilling and cold shock
tolerance in rice by genome-wide association analysis
Yan Lv, Zilong Guo, Xiaokai Li, Haiyan Ye, Xianghua Li, Lizhong Xiong*
National Key Laboratory of Crop Genetic Improvement and National Center of Plant Gene
Research (Wuhan), Huazhong Agricultural University, Wuhan 430070, China
*Corresponding author: [email protected]
Brief summary statement
By association mapping study of 529 rice accessions, we revealed a distinct genetic basis for
natural chilling and cold shock stress tolerance at the seedling stage, and we also found that
the cold adaptability of rice is associated with the subpopulation and latitudinal distribution.
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/pce.12635
This article is protected by copyright. All rights reserved.
ABSTRACT
In order to understand cold adaptability and explore additional genetic resources for the cold
tolerance improvement of rice, we investigated the genetic variation of 529 rice accessions
under natural chilling and cold shock stress conditions at the seedling stage using
genome-wide association studies, a total of 132 loci were identified. Among them, 12 loci
were common for both chilling and cold shock tolerance, suggesting that rice has a distinct
and overlapping genetic response and adaptation to the two stresses. Haplotype analysis of a
known gene OsMYB2, which is involved in cold tolerance, revealed indica-japonica
differentiation and latitude tendency for the haplotypes of this gene. By checking the
subpopulation and geographical distribution of accessions with tolerance or sensitivity under
these two stress conditions, we found that the chilling tolerance group, which mainly
consisted of japonica accessions, has a wider latitudinal distribution than the chilling
sensitivity group. We conclude that the genetic basis of natural chilling stress tolerance in rice
is distinct from that of cold shock stress frequently used for low temperature treatment in the
laboratory, and the cold adaptability of rice is associated with the subpopulation and
latitudinal distribution.
Key-words: rice; chilling tolerance; cold shock tolerance; GWAS; subpopulation distribution;
geographical distribution.
This article is protected by copyright. All rights reserved.
INTRODUCTION
Low temperature is one of the major abiotic stresses limiting plant growth, productivity,
and quality. It has been generally accepted that the breeding of cereal crops should be focused
on the improvement of stress tolerance and photosynthesis by increasing the use efficiencies
of water and nutrients, adjusting to local temperature and precipitation (Sang, 2011). With
global environmental worsening and abnormal climate changes, it is urgent to reveal the
genetic and molecular mechanisms of plant responses to low temperature stress and to search
for useful genetic resources for improving low-temperature tolerance. As one of the major
crops, rice is widely grown in tropical, subtropical, and temporal regions, and temperature is
one of the major environmental factors limiting its geographic distribution. The optimal
temperature for rice growth is 25-30°C (Kim et al., 2014). Previous studies on low
temperature stress in rice mainly concentrated on chilling stress (temperature around 10°C)
which was frequently used to distinguish it from freezing stress (temperature around 0°C),
however there is no clear definition of chilling or cold/freezing stress, and the treatment
temperatures were often different for the same term in many reports (Cheng et al., 2007, Guo
et al., 2006, Ma et al., 2015, Wang et al., 2013, Wang et al., 2014, Yang et al., 2012). To
date, no report has compared different low-temperature stresses such as natural chilling stress
with the acute freezing stress in rice. Recent studies have revealed some mechanisms and
signaling networks involved in the cold stress response in rice (Knight & Knight, 2012, Ma et
al., 2015, Wang et al., 2014, Yang et al., 2012). Zhao et al focused on moderate cold stress
(8°C) under natural environmental conditions using a cold tolerant variety for genome-wide
expression profiling, and they proposed a series of cold response mechanisms: the induction
This article is protected by copyright. All rights reserved.
of OsDREB2A, glutathione peroxidase (GPX), and glutathione S-transferase (GST) serving
as the main reactive oxygen species (ROS) scavenger, and the ABA signaling pathway plays
a dominant role (Zhao et al., 2014b). In addition, this study suggested that the cold stress
response of rice varies with the specific temperature imposed and the rice genotypes utilized
(Zhao et al., 2014b).
Rice has two major subspecies, indica and japonica. The eco-geographical
differentiation of the two subspecies has been reported (Londo et al., 2006, Yu et al., 2003). It
has been generally accepted that japonica rice had higher potential in cold adaptability than
indica rice (Cheng et al., 2007, de Los Reyes et al., 2013, Lu et al., 2014, Ma et al., 2015,
Morsy et al., 2005, Pan et al., 2015), but the genetic evidence for this is very limited.
Dissection of the genetic and molecular basis of cold response and adaptation is the
foundation for the improvement of low-temperature tolerance. Among the previous studies,
QTL mapping based on the molecular marker and linkage maps has always been a common
and classical approach for the genetic study (Fujino et al., 2004, Koseki et al., 2010, Liu et al.,
2013, Yang et al., 2013b). Liu et al combined whole-genome expression profiling analysis of
two parents of a genetic population and QTL mapping of rice under cold stress conditions at
the early seedling stage, and they identified a candidate gene LOC_Os07g22494, which was
further confirmed by genetic transformation approach (Liu et al., 2013). Besides the studies
at the seedling stage, QTL associated with cold tolerance at the germination stage (qLTG3-1,
qLTG11, qLTG9), booting stage (qCTB7, qLTB3), fertilization stage (qCTF7, qCTF8,
qCTF12), and reproductive stage (qPSST-3, qPSST-7, qPSST-9) have also been reported
(Fujino & Sekiguchi, 2011, Fujino et al., 2008, Iwata & Fujino, 2010, Li et al., 2013, Shinada
This article is protected by copyright. All rights reserved.
et al., 2013, Shirasawa et al., 2012, Suh et al., 2010, Zhou et al., 2010).
Recently, genome-wide association study (GWAS), representing one of the newly
developed genetic approaches, has been adopted to investigate the genetic architectures of
various important agronomic traits in rice. Huang firstly performed GWAS for 14 agronomic
traits and identified a number of loci potentially important for rice grain yield and
improvement, which strongly suggested GWAS based on second-generation sequencing
could be a powerful supplement for the traditional linkage mapping (Huang et al., 2010).
Then the same group conducted GWAS on flowering time and grain yield traits by increasing
the population to 950 accessions, and they pointed out that the larger sample could increase
the power to detect variants associated with the traits of interest (Huang et al., 2012). Besides
these, association mapping was conducted for stigma and spikelet characteristics (Yan et al.,
2009), aluminum tolerance (Famoso et al., 2011), grain concentrations of arsenic, copper,
molybdenum, and zinc (Norton et al., 2014), root traits (Courtois et al., 2013), sheath blight
resistance (Jia et al., 2012), grain color, phenolic content, flavonoid content and antioxidant
capacity (Shao et al., 2011), grain quality traits (de Oliveira Borba et al., 2010), grain
metabolites (Lou et al., 2011), harvest index (Li et al., 2012), and silica concentration
in rice hulls (Bryant et al., 2011). Meanwhile, the association analysis approach has also been
used to investigate the genetic variation of candidate genes for important traits in rice (Lu et
al., 2012, Tian et al., 2009, Wu et al., 2013, Yan et al., 2013, Zhao et al., 2011). Despite the
wide application of GWAS in the genetic dissection of agronomic traits in crops, very few
studies have used this approach to investigate low temperature tolerance (Huang et al., 2013,
Strigens et al., 2013). Recently, Pan et al. conducted association mapping for rice cold
This article is protected by copyright. All rights reserved.
tolerance at the germination and booting stages with 174 Chinese rice accessions and 273
SSR markers, and 51 QTLs for cold tolerance were detected (Pan et al., 2015), but the
genetic basis of natural chilling stress was not addressed.
In this study, a panel containing 529 accessions was used to conduct association analysis
of rice tolerance to natural chilling and cold shock stresses with an aim to reveal the genetic
difference of rice in response to the two stresses. Further haplotype analysis of a candidate
gene and the geographical distribution of the chosen accessions revealed indica-japonica
differentiation and latitudinal tendencies of the cold adaptability, which provided valuable
reference for elucidating the genetic basis and differentiation of low temperature tolerance in
rice.
MATERIALS AND METHODS
Materials
A total of 529 rice accessions including 202 from the China Core Collection and 327
from the World Core Collection were used for the association analysis (Supporting
Information Table 1). This panel of rice accessions is essentially the same as the panel of 533
accessions as previously described (Yang et al., 2014b) except three accessions (C126, W196,
and W232) with severe heterozygosity and one (W190) with a low mapping rate (10%)
omitted.
Low-temperature treatment
After germination for 7 days, the seedlings with uniform growth were transplanted to 10
cm x 10 cm pots each containing 9 seedlings. Each accession was planted in 3 pots as three
This article is protected by copyright. All rights reserved.
biological repeats, and the layout of the pots followed a completely randomized block design.
