The RING-Type Domain-Containing Protein GNL44 Is Essential for Grain Size and Quality in Rice (Oryza sativa L.)

Grain size in rice (Oryza sativa L.) shapes yield and quality, but the underlying molecular mechanism is not fully understood. We functionally characterized GRAIN NUMBER AND LARGE GRAIN SIZE 44 (GNL44), encoding a RING-type protein that localizes to the cytoplasm. The gnl44 mutant has fewer but enlarged grains compared to the wild type. GNL44 is mainly expressed in panicles and developing grains. Grain chalkiness was higher in the gnl44 mutant than in the wild type, short-chain amylopectin content was lower, middle-chain amylopectin content was higher, and appearance quality was worse. The amylose content and gel consistency of gnl44 were lower, and protein content was higher compared to the wild type. Rapid Visco Analyzer results showed that the texture of cooked gnl44 rice changed, and that the taste value of gnl44 was lower, making the eating and cooking quality of gnl44 worse than that of the wild type. We used gnl44, qgl3, and gs3 monogenic and two-gene near-isogenic lines to study the effects of different combinations of genes affecting grain size on rice quality-related traits. Our results revealed additive effects for these three genes on grain quality. These findings enrich the genetic resources available for rice breeders.


Introduction
Rice (Oryza sativa L.) is an important food crop that provides the daily dietary calories for much of the human population; as such, its yield is crucial to ensuring food security.Grain weight, grain number per panicle, and panicle number per plant are the most important agronomic components of grain yield.Numerous genes regulating grain number have been identified in rice over the years [1][2][3][4][5].These genes function in various periods of spike development.Nevertheless, knowledge of the genetic and molecular mechanisms that establish grain number in rice is still limited.
In addition to yield, grain size can also affect the quality of rice.In general, large grains are associated with high yield but poor quality, especially the appearance quality of milled rice (grains with the outer hull removed), which includes chalkiness [25,26].Slender rice is popular with consumers for its transparent appearance and lack of chalkiness.The genotype at GW7 can improve rice quality by changing the pattern of cell division, making rice grains more slender and less chalky 18 .Although several grain size genes have been cloned in rice, the genetic and molecular mechanisms that determine grain size remain largely unknown [27].Functional identification of other grain size genes or QTLs is important to further our understanding of the molecular mechanisms regulating grain-related traits and to help meet the demand for high-yielding and superior-quality cultivars.
Although numerous QTLs/genes related to grain size have been functionally characterized, there have been few studies about the genetic interactions of these grain size-related genes [28].Gao et al. (2015) used the near-isogenic lines (NILs) NIL-GS3, NIL-qgl3, and NIL-GS3/qgl3 to study the interaction effects of GS3 and qGL3, finding that GS3 and qGL3 have an additive effect on regulating rice grain length [29].He et al. (2023) used NILs that differed by a single gene or two genes to study the genetic effects of the grain size-related genes gw2, gs3, and qgl3 on grain size, finding that combinations of these three genes had additive effects on grain size [30].These studies focused on the effects of grain size-related genes on grain size and yield.However, less is known about the effects of grain size-related genes on rice quality.
Here, we report that GNL44, encoding a RING-type E3 ubiquitin ligase, regulates grain size and quality in rice.We analyzed the effects of gnl44, qgl3, and gs3 on rice qualityrelated traits using monogenic and two-gene NILs.Our results revealed additive effects for these three genes on grain quality.The findings may facilitate our understanding of the mechanisms underlying the determination of grain size and quality.Our results also provide the necessary theoretical basis and genetic resources for breeding high-yield and superior-quality rice cultivars.

The gnl44 Mutant Produces Fewer but Larger Grains
The gnl44 mutant was isolated from EMS-mutagenized M 2 populations of the WT japonica cultivar Wuyunjing.gnl44 mutant plants were slightly shorter than the WT (Figure 1A,E).Panicle length, the number of primary branches, and the number of secondary branches were also lower in gnl44 compared to the WT (Figure 1B,F-H).The grain number per panicle of gnl44 was about 43.95% lower than that of the WT (Figure 1I).In addition, the gnl44 mutant produced large grains (Figure 1C,D).The length of gnl44 grains increased by 9.8% (Figure 1J) and their width increased by 21.2% compared to the WT (Figure 1K).The thickness of the gnl44 grains increased by 15.3% compared to the WT (Figure 1L).The gnl44 grains were also significantly heavier than the WT grains (Figure 1M).However, the grain yield per plant in gnl44 was decreased by about 21.8% compared to the WT (Figure 1N).These results indicate that gnl44 negatively regulates grain number and grain size in rice.
(Figure 1A,E).Panicle length, the number of primary branches, and the number of secondary branches were also lower in gnl44 compared to the WT (Figure 1B,F-H).The grain number per panicle of gnl44 was about 43.95% lower than that of the WT (Figure 1I).In addition, the gnl44 mutant produced large grains (Figure 1C,D).The length of gnl44 grains increased by 9.8% (Figure 1J) and their width increased by 21.2% compared to the WT (Figure 1K).The thickness of the gnl44 grains increased by 15.3% compared to the WT (Figure 1L).The gnl44 grains were also significantly heavier than the WT grains (Figure 1M).However, the grain yield per plant in gnl44 was decreased by about 21.8% compared to the WT (Figure 1N).These results indicate that gnl44 negatively regulates grain number and grain size in rice.

GNL44 Affects Cell Division and Cell Expansion
Given that gnl44 had longer and wider grains surrounded by the spikelet hull compared to the WT (Figure 2A), we produced paraffin sections from developing grains to analyze the parenchyma cell layer in gnl44 and WT.The length of the outer parenchyma cell layer and the number of cells were significantly increased in gnl44 compared to the WT (Figure 2B-E).In addition, we measured the length and width of cells in the outer and inner glumes of WT and gnl44, revealing that the outer epidermal cells in the gnl44 lemmas were longer and wider than those of the WT.Similarly, the inner epidermal cells of gnl44 were longer and wider than those of the WT (Figure 2F-K).Thus, GNL44 regulates grain size by limiting cell expansion in spikelet hulls.
Given that gnl44 had longer and wider grains surrounded by the spikelet hull compared to the WT (Figure 2A), we produced paraffin sections from developing grains to analyze the parenchyma cell layer in gnl44 and WT.The length of the outer parenchyma cell layer and the number of cells were significantly increased in gnl44 compared to the WT (Figure 2B-E).In addition, we measured the length and width of cells in the outer and inner glumes of WT and gnl44, revealing that the outer epidermal cells in the gnl44 lemmas were longer and wider than those of the WT.Similarly, the inner epidermal cells of gnl44 were longer and wider than those of the WT (Figure 2F-K).Thus, GNL44 regulates grain size by limiting cell expansion in spikelet hulls.

