Natural variation in Glume Coverage 1 causes naked grains in sorghum

Hulled grains of wild cereals protect seeds from pathogens and predation but hinder threshing in agriculture, especially under mechanized harvesting which can damage embryos and reduce grain quality. Naked grains with reduced glume coverage facilitate planting, threshing, and processing, and glume coverage has historically been used to classify sorghum subspecies. Despite its agronomic importance, the genetic basis of glume coverage in sorghum was unknown. Heterotrimeric G proteins (α, β, γ) act as molecular switches in plant signaling, and phospholipases are key regulators within G-protein cycles, influencing defense and growth. However, the relationship between Gγ subunits and phospholipases in plant development had not been established. The study aims to identify the gene(s) controlling glume coverage in sorghum, elucidate the molecular mechanism, and assess signatures of domestication related to naked grains.

Prior work shows G-proteins regulate diverse plant traits, including grain size in rice via atypical Gγ proteins (e.g., GS3, DEP1) acting with Gβ to modulate downstream transcription factors such as OsMADS1. Phospholipases, particularly PLAII, generate lipid-derived signaling molecules and are implicated in cell growth and defense. In cereals, domestication for reduced glume toughness or coverage has occurred (e.g., maize tga1, wheat Tg and Sog loci), and sorghum subspecies classifications use glume coverage; yet the specific gene controlling glume coverage in sorghum was not defined. The potential interaction between Gγ subunits and phospholipases in regulating cell proliferation had been suggested in animal systems but was untested in plants.

• Germplasm and phenotyping: Assessed glume coverage morphology in 915 diverse sorghum accessions; quantified glume coverage and glume dimensions. In a Sorghum Association Panel (SAP) of 352 inbred lines, scored glume coverage in five classes across multi-environment trials (Beijing, Yinchuan, Sanya; two replicates; 2016–2017). Threshing efficiency measured versus glume coverage classes.
• GWAS: Used 82,430 SNPs (CMLM model; Bonferroni threshold P<1e-6) aligned to BTx623 reference. Identified significant loci associated with glume coverage.
• Linkage mapping and fine mapping: Generated an F2 from SN010 (hulled) × M-81E (naked); mapped a major locus on chr1. Developed three residual heterozygous lines and advanced to F3–F5; screened 1678 plants total across generations with flanking markers. Narrowed the locus to ~58 kb using 40 recombinants; evaluated gene expression of candidates in panicles.
• Sequence and haplotype analysis: Sequenced GC1 coding region in 482 accessions; cataloged 257 SNPs and 103 indels; defined five haplotypes (one wild type, four truncations: gc1-a, gc1-b, gc1-c, gc1-d). Performed haplotype-based association and LD analyses for glume and yield-related traits.
• Near-isogenic lines (NILs): Constructed NIL-GC1 and NIL-gc1-a from the SN010 × M-81E cross to assess phenotypic effects under uniform backgrounds.
• Transgenics in sorghum: Generated Wheatland (wild-type GC1 background) lines overexpressing GC1-Myc (GC1-OE), overexpressing truncated gc1-Myc (gc1-OE, deletion of aa141–198), and CRISPR/Cas9 GC1 knockouts (GC1-KO). Validated by qPCR and sequencing.
• Cross-species validation in millet: Identified SiGC1 (78.3% identity). Made SiGC1124-Myc overexpression (truncated tail mimicking sorghum gc1) and SiGC1-KO lines in cultivar Ci846 via Agrobacterium-mediated transformation and CRISPR.
• Protein assays: Western blots of Myc- or GFP-tagged GC1/gc1 variants in panicles and N. benthamiana; cycloheximide (CHX) chase and proteasome inhibitor MG132 treatments to assess protein stability; quantification by band densitometry.
• Histology and in situ hybridization: Sectioned glumes at developmental stages; measured cell number and size; RNA in situ localization of GC1 transcripts in 3 cm panicles.
• Transcriptomics: RNA-seq of NIL panicles (0–3 cm, 3–10 cm, heading); identified DEGs (|log2FC|≥1, adjusted P<0.001); GO/KEGG enrichment focusing on cell cycle pathways; qPCR validation of Cyclin–CDK genes.
• Protein interaction and degradation module: Identified gc1 interactors by IP-MS from gc1-OE panicles; validated GC1/gc1–SbpPLAII-1 interactions via LCI, pull-down, co-IP, and BiFC; subcellular localization by confocal microscopy. Tested whether GC1/gc1 promote SbpPLAII-1 degradation in mixed protein extracts.
• Functional test of SbpPLAII-1: Overexpressed SbpPLAII-1 in millet; assessed glume morphology, cell proliferation, and Cyclin–CDK gene expression.
• Selection and geography: Mapped geographic distribution of GC1 haplotypes across 38 countries (482 accessions). Computed nucleotide diversity (π), π ratios, Tajima’s D, Fst around GC1 in wild vs naked landraces and improved lines to detect positive selection signatures.

