Post-Golgi trafficking of rice storage proteins requires the small GTPase Rab7 activation complex MON1–CCZ1

Abstract

Protein storage vacuoles (PSVs) are unique organelles that accumulate storage proteins in plant seeds. Although morphological evidence points to the existence of multiple PSV-trafficking pathways for storage protein targeting, the molecular mechanisms that regulate these processes remain mostly unknown. Here, we report the functional characterization of the rice (Oryza sativa) glutelin precursor accumulation7 (gpa7) mutant, which over-accumulates 57-kDa glutelin precursors in dry seeds. Cytological and immunocytochemistry studies revealed that the gpa7 mutant exhibits abnormal accumulation of storage prevacuolar compartment-like structures, accompanied by the partial mistargeting of glutelins to the extracellular space. The gpa7 mutant was altered in the CCZ1 locus, which encodes the rice homolog of Arabidopsis (Arabidopsis thaliana) CALCIUM CAFFEINE ZINC SENSITIVITY1a (CCZ1a) and CCZ1b. Biochemical evidence showed that rice CCZ1 interacts with MONENSIN SENSITIVITY1 (MON1) and that these proteins function together as the Rat brain 5 (Rab5) effector and the Rab7 guanine nucleotide exchange factor (GEF). Notably, loss of CCZ1 function promoted the endosomal localization of vacuolar protein sorting-associated protein 9 (VPS9), which is the GEF for Rab5 in plants. Together, our results indicate that the MON1–CCZ1 complex is involved in post-Golgi trafficking of rice storage protein through a Rab5- and Rab7-dependent pathway.

The small GTPases Rab5- and Rab7-dependent pathway is involved in rice storage protein trafficking to vacuoles.

Introduction

Plant cells contain two morphologically distinct types of vacuoles: lytic vacuoles (LVs) and protein storage vacuoles (PSVs). LVs are mainly responsible for degradation, defense, and turgor pressure maintenance in vegetative tissues, whereas PSVs specifically function to store protein reserves in endosperm and embryonic tissues (Paris et al., 1996; Vitale and Raikhel, 1999; Jiang et al., 2001; Shimada et al., 2018). The biogenesis and diverse functions of plant vacuoles are orchestrated by the endosomal membrane system that feeds into the vacuoles (Viotti, 2014; Di Sansebastiano et al., 2018; Cui et al., 2019). In plant vegetative tissues, the cargo targeting to LVs appears to involve at least four distinct trafficking routes (Uemura and Ueda, 2014; Viotti, 2014; Minamino and Ueda, 2019). In seed cells, unique vesicular carriers, including dense vesicles (DVs, 100–200 nm in diameter), precursor accumulating vesicles (PACs, 200–400 nm in diameter), and storage prevacuolar compartments (sPVCs, 200–500 nm in diameter), efficiently transport massive amounts of storage proteins to the PSVs (Jolliffe et al., 2005; Robinson et al., 2005; Vitale and Hinz, 2005; Shen et al., 2011; Shimada et al., 2018). Although multiple PSV trafficking pathways involving these plant-unique vesicles have been proposed (Vitale and Raikhel, 1999; Vitale and Hinz, 2005; Shen et al., 2011), the molecular machineries underlying these pathways remain largely unknown.

Rice (Oryza sativa) endosperm has been used as a model system to dissect the cellular machinery for transport and accumulation of storage proteins because of its agronomic importance and the rich genetic resources available for rice (Ren et al., 2014, 2020). Rice accumulates three major types of storage proteins in developing endosperm: glutelins, α-globulin, and prolamins (Takemoto et al., 2002). Prolamins are synthesized in the endoplasmic reticulum (ER) lumen and eventually bud off from the ER as spherical protein body Is (PBIs). In contrast, glutelins are initially synthesized as 57-kDa precursors at the ER, pass through the Golgi apparatus, follow the DV-mediated post-Golgi trafficking pathway for delivery to PSVs—where they are cleaved into acidic and basic subunits by vacuolar processing enzymes—and eventually develop into irregularly shaped PBIIs together with α-globulin (Yamagata et al., 1982; Krishnan et al., 1986; Wang et al., 2009; Kumamaru et al., 2010; Liu et al., 2013; Ren et al., 2014). Morphological evidence indicates that the Golgi-dependent pathway mediated by DVs is the dominant route for the transport of glutelin and α-globulin in rice endosperm, although a direct ER-to-PSV pathway mediated by PAC vesicles has also been reported (Robinson et al., 2005; Takahashi et al., 2005; Wang et al., 2010; Liu et al., 2013; Ren et al., 2014, 2020).

Mutants, such as glutelin precursor accumulation (gpa) and glutelin precursor (glup), that over-accumulate 57-kDa glutelin precursors (also called 57H mutants) have been extensively used to identify regulators of storage protein trafficking in rice (Ueda et al., 2010; Wang et al., 2010). To date, five genes have been shown to be directly involved in vesicle-mediated glutelin transport. GPA4/GLUP2 encodes GOLGI TRANSPORT 1B (GOT1B), which interacts with Sec23 to facilitate coat protein complex II (COPII) assembly (Fukuda et al., 2016; Wang et al., 2016), whereas GPA1/GLUP4 and GPA2/GLUP6 encode the Rat brain 5a (Rab5a) and its guanine nucleotide exchange factor (GEF) Vacuolar protein sorting-associated protein 9a (VPS9a), respectively, and function in the directional targeting of DVs to the PSVs in rice endosperm cells (Wang et al., 2010; Fukuda et al., 2011, 2013; Liu et al., 2013). In addition to these widely conserved regulators of vesicular transport, two plant-unique factors function in post-Golgi trafficking of glutelins in rice (Ren et al., 2014, 2020). GPA3 encodes a trans-Golgi network (TGN) and DV dually localized protein that forms a functional protein complex with Rab5a through directly interacting with VPS9a, which is postulated to be responsible for DV maturation from the TGN (Ren et al., 2014). GPA5, a phox-homology domain-containing protein, was recently identified as a plant-specific effector of Rab5a that works cooperatively with the tethering factor class C core vacuole/endosome tethering (CORVET) complex and the VAMP727-containing soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) complex to mediate the membrane tethering and fusion of DVs with PSVs in rice endosperm (Ren et al., 2020). A hallmark of the post-Golgi sorting-deficient mutants in rice is the misfusion of DVs with the plasma membrane (Wang et al., 2010; Fukuda et al., 2011, 2013; Liu et al., 2013; Ren et al., 2014, 2020). These data combined with previous morphological evidence suggest that a Rab5a-based regulatory cascade involving GPA3, GPA2/GLUP6/VPS9a, GPA1/GLUP2/Rab5a, GPA5, CORVET, and SNARE may be responsible for direct DV-to-PSV trafficking in rice endosperm (Ren et al., 2020).

In dicots such as Arabidopsis (Arabidopsis thaliana) and pea (Pisum sativum L.), storage proteins move from DVs to pre-vacuolar compartments (PVCs)/multivesicular bodies (MVBs) before reaching the PSVs (Robinson et al., 1998; Otegui et al., 2006). This route differs from the direct DV-to-PSV transport route in rice endosperm (Ren et al., 2020). Molecular and genetic evidence from Arabidopsis points to the existence of PVCs/MVBs during the post-Golgi trafficking of 12S globulin and artificial GFP-CT24 cargo (green fluorescent protein fusion of sorting signals from αʹ-subunit of β-conglycinin). This trafficking involves sequential action of Rab5 and Rab7 mediated by the tethering complex homotypic fusion and protein sorting (HOPS) and the GEF complex MONENSIN SENSITIVITY1–CALCIUM CAFFEINE ZINC SENSITIVITY1 (MON1–CCZ1) for Rab7 (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Takemoto et al., 2018; Minamino and Ueda, 2019). These data suggest that different plant species may employ different pathways for storage protein transport (Minamino and Ueda, 2019; Ren et al., 2020). Whether the Rab5- and Rab7-dependent pathway modulated by the MON1–CCZ1 complex operates in the trafficking of glutelins in rice endosperm remains unknown, although typical PVCs/MVBs have been not observed in rice developing endosperm cells (Fukuda et al., 2011, 2013; Liu et al., 2013; Ren et al., 2014, 2020).