Cloning of GNL44
To determine the genetic basis of the gnl44 phenotype, we crossed the gnl44 mutant with the WT rice cultivar WYJ.All F1 plants showed a wild-type phenotype (normal grains), and the segregation (normal grains: large grains) ratio in the F2 progeny was 3:1

Cloning of GNL44
To determine the genetic basis of the gnl44 phenotype, we crossed the gnl44 mutant with the WT rice cultivar WYJ.All F 1 plants showed a wild-type phenotype (normal grains), and the segregation (normal grains: large grains) ratio in the F 2 progeny was 3:1 (Table 1).These results suggest that the gnl44 mutation is recessive and affects a single nuclear gene.We cloned GNL44 using the MutMap approach.We sequenced genomic DNA from a pool of F 2 plants with the gnl44 phenotype from the gnl44 × WYJ cross.We quantified the frequency of single-nucleotide polymorphisms (SNPs) in the sequencing results, as described previously [2].We detected 2979 SNPs and 227 small insertions and deletions (InDels) between WYJ and the pooled F 2 plants with the gnl44 phenotype.Using these SNPs, we calculated the SNP index in the pooled F 2 plants across the entire genome (Figure 3A).The mapping interval of GNL44 on chromosome 2 is 520 kb.Of all SNPs in this interval, only one located in the coding region of a predicted gene had an SNP index of 1, with clear and tight linkage with neighboring SNPs.This SNP in gnl44 is a mutation from A to G located in the 7th exon of Os02g0244100, causing a premature termination of translation (Figure 3B-D).These results indicate that Os02g0244100 is the candidate gene for GNL44.
We cloned GNL44 using the MutMap approach.We sequenced genomic DNA from a pool of F2 plants with the gnl44 phenotype from the gnl44 × WYJ cross.We quantified the frequency of single-nucleotide polymorphisms (SNPs) in the sequencing results, as described previously [2].We detected 2979 SNPs and 227 small insertions and deletions (InDels) between WYJ and the pooled F2 plants with the gnl44 phenotype.Using these SNPs, we calculated the SNP index in the pooled F2 plants across the entire genome (Figure 3A).The mapping interval of GNL44 on chromosome 2 is 520 kb.Of all SNPs in this interval, only one located in the coding region of a predicted gene had an SNP index of 1, with clear and tight linkage with neighboring SNPs.This SNP in gnl44 is a mutation from A to G located in the 7th exon of Os02g0244100, causing a premature termination of translation (Figure 3B-D).These results indicate that Os02g0244100 is the candidate gene for GNL44.

Complementation Test
To test whether the mutation in Os02g0244100 was responsible for the observed gnl44 phenotypes, we carried out a genetic complementation test.We introduced a WT genomic fragment containing the entire coding region of GNL44 together with 1662 bp of upstream and 787 bp of downstream sequence into the gnl44 mutant by Agrobacterium-mediated transformation, yielding 5 transgenic lines (GNL44COM/gnl44).
We quantified similar values for grain number per panicle, grain length, grain width, and thousand-grain weight in the WT and the complementation lines (Figure 4A-G).These results indicate that GNL44 is Os02g0244100, also named GW2.GNL44 is annotated by The Rice Annotation Project (rapdb.dna.affrc.go.jp, accessed on 6 May 2021) as encoding a RING-type E3 ubiquitin ligase.

Complementation Test
To test whether the mutation in Os02g0244100 was responsible for the observed gnl44 phenotypes, we carried out a genetic complementation test.We introduced a WT genomic fragment containing the entire coding region of GNL44 together with 1662 bp of upstream and 787 bp of downstream sequence into the gnl44 mutant by Agrobacterium-mediated transformation, yielding 5 transgenic lines (GNL44COM/gnl44).
We quantified similar values for grain number per panicle, grain length, grain width, and thousand-grain weight in the WT and the complementation lines (Figure 4A-G).These results indicate that GNL44 is Os02g0244100, also named GW2.GNL44 is annotated by The Rice Annotation Project (rapdb.dna.affrc.go.jp, accessed on 6 May 2021) as encoding a RING-type E3 ubiquitin ligase.To study the molecular mechanism underlying the role of GNL44 in grain-type formation, we performed transcriptome sequencing of panicles at different stages of development.We prepared and sequenced RNA samples from WT and gnl44 panicles at stages P1 (3 cm) and P2 (10 cm).The results indicated that the target genes of GNL44 are mainly involved in protein metabolism pathways such as protein folding, protein self-association, and protein complex oligomerization (Figure 4H,I).Peroxidase activity and oxidoreductase activity were also significantly enriched (Figure 4H,I).These results suggest that the GNL44 mutation has a significant impact on important biological processes such as protein-related metabolic pathways and redox pathways, thereby affecting the growth and development of glumes during the early stage of panicle development.

Expression of GNL44 and Subcellular Localization of GNL44
We measured GNL44 expression in different tissues collected from WT plants and observed that GNL44 was preferentially expressed in panicles and developing seeds (Figure 5A).To assess the subcellular localization of GNL44, we transfected rice protoplasts with a construct encoding a GNL44-green fluorescent protein (GFP) fusion or with 35S:GFP as control (Figure 5B).We detected green fluorescence for GNL44-GFP in the cytoplasm, indicating that GNL44 is a cytosolic protein.
To study the molecular mechanism underlying the role of GNL44 in grain-type formation, we performed transcriptome sequencing of panicles at different stages of development.We prepared and sequenced RNA samples from WT and gnl44 panicles at stages P1 (3 cm) and P2 (10 cm).The results indicated that the target genes of GNL44 are mainly involved in protein metabolism pathways such as protein folding, protein self-association, and protein complex oligomerization (Figure 4H,I).Peroxidase activity and oxidoreductase activity were also significantly enriched (Figure 4H,I).These results suggest that the GNL44 mutation has a significant impact on important biological processes such as protein-related metabolic pathways and redox pathways, thereby affecting the growth and development of glumes during the early stage of panicle development.

Expression of GNL44 and Subcellular Localization of GNL44
We measured GNL44 expression in different tissues collected from WT plants and observed that GNL44 was preferentially expressed in panicles and developing seeds (Figure 5A).To assess the subcellular localization of GNL44, we transfected rice protoplasts with a construct encoding a GNL44-green fluorescent protein (GFP) fusion or with 35S:GFP as control (Figure 5B).We detected green fluorescence for GNL44-GFP in the cytoplasm, indicating that GNL44 is a cytosolic protein.

