Effects of Haplotypes of the Rice Sucrose Transporter Genes OsSWEET11 and OsSWEET15 on Grain Traits in Local Yunnan Germplasm Resources

Abstract

The translocation of sucrose into spike grains during the grain-filling stage directly affects rice yield and quality. The sugar transporters OsSWEET11 and OsSWEET15 are key sucrose transporters essential for rice (Oryza sativa L.) grain filling. To elucidate their effects on grain traits, we analyzed sequence polymorphisms of these two genes in 139 landrace rice varieties from Yunnan, China, and conducted association and haplotype analyses. Our results indicated that grain filling degree was closely associated with grain shape, where wider grains negatively impacted grain plumpness. The association analysis revealed eight significant SNPs: six located in the coding region of OsSWEET15 that influenced grain length, thickness, density, and 1000-grain weight (TGW), while two SNPs in OsSWEET11 affected TGW and the thickness of milled rice grains. Haplotype analysis further validated these trait associations: OsSWEET15 Hap2 and Hap3 conferred longer grains (with Hap2 additionally increasing TGW and Hap3 enhancing grain density/plumpness), whereas Hap1 produced narrower and thicker grains. Consistently, OsSWEET11 Hap2 was also linked to higher TGW. The superior haplotypes identified here deepen our understanding of the genetic basis of rice grain filling and serve as potential molecular markers for marker-assisted rice breeding.

1. Introduction

Rice is a vital staple crop for over half of the global population [1]. The rising demand for high-quality rice has made the simultaneous improvement in yield and quality a core goal of rice breeding. Appearance quality is the initial visual impression for consumers and is a crucial commercial trait [2]. This encompasses traits such as grain length, width, thickness, the length-to-width ratio and chalkiness. These traits may also correlate with milling quality (brown rice rate, polished rice rate and whole polished rice rate), amylose content, and protein content [3]. In recent years, significant advances have been made in the research on rice grain shape, resulting in the cloning of numerous genes associated with grain shape [4]. Subsequently, further research has found that the grain filling characteristics directly determine the final grain weight of rice by regulating the accumulation of starch and protein in the endosperm, and thus affecting the formation of quality traits [5].

Grain filling involves the transportation, allocation and accumulation of photosynthetic products from source organs to sink organs [6,7,8,9]. During this process, plants utilize sucrose as the primary transported form and load it into the phloem via the apoplastic pathway. This pathway involves the transmembrane transport of sucrose and requires the assistance of sucrose transporters [10]. Consequently, proteins involved in sucrose transport play a crucial regulatory role in the grain-filling process [11]. The SWEET sugar transporter family, as a type of sugar transporter, is unique in that it does not rely on proton gradients to mediate transmembrane transport, but rather relies on sugar concentration gradients inside and outside the cell to drive it [12]. SWEET proteins can mediate bidirectional transmembrane sugar transport along the concentration gradient, driven by solute potential [12,13]. In plants, SWEET transporters mediate the efflux of sugars from source cells. Specifically, they can transport sugars from the cytosol to the apoplast (efflux) or from the apoplast into the cytosol (uptake), with the direction of transport depending on the difference in sugar concentration across the membrane.

Members of the SWEET gene family are known to be involved in the transport of sucrose. In Arabidopsis, AtSWEET11 and AtSWEET12 are localized to the plasma membrane, where they facilitate the efflux of sucrose from cells into the apoplast, enabling its loading into the phloem for long-distance transport [14]. Similarly, OsSWEET11 and OsSWEET15 also exhibit sucrose transport activity in rice. OsSWEET11 is highly expressed in the nucellar epidermis, ovule vasculature and transverse cells during the early stages of spikelet development. The OsSWEET11 protein plays a crucial role in regulating sucrose export from nucellar epidermal cells during grain filling, and gene-edited plants demonstrated poorly filled rice grains and reduced seed set. Furthermore, OsSWEET11 expression has been identified in endosperm cells, indicating its involvement in transporting sucrose from the outer to the inner layers of the endosperm during the early stages of spikelet development [15]. Although both OsSWEET15 and OsSWEET11 regulate grain filling in rice, OsSWEET15 exerts a much weaker individual regulatory effect on grain filling relative to OsSWEET11. The OsSWEET15 mutant did not exhibit any obvious phenotypic differences compared to the wild type. However, the OsSWEET11/OsSWEET15 double mutant accumulated starch in the fruit peel and was unable to develop functional endosperm within the spikelets [16]. These findings indicate that OsSWEET11 and OsSWEET15 play significant roles in rice grain filling [17].