In this study, we characterized the rice gpa7 mutant and identified the causal locus as CCZ1, a homolog of Arabidopsis CCZ1a and CCZ1b. Mutation of rice CCZ1 disrupted the trafficking of glutelins, as evidenced by the abnormal accumulation of sPVC-like structures and the mistargeting of glutelins to the apoplast space. We showed that CCZ1 physically interacts with MON1 to form a dimeric complex that functions as the Rab5 effector and the Rab7 GEF. The mutation of CCZ1 significantly enhanced the endosomal localization of the Rab5 GEF, VPS9, in both rice and soybean (Glycine max). These findings indicate an important role of the MON1–CCZ1 complex in rice storage protein trafficking to the PSVs.

Results

The gpa7 mutant is defective in storage protein trafficking

To dissect the molecular machineries involved in rice storage protein trafficking, we characterized a rice 57H mutant named gpa7. The gpa7 mutant developed a floury-white endosperm (Figure 1, A and B), which is similar to the previously reported mutants gpa1–gpa6. Under scanning electron microscopy, the gpa7 endosperm contained round and loosely arranged starch granules compared with the densely packed and polyhedral-shaped starch granules in wild-type (WT) endosperm (Figure 1C). Immunoblotting with glutelin and α-globulin antibodies showed increased accumulation of glutelins in the form of precursors, accompanied by decreased levels of mature glutelin acidic and basic subunits as well as α-globulin in the gpa7 mutant compared with the WT (Figure 1, D and E). These results suggest that the gpa7 mutant is defective in both endosperm development and glutelin deposition.

Figure 1.

The gpa7 mutant has a defect in storage protein transport to the PSV. A and B, Comparison of representative WT and gpa7 dry seeds. Bars, 1 mm. C, Scanning electron microscopy images of transverse sections of WT and gpa7 grains. Bars, 10 μm. D, Protein profiles of WT and gpa7 dry seeds on SDS–PAGE gels stained by Coomassie blue. pGT, 57-kDa proglutelins; αGT, 40-kDa glutelin acidic subunits; αGlb, 26-kDa α-globulin; βGT, 20-kDa glutelin basic subunits; Pro, prolamins. E, Immunoblot analysis of storage proteins in WT and gpa7 dry seeds with anti-glutelin acidic subunits and anti-α-globulin antibodies. Arrowhead denotes 57-kDa proglutelins, while arrow indicates glutelin acidic subunits. F, Immunoblot analysis of WT and gpa7 dry seeds with anti-molecular chaperone (BIP1 and PDI1-1) antibodies. Anti-EF1-α antibodies were used as a loading control in (E) and (F). G, Images of WT and gpa7 seedlings grown for 5 d on half strength MS medium. Bar, 1 cm. H, Light microscopy observation of WT and gpa7 grain sections stained with Coomassie blue. Black and red arrowheads indicate irregularly shaped PBIIs and round-shaped PBIs. SG, starch grains. Bars, 5 μm. I, Immunofluorescence microscopy images showing the distribution of glutelins and prolamins in WT and gpa7 developing subaleurone cells. Secondary antibodies conjugated with Alexa fluor 488 (green) and Alexa fluor 555 (red) were used to trace the antigens recognized by the anti-glutelin acidic subunits and anti-prolamin antibodies. Arrows indicate PBIs, while arrowheads indicate PBIIs. PMB structures are indicated by asterisks. Bars, 5 μm. J and K, Measurement of the diameters of PBIIs (J) and PBIs (K). Values are means ± sd. Asterisks indicate the statistical significance between the WT and gpa7 values, as determined by Student’s t test (**P < 0.01; n > 300).

Elevated accumulation of molecular chaperones, such as ER lumen-BINDING PROTEIN1 (BIP1) and protein disulfide isomerase 1-1 (PDI1-1), is a common feature of rice 57H mutants defective in maturation and/or ER exit of glutelin precursors due to a stimulated unfolded protein response (Takemoto et al., 2002; Ueda et al., 2010; Ren et al., 2014, 2020; Wang et al., 2016). As shown in Figure 1F, the expression levels of BIP1 and PDI1-1 were largely comparable in WT and gpa7 endosperm, suggesting that the gpa7 mutant is more likely defective in post-Golgi trafficking of glutelins. In support of this notion, there were no substantial ultrastructural differences in the morphology of the ER, PBIs, and Golgi between WT and gpa7 developing subaleurone cells (Supplemental Figure S1). Notably, in addition to endosperm defects, the gpa7 mutant exhibited a seedling-lethal phenotype (Figure 1G), suggesting an essential role of GPA7 in vegetative development.

To explore the cellular basis of abnormal glutelin precursor accumulation in the gpa7 mutant, we conducted cytological analyses on the WT and gpa7 developing subaleurone cells at 9 d after flowering (DAF). WT subaleurone cells developed irregularly shaped PBIIs, while many small proteins and clusters of protein clumps accumulated in gpa7 subaleurone cells (Figure 1H). In contrast, there were no detectable morphological differences in PBIs between WT and gpa7 cells (Figure 1H). More strikingly, the gpa7 mutant contained abnormal protein-filled paramural bodies (PMBs) located adjacent to the cell wall, although the number of PMBs was markedly less than that of the previously described gpa1 mutant (Supplemental Figure S2). To further determine the subcellular distribution of storage proteins, we conducted immunofluorescence microscopy studies with glutelin and prolamin antibodies. In the WT , glutelins were deposited within PBIIs, while prolamins were specifically sequestered in PBIs (Figure 1I). However, in the gpa7 mutant, in addition to the distribution in clusters of protein granules in the cytoplasm, glutelins were abnormally accumulated in the PMBs and protein granules adjacent to the cell wall, suggesting that glutelins were partially mistargeted to the PMBs. As a result of glutelin mistargeting, the size of PBIIs was significantly smaller in the gpa7 mutant, but the size of PBIs was similar between the WT and the gpa7 mutant (Figure 1, J and K). These results collectively suggest that the 57H mutant gpa7 is defective in both post-Golgi trafficking of glutelins and vegetative development.

