Fine Mapping of qTGW7b, a Minor Effect QTL for Grain Weight in Rice (Oryza sativa L.)

Grain weight is a key trait that determines rice quality and yield, and it is primarily controlled by quantitative trait loci (QTL). Recently, attention has been paid to minor QTLs. A minor effect QTL qTGW7 that controls grain weight was previously identified in a set of chromosomal fragment substitution lines (CSSLs) derived from Nipponbare (NPB)/93-11. Compared to NPB, the single segment substitution line (SSSL) N83 carrying the qTGW7 introgression exhibited an increase in grain length and width and a 4.5% increase in grain weight. Meanwhile, N83 was backcrossed to NPB to create a separating population, qTGW7b, a QTL distinct from qTGW7, which was detected between markers G31 and G32. Twelve near-isogenic lines (NILs) from the BC9F3 population and progeny of five NILs from the BC9F3:4 population were genotyped and phenotyped, resulting in the fine mapping of the minor effect QTL qTGW7b to the approximately 86.2-kb region between markers G72 and G32. Further sequence comparisons and expression analysis confirmed that five genes, including Os07g39370, Os07g39430, Os07g39440, Os07g39450, and Os07g39480, were considered as the candidate genes underlying qTGW7b. These results provide a crucial foundation for further cloning of qTGW7b and molecular breeding design in rice.


Introduction
Rice represents the primary food consumed by urban and rural residents in China, accounting for more than 60% [1]. Rice production is a crucial strategy for combating climate change and population growth. Rice yield is primarily determined by the following three factors: panicle number per plant, number of grains per panicle, and grain weight. Rice grain weight is affected by grain length (GL), grain width (GW), grain thickness (GT), grain filling, and other factors [2]. Multiple genes synergistically regulate the quantitative trait of grain weight to maintain the dynamic balance of rice yield. Many quantitative trait loci (QTLs) for rice grain weight have been mapped to chromosomes, as a result of the enrichment of molecular markers by whole genome sequencing and the development of novel mapping methods [3][4][5][6][7].
With the advancement of functional genomics and molecular genetics in rice, many genes/QTLs that control grain size have been cloned and functionally studied, including GS3, encoding the G-protein γ subunit, which reduces grain length by competitively interacting with Gβ [8]. DEP1/qPE9-1 also encodes a γ subunit of a G-protein and GS3 and DEP1 can interact with OsMADS1 and function as cofactors in the regulation of common target genes that control grain size [8,9]. GW6 encodes a GAST (GA-stimulated) family protein that regulates gibberellin response and biosynthesis to affect grain width and weight positively [10]. GL3.1/qGL3, encoding a serine/threonine phosphatase, is a negative regulator of grain length that changes cell numbers by directly phosphorylating the cell cycle protein cyclin-T1;3 [11,12]. TGW6, which encodes an IAA-glucose hydrolase, controls the transition of IAA from the syncytial to the cytosolic stage to negatively regulate seed length [13].
LG1 encodes a ubiquitin-specific protease, OsUBP15. OsDA1 interacts with Os-UBP15 to reduce its stability, negatively regulating cell proliferation and rice seed size [14]. GS2/OsGRF4 encodes a growth-regulating factor that suppresses cell elongation and cytokinesis by interacting with transcriptional co-activators OsGRFs, thereby influencing grain size and weight. In addition, OsmiR396 regulates GS2, and GS2 AA mutations can affect OsmiR396-mediated splicing, resulting in the large-grain phenotype [15,16]. GW5 encodes a calmodulin-binding protein and negatively regulates grain width and grain weight. GW5 inhibits the autophosphorylation of GSK2 and the phosphorylation of OsBZR1 and DLT by GSK2, affecting the accumulation of unphosphorylated OsBZR1 and DLT proteins in the nucleus [17]. TGW3/qTGW3/GL3.3, encoding a GSK3/SHAGGY-like family kinase, OsSK41/OsGSK5, interacts with and phosphorylates OsARF4 to regulate rice grain size and weight via auxin-responsive signaling. Loss of function of qTGW3 results in larger rice grains [18][19][20]. These genes act as negative regulators on rice grain size.
In contrast, some genes control grain length as positive regulators. GS5 encodes a serine carboxypeptidase, a quantitative trait gene that controls grain width, fullness and thousand-grain weight in rice. A high level of GS5 expression results in larger seeds [21]. GW2 on chromosome 2 encodes a cyclic E3 ubiquitin ligase that inhibits cell division by anchoring its substrate to the proteasome for degradation to regulate seed size [22,23]. GL7/GW7, which consists of two tandem repeats, GL7-S1 and GL7-S2, regulates grain length by increasing cell division in the longitudinal direction and controls grain width by decreasing cell division in the transverse direction. GL7 interacts with the cell proliferationpromoting regulator OsSPL16/GW8. In rice, a higher level of OsSPL16 expression promotes cell division and grain filling, increasing grain width and yield [24,25]. qLGY3, located on chromosome 3 and encoding the MADS-domain transcription factor OsMADS1, regulates grain size by interacting with GS3 and DEP1 to enhance OsMADS1 transcriptional activity [9]. WG7, which encodes a CW domain-containing zinc finger protein, directly binds to the promoter of OsMADS1 and promotes OsMADS1 expression by enhancing its H3K4me3 enrichment to positively regulate grain size in rice [26].
Nonetheless, most QTLs that control grain size and weight show minor effects and are difficult to clone. To date, only six QTLs with minor effects have been cloned. Five control rice flowering time and one controls rice grain weight. The additive effects of the five QTL related to heading time (DHD4, qHd1, Hd17, Hd18, and DTH2) ranged from 1 to 4 days [27][28][29][30][31]. In different years, the additive effect of the grain weight QTL qTGW1.2b ranged from 0.13 g to 0.19 g [32], which was significantly less than the additive effect of the cloned major effect QTL on grain weight. Minor effect QTL are sensitive to genetic background and strongly influenced by environmental and measurement biases. Consequently, cloning minor effect QTL has proven to be a formidable challenge.
In our previous studies, one minor effect QTL for grain weight qTGW7 was detected on the long arm of chromosome 7, using a set of CSSLs derived from a cross between indica rice cultivars 93-11 and a japonica cultivar NPB [33]. In this study, we fine mapped qTGW7b, a QTL with a minor effect regulating grain weight, close to qTGW7 in an 86.2-kb region. Our research expands the minor effect QTL for grain size regulation in rice and lays the groundwork for future gene cloning and functional studies.