GNL44 Affects Grain Appearance Quality of Milled Rice
To explore the effect of the GNL44 genotype on rice quality, we characterized the appearance of rice from WT and gnl44.The width and length of brown grains from gnl44 were higher than those from the WT (Figure 6A-D).The transparency of the milled rice from gnl44 was lower than that from WT.Some milled rice from gnl44 showed large areas with a white core in endosperm (Figure 6E,F).In agreement, chalky grain rate and chalkiness degree in the gnl44 mutant were significantly higher than in WT (Figure 6G,H).To gain a better picture of grain ultrastructure, we observed the structure of starch granules from WT and gnl44 grains by scanning electron microscopy.The starch granules of WT were closely arranged and regularly organized in contrast to the starch granules of gnl44, which were irregular and loosely arranged (Figure 6E,F).These results suggest that the appearance quality of milled rice is reduced in the gnl44 mutant.

GNL44 Affects Grain Appearance Quality of Milled Rice
To explore the effect of the GNL44 genotype on rice quality, we characterized the appearance of rice from WT and gnl44.The width and length of brown grains from gnl44 were higher than those from the WT (Figure 6A-D).The transparency of the milled rice from gnl44 was lower than that from WT.Some milled rice from gnl44 showed large areas with a white core in endosperm (Figure 6E,F).In agreement, chalky grain rate and chalkiness degree in the gnl44 mutant were significantly higher than in WT (Figure 6G,H).To gain a better picture of grain ultrastructure, we observed the structure of starch granules from WT and gnl44 grains by scanning electron microscopy.The starch granules of WT were closely arranged and regularly organized in contrast to the starch granules of gnl44, which were irregular and loosely arranged (Figure 6E,F).These results suggest that the appearance quality of milled rice is reduced in the gnl44 mutant.

Eating and Cooking Quality of the gnl44 Mutant Is Worse Than That of the WT
To clarify the effects of GNL44 on eating and cooking quality, we performed a rice quality assay.Rice flour produced from gnl44 grains had lower amylose content, higher protein content, and lower gel consistency compared to the WT (Figure 7A-C).
We measured the length distribution of amylopectin chains in rice flour produced from WT and gnl44 grains.WT flour had higher contents of DP 6-12 amylopectin chains than gnl44 flour, except for DP 8 chains.By contrast, WT flour had fewer DP 13-24 chains than gnl44 flour did, except for DP 15.The contents of DP 25-36 chains were slightly lower in the WT than in gnl44, except for DP 36, while aggregates of DP >37 chains accumulated to similar levels in the WT and gnl44 (Figure 7D).These results suggest that gnl44 mainly affects the biosynthesis of fa and fb1 chains and that the grain quality of milled rice is reduced in the gnl44 mutant.
Amylose content, protein content, gel consistency, and length distribution of amylopectin chains are all closely related to eating and cooking quality (ECQ), which prompted us to understand the effect of gnl44 on the texture of cooked rice by carrying out an RVA experiment.The values for peak viscosity and breakdown viscosity of gnl44 were lower than those of the WT.However, cold viscosity, final viscosity, and setback viscosity all increased in gnl44 relative to the WT (Figure 7E, Table S2).Furthermore, the taste value of gnl44 was lower than that of the WT (Figure 7F).These results suggest that the ECQ of gnl44 is decreased compared to that of the WT.

Eating and Cooking Quality of the gnl44 Mutant Is Worse Than That of the WT
To clarify the effects of GNL44 on eating and cooking quality, we performed a rice quality assay.Rice flour produced from gnl44 grains had lower amylose content, higher protein content, and lower gel consistency compared to the WT (Figure 7A-C).
We measured the length distribution of amylopectin chains in rice flour produced from WT and gnl44 grains.WT flour had higher contents of DP 6-12 amylopectin chains than gnl44 flour, except for DP 8 chains.By contrast, WT flour had fewer DP 13-24 chains than gnl44 flour did, except for DP 15.The contents of DP 25-36 chains were slightly lower in the WT than in gnl44, except for DP 36, while aggregates of DP >37 chains accumulated to similar levels in the WT and gnl44 (Figure 7D).These results suggest that gnl44 mainly affects the biosynthesis of fa and fb1 chains and that the grain quality of milled rice is reduced in the gnl44 mutant.
Amylose content, protein content, gel consistency, and length distribution of amylopectin chains are all closely related to eating and cooking quality (ECQ), which prompted us to understand the effect of gnl44 on the texture of cooked rice by carrying out an RVA experiment.The values for peak viscosity and breakdown viscosity of gnl44 were lower than those of the WT.However, cold viscosity, final viscosity, and setback viscosity all increased in gnl44 relative to the WT (Figure 7E, Table S2).Furthermore, the taste value of gnl44 was lower than that of the WT (Figure 7F).These results suggest that the ECQ of gnl44 is decreased compared to that of the WT.
The above results indicate that the mutation in GNL44 in rice leads to poor appearance quality and lower ECQ.To determine which genes might mediate this effect, we explored the downstream target genes of GNL44 by RNA-seq analysis.We collected seeds from WT and gnl44 plants at 5 and 15 days after flowering (DAF) for transcriptome analysis.Protein metabolism pathways such as protein folding, protein catabolic process, and protein complex oligomerization as well as redox pathways such as peroxidase activity and response to reactive oxygen species (ROS) were enriched among the differentially expressed genes (DEGs) (Figure 7G,H).These results indicate that GNL44 affects rice appearance quality and ECQ via protein-related metabolic pathways and redox-related biological processes.
Int. J. Mol.Sci.2024, 25, x FOR PEER REVIEW 9 of 21 The above results indicate that the mutation in GNL44 in rice leads to poor appearance quality and lower ECQ.To determine which genes might mediate this effect, we explored the downstream target genes of GNL44 by RNA-seq analysis.We collected seeds from WT and gnl44 plants at 5 and 15 days after flowering (DAF) for transcriptome analysis.Protein metabolism pathways such as protein folding, protein catabolic process, and protein complex oligomerization as well as redox pathways such as peroxidase activity and response to reactive oxygen species (ROS) were enriched among the differentially expressed genes (DEGs) (Figure 7G,H).These results indicate that GNL44 affects rice appearance quality and ECQ via protein-related metabolic pathways and redox-related biological processes.

