The paralogs GmFBXL87a/b independently and cooperatively regulate 100-seed weight for maintaining genetic robustness in soybean evolution

The paralogs GmFBXL87a/b independently and cooperatively regulate 100-seed weight for maintaining genetic robustness in soybean evolution | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 21 October 2025 V1 Latest version Share on The paralogs GmFBXL87a/b independently and cooperatively regulate 100-seed weight for maintaining genetic robustness in soybean evolution Authors : Siming Wei 0000-0002-4733-9744 , Chaorui Kang , Jiayin Han , Xinyue Liu , Mingliang Yang , Fubin Cao , Jianguo Xie , Shuangzhe Li 0009-0004-9337-5537 , Xue Han 0000-0002-0873-6456 , Hongwei Jiang , Zhaoming Qi 0000-0002-0657-9127 , and Qingshan Chen [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.176103210.07633391/v1 177 views 161 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Gene duplication leading to the formation of paralogous genes serves as a compensatory mechanism that helps ensure the genetic robustness of plants. The 100-seed weight of soybean is a crucial domestication trait that determines yield; yet, relatively few genes have been identified, and its genetic and regulatory mechanisms remain unclear. We mapped two major quantitative trait loci (QTL) for 100-seed weight, qHSW12-2 (0.16 Mb interval) and qHSW11-5 (0.3 Mb interval), using two segregating populations derived from chromosome segment substitution lines (CSSLs). GmFBXL87a and GmFBXL87b , cloned from these two QTL intervals respectively, are paralogs encoding F-box proteins. Transgenic analysis of GmFBXL87a and GmFBXL87b demonstrated that these two paralogs regulate 100-seed weight both independently and cooperatively; based on this dual role, we propose their functional mode as the ‘Brotherhood’ model. Furthermore, GmFBXL87a/b affect cell expansion and ultimately determine seed size through both shared and distinct pathways. The larger 100-seed weight haplotype combinations of GmFBXL87a/b were found in less than 51% of soybean accessions in China. This study reveals that soybean paralogous genes play a crucial role in enhancing the genetic robustness of 100-seed weight regulation, and highlighting the potential of GmFBXL87a/b in improving large-seeded soybean varieties. 1 Introduction Genetic robustness may enable organisms to sustain phenotypic consistency in diverse and fluctuating environments (Félix and Wagner 2008). The expansion of gene families and the presence of functional redundancy among paralogous genes can contribute to an organism’s genetic robustness (Diss et al. 2014; Suzuki et al. 2018; Bailon-Zambrano et al. 2022; Dai et al. 2023). The emergence of paralogous genes via gene family duplication could serve as a compensatory mechanism, which helps sustain the robustness in leguminous plants (Yong et al. 2024). Soybean is an ancient tetraploid that has undergone multiple rounds of whole-genome duplication (WGD), including a recent, independent WGD approximately 10 million years ago (MYA). It has subsequently been domesticated in East Asia over the past 6,000–9,000 years (Wang et al. 2021). Multiple WGDs have led to the presence of large, multigene families in soybean. Such gene duplications are an important mechanism for the emergence of evolutionary innovations, leading to the creation of new genes and functions, and resulted in a large number of redundant genes (Lynch and Conery 2000). However, the functions and genetic mechanisms of these redundant paralogous genes in soybean remain unclear. Soybean 100-seed weight is a complex quantitative trait influenced by multiple developmental processes, which is a crucial domestication trait and a significant determinant of overall soybean yield (Xin et al. 2016; Wang et al. 2020). Both seed size and 100-seed weight are complex quantitative traits controlled by multiple genes (Kato et al. 2014; Karikari et al. 2020) and are readily influenced by environmental variations (Li et al. 2019a; Yang et al. 2019). Several soybean genes associated with domesticated 100-seed weight have been cloned by genetic mapping and genome-wide association studies (Lu et al. 2017; Wang et al. 2020; Nguyen et al. 2021; Duan et al. 2022; Li et al. 2022; Cai et al. 2023). Notably, GmSWEET10a has been shown to enhance both seed size and oil content in soybean while simultaneously reducing protein content. GmSWEET10b exhibits functional redundancy with its homolog GmSWEET10a and is currently under selection in breeding programs, making the elite allele GmSWEET10b a promising target for modern soybean breeding efforts (Wang et al. 2020). Investigating the roles of homologous soybean genes can promote advances in precision gene-editing by revealing the mechanisms that underlie their redundant or dosage-dependent effects. To date, the characterization of genes that control soybean seed size/weight has predominantly relied on reverse genetics and comparative transcriptomics (Zhang et al. 2015; Zhou et al. 2015; Gu et al. 2017; Gao et al. 2018; Zhang et al. 2018). However, only a limited number of genes associated with increased seed size and high or stable yields have been cloned. Among the genes known to influence seed size in soybean and other crop species are several genes encoding F-box proteins (FBPs). FBPs are subunits of the SCF complex, one of the largest families of E3 ubiquitin ligases (Abd-Hamid et al. 2020), and function in the recruitment of target substrates (Lechner et al. 2006). As key enzymes of the ubiquitin-proteasome system, E3 ligases participate in numerous biological processes across all eukaryotes (Hua et al. 2011; Yang et al. 2008; Jin et al. 2004; Xu et al. 2009). The SCF complex is formed when a Cullin (Cul) scaffold protein interacts with RING-box protein (Rbx1) at its carboxyl terminus to create the core catalytic structure (Magori and Citovsky 2011). The amino terminus of Cul then interacts with S-Phase Kinase-Associated Protein 1 (SKP1), which can bind one or more different FBPs. FBPs recognize and bind to substrates through their carboxyl-terminal domain, facilitating the ubiquitination and subsequent degradation of their targets (Lechner et al. 2006; Zhang et al. 2019). They participate in multiple essential biological processes in plants, including phytohormone signaling, cell signaling, development and morphogenesis, the cell cycle, and circadian clock regulation (Stefanowicz et al. 2015). FBPs also have roles in self-incompatibility, biotic and abiotic stress responses, plant–pathogen interactions, and secondary metabolite biosynthesis, as highlighted in previous studies (Abd-Hamid et al. 2020). OsFBK12, an F-box protein with a Kelch repeat motif, has been shown to influence leaf senescence, seed size, and grain number in rice (Chen et al. 2013). Overexpression of SLB1, an F-box protein and an ortholog of sterile apetala (SAP) from Arabidopsis thaliana , increased seed and leaf size in both Medicago truncatula and soybean (Yin et al. 2020). Similarly, overexpression of the FBP gene GmFBXL12 in soybean increased seed size, pod number, and seed number per plant (Hina et al. 2024). Despite the important functions of FBPs, particularly those with C-terminal leucine-rich repeat (LRR) domains, few have been extensively characterized in soybean. In this study, we used two CSSL-derived populations to fine-map the 100-seed weight QTLs qHSW12-2 and qHSW11-5 , with qHSW12-2 overlapping the QTL qHSW-12-3 reported in our previous work (Zheng et al. 2022). Based on genome sequencing and expression analysis, we further identified Glyma.12G088900 ( GmFBXL87a ) and Glyma.11G183600 ( GmFBXL87b ) as candidate genes for 100-seed weight within the intervals of these two QTLs; notably, these two genes are paralogs encoding FBXL proteins. Both GmFBXL87a and GmFBXL87b regulate 100-seed weight by mediating cell expansion and modulating the expression of cell growth-related genes. Additionally, we detected signatures of domestication selection for GmFBXL87a and GmFBXL87b , and confirmed the breeding potential of their optimal haplotype combinations. These results not only deepen our understanding of the genetic mechanisms by which paralogs govern soybean seed traits, but also provide valuable haplotype resources and mechanistic insights for optimizing molecular breeding strategies targeting soybean seed weight improvement. 2 Materials and Methods 2.1 Plant materials Two small-seeded CSSLs, R158 and R86, were selected from a BC₃F₆ CSSL population comprising 213 lines (Zheng et al. 2022). The population was derived from a cross between the wild soybean ( Glycine soja ) accession ZYD06 (donor parent) and the cultivated soybean ( Glycine max ) cultivar SN14 (recurrent parent), both originating from Heilongjiang Province, China. Genetic populations, including R158, R86, SN14, their F₂ progeny, and recombinant homozygous lines (RHLs, F₃ generation), were cultivated in experimental fields at Harbin, China (126°38´E, 45°45´N) across multiple growing seasons (2020–2023). Field trials followed a randomized complete block design with three replicates. Each plot consisted of 5-m-long rows with a 6-cm spacing between plants and a 60-cm spacing between rows. In addition, transgenic soybean plants were grown under controlled greenhouse conditions (25°C, 16/8-h light/dark cycle) as well as in field plots at Harbin, China to evaluate yield-related traits. Standard agronomic practices, including irrigation and pest control, were uniformly applied. Phenotypic data for 100-seed weight were collected from mature seeds of 20 plants per line. RHLs were genotyped using SSR markers to confirm the presence of substituted ZYD06 segments. 2.2 Yield-trait evaluation Seed weight was determined by randomly sampling 100 or 1000 fully mature, undamaged seeds per genotype and weighing them on a precision balance (0.01-g accuracy). The lengths and widths of five randomly selected seeds per line were measured using digital calipers, and the values were averaged across three biological replicates (5 seeds/replicate). Individual plant traits, including total seed weight, pod weight, and plant height (measured from the first growth node to the apical meristem of the main stem), were recorded from 20 representative plants per line. All measurements were performed with three technical replicates to ensure data reliability. 