Grain Size Analysis of NPB and N83
The detection of a minor effect QTL qTGW7 controlling TGW has been reported previously using a set of CSSLs constructed with NPB as the recipient parent and 93-11 as the donor parent [33]. N83, a line containing a 93-11 fragment, has a larger grain size than NPB. The substitution segments of N83 that originate in 93-11 were detected by resequencing, and only a substitution segment from 15.9 M to 23.8 M on the chromosome 7 long arm was identified ( Figure 1). The GL and GW of N83 increased by 4.3% and 5.5% in 2020 and 4.3% and 5.6% in 2021, respectively, compared to the recurrent parent NPB (Figure 2A-C). However, there was no difference in GT between NPB and N83 ( Figure 2D). Furthermore, the dehulled grain of N83 was significantly bigger than that of NPB (Figure 2A), resulting in a 4.4% and 4.6% increase in the TGW of N83 relative to NPB in 2020 and 2021, respectively ( Figure 2E, Supplementary Table S1). We also investigated the mature seed size of 93-11 and NPB. In addition, 9311 had longer GL, narrower GW and GT, and higher TGW than NPB (Supplementary Figure S1, Supplementary Table S2). These results indicate that the differences in grain size and weight between N83 and NPB are stable and that there may be a QTL for N83 that influences grain size and weight.

Cytological Analysis of Spikelet Hull in NPB and N83
The spikelet hull (consisting of palea and lemma) imposes a maximum size restriction on the final grain [22,34]. To investigate the factors responsible for the increase in GL and GW of the substitution line N83, scanning electron microscopy (SEM) was used to evaluate the cell morphology of the lemma by comparing the cell length and width of the outer epidermal cells in spikelet hulls in NPB and N83. As demonstrated in Figure 3, the outer epidermal cells of N83 spikelet hulls were longer and wider than those of NPB spikelet hulls. Counting the number of cells in the spikelet hull in the direction of grain length revealed no difference between N83 and NPB ( Figure 3). These results indicate that the increased cell expansion in spikelet hulls is primarily attributable to the N83 phenotype.