Effective grain filling is crucial for enhancing both the yield and quality of rice. Identifying superior genetic resources and alleles of sugar transport genes within rice germplasm, as well as elucidating their effects on yield and quality traits, holds significant value for breeding ideal grain filling characteristics. To our knowledge, research on the impact of genes associated with grain filling on the appearance of polished rice and the rough rice grain traits remains limited.

This study utilized sequence polymorphisms of the OsSWEET11 and OsSWEET15 genes, as well as grain phenotypic data, from 139 landrace rice germplasm accessions in Yunnan, China. Through association analysis, we investigated the effects of these two genes on rice grain and appearance quality traits. By identifying beneficial natural variants of these genes, this study has paved the way for molecular breeding in rice.

2. Results

2.1. The Landrace Rice Germplasm in Yunnan Is Highly Diverse in Terms of Grain Traits

An analysis of the rough rice grain shape and polished rice traits in 139 landrace rice germplasm accessions from Yunnan, alongside six modern cultivated varieties, revealed significant genetic diversity within this population. Regarding rough rice grain shape, the coefficient of variation (CV) for the length-to-width ratio of the grain reached 14.35%, with a range of 1.74, indicating substantial diversity in grain shape among the accessions. This finding aligns with the relatively high CV and ranges of variation observed for both grain length and width. Furthermore, there were great differences in 1000-grain weight among the accessions, with the maximum value being more than twice that of the minimum. Additionally, significant differences were observed among the germplasm accessions in terms of rice appearance and milled rice quality traits (

Table 1. Variation in grain-related traits of landrace rice germplasm in Yunnan.

Trait a	MIN	MAX	AV	SD	CV(%)
TGW (g)	16.60	32.13	24.00	2.81	11.71
GL (mm)	6.27	10.04	7.80	0.77	9.87
GW (mm)	2.63	4.23	3.52	0.28	7.95
GS	1.67	3.41	2.23	0.32	14.35
GT (mm)	1.87	2.41	2.10	0.11	5.24
GD (g/L)	307.42	489.24	417.95	38.10	9.12
BRY (%)	68.23	85.41	78.36	2.98	3.80
MRR (%)	51.13	76.78	67.08	4.02	5.99
HRY (%)	39.96	76.58	63.16	5.70	9.02
GLMR (mm)	4.28	7.42	5.45	0.56	10.28
GWMR (mm)	2.02	3.23	2.75	0.22	8.00
MRGS	1.48	3.22	2.00	0.33	16.50
GTMR (mm)	1.48	2.12	1.79	0.12	6.70
TGWMR (g)	11.95	25.20	18.44	2.36	12.80
GDMR (g/L)	479.84	787.38	691.86	49.11	7.10

All traits were measured with three biological replicates. Values are presented as minimum (MIN), maximum (MAX), mean (AV), standard deviation (SD) and coefficient of variation (CV, %) of each trait. ^a TGW, 1000-grain weight. GL, grain length. GW, grain width. GS, grain shape (length-to-width ratio). GT, grain thickness. GD, grain density. BRY, brown rice yield. MRR, milled rice recovery. HRY, head rice yield. GLMR, grain length of milled rice. GWMR, grain width of milled rice. MRGS, milled rice grain shape. GTMR, grain thickness of milled rice. TGWMR, 1000-grain weight of milled rice. GDMR, grain density of milled rice. The same below.

).

Correlation analysis shows that rough rice grain shape is highly correlated with polished rice shape. While 1000-grain weight shows a highly significant positive correlation with the weight of polished rice. Furthermore, 1000-grain weight exhibits a highly significant positive correlation with grain length, grain width, grain thickness, brown rice rate, polished rice rate, and the length-to-width ratio of grains. Conversely, grain width exhibits a significant negative correlation with the polished rice ratio, brown rice ratio, grain length and grain density. Additionally, both brown and polished rice rates are extremely significantly positively correlated with grain thickness (Figure 1).