2.3 Genotyping and QTL mapping Genomic DNA was extracted from young leaf tissues of soybean plants using a magnetic bead-based nucleic acid extraction system by Chengdu Hanchen Guangyi Technology (Chengdu, China) according to the manufacturer’s protocol. Whole-genome resequencing of R158 was performed on the Illumina NovaSeq 6000 platform by Biomarker Technologies (Beijing, China). For genotyping analysis, polymorphic SSR markers based on the substituted segments in R158 and R86 were selected from SoyBase (https://www.soybase.org/) (Table S5). PCR amplification was performed using a touchdown protocol, and the products were separated on 8% non-denaturing polyacrylamide gels with silver staining for polymorphism detection. SNP markers were developed by comparing resequencing data from the parental lines SN14 and ZYD06 and were used for fine mapping in the RHLs. QTL analysis was performed using the inclusive composite interval mapping (ICIM) method in IciMapping 4.1 software (https://isbreeding.caas.cn), with a logarithm of odds (LOD) threshold of 3.0 determined by 1000 permutation tests. 2.4 Reverse transcription-PCR (RT–PCR) Total RNA was isolated from soybean seeds at the Cot, EM1, and EM2 developmental stages (10, 20, and 25 days after flowering) using the RNAprep Pure Polysaccharide Polyphenol Plant Total RNA Extraction Kit (Tiangen Biotech, Beijing, China), incorporating an on-column DNase I digestion step to eliminate genomic DNA contamination. First-strand cDNA was synthesized from 1 μg total RNA using HiScript III RT SuperMix for qPCR (Vazyme Biotech, Nanjing, China) according to the manufacturer’s protocol. For RT–PCR analysis, gene-specific primers (Table S5) were designed to amplify target sequences with an annealing temperature gradient optimization. Quantitative real-time PCR (qRT–PCR) was performed on a LightCycler 480 II system (Roche Diagnostics) using ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech). Three biological replicates were performed for each genotype. The soybean Actin11 gene ( Glyma.18G290800 ) (Hu et al. 2009; Li et al. 2012) served as the endogenous control, and relative expression levels were calculated by the 2 −ΔΔCt method (Livak and Schmittgen 2001). Hierarchical clustering of expression patterns was visualized using the pheatmap package (v1.0.12) in R 4.0.2 (R Team 2020). 2.5 Plant transformation To generate GmFBXL87a -overexpressing transgenic soybean plants, the 1260-bp coding sequence (CDS) of GmFBXL87a was amplified from soybean cDNA using gene-specific primers (Table S5) with Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific). The purified PCR product was cloned into the KpnI and BamHI restriction sites of the binary vector pSOY1, creating the 35S::GmFBXL87a overexpression construct. For CRISPR-Cas9-mediated gene editing, two 23-nt single guide RNAs (sgRNAs) targeting conserved exonic regions of both GmFBXL87a and GmFBXL87b were designed (Target1 and Target2) using CRISPR-P 2.0 software and inserted into the pGES401 vector (Bai et al. 2022) to generate the pGES401- GmFBXL87a/b construct. All recombinant plasmids ( 35S::GmFBXL87a and pGES401- GmFBXL87a/b ) were transformed into chemically competent cells of Agrobacterium tumefaciens strain EHA105 (Weidi Biotechnology, Shanghai, China) via the freeze–thaw method. The transformed Agrobacterium cells were used to infect the soybean cultivar DN50 (Zeng et al. 2004) by cotyledonary node transformation. Putative transgenic plants were selected on medium containing 50 mg/L hygromycin B and verified by genomic PCR, Sanger sequencing, and qRT-PCR analysis using gene-specific primers (Table S5). For comparative functional analysis in Arabidopsis thaliana , a T-DNA insertion mutant ( atfbxl87 ) of the Arabidopsis thaliana ortholog AT1G67190 was obtained from the Arabidopsis Biological Resource Center (ABRC) (https://abrc.osu.edu/). Genomic DNA was extracted from well-developed rosette leaves of wild-type Col-0 and the atfbxl87 mutants. Detection primers (LP, BP, RP) were retrieved from the SIGnAL website (http://signal.salk.edu/tdnaprimers.2.html), and PCR amplification was conducted using the extracted DNA as template. Specific primers were used to determine the homozygosity of the mutant (Table S5). Unless otherwise noted, all phenotypic analyses were performed using homozygous T 4 -generation Arabidopsis mutants and T 3 -generation transgenic soybean lines. 2.6 Cytological analyses The parent lines used were SN14, ZYD06, and R158, and the transgenic materials (T 2 ) included DN50, crfbxl87a , crfbxl87b , crfbxl87a/b , and OEFBXL87a . Developing seeds from the Cot, EM1, and EM2 stages were fixed in FAA solution (70% ethanol) for 2 days. Samples underwent graded dehydration in 75%, 85%, 95%, and absolute ethanol for 2 hours each. Transparency was achieved using a 1:1 solution of ethanol and xylene for 2 hours, followed by pure xylene for 2 hours. Paraffin embedding was performed at 42°C, with wax added gradually until the xylene was replaced. Sections (7 μm) were cut, mounted, and stained. De-waxing was performed at 65–70°C for 30 minutes, followed by rinsing and staining with 1% safranin for 30 seconds and 0.5% solid green for 8 seconds. Mature soybean seeds were halved, and one half was mounted on a metal sample holder using conductive tape and gold-coated via ion sputtering. Scanning electron microscopy (SEM) images were captured using a Hitachi S-3400N microscope at 600× magnification, and selected areas were magnified to 1000× for detailed analysis. The total number of epidermal cells on the cotyledon surface for each genotype was calculated by extrapolating from the number of cells per unit area at 600× magnification (3.26E-2 mm 2 ) and the corresponding total area of the cotyledons (Wei et al. 2023). For each genotype, three biological replicates were performed, each consisting of three cotyledons. Cells were observed using an optical microscope, and statistical analysis was performed with Axio Vision Rel. 4.7 software (Carl Zeiss AG, Oberkochen, Germany). 2.7 Subcellular localization The subcellular localization of GmFBXL87a and GmFBXL87b proteins was predicted using WoLF PSORT (https://wolfpsort.hgc.jp/). For experimental validation, the coding sequences of GmFBXL87a and GmFBXL87b were cloned into the pSOY1 vector (lacking stop codons) to generate N-terminal GFP fusion constructs driven by the CaMV 35S promoter ( 35S::GmFBXL87a-GFP and 35S::GmFBXL87b-GFP ). The empty 35S::GFP vector served as a control. For agroinfiltration, all constructs were transformed into Agrobacterium tumefaciens strain GV3101. Bacterial cultures were grown to OD600 = 0.6 in LB medium at 28°C with 220 rpm shaking, harvested by centrifugation (4000 × g , 10 min, 4°C), and washed twice with infiltration buffer (10 mM MgCl 2 , 10 mM MES [pH 5.7]). The pellets were resuspended in the same buffer supplemented with 200 μM acetosyringone to OD600 = 0.6 and incubated at room temperature for 2–3 h. The suspensions were pressure-infiltrated into the abaxial side of Nicotiana benthamiana leaves using a needleless syringe. After 48 h of incubation under controlled conditions (22°C, 16/8-h light/dark), GFP signals were observed using an Olympus FV3000 confocal laser scanning microscope (488 nm excitation). Images were processed with FluoView v5.0 software (Olympus, Tokyo, Japan). 2.8 Luciferase activity assay in tobacco leaves The 2 kb promoter fragments of Glyma.12G088900 SN14pro , Glyma.12G088900 ZYD06pro , Glyma.11G183600 SN14pro , and Glyma.11G183600 ZYD06pro were inserted into the pGreen II-0800-nLUC vector. The resulting constructs were introduced into Agrobacterium tumefaciens strain GV3101 and infiltrated into leaves of 4–5-week-old tobacco plants by needle injection. After 48 hours, photos were taken, and 0.1 g leaf sample was harvested and immediately frozen in liquid nitrogen. Samples were lysed with 300 µL of 1x Cell Lysis Buffer, and luciferase activity was quantified using dual-luciferase assays to measure LUC and REN values. The relative activity was expressed as the LUC/REN ratio. 2.9 Protein structure prediction The protein sequences of GmFBXL87a and GmFBXL87b were analyzed using comprehensive bioinformatics approaches. Homologous sequences were identified from cultivated soybean ( Glycine max ), wild soybean ( Glycine soja ), Arabidopsis thaliana , and Medicago truncatula using Phytozome v13.0 (https://phytozome-next.jgi.doe.gov/). Protein domain architecture was predicted using Pfam v35.0 (https://pfam.xfam.org/). For structural modeling, the amino acid sequences of GmFBXL87a/b were submitted to the AlphaFold3 (https://alphafoldserver.com/). The human F-box protein SKP2 (PDB ID: 2ASS) was selected as a template because it had the highest sequence identity (42%) to GmFBXL87a/b. The predicted 3D structures were visualized and annotated using UCSF ChimeraX 1.4 (https://www.cgl.ucsf.edu/chimerax/), and conserved domains were highlighted on the basis of multiple sequence alignment performed in DNAMAN 9.0 (Lynnon Biosoft) using the BLOSUM62 matrix (https://www.lynnon.com/alignm.html). 2.10 Nucleotide diversity, haplotypes, and geographic distribution A total of 2898 soybean accessions and 547 additional soybean accessions were used for population genetic analysis (Liu et al. 2020). Population genetic parameters were calculated using established bioinformatics tools. Nucleotide diversity (π) and fixation index ( F ST ) values were computed in 500-kb sliding windows with 20-kb steps using DnaSP 5.10 (Librado and Rozas 2009) and VCFtools (Danecek et al. 2011). The relative nucleotide diversity ratio was expressed as 1/θπ, calculated by dividing π values from high-diversity regions by those from low-diversity regions. Neutrality tests were performed by estimating Tajima’s D across 500-kb genomic windows. Linkage disequilibrium patterns were analyzed using PLINK (Purcell et al. 2007), and haplotype network reconstruction was performed in PopART (http://popart.otago.ac.nz) using the TCS algorithm (Clement et al. 2000). Geographical distribution patterns were visualized and analyzed using ArcGIS 10.5 (ESRI) (https://www.esri.com/en-us/home). 