Genetic Analysis of Grain Size-Related Traits of N83
To validate qTGW7 and determine whether this QTL influenced the variation in N83 grain size, we backcrossed N83 with NPB and then self-crossed to create a BC 9 F 2 segregating population. The BC 9 F 2 segregation population contained 289 individual plants for linkage analysis and QTL analysis. One heterozygous plant was selected, covering G6-G32 for self-crossing to generate the BC 9 F 3 population. The BC 9 F 3 population consisted of 297 individual plants. The Anderson-Darling test, D'Agostino and Pearson test, Shapiro-Wilk test, and Kolmogorov-Smirnov test were utilized to analyze the distribution of the grain size-related traits GL, GW, and TGW. Remarkably, phenotypic variation in GL and TGW approximated a normal distribution based on quantile-quantile plot (QQ plot) values (Figures 4 and S4). Transgressive segregation was observed only for GW (Tables 1 and S1).  Combining the genotype and phenotype information of each plant in the BC 9 F 3 population, frequency distributions for each genotype group were plotted as a series, and differences were observed between the GL, GW, and TGW genotype groups. The 93-11 homozygotes were clustered towards the area of higher values, while the NPB homozygotes to the area of lower values ( Figure 5, Table 2), indicating that there are QTL affecting grain size in this population and it is suitable for QTL analysis. In addition, the correlation analysis revealed a relationship between TGW, GL, and GW. All three correlation coefficients were highly significant, including 0.647 between GL and GW, 0.719 between GL and TGW, and 0.709 between GW and TGW (Table 3).

Validation and Fine Mapping of qTGW7b
TGW was selected as the genotype for QTL analysis, since the differences between N83 and NPB in GL and GW were minor, susceptible to environmental influences, and difficult to discern in the field. In contrast, the variation in TGW in the segregated population was as high as 5 g. Therefore, two linkage maps were constructed, combining each plant's genotype and phenotype information in the BC 9 F 2 and BC 9 F 3 populations (Table 4). Significant QTL effects for the TGW were observed in both BC 9 F 2 and BC 9 F 3 populations. For TGW, the enhancing alleles were derived from 93-11 in the two populations. In BC 9 F 2 , the additive effects were 0.69 g for TGW, explaining 19.33% of the phenotypic variances. In BC 9 F 3 , the additive effects were 0.61 g for TGW, explaining 17.17% of the phenotypic variations. The results suggested there might be a QTL in the segregating region. In addition, twelve different types of recombinants were screened from the BC 9 F 3 population based on the positions of recombination breakpoints determined using molecular markers, and individual recombinant BC 9 F 4 plants with homozygous 93-11 genotypes were selected and developed into NILs (R1 to R12) in order to quantify the TGW phenotype. The TGW of plants in groups R1 through R7 was not significantly different from NPB. Similarly, there were no significant differences between plants in groups R8 through R12 and N83. In contrast, the TGW of plants that belonged to groups R1 to R7 and R8 to R12 varied considerably (Table 5). By combining the genotypes and phenotypes of twelve NILs, we were able to map qTGW7 to a 651-kb genomic region flanked by markers G31 and G32 ( Figure 6A). However, this interval differs from the previously reported one, so we designated it as qTGW7b.
Ten BC 9 F 3 heterozygous plants were utilized to establish a self-population for fine mapping in order to narrow down the mapped region further. We identified ten recombinants by screening 3118 BC 9 F 3:4 generation individuals for the G31 and G32 markers ( Figure 6B). Using five additional polymorphism markers, the ten recombinants were separated into five groups based on the recombination breakpoints ( Figure 6B). A test of the progeny of recombinant NPB homozygotes and 93-11 homozygotes from the BC 9 F 3:4 generation revealed a significant difference in TGW between the N2 and N5 groups; however, this was not among the N1, N3, or N4 groups (Table 6). qTGW7b was ultimately narrowed to a~86.2-kb region flanked by markers G72 and G32 ( Figure 6B).