In this study, we designed two treatments to mimic the natural chilling (temperature
gradually declined to a range of 5-12°C) and cold shock (constant temperature at 4°C)
stresses respectively. For the natural chilling stress treatment, the seedlings were grown in a
greenhouse for 3 weeks, at which time the plants went into the 4-leaf-stage, the natural
chilling treatment was carried out in the greenhouse in winter (Wuhan, China) with the
heating and light turned off and the natural low-temperature fluctuating between 5-12°C
depending on the outside temperature. The temperature in the greenhouse was recorded every
half hour by a weather station (Spectrum Technologies, Inc. WatchDog 2000), and a portion
of the record is provided in Supporting Information Fig. S1. For the cold shock treatment, the
4-leaf-stage rice plants was performed in a growth chamber set at a constant 4°C with 14
hours/10 hours of light/darkness.
Determination of electrolyte leakage
The electrolyte leakage (EL) measurement was performed as previously described (Guo
et al., 2006) with minor modifications. Two fully expanded leaves from two plants were cut
into segments of similar sizes, and immersed in 8 ml of double distilled water in a 10 ml test
tube for 24 h at 25°C with continual shaking at a speed of 100 rpm. The initial conductivity
(R1) was measured with a conductivity meter (Model DDS-IIA, Shanghai Leici Instrument
Inc., Shanghai, China). Then, the test tubes were placed in boiling water for 20 min and
cooled naturally to room temperature, and the conductivity (R2) was determined again. The
relative EL was calculated as the ratio of R1 to R2. A total of 7 traits including EL under
normal condition (ELN), EL after natural chilling stress treatment for 3 days (ELC1) and 7
This article is protected by copyright. All rights reserved.
days (ELC2), EL after recovery for 7 days (ELR), EL before cold shock treatment (4°C)
(ELSN), and EL after cold shock treatment for 1 day (ELSC1) and 3 days (ELSC2) were
investigated in this study. Five ratio traits (ELR1, ELR2, ELR3, ELSR1, and ELSR2) were
calculated as the ratios of the EL values under stress conditions to the EL under normal
conditions (see Table 1 for definitions of these parameters or traits).
Other phenotypic data collection
According to the extent of leaf rolling, the survival rate, and the chlorosis conditions, we
divided the accessions into 5 resistance levels (score 1-5, respectively), which is another
feasible method to investigate the cold response phenotype. Score 1 (the most resistant)
indicated that the seedlings had normal leaf color with no damage, while score 5 (the least
resistant) indicated that all of the seedlings were wilting or dead. The resistance level (score)
under natural chilling stress (RLC) and after recovery (RLR) were collected, and the average
resistance level of three repeats was used for further analysis. The survival rate (SRC) as a
commonly used criteria for the chilling tolerance of rice was collected after the natural
chilling stress treatment, and the ratio of fresh vs dry biomass after natural chilling stress
(BMR) was used as another criteria to evaluate the natural chilling tolerance.
Microarray Analysis
GO analysis was performed in a MAS 3.0 molecule annotation system
(http://bioinfo.capitalbio.com/mas3/) and the microarray dataset was collected from our
group based on Affymetrix GeneChip Rice Genome Array (Supporting Information Table 2).
The chip data has been submitted to NCBI GEO database (GSE71680). The chip contained
57381 probes, among these, 1347 probes representing 1260 transcripts from indica, and
This article is protected by copyright. All rights reserved.
54168 probes representing 48564 transcripts from japonica. It should be noted that 9311,
MH63, ZS97, TRAT109 in the chip dada corresponds to C145, C148, C147, C153 in our
panel and C087, C063, C070, C079 corresponds to 4 species Lixingeng, Geng87-304,
Youmangzaogeng and Muguanuo-1 in our panel. Chip data listed the signal ratio of stress
condition (cold shock stress at 4°C for 6 hours and 24 hours) to normal condition. Gene
cluster analysis was conducted using Gene Cluster 3.0 and Java Tree View.
Genome wide association study
A total of 529 accessions were collected to construct this association panel. For GWAS
of the 16 traits, we adopted a mixed-model approach using the factoral spectrally transformed
linear mixed model (FaST-LMM) program, with 4,358,600 SNPs across the entire rice
genome (minor allele frequency ≥ 0.05; the number of accessions with minor alleles ≥ 6). The
suggestive and significant P value thresholds of the entire population were respectively
1.21E-06 and 6.03E-08. The linkage disequilibrium (LD) statistic r2 was calculated by Plink
based on haplotype frequencies. More detailed information about the GWAS analysis was
referenced from the recent study (Yang et al., 2014b).
Subpopulation, geographical distribution and classification of rice accessions
The 529 accessions in our panel were divided into 4 subpopulations, including indica
subset (accounting for 56.77% of the collection), japonica subset (29.48%), aus type (8.70%),
and other subset (6.05%). The information on the subpopulation, and the degrees longitude
and latitude was referred to RiceVarMap (http://ricevarmap.ncpgr.cn/) (Zhao et al., 2014a).
Since our panel included 202 China Core Collection (CCC) accessions and 327 World Core
Collection (WCC) accessions, the selected accessions from different provinces of China and
This article is protected by copyright. All rights reserved.
these were marked on the China Map, and the accessions with a geographical location (116.4,
39.9; China) were marked on the World Map.
The selection criteria for sensitivity or tolerance classification of the accession to the
two stresses were based on the indicative traits, except for the ELN and ELSN in chilling or
cold shock stress. The rice accessions selected for subpopulation or latitudinal distribution
analysis should be from the top or bottom 150 accessions according to the order of the values
for each trait used for evaluating the chilling or cold shock tolerance, and meanwhile the
accessions selected for each type should match the tolerant or sensitive criteria by 80% of the
traits under the same stress condition.
Statistical analysis
Differences in phenotypic and latitude values of accessions in the haplotypes or subgroups
were examined by one-way ANOVA and Duncan multiple comparison if ANOVA result is
significant (P<0.05) (Lu et al., 2013). For the phenotypic values of the four haplotypes of
gene locus 07g44410, we used the Kruskal-Wallis test, which is a non-parametric test for one
factor ANOVA, and multiple comparison was examined if the test result was significant
different (P<0.01). Differences in latitude values between specified groups such as
cold-tolerant and cold-sensitive groups were examined by Student's t-test. Statistical analysis
was run by IBM SPSS statistics 19.0.
This article is protected by copyright. All rights reserved.
RESULTS
Evaluation of the cold response of rice at the seedling stage
In previous studies of cold stress responses in rice, the artificial cold treatment is a cold
shock stress, and very few experiments mimic the natural chilling stress conditions. In this
study, we designed two treatments to mimic the chilling (temperature gradually declined to a
range of 5-12°C) and cold shock (constant temperature at 4°C) stresses respectively. The
temperature record of the natural chilling stress treatment is shown in Supporting Information
Fig. S1. Beside survival rate, a common criteria reflecting the final performance of plants
after stress recovery, electrolyte leakage (EL) in leaves was adopted as a major physiological
parameter in this study since EL can partially reflect the damage of green leaves during the
stress process but its genetic basis was seldom addressed. EL was measured to evaluate the
cold response of rice seedlings under natural chilling stress (ELN, ELC1, ELC2, ELR) and
cold shock stress (ELSN, ELSC1, and ELSC2) (Table 1 for definitions of these parameters or
traits). Meanwhile, other traits such as the fresh vs dry biomass ratio after natural chilling
stress treatment (BMR), resistance level (score) under natural chilling stress treatment (RLC),
resistance level (score) after recovery (RLR), and the survival rate after natural chilling stress
treatment (SRC) were applied in the evaluation of natural chilling stress tolerance since these
indices are more meaningful in rice breeding in natural chilling tolerance.
The results of seven electrolyte leakage indices (ELN, ELC1, ELC2, ELR, ELSN,
ELSC1, and ELSC2) indicated that with the prolonged time duration of stress treatment, large
variation was observed for EL and the relevant ratio traits including ELR1, ELR2, ELR3,
ELSR1, and ELSR2 (Table 1). The range of ELN and ELSN was not exactly the same, which
This article is protected by copyright. All rights reserved.
may be due to the slight differences of seedling growth state in the two experiments. Among
these traits, the variation range of ELC1 is similar to ELN, but the variations of ELC2 and
ELR were obviously increased, suggesting slight cell membrane damage at the early stage of
natural chilling stress treatment, but that the damage became serious and irreversible at the
later stage of the stress treatment. The situation was different in the cold shock stress
treatment, in which ELSN, ELSC1, and ELSC2 were increased since the onset of the stress
treatment (Table 1), indicating that the cell membrane damage was faster and more
significant.