2.11 Phylogenetic tree construction Protein sequences of GmFBXL87a/b and their homologs in Arabidopsis thaliana , Glycine soja and Glycine max were download from phytozome database (https://phytozome-next.jgi.doe.gov/). Multiple sequence alignment was performed using MUSCLE v3.8.31 (Edgar 2004) with default parameters. The resulting alignment was used to reconstruct phylogenetic relationships with the maximum likelihood method implemented in IQ-TREE (Nguyen et al. 2015). Branch support was assessed through 1000 bootstrap replicates. The final phylogenetic tree was visualized and annotated using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/). 2.12 DNA sequencing and primer synthesis All oligonucleotide primers (Tables S1) were synthesized by Taihe Biotechnology (Beijing, China). DNA fragment verification, cDNA sequencing, and plasmid confirmation were performed by Tsingke Biotechnology (Beijing, China) using Sanger sequencing technology. 2.13 Statistical Analysis Statistical significance was assessed by two-tailed Student’s t -tests for pairwise comparisons or one-way ANOVA with post-hoc tests for multiple comparisons using a significance threshold of P < 0.05 in SPSS Statistics 17.0 (IBM, USA). Data were visualized with R 4.0.2, using ggplot2 (Galensa, 2017) for violin plots and ggpubr (Kassambara 2017) for frequency histograms. 3 Results 3.1 Identification and characterization of the small-seeded CSSL R158 and R86 The small-seeded CSSL R158 and R86 were selected from a wild soybean CSSL population whose parents were the cultivated variety ‘Suinong14’ (SN14) and the wild soybean ‘ZYD00006’ (ZYD06) (Zheng et al. 2022). Phenotyping over three years revealed that 100-seed weight, seed length, and seed width were greater in R158 and R86 than in ZYD06 but significantly lower than in SN14 (Figure 1A–D). In addition, there was no significant difference in seed weight per plant among SN14, R158 and R86 (Figure S1). Both R158 and R86 were therefore a stable small-seeded CSSL with reduced seed size and weight. These results suggested that the seed-related phenotypic variation in R158 and R86 were attributable to substituted chromosomal segments of ZYD06 present in the SN14 background. Because seed development is tightly coupled to cellular developmental processes (Li et al. 2019b), we examined paraffin sections of SN14, R158, R86, and ZYD06 seeds collected at different developmental stages. Across all genotypes, the average area of individual cells gradually increased during seed development. Notably, cell area was significantly larger in developing seeds of the wild soybean ZYD06 than in those of R158, R86 and SN14 (Figure S2). These differences may stem from complex genomic variations between wild and cultivated soybeans, suggesting that comparisons among SN14, R158 and R86, most of whose genomes are identical, could offer valuable insights into the role of the substituted chromosomal segments in cellular regulation. Within developing cotyledons, the average cell area was larger in SN14 than in R158 and R86, whereas cell number in the same cotyledon area was higher in R158 and R86 than in SN14 (Figure S2). Consistent with these findings, SEM observations of mature seeds revealed comparable structural variations (Figure 1E–K). The inner area of the cotyledon central surface was approximately 13.35%, 34.81% larger in SN14 than in R158 and R86, respectively (Figure 1E). Compared to SN14, R158 and R86 had a 15.6% and 34.5% smaller average cell area, respectively, but a 4.5% and 0.5% increase in cell number (Figure 1F–K). These results suggest that the substituted chromosomal segments in both R158 and R86 primarily promote cell expansion. 3.2 Fine mapping the QTL qHSW12-2 and qHSW11-5 for 100-seed weight Whole-genome resequencing of both R158 and R86 identified substituted segments from ZYD06 on 11 chromosomes (Figure S3). To map QTLs associated with reduced 100-seed weight in R158 and R86, we constructed an F 2 segregating population of 553 lines and 647 lines by crossing SN14 and R158, SN14 and R86, respectively (Figure 2, Figure 3). The QTL on chromosome 12 spanned a 3.47-Mb region between markers M157 and M180 mapped from SN14 and R158 backcrossed population (Figure 2A) and overlapped with another 100-seed weight QTL mapped in our previous work using CSSLs (Zheng et al. 2022). Simultaneously, the QTL on chromosome 11 spanned a 2.15-Mb region between markers M13 and M17 mapped from SN14 and R86 backcrossed population (Figure 3A). We therefore focused our subsequent work on the QTL on chromosome 11 and chromosome 12. 3.3 Positional cloning of qHSW12-2 and qHSW11-5 To fine-map the QTL and identify the candidate gene(s) associated with 100-seed weight, we selected two residual heterozygous inbred line (RHL) populations of 224 lines and 173 lines from the F 2 populations, respectively, which segregated only at the QTL on chromosome 11 and chromosome 12 (Figure 2A, B, Figure 3A, B). By combining phenotype and genotype data for the RHLs, we fine mapped the two QTLs to a 0.16-Mb and 0.3-Mb interval, which we named qHSW12-2 and qHSW11-5 (Figure 2B, Figure 3B). Seventeen and twenty-four candidate genes were annotated on the qHSW12-2 and qHSW11-5 interval (Figure 2A, Table S1), referring to ‘Williams 82 ’ as the reference genome (Schmutz et al. 2010). Whole-genome resequencing data revealed that eight of these genes had nonsynonymous variations between SN14 and ZYD06 on qHSW12-1 (Figure 2B), and four of these genes had nonsynonymous variations on qHSW11-5 (Figure 3B). Three of candidate genes were highly expressed specifically in soybean seeds on both qHSW12-1 and qHSW11-5 , but only Glyma.12G088900 and Glyma.11G193600 was differentially expressed between seeds of SN14 and ZYD06 (Figures 2B, 3B; Figure S4) (Libault et al. 2010; Sreedasyam et al. 2023). The functions of its homologs in other species and a genotype analysis also supported the identification of Glyma.12G088900 and Glyma.11G183600 as the candidate gene associated with 100-seed weight (Figure 2B). The coding sequence of Glyma.12G088900 and Glyma.11G183600 did not differ between SN14 and ZYD06 (Figures 2C, 3C) . H owever, there were six variant sites (five SNPs and one Indel) between SN14 and ZYD06 on the Glyma.12G088900 promoter region, and fourteen variant sites between SN14 and ZYD06 on the Glyma.11G183600 promoter region (Figures 2C, 3C, Table S2). Two SNP/Indel markers were designed on the Glyma.12G088900 and Glyma.11G183600 promoter region and used to genotype the RHL population. The results indicated that the 100-seed weight of the homozygous SN14 genotype was significantly greater than that of the heterozygous and homozygous ZYD06 genotypes (Figures 2D, 3D). Moreover, transient expression assays revealed that the promoter activity of Glyma.12G088900 ZYD06pro and Glyma.11G183600 ZYD06pro were significantly greater than that of Glyma.12G088900 SN14pro and Glyma.11G183600 SN14pro , respectively, suggesting that natural genetic variation in Glyma.12G088900 and Glyma.11G183600 between SN14 and ZYD06 caused differential expression of the two genes between the two haplotypes (Figure 2E–G; Figure 3E–G; Figure S4). These results provided further evidence that Glyma.12G088900 and Glyma.11G183600 were the casual 100-seed weight candidate genes. 3.4 The GmFBXL87a and GmFBXL87b were paralogs in soybean Functional annotation and phylogenetic analysis revealed that both Glyma . 12G088900 and Glyma . 12G183600 encoded F-box proteins from the F-box and leucine-rich repeat (FBXL) family, and the two paralogs have a closer phylogenetic relationship with Glyma.19G194600 (Figure 4A, B). Notably, Glyma . 12G088900 and Glyma . 11G183600, showed 98% sequence similarity and contained a complete FBXL domain ( Figure 4C, Figures S5, S6 ). We therefore named the two paralogs GmFBXL87a (Glyma . 12G088900) and GmFBXL87b ( Glyma . 11G183600). Protein tertiary structure predictions indicated that there were no major differences in the 3D structures of their encoded proteins ( Figure 4D ). The tissue expression patterns of GmFBXL87a and GmFBXL87b (GmFBXL87a/b) were similar. Both showed high expression in the root and during the late stages of seed development, and expression levels of GmFBXL87a/b were higher in R158 and ZYD06 than in SN14 ( Figure 4E ). The results of subcellular localization assays indicated that both GmFBXL87a and GmFBXL87b were present in the nucleus and cytoplasm ( Figure S7 ). Thus, the two paralogs were likely to perform similar functions . 