Blocking Rice Shoot Gravitropism by Altering One Amino Acid in LAZY1

Tiller angle is an important trait that determines plant architecture and yield in cereal crops. Tiller angle is partially controlled during gravistimulation by the dynamic re-allocation of LAZY1 (LA1) protein between the nucleus and plasma membrane, but the underlying mechanism remains unclear. In this study, we identified and characterized a new allele of LA1 based on analysis of a rice (Oryza sativa L.) spreading-tiller mutant la1G74V, which harbors a non-synonymous mutation in the predicted transmembrane (TM) domain-encoding region of this gene. The mutation causes complete loss of shoot gravitropism, leading to prostrate growth of plants. Our results showed that LA1 localizes not only to the nucleus and plasma membrane but also to the endoplasmic reticulum. Removal of the TM domain in LA1 showed spreading-tiller phenotype of plants similar to la1G74V but did not affect the plasma membrane localization; thus, making it distinct from its ortholog ZmLA1 in Zea mays. Therefore, we propose that the TM domain is indispensable for the biological function of LA1, but this domain does not determine the localization of the protein to the plasma membrane. Our study provides new insights into the LA1-mediated regulation of shoot gravitropism.


Introduction
Plant architectural traits are important targets for crop improvement. For example, in rice (Oryza sativa L.), the tiller angle and overall orientation of plants are important factors that influence planting density, light interception, photosynthetic efficiency, disease resistance, and grain yield [1,2]. Rice plants with wide tiller angles have a spreading growth habit yielding favorable access to resources, and they experience less pressure from pathogen attacks. However, this type of prostrate plant occupies more space; thus, leading to a reduction in grain productivity per unit area. In contrast, narrow tiller angles yield compact plants that are inefficient at capturing light and more susceptible to diseases or pests, reducing grain yield [3]. In practice, tiller angle must be balanced between plant performance and grain yield [4][5][6]. Therefore, understanding of the genomic regulatory mechanisms of tiller angle is important for crop genetic improvement that can be achieved through molecular breeding.
LAZY1-like genes (designated as LZYs) play a highly conserved role in plant gravitropism, but they act on different organs in different plant lineages or at different times during growth, which is likely due to their distinct expression patterns [19][20][21][22]. For example, in rice, LA1 functions in both coleoptile circumnutation and angling of leaves and tillers [7,8], while its three homologs (namely OsDRO1, OsDRL1, and OsDRL2) primarily affect root architecture [23,24]. In contrast, in Zea mays, ZmLA1 mainly controls shoot gravitropism and inflorescence development [21], and in Arabidopsis thaliana, five AtLZYs have been identified with distinct functions [22]. In particular, AtLZY1 is required for gravitropism in the rachis of the inflorescence and in the seedling hypocotyl [25], while AtLZY2, AtLZY3, and AtLZY4 redundantly function in root gravitropism, and contribute less to shoot gravitropism [22,26]. Recent studies have shown that although LZYs are uncharacterized proteins, they share five conserved domains, including a predicted transmembrane (TM) domain which was thought to determine their plasma membrane localization [2,7,21]. Despite extensive studies of LZYs, our understanding of their protein structures and the molecular mechanisms governing gravitropism remains limited.
Plant LZYs act as negative regulators of polar auxin transport (PAT), which establishes auxin redistribution after perceiving gravitropic stimulation [7,8,21,22,[27][28][29][30][31]. Loss-offunction of LA1 results in enhanced PAT, and consequently alters the asymmetric distribution of auxin in shoots, leading to a spreading tiller phenotype of rice plants [29]. Expression levels of LA1 are positively controlled by the gravitropic response factor HEAT STRESS TRANSCRIPTION FACTOR 2D (HSFA2D) [32], which can be directly repressed by two homeobox-leucine zipper transcription factors, OsHOX28 and OsHOX1 [33]. The actions of LZYs and auxin upon gravistimulation have also been confirmed in maize [21], Arabidopsis [22], and Lotus japonicus [34]. Intriguingly, the dynamic re-allocation of the LA1 protein between the plasma membrane and nucleus is essential for auxin redistribution upon gravistimulation. The interaction of Brevis Radix Like 4 (OsBRXL4) and LA1 at the plasma membrane adjusts the relative proportion of plasma-membrane-localized and nucleus-localized LA1 proteins, leading to asymmetric redistribution of auxin and thus dynamic reorientation of tiller angles [35]. Therefore, elucidating the regulatory mechanisms of LA1 subcellular localization is important for revealing its mode of action upon gravistimulation.
Here, we reported the identification and characterization of a new allele of LA1 from a spreading-tiller mutant of rice, which was caused by a substitution from glycine to valine occurring in the 74th amino acid position of LA1 that is within its TM domain. The new mutant allele is accordingly named la1 G74V . To elucidate the mechanism underlying the impaired tiller growth in la1 G74V , we first confirmed that the mutation in LA1 was responsible for the altered architecture via overexpression of LA1 and la1 G74V . Since the mutation occurred in the TM domain of LA1, we further analyzed the biological function of the TM domain by creating gene-knockout and TM-lacking mutants. We found that these new mutations resulted in altered tiller angle of plants but did not affect the subcellular localization of LA1. Overall, these findings reveal the new insight in the functional domains of LA1 on controlling the tiller angle growth.