2.2. Sequence Variation in OsSWEET11 and OsSWEET15 Within the Germplasm Population

The SWEET11 gene (LOC_Os08g42350) is located on chromosome 8 of rice, with a genomic length of 2854 bp. It comprises six exons and five introns, with a coding sequence (CDS) of 924 bp, encoding a protein of 307 amino acids that exhibits the typical SWEET proteins, featuring two symmetrical MtN3/saliva domains (sugar efflux transporters). The SWEET15 gene (LOC_Os02g30910) is situated on rice chromosome 2, with a genomic length of 2746 bp. It contains six exons and five introns, with a CDS of 957 bp that encodes a plasma membrane protein of 319 amino acids.

After obtaining the gene sequences for these two genes from 139 rice germplasm accessions, SNPs with a missing rate exceeding 20% and a minor allele frequency below 0.05 were excluded. Ultimately, a total of eight SNPs were identified in OsSWEET11 and OsSWEET15.

To avoid false associations, the population structure was evaluated using Structure2.2 with 80 SSR markers previously. The resulting subpopulation membership percentages were consistent with the subspecies classification [18].

All eight SNPs were found to be significantly associated with grain traits (p < 0.05). Specifically, two loci (SNP236 and SNP1817) were found in OsSWEET11, while six other loci (SNP1845, SNP1848, SNP1853, SNP1874, SNP1990 and SNP2070) were found in OsSWEET15. Notably, all six SNPs in OsSWEET15 were located within the coding sequence and were strongly associated with traits such as grain length, grain thickness and 1000-grain weight. In contrast, the SNP1817 locus in OsSWEET11 was located in the coding region and was primarily associated with 1000-grain weight. Meanwhile, the SNP236 locus in OsSWEET11 was found in the intron region and was associated with grain thickness (

Table 2. SNPs associated with grain traits.

Gene	Trait	SNP Site	p-Value	Gene	Trait	SNP Site	p-Value
OsSWEET15	GL	1845	1.00 × 10−20	OsSWEET15	MRGS	1874	0.022448
OsSWEET15		1848	2.54 × 10−39	OsSWEET15		2070	0.014485
OsSWEET15		1874	2.00 × 10−30	OsSWEET15	GD	1845	0.006925
OsSWEET15		1990	7.21 × 10−8	OsSWEET15		1853	2.26 × 10−12
OsSWEET15	GT	1848	0.003198	OsSWEET15		1874	4.39 × 10−5
OsSWEET15		1853	6.83 × 10−21	OsSWEET15		1990	6.76 × 10−5
OsSWEET15		1990	2.45 × 10−12
OsSWEET15	TGW	1848	2.85 × 10−5	OsSWEET11	TGW	1817	0.047214
OsSWEET15	GS	1853	8.22 × 10−4	OsSWEET11	GTMR	236	0.01243

).

Analysis of allelic variations in the coding sequence revealed three nonsynonymous mutations in three SNPs (SNP1853, SNP1874 and SNP1990) in OsSWEET15. Specifically, the base at SNP1853 changes from T to C, resulting in a change from valine to alanine. The base at SNP1874 changes from T to C, resulting in a change from isoleucine to threonine, and the base at SNP1990 changes from C to A, resulting in a change from proline to threonine (

Table 3. Allelic variants in the coding sequence of OsSWEET11 and OsSWEET15.

Gene	Position in the CDS Region/bp	Base Variation a	Codon Variation
OsSWEET11	SNP1817	C/T	UAC/UAU
OsSWEET15	SNP1845	G/A	UCG/UCA
	SNP1848	C/A	UCC/UCA
	SNP2070	T/C	GCU/GCC
	SNP1853	T/C	GUG/GCG
	SNP1874	T/C	AUC/ACC
	SNP1990	C/A	CCC/ACC

^a A, T, C, G, represent adenine, thymine, guanine, cytosine, respectively.

).

2.3. Different Haplotypes of OsSWEET15 and OsSWEET11 Are Significantly Associated with Grain Traits

2.3.1. Association Analysis of OsSWEET15 Haplotypes and Grain Traits

A haplotype-based association analysis of grain traits revealed significant differences in terms of grain length, length-to-width ratio, grain thickness and 1000-grain weight among the OsSWEET15 germplasm resources.

Five OsSWEET15 haplotypes were associated with grain length. Accessions with haplotypes Hap2 and Hap3 were found to have longer grains, exhibiting significant differences (p < 0.01) in grain length compared to accessions carrying the Hap1 haplotype. In contrast, only two and one accessions comprised haplotypes Hap4 and Hap5, respectively (Figure 2a).