The variation distributions of the cold response indices or traits are shown in Supporting
Information Fig. S2, which can be roughly classified into three categories: three traits (ELN,
ELC1, and ELSN) showed typical normal distribution; eight traits (BMR, RLR, ELR1, ELR2,
ELR3, ELSC1, ELSC2, ELSR1, and ELSR2) showed skewed distribution; while the other
four traits (RLC, SRC, ELC2, and ELR) showed bimodal distribution.
Correlations among the cold response traits
The correlation analysis among the cold response traits was performed, and the results
are presented in Table 2. Eleven traits under the natural chilling stress treatment and five
traits under the cold shock stress treatment had significant correlations with r which ranged
from 0.3 to 0.8 between each other. Significant correlation was observed between the green
leaf area or biomass-related traits (such as BMR, RLR, and RLC) and the EL traits (Table 2),
suggesting that the EL indices can largely reflect the damage of green leaves during cold
stress process. In addition, the ratio trait ELR2 was strongly correlated with the
This article is protected by copyright. All rights reserved.
corresponding trait under natural chilling stress conditions (ELC2; r >0.8), as well as ELR
and ELR3, ELR1 and ELC1, ELR2 and ELC2. However, the 11 traits and 5 traits under
different stress conditions had no significant correlations, indicating that the rice seedlings
exhibited different responses under these two types of low-temperature stress conditions at
the physiological or biochemical levels.
Loci associated with low-temperature tolerance identified by GWAS
We selected 4,358,600 SNPs across the entire rice genome for GWAS for the cold
response traits described above. Any two leading SNPs within a 200 kb range were
considered as one association locus. The association analysis for the whole panel ultimately
identified 132 loci associated with natural chilling stress and cold shock stress with the
suggestive threshold value at 1.21E-06 (Supporting Information Table 3). We also performed
GWAS on indica and japonica subpopulations and a large number of peaks were also
detected (Supporting Information Table 4, 5). More detailed information for SNPs, physical
positions, P values are listed in Supporting Information Table 3 (the whole panel), Supporting
Information Table 4 (the indica panel), Supporting Information Table 5 (the japonica panel).
A quantile-quantile plot of all 16 traits is provided in Supporting Information Fig. S3, S4, S5.
Manhattan plots for the association analysis of these traits are displayed in Fig. 1 and
Supporting Information Fig. S6, S7, S8. Since FaST-LMM program used here could reduce
the effect of population structure (Yang et al., 2014a), and the quantile-quantile plot of all 16
traits for the whole panel showed satisfied effect in control of population structure
(Supporting Information Fig. S3), we focused on the GWAS results from the whole panel in
This article is protected by copyright. All rights reserved.
the following analyses.
For a better view of the comprehensive association results, all of the detected association
loci in the whole panel were marked on the 12 chromosomes according to their physical
distance on the rice genome (Fig. 2), and the loci for natural chilling stress (57 loci) and cold
shock stress (63 loci) are shown in blue and green respectively. The loci detected for the traits
under the two stress conditions (12 loci) are marked in red. From the graphic view, we
noticed that 132 of the loci are distributed widely in the rice genome, with chromosomes 2
and 10 exhibiting relatively fewer detected loci.
Among the 132 association loci, 39 were detected for two or three different traits, and 24
loci had a significant P value (threshold value at 6.03E-08) for at least one trait. Examples of
loci (L18, L27, L63, L79, and L104) which were detected for two traits are shown in Fig. 1, L
is short for locus. In addition, 18 of the 39 loci were associated with EL traits under the stress
condition and the relevant ratio traits as well, which were consistent with the correlation
results in Table 2. It was noted that six loci (L18, L39, L51, L79, L96, and L121) in Fig. 2
were detected for three traits which were correlated with each other, indicating that these loci
probably have important roles in cold tolerance.
Comparison of cold tolerance loci from GWAS and genetic mapping
There were many overlaps between the loci detected by GWAS in this study and the
reported QTL related to low-temperature stress tolerance. A total of 68 loci from this study
were located in or overlapped with the reported QTL (shown in grey in Fig. 2 with the
corresponding references in Supporting Information Table 3) in which half of them were
This article is protected by copyright. All rights reserved.
detected for the cold tolerance at the seedling stage, while the other QTL were detected at the
germination stage or the reproductive stage. Although no significant loci in this study were
found with overlapping to the reported QTL qCOLD1 with causal gene COLD1 characterized
recently (Ma et al., 2015), overlapping loci were found for the other two QTLs, qCOLD4
(overlapped with L59 and L60) and qCOLD2 (overlapped with L87 and L88), reported in
their study. Such a significant portion of overlapping further supported our GWAS results.
The co-localization results of the 68 loci suggested our GWAS on cold stress in rice was
feasible and efficient, while the other 64 loci without overlap to reported QTL may be
potential novel loci for chilling and/or cold shock tolerance in rice.
Cold-responsive genes within the significant association loci
By checking our whole genome expression profiling data, all of the cold
stress-responsive genes (potential candidates) within 200 kb (100 kb upstream and 100 kb
downstream of the leading SNPs) for the 132 loci were selected and listed in Supporting
Information Table 6. Five reported genes, OsMYB2 (Os03g20090), Ctb1 (Os04g52830),
OsRAN2 (Os05g49890), OsiSAP8 (Os06g41010), and OsLti6a (Os07g44180), have been
confirmed to participate in cold response (Chen et al., 2011, Kanneganti & Gupta, 2008, Ma
et al., 2015, Morsy et al., 2005, Saito et al., 2004, Yang et al., 2012), and these genes are
indicated with black arrows in the Manhattan plots (Fig. 1). Besides these five genes,
additional stress-related genes such as SRWD5, OsPR4a (Huang et al., 2008, Wang et al.,
2011), SLAC1, OsACO7, OsACS2, and CRR6 that have been reported to be involved in
stomatal conductance, ethylene biosynthesis and metabolism (Iwai et al., 2006, Kusumi et al.,
This article is protected by copyright. All rights reserved.
2012, Yamori et al., 2011), were also induced by cold stress according to the expression
profiling data (Supporting Information Fig. S9), suggesting that they may also be involved in
cold response.
We performed cluster analysis for the cold responsive genes located in the association
loci using the gene chip expression profiles from the seedlings of 8 rice accessions (3
cold-sensitive: C148, C147, C145; and 4 cold-tolerant: C087, C063, C070, C079; and one
intermediate type C153) treated with cold shock stress (4°C) for 6 hours and 24 hours
(Supporting Information Table 2). The analysis revealed two main categories of expression
patterns: 53 genes were down-regulated in at least 4 accessions by cold shock stress, and 49
genes were up-regulated in different levels (Supporting Information Fig. S9). The 5 reported
genes involved in cold stress also exhibited different expression patterns. Ctb1 and OsLti6a
were both weakly (less than two-fold) down-regulated in these 8 accessions after cold shock
treatment for 6 hours and 24 hours, while OsRAN2 was weakly up-regulated in the 24 hour
cold shock treatment. OsMYB2 was highly induced after stress in the rice accessions C148,
C153, C063, C070, and C079. The differential expression patterns of these genes suggested
that these rice accessions may have different mechanisms in response to the cold stress.
Haplotype analysis for the reported gene OsMYB2
Further haplotype analysis was focused on the reported gene OsMYB2 since it was
reported to participate in abiotic stress (including cold, salt, and drought) response at the
seedling stage (Yang et al., 2012), and the L33 containing OsMYB2 was overlapped with the
known QTL qLVG3 (Han et al., 2006). The SNP data was referred to RiceVarMap including
This article is protected by copyright. All rights reserved.
its intragenic region and 2 kb upstream (Zhao et al., 2014a). Four major haplotypes were
observed, with low frequency haplotypes (less than 5 accessions) being omitted (Fig. 3a). We
conducted multiple comparison tests of the traits with log transformation for equal variances,
and the results showed that Hap1 and Hap2 had lower ELSR2 values (corresponding to low
cold shock sensitivity) than Hap3 (corresponding to high cold shock sensitivity) (P<0.05),
while Hap4 was an intermediate type. At the OsMYB2 locus, there was one synonymous SNP
(11325754) and three nonsynonymous SNPs (11325395, 11325497, and 11325747) in the
second exon, six SNPs in 5’ and 3’ untranslated regions, one SNP in the intron, and 35
substitutions in the 2 kb cis-regulatory region. We noticed that two nonsynonymous SNPs led
to changes in amino acids (one at 11325395 caused a C in Hap1 to a Y in the other 3
haplotypes, and the other at 11325747 caused a W in Hap1 to an R in the other 3 haplotypes),
which suggests that these SNPs may be associated with the gene function to a certain degree.
