The yield of rice F₁ hybrid seed production is influenced by parental line traits, including stigma exsertion rate (SER), which impacts seed pricing and utilization (Marathi and Jena, 2015). SER is highly susceptible to environmental fluctuations, phenotypic and complex genetic factors (Miyata et al., 2007). Although over 40 QTLs related to SER have been identified, none have been molecularly characterized due to differences in genetic background and small additive effects (Zhu et al., 2023).

We have demonstrated a positive correlation between stigma size and SER previously (Tan et al., 2023). Then we previously located the stigma size gene SER1 within a 470 kb interval on chromosome 2 based on SSSL-42, a Single Segment Substitution Line with Huajingxian74 (HJX74) as the recipient parent (Tan et al., 2021). In this study, homozygous recombinant lines derived from the crossing of SSSL-42 and HJX74 allowed the region of SER1 to be narrowed down to a 29.48 kb stretch flanked by markers QY18 and LST2 (Figure 1a, Table S1). The line R4, with the shortest substitution segment, was identified as a near-isogenic line for SER1 (NIL-SER1), while the HJX74 was referred to as NIL-ser1 (Figure 1a). The NILs did not differ from one another in many agronomic traits, but the significant difference in SER and stigma size were detected between NILs (Figures 1b–e and S1).

There are three candidate genes (OsSPL5, OsCH240 and OsSm-F) detected related to the mapped interval. Variant analysis revealed that OsSPL5 harbours two nucleotide polymorphisms in the third exon in the mapped interval, resulting in amino acid substitutions (Figure S2). The transcriptome assays of the stigma revealed no significant differences in the gene expression among these three genes between the NILs (Figure S3). To investigate the candidate gene for SER1, we obtained over-expression lines and knockout lines for the three candidate genes. Either the gene-edited lines in the NIL-SER1 background or over-expression lines in the NIL-ser1 background, the transgenic lines of OsCH240 and OsSm-F exhibited no phenotypic changes in stigma size (Figure S4). However, the over-expression of OsSPL5 resulted in enlarged stigmas, whereas the knockout of OsSPL5 led to smaller stigmas (Figure 1g–j). Furthermore, the stigma exertion rate changed accordingly in different transgenic lines of OsSPL5 (Figure 1e,g,h). Thus, the candidate gene for SER1 is OsSPL5. The further RT-qPCR assay detected no variation in transcriptional levels across different tissues (Figure S5). Furthermore, the stigma size dramatically decreased in the gene-edited lines, KO-SER1-3rd exon (Figure S6). It strongly suggests that the sequence variation located in the third exon of OsSPL5 is the primary cause of the phenotypic differences between NILs.

Scanning electron microscopy (SEM) indicated a significant increase in the epidermal cell length of both stigma brush (SB) and no-brush parts (SNB) in NIL-SER1 (Figures 1f and S7). Anatomical observations revealed that stigma size differences between NILs gradually increased as the spikelet developed (Figure S8). These results indicated that OsSPL5 regulates stigma size in rice by affecting cell size, which subsequently impacts SER. OsSPL5 is an important member of the SPL family, and subcellular localization indicates that it is primarily located in the nucleus (Figure S9). Transcriptional activation assays elucidated that the activation domain of OsSPL5 is located in the N-terminal region (Figure S10). Transcriptome analysis identified 3331 and 759 significant different expression genes in NIL-SER1 vs NIL-ser1 and NIL-ser1 vs KO-SER1, respectively (Table S2). Single transcription factor differentially regulate downstream target genes based on their functional strength in diverse allelic backgrounds, while gene knockout typically results in loss of function. A total of 379 genes exhibited increased expression in NIL-SER1 compared to NIL-ser1, while exhibiting decreased expression in KO-SER1 compared to NIL-SER1 (Figure 1k and Table S2).