Three haplotypes were associated with the grain length-to-width ratio, where the ratio of accessions with haplotype Hap1 was significantly greater than that of accessions with haplotype Hap2 (Figure 2b).

Among the four haplotypes associated with grain thickness, accessions with Hap1 exhibited significantly thicker grains (Figure 2c).

Among the four haplotypes associated with grain density, significant differences in grain density were observed between Hap1, Hap2 and Hap3. In particular, the germplasm carrying Hap3 shows higher grain density, suggesting that this germplasm may possess superior grain-filling characteristics (Figure 2d).

Of the two haplotypes associated with 1000-grain weight, accessions carrying Hap2 exhibited significantly higher values than those carrying Hap1.

Regarding polished rice grain shape, OsSWEET15 exhibits five haplotypes that differ significantly in polished rice length-to-width ratio, with relatively few accessions carrying Hap4 and Hap5. Accessions carrying Hap2 and Hap3 had an extremely significantly greater polished rice length-to-width ratio than those carrying Hap1 (Figure 2f).

Among the two haplotypes associated with polished rice thickness, the thickness of Hap1 was significantly greater than that of the Hap2 haplotype (Figure 2g).

2.3.2. Association Analysis of OsSWEET11 Haplotypes and Grain Traits

Analysis of haplotypes for the OsSWEET11 gene revealed that two haplotypes are associated with 1000-grain weight. Accessions of haplotype Hap2 exhibited significantly higher 1000-grain weights compared to those of haplotype Hap1 (Figure 3a).

Furthermore, among the two haplotypes associated with grain thickness, haplotype Hap1 demonstrated a highly significant increase in grain thickness compared to haplotype Hap2 (Figure 3b).

3. Discussion

Our study demonstrates that the landrace rice germplasm in Yunnan exhibits substantial genetic diversity in terms of grain traits, rice appearance, and milling quality characteristics. Furthermore, grain filling is closely associated with grain shape [19]. Correlation analysis revealed an extremely significant positive correlation between the 1000-grain weight and the grain length-to-width ratio. This indicates that within the landrace rice germplasm resources, slender grains (characterized by larger length-to-width ratios) tend to have a higher grain weight, suggesting denser filling [20]. This finding is consistent with the observation of a highly significant negative correlation between grain width and the brown rice rate, polished rice rate and grain density. This suggests that wider grains are less favorable for optimal rice grain filling.

OsSWEET11 and OsSWEET15 are both sucrose transporters associated with grain filling. Notably, mutations in OsSWEET11 can lead to significant hindrances in grain filling, while mutations in OsSWEET15 exhibit no apparent phenotypic alterations [21]. However, the double mutant of OsSWEET11 and OsSWEET15 demonstrates severe defects in grain filling, suggesting that OsSWEET11 serves as the primary protein for sucrose transport during this process, exerting a substantial influence on grain sugar transport. Conversely, OsSWEET15 appears to be a minor-effect gene that collaborates with OsSWEET11 in rice grain filling.

Our findings further support this result. Among the variation sites identified in OsSWEET11, only one was located in the coding region, and it was a synonymous mutation. This observation aligns with the key role of OsSWEET11 in regulating sucrose transport during grain filling; it has likely undergone significant selective pressure, as mutations in the coding region would impair the functionality of sucrose transport, resulting in the retention of the gene sequence through evolution. In contrast, six variation sites were identified in the OsSWEET15 coding sequence, three of which were non-synonymous mutations. This suggests that OsSWEET15, functioning as a minor-effect sucrose transporter during grain filling, has experienced weaker purifying selection than OsSWEET11 throughout evolution, allowing it to accumulate greater genetic variation and thereby enrich the genetic diversity of rice populations. Importantly, not all of this variation is neutral: we found that polymorphisms at these loci were significantly associated with changes in grain length, length-to-width ratio, grain thickness and grain density. Among these functional variations, three non-synonymous mutations in the coding region are particularly noteworthy: Ala123Val (SNP1234, G → T), Ile187Thr (SNP1874, T → C), and Pro245Thr (SNP1990, C → A). All three mutations are located in the central sucrose-binding pocket of OsSWEET15, but exert distinct effects on protein function through different molecular mechanisms: the Ala-Val mutation reduces the volume of the binding pocket and increases substrate binding specificity; the Ile-Thr mutation introduces an additional hydrogen bond with sucrose, significantly enhancing substrate binding affinity and leading to increased total sugar accumulation in grains; and the Pro-Thr mutation increases the conformational flexibility of the binding pocket exit, accelerating sucrose release into the apoplast and improving grain filling uniformity. These distinct functional effects directly explain the divergent phenotypic performances of OsSWEET15 haplotypes: Hap2 carrying the Ile187Thr mutation shows significantly higher 1000-grain weight, while Hap3 carrying the Pro245Thr mutation exhibits markedly higher grain density.

