The genetic dissection and cloning of quantitative trait loci (QTLs) are prerequisites for implementing genomics-based applications in breeding programs. However, traditional positional cloning for QTLs is difficult, laborious and time-consuming. For example, more than 250 and 600 QTLs for cold tolerance have been detected in rice using genetic mapping and genome-wide association studies, respectively, but only 15 of these have been positionally cloned (Li et al., 2022).

We previously reported the identification of 5636 non-synonymous SNPs in protein-coding genes fixed in at least one rice subspecies, based on a genome-wide selective sweep analysis of 655 japonica and 1205 indica accessions. These accessions were selected from 2673 landraces through SNP-assisted principle component analysis, excluding 813 admixed accessions. We classified these selective sweep SNPs (SSNPs) into three types, jS (japonica-selected), iS (indica-selected) and jiS (japonica- and indica-selected) (Wang et al., 2022). As the two subspecies differ markedly in their adaptation to different environmental temperatures, SSNPs could thus serve as markers assisting candidate-gene selection for cloning cold-tolerance QTL genes. To efficiently identify new cold-tolerance genes, we built a comparative genomics analysis and QTL-associated cloning platform, called SSNP-guided cloning (SSNPC). As a case study for this method, we attempted to clone causal genes of 11 primarily mapped cold-tolerance QTLs (Figure 1a).

The 11 QTL regions (~2 Mbp each) encompass a total of 4588 predicted genes, 2901 of which have non-synonymous SNPs between the japonica and indica groups. To detect signals of strong selection, we used a lab-developed software package (RGVD, REAG and RCRAP) to calculate pooled heterozygosity (H_P = 2∑n_MAJ∑n_MIN/(∑n_MAJ + ∑n_MIN)², where n_MAJ and n_MIN are counts of the most and least abundant allele for each SNP, respectively; Axelsson et al., 2013) and fixation index (F_ST) values in sliding 10-kb windows for all these SNPs. In rice, we applied the cutoffs of 0.001 and 0.95 for H_P and F_ST, respectively, to define selective sweeps with reduced H_P and/or increased F_ST, which reflect strong selection during the evolution and domestication of japonica and indica rice populations. Thus, the SNPs within the selective sweeps, namely SSNPs, may assist in picking candidate genes with ‘large’ effects on cold tolerance. Indeed, using this algorithm we detected SSNPs in 11 of the 15 positionally cloned cold-tolerance genes in rice (Table S1). We therefore selected genes containing one or more SSNPs of the jS and jiS types (H_P-japonica < 0.001 and/or F_ST > 0.95) as cold-tolerance gene candidates in these 11 QTL regions. In this way, 27 candidate genes stood out from 2901 genes with non-synonymous SNPs; these candidate genes were denoted in numerical order as Cold Tolerance-1 (CT-1) to CT-27 (Figure 1a; Tables S2 and S3). The reduced H_P scores of divergent haplotypes of gene loci may also reflect the natural selection of the genes. We analysed the H_P scores for the haplotype SNPs in the coding regions of the 27 SSNP-selected candidate genes and found that only three of them (CT-1, CT-6 and CT-21) had H_p-japonica scores of <0.001, as compared to 18 genes with the SSNPs having these low H_p-japonica scores (Tables S3 and S4). Therefore, SSNPs should be more suitable than haplotypes for candidate selection.

Using the plant CRISPR/Cas9 system (Ma et al., 2015; Tables S5 and S6), we knocked out each of these 27 candidate genes in a japonica variety, Nipponbare and obtained homozygous mutant lines (ct-1 to ct-27, T₁ and T₂ generations, ct-23 was male-sterile). We investigated their cold sensitivities by subjecting 15-day-old Nipponbare (WT) and mutant seedlings to 4 °C for 48 h, then returned them to room temperature (RT) for 5 days to allow growth recovery. Seven mutant lines, ct-5, ct-9, ct-10, ct-17, ct-19, ct-24 and ct-26, showed significantly decreased seedling survival rates compared to the WT (Figure 1b). Six of these seven CT genes (except CT-5) were cold-responsive at the mRNA level (Figure 1c; Table S7).