3.5 The paralogs GmFBXL87a and GmFBXL87b independently and cooperatively control 100-seed weight We next investigated the functions of GmFBXL87a/b in soybean by CRISPR/Cas9-mediated gene editing in the ‘Donnong50’ (DN50) background, which has the same GmFBXL87a/b haplotype as SN14 (Figure S8). We obtained three independent homozygous mutant lines for GmFBXL87a ( crfbxl87a -1 to crfbxl87a -3) and GmFBXL87b ( crfbxl87b -1 to crfbxl87b -3), as well as three independent homozygous double-mutant lines for GmFBXL87a/b ( crfbxl87a/b -1 to crfbxl87a/b -3) (Figure S9A–E). The crfbxl87a , crfbxl87b , and crfbxl87a/b mutants carried different edits in GmFBXL87a/b , which caused frameshifts and early termination of GmFBXL87a/b translation (Figure S9F). The 100-seed weight of the GmFBXL87a -edited and GmFBXL87b -edited single mutants was approximately 16.62% and 22.07% greater than that of DN50, respectively (Figure 5A, B). Most interestingly, the 100-seed weight of the GmFBXL87a/b double mutants was 33.54% greater, an improvement larger than that observed in the single-gene mutants (Figure 5A, B). Accordingly, seed length and width were also significantly greater in the mutants (Figure 5C, D). Plant height, pod number, pod width, and yield per plant did not differ between the mutants and DN50 (Figure S10A–D), but pod width was greater and seed number per plant was lower in the mutants (Figure S10E, F). Together, these differences could explain why yield per plant was unchanged in the mutants. We next examined an Arabidopsis mutant of the GmFBXL87a ortholog AtFBXL87a ( AT1G67190 ) and found that it also had a significantly higher 1000-seed weight than the wild type (Figure S11), consistent with the phenotype of the soybean crfbxl87a mutant. Moreover, GmFBXL87b was upregulated in the in crfbxl87a mutants, and GmFBXL87a was upregulated in the crfbxl87b mutants (Figure S12), suggesting that loss of one paralog is compensated by upregulation of the other. These results indicate that GmFBXL87a and GmFBXL87b have independent effects on 100-seed weight but also display partial functional complementarity. To confirm the role of GmFBXL87a in soybean seed development, we generated six transgenic soybean lines overexpressing GmFBXL87a ( OEGmFBXL87a-1 to OEGmFBXL87a-6 ) in the DN50 background (Figure S13A, B). The 100-seed weight of the OEGmFBXL87a transgenic plants was 18.41% lower than that of DN50 (Figure 5E, F). Consistent with their 100-seed weight, the seed length and width of OEGmFBXL87a plants were 14.02% and 16.12% lower, respectively, than those of DN50 (Figure 5G, H). Plant height, seed weight per plant, pod number, and pod width of the OEGmFBXL87a plants did not differ significantly from those of DN50 (Figure S13C–F), but their pod length was reduced and their 3seed number per plant was greater (Figure S15G, H). These results demonstrate that GmFBXL87a functions as a suppressor of 100-seed weight, seed size, and pod length but positively regulates seed number per plant; thus, its mutation or overexpression does not significantly alter the total yield per plant. 3.6 GmFBXL87a and GmFBXL87b mainly act as negative regulators of cell expansion Seed size increased rapidly from the Cot stage to the EM1 and EM2 stages in SN14, R158, and ZYD06, and GmFBXL87a expression also increased significantly from Cot to EM1 and EM2 in all three materials ( Figure S 14). The expression pattern of GmFBXL87b was similar to that of GmFBXL87a ( Figure 4E ). Therefore, both GmFBXL87a and GmFBXL87b appeared to be involved in seed development, primarily during the later stages. We next examined paraffin sections of cotyledons from DN50 and GmFBXL87a/b transgenic seeds collected at different developmental stages. At the Cot and EM1 stages, there were no significant differences in the cell number or the average area of individual cells among any of the materials (Figure 6A–E). However, at EM2, cell number was 17.01%, 17.94%, and 21.12% lower in the crfbxl87a , crfbxl87b , and crfbxl87a/b mutants than in DN50, respectively, whereas the average area of individual cells was 20.67%, 22.60%, and 27.36% greater (Figure 6F, G). By contrast, cell number was higher in OEGmFBXL87a than in DN50, but average cell area was lower (Figure 6F, G). To further examine the effects of GmFBXL87a/b on 100-seed weight, we examined mature seeds of DN50 and GmFBXL87a/b transgenic plants by SEM. The area of the ventral seed surface was 23.15%, 28.21%, and 39.40% greater in crfbxl87a , crfbxl87b , and crfbxl87a/b mutants seeds than in DN50, respectively, whereas that of the OEGmFBXL87a seeds was 17.64% smaller (Figure 6H). Consistent with the results obtained from developing seeds, the average cell area in mature seeds was 27.35%, 30.03%, and 35.89% greater in the crfbxl87a , crfbxl87b , and crfbxl87a/b mutants than in DN50, whereas the total cell number was 7.71%, 8.19%, and 10.00% lower (Figure 6I–M). By contrast, cell number in mature seeds was 37.11% higher in OEGmFBXL87a than in DN50, whereas the average cell area was 19.75% lower (Figure 6N, O). The increased cell area of the crfbxl87a , crfbxl87b , and crfbxl87a/b mutants more than compensated for their reduced cell numbers, leading to an increase in total seed area. These results demonstrate that GmFBXL87a/b determine 100-seed weight mainly through their effects on cell expansion. Moreover, both paralogs contribute to 100-seed weight through a combination of independent and coordinated regulation of cell expansion. 3.7 GmFBXL87a/b influence the expression of genes that control cell growth and seed size To further investigate the involvement of GmFBXL87a/b in cell growth, we analyzed the expression patterns of published genes related to cell division and cell expansion in GmFBXL87a/b transgenic soybean lines and DN50 (Table S3). qRT–PCR results showed that Growth-Regulating Factor 5 ( GRF5) was downregulated in all crfbxl87a , crfbxl87b , and crfbxl87a/b lines but upregulated in OEGmFBXL87a lines compared to DN50 at the EM2 stage. By contrast, Target of Rapamycin ( TOR ) and Auxin Response Factor 2 ( ARF2 ) were upregulated in crfbxl87a and crfbxl87a/b mutants but downregulated in gmfbxl87b and OEGmFBXL87a lines (Figure 7A). We also examined the expression of twelve genes involved in pathways known to regulate soybean seed size (Table S3). Among them, expression of Gibberellin 20-Oxidase ( GmGA20OX), Protein Phosphatase 2C-1 (GmPP2C-1), and Seed Thickness 05 ( GmST05) was higher in the crfbxl87a , crfbxl87b , and crfbxl87a/b mutants than in DN50 at the EM2 stage but lower in OEGmFBXL87a lines. Notably, expression of these three genes was higher in crfbxl87a/b mutants than in the single mutants (Figure 7A). These expression results suggest that GmFBXL87a/b appear to be integrated into the molecular pathways that control cell growth and seed development. Although 100-seed weight increased sequentially in the crgmfbxl87a , crgmfbxl87b , and crgmfbxl87a/b mutants, the differences in cell area and 100-seed weight of the crgmfbxl87a/b mutants were less than the sum of the differences in the crgmfbxl87a and crgmfbxl87b mutants. Moreover, and the expression levels of crgmfbxl87a were elevated in the crgmfbxl87b mutant, and those of crgmfbxl87b in the crgmfbxl87a mutant. These results indicate that GmFBXL87a and GmFBXL87b can independently regulate 100-seed weight but also exhibit functional coordination. Their interaction is consistent with a gene dosage effect typical of duplicated genes, resembling a ‘brotherly’ relationship in which each function independently yet supports the other. We therefore propose the ‘Brotherhood’ seed-size-control model, which describes how the two paralogs jointly regulate 100-seed weight (Figure 7B). Because mutation or overexpression of GmFBXL87a/b influenced the expression of three genes associated with cell division and expansion ( ARF2 , GRF5 , and TOR ) and three genes involved in the regulation of soybean seed size ( GmGA20OX , GmPP2C-1 , and GmST05 ), GmFBXL87a/b influence cell growth and ultimately determining seed size through both shared and distinct pathways (Figure 7C). 3.8 GmFBXL87a/b were selected during soybean domestication Using published genomic resequencing data from 2898 soybean accessions (Liu et al. 2020), we calculated nucleotide diversity along chromosome 12 and obtained θπ values of 0.16–0.39, 0.26–0.53, and 0.59–0.74 in a 500-kb window around GmFBXL87a for cultivar/wild, landrace/wild, and cultivar/landrace, respectively ( Figure S 15A). Analysis of population genetic differentiation revealed F ST values of 0.34–0.35, 0.26–0.27, and 0.013–0.014 for cultivar/wild, landrace/wild, and cultivar/landrace in the same interval ( Figure S 15B), and Tajima’s D values were –0.12– –0.92, –1.25– –0.06, and –0.12–0.38 for cultivar, landrace, and wild ( Figure S 15C). Likewise, in the 500-kb region surrounding GmFBXL87b on chromosome 11, we obtained θπ values of 0.39–0.55, 0.64–0.87, and 0.61–0.64 for cultivar/wild, landrace/wild, and cultivar/landrace, respectively ( Figure S 16A). F ST values were approximately 0.21, 0.15, and 0.02 for cultivar/wild, landrace/wild, and cultivar/landrace in the same region ( Figure S 16B), and Tajima’s D values were –1.37– –0.85, 0.44–0.61, and –0.92– –0.69 for Cultivar, Landrace, and Wild ( Figure S 16C). Furthermore, the paralogs GmFBXL87a/b with a K A /K S value of 0.15, indicating that they have undergone purifying selection ( Figure S 16D). These results suggest that the genomic regions containing GmFBXL87a/b may have both undergone selection during soybean domestication. We performed a haplotype analysis using 2898 soybean accessions and identified 11 haplotypes for GmFBXL87a . Hap1, Hap2, and Hap3 were the primary haplotypes in wild soybean, whereas the main haplotypes in landraces and improved soybeans were (in order) Hap7, Hap15, Hap16, and Hap17 (Figure S17A, B). Hap15 was by far the most common haplotype in landraces and improved soybean, and it was also associated with the largest 100-seed weight ( Figure S17B–D ). Hap15 was the major haplotype (≥80% frequency) in all four studied regions of China (NER, Northeast region; NR, North region; HR, Huanghuai region; and SR, South region ) ( Figure S 17E), and it had the highest 100-seed weight across all four regions ( Figure S 17F–I). We identified eight haplotypes for GmFBXL87b : Hap1and Hap2 were the primary haplotypes in wild soybean, whereas the major haplotypes in landraces and improved soybeans were (in order) Hap6, Hap8, Hap1, and Hap2 (Figure S18A, B). Hap6 and Hap8 were much more common than the other two haplotypes in landraces and improved soybeans, and they were associated with the largest 100-seed weights ( Figure S18B–D ). Hap6 and Hap8 were also the major haplotypes in NER, NR, HR, and SR ( Figure S 18E) and consistently had the highest 100-seed weights across all locations ( Figure S 18F – I). The proportion of Hap6 gradually decreased from NER to SR. Therefore, in northern regions, Hap6 haplotypes can be used as genetic resources for breeding superior large-seeded varieties, whereas in southern regions, GmFBXL87b shows greater breeding potential. 