Characterization of the Rice Spreading-Tiller Mutant la1 G74V
To identify key genetic components involved in the control of rice plant architecture, we investigated an EMS-induced mutant library with altered plant morphology [36] and found a mutant that exhibited a spreading-tiller phenotype ( Figure 1A-D). Later genetic mapping and functional analysis revealed that this prostrate phenotype was caused by a point mutation located in the LA1 gene (see below); we therefore named this mutant la1 G74V . ulation treatment, the la1 G74V seedlings maintained the horizontal-direction growth under both light and dark conditions ( Figure 1F), indicating that their gravitropic response was drastically reduced in la1 G74V mutant seedlings. We further examined the gravitropic response of WT and la1 G74V coleoptiles under dark conditions and found that the coleoptiles of la1 G74V had a reduced gravitropic response compared with the WT ( Figure S1). Thus, we believed that the causal gene of the la1 G74V mutant might function in the gravitropic response of coleoptiles and shoots.

Map-Based Cloning of la1 G74V
To isolate the mutated gene conferring the prostrate growth, we generated an F2-segregating population (784 progenies), which was derived from a cross between la1 G74V and O. sativa ssp. indica ZS97. Among the F2 plants, 605 individuals grew normally, and 179 individuals were prostrate, matching a 3:1 Mendelian ratio (χ 2 = 1.96 < χ 2 df=1 = 3.84, p = 0.05), indicating that the spreading-tiller phenotype of la1 G74V is caused by a single recessive locus. Compared with the wild-type (WT) plants, the la1 G74V plants exhibited an extremely wide tiller angle ( Figure 1E). Directional growth of plants is most often associated with either phototropism or gravitropism [37]. Therefore, we first sought to determine whether the wide tiller angle of la1 G74V was caused by impaired phototropism or shoot gravitropism. We planted seedlings of the mutant in a horizontal direction and monitored their response to gravity under light and dark conditions. After 12 h of gravistimulation treatment, the la1 G74V seedlings maintained the horizontal-direction growth under both light and dark conditions ( Figure 1F), indicating that their gravitropic response was drastically reduced in la1 G74V mutant seedlings. We further examined the gravitropic response of WT and la1 G74V coleoptiles under dark conditions and found that the coleoptiles of la1 G74V had a reduced gravitropic response compared with the WT ( Figure S1). Thus, we believed that the causal gene of the la1 G74V mutant might function in the gravitropic response of coleoptiles and shoots.