The Cut&Tag-Seq assays with GFP-SER1 fusion-transformed protoplasts revealed 2153 promoter- (−1000 to 0) related peaks associated with 1852 genes (Figure 1l; Table S3). Subsequent association analysis pinpointed 27 candidate downstream target genes of SER1, which displayed upregulation in NIL-SER1 and down regulation in KO-SER1 (Figure 1k; Table S4). Notably, DEP1, a gene that encodes the γ subunit of the heterotrimeric G-protein and is known as a crucial factor for spikelet and flower development in rice (Huang et al., 2009, 2022), was identified as one of the key candidate genes. The enrichment of SBP binding motifs (GTAC) was observed in its promoter, overlapping with the Cut&Tag peak summit (Figure 1k,l). Alphafold3 docking illustrated a strong binding interaction between the DEP1 promoter motif and SER1 (Figure S11). The combination of data from Cut&Tag-Seq and further assays of DAP-Seq illustrated a ~230 bp binding window of SER1 in the DEP1 promoter (Figure S12). The further promoter-LUC assay confirmed that SER1 binding to the DEP1 promoter enhanced downstream gene expression (Figure 1m). This finding was reinforced by Y1H assays, indicating a positive regulatory relationship between SER1 and DEP1 (Figure 1n). The investigation of relationship between DEP1 alleles and stigma size elucidated that the NIL-dep1-ser1 shows an elevation in stigma width compared to NIL-ser1 (NIL-DEP1-ser1) (Figure 1o,p). Furthermore, DEP1 knockout lines derived from NIL-SER1 showed a significant reduction in stigma size (Figure 1q,r). Additionally, the over-expression of SER1 resulted in shorter panicles, while KO-SER1 exhibited elongated panicles (Figures S13 and S14), consistent with known DEP1 functions. These findings suggest that SER1 exerts a positive regulatory effect on DEP1, thereby modulating stigma and panicle development in rice through the G-protein signalling pathway.

To explore the natural variation of the SER1 gene, a total of 2042 Oryza accessions displaying extensive genetic diversity were analysed (Yao et al., 2019). SNP analysis of OsSPL5 indicated that SER1 and ser1 are the two predominant haplotypes. These two haplotypes are widespread in wild rice, but in cultivated rice, nearly all varieties of the indica subspecies carry ser1, while the japonica subspecies predominantly carry SER1. This indicates near complete differentiation between indica and japonica rice at this locus (Figure 1s; Table S5). The nucleotide diversity (π) of the SER1 gene is extremely low within the two cultivated rice subspecies (Figure 1t). This suggests significant potential for SER1 in indica rice breeding programs. The downstream DEP1 gene also shows strong inter-subspecific differentiation, suggesting co-selection during domestication (Figure S15; Table S5). We further introgressed the SER1 gene into lines of P132-16A and P132-16B, which is an indica male sterile line and its corresponding restorer line derived from HJX74 with a ser1 genetic background (Figure 1u–w). The restorer line, P132-16B-SER1, exhibited a heritably higher SER compared to both HJX74 and P132-16B (Figure 1x,y). The generated P132-16A-SER1 lines, which maintained pollen sterility (Figure 1w), showed a higher seed-setting rate than P132-16A in outcrossing rates analysis (Figure 1z). These results indicate the potential for effective application of SER1 in rice breeding programs aimed at improving hybrid seed production.

In this study, we identified SER1 (synonymous with OsSPL5) as a pivotal regulator of stigma exertion rate and stigma size in rice. Previous research has demonstrated that the SPL family influences inflorescence morphology (Wang and Zhang, 2017). The potential interaction between SER1 and DEP1 suggests an intricate regulatory network. The differential expression of genes, such as OsRAC3, OsBMY4 and OsGASR2, further suggests extensive genetic interactions (Table S4). Previous studies have elucidated that stigma exsertion is crucial for hybrid seed production and significantly impacts rice domestication. The transition from the high SER and outcrossing behaviour of wild rice to the low SER and predominantly self-pollinating behaviour in cultivated rice has been reported (Zhu et al., 2023). Haplotype analysis revealed higher nucleotide diversity for SER1 in wild rice, suggesting additional functional roles and highlighting the evolutionary significance of SER in rice domestication.