The gpa7 mutation induces the formation of enlarged sPVCs and PMBs

To further explore the trafficking defects of glutelins in the gpa7 mutant, we performed immunogold labeling of glutelins with ultrathin sections prepared from high-pressure frozen/freeze-substituted (HPF/FS) subaleurone cells at different developmental stages (Figure 2). Morphological and molecular studies have demonstrated that DVs are the major carrier vesicles for post-Golgi trafficking of rice storage proteins to the PSVs (Krishnan et al., 1986, 1992; Ren et al., 2014, 2020). As shown in Supplemental Figure S1C, glutelin-containing DVs budded off from the TGN in the gpa7 endosperm as in the WT endosperm. There was no significant difference in the size of DVs between WT and gpa7 cells (Figure 2, A, B, and F). In addition to the typical DVs with a diameter of 150–200 nm, larger DVs with a diameter of 300–450 nm were occasionally observed in the WT (Figure 2, C, E, and F). Those larger DVs are similar to the previously reported sPVCs (Shen et al., 2011, Figure 2D). Therefore, we also termed these unique organelles as sPVCs. In contrast, such sPVC-like structures were even larger (with a diameter of 300–600 nm) and readily observed in gpa7 endosperm (Figure 2, D–F). Strikingly, these sPVC contained intraluminal vesicles as typical PVCs did in vegetative tissues (Figure 2D). As a result of the abnormal accumulation of sPVC-like structures, PBIIs/PSVs were not efficiently filled by glutelins in the gpa7 mutant, and many small glutelin-containing protein granules were separately located within PBIIs/PSVs (Figure 2, G–I). In addition, we observed possible fusion between small PBIIs/PSVs in the WT (Figure 2J), while the fusion of PBIIs/PSVs appeared to be blocked in gpa7 endosperm subaleurone cells (Figure 2, K and L). Consistent with the immunofluorescence staining (Figure 1I), some glutelin-containing protein granules were observed in the apoplast space in the gpa7 mutant but not in the WT endosperm (Figure 2, M–O). During endosperm development, partial glutelins appeared to be continuously discharged into the apoplast space, and eventually formed the PMB structure (Figure 2O). Together, these observations suggest that GPA7 may function in the sPVC-mediated post-Golgi trafficking of glutelins to PSVs.

Figure 2.

Immunoelectron microscopy localization of glutelins in subaleurone cells of WT and gpa7 developing endosperm. Ultrathin sections were prepared from HPF/FS-fixed samples of WT and gpa7 developing subaleurone cells, followed by immunogold labeling using anti-glutelin acidic subunit antibodies. CW, cell wall. Bars in (A–D), 200 nm; bars in the inset in (D), 50 nm; bars in (G–H) and (M–N), 500 nm; bars in (I–L) and (O), 1 μm. A and B, Electron micrographs showing the morphology of DVs in the WT (A) and the gpa7 mutant (B). C and D, Electron micrographs showing the morphology of sPVCs in the WT (C) and the gpa7 mutant (D). The inset on the lower left represents the magnified images of the selected area in (D). Red arrowheads indicate intraluminal vesicles. E, Quantitative analysis of the relative frequency of DVs and sPVCs within the same area, expressed as the ratio of the number of sPVCs to the number of DVs. Values are means ± sd. Asterisks indicate the statistical significance between WT and gpa7 values, as determined by Student’s t test (**P < 0.01; n = 3). F, Quantitative analysis of the diameters of DVs and sPVCs in the WT and the gpa7 mutant. Asterisks indicate significant differences as determined by Student’s t test (**P < 0.01; n > 20). Boxes represent the median values, the first, and third quartiles; whiskers represent the minimum and maximum values. G–I, Electron micrographs showing the filling status of PBIIs in the WT (G) and the gpa7 mutant (H and I). J–L, Electron micrographs showing the possible fusion between PBIIs in the WT (J) and the gpa7 mutant (K and L). Red arrowheads indicate PBII membrane. M–O, Compared with the WT (M), glutelins are mistargeted to the apoplast space (N), which induces the formation of a PMB structure (O) in the gpa7 mutant.

gpa7 is mutated in a homolog of Arabidopsis CCZ1 that is broadly expressed in rice

We performed gene mapping using an F₂ population derived from a cross between the gpa7 heterozygous mutant and the japonica var. Nipponbare. Based on the analysis of 182 F₂ mutant individuals, the causal locus was delimited to a 106-kb genomic region on rice chromosome 8 that harbors nine putative candidate genes (Figure 3A). Sequencing analysis revealed a four-base deletion within the seventh exon of LOC_Os08g33076 that was predicted to produce a truncated protein containing the N-terminal 277 amino acid residues of the putative full-length protein (Figure 3, B and C). To test whether the mutation in LOC_Os08g33076 caused the gpa7 phenotypes, a complementation test was performed by introducing the WT coding sequence and a 2,222-bp upstream regulatory region of LOC_Os08g33076 into the gpa7 homozygous calli. The five positive T₁ generation transgenic lines survived and displayed WT phenotypes including grain appearance, storage protein profile, and subcellular structures (Figure 3, D–F). Therefore, we conclude that LOC_Os08g33076 is the underlying gene responsible for the gpa7 mutant phenotypes.

Figure 3.

Map-based cloning of GPA7. A, Fine mapping of the GPA7 locus. The GPA7 locus was narrowed to a 106-kb physical region between markers PT4 and PT7 on rice chromosome 8. This region includes nine predicted open reading frames. The molecular markers and the number of recombinants (Rec.) are indicated. B, Exon/intron structure and the mutation site of GPA7. The GPA7 gene comprises 10 exons (closed boxes) and 9 introns (open boxes). A 4-bp deletion in GPA7 led to early termination of translation. C, Molecular identification of gpa7 by an insertion/deletion marker. The larger-molecular-weight band represents the WT allele, and the smaller-molecular-weight band represents the gpa7 allele. D–F, A WT full-length GPA7 gene driven by its endogenous promoter (2,000 bp) completely rescues the grain appearance (D), storage protein composition pattern (E), and storage protein sorting defects (F) of the gpa7 mutant. L1 and L2 denote grains from two independent T₁ transgenic lines. Arrowhead in (E) indicates the 57-kD proglutelins. Black and red arrowheads in (F) indicate PBIIs and PBIs. SG, starch grains. Bar in (D), 1 mm; bars in (F), 5 μm. G, Schematic domain structure of the GPA7 protein.

LOC_Os08g33076 encodes a predicted DUF1712 domain-containing protein composed of 492 amino acid residues, with a calculated molecular mass of 54.86 kDa (Figure 3G). A BLAST search verified that LOC_Os08g33076 represents a single-copy gene in the rice genome. Furthermore, phylogenetic analysis showed that homologs are widely present in eukaryotes (Supplemental Figure S3). Notably, the encoded protein was homologous to Arabidopsis CCZ1a and CCZ1b (Supplemental Figure S4; Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014). Therefore, for simplicity, we named the WT rice gene CCZ1.

We next investigated the spatial expression profile of CCZ1 using reverse transcription quantitative polymerase chain reaction (RT-qPCR). As shown in Supplemental Figure S5, CCZ1 was expressed in all tissues examined, including root, stem, leaf, sheath, panicle, and in endosperm tissues at different developmental stages, with the highest expression in young panicles. During endosperm development, the expression of CCZ1 was relatively low at the early stage, peaked at 12 DAF, and then decreased, consistent with online predictions in ePlant Rice (http://bar.utoronto.ca/eplant_rice/). Taken together, we identified the gpa7 mutant causal locus as CCZ1, which is broadly expressed in rice.

CCZ1 localizes to the PVCs in transgenic rice root cells

To determine the subcellular localization of CCZ1, we generated transgenic rice plants expressing a GFP–CCZ1 fusion protein under the control of a 35S promoter in the ccz1 homozygous background. All transgenic plants displayed the WT phenotypes (Supplemental Figure S6), suggesting that GFP–CCZ1 represents a functional fusion protein in vivo. Confocal imaging showed that GFP–CCZ1 was localized to the cytosol and punctate compartments in the cytoplasm in complemented rice root tip cells (Supplemental Figure S7). To determine the nature of CCZ1-positive punctate compartments in rice plants, we transformed the Discosoma red fluorescent protein (DsRed)-tagged Arabidopsis organelle markers (Geldner et al., 2009) into the GFP–CCZ1-expressing rice plants (in the ccz1 homozygous background). As shown in Figure 4, A–C, GFP–CCZ1 colocalized with DsRed-AtRHA1 (the PVC marker) but not with DsRed-AtSYP32 (the Golgi marker) or DsRed-AtVTI12 (the TGN marker). Furthermore, immunogold microscopy examination of ultrathin sections verified CCZ1 signals on the PVCs/MVBs in the GFP–CCZ1 complemented transgenic rice root tip cells (Supplemental Figure S8). These results collectively suggest that CCZ1 is a PVC-localized protein in rice root cells.