Map-Based Cloning of la1 G74V
To isolate the mutated gene conferring the prostrate growth, we generated an F 2segregating population (784 progenies), which was derived from a cross between la1 G74V and O. sativa ssp. indica ZS97. Among the F 2 plants, 605 individuals grew normally, and 179 individuals were prostrate, matching a 3:1 Mendelian ratio (χ 2 = 1.96 < χ 2 df=1 = 3.84, p = 0.05), indicating that the spreading-tiller phenotype of la1 G74V is caused by a single recessive locus.
Based on the genetic linkage analysis with 96 F 2 prostrate progenies, we preliminarily mapped the candidate locus to a 2.34 Mb region between the molecular markers 16,103 and 18,443 on chromosome 11 and found that no recombination was identified between the 17,323 marker and the la1 G74V locus ( Figure 1G). The 17,323 marker was located in the fourth exon of LOC_Os11g29840 (LA1), which encodes a plant-specific protein to control tiller angle upon gravitropism [7,8]. Therefore, we sequenced LA1 in the mutant and identified a single nucleotide substitution of G-to-T in the third exon of this gene. This substitution occurred at 220 bp downstream of the start codon (ATG) and caused a change in the 74th amino acid, glycine (G), to a valine (V), therefore named this mutant allele la1 G74V (Figures 1G and S2).
To determine whether la1 G74V was the cause of the prostrate phenotype, we constructed the overexpression vectors of the WT LA1 and mutant la1 G74V driven by the maize Ubiquitin promoter, and transformed them into the la1 G74V plants, generating four and three independent transgenic lines, respectively. All transgenic lines (T 1 ) carrying the WT LA1 transgene were rescued and showed a normal tiller angle, while the transgenic lines (T 1 ) overexpressing la1 G74V failed to rescue the la1 G74V mutant phenotype (Figure 2A,B). Additionally, we observed no significant difference in the expression level of LA1 between the WT and la1 G74V ( Figure 2B), indicating that the point mutation did not affect the expression of LA1. Thus, we concluded that the G-to-T mutation in the LA1 locus was responsible for the spreading-tiller phenotype of la1 G74V . Based on the genetic linkage analysis with 96 F2 prostrate progenies, we preliminarily mapped the candidate locus to a 2.34 Mb region between the molecular markers 16,103 and 18,443 on chromosome 11 and found that no recombination was identified between the 17,323 marker and the la1 G74V locus ( Figure 1G). The 17,323 marker was located in the fourth exon of LOC_Os11g29840 (LA1), which encodes a plant-specific protein to control tiller angle upon gravitropism [7,8]. Therefore, we sequenced LA1 in the mutant and identified a single nucleotide substitution of G-to-T in the third exon of this gene. This substitution occurred at 220 bp downstream of the start codon (ATG) and caused a change in the 74th amino acid, glycine (G), to a valine (V), therefore named this mutant allele la1 G74V (Figures 1G and S2).
To determine whether la1 G74V was the cause of the prostrate phenotype, we constructed the overexpression vectors of the WT LA1 and mutant la1 G74V driven by the maize Ubiquitin promoter, and transformed them into the la1 G74V plants, generating four and three independent transgenic lines, respectively. All transgenic lines (T1) carrying the WT LA1 transgene were rescued and showed a normal tiller angle, while the transgenic lines (T1) overexpressing la1 G74V failed to rescue the la1 G74V mutant phenotype ( Figure  2A,B). Additionally, we observed no significant difference in the expression level of LA1 between the WT and la1 G74V ( Figure 2B), indicating that the point mutation did not affect the expression of LA1. Thus, we concluded that the G-to-T mutation in the LA1 locus was responsible for the spreading-tiller phenotype of la1 G74V .

The Effect of G74V Mutation on the Conformation and Subcellular Localization of LA1 Protein
Previous studies indicated that the LA1 protein contains a predicted TM domain located from the 62nd to the 83rd amino acid residues, and a predicted nuclear localization signal (NLS) domain ( Figure 3A) [35]. Considering that the G74V substitution is located in the TM region, we speculated that this variation might affect the conformation or localization of the LA1 protein. Prediction of 3D protein structure of LA1 showed that the TM domain folded inward, suggesting that this amino acid mutation from the polar glycine to the nonpolar valine might result in a drastic change in the protein conformation ( Figure 3B).

