OsPEX1, a leucine-rich repeat extensin protein, plays an important role in the regulation of caryopsis development in rice

Rice caryopses are enclosed by the outside glumes. It is long been acknowledged that the size and dimension of the outer glume is the main determinant that dictates the caryopsis size. However, it is unclear whether the development of caryopsis is completely dependent on the size of the glume, or it can grow and expand autonomously in spite of the constraint of glume enclosure. Here we report the identication of a mutant line that produces normal-sized glumes with smaller mature caryopsis that do not ll up the entire glume cavity. The mutant phenotype is caused by ectopic expression of a leucine-rice repeat extensin gene, OsPEX1 . The caryopsis phenotype in pex1 was caused by a reduction in cell size. OsPEX1 is highly expressed in the developing caryopsis. Over expression of the OsPEX1 gene driven by a constitutive promoter recapitulates the mutant phenotype, conrming that the small caryopsis phenotype is caused by ectopic expression of the OsPEX1 gene. Our results suggest that caryopsis development can be genetically uncoupled from maternally controlled glume development.

identi cation of a mutant line that produces normal-sized glumes with smaller mature caryopsis that do not ll up the entire glume cavity. The mutant phenotype is caused by ectopic expression of a leucine-rice repeat extensin gene, OsPEX1 . The caryopsis phenotype in pex1 was caused by a reduction in cell size. OsPEX1 is highly expressed in the developing caryopsis. Over expression of the OsPEX1 gene driven by a constitutive promoter recapitulates the mutant phenotype, con rming that the small caryopsis phenotype is caused by ectopic expression of the OsPEX1 gene. Our results suggest that caryopsis development can be genetically uncoupled from maternally controlled glume development.

Background
Rice is one of the most important staple food crops in the world, and is primarily consumed by human as polished white or unpolished brown rice. Grain size, usually measured by grain length, width, thickness and weight, is an important quantitative trait for rice because it affects the yield and the quality of rice; and has attracted considerable attention as one major topic for research and breeding. The rice grain is derived from a oret with caryopsis that is encased by the glume (lemma and palea). The caryopsis consists of the diploid maternal tissues (pericarp, testa, and nucellus), the triploid endosperm, and the diploid embryo. During ower development, the glume expands its volume and grows to maximum size prior to anthesis. Inside the mature glume, the caryopsis starts to grow and increase in size after fertilization, and reaches its nal size around twenty days after fertilization (Yoshida and Nagato, 2011).
Although the glume and caryopsis grow and develop at different time points, the mature caryopses perfectly ll the inner chamber enclosed by the glumes. Thus, it has long been accepted that the size and dimension of the outer glume is the main determinant that dictates the caryopsis size.
Recent studies have begun to shed light into the regulatory mechanism of glume development through identi cation and functional characterization of genes associated with grain size and yield. To date, nearly one hundred genes associated with grain shape and weight have been isolated in rice Li et al., 2019). However, almost all of the genes identi ed so far affect glume size. For example, natural mutants of GWs (grain width) produce wider glumes and widened caryopses. Other genes affecting glume size that have been studied in uence glume size either by affecting cell number and size in lemmas and palea, or cell cycle regulation Kitagawa et al., 2010;Heang and Sassa, 2012;Nakagawa et al., 2012;Qi et al., 2012;Segami et al., 2012;Zhang et al., 2012;Luo et al., 2013).
With respect to the regulatory mechanism of glume development, phytohormone and G-protein signaling have been shown to play important roles in regulating grain size and yield in rice. For instance, several dwarf mutants in brassinosteroid (BR) hormone biosynthesis and signaling pathways affect grain and panicle sizes, including d61, d2, d11, sg1 and RAV6 (Yamamuro et al., 2000;Mori et al., 2002;Hong et al., 2003;Tanabe et al., 2005;Nakagawa et al., 2012;Zhang et al., 2015). Mutations in the Dwart1 (D1), also known as the rice heterotrimeric G protein alpha subunit (RGA1), affect multiple signaling pathways such as BR and GA, and have pleiotropic effects on organ growth and reduction in glum and seed size (Ashikari et al., 1999;. Furthermore, GW6, encoding a GAST family protein, has been identi ed that control grain size through gibberellin pathway (Shi et al., 2020). Moreover, recent studies indicated that ubiquitin-proteasome pathway and mitogen-activated protein kinase (MAPK) signaling are also involved in the regulation of glume size (Li et al., 2019). However, little is known about how caryopsis size is controlled and regulated. In addition, it is unclear whether the development of caryopsis is completely dependent on the size of the glume, or it can grow and expand autonomously, independent of the constraint of glume enclosure.
Mutation in OsKinesin-13A (sar1) causes small glumes and rounded grains due to defects in cell elongation. Although glumes and caryopses were both reduced in length, sar1 speci cally affects glume elongation and length and the reduction in caryopses size was an indirect effect of the reduced space of the shortened glumes as the sar1 caryopses can grow to the WT length after removal of the glume height constrain (Deng et al., 2015). This observation suggests a model that the cell length in glume and caryopsis can be regulated independently. However, very little genetic evidence has been uncovered to support such model.
However, little is known about the functions of LRXs protein in rice. Here we reported the identi cation of a rice mutant that affects caryopsis size, instead of the development of glume. The mutant phenotype is caused by ectopic expression of the OsPEX1 gene, encoding an LRX protein with LRR domains and extensin domain. Our result provides a novel example that caryopsis growth and cell expansion can be separated from the constraint of pre-determined glume size.