Analysis of qTGW7b Candidate Genes
According to the Rice Genome Annotation Project (http://rice.uga.edu/, accessed on 20 January 2022), the 86.2 kb region contains 14 predicted genes ( Table 7). These genes encode four retrotransposon proteins (Os07g39380, Os07g39390, Os07g39410, and Os07g39420), three transporter family proteins (Os07g39350, Os07g39360, and Os07g39460), two expressed proteins (Os07g39370 and Os07g39450), a polyol transporter protein (Os07g39400), an mTERF family protein (Os07g39430), a zinc finger protein (Os07g39440), a chitin-inducible gibberellin-responsive protein (Os07g39470), and a WRKY transcription factor (Os07g39480). First, the retrotransposon proteins were eliminated as candidates for qTGW7b. Genomic sequencing of the last ten candidate genes in NPB and N83 revealed that amino acid diversity occurred, except for Os07g39350, Os07g39360, Os07g39460, and Os07g39470. Therefore, the remaining six candidate genes (Os07g39370, Os07g39400, Os07g39430, Os07g39440, Os07g39450, and Os07g39480) were prioritized for further investigation. Os07g39370 had a 4-bp insertion, a 17-bp deletion and a single SNP, which resulted in a code shift beginning at the 47th amino acid. Os07g39400 contained a 12-bp insertion and four SNPs, which caused four amino acid insertions and one nonsynonymous mutation. Three SNPs were present on Os07g39430, resulting in one synonymous and two nonsynonymous mutations. Four SNPs were found in both Os07g39440 and Os07g39450, resulting in one synonymous and three nonsynonymous mutations. Os07g39480 contained a 42-bp deletion, a 4-bp deletion and three SNPs, which led to four nonsynonymous mutations and the deletion of amino acids at positions 479 to 492 and 543 (Table 7). Table 7. Candidate genes in the target region of qTGW7b.

LOC_Os07g39350
Transporter family protein None None LOC_Os07g39360 Transporter family protein None None  We further investigated the expression of these six genes in developing young panicles and found that only Os07g39400 exhibited a non-significant difference in expression between NPB and N83. Os07g39480 had the highest expression level among the remaining differentially expressed genes. Os07g39370 and Os07g39480 had significantly lower expression levels in N83 than in NPB, whereas Os07g39430, Os07g39440, and Os07g39450 had significantly higher expression levels in N83 than in NPB (Figure 7). These results indicate that five genes, including Os07g39370, Os07g39430, Os07g39440, Os07g39450, and Os07g39480, were considered the candidate genes underlying qTGW7b.