We further checked the latitude distribution of the haplotypes by a scatter diagram of the
latitude of origin of these accessions (Fig. 3b). The accessions in Hap1 were from higher
latitude regions compared to the accessions in the other three haplotypes (P<0.05) (Fig. 3c).
We also checked the subpopulation and geographical distribution of all 412 accessions
in relation to the four haplotypes of OsMYB2 (Fig. 4). It was noticed that 97.75% of the
accessions in Hap1 group belong to japonica, while 89.03% of the accessions in Hap2 belong
to indica. Meanwhile, 77.27% of the accessions in Hap3 group belong to aus subgroup, and
67.81% and 23.29% of the accessions in Hap4 group belong to indica and japonica
respectively (Fig. 4b), which may partially explain why Hap4 exhibits an intermediate type in
terms of cold sensitivity. We also found that accessions in Hap1 in red distributed widely no
This article is protected by copyright. All rights reserved.
matter around the world or China (Fig. 4a,c). These results suggest that the rice accessions in
different haplotypes of OsMYB2 had indica-japonica differentiation and differential
latitudinal distribution tendency.
Indica-japonica differentiation and latitudinal distribution are associated with the cold
adaptability of rice
From the results of the OsMYB2 haplotype analysis above, we hypothesized that
indica-japonica differentiation and the latitudinal distribution of rice may generally be
associated with cold adaptability. To test this hypothesis, we further checked if there existed
such differences between the accessions in our panel based on the relative chilling tolerance
or cold shock sensitivity. We selected 114 chilling tolerant (CT), 143 chilling sensitive (CS),
123 cold shock insensitive (CSI), and 131 cold shock sensitive (CSS) accessions from the
panel according to their comprehensive performance under the two different stress conditions,
and checked their subspecies classification and latitudinal distribution (Supporting
Information Fig. S10, S11). The results showed that 82% of the CT accessions belong to
japonica rice while 76% of the CS accessions belong to indica rice. However, for the cold
shock stress tolerance, the indica-japonica subpopulation distribution for both CSI and CSS
had no obvious tendencies (Supporting Information Fig. S10b, S11b).
The geographic origins of these accessions were also marked on a world map and a map
of China (Supporting Information Fig. S10a, c, and Fig. 11a, c). For the accessions under
natural chilling stress condition, when compared with the CS accessions (marked in blue), the
CT accessions (red) were distributed in a much larger geographic region both in China and
This article is protected by copyright. All rights reserved.
the world (Supporting Information Fig. S10). However, the geographic distribution
differences between CSI and CSS accessions were not obvious. As latitude has a great
influence on temperature which limits rice growth, we also performed scatter diagram and a
statistical analysis for the latitudes of all of the chosen accessions (Fig. 5a, b, Supporting
Information Fig. S12a, b). The results showed that the latitudinal distribution of the CT group
is significantly higher than that of CS group (P=1.28E-03) (Fig. 5a, b), while the two groups
under cold shock stress condition (CSI and CSS) exhibited no differences in latitudinal
distribution (Supporting Information Fig. S12a, b). Among the four types (CT, CS, CSI, CSS),
there are 32 accessions classified as both CT and CSI (tolerant to both stresses) and 38
accessions classified as both CS and CSS (sensitive to both stresses) (Fig. 5c). The latitudes
of the CT and CSI accessions are also generally higher than that of the CS and CSS
accessions (P=8.27E-04) (Fig. 5d, e), while the latitudinal distributions of the overlapped 25
accessions between CT and CSS and the overlapped 32 accessions between CS and CSI
showed no obvious difference (Supporting Information Fig. S12c, d). The subpopulation and
the geographical distribution of the overlapped accessions between any two of the four types
(Supporting Information Fig. S13) showed no obvious trends probably because of the small
sample number. These results together indicate that indica-japonica differentiation and the
latitudinal distribution of rice is associated with cold adaptability, which is essentially
determined by the natural chilling tolerance rather than by cold shock tolerance.
This article is protected by copyright. All rights reserved.
DISCUSSION
Comparison of association and QTL mapping of cold tolerance
Association mapping was applied to a vast range of complex traits which are important
in the agronomy and breeding improvement of many fundamental crops (Hall et al., 2010,
Shao et al., 2011). According to our results, association mapping was more efficient to detect
the number of loci controlling complex traits such as chilling or cold tolerance compared to
the traditional QTL mapping. Previous studies identified quite a few QTLs which contributed
to chilling tolerance either at the seedling stage or the germination stage in rice, such as
qCTS2, qCTS7, qLTG-3, qLTG-11-1, and clr9, which are shown in Fig. 2 (Jiang et al. 2006;
Liu et al. 2013; Oh et al. 2004). However, these QTLs have large physical intervals, making
subsequent fine mapping difficult. When utilizing QTL mapping one always needs to
generate high quality mapping populations, which is costly and time consuming (Hall et al.,
2010, Holland, 2007, Paterson, 1995). Certainly there are successful examples utilizing a
QTL mapping strategy for fine mapping of low temperature tolerance such as qLTG-9 and
qLTG3-1 (Fujino et al., 2008, Fujino et al., 2004, Li et al., 2013, Ma et al., 2015), and even
for cloning of cold tolerance genes such as COLD1 (Ma et al., 2015).
It should be noted that association mapping may lead to false positive associations. In
this study, although most of the 16 traits used for the association analysis were correlated
with each other, 93 of the 132 loci were detected only for one trait, which hinted that some of
these loci (especially those with a high suggestive P-value) may be false associations. In
addition, one association locus in this study (and other studies in rice as well) was defined as
a 200 kb region containing 10 or more genes, which makes further confirmation of these loci
This article is protected by copyright. All rights reserved.
more difficult. For comparison, species such as maize exhibit a small linkage disequilibrium
(LD), which determines the resolution at 50-100 kb for association mapping, as the average
gene number is 1-2 genes within a 100 kb region of the maize genome (Huang et al., 2013).
Therefore, reducing the false positives as much as possible is important for the subsequent
verification of GWAS results and the cloning of causal genes. Besides a low P-value, the loci
with pleiotropic effects or overlapping with known QTL for similar traits can be considered
as positives with high confidence. In this study, 39 loci were detected for two or three traits,
such as L18, L27, L63, L79, and L104 which are indicated in Fig. 1, and 68 loci were
co-localized with known QTL, and these loci may be considered with high priority for further
validation. For example, the L18 in Fig. 2 with the most significant P value was co-located
with the reported QTL qSCT1a/qSCT1B (Kim et al., 2014), which was also detected at the
seedling stage under chilling stress conditions, even though the treatment temperature was
different. For preliminary validation of the association results, we selected a few candidate
genes (Supporting Information Fig. 9, Supporting Information Table 6) for haplotype analysis.
Besides the OsMYB2 that had been reported in cold stress response, LOC_Os07g44410
which encodes a WD-40 protein and was induced by cold stress in the 8 accessions, is
another example. This gene is located in L77 (P = 5.30E-08) which was detected for the trait
ELR and overlapped with the reported QTL qLVG7-2 (Han et al., 2006) . The mean values of
ELR of the accessions in Hap1 and Hap3 of this gene were significantly smaller than that of
Hap2 and Hap4, and the four haplotypes also showed indica-japonica distribution tendency
(Supporting Information Fig. S14a) and difference in latitude distribution (Supporting
Information Fig. S14b, c), which is similar to the haplotype analysis results of OsMYB2.
This article is protected by copyright. All rights reserved.
Therefore, combining the whole genome expression profiling data with haplotype analysis
may narrow down the number of candidate genes.
Rice has low LD decay which made it easy to produce the false positives (Han & Huang,
2013). Actually, LD can be affected by genetic drift, natural selection and population
stratification, and the last one was considered as the major factor (Cardon & Palmer, 2003).
Researchers have explored analytical methods to reduce false positives including structured
association (SA), and genomic control (GC) for population structure (Yu et al., 2006).
Recently, GWAS was successfully adopted to dissect the genetic basis of metabolites by
integrating genomic data and comprehensive metabolic profiles in rice, which allows for the
large-scale identification of candidate genes, the elucidation of metabolic pathways, and for
notably improving the resolution of association mapping (Chen et al., 2014). The
combination of association studies of traits and metabolic changes may be adopted for
dissecting the genetic architecture of complex traits such as chilling or cold tolerance in
future studies.