水稻 F₁ 杂交种子生产的产量受亲本品系性状的影响，包括柱头脱落率 （SER），这会影响种子的价格和利用（Marathi 和 Jena，2015）。SER 极易受到环境波动、表型和复杂遗传因素的影响（Miyata et al.， 2007）。尽管已经鉴定出 40 多个与 SER 相关的 QTL，但由于遗传背景的差异和较小的加性效应，没有一个被分子表征（Zhu等人 ，2023 年）。

我们之前已经证明了柱头大小与 SER 之间存在正相关（Tan et al.， 2023）。然后我们之前基于 SSSL-42 将柱头大小基因 SER1 定位在 2 号染色体上 470 kb 的间隔内，SSSL-42 是以华晶贤 74 （HJX74） 为受体亲本的单段替换系 （Tan et al.， 2021）。在这项研究中，源自 SSSL-42 和 HJX74 杂交的纯合重组系允许将 SER1 的区域缩小到 29.48 kb 的延伸，两侧是标记 QY18 和 LST2（图 1a，表 S1）。具有最短替换片段的品系 R4 被确定为 SER1 （NIL-SER1） 的近同基因品系，而 HJX74 被称为 NIL-ser1 （图 1a）。NILs 在许多农艺性状上彼此没有差异，但在 NILs 之间检测到 SER 和柱头大小的显着差异（图 1b-e 和 S1）。

检测到三个候选基因 （OsSPL5 、 OsCH240 和 OsSm-F ） 与映射区间相关。变异分析显示，OsSPL5 在定位区间的第三个外显子中含有两个核苷酸多态性，导致氨基酸取代（图 S2）。柱头的转录组测定显示 NIL 之间这三个基因之间的基因表达没有显着差异（图 S3）。为了研究 SER1 的候选基因，我们获得了三个候选基因的过表达系和敲除系。无论是 NIL-SER1 背景中的基因编辑系还是 NIL-ser1 背景中的过表达系，OsCH240 和 OsSm-F 的转基因系在柱头大小上都没有表现出表型变化（图 S4）。然而，OsSPL5 的过表达导致柱头扩大，而 OsSPL5 的敲除导致较小的柱头（图 1g-j）。此外，OsSPL5 的不同转基因品系的柱头消耗率也发生了相应的变化 （图 1e，g，h）。因此，SER1 的候选基因是 OsSPL5。进一步的 RT-qPCR 测定检测到不同组织中转录水平没有变化（图 S5）。此外，基因编辑系 KO-SER1-3rd 外显子的柱头大小显着减小 （图 S6）。它强烈表明，位于 OsSPL5 第三个外显子中的序列变异是 NILs 之间表型差异的主要原因。

扫描电子显微镜 （SEM） 表明 NIL-SER1 中柱头刷 （SB） 和无刷部分 （SNB） 的表皮细胞长度显着增加（图 1f 和 S7）。解剖观察显示，随着小穗的发育，NIL 之间的柱头大小差异逐渐增加（图 S8）。这些结果表明，OsSPL5 通过影响细胞大小来调节水稻中的柱头大小，从而影响 SER。OsSPL5 是 SPL 家族的重要成员，亚细胞定位表明它主要位于细胞核中（图 S9）。转录激活测定阐明了 OsSPL5 的激活结构域位于 N 末端区域（图 S10）。转录组分析在 NIL-SER1 与 NIL-ser1 和 NIL-ser1 与 KO-SER1 中分别发现了 3331 和 759 个显著不同的表达基因（表 S2）。单个转录因子根据下游靶基因在不同等位基因背景中的功能强度对下游靶基因进行差异调节，而基因敲除通常会导致功能丧失。共有 379 个基因在 NIL-SER1 中表现出与 NIL-ser1 相比增加的表达，而在 KO-SER1 中表现出与 NIL-SER1 相比降低的表达（图 1k 和表 S2）。