The Effect of G74V Mutation on the Conformation and Subcellular Localization of LA1 Protein
Previous studies indicated that the LA1 protein contains a predicted TM domain located from the 62nd to the 83rd amino acid residues, and a predicted nuclear localization signal (NLS) domain ( Figure 3A) [35]. Considering that the G74V substitution is located in the TM region, we speculated that this variation might affect the conformation or localization of the LA1 protein. Prediction of 3D protein structure of LA1 showed that the TM domain folded inward, suggesting that this amino acid mutation from the polar glycine to the nonpolar valine might result in a drastic change in the protein conformation ( Figure 3B). We further analyzed subcellular localization of LA1 with the PSORT Prediction program and found that this protein was predicted to localize not only to the nucleus and plasma membrane but also to the endoplasmic reticulum, which plays an important role in gravitropic response [38,39]. To test the subcellular localization of LA1, we performed transient expression analysis using fusion of LA1 to GFP (LA1-GFP) in rice protoplasts and found that the green fluorescence signal of LA1-GFP completely overlapped with the red fluorescence signal markers for the nucleus, plasma membrane, and endoplasmic reticulum ( Figure 3C).
Given that the TM domain determines plasma membrane localization of LZYs [7,21,35], we compared the plasma membrane localization of LA1-GFP to that of la1 G74V -GFP fusion proteins in rice protoplasts. The results showed that both LA1-GFP and la1 G74V -GFP fusion proteins localized to the plasma membrane with no obvious differences between them ( Figure 4A). These results suggest that the G74V variation in the TM domain of LA1 negligibly affects its subcellular localization, but likely eliminates its functional conformation. (C) Subcellular localization of LA1. The LA1-GFP construct was co-expressed respectively with the nucleus-localized NLS-mCherry construct, plasma-membrane-localized PM-mCherry construct, and endoplasmic-reticulum-localized ER-mCherry in rice protoplasts. Scale bars, 10 µm.
We further analyzed subcellular localization of LA1 with the PSORT Prediction program and found that this protein was predicted to localize not only to the nucleus and plasma membrane but also to the endoplasmic reticulum, which plays an important role in gravitropic response [38,39]. To test the subcellular localization of LA1, we performed transient expression analysis using fusion of LA1 to GFP (LA1-GFP) in rice protoplasts and found that the green fluorescence signal of LA1-GFP completely overlapped with the red fluorescence signal markers for the nucleus, plasma membrane, and endoplasmic reticulum ( Figure 3C).
Given that the TM domain determines plasma membrane localization of LZYs [7,21,35], we compared the plasma membrane localization of LA1-GFP to that of la1 G74V -GFP fusion proteins in rice protoplasts. The results showed that both LA1-GFP and la1 G74V -GFP fusion proteins localized to the plasma membrane with no obvious differences between them ( Figure 4A). These results suggest that the G74V variation in the TM domain of LA1 negligibly affects its subcellular localization, but likely eliminates its functional conformation.

The TM Domain of LA1 Is Essential for Its Biological Function Rather Than Its Plasma Membrane Localization
To further determine the biological function of the TM domain, we constructed an LA1 knockout (designated as la1) and an LA1-TM-absence (designated as LA1ΔTM) mutant in the Nipponbare cultivar via CRISPR/Cas9-mediated genome editing. To generate the LA1ΔTM mutant, two editing target sites were designed to delete the TM coding region on the third exon of LA1 ( Figure S2). Several T1 independent lines with homozygous mutations in LA1 were obtained, among which one line carried an LA1ΔTM type allele with a 63 bp deletion that did not cause frame shift but led to the loss of the TM domain ( Figure S2; Table S1). Compared to the WT, the LA1ΔTM mutant also showed prostrate growth of plants, mimicking the phenotypes of la1 and la1 G74V mutants ( Figure 5A).
We further wondered whether the absence of the TM domain could alter the subcellular localization of LA1ΔTM. Thus, we generated a P35S::LA1ΔTM-GFP fusion construct for the transient expression assay in rice protoplasts. Surprisingly, the results showed that removal of the TM domain did not affect the plasma membrane localization of LA1ΔTM-GFP, compared with that of LA1-GFP ( Figure 5B,C). These results support the notion that the TM domain of LA1 is essential for its biological function but not for its plasma membrane localization.