Distinct genetic basis of natural chilling and cold shock tolerance
The low temperatures action on rice under natural conditions mainly includes two types:
1) natural chilling stress during which the temperature gradually declines to an unfavorable
range for rice growth; 2) cold shock stress caused by the rapid (even overnight) decline of the
temperature nearly to the freezing point, which occasionally occurs in temporal regions at the
rice seedling stage. In most studies of cold stress responses in rice, the artificial cold
treatment is very close to the cold shock, and very few experiments mimic natural chilling
stress conditions. Plants have evolved diverse mechanisms in response to chilling and
This article is protected by copyright. All rights reserved.
freezing temperatures (Zhu et al., 2007). Although some studies on chilling or cold shock
responses have been reported (Chaikam & Karlson, 2008, Chawade et al., 2013, de Los
Reyes et al., 2013, Mao & Chen, 2012, Yang et al., 2013a), the differences between the
genetic basis of chilling and cold tolerance were seldom addressed. To the best of our
knowledge, this study was the first attempt to address the difference in genetic basis of
chilling and cold shock stress tolerance. We found that the performance of various traits
under chilling and cold shock stress conditions exhibited no correlations (Table 2), suggesting
that rice may have evolved different mechanisms to cope with chilling and cold shock
stresses. Taking EL for example, which were the main traits analyzed here, it is generally
accepted that the cell membrane is one of the first targets of abiotic stresses, especially when
different low temperature stress occurs, and the maintenance of their integrity and stability
under stress conditions is a major component of resistance (Morsy et al., 2005, Thomashow,
1999, Whitlow et al., 1992). We observed that most of the loci were unique for chilling or
cold shock stress, except for 12 loci (Fig. 2, Supporting Information Table 3) which were
simultaneously detected for some of the traits under the two stress conditions. Half of the 12
common loci were detected for EL traits, suggesting the genetic basis of EL change under
chilling and cold shock stress tolerance may be partially common. These common loci for
both chilling and cold shock tolerance may provide referential information to unveil the
common molecular mechanisms in response to low temperature stresses.
Another notable result is that 68 of 132 loci were co-localized with known QTL for low
temperature tolerance at different stages of rice development (Fig. 2). Most of the QTL were
detected in rice at the seedling stage, while quite a few were detected in rice at the germination
This article is protected by copyright. All rights reserved.
and reproductive stages, indicating the genetic basis of low temperature tolerance at different
growth periods may have an overlap. Such overlap can be also reflected by the reported QTL
listed in Fig. 2. qGR2, qLTG5-2, and qLTG5-1 which were located at the germination stage
exhibited overlap with qCSH2, qCTS2, qCTSS2B, qCTS5, and qSV-5 detected at the seedling
stage, and this phenomenon was also found in chromosomes 6, 7, 9, 10, and 11. However, an
association study in maize showed that none of the 43 identified SNPs were simultaneously
detected for chilling tolerance at the seedling stage and the germination stage, and the
correlation analysis of these traits also suggested that the genetic basis of chilling tolerance at
the germination and the seedling stages were different (Huang et al., 2013).
Relationship of cold adaption, subpopulation and latitudinal distribution in rice
Tropical species are generally sensitive to cold stress conditions, so the distribution of
plants is partly determined by the sensitivity to low temperatures (Morsy et al., 2005). There
have been reports discussing the cold response difference between indica and japonica rice,
and one study concluded that the initial event was oxidative stress induced by chilling, which
partly explains the differential sensitivities of indica and japonica rice to chilling stress
(Cheng et al., 2007). Another group reported that three japonica cultivars exhibited higher
vigor than the two indica cultivars at the germination and the seedling stages under cold
stress conditions (Morsy et al., 2005). A recent report on the differentiation of COLD1 also
suggested that cultivars with COLD1jap
genotype showed stronger cold tolerance than
COLD1Ind
cultivars (Ma et al., 2015). However, there were few reports focusing on
indica-japonica differentiation of cold tolerance based on a large sample of natural
germplasms. A recent study using a mini core collection of 174 Chinese rice accessions
This article is protected by copyright. All rights reserved.
suggested that japonica rice had a stronger tolerance than indica rice at the germination stage
and the booting stage (Pan et al., 2015). Such difference was also observed in this study. We
observed that 94 japonica accessions displayed a stronger cold tolerance than 109 indica
accessions (Supporting Information Fig. S10). Even though there are some japonica and
indica accessions in our panel which displayed an intermediate type of cold tolerance, our
results in general indicated that japonica rice exhibited stronger cold tolerance capabilities
than indica rice.
From the haplotype analysis, we also found that the latitudinal distribution of rice
germplasms as well as the OsMYB2 haplotypes was associated with cold tolerance. We
checked the geographical distribution of 529 accessions which were divided into 4 subsets
according to their origins (Supporting Information Fig. S15). Although indica subset comes
from lower latitude regions including several major rice growing belts while the japonica
subset distributed more widely, the latitudinal distribution was not significantly different
between indica and japonica rice. Therefore, we propose that the geographical differences
between the CT and CS accessions were likely associated with the differences in cold
tolerance (Fig. 5, Supporting Information Fig. S10).
During the long-term evolution, rice may have evolved with adaptability to different
environments and latitudes. To further prove the association between chilling tolerance and
latitude, we compared the latitudinal distribution of the accessions with similar phenotypic
performance under the two stress conditions: i.e. overlapped accessions between the CT with
the CSI groups, or between the CS and the CSS groups (Fig. 5c,d,e), and the subpopulation
and the geographical distribution of such accessions were summarized in Supporting
This article is protected by copyright. All rights reserved.
Information Fig. S13. Most of the overlapping accessions between CT and CSI groups
(71.88%) are japonica rice with a much wider latitudinal distribution. On the other hand,
almost all (94.72%) of the 38 overlapping accessions between CS and CSS groups are aus
and indica rice with narrow latitudinal distributions (Fig. 5d). Therefore, it can be generally
predicted that japonica rice distributed in higher latitude regions will exhibit stronger
adaptability under cold stress conditions.
In conclusion, we employed a genome wide association strategy with 529 accessions for
rice cold tolerance at the seedling stage, and 68 of all loci from the whole panel were located
or overlapped with reported QTL. We found that the two types of cold stresses (chilling and
cold shock) had a distinct genetic basis. The rice accessions have indica-japonica
differentiation and differential latitudinal distribution with cold adaptability, which is also the
case for the haplotypes of the reported OsMYB2 gene. The loci detected in this study may
provide valuable information for the ecological adaptability of rice, and the accessions
classified as CT or CS would be potential genetic resources for rice improvement.
ACKNOWLEDGEMENTS
This work was supported by grants from the National Program on High Technology
Development (2014AA10603, 2012AA10A303) and the National Natural Science
Foundation (31271316).
This article is protected by copyright. All rights reserved.
REFERENCES
Bryant R., Proctor A., Hawkridge M., Jackson A., Yeater K., Counce P., ..., Fjellstrom R. (2011) Genetic
variation and association mapping of silica concentration in rice hulls using a germplasm collection.
Genetica 139, 1383-1398.
Cardon L.R. & Palmer L.J. (2003) Population stratification and spurious allelic association. Lancet 361,
598-604.
Chaikam V. & Karlson D. (2008) Functional characterization of two cold shock domain proteins from Oryza
sativa. Plant, cell & environment 31, 995-1006.
Chawade A., Lindlof A., Olsson B. & Olsson O. (2013) Global expression profiling of low temperature induced
genes in the chilling tolerant japonica rice Jumli Marshi. PloS one 8, e81729.
Chen N., Xu Y., Wang X., Du C., Du J., Yuan M., ..., Chong K. (2011) OsRAN2, essential for mitosis, enhances
cold tolerance in rice by promoting export of intranuclear tubulin and maintaining cell division under cold
stress. Plant, cell & environment 34, 52-64.
Chen W., Gao Y., Xie W., Gong L., Lu K., Wang W., ..., Luo J. (2014) Genome-wide association analyses
provide genetic and biochemical insights into natural variation in rice metabolism. Nature genetics 46,
714-721.
Cheng C., Yun K.Y., Ressom H.W., Mohanty B., Bajic V.B., Jia Y., ..., de los Reyes B.G. (2007) An early
response regulatory cluster induced by low temperature and hydrogen peroxide in seedlings of
chilling-tolerant japonica rice. BMC genomics 8, 175.
Courtois B., Audebert A., Dardou A., Roques S., Ghneim-Herrera T., Droc G., ..., Dingkuhn M. (2013)
Genome-wide association mapping of root traits in a japonica rice panel. PloS one 8, e78037.
de Los Reyes B.G., Yun S.J., Herath V., Xu F., Park M.R., Lee J.I. & Kim K.Y. (2013) Phenotypic, physiological,
and molecular evaluation of rice chilling stress response at the vegetative stage. Methods in molecular
biology 956, 227-241.
de Oliveira Borba T.C., Brondani R.P., Breseghello F., Coelho A.S., Mendonca J.A., Rangel P.H. & Brondani C.
(2010) Association mapping for yield and grain quality traits in rice (Oryza sativa L.). Genetics and
molecular biology 33, 515-524.