Rice Quality of NIL-gnl44, NIL-gs3, and NIL-qgl3
High yield and superior quality are the primary goals of rice breeders and can be affected by grain size [35].We previously reported that the large-grain rice cultivar TD70 (thousand-grain weight of 68.6 g) carries superior alleles at the loci qGL3 and GS3, which control grain size, while Kasalath (thousand-grain weight of 19.1 g) has normal alleles at these loci.TD70 is therefore an ideal material for studying the effects of genetic interactions between different grain size genes on milled rice quality [36].
We wished to explore the effect of the interaction between GNL44 and qGL3 and GS3 on rice quality by constructing monogenic NILs of qgl3 and gs3 with Kasalath as the receptor and TD70 as the donor.We also constructed NIL-gnl44 with Kasalath as the receptor and the gnl44 mutant as the donor.
The width of the milled grain from NIL-gnl44 and the length of the milled grain from NIL-qgl3 were significantly higher compared to that of Kasalath (Figure 8A-C).The chalky grain ratio and chalkiness degree of NIL-gnl44 were higher than those of Kasalath, and those of NIL-qgl3 were significantly lower than those of Kasalath (Figure 8D,E).A rice quality assay showed that the amylose contents of NIL-gnl44 and NIL-gs3 were higher than that of Kasalath.The protein contents of NIL-gnl44 and NIL-qgl3 were also higher than that of Kasalath, while that of NIL-gs3 was lower than that of Kasalath (Figure 8F,G).The Gel consistency of the three NILs was lower than that of Kasalath (Figure 8H).Moreover, the RVA experiments showed that the peak viscosity of NIL-gnl44 and NIL-gs3 were higher, and that of NIL-qgl3 was lower, compared to that of Kasalath.The cold viscosity of NIL-gs3 was elevated compared to that of Kasalath but was similar to that of Kasalath in NIL-gnl44 and NIL-gs3.The breakdown viscosities of NIL-gs3 and NIL-qgl3 were lower than that of Kasalath.The final viscosity and setback viscosity were higher in all three monogenic NILs relative to Kasalath.The consistency viscosity of NIL-gnl44 was elevated, similar in NIL-gs3, and slightly increased in NIL-qgl3 compared to that of Kasalath (Figure 8I,J).These results suggest that GNL44 and qGL3 mainly affect the chalkiness and gel consistency and that GS3 mainly affects protein content.

Additive Effects of gnl44, gs3, and qgl3 on Rice Quality
We also constructed two-gene NILs to evaluate the effects of different combinations of grain size genes on rice quality.The length and width of milled grain from these twogene NILs increased in general compared to those of Kasalath (Figure 9A-C).The chalky grain ratio and chalkiness degree of the two-gene NILs NIL-gnl44/gs3 and NIL-gnl44/qgl3 in which gnl44 was pyramided were higher than those of the corresponding monogenic NILs NIL-gs3 and NIL-qgl3.However, the chalky grain ratio and chalkiness degree of the two-gene NILs NIL-gnl44/qgl3 and NIL-gs3/qgl3 in which qgl3 was pyramided were lower than those of the corresponding monogenic NILs NIL-gnl44 and NIL-gs3 (Figure 9D,E).The amylose content of all three two-gene NILs was decreased compared to that of Kasalath (Figure 9F).The protein content of the two-gene NILs NIL-gnl44/gs3 and NIL-gs3/qgl3 was higher than that of the corresponding monogenic NILs NIL-gnl44 and NIL-qgl3 (Figure 9G).The gel consistency of NIL-gnl44/gs3 and NIL-gs3/qgl3 was lower than that of the corresponding monogenic NILs NIL-gnl44 and NIL-qgl3.The gel consistency of NIL-gnl44/qgl3 was the highest among the seven lines (three two-gene NILs, three monogenic NILs, and Kasalath) (Figure 9H).

Additive Effects of gnl44, gs3, and qgl3 on Rice Quality
We also constructed two-gene NILs to evaluate the effects of different combinations of grain size genes on rice quality.The length and width of milled grain from these twogene NILs increased in general compared to those of Kasalath (Figure 9A-C).The chalky grain ratio and chalkiness degree of the two-gene NILs NIL-gnl44/gs3 and NIL-gnl44/qgl3 in which gnl44 was pyramided were higher than those of the corresponding monogenic NILs NIL-gs3 and NIL-qgl3.However, the chalky grain ratio and chalkiness degree of the two-gene NILs NIL-gnl44/qgl3 and NIL-gs3/qgl3 in which qgl3 was pyramided were lower than those of the corresponding monogenic NILs NIL-gnl44 and NIL-gs3 (Figure 9D,E).The amylose content of all three two-gene NILs was decreased compared to that of Kasalath (Figure 9F).The protein content of the two-gene NILs NIL-gnl44/gs3 and NIL-gs3/qgl3 was higher than that of the corresponding monogenic NILs NIL-gnl44 and NIL-qgl3 (Figure 9G).The gel consistency of NIL-gnl44/gs3 and NIL-gs3/qgl3 was lower than that of the corresponding monogenic NILs NIL-gnl44 and NIL-qgl3.The gel consistency of NIL-gnl44/qgl3 was the highest among the seven lines (three two-gene NILs, three monogenic NILs, and Kasalath) (Figure 9H).RVA results showed that the peak viscosity was elevated in all two-gene NILs relative to their corresponding monogenic NILs and Kasalath.The cold viscosity of NIL-gnl44/qgl3 was higher, of NIL-gs3/qgl3 was similar, and of NIL-gnl44/gs3 was slightly decreased relative to that of Kasalath (Figure 9I,J).The breakdown viscosity increased in NIL-gnl44/gs3 and slightly decreased in NIL-gnl44/gs3 compared to Kasalath.The final viscosity and setback viscosity increased for all three two-gene NILs relative to Kasalath.The consistency viscosity of NIL-gnl44/gs3 and NIL-gs3/qgl3 also increased (Figure 9I,J).Taken together, these findings suggest that GNL44, GS3, and qGL3 differentially affect rice quality.
We performed a two-way analysis of variance (ANOVA) to dissect the genetic interactions between gnl44, qgl3, and gs3 (Tables 2 and S3).We determined that the combination of gnl44 and qgl3 leads to a genetic interaction that influences chalky grain ratio, protein content, gel consistency, and amylose content.gnl44 and gs3 had significant genetic interactions for chalky grain ratio, chalkiness degree, protein content, and amylose content.Conversely, gs3 and qgl3 exhibited genetic interactions for milled grain length, chalky grain ratio, and chalkiness degree (Tables 2 and S3).RVA results showed that the peak viscosity was elevated in all two-gene NILs relative to their corresponding monogenic NILs and Kasalath.The cold viscosity of NIL-gnl44/qgl3 was higher, of NIL-gs3/qgl3 was similar, and of NIL-gnl44/gs3 was slightly decreased relative to that of Kasalath (Figure 9I,J).The breakdown viscosity increased in NIL-gnl44/gs3 and slightly decreased in NIL-gnl44/gs3 compared to Kasalath.The final viscosity and setback viscosity increased for all three two-gene NILs relative to Kasalath.The consistency viscosity of NIL-gnl44/gs3 and NIL-gs3/qgl3 also increased (Figure 9I,J).Taken together, these findings suggest that GNL44, GS3, and qGL3 differentially affect rice quality.
We performed a two-way analysis of variance (ANOVA) to dissect the genetic interactions between gnl44, qgl3, and gs3 (Tables 2 and S3).We determined that the combination of gnl44 and qgl3 leads to a genetic interaction that influences chalky grain ratio, protein content, gel consistency, and amylose content.gnl44 and gs3 had significant genetic interactions for chalky grain ratio, chalkiness degree, protein content, and amylose content.Conversely, gs3 and qgl3 exhibited genetic interactions for milled grain length, chalky grain ratio, and chalkiness degree (Tables 2 and S3).