3.9 The larger haplotype combination of GmFBXL87a/b for 100-seed weight has significant breeding potential We next examined the frequency and distribution of eleven different GmFBXL87a/b haplotype combinations. Hap15 a &Hap6 b was the predominant haplotype combination in landraces and improved soybeans, accounting for 52% and 68% of these accessions, respectively ( Figure 8A). Hap15 a &Hap8 b , Hap15 a &Hap6 b , and Hap17 a &Hap6 b had significantly higher 100-seed weights than the other haplotype combinations ( Figure 8B, C). We classified the haplotype combinations into three groups on the basis of their 100-seed weights, from highest to lowest ( Figure 8C). In both landraces and improved soybeans, the 100-seed weight of Group 1 was significantly higher than those of Groups 2 and 3, and the 100-seed weight of Group 2 was significantly higher than that of Group 3 ( Figure 8D, E). Group 1 combinations were present at frequencies of 48.04%, 43.37%, 45.2%, and 13.56% in NER, NR, HR, and SR, respectively ( Figure 8F), and Group 2 combinations were present at frequencies of 50.2%, 50%, 50%, and 16.93% in the same regions ( Figure 8F). The 100-seed weight of Group 1 was higher than that of Groups 2 and 3 in NR, HR, and SR ( Figure 8G –J ). These results indicate that haplotype combinations in Group 1 produce the greatest 100-seed weight and therefore have significant breeding potential in all four regions of China. 4 Discussion 4.1 CSSLs enable cloning of the qHSW12-2 and qHSW11-5 QTL for 100-seed weight At present, 545 QTLs related to seed weight are recorded in the SoyBase database. (https://www.soybase.org/), 17 of which are mapped to chromosome 12 and 26 QTLs are mapped on chromosome 11 (Wei et al. 2023). However, most seed-weight QTLs have not been fine mapped owing to a lack of suitable genetic segregating populations. Primary genetic-mapping populations are only suitable for initial QTL mapping (Liu et al. 2013). Because 100-seed weight is controlled by numerous genes, it is susceptible to interference from genetic backgrounds during fine mapping ( Zhang et al. 2015; Nguyen et al. 2021 ). Such interference can be minimized using CSSL population, which have a relatively consistent genetic background, making them suitable for the fine mapping of complex quantitative traits ( Yano et al. 2001 ). This approach can significantly enhance the efficiency and accuracy of fine-mapping results (Liu et al. 2004; Thomson 2006). The first CSSL population in crops was established in tomato ( Eshed and Zamir 1995 ), and CSSL populations have subsequently been developed in crops such as rice, cotton, and barley (Pillen et al. 2003; Shen and Xing 2014; Zhu et al . 2020). Currently, there are only two CSSL populations in soybean (Yang et al. 2019; Zheng et al. 2022). The construction of CSSL-derived populations that segregate for a single trait is particularly useful for fine mapping of complex traits, and the use of wild resources enables the discovery of new superior alleles (Xin et al. 2016; Yang et al. 2019; Zheng et al. 2022). Here, we used CSSL R158 and R86 from a wild soybean CSSL population that showed reduced 100-seed weight and crossed it with one of the parents, SN14, to construct a segregating population for 100-seed weight. We used this population to fine map qHSW12- 2 and qHSW11- 5 , two QTLs for 100-seed weight, to a 160-kb and 300-kb interval on chromosome 12 and chromosome 11, respectively. Ultimately, we identified two FBXL paralogs, Glyma.12G088900 and Glyma.11G183600 , within the qHSW12-2 and qHSW11- 5 region. Notably, qHSW12-2 overlapped with a previously published QTL for 100-seed weight ( qHSW12-3 ), which has an interval size of only 64.4 kb and was fine-mapped using another CSSL, R183 (Zheng et al. 2022). Therefore, qHSW12-2 and qHSW12-3 may ultimately map to the same causal gene. 4.2 ‘Brotherhood’, a novel soybean model for the control of 100-seed weight, and has been retained after gene duplication to ensure genetic robustness Gene duplication is a key driver of genome evolution (Lan and Pritchard, 2016), and WGD events can lead to typical polyploidy (Comai 2005). All angiosperms appear to share two rounds of ancient whole-genome duplication, one that occurred in the common ancestor of extant seed plants and the other in the common ancestor of modern angiosperms (Jiao et al. 2011). Soybean has also undergone two additional WGD events, one approximately 59 MYA and the other between 5 and 13 MYA (Schmutz et al . 2010). Approximately 75% of soybean genes are present in multiple copies in the genome (Schmutz et al. 2010). Owing to functional redundancy, one member of a duplicated gene pair often loses function or undergoes differentiation after a WGD event. Duplicated genes have at least four evolutionary fates: pseudogenization, neofunctionalization, subfunctionalization, or sharing of the ancestral function by both copies (Force et al. 1999; Panchy et al. 2016; Fang et al. 2023). After WGDs and small-scale duplications, dosage balance can have different effects on the evolutionary dynamics of subfunctionalization and non-functionalization (Wilson et al. 2023). Here, the paralogs GmFBXL87a/b displayed highly similar structural features and tissue-specific expression patterns, and have undergone purifying selection . Each gene can independently reduce 100-seed weight by inhibiting cell expansion, while their coordinated expression further enhances this suppression. Interestingly, GmFBXL87a/b had different effects on the expression of several genes related to cell growth and seed size. Although their effects on GRF5 , GmGA20OX , GmPPC2-1 , and GmST05 expression were similar, GmFBXL87a suppressed expression of TOR and ARF2 , whereas GmFBXL87b promoted their expression, suggesting that GmFBXL87a/b may balance each other’s effects on TOR and ARF2 expression. Thus, we suggest that GmFBXL87a/b Likely arose from a common ancestor and emerged after duplication of the soybean genome. They were probably retained through dosage balance or impeded subfunctionalization and were subsequently subjected to purifying selection. These genes function independently while also collaboratively regulating 100-seed weight. This model of functional redundancy is revealed for the first time in the regulation of 100-seed weight in soybean. The functional dissection of GmFBXL87a/b paralogs not only reveals fundamental issues related to gene duplication in polyploid crops but also provides assurance for the genetic robustness of 100-seed weight in soybean. 4.3 GmFBXL87a/b may influence seed regulatory pathways through formation of SCF complexes As FBXL proteins, GmFBXL87a/b are likely to bind to different SKP1 proteins, regulating seed weight through the formation and activity of SCF complexes. Indeed, the Arabidopsis homolog of GmFBXL87a/b, AT1G67190, was shown to interact with 8 out of 20 tested SKP-like proteins in yeast two-hybrid assays (Kuroda et al. 2012). Soybean itself contains three SKP1 homologs ( Glyma.01G163900 , Glyma.16G123700 , and Glyma.11G079600 ), and the F-box protein GmFBL144 interacts with Glyma.11G079600 to form SCF complexes that enhance drought tolerance by promoting HSP70 degradation (Xu et al. 2022). FBXL proteins in other plant species play crucial roles in hormone signaling pathways (e.g., TIL1 in auxin signaling (Dharmasiri et al. 2005) and COI1 in the jasmonate pathway (Thines et al. 2007). Recently, overexpression of another soybean FBXL gene, GmFBXL12 , was shown to alter seed weight in ‘Tianlong No. 1’ soybean, although the mechanism underlying this effect is not yet clear (Hina et al. 2024). These results suggest that GmFBXL87a/b might function similarly, and our subsequent work will focus on determining whether GmFBXL87a/b regulates cell development and 100-seed weight in soybean by forming SCF complexes through binding to the same or different SKP1 proteins. Here, we preliminarily found that GmFBXL87a/b can co-regulate the expression of genes in cell-growth and seed-development pathways (e.g., GRF5 , TOR , ARF2 , GmGA20OX , GmPPC2-1 , and GmST05 ), suggesting that GmFBXL87a/b control 100-seed weight, at least in part, through participation in these pathways. 4.4 The larger 100-seed weight haplotype combination of GmFBXL87a/b have significant breeding potential Understanding the genetic mechanisms of domestication is crucial for addressing the current demand for high soybean yields (Lu et al. 2020). Most domestication-related traits are controlled by one or two major QTLs, along with several genotype-dependent minor QTLs (Liu et al. 2007). Yield traits in soybean, specifically seed traits, appear to be regulated by multiple QTLs/genes (Lu et al. 2017; Wang et al. 2020; Nguyen et al. 2021; Duan et al. 2022; Li et al. 2022; Cai et al. 2023). 