Figure 4.

Subcellular localization of CCZ1 in transgenic rice root tip cells. A–C, Confocal microscopy images showing that GFP–CCZ1 is localized to the cytosol and punctate organelles in the cytoplasm that are obviously distinct from the Golgi marker (DsRed-AtSYP32 [A]) and the TGN marker (DsRed-AtVTI12 [B]) but colocalize with the PVC marker (DsRed-AtRHA1 [C]) in rice root tip cells. Pearson–Spearman correlation coefficients (r_s) between GFP–CCZ1 and each marker are shown in the right panels. Values are mean ± sd. n = 3. Bars, 5 μm.

CCZ1 and MON1 can form a dimeric protein complex in rice

Previous studies in animal and yeast cells revealed that MON1 forms a heterodimer with CCZ1 that functions in endosome-to-lysosome/vacuole trafficking (Nordmann et al., 2010; Gerondopoulos et al., 2012). A comparable MON1–CCZ1 complex has also been functionally characterized in Arabidopsis (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014). In addition to the single-copy CCZ1 gene, the rice genome contains one MON1 homolog (LOC_Os01g74460). To test whether CCZ1 and MON1 can form a dimeric complex in rice, we analyzed the interaction between CCZ1 and MON1 using the yeast two-hybrid (Y2H) assay. CCZ1 strongly interacted with MON1 (Figure 5A). We constructed two domain-deletion variants of MON1 and observed that the MON1 domain is necessary for the interaction between CCZ1 and MON1 (Figure 5B). The mutated gpa7/ccz1 protein failed to interact with MON1 (Figure 5, A and B), indicating that the C-terminal 215 amino acids of CCZ1 which lost in the gpa7 mutant play an essential role in mediating the interaction with MON1.

Figure 5.

CCZ1 physically interacts with MON1. A and B, Y2H assay showing that CCZ1 can interact with the full-length MON1 (A) and the MON1 domain in MON1 is necessary for the interaction (B). AD, activation domain; BD, binding domain; DDO, SD/–Trp/–Leu; QDO, SD/–Trp/–Leu/–His/–Ade. C, LCI assay showing that CCZ1 interacts with MON1 in leaf cells of N. benthamiana. Colored scale bar indicates the luminescence intensity in counts per second (cps). CL, C terminus of LUC; NL, N terminus of LUC. D, CoIP assay showing that FLAG-CCZ1 can be immunoprecipitated by GFP-MON1 in the total leaf extract of N. benthamiana with GFP-Trap beads. IB, immunoblotting.

We verified the interaction between CCZ1 and MON1 using an in vivo firefly luciferase complementation imaging (LCI) assay in Nicotiana benthamiana leaf epidermal cells (Figure 5C). Furthermore, an in vivo coimmunoprecipitation (CoIP) assay confirmed that Flag-CCZ1 can be coimmunoprecipitated by GFP–MON1 but not by free GFP in the total leaf extract of N. benthamiana with GFP-Trap (Figure 5D). Taken together, these results suggest that CCZ1 is able to form a heterodimer with MON1 in rice.

The MON1–CCZ1 complex acts as the GEF for Rab7 in rice

In yeast and animal cells, the MON1–CCZ1 complex serves as the GEF for Rab7 family members (Nordmann et al., 2010; Gerondopoulos et al., 2012; Cabrera et al., 2014). The rice genome contains five Rab7 homologs, which are further divided into two subgroups: Rab7a and Rab7b (Pitakrattananukool et al., 2012). Given that Rab7b3 is highly expressed in various tissues, including developing seeds, based on the analysis from the rice expression profile database (ePlant Rice), we selected Rab7b3 for further experiments. In Y2H assays, neither MON1 nor CCZ1 alone interacted with various forms of Rab7b3 (Figure 6A;Supplemental Figure S9A). We therefore employed yeast three-hybrid (Y3H) analysis with CCZ1 as a bridging protein to investigate the potential interaction between MON1 and Rab7b3 (Figure 6B;Supplemental Figure S9, B and C). In the presence of CCZ1, MON1 specifically interacted with the inactive GDP-fixed (T22N) form, but not with the WT and active GTP-fixed (Q67L) forms of Rab7b3 (Figure 6B). In the negative control, the mutated ccz1 protein abolished the interaction between MON1 and the GDP-fixed form of Rab7b3 (Supplemental Figure S9B). We verified the interaction of CCZ1 with the GDP-fixed form of Rab7b3 when coexpressed with MON1 using an in vivo LCI assay (Figure 6C) and bimolecular fluorescence complementation (BiFC) assay (Figure 6D;Supplemental Figure S9D) in leaf epidermal cells of N. benthamiana. In the absence of MON1, CCZ1 failed to interact with the inactive form of Rab7b3 (Figure 6, C and D). These results indicate that MON1–CCZ1 complex interact with inactive form of Rab7 in rice. The weaker interactions of MON1–CCZ1 with WT Rab7b3 most likely resulted from the high ratio of GTP:GDP in yeast cells, as was previously reported for Rab5 and its GEF VPS9a combination in plants (Goh et al., 2007; Liu et al., 2013; Wei et al., 2020b).

Figure 6.

The MON1–CCZ1 complex serves as the GEF of Rab7. A, Y2H assay showing that neither MON1 nor CCZ1 alone can interact with Rab7b3. Rab7b3^T22N, the inactive GDP-bound form of Rab7b3; Rab7b3^Q67L, the constitutively active GTP-bound form of Rab7b3; AD, activation domain; BD, binding domain; DDO, SD/–Trp/–Leu; QDO, SD/–Trp/–Leu/–His/–Ade. B, Y3H assay showing that MON1 can interact with the GDP-bound form of Rab7b3 when CCZ1 expression was induced by –Met. MON1 was constructed into the pGADT7 vector, and the different forms of Rab7b3 and CCZ1 were separately cloned into multiple cloning sites I (MCSI) and MCSII of the pBridge vector, respectively. –TLM, SD/–Trp/–Leu/–Met; –TLHA, SD/–Trp/–Leu/–His/–Ade; –TLHAM, SD/–Trp/–Leu/–His/–Ade/–Met. C, LCI assay showing that CCZ1 specifically interacts with the WT and inactive form of Rab7b3 when coexpressed with GFP-MON1 in leaf cells of N. benthamiana. Colored scale bar indicates the luminescence intensity in counts per second (cps). CL, C terminus of LUC; NL, N terminus of LUC. D, BiFC assay showing that CCZ1 specifically interacts with the inactive variant of Rab7b3 when coexpressed with mCherry-MON1 in punctate compartments in leaf epidermal cells of N. benthamiana. Bars, 20 μm. E and F, In vitro GEF activity assay of the MON1–CCZ1 complex. Nucleotide exchange on Rab7b3 (E) and Rab5a (F) was measured by monitoring tryptophan autofluorescence in the presence of 0 μM (black), 0.25 μM (red), 0.5 μM (blue), or 1 μM (green) GST-CCZ1-12×His-MON1. Rab5a was used as a negative control. I₃₄₀, relative level of tryptophan intrinsic fluorescence of Rab proteins at 340 nm. Representative data from three biological replicates are shown. The jagged lines represent the raw data, and the bold lines represent the trend lines. Values are mean ± sd.