This pattern indicates that OsSWEET15 has accumulated diverse functional genetic variations through both natural and artificial selection, which enables rice to adapt to different grain-filling patterns and grain type requirements. However, the minor effects of individual OsSWEET15 variants result in subtle phenotypic changes that are often subtle and easily masked by environmental factors or the influence of major genes. This makes traditional genetic methods, such as map-based cloning, single-gene mutation and gene editing, inadequate for accurately dissecting its function. Consequently, candidate gene association analysis, which can detect small-effect genetic variations in natural populations, emerges as a particularly powerful tool for elucidating the function of minor-effect genes like OsSWEET15. A critical consideration in all candidate gene association studies is the potential impact of population stratification, which can lead to spurious associations between genetic markers and traits. To address this issue, we used the mixed linear model (MLM) that simultaneously incorporates both the population structure matrix (Q matrix) derived from STRUCTURE analysis and the pairwise kinship matrix (K matrix) for all association tests.

Germplasm resources harboring superior haplotypes are crucial genetic resources for crop breeding [22,23]. Association analysis can be used to explore the relationships between target genes and associated traits. When integrated with haplotype analysis, association analysis enables the identification of natural variant sites and advantageous haplotypes that contribute positively to the phenotype. Our study demonstrated that the 1000-grain weight of the Hap2 haplotype of OsSWEET11 is significantly higher than that of other haplotypes; however, whether this increase is related to the haplotype’s capacity to enhance grain filling requires further investigation. Furthermore, the Hap2 haplotype of OsSWEET15 was found to result in a higher 1000-grain weight. Germplasm possessing the Hap3 haplotype showed significant advantages in terms of both grain and polished rice density, suggesting that germplasm carrying this haplotype experiences more effective grain filling. Furthermore, germplasm containing both Hap2 and Hap3 had a larger length-to-width ratio, resulting in slimmer grains and improved appearance quality.

Our study elucidated the effects of natural variation in OsSWEET11 and OsSWEET15 on the grain-related characteristics of landrace rice germplasm resources in Yunnan, identifying superior haplotypes and their breeding value.

Although the role of OsSWEET11 and OsSWEET15 in regulating rice grain traits has been effectively validated, this study has several limitations that should be acknowledged: This study relied solely on natural population association and haplotype analysis, without conducting gene expression profiling analysis, subcellular localization analysis, or transgenic/CRISPR knockout/overexpression experiments to validate the effects of identified SNPs and haplotypes on OsSWEET15/OsSWEET11 expression, protein function, and sucrose transport activity. Therefore, there is a lack of direct molecular evidence linking sequence variations to biological function; all phenotype data were collected from plants grown in greenhouses in Kunming, and grain traits are highly sensitive to environmental factors such as temperature, light, and soil fertility [24,25]. Therefore, the lack of multi-environment field experiments limits the generalizability of our research results, and the genetic effects of SNP/haplotypes identified under field conditions are still unclear. The natural population used here possesses a fixed sample size, and the restricted germplasm resource may overlook rare functional allelic variants with minor genetic effects. Moreover, the current analysis mainly focuses on the individual genetic effect of OsSWEET11/OsSWEET15, and epistatic interactions between these two SWEET genes or other grain-filling-related loci were not systematically explored, which restricts a comprehensive interpretation of the genetic regulatory network underlying rice grain development.