The TM Domain of LA1 Is Essential for Its Biological Function Rather Than Its Plasma Membrane Localization
To further determine the biological function of the TM domain, we constructed an LA1 knockout (designated as la1) and an LA1-TM-absence (designated as LA1∆TM) mutant in the Nipponbare cultivar via CRISPR/Cas9-mediated genome editing. To generate the LA1∆TM mutant, two editing target sites were designed to delete the TM coding region on the third exon of LA1 ( Figure S2). Several T 1 independent lines with homozygous mutations in LA1 were obtained, among which one line carried an LA1∆TM type allele with a 63 bp deletion that did not cause frame shift but led to the loss of the TM domain ( Figure S2; Table S1). Compared to the WT, the LA1∆TM mutant also showed prostrate growth of plants, mimicking the phenotypes of la1 and la1 G74V mutants ( Figure 5A).
We further wondered whether the absence of the TM domain could alter the subcellular localization of LA1∆TM. Thus, we generated a P 35S ::LA1∆TM-GFP fusion construct for the transient expression assay in rice protoplasts. Surprisingly, the results showed that removal of the TM domain did not affect the plasma membrane localization of LA1∆TM-GFP, compared with that of LA1-GFP ( Figure 5B,C). These results support the notion that the TM domain of LA1 is essential for its biological function but not for its plasma membrane localization.

Discussion
LZYs have conserved molecular functions in gravitropic responses, although they share low protein sequence similarity among plant lineages [20,23,40]. For example, At-LZY1 and AtLZY2 share only 19.8% sequence identity, yet AtLZY2 driven by the promoter of AtLZY1 can rescue the phenotype of the atlzy1 mutant [41]. These data support the notion that LZYs may share a conserved functional protein conformation. The functional redundance of LZYs in plant gravitropism is implicated in their five short conserved sequence structures, especially the unique motif "GφL(A/T)IGT" in the TM domain [25]. This is consistent with our findings that a single amino acid mutation is sufficient to eliminate shoot gravitropism (Figure 1). A similar phenomenon was observed in an AtLZY1 L92A/I94A mutant, which contained two amino acid mutations within the TM domain [42]. In addition, prediction of 3D protein structures revealed that the TM domain may be vital for the functional conformation of LA1 ( Figure 3B). In view of that LZYs are novel proteins that show no homology to any other functionally characterized proteins [7,20,23,40]; future studies will be pivotal to clarify their nature characteristics and modes of action.
Subcellular localization of a protein is essential for understanding its molecular mode of action. Emerging studies have demonstrated that subcellular localizations of LZYs are divergent. For example, while LA1, ZmLA1, and AtLZY1 localize to both the plasma membrane and nucleus, OsDRO1, OsDRL1, AtLZY2, AtLZY3, and AtLZY4 pri-