To directly investigate GEF activity of the MON1–CCZ1 complex toward Rab7b3, we performed an in vitro GEF activity assay by monitoring the variation in intrinsic tryptophan fluorescence upon nucleotide exchange (Pan et al., 1995; Goh et al., 2007; Nordmann et al., 2010; Wei et al., 2020b). Since neither MON1 nor CCZ1 alone functioned as the Rab7b3 GEF, we expressed MON1–CCZ1 as a recombinant protein to perform the in vitro GEF assay (Supplemental Figure S9E; Cui et al., 2014). The MON1–CCZ1 fusion protein interacted with Rab7b3 (Supplemental Figure S9F), suggesting that the MON1–CCZ1 recombinant protein represents a biologically functional protein. As shown in Figure 6, E and F, the purified GST-tagged MON1–CCZ1 protein showed GEF activity for Rab7b3 in a dose-dependent manner, but not for Rab5a. Together, our results collectively suggest that the MON1–CCZ1 complex serves as a GEF to activate Rab7 in rice.

MON1 physically interacts with Rab5a

In yeast and Arabidopsis cells, the MON1–CCZ1 complex is recruited to the PVC membrane by Rab5 (Nordmann et al., 2010; Cui et al., 2014; Singh et al., 2014). To investigate the regulatory relationship between the MON1–CCZ1 complex and Rab5a, we analyzed the interaction of MON1 and CCZ1 with Rab5a using Y2H assays. MON1 directly interacted with the WT and active GTP-fixed (Q70L) forms but not with the inactive GDP-fixed (S25N) form of Rab5a, while CCZ1 failed to interact with all three forms of Rab5a (Figure 7A;Supplemental Figure S10A). Thus, MON1 appears to be an effector of GTP-fixed Rab5a.

Figure 7.

MON1 physically interacts with GTP-bound form of Rab5a. A, Y2H assay showing that MON1 specifically interacts with the WT and the constitutively active variant of Rab5a. Rab5a^S25N, inactive GDP-bound form of Rab5a; Rab5a^Q70L, constitutively active GTP-bound form of Rab5a; AD, activation domain; BD, binding domain; DDO, SD/−Trp/−Leu; QDO, SD/−Trp/−Leu/−His/−Ade. B, LCI assay showing that MON1 specifically interacts with the WT and the GTP-bound form of Rab5a in leaf cells of N. benthamiana. Colored scale bar indicates the luminescence intensity in counts per second (cps). CL, C terminus of LUC; NL, N terminus of LUC. C, BiFC assay showing that MON1 specifically interacts with the GTP-bound form of Rab5a in PVCs in leaf epidermal cells of N. benthamiana. Bars, 20 μm. D, CoIP assay showing that MON1 can be immunoprecipitated in the total protein extract of GFP-Rab5a transgenic rice endosperm with GFP-Trap beads. IB, immunoblotting.

We verified the interaction specificities of MON1 with Rab5a in the presence or absence of CCZ1 using an in vivo LCI assay in leaves of N. benthamiana (Figure 7B). Similar to previous reports (Cui et al., 2014; Singh et al., 2014), our BiFC assay in leaf epidermal cells of N. benthamiana showed that MON1 can interact with the active form of Rab5a in the PVC (Figure 7C), suggesting that MON1 might be recruited to PVCs by the active form of Rab5a. Furthermore, in vivo CoIP assays confirmed the interaction of MON1 alone and Rab5a in leaf epidermal cells of N. benthamiana and developing rice endosperm cells (Figure 7D;Supplemental Figure S10, B and C). Together, these data suggest that the MON1–CCZ1 complex acts as an effector for Rab5a, with MON1 acting as the direct interaction partner of active Rab5a.

The CCZ1 mutation promotes the endosomal localization of VPS9a in plants

Rab5a and its GEF VPS9a function cooperatively in the post-Golgi trafficking of rice storage proteins (Wang et al., 2010; Liu et al., 2013). To investigate the effect of the CCZ1 mutation on Rab5a and VPS9a, we expressed VPS9a-GFP in WT and gpa7 backgrounds. Notably, mutation of CCZ1 enhanced the endosomal localization of VPS9a in rice root cells (Figure 8A). Furthermore, immunogold microscopy examination of ultrathin sections verified many more VPS9a signals on the enlarged PVCs/MVBs in gpa7 root tip cells, in contrast to the cytosolic location of VPS9a in WT (Supplemental Figure S11). Using a transient RNA interference (RNAi) system in rice protoplasts, we verified the endosomal recruitment effect of CCZ1 on VPS9a and identified the punctate compartments as PVCs (Figure 8B;Supplemental Figure S12). A similar recruitment effect was also observed in developing soybean cotyledon cells (Figure 8C).

Figure 8.

The CCZ1 depletion promoted VPS9 recruitment onto PVCs. A, Subcellular localization of OsVPS9a-GFP in WT and gpa7 root tip cells. Bars, 5 μm. B, RNAi of CCZ1 enhanced the recruitment of VPS9a from the cytosol to PVCs labeled by mCherry-VSR2. Bars, 5 µm. C, Confocal images showing subcellular localization pattern variation of GmVPS9b-GFP in developing soybean cotyledon cells after transient expression alone or with RNAi-GmCCZ1. Bars, 5 μm. D and E, Total protein extracts from 1-week-old WT and gpa7 seedlings were ultracentrifuged at 100,000g for 1 h to obtain the pellet (P100) and supernatant (S100). Immunoblot analysis of total protein with anti-VPS9a and anti-EF1-α (D); immunoblot analysis of P100 and S100 with anti-VPS9a and specific antibodies for organelle markers: anti-TIP3-1 (for tonoplast) and anti-UGPase (for the cytosol) (E).

To rule out an artifact resulting from VPS9a-GFP overexpression, we detected the subcellular distribution of endogenous VPS9a using anti-VPS9a antibodies. As shown in Figure 8D, the CCZ1 mutation slightly but obviously increased the accumulation of VPS9a. Furthermore, we conducted a subcellular fractionation assay. Consistent with the confocal microscopy of the VPS9a-GFP transgenic line in the WT background, most VPS9a was localized to the soluble fraction in WT, while only a small proportion of VPS9a was localized in the membrane fraction (Figure 8E). Notably, considerable amounts of VPS9a were translocated to the membrane fraction in the gpa7 homozygous mutant background (Figure 8E). Taken together, these results suggest that CCZ1 depletion promotes the recruitment of VPS9a onto PVCs in plants.

Discussion

The gpa7 mutant is a 57H mutant defective in post-Golgi trafficking of glutelins with different characteristics from other 57H mutants

In this study, we identified the gpa7 mutant as a 57H mutant in rice. Similar to the previously reported gpa mutants, the gpa7 mutant displayed a floury-white endosperm (Figure 1, A–C). This characteristic indicates that the gpa7 mutant likely processes glutelin normally; in contrast the W379/glup3 mutant, a loss-of-function mutant of VACUOLAR PROCESSING ENZYME1 that is defective in glutelin processing, develops a translucent endosperm (Wang et al., 2009; Kumamaru et al., 2010). Using map-based cloning, we located gpa7 on chromosome 8 (Figure 3), which is separate from the location of the previously reported rice 57H mutant loci, including Glup1–glup7, gpa1–gpa6, and gpa8 (Ueda et al., 2010; Wang et al., 2010, 2016; Liu et al., 2013; Ren et al., 2014, 2020; Zhu et al., 2019, 2021). The comparable accumulation level of BIP1 and PDI1-1 between the WT and the gpa7 mutant eliminated the possibility that gpa7 is defective in maturation and/or ER exit of glutelins (Figure 1F). Together, these lines of evidence indicate that gpa7 is a new 57H mutant with a defect in post-Golgi trafficking of glutelins.