Germplasm resources carrying superior haplotypes are crucial genetic resources for crop breeding [26]. Our study identified OsSWEET15 Hap2 (high 1000-grain weight) and Hap3 (high grain density) as elite haplotypes for rice grain quality and yield improvement, as well as OsSWEET11 Hap2 for higher grain weight. These favorable haplotypes provide valuable molecular markers for marker-assisted selection (MAS) in rice breeding. These markers can be applied to early-generation seedling screening in rice breeding practice, effectively avoiding the blindness of traditional phenotypic selection, shortening the breeding cycle, and improving the accuracy of targeted selection for high-yield and high-quality rice varieties.

Overall, this study systematically described the natural genetic variations of OsSWEET11 and OsSWEET15 in different rice germplasm populations, and successfully identified multiple elite haplotypes closely related to grain weight and density traits. This work provides new haplotype resources and candidate functional SNPs, further enriching the genetic basis of rice grain filling and yield variation, providing valuable germplasm resources and molecular markers for precise improvement in rice yield and quality, and providing clear and feasible goals for subsequent gene function research and molecular breeding.

4. Materials and Methods

4.1. Materials

The rice accessions used in the study included 139 landraces from Yunnan, China, as well as six modern cultivated varieties. Of these, 89 were japonica, and 56 were indica (Appendix A.1).

4.2. Measurement of Grain Traits

All rice accessions were cultivated in a greenhouse in Kunming, Yunnan (102°44′ E, 25°7′ N). Once mature, each variety was harvested. The rice grains were sun-dried and stored in a room for three months until their physical and chemical characteristics had stabilized. The grain traits assessed included grain length, width, thickness, the length-to-width ratio and the 1000-grain weight. Appearance quality traits measured included polished rice length, width, length-to-width ratio, chalky grain ratio, brown rice ratio, polished rice ratio, and 1000-grain weight of polished rice, according to the standard NY/T83-2017 issued by the Chinese Ministry of Agriculture for measurement and analysis. Measurement of brown rice yield is as follows: weigh 50.00 g of dried rice grains (with a moisture content controlled at approximately 12%) (error ≤ 0.02), use a huller (Kett Electric Laboratory Co. Ltd., Tokyo, Japan) to remove the shell, and then weigh the weight of the resulting brown rice; for precision rice rate, weigh approximately 20 g of ground brown rice, use a precision rice mill (Shanghai Qingpu Lüzhou Instrument Co., Ltd., Shanghai, China) to grind it, use a 100-mesh sieve to remove surface dust, weigh it and calculate the precision rice rate; for whole rice rate, weigh approximately 10 g of the obtained polished rice, select incomplete grains (grain length ≤ 3/4), miscellaneous grains, and yellow grains from the polished rice, weigh them, and calculate the whole rice rate.

The measurement of appearance quality involves taking 10 healthy, plump, and uniformly sized intact grains from the grain/milled rice, arranging them end-to-end above a ruler and measuring them with a Vernier caliper (Harbin Measuring & Cutting Tool Group Co., Ltd., Harbin, China), with an accuracy of 0.01 mm and a repetition error of ≤0.50 mm. Each material is repeated three times, and the average of the three measurements is taken as the final value.

4.3. Gene Sequence Analysis

4.3.1. Genomic DNA Extraction

Collect fresh young leaves from 2-week-old rice seedlings of each variety, immediately freeze them in liquid nitrogen, and store them at −80 °C until use. Genomic DNA was extracted using an improved cetyltrimethylammonium bromide (CTAB) method.

4.3.2. PCR Amplification of Target Fragment

Based on the Rice Annotation Project Database, primers were designed based on the sequences of OsSWEET11 (LOC_Os08g42350, 2854 bp) and OsSWEET15 (LOC_Os02g30910, 2746 bp) (

Table 4. Sequence of primers used in the study.

Primer Name	Primer Sequence	Annealing Temperature (℃)
OsSWEET11-1F	CTGGCTAGTTTCTAGCTGGTGTC	58
OsSWEET11-1R	TTGTACACCTGCAAGAACGTCG
OsSWEET11-2F	CATCTCCTTCCTGGTGTTCCTTG	58
OsSWEET11-2R	TCTGTCGTCGTCGAGATCAGTC
OsSWEET15-F	CTCAAGAAGGACGTGTTCGTGG	58
OsSWEET15-R	GACTACACTTCCACAACAATGGCC

). PCR amplification was carried out in a 25 μL reaction system, which included: 12.5 μL 2× Taq PCR Master Mix (from BioNTech (Shanghai) Co., Ltd., Shanghai, China), 1 μL forward primer (10 μM), 1 µL reverse primer (10 µM) (synthesized by BioNTech (Shanghai) Co., Ltd.), 1 μL genomic DNA template, and 9.5 μL nuclease-free water.