• Trait distribution and correlation: Wild accessions had fully hulled grains; ~60% of cultivars exhibited low/very low glume coverage. Glume coverage strongly correlated with glume length (R²=0.98) and inversely with threshing efficiency.
• Genetic mapping: GWAS identified three major loci on chromosomes 1, 2, and 3, with a strong signal on chr1. Biparental mapping co-localized the major locus, fine-mapped to ~58 kb containing five genes; expression and association pinpointed Sobic.001g341700 (GC1) as the candidate.
• Causal variation: Within exon 5, four truncating variants were identified: gc1-a (GTGGC insertion), gc1-b (G deletion), gc1-c (C→A nonsense), gc1-d (165 bp insertion). Haplotype analysis (n=482) showed strongest association at +4158 (P=5.27E-14), with linked sites +4151 and +4285 (LD r²>0.9). No strong association with grain yield-related traits.
• Gene identity and conservation: GC1 encodes an atypical Gγ-like protein (198 aa) with N-terminal Gγ-like domain and C-terminal transmembrane region; orthologous to rice GS3 (50.86% similarity), with highly conserved N-terminus and variable C-terminus across grasses.
• Functional genetics in sorghum: NIL-gc1-a had reduced glume length, slightly reduced width, much lower coverage, and >60% increased threshing rate vs NIL-GC1. In Wheatland, GC1-OE mildly reduced glume coverage with ~13% threshing increase; GC1-KO caused long, hard-threshing glumes with full coverage; gc1-OE (C-terminal truncation) strongly reduced glume length and coverage with ~55% threshing rate increase.
• Cross-species validation: In millet, overexpressing truncated SiGC1124 reduced glume length and coverage; SiGC1-KO increased glume length by ~30%, mirroring sorghum results.
• Protein stability mechanism: Truncated gc1 variants accumulated 3.0–7.6× more than full-length GC1; CHX chases showed rapid loss of full-length but persistence of truncated proteins. MG132 stabilized GC1, indicating 26S proteasome-dependent degradation; truncated proteins were more stable and less sensitive to C-terminus-mediated proteolysis.
• Cellular basis: GC1 is highly expressed in early panicles (3–6 cm), localized in spikelet organs. NIL-gc1-a glumes had ~35% fewer cells in the longitudinal inner layer without change in cell size; reduced cell number evident from primordia stage onward.
• Transcriptional changes: RNA-seq revealed DEGs enriched in cell proliferation and kinase activity. Cyclin–CDK pathway genes (SbCYCA2;3, SbCYCB2;2, SbCDKB1;1) were downregulated in NIL-gc1-a; qPCR confirmed significant reductions in early panicles.
• Interaction module: IP-MS identified SbpPLAII-1 (patatin-like phospholipase AII-1) as a candidate interactor. GC1 and gc1-a physically interacted with SbpPLAII-1 (LCI, pull-down, co-IP, BiFC), co-localized at the plasma membrane. Overexpressing SbpPLAII-1 in millet increased glume cell number and upregulated Cyclin–CDK genes, indicating it positively promotes cell proliferation.
• Degradation of SbpPLAII-1: Increasing GC1/gc1-a protein amounts promoted degradation of SbpPLAII-1 in vitro/in vivo; truncated gc1-a enhanced SbpPLAII-1 degradation more than full-length GC1.
• Domestication and selection: Truncated gc1 alleles occurred in 31.8% of accessions (gc1-a in 26.2%) across 22 countries, with rare alleles (gc1-b/c/d) concentrated near the Sahelian zone, suggesting a domestication origin for naked sorghum there (notably Nigeria). Sequence analyses showed reduced π in exon 5 and 3' UTR, significantly negative Tajima’s D, and elevated Fst (max 0.34) between wild and naked cultivars, consistent with positive selection targeting GC1.

The study demonstrates that GC1, an atypical Gγ-like subunit, is a negative regulator of sorghum glume coverage and that natural C-terminal truncations stabilize the protein, intensify its function, and thereby reduce glume size by suppressing cell proliferation. This mechanistic role aligns with known Gβγ-mediated signaling pathways affecting grain traits in rice (e.g., GS3), yet here specifically impacts glume coverage rather than grain size, a key domestication trait in sorghum. The identification of SbpPLAII-1 as a positive regulator of glume cell proliferation integrates phospholipase signaling into a G-protein module in cereals; GC1/gc1-a promotes SbpPLAII-1 degradation, reducing Cyclin–CDK transcript levels and cell division. The presence and geographic enrichment of truncated GC1 alleles, combined with strong population genetic signatures, indicate these alleles were positively selected during domestication to produce naked grains that facilitate threshing and mechanization. Compared to rice, where GS3 has a large effect on grain size, GC1 appears to exert minor effects on yield traits in sorghum, implying that breeding for naked grains via GC1 truncations can be combined with other loci to maintain or improve yield.

This work identifies GC1 as the major gene controlling glume coverage in sorghum and elucidates a mechanism whereby natural C-terminal truncations stabilize GC1, enhance degradation of the positive cell-proliferation factor SbpPLAII-1, downregulate Cyclin–CDK genes, and reduce glume size, yielding naked grains with improved threshability. Cross-species validation in millet supports a conserved role of truncated Gγ-like subunits in glume development. Population genetics and geographic analyses reveal strong positive selection on truncated GC1 alleles, implicating them in sorghum domestication for naked grains. These findings provide actionable targets for breeding and gene editing to develop naked-grain sorghum varieties, potentially coupled with major yield loci (e.g., qGW1a) to offset any minor yield penalties. Future research should dissect the precise biochemical pathway linking Gβγ to SbpPLAII-1 turnover, identify E3 ligases or other components mediating C-terminus-dependent proteolysis, and generate double-mutant combinations to genetically validate the GC1–SbpPLAII-1 interaction in planta.

• Genetic interaction was inferred biochemically; double loss-of-function mutants for GC1 and SbpPLAII-1 were not available to confirm epistasis in planta.
• Mechanistic details of how GC1/gc1 promotes SbpPLAII-1 degradation (e.g., specific ubiquitination machinery) remain unresolved.
• Associations with grain yield-related traits were weak in the diversity panel; the broader agronomic impact of GC1 truncations on yield components under varied environments warrants further evaluation.