Plant Materials and Growth Conditions
Rice cultivar 'Zhonghua 11' was used as the wild type. The pex1 mutant described in this paper was derived from an Ac/Ds transposon-tagging population in the japonica rice variety Zhonghua 11 (Liu et al., 2007;Ke et al., 2019). All rice seeds in this study were propagated in the paddy eld in Guangzhou, China.

Plasmid Construction and Rice Transformation
To produce PEX1 overexpression transgenic plants, the full-length PEX1 coding sequences was ampli ed from cDNA derived from rice 'Zhonghua 11' using corresponding primer pairs listed in Table S1. After con rmation by DNA sequencing, the ampli ed sequence was cloned into the binary vector pCUbi1390 for PEX1overexpression in rice. The nal constructs were electroporated into Agrobacterium tumefaciens strain EHA105.
For promoter analysis, a1572 bp DNA fragment immediately upstream of OsPEX1 start codon was ampli ed using primers LRXPF and LRXPF (Table S1) and inserted into the EcoRI and NcoI sites of the pCAMBIA1305.1 vector. The resulting plasmid was introduced into wild-type plant by Agrobacterium tumefaciens-mediated transformation. Transgenic plants were selected on Hygromycin medium, and T 2 transgenic plants were used to analyse GUS activity.
Quantitative RT-PCR Total RNA was extracted from frozen samples with TRIzol reagent (Invitrogen) according to the manufacturer's instructions. The RNA was pre-treated with DNase I, and rst-strand cDNA was generated using a RevertAid First Strand cDNA Synthesis Kit (Thermo Scienti c). Quantitative RT-PCRs were performed using a SYBR Premix Ex Taq™ RT-PCR kit (Takara) following the manufacturer's instructions. Investigated genes and their speci c primer sets are shown in Supplemental Table S1. The relative expression level of a target gene was normalized to that of rice 25sRNA.

Histochemical analysis
The mature caryopses are soaked in FAA solution (50% ethanol, 5% glacial acetic acid and 5% formaldehyde) overnight at 4°C for cell xation and dehydrated in a graded alcohol series. Then the samples were sectioned transversely into 2-mm slices using a razor blade and embedded in para n. Semi-thin sections were cut using a glass knife in a Leica ultra-microtome, stained with 0.05% (w/v) Fast Green, and 0.1% (w/v) Safranin. The histochemical slices were observed under a light microscope. By using Image J software, the cell size was calculated.
Results pex1 produces normal size of glumes but smaller caryopsis In previous study, we characterized a rice semi-dominant dwarf mutant pex1 (Ke et al., 2019), which was obtained from a transposon-tagged population in the japonica rice variety Zhonghua 11 (Liu et al., 2007). Interestingly, this mutant has almost the same size of glumes as the wild-type (WT) plants, but with obviously smaller caryopsis (Fig. 1a). Co-segregation analysis indicated that the dwar sm of pex1 was companied by smaller caryopsis (data no shown). Compared with the wild type, caryopses width and the length of the heterozygous pex1 mutant (pex1/+) were reduced by 18.5% and 4.0%, respectively, while the homozygous pex1 mutant (pex1/pex1) had a 25.3% reduction in caryopses width and a 7.2% reduction in caryopses length (Fig. 1b). As a result, the 1000-grain weight of pex1/+ and pex1/pex1 were reduced by 25.0% and 33.8%, respectively (Fig. 1b).

Developmental characteristics of pex1 caryopsis
Morphological analysis was carried out for the grain and caryopses of wild type and pex1 mutant grown in a paddy eld. The anthesis date of each oret was marked on the surface of the lemma using a marker, and samples were collected. The glumes of pex1 plants were almost indistinguishable from wild type (WT) plants (Fig. 2a). However, the pex1 mutant produced signi cantly smaller caryopses from 7 days after pollination (DAP) compared with the wild type (Fig. 2 b, c). The major extension of the rice caryopsis along the long axis occurred from 1 to 7 DAP, while major expansion along the transversal axis occurred between 1 and 15 DAP (Fig. 2 b, c). The results suggested that the growth pattern for caryopsis width differed from that of caryopsis length in rice.