Seven 57H mutants defective in post-Golgi trafficking of rice storage proteins have been functionally characterized, including gpa1/glup4, gpa2/glup6, gpa3, gpa5, gpa6, and gpa8 (Wang et al., 2010; Fukuda et al., 2011, 2013; Liu et al., 2013; Ren et al., 2014, 2020; Zhu et al., 2019, 2021). Although a conspicuous phenotype of these 57H mutants, including gpa7, is the mistargeting of glutelins and the formation of PMB structures, several properties distinguish the gpa7 mutant from other post-Golgi trafficking-deficient mutants. First, the amount of PMB structures in the gpa7 mutant was significantly less than that of the gpa1 mutant (Supplemental Figure S2). Conversely, many more enlarged sPVC-like structures appeared in the gpa7 mutant but not in other post-Golgi trafficking-deficient mutants, such as gpa3 and gpa5 (Ren et al., 2014, 2020). These phenotypic defects suggested that the transport from sPVCs to the next destination may be blocked in the gpa7 mutant. Second, we observed some small PBIIs that were not fully fused in the gpa7 mutant, which was rarely observed in the WT and other gpa mutants (Figure 2, J–L), suggesting that the homotypic fusion of small PBIIs may be impaired in the gpa7 mutant. Finally, PBIIs were insufficiently filled in the gpa1–gpa8 mutants due to the mistargeting of glutelins. In the previously reported gpa1–gpa3 mutants, glutelins are deposited as large and round protein lumps inside PBIIs, while PBIIs of the gpa7 mutant were filled with many small protein granules (Figure 2H). Together, these results suggest that the gpa7 mutation in CCZ1 conferred distinct effects on PSV-trafficking pathways.

The MON1–CCZ1 complex functions as the Rab5a effector and the Rab7 GEF in rice

In membrane trafficking systems of eukaryotic cells, Rab GTPases play vitally important roles in the accurate targeting and tethering of vesicular carriers to target membranes by acting as molecular switches cycling between active and inactive forms (Stenmark, 2009; Minamino and Ueda, 2019). The functions of Rab GTPases are fulfilled by GEF activation and successive effector recruitment. The Rab5 and Rab7 subfamilies of Rab GTPases are evolutionarily conserved in eukaryotes. Like their counterparts in animals and yeast, plant Rab5 and Rab7 proteins are required for vacuolar trafficking (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Minamino and Ueda, 2019). Loss-of-function mutations of Rab5a and its GEF VPS9a cause mistargeting of glutelins to the extracellular space in rice endosperm cells (Wang et al., 2010; Liu et al., 2013). Recently, we showed that GPA5 acts as a downstream effector of Rab5a to mediate the tethering and fusion of DVs with PSVs through its interaction with CORVET and SNARE complexes (Ren et al., 2020). In Arabidopsis, the MON1–CCZ1 complex has been reported to modulate the Rab5-to-Rab7 conversion via sequentially acting as the Rab5 effector and Rab7 GEF (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014). However, the underlying role of MON1 and CCZ1 in vacuolar trafficking remains unknown in rice.

In the rice genome, MON1 and CCZ1 exist as single-copy genes. Several lines of evidence substantiated that MON1 and CCZ1 can form a functional heterodimer in vitro and in vivo (Figure 5). Further, our combined biochemical experiments together with the GEF activity assay showed that the MON1–CCZ1 complex acts as the Rab7 GEF (Figure 6). MON1 specifically interacted with the active form of Rab5a (Figure 7), suggesting the Rab5a effector nature of the MON1–CCZ1 complex. Together, these results suggest that the Rab5 and Rab7-dependent vacuolar trafficking pathway, which is mediated by the MON–CCZ1 complex, is highly conserved in plants.

The specific subcellular distribution of GEFs greatly influences the spatiotemporal activities of Rab GTPases in the membrane trafficking system (Shideler et al., 2015). VPS9 has long been known as the common GEF for Rab5 members in plants (Goh et al., 2007; Fukuda et al., 2013; Liu et al., 2013; Wen et al., 2015; Wei et al., 2020b), although how VPS9 proteins are specifically targeted to the endosomal membrane remains poorly understood in plants. In yeast, the endosomal localization of VPS9 is promoted by its ubiquitin-binding CUE domain (Shideler et al., 2015), which is missing in the plant homologs of VPS9. A recent study in Arabidopsis showed that PLANT-UNIQUE RAB5 EFFECTOR 2 (PUF2) promotes the recruitment of VPS9a onto the endosomes, although a loss-of-function mutation of PUF2 caused no detectable defects in vacuolar transport of storage proteins in Arabidopsis (Ito et al., 2018). Notably, our studies showed that mutation of CCZ1 promoted the endosomal localization of VPS9a in rice (Figure 8, A and B). Using a transient RNAi strategy in developing soybean cotyledons, we also confirmed that RNAi of GmCCZ1s enhanced the endosomal localization of GmVPS9b (Figure 8C). The endosomal sorting complex required for transport (ESCRT) dysfunction in yeast causes the hyperactivation of Vps21 (a yeast Rab5 homolog) and the enlargement of class E compartments due to enhanced Vps9 activity (Russell et al., 2012). The CCZ1 mutation described herein promoted the endosomal recruitment of VPS9, opening the door to examine the determinants of VPS9 localization in the future.

The MON1–CCZ1 complex plays an important role in the post-Golgi trafficking of rice storage proteins to the PSVs

In Arabidopsis, multiple post-Golgi trafficking routes have been suggested for different cargo delivery (Minamino and Ueda, 2019). Of them, a Rab5 and Rab7-dependent route involving MON1–CCZ1 is responsible for the transport of 12S globulin, artificial vacuolar cargo GFP-CT24, and the inositol transporter INT1, while a Rab5-dependent but Rab7-independent route is required for the SNARE subunits SYP22 and VTI11 transport. In addition to the Rab5-mediated pathways, an AP-3-dependent pathway has been shown to be responsible for the transport of VAMP711, VAMP713, vacuolar sucrose transporter SUC4, and PROTEIN S-ACYL TRANSFERASE10. In addition to the genetic strategy, membrane traffic events have been dissected pharmacologically (von Kleist and Haucke, 2012; Zhu et al., 2020). For example, a recent study identified the existence of a heterogeneous population of Golgi-independent TGNs (De Caroli et al., 2020). Although these extensive efforts, our understanding of post-Golgi traffic of seed storage proteins remains limited and fragmented.

Our previous studies demonstrated that GPA3, GPA2/VPS9a, and GPA1/Rab5a function synergistically to regulate post-Golgi trafficking of glutelins to the PSVs in rice endosperm (Ren et al., 2014). The stronger sorting defects of glutelins, as was evidenced by the accumulation of glutelin precursors in gpa3 gpa1 and gpa3 gpa2 double mutants, implied an essential role for Rab5a-centered molecular machinery in glutelin targeting to the PSVs. Our recent study revealed a consecutive post-Golgi storage protein trafficking pathway through coordinated and sequential action of Rab5a, GPA5, CORVET, and VAMP727-containing SNARE complexes, which are responsible for the fusion of DVs with PSVs (Ren et al., 2020). However, the moderate sorting defect resulting from the loss-of-function GPA5 mutation suggested that other Rab5-dependent pathways also contribute to the trafficking of glutelins to the PSVs in rice endosperm.