Famoso A.N., Zhao K., Clark R.T., Tung C.W., Wright M.H., Bustamante C., ..., McCouch S.R. (2011) Genetic
architecture of aluminum tolerance in rice (Oryza sativa) determined through genome-wide association
analysis and QTL mapping. PLoS genetics 7, e1002221.
Fujino K. & Sekiguchi H. (2011) Origins of functional nucleotide polymorphisms in a major quantitative trait
locus, qLTG3-1, controlling low-temperature germinability in rice. Plant molecular biology 75, 1-10.
Fujino K., Sekiguchi H., Matsuda Y., Sugimoto K., Ono K. & Yano M. (2008) Molecular identification of a
major quantitative trait locus, qLTG3-1, controlling low-temperature germinability in rice. Proceedings of
the National Academy of Sciences of the United States of America 105, 12623-12628.
Fujino K., Sekiguchi H., Sato T., Kiuchi H., Nonoue Y., Takeuchi Y., ..., Yano M. (2004) Mapping of
quantitative trait loci controlling low-temperature germinability in rice (Oryza sativa L.). Theoretical and
applied genetics 108, 794-799.
Guo Z., Ou W., Lu S. & Zhong Q. (2006) Differential responses of antioxidative system to chilling and drought
in four rice cultivars differing in sensitivity. Plant physiology and biochemistry 44, 828-836.
Hall D., Tegstrom C. & Ingvarsson P.K. (2010) Using association mapping to dissect the genetic basis of
complex traits in plants. Briefings in functional genomics 9, 157-165.
This article is protected by copyright. All rights reserved.
Han B. & Huang X. (2013) Sequencing-based genome-wide association study in rice. Current opinion in plant
biology 16, 133-138.
Han L.Z., Zhang Y.Y., Qiao Y.L., Cao G.L., Zhang S.Y., Kim J.H. & Koh H.J. (2006) Genetic and QTL analysis
for low-temperature vigor of germination in rice. Yi chuan xue bao 33, 998-1006.
Holland J.B. (2007) Genetic architecture of complex traits in plants. Current opinion in plant biology 10,
156-161.
Huang J., Wang M.M., Bao Y.M., Sun S.J., Pan L.J. & Zhang H.S. (2008) SRWD: a novel WD40 protein
subfamily regulated by salt stress in rice (OryzasativaL.). Gene 424, 71-79.
Huang J., Zhang J., Li W., Hu W., Duan L., Feng Y., ..., Yue B. (2013) Genome-wide association analysis of ten
chilling tolerance indices at the germination and seedling stages in maize. Journal of integrative plant
biology 55, 735-744.
Huang X., Wei X., Sang T., Zhao Q., Feng Q., Zhao Y., ..., Han B. (2010) Genome-wide association studies of
14 agronomic traits in rice landraces. Nature genetics 42, 961-967.
Huang X., Zhao Y., Wei X., Li C., Wang A., Zhao Q., ..., Han B. (2012) Genome-wide association study of
flowering time and grain yield traits in a worldwide collection of rice germplasm. Nature genetics 44,
32-39.
Iwai T., Miyasaka A., Seo S. & Ohashi Y. (2006) Contribution of ethylene biosynthesis for resistance to blast
fungus infection in young rice plants. Plant physiology 142, 1202-1215.
Iwata N. & Fujino K. (2010) Genetic effects of major QTLs controlling low-temperature germinability in
different genetic backgrounds in rice (Oryza sativa L.). Genome 53, 763-768.
Jia L., Yan W., Zhu C., Agrama H.A., Jackson A., Yeater K., ..., Wu D. (2012) Allelic analysis of sheath blight
resistance with association mapping in rice. PloS one 7, e32703.
Jung K.H., Gho H.J., Giong H.K., Chandran A.K., Nguyen Q.N., Choi H., ..., An G. (2013) Genome-wide
identification and analysis of Japonica and Indica cultivar-preferred transcripts in rice using 983 Affymetrix
array data. Rice 6, 19.
Kanneganti V. & Gupta A.K. (2008) Overexpression of OsiSAP8, a member of stress associated protein (SAP)
gene family of rice confers tolerance to salt, drought and cold stress in transgenic tobacco and rice. Plant
molecular biology 66, 445-462.
Kim S.M., Suh J.P., Lee C.K., Lee J.H., Kim Y.G. & Jena K.K. (2014) QTL mapping and development of
candidate gene-derived DNA markers associated with seedling cold tolerance in rice (Oryza sativa L.).
Molecular genetics and genomics 289, 333-343.
Knight M.R. & Knight H. (2012) Low-temperature perception leading to gene expression and cold tolerance in
higher plants. The New phytologist 195, 737-751.
Koseki M., Kitazawa N., Yonebayashi S., Maehara Y., Wang Z.X. & Minobe Y. (2010) Identification and fine
mapping of a major quantitative trait locus originating from wild rice, controlling cold tolerance at the
seedling stage. Molecular genetics and genomics 284, 45-54.
Kusumi K., Hirotsuka S., Kumamaru T. & Iba K. (2012) Increased leaf photosynthesis caused by elevated
stomatal conductance in a rice mutant deficient in SLAC1, a guard cell anion channel protein. Journal of
experimental botany 63, 5635-5644.
Li L., Liu X., Xie K., Wang Y., Liu F., Lin Q., ..., Wan J. (2013) qLTG-9, a stable quantitative trait locus for
low-temperature germination in rice (Oryza sativa L.). TAG. Theoretical and applied genetics. Theoretische
und angewandte Genetik 126, 2313-2322.
Li X., Yan W., Agrama H., Jia L., Jackson A., Moldenhauer K., ..., Wu D. (2012) Unraveling the complex trait of
harvest index with association mapping in rice (Oryza sativa L.). PloS one 7, e29350.
This article is protected by copyright. All rights reserved.
Liu F., Xu W., Song Q., Tan L., Liu J., Zhu Z., ..., Sun C. (2013) Microarray-assisted fine-mapping of
quantitative trait loci for cold tolerance in rice. Molecular plant 6, 757-767.
Londo J.P., Chiang Y.C., Hung K.H., Chiang T.Y. & Schaal B.A. (2006) Phylogeography of Asian wild rice,
Oryza rufipogon, reveals multiple independent domestications of cultivated rice, Oryza sativa. Proceedings
of the National Academy of Sciences of the United States of America 103, 9578-9583.
Lou Q., Ma C., Wen W., Zhou J., Chen L., Feng F., ..., Xu G. (2011) Profiling and association mapping of grain
metabolites in a subset of the core collection of Chinese rice germplasm (Oryza sativa L.). Journal of
agricultural and food chemistry 59, 9257-9264.
Lu G., Wu F.Q., Wu W., Wang H.J., Zheng X.M., Zhang Y., ..., Wan J. (2014) Rice LTG1 is involved in adaptive
growth and fitness under low ambient temperature. The Plant journal : for cell and molecular biology 78,
468-480.
Lu L., Shao D., Qiu X., Sun L., Yan W., Zhou X., ..., Xing Y. (2013) Natural variation and artificial selection in
four genes determine grain shape in rice. The New phytologist 200, 1269-1280.
Lu L., Yan W., Xue W., Shao D. & Xing Y. (2012) Evolution and association analysis of Ghd7 in rice. PloS one
7, e34021.
Ma Y., Dai X., Xu Y., Luo W., Zheng X., Zeng D., ..., Chong K. (2015) COLD1 Confers Chilling Tolerance in
Rice. Cell 160, 1209-1221.
Mao D. & Chen C. (2012) Colinearity and similar expression pattern of rice DREB1s reveal their functional
conservation in the cold-responsive pathway. PloS one 7, e47275.
Morsy M.R., Almutairi A.M., Gibbons J., Yun S.J. & de Los Reyes B.G. (2005) The OsLti6 genes encoding
low-molecular-weight membrane proteins are differentially expressed in rice cultivars with contrasting
sensitivity to low temperature. Gene 344, 171-180.
Norton G.J., Douglas A., Lahner B., Yakubova E., Guerinot M.L., Pinson S.R., ..., Price A.H. (2014) Genome
wide association mapping of grain arsenic, copper, molybdenum and zinc in rice (Oryza sativa L.) grown at
four international field sites. PloS one 9, e89685.
Pan Y., Zhang H., Zhang D., Li J., Xiong H., Yu J., ..., Li Z. (2015) Genetic analysis of cold tolerance at the
germination and booting stages in rice by association mapping. PloS one 10, e0120590.
Paterson A.H. (1995) Molecular dissection of quantitative traits: progress and prospects. Genome research 5,
321-333.
Saito K., Hayano-Saito Y., Maruyama-Funatsuki W., Sato Y. & Kato A. (2004) Physical mapping and putative
candidate gene identification of a quantitative trait locus Ctb1 for cold tolerance at the booting stage of rice.