OsPEX1 affects caryopsis size by in uencing cell expansion
To test whether cell size is affected in those caryopses, aleurone cells of the caryopses from the WT and pex1 plants were examined and indeed, smaller size aleurone cells were observed in the caryopses of the pex1 plants (Fig. 3a-c). In order to examine the phenotype by molecular analyses, expression of several representative genes involved in cell expansion was investigated by quantitative RT-PCR (Fig. 3d). The cell expansion-related gene such as OsEXP6 (LOC_Os03g21820) and xyloglucan endotransglycosylase/hydrolase (XTH) genes including XTH10 (LOC_Os06g48200) and XTH17 (LOC_Os08g13920) were signi cantly downregulated in the caryopses of pex1 mutants compared with the WT control. These results suggested the pex1 mutation affected caryopsis size by decreased cell size.
Overexpression of PEX1 Leads to a pex1-like Phenotype Molecular and genetic analyses revealed that the dwarf phenotype of pex1 was caused by ectopic expression of a leucine-rich repeat extension-like gene, OsPEX1 (Ke et al., 2019). To con rm that ectopic expression of OsPEX1 (LOC_Os11g43640) is also responsible for the small caryopsis in pex1, we overexpressed LOC_Os11g43640 in the wild type plant under the control of the maize (Zea mays) Ubiquitin promoter. Compared to WT control, all positive transformants displayed a normal size of glumes but smaller caryopsis, similar to the phenotype characteristic of the pex1 mutants (Fig. 4). Thus, we concluded that the abnormal phenotypes of the pex1 mutant results from ectopic expression of OsPEX1 (LOC_Os11g43640).

Expression pattern of OsPEX1
Given that pex1 exhibits the normal size of glumes but smaller caryopsis than WT, we compared the transcriptional level of OsPEX1 in glumes and caryopsis between WT and pex1 mutant. We found that in the caryopsis OsPEX1 gene in pex1 mutants was transcribed approximately fourfold more abundant than in WT, whereas in the glumes OsPEX1 was not differentially expressed between the WT and pex1 mutants (Fig. 5a). The ndings indicate that OsPEX1 negatively modulates caryopsis size in rice. To elucidate the role of OsPEX1 in caryopsis development, we further investigate the expression pattern of OsPEX1during caryopsis development. As shown in Fig. 5b, the transcript of OsPEX1 was rapidly increased after owering, reaching peak (approximately seven folds) 7 DAP, and then reducing but retain higher level at 10 DAP.
Further promoter-β-glucuronidase (GUS) fusion analysis revealed that OsPEX1 was expressed in the outer surface of the endosperm, including the aleurone layer of developing caryopses (Fig. 5c), especially the dorsal vascular bundle of developing caryopses with high levels of GUS expression (Supplemental Fig.   S1). Interestingly, GUS expression was restricted to embryo at mature seed, no GUS staining was detected at the aleurone layer of mature seed.
We also conducted the expression analysis of several key genes involved in glume regulation in WT and pex1 mutant. As expected, qRT-PCR (Quantitative reverse transcription PCR) results showed that the transcriptional level of eight genes tested in pex1 mutant were similar to that in WT (Supplemental Fig.  S2), suggesting the ectopic expression of OsPEX1 have less effects on genes related to glume size, consistent with the pex1 phenotype with normal glume compared to WT plants.

Phylogenetic analyses of LRXs family in rice and Arabidopsis
We investigated the phylogenetic relationship between rice and Arabidopsis LRXs genes by comparing the putative full-length protein sequences, which can be divided into two sub-groups, LRX sub-group and PEX one (Fig. 6a), as previous reported by Baumberger et al . All of the putative OsLRX proteins contain the conserved leucine-rich repeat (LRR) motifs for protein-protein interactions and extensin domain probably involved in cross-linking to cell wall components (Fig. 6B). A cysteine-rich domain (CRD) might act to stabilize the structure of the LRX protein through forming disul de bonds with cysteine residues of the LRR domain (Herger et al., 2019;Moussu et al., 2020). None of the OsLRXs genes except for OsPEX2 has introns (Fig. 6b). OsPEX1 has the longest coding sequence (CDS) in the LRXs members of rice (Fig. 6b).