The gnl44 Mutant Has Fewer but Larger Grains
In this study, we identified the mutant gnl44 with a larger grain size from an EMS mutant library of the japonica rice cultivar Wuyunjing.Compared to WT, the gnl44 mutant had a 20.2% increase in grain width and a 9.8% increase in grain length (Figure 1E,F).However, due to a concomitant drop in grain number per panicle, the final grain yield per plant decreased significantly compared to WT (Figure 1N).Using the MutMap approach for gene mapping, together with complementation experiments, we established that a G-to-A mutation in exon 7 of Os02g0244100 led to the early termination of translation and was responsible for the mutant phenotype (Figures 3 and 4).Subcellular localization revealed that GNL44 localized to the cytoplasm (Figure 5B).The results of tissue model expression showed that gnl44 was highly expressed in the panicle and developing endosperm (Figure 5A).The Rice Annotation Project (rapdb.dna.affrc.go.jp, accessed on 6 May 2021) notes that GNL44 encodes the RING-type E3 ubiquitin ligase GW2; the gnl44 mutant therefore defines a new allele of GW2.
Lu et al. ( 2013) analyzed the GW2 sequence across 127 rice varieties and reported that most SNPs/InDels were located in the intronic and promoter regions.Moreover, the variation in the promoter region did not affect GW2 expression [37].Zhang et al. (2015) and Dixit et al. (2013) analyzed GW2 haplotypes in different rice resources and obtained similar results [36,38].In this study, using EMS mutagenesis, we obtained a new mutant allele of GW2 named gnl44.The gnl44 mutant exhibits a severe reduction in grain number but also a larger grain phenotype, making gnl44 a rare new allele of GW2.Song et al. ( 2007) cloned rice GW2 using the cultivars WY3 and FAZ1.The authors determined that grain length, grain width, grain thickness, and thousand-grain weight were higher in NIL-gw2 compared to FAZ1.However, grain number per panicle decreased and grain yield per plant increased in NIL-gw2 [7].Huang et al. (2022) independently cloned GW2 using the cultivar HZ and observed that NIL-gw2 had larger grains and a higher thousand-grain weight compared to HZ.However, there was no significant change in the grain number per panicle of NIL-gw2, and the yield per plant increased [39].In the current study, we isolated the new GW2 EMS mutant gnl44.Although gnl44 grains were wider, longer, and heavier compared to WT grains, the number of grains per panicle decreased by 43.95% in gnl44 compared to WT, resulting in a 21.8% decrease in grain yield per plant.Among the lines examined in these three studies, the phenotype of the NIL generated by Song et al. resulted from premature termination due to mutations in exon 4 of GW2, that of gnl44 isolated in the current study resulted from premature termination due to mutations in exon 7, and that of the NIL isolated by Huang et al. resulted from a single base substitution in this gene.These findings suggest that exons 7 and 8 of GW2 might be related to the establishment of grain number per panicle in rice.
GO enrichment analysis of the transcriptome data from young panicles at different developmental stages indicated that protein folding and other protein-related pathways were significantly enriched among the DEGs (Figure 4H,I).This finding is consistent with the observation that GNL44, an E3 ubiquitin ligase, is involved in protein degradation.Hao et al. (2021) found that GNL44/GW2 ubiquitinates the glutaredoxin protein OsGRX8 and targets it for degradation to regulate grain size.The modified expression of OsGRX8 affects plant responses to various abiotic stresses [40].Indeed, we found that many abiotic stress-related pathways were enriched in the transcriptome results (Figures 4H,I and 7G,H).These results suggest that GNL44/GW2 may be involved in abiotic stress responses.

gnl44 Affects Appearance Quality and ECQ
GW2 is a major gene affecting grain size, and previous research has mostly focused on its regulation of grain size [7,39,41,42].Few studies have explored the regulation of rice quality by GW2 in depth.In this study, we investigated the appearance and cooking qualities of the gnl44 mutant.Compared to WT, chalky grain ratio and chalkiness percentage of the gnl44 mutant increased (Figure 6G,H).The starch in gnl44 mutant grains was loosely arranged (Figure 6E,F).Moreover, the observed distribution of amylopectin chain length suggests that GNL44 mainly affected the starch biosynthesis of fa and fb1 chains (Figure 7D).The appearance quality of milled gnl44 grain was worse than that of WT.In addition, the amylose and protein contents of gnl44 grains was significantly decreased compared to WT (Figure 7A).The results of RVA showed that the peak viscosity and breakdown viscosity of gnl44 were also decreased (Figure 7D,E).The taste value of the gnl44 mutant was lower than that of WT (Figure 7F).These results indicate that GNL44 mutation leads to poor appearance and taste quality in milled rice.Song et al. (2007) [7] previously showed that amylose content, protein content, and gel consistency did not differ between NIL-gw2 and FAZ1.However, the chalky grain ratio of NIL-gw2 was increased compared to FAZ1.Based on these results, Song et al. [7] postulated that the GW2 allele from cultivar WY3 had little influence on appearance and would not reduce cooking quality.Huang et al. (2022) [39] showed that amylose content, chalkiness degree, and chalky grain ratio similarly did not differ between NIL-gw2 and HZ.However, the gel consistency of NIL-gw2 was higher than that of HZ, while its protein content was lower than that of HZ.This observation was taken to indicate that the jf42-GW2 allele did not affect the appearance quality of milled rice.
Wx is the major QTL affecting the appearance quality, eating quality, and cooking quality of rice [43,44].To date, at least eight Wx alleles have been functionally characterized, including Wx lv , Wx a , Wx in , Wx b , Wx op/hp , Wx mq , Wx mp , and wx [44][45][46][47][48].In this study, the Wx genotype of WYJ, gnl44, and TD70 was Wx b , and the Wx genotype of Kasalath, the monogenic NILs, and the two-gene NILs was Wx lv .We determined that in the Wx b background, the AC content decreased in the presence of the GNL44/GW2 mutation (Figure 7A), while the same mutation increased the AC content in the Wx lv background (Figure 8F).These results indicate that although GNL44/GW2 can affect AC, Wx is still the major gene determining AC.
Rice grain length and grain width are important factors affecting appearance quality.We investigated the grain length-to-width ratio in all three GW2 studies.The grain lengthto-width ratio was reported to be about 3.3 (Song et al., 2007 [7]), 2.91 (Huang et al., 2022 [39]), or 1.98 (this work).Therefore, the grain length-to-width ratio of gnl44 in this study was significantly lower than that of gw2 alleles reported in the other two studies, offering a possible explanation for the deterioration in appearance quality and cooking quality of gnl44.
Achary et al. [49] used clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated nuclease 9 (Cas9)-mediated gene editing to target the fourth exon of GW2.The endosperm of the GW2-KO mutant seeds showed a thicker aleurone layer, with higher protein content in the seeds.The accumulation of essential dietary minerals (Fe, Zn, K, P, Ca) in the rice endosperm of the OsGW2-KO mutant also increased [49].This result indicates that GW2 not only affects the appearance and cooking quality but also the mineral content of rice.
The excessive accumulation of ROS in rice endosperm can trigger programmed cell death (PCD), thereby inhibiting the accumulation of storage substances and leading to chalkiness.The imbalance of redox homeostasis caused by oxidative stress is the main cause of grain chalkiness [50,51].In this study, analysis of transcriptome data revealed the significant enrichment of genes involved in the oxidative stress response and protein metabolism-related pathways in gnl44 grains (Figure 7G,H).These results suggest that GNL44 may regulate grain appearance and ECQ by affecting the metabolism of rice qualityrelated proteins and redox homeostasis.