Discussion
Since the beginning of this century, great progress has been made in cloning major QTLs for grain size-related traits in rice, allowing for the possibility of novel breeding strategies based on genotype information [35,36]. Nevertheless, the QTLs that have been cloned and fine mapped represent only a small proportion of the QTLs detected in the primary populations [37]. It has been assumed that the primary-mapped QTLs are far below the actual basis of trait variation, as most experiments have a low capacity for detecting minor effect QTLs [38]. In this study, we fine mapped a new minor effect QTL qTGW7b that controls grain weight in an 86.2-kb region in chromosome 7.
According to the report, our group has previously identified a minor effect QTL qTGW7 that controls rice grain weight on chromosome 7 between 19.5 M and 22.5 M using the same set of CSSLs [33]. In contrast, the fine mapping interval of qTGW7b in this study deviates from the primary mapping interval, which may be attributable to a number of factors. First, minor effect QTLs have minor effects and can be misidentified phenotypically. Additionally, marker coverage can influence the precision of QTL analysis. Any single contradictory plant during the minor effect QTL detection procedure can result in bias in the QTL mapping region. The above concerns could be why the qTGW7 detection region is 0.9 M to the left of qTGW7b on the chromosome. Additionally, qGL7-2, a major effect QTL for grain length, was also detected in the region adjacent to qTGW7b [39]. Minor effect QTL are often difficult to detect in such a background.
CSSL is a series of continuous target interval hybrid background homozygous NIL obtained by hybridization, self-crossing, backcrossing, and marker-assisted selection (MAS). Using CSSL can eliminate the interference of background and environment on QTL mapping and enable multi-point repeated use over many years, which is widely utilized in QTL mapping [40,41]. In this study, N83 was a single-segment substitution line (Figure 1). The individual plants in the BC 9 F 2 or BC 9 F 3 segregating population of the N83/NPB backcross differed only within the introgression segment of chromosome 7, without background differences. Thereby, phenotypic variations were only caused by QTL within this segment.
The development of functional genomics and molecular genetics in rice will permit the functional validation of candidate genes and molecular mechanistic studies. Consequently, the cloning of QTL for critical agronomic traits in rice is highly dependent on the precision of fine mapping. Our research offers a novel strategy that only requires the detection of a small number of recombinants based on DNA markers in the region of interest. In the beginning, twelve BC 9 F 3 plants with distinct breakpoints in the 8-Mb target region were initially identified. One QTL, designated qTGW7b, was identified within a 651-kb interval using segregating populations derived from self-crossing progeny ( Figure 6A). Afterwards, five BC 9 F 3:4 plants with distinct breakpoints in the putative qTGW7b region were chosen and self-crossed to generate populations that segregated. As significant TGW differences between NPB and 93-11 homozygotes were detected in N2 and N3 that were segregated in different regions, the qTGW7b was narrowed to an 86.2-kb interval ( Figure 6B).
For qTGW7b that has been delimitated within an 86.2-kb interval, the QTL region contained 14 annotated genes. Sequencing and expression analysis identified the candidate genes for qTGW7b as Os07g39370, Os07g39480, Os07g39430, Os07g39440, and Os07g39450. Both Os07g39370 and Os07g39450 encode unidentified proteins. Os07g39430 encodes a mitochondrial transcription termination factor, regulating the transcription and translation of genes. Os07g39440 encodes a zinc finger protein. Several studies indicate that zinc finger proteins are also involved in regulating the size and weight of rice grains. Drought and salt tolerance (DST), which encodes a zinc-finger transcription factor, regulates the expression of Gn1a/OsCKX2 in the reproductive meristem (SAM) directly. DST reg1 , a semi-dominant allele of the DST gene, interferes with DST-directed regulation of OsCKX2 expression and raises CK levels in the SAM, resulting in increased panicle branching, grain number, and grain weight [42]. Lacking rudimentary glume 1 (LRG1) encodes a zinc-finger protein that plays a crucial role in the regulation of spikelet organ identity and grain size. LRG1 mutation results in smaller grain size and diminished grain weight [43]. Os07g39480 encodes a WRKY transcription factor, OsWRKY78, which is involved in the regulation of grain size in rice, and OsWRKY78-RNAi has a smaller grain size and decreased grain weight [44].
According to quantitative genetic theory and modern QTL mapping progress, minor effect QTL play an important role in the regulation of important agronomic traits in rice [45], and this QTL should not be neglected in mechanistic analysis or breeding applications. In this study, the qTGW7b allele from 93-11 significantly increased GL and GW compared to that of qTGW7b NPB , resulting in a 4.5% increase in TGW ( Figure 2, Table 1). The morphology of brown rice became larger without any changes in the rice appearance quality (Figure 2A). Additionally, plants with the qTGW7b 93−11 had more GN and longer GL (Supplementary Figure S2, Supplementary Table S2). Furthermore, qTGW7b 93−11 plants have thicker and stronger stems (Supplementary Figure S3), which contribute to their resistance to lodging. These are all beneficial traits for yield composition. Therefore, qTGW7b can be a valuable genetic resource for grain and plant improvement in rice.

Plant Materials and Growth Conditions
A set of CSSLs was generated by using NPB, a japonica cultivar, as the recipient and 93-11, an indica cultivar, as the donor, as previously reported [46]. N83, a BC 8 F 6 line, was backcrossed with NPB to generate a segregating BC 9 F 2 population for grain weight QTL analysis and fine mapping. Twelve sets of NILs originated from the BC 9 F 3 population, and five NILs populations were derived from a large BC 9 F 3:4 population developed for fine mapping (Figure 8). All plants were grown in paddy fields at Huaisi county of Yangzhou (Jiangsu province, China) or Lingshui (Hainan province, China) under normal cultivation conditions.