In Arabidopsis seed cells, PVCs/MVBs acts as intermediate compartments from DVs to the PSVs (Otegui et al., 2006). Recently, a Rab5 and Rab7-dependent pathway that is mediated by the MON1–CCZ1 complex was reported to be involved in the trafficking of 12S globulin (Ebine et al., 2014). It is noted that the mon1 mutant, in addition to the partial mis-secretion of 12S globulin to the extracellular space, accumulates considerable amounts of 12S globulin in the PSVs (Cui et al., 2014). Similar phenomenon has been observed in the gpa7 mutant (Figures 1 and 2). These lines of evidence from studies in Arabidopsis and rice suggest that multiple transport pathways may contribute to the transport of storage proteins in plants. It is tempting to suppose that Rab5 employs different downstream effectors, such as GPA5 and the MON1–CCZ1 complex, to execute storage protein delivery to PSVs in rice seeds (Supplemental Figure S13). Future studies should investigate how Rab5 recruits these different downstream effectors to activate distinct post-Golgi transport pathways in rice.

Materials and methods

Plant materials

The gpa7 mutant was derived from a pool of ⁶⁰Co-irradiated lines of the indica rice (Oryza sativa) variety N22. All rice plants were grown in paddy fields during the normal growing season or in a greenhouse at the Chinese Academy of Agricultural Sciences in Beijing.

Antibodies

Anti-glutelin, anti-α-globulin, anti-BIP1, and anti-PDI1-1 antibodies were described previously (Wang et al., 2010, 2016; Ren et al., 2014, 2020). Anti-His (dilution 1:2,000; Sigma-Aldrich, H1029), anti-GFP (dilution 1:3,000; Roche, 11814460001), anti-Flag (dilution 1:3,000; Sigma-Aldrich, F1804), anti-EF-1α (dilution 1:3,000; Agrisera, AS10 934), and anti-UGPase (dilution 1:3,000; Agrisera, AS05 086) antibodies are commercially available.

Map-based cloning

An F₂ segregating population was generated from the cross between the gpa7^+/− heterozygous plant and the japonica variety Nipponbare. F₂ seeds with floury endosperm were used for DNA extraction and phenotypic analysis, as described previously (Ren et al., 2014, 2020). The GPA7 locus was narrowed to a 106-kb physical interval between insertion/deletion markers PT4 and PT7 on chromosome 8. Molecular markers used for fine mapping are shown in Supplemental Table S1.

Seed protein extraction, SDS–PAGE, and immunoblot analyses

Total seed protein extraction and sodium dodecyl sulfate-polyacrylamide gel electrophoresis （SDS–PAGE） analysis were performed as described previously (Ren et al., 2014, 2020; Wang et al., 2016). Briefly, proteins were extracted from developing or dry seeds using 0.125 M Tris–HCl, 4% (w/v) SDS, 4 M urea, and 5% (v/v) β-mercaptoethanol, pH 6.8. The proteins were resolved by SDS–PAGE on a 12.5% (w/v) uniform gel, followed by Coomassie brilliant blue staining or electrotransfer to nitrocellulose membranes. Antigen-antibody reactions were detected with the ECL Detection Kit (Thermo Fisher Scientific), followed by visualization with the ECL detection system (Odyssey-Fc, LI-COR).

Microscopy observation

Scanning electron microscopy, immunofluorescence microscopy, transmission electron microscopy, high-pressure frozen/freeze substitution, and immunogold labeling experiments were performed as described previously (Ren et al., 2014, 2020).

Briefly, for scanning electron microscopy, transverse sections of dry seeds from WT and gpa7 were sputter-coated with gold palladium, and then were observed with a Hitachi S-3400N scanning electron microscopy.

For immunofluorescence analysis, 1-mm thickness endosperm semithin sections were blocked in Tris-buffered saline-Tween (TBST, 10 mM Tris–HCl, 150 mM NaCl, and 0.05% (v/v) Tween 20, pH 7.4) containing 3% bovine serum albumin (BSA) for 1 h and then incubated with the primary antibodies (1:100 dilution) for 1 h in TBST containing 1% BSA. After washing three times with TBST, the sections were incubated with the Alexa fluor 488 (green) or Alexa fluor 555 (red) conjugated secondary antibodies (Invitrogen) for 1 h and then washed three times in TBST, followed by observation using a Zeiss LSM700 laser scanning confocal microscope. Alexa fluor 488 (green) signals with white light laser at emission wavelength of 495–530 nm were recorded with the excitation wavelength at 488 nm, Alexa fluor 555 (red) signals with white light laser at emission wavelength of 561–596 nm were recorded with the excitation wavelength at 555 nm.

For immunogold electron microscopy, developing endosperm of the WT and gpa7 were fixed by high-pressure freezing (EMPACT2, Leica) and freeze substituted with 0.2% (w/v) uranyl acetate in acetone at –85°C for 24 h, followed by a series of gradient dehydration. Then, the samples were embedded in LOWICRYL HM20 resin by UV light irradiation (AFS2, Leica). Ultrathin sections of 70-nm thickness were prepared with a Leica EM UC7 microtome. Immunogold labeling on sections was performed with primary antibodies at 50 mg/mL and 5- or 15-nm gold-conjugated secondary antibodies at 1:50 dilution (Abcam). After post-staining with aqueous uranyl acetate/lead citrate, the samples were examined using an Hitachi H7700 transmission electron microscope.

Vector construction and rice transformation

For the genetic complementation test, the WT CCZ1 coding sequence and 2,222 bp of upstream regulatory sequences were cloned into the binary vector pCUbi1390 to generate the pCCZ1:CCZ1 construct. For the GFP-tagged CCZ1 transgenic plant, the CCZ1 coding sequence was cloned into the pCAMBIA1305-GFP vector to produce the p35S:GFP-CCZ1 construct. For localization of the TIP3 protein, the coding sequence of TIP3 was cloned into a modified pCAMBIA1305-GFP vector driven by a 980-bp α-globulin promoter (Ren et al., 2014). For colocalization analyses, the coding sequences of AtSYP32, AtVTI12, and AtRHA1 were separately cloned into a modified pCAMBIA2300 vector (Ren et al., 2020) containing a DsRed tag driven by the maize (Zea mays) Ubiquitin promoter, then transformed into the p35S:GFP-CCZ1 transgenic plants. Unless indicated otherwise, all constructs were produced using an infusion cloning kit (Clontech). Constructed vectors were individually introduced into the Agrobacterium tumefaciens strain EHA105, which were used to infect calli from N22 or gpa7 heterozygous seeds.

Y2H assays and Y3H assays

The Y2H assay was performed with the MatchMaker GAL4 Two-Hybrid System (Clontech) following the manufacturer’s instructions. Briefly, the full-length coding sequences of genes of interest were cloned into pGADT7 and pGBKT7 (Clontech), and different combinations of constructs were cotransformed into the yeast (Saccharomyces cerevisiae) AH109 strain. Positive transformants were selected on synthetic dropout (SD/−Leu/−Trp, DDO) nutrient media, while the interactions were screened on SD medium (SD/−Leu/−Trp/−His/−Ade, QDO). The experiments were performed at least three times independently with similar results.

For the Y3H assay, the pBridge vector (Clontech, Cat. No. 630404) with multiple cloning site I (MCSI) fused with the GAL4 binding domain and MCSII driven by a Met-responsive promoter was used following the manufacturer’s instructions. Briefly, the full-length coding sequences for different forms of Rab7b3 were cloned into MCSI, while the full-length coding sequence of CCZ1 was cloned into MCSII. The pBridge vector together with pGADT7-MON1 were cotransformed into the yeast strain AH109. Positive transformants were selected on synthetic dropout (SD/−Leu/−Trp/−Met) nutrient media, while the interactions were screened on a synthetic dropout (SD/−Leu/−Trp/−His/−Ade/−Met) nutrient medium. The experiments were performed at least three times independently with similar results.