To evaluate cold tolerance in the mutant lines, we subjected four representative mutant lines (ct-5, ct-9, ct-19 and ct-26) to 4 °C and measured the physiological indices including soluble sugars, proline, malondialdehyde (MDA), peroxidase (POD), chlorophyll and various chlorophyll fluorescence parameters (the maximal quantum yield of dark-adapted leaves F_v/F_m, the effective quantum yield of illuminated leaves Y(II), photochemical quenching qP and nonphotochemical quenching NPQ). After the cold treatment, all mutants showed lower soluble sugar contents than WT plants and some also displayed lower proline amount, more MDA accumulations and lower POD activities than the WT (Figure 1d). All mutants also demonstrated significant decreases in chlorophyll a, chlorophyll b, total chlorophyll and chlorophyll fluorescence parameter values (Figure 1e), indicative of impaired photosynthesis. These observations confirmed the involvement of these mutants in cold signalling.

Our SSNPC method is more effective and enables higher throughput than traditional positional cloning of adaptive QTLs. It does not require the breeding of mapping populations and laborious genetic mapping to narrow down candidate intervals but uses SSNPs instead identified from genomics data of (sub)species/ecotypes for candidate selection. Our test study showed that this technique can identify and functionally validate multiple authentic cold-tolerance genes in rice within a short timeframe (<2 years), as compared to many years spent in map-based cloning of one single QTL. Our SSNPC could be leveraged into QTL cloning systems for other adaptive traits in rice and other crops. Our selective-sweep analysis also identified numerous SSNPs fixed in gene promoter regions (Wang et al., 2022). Recent studies indicated that natural variations in the HAN1 promoter (Mao et al., 2019) and codon repeat variation in COLD11 (Li et al., 2023) confer cold tolerance in rice. Therefore, the combined use of SSNPs in coding regions, promoter regions and structural variations may further improve the efficiency of the SSNPC method.

数量性状基因座（QTL）的遗传解剖和克隆是在育种计划中实施基于基因组学的应用的先决条件。然而，传统的QTL定位克隆方法困难、费力且耗时。例如，利用遗传作图和全基因组关联研究，分别在水稻中检测到了超过 250 个和 600 个耐冷 QTL，但其中只有 15 个被定位克隆（Li等，  2022）。

我们之前报道过，基于对 655 个粳稻和 1205 个籼稻品种的全基因组选择性扫描分析，在至少一个水稻亚种固定的蛋白质编码基因中鉴定出了 5636 个非同义 SNP 。这些材料是通过SNP辅助主成分分析从2673个地方品种中筛选出来的，排除了813个混合材料。我们将这些选择性清除 SNP (SSNP) 分为三种类型：j S（粳稻选择）、i S（籼稻选择）和ji S（粳稻和籼稻选择）（Wang等人，  2022）。由于两个亚种对不同环境温度的适应能力存在显着差异，因此SSNPs可以作为辅助候选基因选择的标记，用于克隆耐冷QTL基因。为了有效地识别新的耐冷基因，我们建立了一个比较基因组学分析和QTL相关克隆平台，称为SSNP引导克隆（SSNPC）。作为该方法的案例研究，我们尝试克隆 11 个主要定位的耐冷 QTL 的因果基因（图 1a）。