Theoretical and applied genetics 109, 515-522.
Sang T. (2011) Toward the domestication of lignocellulosic energy crops: learning from food crop domestication.
Journal of integrative plant biology 53, 96-104.
Shao Y., Jin L., Zhang G., Lu Y., Shen Y. & Bao J. (2011) Association mapping of grain color, phenolic content,
flavonoid content and antioxidant capacity in dehulled rice. Theoretical and applied genetics 122,
1005-1016.
Shinada H., Iwata N., Sato T. & Fujino K. (2013) Genetical and morphological characterization of cold
tolerance at fertilization stage in rice. Breeding science 63, 197-204.
Shirasawa S., Endo T., Nakagomi K., Yamaguchi M. & Nishio T. (2012) Delimitation of a QTL region
controlling cold tolerance at booting stage of a cultivar, 'Lijiangxintuanheigu', in rice, Oryza sativa L.
Theoretical and applied genetics 124, 937-946.
This article is protected by copyright. All rights reserved.
Strigens A., Freitag N.M., Gilbert X., Grieder C., Riedelsheimer C., Schrag T.A., ..., Melchinger A.E. (2013)
Association mapping for chilling tolerance in elite flint and dent maize inbred lines evaluated in growth
chamber and field experiments. Plant, cell & environment 36, 1871-1887.
Suh J.P., Jeung J.U., Lee J.I., Choi Y.H., Yea J.D., Virk P.S., ..., Jena K.K. (2010) Identification and analysis of
QTLs controlling cold tolerance at the reproductive stage and validation of effective QTLs in cold-tolerant
genotypes of rice (Oryza sativa L.). Theoretical and applied genetics. 120, 985-995.
Thomashow M.F. (1999) PLANT COLD ACCLIMATION: Freezing Tolerance Genes and Regulatory
Mechanisms. Annual review of plant physiology and plant molecular biology 50, 571-599.
Tian Z., Qian Q., Liu Q., Yan M., Liu X., Yan C., ..., Li J. (2009) Allelic diversities in rice starch biosynthesis
lead to a diverse array of rice eating and cooking qualities. Proceedings of the National Academy of
Sciences of the United States of America 106, 21760-21765.
Wang C., Wei Q., Zhang K., Wang L., Liu F., Zhao L., ..., Su Z. (2013) Down-regulation of OsSPX1 causes high
sensitivity to cold and oxidative stresses in rice seedlings. PloS one 8, e81849.
Wang N., Xiao B. & Xiong L. (2011) Identification of a cluster of PR4-like genes involved in stress responses in
rice. Journal of plant physiology 168, 2212-2224.
Wang S.T., Sun X.L., Hoshino Y., Yu Y., Jia B., Sun Z.W., ..., Zhu Y.M. (2014) MicroRNA319 positively
regulates cold tolerance by targeting OsPCF6 and OsTCP21 in rice (Oryza sativa L.). PloS one 9, e91357.
Whitlow T.H., Bassuk N.L., Ranney T.G. & Reichert D.L. (1992) An improved method for using electrolyte
leakage to assess membrane competence in plant tissues. Plant physiology 98, 198-205.
Wu W., Zheng X.M., Lu G., Zhong Z., Gao H., Chen L., ..., Wan J. (2013) Association of functional nucleotide
polymorphisms at DTH2 with the northward expansion of rice cultivation in Asia. Proceedings of the
National Academy of Sciences of the United States of America 110, 2775-2780.
Yamori W., Sakata N., Suzuki Y., Shikanai T. & Makino A. (2011) Cyclic electron flow around photosystem I
via chloroplast NAD(P)H dehydrogenase (NDH) complex performs a significant physiological role during
photosynthesis and plant growth at low temperature in rice. The Plant journal : for cell and molecular
biology 68, 966-976.
Yan W., Liu H., Zhou X., Li Q., Zhang J., Lu L., ..., Xing Y. (2013) Natural variation in Ghd7.1 plays an
important role in grain yield and adaptation in rice. Cell research 23, 969-971.
Yan W.G., Li Y., Agrama H.A., Luo D., Gao F., Lu X. & Ren G. (2009) Association mapping of stigma and
spikelet characteristics in rice (Oryza sativa L.). Molecular breeding 24, 277-292.
Yang A., Dai X. & Zhang W.H. (2012) A R2R3-type MYB gene, OsMYB2, is involved in salt, cold, and
dehydration tolerance in rice. Journal of experimental botany 63, 2541-2556.
Yang C., Li D., Mao D., Liu X., Ji C., Li X., ..., Zhu L. (2013a) Overexpression of microRNA319 impacts leaf
morphogenesis and leads to enhanced cold tolerance in rice (Oryza sativa L.). Plant, cell & environment 36,
2207-2218.
Yang J., Zaitlen N.A., Goddard M.E., Visscher P.M. & Price A.L. (2014a) Advantages and pitfalls in the
application of mixed-model association methods. Nature genetics 46, 100-106.
Yang W., Guo Z., Huang C., Duan L., Chen G., Jiang N., ..., Xiong L. (2014b) Combining high-throughput
phenotyping and genome-wide association studies to reveal natural genetic variation in rice. Nature
communications 5, 5087.
Yang Z., Huang D., Tang W., Zheng Y., Liang K., Cutler A.J. & Wu W. (2013b) Mapping of quantitative trait
loci underlying cold tolerance in rice seedlings via high-throughput sequencing of pooled extremes. PloS
one 8, e68433.
This article is protected by copyright. All rights reserved.
Yu J., Pressoir G., Briggs W.H., Vroh Bi I., Yamasaki M., Doebley J.F., ..., Buckler E.S. (2006) A unified
mixed-model method for association mapping that accounts for multiple levels of relatedness. Nature
genetics 38, 203-208.
Yu S.B., Xu W.J., Vijayakumar C.H., Ali J., Fu B.Y., Xu J.L., ..., Li Z.K. (2003) Molecular diversity and
multilocus organization of the parental lines used in the International Rice Molecular Breeding Program.
Theoretical and applied genetics 108, 131-140.
Zhao H., Yao W., Ouyang Y., Yang W., Wang G., Lian X., ..., Xie W. (2014a) RiceVarMap: a comprehensive
database of rice genomic variations. Nucleic acids research 43, 1018-1022.
Zhao J., Zhang S., Yang T., Zeng Z., Huang Z., Liu Q., ..., Liu B. (2014b) Global transcriptional profiling of a
cold-tolerant rice variety under moderate cold stress reveals different cold stress response mechanisms.
Physiologia plantarum 29.
Zhao K., Tung C.W., Eizenga G.C., Wright M.H., Ali M.L., Price A.H., ..., McCouch S.R. (2011) Genome-wide
association mapping reveals a rich genetic architecture of complex traits in Oryza sativa. Nature
communications 2, 467.
Zhou L., Zeng Y., Zheng W., Tang B., Yang S., Zhang H., ..., Li Z. (2010) Fine mapping a QTL qCTB7 for cold
tolerance at the booting stage on rice chromosome 7 using a near-isogenic line. Theoretical and applied
genetics 121, 895-905.
Zhu J., Dong C.H. & Zhu J.K. (2007) Interplay between cold-responsive gene regulation, metabolism and RNA
processing during plant cold acclimation. Current opinion in plant biology 10, 290-295.
Tab
le 1
. P
erfo
rman
ces
of
the
16 t
rait
s in
this
ass
oci
atio
n s
tud
y.