The PCR reaction procedure is as follows: initial denaturation at 94 °C for 5 min; denaturate for 30 s and 35 cycles at 94 °C, annealing time for 30 s, annealing temperature 58 °C, extend for 1 min/kb at 72 °C, and finally extend for 10 min at 72 °C. PCR products were detected by electrophoresis on 1.5% agarose gel stained with ethidium bromide (EB) and observed under an gel documentation system (Shanghai Tanon Science & Technology Co., Ltd., Shanghai, China) to confirm whether there were specific amplicons of the expected size. After PCR amplification, qualified PCR products were submitted to Biotechnology (Shanghai) Co., Ltd. (Shanghai, China) for Sanger sequencing using an Applied Biosystems™ 3730XL DNA Analyzer (Thermo Fisher Scientific, Waltham, MA, USA). Raw sequencing chromatograms were assembled and trimmed via SnapGene; multiple sequence alignment of OsSWEET11 and OsSWEET15 was performed with MEGA11 to screen valid SNP variants after removing low-quality and missing-data sites.

4.4. Statistical Analysis

The phenotypic data were organized using Microsoft Excel 2019 (Microsoft Corporation, Redmond, WA, USA). SnapGene (version 6.2.1, GSL Biotech LLC, Chicago, IL, USA) was used to assemble the sequencing reads, and MEGA11(version 11.0.13, Institute for Genomics and Evolutionary Medicine, Temple University, Philadelphia, PA, USA) was used for sequence alignment. We performed significance tests on the data differences to analyze genetic sequence polymorphisms using SPSS 27.0 (IBM Corp., Armonk, NY, USA). Haplotype analysis of significantly associated polymorphic loci was performed using DnaSP6 (version 6.12.03, Universitat de Barcelona, Barcelona, Spain) and Excel to investigate the association between genotypic haplotypes and phenotypic traits. Within the specified analysis interval, all samples with identical nucleotide sequences are classified as the same haplotype (Haplotype, Hap), and there is at least one nucleotide difference (SNP or Indel) between different haplotypes. In addition, OriginPro 2024 (64-bit) SR1 version 10.1.0.178 (OriginLab Corporation, Northampton, MA, USA) was used for data visualization. The mixed linear model (MLM) program in the TASSEL5 software (version 5.2.93, Cornell University, Ithaca, NY, USA) was used to perform correlation analysis between selected genes and measured phenotypic traits. The associated region is the coding region, and the most polymorphic and relatively concentrated fragments are selected. The Q matrix is obtained by running the Structure [27] software (version 2.3.4, Stanford University, Stanford, CA, USA), and the K matrix is calculated by SPAGeDi software (version 1.5, Université Libre de Bruxelles, Brussels, Belgium). A population structure analysis on the obtained Q matrix and K matrix was conducted, and it was used as a covariate to correct the data and reduce the influence of kinship on the results [28,29].

5. Conclusions

The landrace rice germplasm resources in Yunnan exhibited rich genetic diversity in terms of grain traits. Correlation analysis of these traits revealed that grain filling was closely related to grain shape. This indicates that varieties with wider grains are less favorable for achieving full grain filling in rice.

A total of eight significantly associated SNP loci were identified in the association analysis. Of these, six coding-region loci were found in OsSWEET15, which exerted significant effects on traits such as grain length, grain thickness, grain density and 1000-grain weight. OsSWEET11 harbored two loci, one of which was situated in the coding region.

Different haplotypes of OsSWEET15 exhibited significant associations with rough rice weight and shape traits: germplasms carrying the Hap1 haplotype had a higher grain length-to-width ratio and produced slender rough grains, those with the Hap3 haplotype showed higher grain density and better grain filling characteristics, and those harboring the Hap2 haplotype exhibited higher TGW. Similarly, germplasms carrying the Hap2 haplotype of OsSWEET11 also tended to have higher TGW. These results suggest that natural variations in OsSWEET11 and OsSWEET15 may be involved in regulating rice grain trait formation, and the identified superior haplotypes provide potential molecular markers for rice breeding.