100-seed weight is one such critical domestication trait that directly determines soybean yield. Here, population genetic analyses suggested that the seed-weight-related genes GmFBXL87a and GmFBXL87b were targets of selection during soybean domestication. The origin of the GmFBXL87a/b genes cannot be confirmed, but they likely originated from a common ancestor during soybean WGD event and were retained throughout the domestication process; their superior haplotypes for 100-seed weight were then selected during soybean improvement, becoming the predominant haplotypes in current landraces and improved soybean varieties. Analysis of their geographic distributions indicated that superior GmFBXL87a/b haplotypes for seed weight have been widely selected in the NER, NR, HR, and SR regions of China. Soybean germplasms carrying the best haplotypes of both paralogs (Group 1) had higher 100-seed weights than those with other haplotype combinations, but the frequency of such germplasm did not exceed 51% in any of the four regions, indicating that Group 1 haplotypes have significant breeding potential. Unfortunately, GmFBXL87a/b do not appear to affect individual plant yield. Therefore, future efforts should focus on incorporating Group 1 germplasm, which has the largest seeds, into breeding programs for high-yielding soybean in combination with superior genes associated with high seed numbers or other yield traits ( Table S 4). Acknowledgements This work was funded by the Natural Science Foundation of China (Grant numbers:32272072), Natural Science Foundation of Heilongjiang Province of China (ZL2024C007), the National Key R&D Program of China (2023ZD0403201-03), the National Natural Science Foundation of China (32472108, 32201755), the China Postdoctoral Science Foundation (2023MD744203), the Heilongjiang Postdoctoral Science Foundation (LBH-Z23011), and the China Agriculture Research System (CARS-04-PS14). We would like to thank A&L Scientific Editing (www.alpublish.com) for its linguistic assistance during the preparation of this manuscript. Conflicts of interest The authors declare no conflict of interest. Data availability statement All experimental data generated in this study are included in the main manuscript or supplementary materials. All plant materials, including transgenic lines and mutant stocks, are available for non-commercial research purposes upon request, subject to completion of a standard Material Transfer Agreement (MTA). Researchers may contact the corresponding author for material requests. Author contributions Q.S.C., Z.M.Q., and H.W.J. conceived and designed the project. S.M.W., R.C.K., J.Y.H. and X.Y.L performed all experiments together. J.Y.H. made important contributions to population construction, field sampling, and genetic analysis. X.Y.L., X.H. and J.Y.H undertook part of the work in the management and molecular experiments of genetically modified soybean materials. M.L.Y., F.B.C., S.Z.L. and J.G.X. mainly undertook the field management of the population. Q.S.C., Z.M.Q, H.W.J. and S.M.W. mainly put forward valuable opinions on the subject design. S.M.W. and C.R.K. analyzed the data, and S.M.W. wrote the manuscript. All authors have read and approved the manuscript. Supporting information Supplemental Figures Supplementary Figures 1–18 are presented in one PDF file Figure S1 Seed weight per plant of SN14, R158 and R86 Figure S2 Cotyledon size, cell number, and cell area of SN14, R158, R86 and ZYD06 seeds at three developmental stages Figure S3 Distribution of substituted segments on the genome of R158 and R86 Figure S4 Expression of candidate genes for control of 100-seed weight as measured by RNA-seq data from SoyBase database Figure S5 Sequence alignment of GmFBXL87a/b in SN14, R158, R86 and ZYD06 Figure S6 The amino acid sequence alignment of GmFBXL87a/b Figure S7 Subcellular localization of GmFBXL87a and GmFBXL87b Figure S8 Alignment of the genomic sequences of GmFBXL87a/b in SN14, R158 or R86, ZYD06, and DN50 Figure S9 Creation of soybean mutants for GmFBXL87a and GmFBXL87b by CRISPR-Cas9 gene editing Figure S10 Phenotype statistical analysis of yield-related traits for GmFBXL87a/b transgenic soybean mutants Figure S11 Phenotypic analysis of 1000-seed weight in Arabidopsis mutant plants atfbxl87 T 3 and T 4 generations Figure S12 Expression of GmFBXL87a/b in seeds of their respective mutants Figure S13 Plant phenotypes of yield-related traits for transgenic soybean overexpressing GmFBXL87a Figure S14 Seed size, seed width, and expression of GmFBXL87a during seed development in SN14, R158, and ZYD06, together with K A /K S analysis of GmFBXL87a/b Figure S15 Selection analysis of the GmFBXL87a on chromosome 12 Figure S16 Selection analysis of the GmFBXL87b on chromosome 11 Figure S17 Haplotype analysis of GmFBXL87a Figure S18 Haplotype analysis of GmFBXL87b Supplemental Tables Supplemental Tables 1–5 are presented in one EXCEL file. Table S1. Functional annotation of all genes in the QTL interval Table S2. Alignment of Glyma.11G183600 upstream 2 kb genome between SN14 and ZYD06 Table S3. Primers for published genes associated with cell-growth pathways and seed size in soybean Table S4. Haplotype combination of GmFBXL87a/b in 2898 germplasm resources Table S5. Primers used in this work References Abd-Hamid, N. A., M. I.Ahmad-Fauz, Z. Zainal, and I. Ismail. 2020. “Diverse and dynamic roles of F-box proteins in plant biology.” Planta 251 : 68. Bai, M., C. Yuan, H. Kuang, et al. 2022. “Combination of two multiplex genome-edited soybean varieties enables customization of protein functional properties.” Mol. Plant 15 : 1081–1083. Bailon-Zambrano, R., J. Sucharov, A. Mumme-Monheit, et al. 2022. “Variable paralog expression underlies phenotype variation.” eLife 11 : e79247. Cai, Z., P. Xian, Y. Cheng, et al. 2023. “MOTHER-OF-FT-AND-TFL1 regulates the seed oil and protein content in soybean.” New Phytol. 239 : 905–919. Chen, Y., Y. Xu, W. Luo, et al. 2013. “The F-box protein OsFBK12 targets OsSAMS1 for degradation and affects pleiotropic phenotypes, including leaf senescence, in rice.” Plant Physiol. 163 : 1673–1685. Clement, M., D. Posada, and K. A. Crandal. 2000. “TCS: a computer program to estimate gene genealogies.” Mol. Ecol. 9 : 1657–1659. Comai, L. 2005. “The advantages and disadvantages of being polyploid.” Nat. Rev. Genet. 6 : 836–846. Dai, Y., L. Luo, and Z. Zhao. 2023. “Genetic robustness control of auxin output in priming organ initiation.” Proc. Natl. Acad. Sci. U. S. A. 120 : e2221606120. Danecek, P., A. Auton, G. Abecasis, et al. 2011. “The variant call format and VCFtools.” Bioinformatics 27 : 2156–2158. Dharmasiri, N., S. Dharmasiri, and M. Estelle. 2005. “The F-box protein TIR1 is an auxin receptor.” Nature 435 : 441–445. Diss, G., D. Ascencio, A. DeLuna, and C. R. Landry. 2014. “Molecular mechanisms of paralogous compensation and the robustness of cellular networks.” J Exp. Zool. B. Mol. Dev. Evol. 322 : 488–499. Duan, Z., M. Zhang, Z. Zhang, et al. 2022. “Natural allelic variation of GmST05 controlling seed size and quality in soybean.” Plant Biotechnol. J. 20 : 1807–1818. Edgar, R. C. 2004. “MUSCLE: a multiple sequence alignment method with reduced time and space complexity.” BMC Bioinformatics 19 : 113. Eshed, Y., and D. Zamir. 1995. “An introgression line population of Lycopersicon pennellii in the cultivated tomato enables the identification and fine mapping of yield-associated QTL.” Genetics 141 :1147–1162. Fang, C., M. Yang, Y. Tang, et al. 2023. “Dynamics of of cis -regulatory sequences and transcriptional divergence of duplicated genes in soybean.” Proc. Natl. Acad. Sci. U. S. A. 120 : e2303836120. Félix, M-A., and A. Wagner. 2008. “Robustness and evolution: concepts, insights and challenges from a developmental model system.” Heredity (Edinb) 100 : 132–140. Force, A., M. Lynch, F. B. Pickett, A. Amores, Y. L. Yan, and J. Postlethwait. 1999. “Preservation of duplicate genes by complementary, degenerative mutations.” Genetics 151 : 1531–1545. Galensa, K. 2017. “ggplot2: elegant graphics for data analysis (2nd ed.).” Computing Reviews 58 : 457–458. Gao H. H., Y. Wang, W. Li, et al. 2018. “Transcriptomic comparison reveals genetic variation potentially underlying seed developmental evolution of soybeans.” J. Exp. Bot. 69 : 5089–5104. Gu, Y. Z., W. Li, H. W. Jiang, et al. 2017. “Differential expression of a WRKY gene between wild and cultivated soybeans correlates to seed size.” J. Exp. Bot. 68 : 2717–2729. Hina, A., N.Khan, K. Kong, et al. 2024. “Exploring the role of FBXL gene family in soybean: implications for plant height and seed size regulation.” Physiol. Plant 176 : e14191. Hu, R. B., C. M. Fan, H. Y. Li, Q. Zhang, and Y. F. Fu. 2009. “Evaluation of putative reference genes for gene expression normalization in soybean by quantitative real-time RT-PCR.” BMC Mol. Biol. 10 : 93–104. Hua, Z., C. Zou, S. H Shiu, and R. D.Vierstra. 2011. “Phylogenetic compari son of F-Box ( FBX ) gene superfamily within the plant kingdom reveals divergent evolutionary histories indicative of genomic drift.” PLoS ONE 6 : e16219. Jiao, Y., N. J. Wickett, S. Ayyampalayam, et al. 2011. “Ancestral polyploidy in seed plants and angiosperms.” Nature 473 : 97–100. Jin, J., T. Cardozo, R. C. Lovering, S. J. Elledge, M. Pagano, and J. W. Harper. 2004. “Systematic analysis and nomenclature of mammalian F-box proteins.” Genes Dev. 18 : 2573–2580. Karikari, B., Z. L. Wang, Y. L. Zhou, W. L. Yan, J. Y. Feng, and T. J. Zhao. 2020. “Identification of quantitative trait nucleotides and candidate genes for soybean seed weight by multiple models of genome-wide association study.” BMC Plant Biol. 20 : 404. Kassambara, A. 2017. “ ’ggplot2’ Based Publication Ready Plots [R package ggpubr version 0.1.0].” Kato, S., T. Sayama, K. Fujii, et al. 2014. “A major and stable QTL associated with seed weight in soybean across multiple environments and genetic backgrounds.” Theor. Appl. Genet. 127 : 1365–1374. Kuroda, H., Y. Yanagawa, N. Takahashi, Y. Horii, and M. Matsui. 2012. “A comprehensive analysis of interaction and localization of Arabidopsis SKP1-like (ASK) and F-box (FBX) proteins.” PLoS One 7 : e50009. Lan, X., and J. K. Pritchard. 2016. “Coregulation of tandem duplicate genes slows evolution of subfunctionalization in mammals.” Science 352 : 1009–1013. Lechner, E., P.Achard, A. Vansiri, T. Potuschak, and P. Genschik. 2006. “F-box proteins everywhere.” Curr. Opin. Plant Biol. 9 : 631–638. Li, C. H., X. Wu, Y. X. Li, et al. 2019a. “Genetic architecture of phenotypic means and plasticities of kernel size and weight in maize.” Theor. Appl. Genet. 132 : 3309–3320. Li, J., Y. Zhang, R. Ma, et al. 2022. “Identification of ST1 reveals a selection involving hitchhiking of seed morphology and oil content during soybean domestication.” Plant Biotechnol. J. 20 : 1110–1121. Li, N., R. Xu, and Y. Li. 2019b. “Molecular networks of seed size control in plants.” Annu. Rev. Plant Biol . 70 : 435–463. Li, Q., Fan, C. M., X. M. Zhang, and Y. F. Fu. 2012. “Validation of reference genes for real-time quantitative PCR normalization in soybean developmental and germinating seeds.” Plant Cell Rep . 31 : 1799–1799. Libault, M., A. Farmer, T. Joshi, et al. 2010. “An integrated transcriptome atlas of the crop model Glycine max , and its use in comparative analyses in plants.” Plant J. 63 : 86–99. Librado, P., and J. Rozas. 2009. “DnaSP v5: a software for comprehensive analysis of DNA polymorphism data.” Bioinformatics 25 : 1451–1452. Liu, B.H., T. Fujita, Z. H. Yan, S. Sakamoto, D. H. Xu, and J. Abe. 2007. “QTL mapping of domestication-related traits in soybean ( Glycine max ).” Ann. Bot. 100 : 1027–1038. Liu, G. M., W. T. Li, R. Z. Zeng, Z. M. Zhang, and G. Q. Zhang. 