Potential Value of gnl44 for Rice Quality Improvement
In rice breeding, it is necessary to select the appropriate grain size gene to achieve a desired improvement in yield and quality [35].Haplotype analysis indicates that the functional variation in GW2 has largely not been exploited by breeders [37,38,52].Therefore, evaluating the effects of different grain size genes, especially gw2 and others, is of great significance for rice breeding in general and for rice quality in particular.
In this study, we isolated a rare new allele of gw2-gnl44.To assess the potential breeding value of gnl44, we used monogenic NILs and two-gene NILs carrying gnl44, gs3, or qgl3 alleles to study their additive effects on traits related to rice quality.The amylose content of all three two-gene NILs decreased compared to that of Kasalath (Figure 9F).The amylose and protein contents of NIL-gnl44/qgl3 were lower than those of Kasalath, but its gel consistency was higher than that of Kasalath (Figure 9F-H).Milled rice is reported to have a low amylose content, low protein content, and high gel consistency with good cooking quality [53].We hypothesize that pyramiding of gnl44 and qgl3 can be a useful approach to improving the cooking quality of rice.However, the grain yield per plant of NIL-gnl44/qgl3 was lower than that of Kasalath (Table S4), making it unsuitable for breeding rice varieties with high yield and good quality.The chalkiness degree, chalky grain ratio, and gel consistency of NIL-gs2/qgl3 were lower than those of Kasalath, while its amylose and protein contents were higher than those of Kasalath (Figure 9D-H).Moreover, the grain yield per plant of NIL-gs2, qgl3 was comparable to that of Kasalath (Table S4).Therefore, we believe that these two genes can be applied to the breeding of new rice varieties with high yield and good quality.The chalkiness grain rate and chalkiness degree of NIL-gnl44/gs3 were much higher than those of Kasalath, and the highest among the three two-gene NILs were characterized here (Figure 9D,E).The appearance quality of NIL-gnl44, gs3 was poor.Compared to Kasalath, NIL-gnl44/gs3 had a lower amylose content, lower gel consistency, and higher protein content (Figure 9F-H).The grain yield per plant of NIL-gnl44 and gs3 was also lower than that of Kasalath (Table S4).Therefore, gnl44 and gs3 are not suitable for breeding new rice varieties with high yield and good quality.

Plant Materials and Growth Conditions
The gnl44 mutant was obtained by mutagenizing the japonica rice (Oryza sativa) cultivar Wuyunjing (WYJ) with 0.8% ethyl methane sulfonate (EMS).We selected the gnl44 mutant from the M2 generation.After 5 years of cultivation, its phenotype is stable and was used in this study.All plants in this study were cultivated at the Jiangsu Academy of Agricultural Sciences located in Nanjing, Jiangsu province (118 • 5 ′ E, 32 • 0 ′ N).The following agronomic traits were determined for field-grown plants at maturity: plant height, panicle length, grain number per panicle, primary and secondary branches per panicle, grain width, grain length, grain thickness, grain yield per plant, and thousand-grain weight [7].

Positional Cloning of GNL44
An F 2 population was produced by crossing gnl44 with the WT cultivar.WYJ was used to clone GNL44.The whole genomes of a mixed pool of 50 plants with the gnl44 phenotype and wild-type WYJ were sequenced using a NextSeq 500 system (Illumina, San Diego, CA, USA).The MutMap was used to isolate GNL44 as previously described [55].The reference rice genome IRGSP1.0 was used in this study.
For complementation of the gnl44 mutant, a 8330 bp genomic fragment of the candidate GNL44 gene, which contains the entire genomic coding region of GNL44; 1662 bp of upstream sequence; and 787 bp of downstream sequence were amplified with the GNL44-COMF/R primers and sequenced (Table S1).The amplified fragment was cloned into the BamHI site of the pCAMBIA1300 binary vector.The resulting pCAMBIA1300-GNL44 plasmid was introduced into Agrobacterium (Agrobacterium tumefaciens) strain EHA105.gnl44 calli were produced by the induction of mature embryos according to the descriptions of Toki (1997) and Toki (2006) [56,57].Then, gnl44 calli were transformed by an Agrobacterium-mediated transformation method.

Subcellular Localization of GNL44
The full-length GNL44 coding sequence without the stop codon was amplified and cloned in-frame and upstream of the GFP coding sequence into the Sal I site of the p35S::GFP vector.The inserted fragment was amplified by PCR using the specific primer pairs shown in Table S1.The resulting construct was transfected into protoplasts isolated from 20-day-old 93-11 rice seedlings, which were incubated overnight in the dark as described previously [58].GFP fluorescence was visualized with a confocal laser scanning microscope (Zeiss LSM 710).

RT-qPCR and RNA-seq Analysis
Total RNA was extracted from rice tissues using a Total RNA Miniprep kit (Axygene, Hangzhou, China) for RT-qPCR.First-strand cDNA was synthesized from total RNA using a ReverTra Ace qPCR-RT kit (Toyobo, Osaka, Japan), according to the manufacturer's instructions.qPCR was performed using SYBR premix Ex Taq II (Takara, Kusatsu, Japan) on an Applied Biosystems 7900HT instrument.The primers used for qPCR are listed in Table S1.Rice UBQ5 was used as an internal control.Data were analyzed following the relative quantification method [59].Values are means ± SD of three biological replicates.Student's t-test was used for statistical analysis.
For RNA-seq analysis, RNA samples were prepared from 3 cm (P1) and 10 cm (P2) panicles and the seeds of WT and gnl44 plants at 5 and 10 days after flowering (DAF).RNAseq analysis and Gene Ontology (GO) enrichment analysis were performed as described previously [60].