Firefly LCI assays

The coding sequences of MON1, CCZ1, and the different forms of Rab5a and Rab7b3 were cloned into the pCAMBIA-nLUC or pCAMBIA-cLUC vectors, and the MON1 or CCZ1 coding sequences were separately cloned into the pCAMBIA1305-GFP vector as the third protein. These constructed vectors were introduced into Agrobacterium tumefaciens strain EHA105 and then various combinations of EHA105 strains were used to infiltrate N. benthamiana leaves. The relative LUC activity was measured by using BERTHOLD TECHNOLOGIES NightSHADE LB 985, as described previously (Chen et al., 2008). The different forms of Rab5a and Rab7b3 acted as controls for each other. Each data point contains three replicates.

BiFC assays

The coding sequences of MON1, CCZ1, and the different forms of Rab5a and Rab7b3 were separately cloned into pYN1 or pYC1 vectors (Wang et al., 2016). MON1 or CCZ1 coding sequences were separately cloned into the pCAMBIA1305-mCherry vector as the third protein. These constructs were introduced into Agrobacterium strain EHA105 and then used to infiltrate N. benthamiana leaves, as described previously (Waadt and Kudla, 2008). The fluorescent signals were captured using a Zeiss LSM700 laser scanning confocal microscope. The different forms of Rab5a and Rab7b3 acted as controls for each other. GFP signals with white light laser at emission wavelength of 505–530 nm were recorded with the excitation wavelength at 488 nm, mCherry signals with white light laser at emission wavelength of 600–630 nm were recorded with the excitation wavelength at 587 nm.

CoIP assays in N. benthamiana

For CoIP assays in N. benthamiana, the coding sequences of CCZ1 and MON1 were cloned into pCAMBIA1300-221-Flag and pCAMBIA1305-GFP vectors, respectively (Ren et al., 2014). The subsequent experiments were performed as reported previously (Liu et al., 2013; Ren et al., 2020). Briefly, introducing the above construct into the A. tumefaciens EHA105 strain and infiltrating strain combination into N. benthamiana leaves. After 2–3 d, total leaf lysates were prepared in an ice-cold IP buffer (50 mM Tris-MES, pH 7.5, 1 mM MgCl2, 0.5 M Sucrose, 10 mM EDTA, 5 mM DTT, 0.1% [v/v] Nonidet P-40, and 1×Complete Protease Inhibitor Cocktail) and were incubated with GFP-Trap magnetic beads (ChromoTek) for 2 h at 4°C with shaking. After washing with IP buffer five times, the IP samples were boiled in the 5× loading buffer for SDS–PAGE use. The immunoblotting analyses of immunoprecipitated samples then were performed using anti-Flag (dilution 1:3,000) and anti-GFP (dilution 1:3,000) antibodies, respectively.

Guanine nucleotide exchange assay

We separately transformed pGEX-GST-CCZ1-12×His-MON1 for coexpression of CCZ1 and MON1 (Cui et al., 2014), pGEX-GST-Rab7b3, and pGEX-GST-Rab5a into Escherichia coli BL21 Rosetta strain (TransGen). Expression of recombinant proteins was induced with 0.5 mM isopropyl b-d-1-thiogalactopyranoside overnight at 28°C, followed by purification using the Beads^TM GSH (Beaver, Suzhou). Then the GST-tagged CCZ1-12×His-MON1 was purified again by using Ni-NTA His-bind Resin (Millipore).

Intrinsic tryptophan fluorescence measurements were performed as described previously (Goh et al., 2007; Cui et al., 2014; Wei et al., 2020b). Briefly, each purified GST–Rab protein was preloaded with GDP incubated with or without GST-CCZ1-12×His-MON1 in reaction buffer (20 mM Tris–HCl, pH 8.0, 150 mM NaCl, and 0.5 mM MgCl₂) for 100 s at 25°C, Then, GMP-PNP was added to 0.1 mM at time 0 to start the nucleotide exchange reaction. Tryptophan fluorescence was recorded at 340 nm upon excitation at 298 nm with a fluorescence spectrophotometer (model F-4500; Hitachi High Technologies). Tryptophan fluorescence at time 0 was set at 1, and the trend lines were created using Origin 2018 (OriginLab). The assay was repeated at least three times for each Rab GTPase.

Transient expression assay in developing soybean cotyledon cells

Transient expression assays in developing soybean (Glycine max) cotyledon cells were performed as described previously (Maruyama et al., 2006; Wei et al., 2020a; 2020b). Briefly, developing soybean cotyledons (30 DAF) were immersed in 70% (v/v) ethanol for 7 min for surface sterilization, followed by several rinses with sterile water. The cotyledons were placed on Petri plates containing MS agar medium. Particle bombardment was carried out using a Biolistic PDS-100/He (Bio-Rad). Each sample was bombarded at least twice. After bombardment, the soybean cotyledons were incubated on MS agar plates at 25°C in darkness for 34 h. Thin sections from surfaces of bombarded soybean cotyledons were cut with a razor blade, followed by confocal imaging using a Zeiss LSM700 laser scanning confocal microscope. GFP signals with white light laser at emission wavelength of 505–530 nm were recorded with the excitation wavelength at 488 nm, mCherry signals with white light laser at emission wavelength of 600–630 nm were recorded with the excitation wavelength at 587 nm.

Accession numbers

Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: BiP1 (Os02g0115900), PDI1-1 (Os11g0199200), CCZ1 (Os08g0427300), MON1 (Os01g0976000), Rab5a (Os12g0631100), Rab7b3 (Os05g0516600), and VPS9a (Os03g0262900).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Comparisons of ER, Golgi, and PBI morphology in developing WT and gpa7 subaleurone cells.

Supplemental Figure S2. Cell wall components were visualized by Calcofluor white staining in WT, gpa7, and gpa1 endosperm.

Supplemental Figure S3. Phylogenic tree of the GPA7 protein and its homologs in eukaryotes.

Supplemental Figure S4. Amino acid sequence alignment of GPA7 with its homologs.

Supplemental Figure S5. Expression pattern of CCZ1.

Supplemental Figure S6. Expression of GFP-tagged CCZ1 rescues the phenotype of the gpa7 mutant.

Supplemental Figure S7. Subcellular localization of CCZ1 protein in complemented rice root tip cells.

Supplemental Figure S8. Immunogold localization of GFP-CCZ1 in root tip cells.

Supplemental Figure S9. Interaction assays of MON1, CCZ1, and Rab7b3.

Supplemental Figure S10. MON1 interacts with GTP-bound Rab5a directly.

Supplemental Figure S11. Immunogold localization of OsVPS9a-GFP in WT and gpa7 root tip cells.

Supplemental Figure S12. Evaluation of the effects of transient RNAi with CCZ1 expression in rice protoplasts and developing soybean cotyledon cells.

Supplemental Figure S13. A working model depicting the role of MON1–CCZ1 complex in rice storage protein transport.

Supplemental Table S1. Primer pairs used in this study.

 

J.M.W., Y.L.R., and T.P. designed the research. T.P. and Y.H.W. screened the mutant material and cloned the gene. T.P., Y.L.R., and Y.F.W. performed HPF-FS and immunogold labeling experiments. G.X.W., Y.C., W.K.Y., and Y.Z. constructed some vectors. T.P., Y.L.R., Y.H.W, R.N.J, Z.Y.W., and Y.Z.Q. performed all other experiments. B.L.Z., C.L.L., F.W., X.H.B., Y.Z., P.C.Z., M.Y.Y., M.Z.Y., J.P.Z, S.S.Z., Z.J.C., X.Z., and L.J. provided technological assistance. Y.L.R. and T.P. analyzed the data and wrote the article.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions of Author (https://academic.oup.com/plphys/pages/general-instructions) is: Jianmin Wan (wanjm@njau.edu.cn, wanjianmin@caas.cn).