2004. “Identification of QTLs on substituted segments in single segment substitution lines of rice.” Yi Chuan Xue Bao 31 : 1395–1400. Liu, Y. L., Y. H. Li, J. C. Reif, et al. 2013. “Identification of quantitative trait loci underlying plant height and seed weight in soybean.” Plant Genome 6 : 841–856. Liu, Y., H. Du, P. Li, et al. 2020. “Pan-genome of wild and cultivated soybeans.” Cell 182 : 162–176.e13. Livak, K. J., and T. D. Schmittgen. 2001. “Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T))” Method. Methods 25 : 402–408. Lu, S., L. Dong, C. Fang, et al. 2020. “Stepwise selection on homeologous PRR genes controlling flowering and maturity during soybean domestication.” Nat. Genet. 52 : 428–436. Lu, X., Q. Xiong, T. Cheng, et al. 2017. “A PP2C-1 allele underlying a quantitative trait locus enhances soybean 100-seed weight.” Mol. Plant 10 : 670–684. Lynch, M., and J. S. Conery. 2000. “The evolutionary fate and consequences of duplicate genes.” Science 290 : 1151–1155. Magori, S., and V. Citovsky. 2011. “Hijacking of the host SCF ubiquitin ligase machinery by plant pathogens.” Front. Plant Sci. 2 : 87. Nguyen, C. X., K. J. Paddock, Z. Zhang, and M. G. Stacey. 2021. “ GmKIX8-1 regulates organ size in soybean and is the causative gene for the major seed weight QTL qSw17-1 .” New Phytol. 229 : 920–934. Nguyen, L. T., H. A. Schmidt, H. A. Von, and B. Q. Minh. 2015. “IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies.” Mol. Biol. Evol. 32 : 268–274. Panchy, N., M. Lehti-Shiu, and S. H. Shiu. 2016. “Evolution of gene duplication in plants.” Plant Physiol. 171 : 2294–2316. Patikoglou, G. A., J. L. Kim, L. Sun, S. H. Yang, T. Kodadek, and S. K. Burley. 1999. “TATA element recognition by the TATA box-binding protein has been conserved throughout evolution.” Genes Dev. 13 : 3217–3230. Pillen, K., A. Zacharias, and J. Léon. 2003. “Advanced backcross QTL analysis in barley ( Hordeum vulgare L.).” Theor. Appl. Genet. 107 : 340–352. Purcell, S., B. Neale, K. Todd-Brown, L. Thomas, et al. 2007. “PLINK: a tool set for whole-genome association and population-based linkage analyses.” Am. J. Hum. Genet. 81 :559–575. Schmutz, J., S. B. Cannon, J. Schlueter, et al. 2010. “Genome sequence of the palaeopolyploid soybean.” Nature 463 : 178–183. Shen, G., and Y. Xing. 2014. “Two novel QTLs for heading date are identified using a set of chromosome segment substitution lines in rice ( Oryza sativa L.).” J. Genet Genomics 41 : 659–662. Sreedasyam, A., C. Plott, M. S. Hossain, et al. 2023. “JGI Plant Gene Atlas: an updateable transcriptome resource to improve functional gene descriptions across the plant kingdom.” Nucleic Acids Res. 51 : 8383–8401. Stefanowicz, K., N. Lannoo, and E. J. M. Van Damme. 2015. “Plant F-box proteins-judges between life and death.” Crit Rev. Plant Sci. 34 : 523–552. Suzuki, I. K., D. Gacquer, R. Van Heurck, et al. 2018. “Human-specific NOTCH2NL genes expand cortical neurogenesis through Delta/Notch regulation.” Cell 173 : 1370–1384.e16. Team, R. C. 2020. “R: A language and environment for statistical computing.” R Foundation for Statistical Computing. Thines, B., L. Katsir, M. Melotto, et al. 2007. “JAZ repressor proteins are targets of the SCF(COI1) complex during jasmonate signalling.” Nature 448 : 661–665. Thomson, M. J., J. D Edwards, E. M. Septiningsih, S. E Harrington, and S. R. McCouch. 2006. “Substitution mapping of dth1.1 , a flowering-time quantitative trait locus (QTL) associated with transgressive variation in rice, reveals multiple sub-QTL.” Genetics 172 : 2501–2514. Wang, L., G. Jia, X. Jiang, S. Cao, Z. J. Chen, and Q. Song. 2021. “Altered chromatin architecture and gene expression during polyploidization and domestication of soybean.” Plant Cell 33 : 1430–1446. Wang, S. D., Liu, S. L., J. Wang, et al. 2020. “Simultaneous changes in seed size, oil content and protein content driven by selection of SWEET homologues during soybean domestication.” Natl. Sci. Rev. 7 : 1776–1786. Wei, S., B. Yong, H. Jiang, et al. 2023. “A loss-of-function mutant allele of a glycosyl hydrolase gene has been co-opted for seed weight control during soybean domestication.” J. Integr. Plant Biol. 65 : 2469–2489. Wilson, A. E., and D. A. Liberles. 2023. “Dosage balance acts as a time-dependent selective barrier to subfunctionalization.” BMC Ecol. Evol. 23 : 14. Xin, D., Z. Qi, H. Jiang, et al. 2016. “QTL location and epistatic effect analysis of 100-seed weight using wild soybean ( Glycine soja Sieb. & Zucc.) chromosome segment substitution lines.” PLoS One 11 : e0149380. Xu, G., H. Ma, M. Nei, and H. Kong. 2009. “Evolution of F-box genes in plants: different modes of sequence divergence and their relationships with functional diversification.” Proc. Natl. Acad. Sci. U. S. A. 106 : 835–840. Xu, K., Y. Zhao, Y. Zhao, et al. 2022. “Soybean F-Box-Like protein GmFBL144 interacts with small heat shock protein and negatively regulates plant drought stress tolerance.” Front. Plant Sci. 13 : 823529. Yang, H. Y., W. B. Wang, Q. Y. He, S. H. Xiang, and J. Y. Gai. 2019. “Identifying a wild allele conferring small seed size, high protein content and low oil content using chromosome segment substitution lines in soybean.” Theor. Appl. Genet. 132 : 2793–2807. Yang, X., U. C. Kalluri, S. Jawdy, et al. 2008. “The F-box gene family is expanded in herbaceous annual plants relative to woody perennial plants.” Plant Physiol. 148 : 1189–1200. Yano, M., S. Kojima, Y. Takahashi, H. Lin, and T. Sasaki. 2001. “Genetic control of flowering time in rice, a short-day plant.” Plant Physiol. 127 : 1425–1429. Yin, P., Q. Ma, H. Wang, et al. 2020. “SMALL LEAF AND BUSHY1 controls organ size and lateral branching by modulating the stability of BIG SEEDS1 in Medicago truncatula .” New Phytol. 226 : 1399–1412. Yong, B., J. Balarynová, B. Li, et al. 2024. “Paralogous gene recruitment in multiple families constitutes genetic architecture and robustness of pod dehiscence in legumes.” Genome Biol. Evol. 16 : evae267. Zeng, P., D. A.Vadnais, Z. Zhang, and J. C. Polacco. 2004. “Refined glufosinate selection in Agrobacterium -mediated transformation of soybean [ Glycine max (L.) Merrill].” Plant Cell Rep. 22 : 478–482. Zhang, J. P., X. Z. Wang, Y. Lu, et al. 2018. “Genome-wide scan for seed composition provides insights into soybean quality improvement and the impacts of domestication and breeding.” Mol. Plant 11 : 460–472. Zhang, X., Z. H. Gonzalez-Carranza, S. Zhang, Y. C Miao, C. J. Liu, and J. A. Roberts. 2019. “F-Box proteins in plants.” ̵‌ Annu. Rev. Plant Biol. 2 : 307–328. Zhang, Y., J. He, Y. Wang, et al. 2015. “Establishment of a 100-seed weight quantitative trait locus-allele matrix of the germplasm population for optimal recombination design in soybean breeding programmes.” J Exp Bot. 66 : 6311–6325. Zhang, Y., M. F. Iqbal, Y. Wang, et al. 2022. “OsTBP2.1, a TATA-binding protein, alters the ratio of OsNRT2.3b to OsNRT2.3a and improves rice grain yield.” Int J Mol Sci. 23 : 10795. Zheng, H., L. Hou, J. Xie, et al. 2022. “Construction of chromosome segment substitution lines and inheritance of seed-pod characteristics in wild soybean.” Front. Plant Sci. 13 : 869455. Zhou, Z.K., Y. Jiang, Z. Wang, et al. 2015. “Resequencing 302 wild and cultivated accessions identifies genes related to domestication and improvement in soybean.” Nat. Biotechnol. 33 : 408–414. Zhu, D., X. Li, Z. Wang, et al. 2020. “Genetic dissection of an allotetraploid interspecific CSSLs guides interspecific genetics and breeding in cotton.” BMC Genomics 21 : 431. Figure legends FIGURE 1 Seed-related traits and cell morphology of the small-seeded CSSL R158, R86 and its parents (A) Seed-related traits of the small-seeded CSSL R158 and R86, its receptor parent SN14 (cultivated soybean), and its donor parent ZYD06 (wild soybean). Bar = 1 cm. (B–D) Seed-size-related traits in SN14, R158, R86 and ZYD06. (B) 100-seed weight, (C) seed length, and (D) seed width. (E) Area of the ventral cotyledon surface in SN14, R158, R86 and ZYD06. Bar = 5 mm. Data were derived from three biological replicates (5 seeds per replicate). Different lowercase letters denote statistically significant differences ( p < 0.05, one-way ANOVA with Tukey’s HSD test). (F–H) High-magnification SEM images showing cell morphology on the ventral cotyledon surface of mature dry seeds from SN14 (F), R158 (G), R86 (H) and ZYD06 (I). Bar = 50 μm. (J, K) Mean cell area (μm²) (J) and total cell number on the cotyledon ventral surface (K) of R158, R86 and its parents. Values are mean ± SD ( n > 100 cells per genotype). Different lowercase letters denote statistically significant differences ( p < 0.05, one-way ANOVA with Tukey’s HSD test). FIGURE 2 Fine-mapping and characterization of the QTL qHSW12-2 for 100-seed weight (A, B) Process of QTL mapping: (A) preliminary mapping and (B) fine-mapping and candidate-gene analysis. qHSW12-2 was ultimately mapped to a 160-kb interval, and its candidate causal gene was identified as Glyma . 