Ab
bre
via
tio
n
Ra
ng
e
Mea
n
SD
C
V
Tra
it
BM
R
0.8
74
-4.9
29
2.0
91
0.9
86
0.4
71
Fre
sh v
s d
ry b
iom
ass
rat
io a
fter
nat
ura
l ch
illi
ng s
tres
s
RL
C
1.0
00
-5.0
00
3.7
93
1.4
54
0.3
83
Res
ista
nce
level
(sc
ore
) u
nd
er n
atura
l chil
lin
g s
tres
s
RL
R
1.0
00
-5.0
00
3.5
22
0.8
83
0.2
51
Res
ista
nce
level
(sc
ore
) af
ter
reco
ver
y
SR
C
0.0
00
-1.0
00
0.3
54
0.3
75
1.0
61
Surv
ival
rat
e aft
er n
atu
ral
chil
ling s
tres
s
EL
N
0.0
53
-0.2
01
0.0
93
0.0
17
0.1
80
Ele
ctro
lyte
lea
kage
und
er n
orm
al c
ond
itio
n
EL
C1
0
.06
2-0
.19
0
0.1
11
0.0
20
0.1
82
Ele
ctro
lyte
lea
kage
afte
r nat
ura
l ch
illi
ng s
tres
s fo
r 3
day
s
EL
C2
0
.08
3-1
.13
0
0.3
12
0.3
31
1.0
61
Ele
ctro
lyte
lea
kage
afte
r nat
ura
l ch
illi
ng s
tres
s fo
r 7
day
s
EL
R
0.0
59
-1.1
59
0.6
30
0.4
27
0.6
79
Ele
ctro
lyte
lea
kage
afte
r re
cover
y f
or
7 d
ays
EL
R1
0
.65
9-2
.14
7
1.2
11
0.2
44
0.2
02
Rat
io o
f el
ectr
oly
te l
eakage
un
der
mil
d n
atura
l chil
lin
g s
tres
s to
no
rmal
cond
itio
n
EL
R2
0
.84
6-1
5.0
97
3.4
26
3.6
39
1.0
62
Rat
io o
f el
ectr
oly
te l
eakage
un
der
sev
ere
nat
ura
l chil
lin
g s
tress
to
no
rmal
co
nd
itio
n
EL
R3
0
.75
2-1
9.2
59
6.9
52
4.9
45
0.7
11
Rat
io o
f el
ectr
oly
te l
eakage
un
der
str
ess
reco
ver
y t
o n
orm
al c
ond
itio
n
EL
SN
0
.07
4-0
.29
0
0.1
43
0.0
38
0.2
67
Ele
ctro
lyte
lea
kage
bef
ore
clo
d (
4°C
) sh
ock
(no
rmal
cond
itio
n)
EL
SC
1
0.0
73
-0.4
28
0.1
63
0.0
39
0.2
41
Ele
ctro
lyte
lea
kage
afte
r co
ld s
ho
ck f
or
1 d
ays
EL
SC
2
0.0
80
-0.7
39
0.1
92
0.0
68
0.3
56
Ele
ctro
lyte
lea
kage
afte
r co
ld s
ho
ck f
or
3 d
ays
EL
SR
1
0.5
59
-3.7
30
1.1
96
0.4
02
0.3
36
Rat
io o
f el
ectr
oly
te l
eakage
un
der
1-d
ay c
old
sho
ck s
tres
s to
no
rmal
co
nd
itio
n
EL
SR
2
0.5
46
-8.3
56
1.4
03
0.6
69
0.4
77
Rat
io o
f el
ectr
oly
te l
eakage
un
der
3-d
ay c
old
sho
ck s
tres
s to
no
rmal
co
nd
itio
n
This article is protected by copyright. All rights reserved.
Tab
le 2
. C
orr
elat
ion c
oef
fici
ents
of
pai
red t
rait
s of
all
trai
ts i
nves
tigat
ed.
B
MR
R
LC
R
LR
S
RC
E
LN
E
LC
1
EL
C2
E
LR
E
LR
1
EL
R2
E
LR
3
EL
SN
E
LS
C1
E
LS
C2
E
LS
R1
E
LS
R2
BM
R
1.0
00
RL
C
0.5
21
1.0
00
RL
R
0.5
86
0.6
63
1.0
00
SR
C
0.5
65
0.6
71
0.6
77
1.0
00
EL
N
0.6
12
0.6
90
0.7
87
0.7
03
1.0
00
EL
C1
0
.56
7
0.6
28
0.7
53
0.6
56
0.7
80
1.0
00
EL
C2
0
.41
6
0.4
71
0.5
46
0.5
14
0.5
71
0.6
18
1.0
00
EL
R
0.4
72
0.6
23
0.6
07
0.6
44
0.6
90
0.6
43
0.5
04
1.0
00
EL
R1
0
.34
8
0.3
79
0.4
62
0.4
12
0.5
25
0.5
27
0.4
00
0.4
28
1.0
00
EL
R2
0
.40
3
0.4
71
0.5
14
0.4
81
0.6
70
0.5
84
0.9
49
0.5
02
0.4
18
1.0
00
EL
R3
0
.46
6
0.6
04
0.5
79
0.6
13
0.7
59
0.6
14
0.4
73
0.9
59
0.4
36
0.5
47
1.0
00
EL
SN
0
.22
9
0.2
22
0.2
59
0.2
18
0.2
71
0.2
56
0.2
02
0.1
94
0.1
30
0.1
86
0.1
82
1
.00
0
EL
SC
1
0.0
73
0.0
73
0.1
25
0.0
90
0.1
14
0.1
21
0.1
04
0.0
76
0.0
65
0.0
87
0.0
66
0
.29
1
1.0
00
EL
SC
2
0.1
55
0.1
83
0.2
07
0.1
91
0.2
02
0.2
21
0.1
47
0.1
61
0.1
18
0.1
28
0.1
45
0
.43
2
0.4
25
1.0
00
EL
SR
1
0.0
79
0.0
60
0.1
13
0.0
77
0.1
02
0.1
09
0.0
99
0.0
62
0.0
48
0.0
81
0.0
52
0
.39
9
0.9
70
0.4
09
1.0
00
EL
SR
2
0.1
50
0.1
52
0.1
78
0.1
61
0.1
74
0.1
91
0.1
28
0.1
31
0.0
83
0.1
08
0.1
15
0
.60
4
0.4
04
0.9
31
0.4
65
1.0
00
(Lig
ht
gre
y b
ack
gro
und m
eans
corr
elat
ion, 0.3<
r<0.8
; dar
k g
rey b
ack
gro
und
mea
ns
stro
ng c
orr
elat
ion
, 0.8≤
r)
This article is protected by copyright. All rights reserved.
This article is protected by copyright. All rights reserved.
Figure 1. Manhattan plots for six traits: (a): ELR2, (b): ELC2, (c): RLC, (d): ELSR2, (e): ELSC1, (f):
SRC. The negative log10 transformed p-values of genome-wide scan are plotted against the marker
position in the genome. Dotted line: P=1e-06. The positions of 5 reported genes were indicated with
black arrow. Examples of some loci were indicated; L27, L104, L63 in red were detected by EL
related traits under the two cold stress conditions while locus L18 and L79 in blue were identified by
two traits under natural chilling stress condition. L is short for locus.
This article is protected by copyright. All rights reserved.
Figure 2. Distribution of 132 loci on 12 chromosomes according to physical distance. The loci and its
associated traits were marked on the right of chromosomes with the relative position of each locus
(200 kb) shown by its own front physical position; the overlapping known QTL were shown in grey
column on the left while the markers of QTL were shown on the right. The 57 loci in blue were
identified under natural chilling stress condition, the 63 loci in green were identified under cold shock
stress condition, and the 12 loci in red were detected under both stress conditions.
This article is protected by copyright. All rights reserved.
Figure 3. Haplotype analysis of OsMYB2. (a) Haplotypes in 412 accessions (haplotypes with
less than 5 accessions was omitted) according to SNP data from RiceVarMap based on
MSU6.1 annotation. The region contained coding region and 2 kb upstream of the gene.
Letters on the right of the average are ranked by Duncan test at P<0.05, different letters
indicate significant difference. (b) Scatter diagram for the latitudes (ascending sorted) of the
origins of accessions of the four haplotypes at OsMYB2 locus. (c) Comparison of latitude
distribution between accessions of the four haplotypes. Letters above the bars are ranked by
Duncan test at P<0.05; different letters indicate significant difference.
This article is protected by copyright. All rights reserved.
Figure 4. Geographic and subpopulation distribution of accessions of the four OsMYB2
haplotypes. Geographic distribution of accessions in the four haplotypes on world map (a)
and map of China (c), the solid circle in different colors (red, light blue, dark blue, green)
represents accession numbers of the four haplotypes (Hap1−4 respectively), except for 5
accessions in Hap1, 16 accessions in Hap2 and 14 accessions in Hap4 with unknown
geographic location. (b) The subpopulation distribution of 89 accessions in Hap1, 155
accessions in Hap2, 22 accessions in Hap3 and 146 accessions in Hap4.
This article is protected by copyright. All rights reserved.
Figure 5. Latitudinal distributions of CT, CS accessions and the overlapped accessions under the two
stress conditions. (a) Scatter diagram for the latitudinal distribution of 108 CT accessions (red) and
131 CS accessions (blue), sorted by ascending latitude. (b) Comparison of latitude distribution
between CT and CS accessions. (c) Scatter diagram for the latitude distribution of 32 overlapped
accessions between CT and CSI (red) and 38 overlapped accessions between CS and CSS (blue). (e)
Comparison of latitude distribution between accessions overlapped by CT, CSI and CS, CSS. CT=
chilling tolerant accessions; CS= chilling sensitive accessions; CSS= cold shock sensitive accessions;
CSI= cold shock insensitive accessions. Differences in of latitude distribution between accessions
were examined by Student's t-test. *, P<0.05,**, P<0.01.