DNA Extraction, PCR, and Marker Development
Total DNA was extracted from young rice leaves of each plant with the cetyltrimethylammonium bromide (CTAB) method [47]. The polymerase chain reaction (PCR) was conducted in a 20 µL volume with 10 µL of 2 × Taq Master Mix (Vzayme, Nanjing, China), 2 µL of DNA template, and 6 µL of ddH 2 O. The PCR products were separated by electrophoresis on a 3% agarose gel. All insertion/deletion (Indel) markers used for QTL mapping were developed in accordance with the RiceVarmap 2 website (http://ricevarmap.ncpgr.cn/, accessed on 15 March 2021) by comparing the sequences in the target region between NPB and 93-11 and were detected using agarose gel electrophoresis. All of the Indel markers are listed in Supplementary Table S3.

Scanning Electron Microscope (SEM)
Young spikelets about to be extracted from flag leaf sheaths were collected, fixed with FAA, and vacuumized in a vacuum dryer for 30 min. After dehydration in a gradient of ethanol (30% alcohol, 60% alcohol, 75% alcohol, 85% alcohol, 90% alcohol, 96% alcohol, and 100% ethanol) once for 30 min, three times in pure ethanol, the sample was transferred to the sample box with a critical point dryer for drying treatment. The treated samples were fixed on the pedestal and sprayed with gold, and then the three different parts of the grain were observed by scanning electron microscopy.

Sequence Analysis
The RGAP (Rice Genome Annotation Project, http://rice.uga.edu/ accessed on 20 January 2022) database was used to annotate the 14 open reading frames (ORFs) and their putative functions in the 86.2-kb region between G72 and G32. The sequencing primers for the ORFs were designed based on the NPB sequences in the Phytozome database (https://phytozome-next.jgi.doe.gov/, accessed on 20 January 2022). Phanta Max Master Mix was utilized to formulate the PCR reaction mixture (Vazyme, Nanjing, China). DNAS-TAR software was used to analyze the sequencing results. Supplementary Table S2 of the supplementary materials details the amplifying primers for candidate genes.

Phenotype Investigation
The agronomic traits included plant height (PH), panicle number per plant (PN), panicle length (PL), grain number per panicle (GN), grain length (GL), grain width (GW), grain thickness (GT), thousand-grain weight (TGW), and grain yield per plant (GYP). PH represents the height from the ground to the tip of the highest panicle. PL was measured from panicle neck to panicle tip. Ten fully filled grains were selected for investigation of GL, GW, GT, and TGW. Twenty individual plants of NPB and N83 were chosen randomly for PH, PN, PL, GN, GL, GW, GT, TGW, and GYP. Linkage and QTL analysis were performed on TGW of BC 9 F 2 and BC 9 F 3 segregation populations.

Quantitative Real-Time PCR (qRT-PCR)
For the candidate gene analysis in the target region, young panicles from NPB and N83 were collected. Three plants were harvested. After total RNA was extracted using an RNA extraction kit (Tiangen, Beijing, China), 20 µL of cDNA was synthesized by a one-step RT-qPCR kit (Tiangen, Beijing, China). qRT-PCR was performed using ChamQ Universal SYBR qRT Master Mix (Vazyme, Nanjing, China) in 20 µL. All qRT-PCR primers are listed in Supplementary Table S4.

Data Analysis
The phenotypic differences between NPB and N83 and various NILs were evaluated with Student's t-tests. SPSS Statistics (Versions 25.0, IBM Corporation, Armonk, New York, NY, USA) software was used to analyze the correlation between GL, GW, and TGW. All agronomic traits were investigated at maturity. Paddy grains were dried naturally after harvest for at least one week before testing.

Conclusions
As an essential trait of yield, grain size has a direct effect on the production of rice. There are few reports on the mapping of grain size QTL with a minor effect. In this study, a minor effect QTL qTGW7b that controls grain weight was finely mapped in an 86.2-kb region between markers G72 and G32 on chromosome 7 by constructing a backcross segregating population and NILs, using a set of CSSLs produced by NPB/93-11. Cytological analysis revealed that the qTGW7b allele from 9311 increased cell proliferation in the spikelet, resulting in a 4.5% increase in grain weight. qTGW7b 93−11 also increased stem thickness, GN, GL, and yield. Fine mapping of qTGW7b laid the foundations for further gene cloning and provided a new genetic resource for grain and plant architecture improvement.