12G088900 . n , numbers of F 2 and RHL lines. (C) Genomic sequence comparison of the Glyma.12G088900 candidate gene in SN14 and ZYD06. (D) 100-seed weights of plants with different Glyma.12G088900 genotypes in the RHL population. Genotyping was performed using a SNP marker located at −458 in Glyma.12G088900 that differentiated between SN14 and ZYD06. Violin plots show the 100-seed weights for each genotype. Glyma.12G088900 SS , homozygous SN14 genotype; Glyma.12G088900 SZ , heterozygous genotype; Glyma.12G088900 ZZ , ZYD06 homozygous genotype. The box edges represent the interquartile ranges, the center line is the median, and the individual dots are outliers. Different lowercase letters denote statistically significant differences ( p < 0.05, one-way ANOVA with Tukey’s HSD test). (E) Schematic diagrams of Glyma.12G088900 promoter constructs used to measure promoter activity in a tobacco transient expression assay. Glyma.12G088900 SN14pro , upstream 2000-bp promoter region of Glyma.12G088900 from SN14. Glyma.12G088900 ZYD06pro , upstream 2000-bp promoter region of Glyma.12G088900 from ZYD06. (F) Luciferase activity in tobacco leaves transiently transformed with the constructs shown in (E). (G) Quantitative analysis of the luminescence intensity shown in (F). Values are mean ± SD ( n = 3). Significance was analyzed using a two-tailed Student’s t -test (*** p < 0.001). FIGURE 3 Fine-mapping process of the 100-seed weight QTL, qHSW11-5 (A) Preliminary mapping and (B) Fine-mapping and candidate-gene analysis. qHSW11-5 was ultimately mapped to a 300-kb interval, and Glyma.11G183600 was pinpointed as its potential causal candidate gene. n, numbers of F 2 and RHL lines. (C) Alignment of the genomic sequence of the candidate gene Glyma.11G0183600 between the two parental lines (SN14 and ZYD06), highlighting sequence polymorphisms. (D) Variations in 100-seed weight among RHL individuals carrying different genotypes of Glyma.11G0183600 . Genotyping was conducted using a SNP marker positioned at the −503 bp locus of Glyma.11G0183600 , a site that distinguishes SN14 from ZYD06. Violin plots illustrate the distribution of 100-seed weight for each genotype: Glyma.11G0183600 SS (homozygous for the SN14 allele), Glyma.11G0183600 SZ (heterozygous genotype), and Glyma.11G0183600 ZZ (homozygous for the ZYD06 allele). The box edges represent the interquartile ranges, the center line is the median, and the individual dots are outliers. Different lowercase letters denote statistically significant differences ( p < 0.05, one-way ANOVA with Tukey’s HSD test). (E) Schematic diagrams of Glyma.11G0183600 promoter constructs used to measure promoter activity in a tobacco transient expression assay. Glyma.11G0183600 SN14pro , upstream 2000-bp promoter region of Glyma.11G0183600 from SN14. Glyma.11G0183600 ZYD06pro , upstream 2000-bp promoter region of Glyma.11G0183600 from ZYD06. (F) Luciferase activity in tobacco leaves transiently transformed with the constructs shown in (E). (G) Quantitative analysis of the luminescence intensity shown in (F). Values are mean ± SD ( n = 3). Significance was analyzed using a two-tailed Student’s t -test (*** p < 0.001). FIGURE 4 Phylogenetic analysis and collinearity analysis of GmFBXL87a and GmFBXL87b (A) The phylogenetic tree constructed by FBXL genes. GmFBXL87a and GmFBXL87b its homologous genes are highlighted in red font. (B) The collinearity analysis of GmFBXL87a and GmFBXL87b ( GmFBXL87a/b ) in soybean. (C) Alignment of amino acid sequence of GmFBXL87a/b. The genomic sequences of GmFBXL87a/b were cloned from SN14 and ZYD and translated into amino acid sequences using DNAMAN software. (D) Structural prediction and comparison of GmFBXL87a/b. The results of the structural prediction are derived from AlphaFold 3 (https://alphafoldserver.com). (E) Tissue expression analysis of GmFBXL87a/b . Three biological replicates were used for each sample. The mean and SD are presented. Different lowercase letters (A–G) indicate significant differences, and identical letters suggest insignificant differences in one-way ANOVA. FIGURE 5 Phenotypic variation in seed-related traits of GmFBXL87a/b transgenic soybean mutants (A) Seed morphology of crfbxl87a , crfbxl87b , and crfbxl87a/b T 2 mutants and DN50. Bar = 1 cm. (B–D) Seed-related traits of crfbxl87a , crfbxl87b , and crfbxl87a/b T 2 mutants and DN50: (B) 100-seed weight, (C) seed length, and (D) seed width. (E) Seed morphology of six OEGmFBXL87a T 2 lines and DN50. Bar = 1 cm. (F–H) Seed-related traits of OEGmFBXL87a T 2 lines and DN50: (F) 100-seed weight, (G) seed length, and (H) seed width. In (B–D) and (F–H), the percentages indicate the relative increase or decrease in the transgenic lines relative to DN50. Significance was analyzed by two-tailed Student’s t -tests (comparing each transgenic line to DN50; * p < 0.05, ** p < 0.01, and *** p < 0.001). FIGURE 6 GmFBXL87a/b influence seed size by restricting cell expansion in soybean (A) Cotyledon cells of GmFBXL87a/b transgenic soybean lines and DN50 at the Cot, EM1, and EM2 stages. (B–G) Cell number and average cell area of cotyledons from GmFBXL87a/b transgenic soybean lines and DN50 at the Cot, EM1, and EM2 stages: (b) cell number at Cot, (C) cell area at Cot, (D) cell number at EM1, (E) cell area at EM1, (F) cell number at EM2, and (G) cell area at EM2. (H) Ventral cotyledon surfaces of mature dry seeds from GmFBXL87a/b mutant and overexpression lines . Bar = 1 mm. Data were derived from three biological replicates (5 seeds per replicate). (I–M) High-magnification SEM views showing cell morphology on the ventral cotyledon surface of mature dry seeds from GmFBXL87a/b mutant and overexpression lines: (I) DN50, (J) crfbxl87a -1, (K) crfbxl87b -1, (L) crfbxl87a/b -1, and (M) OEGmFBXL87a . Bars = 50 μm. (N, O) Mean cell area (μm²) (N) and total cell number on the cotyledon ventral surface (O) of seeds from GmFBXL87a/b mutant and overexpression lines. Values represent mean ± SD ( n > 100 cells per genotype). The percentages indicate the relative increase or decrease in the transgenic lines relative to DN50. Significance was analyzed using two-tailed Student’s t -tests of transgenic lines versus DN50 (* p < 0.05, ** p < 0.01 and *** p < 0.001). FIGURE 7 Potential regulatory pathways and working model for the control of 100-seed weight by GmFBXL87a/b (A) Expression of genes related to cell growth and seed size in GmFBXL87a/b transgenic soybean lines and DN50 at the EM2 stage. Data were obtained from three replicates per genotype. (* p < 0.05; ** p <0.01; *** p < 0.001; two-tailed Student’s t -tests comparing transgenic materials to DN50). (B) The ‘Brotherhood’ model for the control of seed size by GmFBXL87a/b . Expression of GmFBXL87a/b was lower in R158 and SN14 than in ZYD06 , resulting in greater cell area and increased seed size. Cell area and s eed size were also greater in the crfbxl87a , crfbxl87b, and crfbxl87a/b mutants than in DN50, and seeds of crfbxl87a/b were larger than those of the single-mutant lines. (C) Putative integration of GmFBXL87a/b into the molecular pathways that control cell growth and seed size to coordinate cell development and seed size. FIGURE 8 Haplotype combinations of GmFBXL87a/b in 2898 soybean accessions (A) Percentages of GmFBXL87a/b haplotype combinations in landraces and improved soybeans. (B) Distribution of different GmFBXL87a/b haplotype combination types. Dark orange circles indicate haplotype combinations with the largest 100-seed weight (Hap15 a or Hap8 b ). Light orange circles indicate haplotype combinations with relatively large 100-seed weight (Hap17 a or Hap6 b ). Light green circles indicate haplotype combinations with relatively small 100-seed weight (Hap16 a or Hap1 b ). Dark green circles indicate haplotype combinations with smallest 100-seed weight (Hap7 a or Hap2 b ). The superscripts “a” and “b” denote haplotypes of GmFBXL87a and GmFBXL87b , respectively. The number of germplasm accessions carrying each haplotype combination (No.) is shown in the right-hand column. (C) The 100-seed weights of different haplotype combinations in (C). Haplotypes are divided into three groups on the basis of their 100-seed weights: Group1, (Hap15 a &Hap8 b , Hap15 a &Hap6 b , Hap17 a &Hap6 b , and Hap17 a &Hap8 b ), Group 2 (Hap16 a &Hap8 b , Hap16 a &Hap6 b , Hap16 a &Hap1 b , and Hap15 a &Hap1 b ), and Group 3 (Hap15 a &Hap2 b , Hap7 a &Hap8 b , and Hap7 a &Hap6 b ) (D, E) 100-seed weights of accessions from Groups1–3 in landraces (D) and improved soybeans (E). The box edges represent the interquartile ranges, the center line is the median, and the individual dots are outliers. (F) Proportions of Groups 1–3 in soybean germplasm resources from different regions of China. NER, Northeast region. NR, North region. HR, Huanghuai region. SR, South region. (G–J) 100-seed weights of accessions from Groups 1–3 in soybean accessions from four regions: (G) NER, (H) NR, (I) HR, and (J) SR. The box edges represent the interquartile ranges, and the center line is the median. In ( G–I), different lowercase letters denote statistically significant differences ( p < 0.05, one-way ANOVA with Tukey’s HSD test; n.s., not significant). In (J), * p < 0.05 (two-tailed Student’s t- test). Information & Authors Information Version history V1 Version 1 21 October 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords 100-seed weight development gene pyramiding genetic robustness genetic variation soybean Authors Affiliations Siming Wei 0000-0002-4733-9744 Northeast Agricultural University View all articles by this author Chaorui Kang Northeast Agricultural University View all articles by this author Jiayin Han Northeast Agricultural University View all articles by this author Xinyue Liu Northeast Agricultural University View all articles by this author Mingliang Yang Northeast Agricultural University View all articles by this author Fubin Cao Northeast Agricultural University View all articles by this author Jianguo Xie Jilin Academy of Agricultural Sciences View all articles by this author Shuangzhe Li 0009-0004-9337-5537 Northeast Agricultural University View all articles by this author Xue Han 0000-0002-0873-6456 Northeast Agricultural University View all articles by this author Hongwei Jiang Jilin Academy of Agricultural Sciences View all articles by this author Zhaoming Qi 0000-0002-0657-9127 Northeast Agricultural University View all articles by this author Qingshan Chen [email protected] Northeast Agricultural University View all articles by this author Metrics & Citations Metrics Article Usage 177 views 161 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Siming Wei, Chaorui Kang, Jiayin Han, et al. The paralogs GmFBXL87a/b independently and cooperatively regulate 100-seed weight for maintaining genetic robustness in soybean evolution. Authorea . 21 October 2025. DOI: https://doi.org/10.22541/au.176103210.07633391/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.176103210.07633391/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ffd6469ac341640',t:'MTc3OTQ2OTM4NQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();