Determination of Taste Value and Rapid Visco Analyzer (RVA) Profiles
An RCTA-11A Taste Analyzer (Satake, Hiroshima, Japan) was used to analyze the taste value of milled rice according to the description of Chen et al. [61].A 30 g aliquot of washed milled rice was transferred into a 50 mL aluminum box containing 40 mL distilled water and cooked in an electric rice cooker.After 20 min at equilibrium, the taste value of cooked rice was evaluated.
The RVA profiles of rice flour were investigated with a Rapid Visco Analyzer (Techmaster, Newport Scientific; Warriewood, Sydney, Australia).A 2.0 g aliquot of rice flour and 25 mL distilled water were mixed well in the RVA aluminum canister.The sample was heated according to the method described by Li et al. [62].Finally, peak viscosity, cold viscosity, final viscosity, breakdown viscosity, setback viscosity, and pasting temperature were measured.

Measurement of Physicochemical Characteristics
The amylose content of rice flour was determined as described by Shi et al. [63].The protein content was estimated from the nitrogen content using the Kjeldahl method with a conversion factor of 5.95 [64].The gel consistency was measured following the method of Tan et al. [65].

Amylopectin Branch Chain Length Distribution Analysis
A high-performance anion-exchange chromatography system (ICS-6000, Thermo Fisher Scientific, Waltham, MA, USA) coupled with a Dionex™ CarboPac™ PA10 anionexchange column was used following the method of Li et al. [62] with modifications to analyze the length distribution of amylopectin chains.
Starch samples (5 mg) were suspended in 1 mL double-distilled water and boiled for 30 min.Gelatinized polyglucan (500 µL) was mixed with 10 µL 1 m sodium acetate buffer, 2 µL sodium azide (2% solution, w/v), and 5 µL isoamylase (1000 U/µL 15284, Sigma Aldrich Corporation, St. Louis, MO, USA) and stored in a thermotank at 40 • C for 24 h.Samples were transferred to a boiling water bath for 10 min to terminate the reaction.After ultra-filtration, 500 µL of the solution was taken for analysis of the distribution of amylopectin branch chain length.The branching degree of starch was measured using a Bruker BioSpin GmbH nuclear magnetic resonance spectrometer (Bruker, Rheinstetten, Germany) following the method of Zou et al. [66].

Construction of NILs
All NILs in this study used Kasalath as the recipient material.For NIL-gnl44, the gnl44 mutant was used as a donor; for NIL-gs3 and NIL-qgl3, TD70 was used as donor.After five generations of repeated backcrossing, the BC 5 F 1 population was obtained, from which the BC 5 F 3 population was obtained after two generations of selfing.
From the BC 5 F 3 generation, we developed near-isogenic lines for qgl3(NIL-qgl3) and gs3((NIL-gs3)), which carry an ∼103.5and∼118.4-kbsegment of TD70 containing the qgl3 and gs3 loci in the Kasalath genetic background, respectively, using molecular markerassisted selection (MAS).NIL-gnl44 was constructed with a very small gnl44 chromosomal region (~123.6 kb) containing the gnl44 locus in the Kasalath genetic background using MAS.An analysis of the genes in these segments revealed that they do not contain known grain size-or quality-related genes.
For the two-gene NILs, NIL-gnl44, NIL-gs3, and NIL-qgl3 were crossed to obtain F 1 seeds.In 2018, F 1 plants were grown and F 2 seeds were harvested in Nanjing, Jiangsu Province, China.F 2 plants were planted in Hainan Province, China, in 2018, and plants homozygous for gnl44 and gs3 (NIL-gnl44/gs3) or gnl44 and qgl3 (NIL-gnl44/qgl3) were selected from the F 2 family using MAS.

Conclusions
In this study, we functionally characterized the effects of GNL44 on grain size, appearance quality, and ECQ in rice.The gnl44 mutant had fewer but larger grains than the WT.The appearance quality and taste value of the gnl44 mutant were reduced.The phenotype of gnl44 is caused by a mutation in the Os02g0244100 gene.We also explored the effects of gnl44, gs3, and qgl3 on rice quality using monogenic NILs and two-gene NILs.Analysis of the monogenic NILs showed that gnl44 and qgl3 mainly affected chalkiness and gel consistency, and gs3 mainly affected protein content.Analysis of the grain quality and ANOVA of the two-gene NILs showed that gnl44 and gs3 had significant genetic interactions for chalky grain ratio, chalkiness degree, protein content, and amylose content.The genetic interaction of gnl44 and qgl3 influences chalky grain ratio, protein content, gel consistency, and amylose content.The genetic interaction of gs3 and qgl3 affects milled grain length, chalky grain ratio, and chalkiness degree.Our results lay the foundation for breeding rice cultivars with high yields and superior quality.

Figure 1 .
Figure 1.Phenotypes of WT and gnl44 rice plants.(A) Morphology of WT and gnl44 plants.Scale bars: 10 cm.(B) Panicles of the WT and gnl44.Scale bars: 3 cm.(C,D) Morphology of mature grains from the WT and gnl44.(E-I) Plant height (E), panicle length (F), number of primary branches (G), number of secondary branches (H), and grain number per panicle (I) (n = 10).(J-N) Grain length (J), grain width (K), grain thickness (L), and thousand-grain weight (M) of mature grains and grain yield per plant (N) from the WT and gnl44 (n = 10).Values are means ± SD; Student's t-test was used to calculate the p values; ** p < 0.01.

Figure 1 .
Figure 1.Phenotypes of WT and gnl44 rice plants.(A) Morphology of WT and gnl44 plants.Scale bars: 10 cm.(B) Panicles of the WT and gnl44.Scale bars: 3 cm.(C,D) Morphology of mature grains from the WT and gnl44.(E-I) Plant height (E), panicle length (F), number of primary branches (G), number of secondary branches (H), and grain number per panicle (I) (n = 10).(J-N) Grain length (J), grain width (K), grain thickness (L), and thousand-grain weight (M) of mature grains and grain yield per plant (N) from the WT and gnl44 (n = 10).Values are means ± SD; Student's t-test was used to calculate the p values; ** p < 0.01.