使用 GFP-SER1 融合转化原生质体的 Cut&Tag-Seq 分析揭示了与 1852 个基因相关的 2153 个启动子-（-1000 至 0）相关峰（图 1l;表 S3）。随后的关联分析确定了 SER1 的 27 个候选下游靶基因，这些基因在 NIL-SER1 中上调，在 KO-SER1 中下调（图 1k;表 S4）。值得注意的是，DEP1 是一种编码异源三聚体 G 蛋白 γ 亚基的基因，被称为水稻小穗和花发育的关键因素（Huang et al.， 2009， 2022），被确定为关键候选基因之一。在其启动子中观察到 SBP 结合基序 （GTAC） 的富集，与 Cut&Tag 峰顶重叠（图 1k，l）。Alphafold3 对接表明 DEP1 启动子基序与 SER1 之间存在很强的结合相互作用（图 S11）。来自 Cut&Tag-Seq 的数据与 DAP-Seq 的进一步检测相结合，表明 SER1 在 DEP1 启动子中的结合窗口为 ~230 bp（图 S12）。进一步的启动子-LUC 测定证实 SER1 与 DEP1 启动子的结合增强了下游基因表达（图 1m）。Y1H 测定加强了这一发现，表明 SER1 和 DEP1 之间存在正向调节关系（图 1n）。对 DEP1 等位基因与柱头大小之间关系的调查阐明，与 NIL-ser1 （NIL-DEP1-ser1） 相比，NIL-dep1-ser1 显示柱头宽度升高（图 1o，p）。此外，源自 NIL-SER1 的 DEP1 敲除细胞系显示柱头大小显着减小（图 1q，r）。 此外，SER1 的过表达导致穗短，而 KO-SER1 表现出细长的穗（图 S13 和 S14），与已知的 DEP1 功能一致。这些发现表明 SER1 对 DEP1 发挥正向调节作用，从而通过 G 蛋白信号通路调节水稻的柱头和穗发育。

为了探索 SER1 基因的自然变异，共分析了 2042 个 Oryza 种质，显示出广泛的遗传多样性（Yao等 人，2019 年）。OsSPL5 的 SNP 分析表明，SER1 和 ser1 是两种主要的单倍型。这两种单倍型在野生稻中广泛存在，但在栽培稻中，几乎所有种类的籼稻亚种都携带 ser1，而粳稻亚种主要携带 SER1。这表明籼稻和粳稻在该基因座上几乎完全分化（图 1;表 S5）。SER1 基因的核苷酸多样性 （π） 在两个栽培稻亚种中极低（图 1t）。这表明 SER1 在籼稻育种计划中具有巨大潜力。下游 DEP1 基因也显示出强烈的亚种间分化，表明驯化过程中存在共选择（图 S15;表 S5）。我们进一步将 SER1 基因引入 P132-16A 和 P132-16B 品系，这是籼稻雄性不育品系及其相应的恢复系，来源于 HJX74，具有 ser1 遗传背景（图 1u-w）。与 HJX74 和 P132-16B 相比，恢复系 P132-16B-SER1 表现出更高的遗传性 SER（图 1x，y）。生成的 P132-16A-SER1 品系保持花粉不育（图 1w），在异交率分析中显示出比 P132-16A 更高的坐籽率（图 1z）。 这些结果表明 SER1 在旨在提高杂交种子生产的水稻育种计划中有效应用的潜力。

在这项研究中，我们确定 SER1 （与 OsSPL5 同义） 是水稻柱头消耗率和柱头大小的关键调节因子。以前的研究表明，SPL 家族影响花序形态（Wang 和 Zhang，2017）。SER1 和 DEP1 之间的潜在相互作用表明存在错综复杂的监管网络。OsRAC3、OsBMY4 和 OsGASR2 等基因的差异表达进一步表明广泛的遗传相互作用（表 S4）。以前的研究已经阐明，柱头脱落对于杂交种子生产至关重要，并显着影响水稻驯化。据报道，野生稻的高 SER 和异交行为转变为栽培稻中的低 SER 和主要的自花授粉行为（Zhu et al.， 2023）。单倍型分析显示野生稻中 SER1 的核苷酸多样性较高，表明了额外的功能作用，并突出了 SER 在水稻驯化中的进化意义。