Discussion
LZYs have conserved molecular functions in gravitropic responses, although they share low protein sequence similarity among plant lineages [20,23,40]. For example, AtLZY1 and AtLZY2 share only 19.8% sequence identity, yet AtLZY2 driven by the promoter of AtLZY1 can rescue the phenotype of the atlzy1 mutant [41]. These data support the notion that LZYs may share a conserved functional protein conformation. The functional redundance of LZYs in plant gravitropism is implicated in their five short conserved sequence structures, especially the unique motif "GϕL(A/T)IGT" in the TM domain [25]. This is consistent with our findings that a single amino acid mutation is sufficient to eliminate shoot gravitropism (Figure 1). A similar phenomenon was observed in an AtLZY1 L92A/I94A mutant, which contained two amino acid mutations within the TM domain [42]. In addition, prediction of 3D protein structures revealed that the TM domain may be vital for the functional conformation of LA1 ( Figure 3B). In view of that LZYs are novel proteins that show no homology to any other functionally characterized proteins [7,20,23,40]; future studies will be pivotal to clarify their nature characteristics and modes of action.
Subcellular localization of a protein is essential for understanding its molecular mode of action. Emerging studies have demonstrated that subcellular localizations of LZYs are divergent. For example, while LA1, ZmLA1, and AtLZY1 localize to both the plasma membrane and nucleus, OsDRO1, OsDRL1, AtLZY2, AtLZY3, and AtLZY4 primarily localize only to the plasma membrane [21,23,26,34,41,42]. Recent studies showed that the NLS domain determines localization of LA1 to the nucleus [35,43], while removal of the first 100 N-terminal (containing the TM domain) or last 96 C-terminal amino acids (not containing the TM domain) eliminates its localization to the plasma membrane [7,35]. Nevertheless, Zhu et al. [43] showed that the absence of 160 C-terminal amino acids of LA1 does not impair its localization to the plasma membrane localization, while in maize, the TM domain was found to determine ZmLA1 localization to the plasma membrane [21]. Despite the high amino acid similarity of TM domains among LZYs (Figure 3A), the evidence that deleting the TM domain of LA1 did not impair its localization to the plasma membrane ( Figure 5B,C) supports the possibility that functions of the TM domain within this protein family may be highly divergent. Similarly, OsBRXL4 regulates shoot gravitropism by affecting PAT as did LA1, and its interaction with LA1 at the plasma membrane enhances the relative proportion of LA1 protein localized to the nucleus [35]. Subcellular localization of LZYs seems to be dynamic and regulated by their signal peptides as well as interactors; for example, Dong et al. [21] found that ZmLA1 in maize could interact with the auxin transport regulator protein kinase, ZmPKC, at the plasma membrane and the AUX/IAArelated protein, ZmIAA17, in the nucleus. Intriguingly, the TM domain of ZmLA1 is associated with its interaction with ZmPKC at the plasma membrane, but seemingly does not affect its interactors in the nucleus [21], suggesting that the TM domain may specifically participate in the localization of its auxin-related interactors to the plasma membrane. Taken together, these observations regarding an association between LZYs and proteins involved in auxin redistribution at the plasma membrane compared with the nucleus invite further investigation.
The endoplasmic reticulum serves as an important Ca 2+ storage organelle and is thought to regulate gravity perception and signal transduction through triggering release of Ca 2+ [38,39,[44][45][46][47]. A recent study found that the regulatory pathway "protein processing in endoplasmic reticulum" is activated early in the gravitropic response, and that the Ca 2+ response factor HSFA2D acts as an upstream positive regulator of the gravitropic response mediated by LA1 [32]. Consistent with these findings, our results showed that LA1 could localize to the endoplasmic reticulum ( Figure 3C). Thus, it appears that LA1 may play an important role in Ca 2+ -mediated signal transduction upon gravistimulation, and this merits further study. Natural genetic variations in many genes affect the response of plants to various environmental conditions by altering their expression or producing protein variants [48]. For example, genetic variations in some rice genes contribute greatly to the change in growth habit from prostrate in wild rice to the erect in cultivated rice, which consequently produces higher grain yields [9,[12][13][14]49]. The differences in the tiller angles between indica and japonica cultivars are due to the four-nucleotide substitutions in the 3 -untranslated region of TAC1, which reduces TAC1 expression in japonica and reduces tiller angle [9]. Previous studies have indicated that LA1 barely affects other agronomic traits, making it a desirable target for improving yield without unwanted consequences [7,8]. For genes such as LA1, precise genomic editing in crops can accelerate crop improvement in a targeted manner [40,50,51]. Therefore, it is feasible and worthwhile to obtain favorable alleles of LA1 for breeding improvement through editing of vital cis-elements or amino acids.