Discussion
Zhu and colleagues reported that OsKinesin-13A, a microtubule depolymerase plays a role in the regulation of glume length by affecting cell elongation (Deng et al., 2015). The authors showed that the mutant of OsKinesin-13A, sar1, displayed length reduction in the glume. However, WT and sar1 caryopses matured under glume-cutting conditions were similar in length. The observations indicated that the grain phenotype was caused by reduction in glume length, which indirectly restricted caryopsis size. Although some studies of the relationship between glume and caryopsis development were discussed based on notched grains in rice (Takeda et al., 1981;Xiong et al., 1986), the observations by Deng and colleagues (Deng et al., 2015) suggest that the size and length of oral glume and caryopsis can be regulated independently.
The length reduction in sar1 caryopses was not a direct consequence of mutation in OsKinesin-13A, but resulted from the space restrictions due to shortened glumes (Deng et al., 2015), indicating that OsKinesin-13A controls glume size but not caryopsis size. By contrast, mutation in OsPEX1 directly affected caryopsis size but had little effect on glume size. These results indicate that OsPEX1 is directly involved in regulating caryopsis size. Thus, our ndings strongly support the model that caryopsis development can be separated from maternally controlled glume development.
The OsPEX1 gene is a member of LRX (LRR/EXTENSIN) gene family. The LRX proteins are characterized by a domain with leucine-rich repeats in addition to the extension domain. LRRs are frequently implicated in protein-protein interactions and signal transduction during development or in pathogen recognition and defense (Song et al., 1995;Torii et al., 1996;Jinn et al., 2000). Furthermore, one proposed function of extensins is to lock-in cell shape upon cell expansion (Carpita and Gibeaut, 1993;Hall and Cannon, 2002). Together, the properties of LRR and extensins, LRXs could potentially be involved in the regulation of cell wall expansion in response to signals. Such function in cell morphogenesis is highlighted by the nding that the Arabidopsis lrx1 mutant developed aberrant root hairs (Baumberger et al., 2001). The idea is also supported by the observation in this study that pex1 had smaller aleurone cells than WT (Fig. 3), consistent with a role of OsPEX1 in cell wall expansion. Together with the observations that overexpression of OsPEX1 resulted in reduction of smaller caryopsis, our results support the existence of a negative regulatory role of OsPEX1 on caryopsis development.
According to expression pattern, LRX genes of higher plants can be classi ed as vegetatively expressed or predominantly expressed in reproductive tissue, two categories that almost completely overlap with the phylogenetic clades Herger et al., 2019). Interestingly, in spite of the fact that OsPEX1 belong to PEX (pollen expression LRX) subfamily, it is highly expressed in root and stem as well as reproductive tissue such as developing caryopsis, but is expressed at much lower level in leaf and glume, consist with the pleiotropic phenotypes of the pex1 plants with dwar sm (Ke et al., 2019)and small caryopsis (Fig. 1), but have less effect on leaves (Ke et al., 2019) and glume developments (Fig. 1).
Our results indicate that OsPEX1 plays an important role in caryopsis development. However, the regulatory mechanisms of caryopsis development by OsPEX1 remain to be elucidated. It has been shown that the LRR domain of LRX proteins, which is highly conserved in higher plants, can directly interact with RALFs (rapid alkalinization factors) peptides hormone and CrRLK1L (Catharanthus roseus receptor-like kinase1-like) transmembrane receptors, and hence LRXs have a regulatory or signaling function during cell growth in Arabidopsis (Herger et al., 2019). Rice RALF and CrRLK1L gene family contain 46 and 16 members, respectively (http://rice.plantbiology.msu.edu/). Future research is need to investigate which RALF or CrRLK1L members in uence the functions of OsPEX1, and whether they can physically interaction with OsPEX1 protein.

Conclusion
In the present study, we propose that OsPEX1 negatively regulates caryopsis size through in uencing cell expansion in rice. In addition, our results suggest caryopsis development can be separated from maternally controlled glume development.

Figure 2
Dynamic characteristics of the pex1caryopsis development. Rice grain (a) and caryopsis (b) development from 0 to 25 days after pollination (DAP). Numbers above denote the DAP. The rapid elongation of the caryopsis occurred in the rst 5 DAP; The major period of expansion occurred in the rst 15 DAP. c comparison of caryopsis length and caryopsis width of WT and pex1 mutant. Scale bar = 5mm  Values are means ± SD. In each graph, statistically signi cant differences are indicated by the asterisks (*P<0.05, **P<0.01).  Identi cation and analysis of putative rice LRX genes. a Phylogenic analysis of OsLRXs. The phylogenetic trees were constructed using Mega7.0 program with neighbor-joining method; bootstrap analysis was performed with 1000 replicates and excluding positions with gaps. Based on the analysis, OsLRXs were divided into two sub-groups. b Gene structures and conserved domains of the OsLRXs. Thin lines represent introns, dark bars refer to putative protein-coding sequence (CDS), the untranslated regions are shown in gray rectangles. LRX proteins contain a signal peptide for protein export, an LRR domain, a cysteine-rich linker domain (CRD) and an extensin domain.