11 个 QTL 区域（每个约 2 Mbp）总共包含 4588 个预测基因，其中 2901 个在粳稻组和籼稻组之间具有非同义 SNP 。为了检测强选择信号，我们使用实验室开发的软件包（RGVD、REAG 和 RCRAP）来计算合并杂合度 ( H _P  = 2Σ n _MAJ Σ n _MIN /(Σ n _MAJ  + Σ n _MIN ) ²，其中n _MAJ和n _MIN分别是每个 SNP 的最丰富和最不丰富等位基因的计数；Axelsson等人，  2013）以及所有这些 SNP 滑动 10 kb 窗口中的固定指数 ( F _{ST ) 值。}在水稻中，我们分别对HP和F _ST应用 0.001 和 0.95 的截止值_，以定义降低HP和/或增加F ST_的选择性扫描，这反映了粳稻_和籼稻种群进化和驯化过程中的强烈选择。因此，选择性扫描中的 SNP（即 SSNP）可能有助于挑选对耐冷性具有“大”影响的候选基因。事实上，使用该算法，我们在水稻 15 个定位克隆的耐寒基因中的 11 个中检测到了 SSNP（表 S1）。因此，我们选择含有j S 和ji S 型（HP - _japonica  < 0.001 和/或F _{ST >} 0.95 ）的一个或多个 SSNP 的基因 作为这 11 个 QTL 区域中的耐冷基因候选。这样，27个候选基因从2901个具有非同义SNP的基因中脱颖而出；这些候选基因按数字顺序表示为Cold Tolerance-1 ( CT-1 ) 至CT-27（图 1a；表 S2 和 S3）。_{基因位点不同单倍型的 HP}分数降低也可能反映了基因的自然选择。我们分析了27 个 SSNP 选择的候选基因编码区中单倍型 SNP 的HP得分，发现只有其中 3 个（ CT-1、CT-6和CT-21）的H _{p- japonica}得分< 0.001，与具有这些低H _p- japonica的 SSNP 的 18 个基因相比分数（表 S3 和 S4）。因此，SSNP 应该比单倍型更适合候选选择。

使用植物CRISPR/Cas9系统（Ma等，  2015；表S5和S6），我们敲除粳稻品种日本晴中的这27个候选基因中的每一个，并获得纯合突变系（ct-1至ct-27， T ₁和T ₂代，ct-23是雄性不育）。我们通过将 15 天大的日本晴 (WT) 和突变幼苗置于 4°C 下 48 小时，然后将其放回室温 (RT) 5 天以使其生长恢复，从而研究了它们的冷敏感性。与 WT 相比，七个突变系ct-5、ct-9、ct - 10、ct -17 、ct-19、ct-24和ct-26 的幼苗存活率显着降低（图 1b）。这七个CT基因中的六个（CT-5除外）在 mRNA 水平上具有冷响应性（图 1c；表 S7）。

为了评价突变株系的耐冷性，我们将四个代表性突变株系（ct-5、ct-9、ct-19和ct-26）置于4℃下，测量了可溶性糖、脯氨酸、丙二醛（MDA）等生理指标。 ）、过氧化物酶（POD）、叶绿素和各种叶绿素荧光参数（暗适应叶片的最大量子产率F _v / F _m、光照叶片的有效量子产率 Y(II)、光化学猝灭 qP 和非光化学猝灭 NPQ）。冷处理后，所有突变体均表现出比野生型植物更低的可溶性糖含量，并且一些突变体还表现出比野生型植物更低的脯氨酸含量、更多的MDA积累和更低的POD活性（图1d）。所有突变体还表现出叶绿素a、叶绿素b、总叶绿素和叶绿素荧光参数值显着降低（图 1e），表明光合作用受损。这些观察证实了这些突变体参与冷信号传导。

我们的 SSNPC 方法比传统的适应性 QTL 定位克隆更有效，并且具有更高的通量。它不需要育种绘图种群和费力的遗传绘图来缩小候选区间，而是使用从（亚）物种/生态型的基因组数据中识别的 SSNP 来进行候选选择。我们的测试研究表明，这项技术可以在短时间内（<2 年）内鉴定水稻中多个真实的耐寒基因并对其进行功能验证，而基于图谱克隆一个 QTL 则需要花费很多年的时间。我们的 SSNPC 可以用于水稻和其他作物的其他适应性性状的 QTL 克隆系统。我们的选择性扫描分析还发现了许多固定在基因启动子区域的 SSNP（Wang等人，  2022）。最近的研究表明，HAN1启动子的自然变异（Mao等，  2019 ）和COLD11的密码子重复变异（Li等，  2023）赋予水稻耐冷性。因此，编码区、启动子区和结构变异中SSNP的组合使用可能会进一步提高SSNPC方法的效率。