Plant Materials and Growth Conditions
The la1 G74V mutant was obtained from an ethyl methane sulfonate (EMS)-induced O. sativa ssp. japonica Nipponbare mutant library [36]. We generated an F 2 -segregating population derived from a cross between la1 G74V and an O. sativa ssp. indica Zhenshan 97B (ZS97) for fine mapping. The plants were planted in paddies under natural growing conditions in Guangzhou. The primers used for map-based cloning are listed in Table S2.

Analysis of Plant Gravity Response
We performed a gravity response assay described by Li et al. [7]. In brief, the gravity response of rice was investigated using five-day-old seedlings planted in plates with 1/2 MS medium (pH 5.8) under light and at 28 • C, and the plates were placed in an orientation to keep the seedlings vertical to the direction of gravity for at least 12 h.

RNA Extraction and qRT-PCR
Total RNA was extracted from rice seedling leaves with TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA), and reverse-transcribed into cDNA using the Tran-Script One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgen, Beijing, China). qRT-PCR was performed using a ChamQ Universal SYBR qPCR Master Mix kit (Vazyme, Nanjing, China) and the CFX Connect Real-time PCR Detection System (Bio-rad, Hercules, CA, USA) with three biological replicates. For downstream analyses of qRT-PCR results, UFC1 (LOC_Os10g13800) was used as an internal reference based on findings by Zhao et al. [52]. Relative expression was calculated using the 2 −∆Ct method [53]. The primers used for qRT-PCR are listed in Table S2.

Vector Construction and Rice Transformation
To generate P Ubi ::LA1 and P Ubi ::la1 G74V constructs for the genetic complementary analysis, the coding sequences of LA1 and la1 G74V were amplified using the primer pair LA1-OE-F/LA1-OE-R (Table S1) and cloned into an overexpression vector. Recombined vectors driven by the maize Ubiquitin promoter were introduced into Agrobacterium and then transformed into the la1 G74V mutants. The design of primers for the target sequences, and a vector for CRISPR/Cas9-mediated genomic editing of LA1, and the decoding of targeted mutation events, were performed with CRISPR-GE (http://skl.scau.edu.cn/, accessed on 16 May 2021) [54]. For deleting the TM-coding region of LA1 or knocking out LA1, we constructed a CRISPR/Cas9 editing vector as described by Ma et al. [55]. We introduced the recombined vector into Agrobacterium and then transformed into Nipponbare. The primers used to generate the transformation vectors are listed in Table S2.

Subcellular Localization
Subcellular localization prediction of a protein was performed with PSORT Prediction (http://psort1.hgc.jp/form.html). The coding sequences of LA1, la1 G74V , and LA1∆TM were amplified using the primer pair LA1-GFP-F/LA1-GFP-R (Table S2) and cloned into the N-terminus of the green fluorescent protein (GFP) gene in the pLYd1GFP vector. The recombinant vectors, which were driven by the 35S promoter, were co-expressed respectively with the nucleus-localized construct, P 35S ::NLS-mCherry, the plasma-membranelocalized construct, P 35S ::PM-mCherry or the endoplasmic-reticulum-localized construct, P 35S ::ER-mCherry in protoplasts of rice leaf sheath. This procedure was described in greater detail in Han et al. [56]. Following the transformation with GFP, we visualized and imaged the fluorescent signals of the transformed cells using a confocal microscope (Zeiss, Oberkochen, Germany).

Conclusions
In the present work, we identified and characterized a novel allele of LA1, la1 G74V , which harbors one non-synonymous mutation in the TM-domain coding region, and we investigated the biological role of the TM domain using this allele. Our results showed that mutation or removal of the TM domain resulted in loss of shoot gravitropism, but negligibly affected the localization of LA1 to the plasma membrane [7,21]. We also found that LA1 could localize to the endoplasmic reticulum, an important organelle that controls gravitropic response of plants via triggering Ca 2+ release. Our findings provide new insights into subcellular localization of LA1 and suggest that the dynamic allocation of LA1 between organelles may contribute greatly to plant gravitropism control.