Figure 3 .
Figure 3. Positional cloning of GNL44.(A) SNP index and corresponding two-sided confidence intervals: 95% (blue line) and 99% (red lines).Different color dots are the original values of the SNP index of different chromosomes, and the black curve are the fitted values of SNP index.The regions indicated in blue shading correspond to the candidate genomic interval.(B) GNL44 gene structure.Black boxes: exons; white boxes: untranslated regions; black lines: introns.ATG and TGA represent the start codon and the stop codon, respectively.(C) Sanger sequencing electropherograms for the WT and gnl44.The mutation site is highlighted by the red box.(D) Mutation in GNL44 genomic DNA, the GNL44 cDNA, and in the predicted GNL44 or gnl44 protein in the WT and gnl44, arrow indicated the mutation site, * indicated the stop codon.

Figure 3 .
Figure 3. Positional cloning of GNL44.(A) SNP index and corresponding two-sided confidence intervals: 95% (blue line) and 99% (red lines).Different color dots are the original values of the SNP index of different chromosomes, and the black curve are the fitted values of SNP index.The regions indicated in blue shading correspond to the candidate genomic interval.(B) GNL44 gene structure.Black boxes: exons; white boxes: untranslated regions; black lines: introns.ATG and TGA represent the start codon and the stop codon, respectively.(C) Sanger sequencing electropherograms for the WT and gnl44.The mutation site is highlighted by the red box.(D) Mutation in GNL44 genomic DNA, the GNL44 cDNA, and in the predicted GNL44 or gnl44 protein in the WT and gnl44, arrow indicated the mutation site, * indicated the stop codon.

Figure 4 .
Figure 4. Complementation of the gnl44 mutant.(A) Photographs of panicle from the WT, gnl44, and two independent GNL44 complementation lines (COM-1, COM-2).(B) Morphology of mature grains from the WT, the gnl44 mutant, and the two complementation lines.Scale bar: 1 cm.(C) Sanger sequencing electropherograms for the WT, gnl44, and COM-1.The mutation site is highlighted by the red box.(D-G) Grain number per panicle (D), grain length (E), grain width (F), and thousand-grain weight (G) of mature grains of A (n = 10).Values are means ± SD.Different letters indicate significant differences as determined by analysis of variance (ANOVA) followed by an LSD test for the comparison of means (p < 0.05).(H,I) Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) between WT and gnl44 at different stages of panicle development.

Figure 4 .
Figure 4. Complementation of the gnl44 mutant.(A) Photographs of panicle from the WT, gnl44, and two independent GNL44 complementation lines (COM-1, COM-2).(B) Morphology of mature grains from the WT, the gnl44 mutant, and the two complementation lines.Scale bar: 1 cm.(C) Sanger sequencing electropherograms for the WT, gnl44, and COM-1.The mutation site is highlighted by the red box.(D-G) Grain number per panicle (D), grain length (E), grain width (F), and thousandgrain weight (G) of mature grains of A (n = 10).Values are means ± SD.Different letters indicate significant differences as determined by analysis of variance (ANOVA) followed by an LSD test for the comparison of means (p < 0.05).(H,I) Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) between WT and gnl44 at different stages of panicle development.

Figure 5 .
Figure 5. GNL44 expression pattern in the WT and subcellular localization of GNL44-GFP.(A) Relative GNL44 transcript levels in various organs of WT plants.L1-L3: first-third leaf of rice seedlings; DAF: days after fertilization.The expression in roots was set to 1. (B) Subcellular localization of GNL44-GFP in transfected rice protoplasts.Free GFP from the p35S:GFP construct served as control.GFP: GFP fluorescence; chlorophyll: chlorophyll autofluorescence.Scale bar: 10 µm.

Figure 5 .
Figure 5. GNL44 expression pattern in the WT and subcellular localization of GNL44-GFP.(A) Relative GNL44 transcript levels in various organs of WT plants.L1-L3: first-third leaf of rice seedlings; DAF: days after fertilization.The expression in roots was set to 1. (B) Subcellular localization of GNL44-GFP in transfected rice protoplasts.Free GFP from the p35S:GFP construct served as control.GFP: GFP fluorescence; chlorophyll: chlorophyll autofluorescence.Scale bar: 10 µm.

Figure 6 .
Figure 6.Grain appearance quality of WT and gnl44 milled rice.(A,B) Photographs showing the length (A) and width (B) of brown rice from the WT and gnl44.Scale bars: 10 mm.(C,D) Length (C) and width (D) of brown rice from the WT and gnl44.Values are means ± SD (n = 3).(E,F) Increased chalkiness in milled rice of gnl44 compared to the WT.Scale bars: 10 mm.Scanning electron microscopy of cross-sections of milled rice from the WT and gnl44 are shown in upper right corners (scale bars, 30 µm). (G,H) Percentage of milled rice with chalkiness (G) and chalkiness degree of milled rice from the WT and gnl44 (H).Student's t-test was used to calculate the p values; ** p < 0.01, * p < 0.05.

Figure 6 .
Figure 6.Grain appearance quality of WT and gnl44 milled rice.(A,B) Photographs showing the length (A) and width (B) of brown rice from the WT and gnl44.Scale bars: 10 mm.(C,D) Length (C) and width (D) of brown rice from the WT and gnl44.Values are means ± SD (n = 3).(E,F) Increased chalkiness in milled rice of gnl44 compared to the WT.Scale bars: 10 mm.Scanning electron microscopy of cross-sections of milled rice from the WT and gnl44 are shown in upper right corners (scale bars, 30 µm). (G,H) Percentage of milled rice with chalkiness (G) and chalkiness degree of milled rice from the WT and gnl44 (H).Student's t-test was used to calculate the p values; ** p < 0.01, * p < 0.05.

Figure 7 .
Figure 7. ECQ of the WT and gnl44.(A-C) Amylose content, protein content, and gel consistency from WT and gnl44 flour.(D) Chain length distribution of amylopectin in the WT and gnl44.(E) RVA spectra of WT and gnl44 flour.(E) Rapid Visco Analyzer profile characteristics of WT and gnl44.(F) Taste value of WT and gnl44 grains.Values are means ± SD (n = 3 in (A-C,F,G)).Student's t-test was used to calculate the p values; ** p < 0.01.(G,H) Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) between WT and gnl44 at different stages after flowering.

Figure 7 .
Figure 7. ECQ of the WT and gnl44.(A-C) Amylose content, protein content, and gel consistency from WT and gnl44 flour.(D) Chain length distribution of amylopectin in the WT and gnl44.(E) RVA spectra of WT and gnl44 flour.(E) Rapid Visco Analyzer profile characteristics of WT and gnl44.(F) Taste value of WT and gnl44 grains.Values are means ± SD (n = 3 in (A-C,F,G)).Student's t-test was used to calculate the p values; ** p < 0.01.(G,H) Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) between WT and gnl44 at different stages after flowering.

Table 2 .
Two-way analysis of variance for genetic interactions among gnl44, qgl3, and gs3.