Arabidopsis HOPS subunit VPS41 carries out plant-specific roles in vacuolar transport and vegetative growth

Abstract

The homotypic fusion and protein sorting (HOPS) complex is a conserved, multi-subunit tethering complex in eukaryotic cells. In yeast and mammalian cells, the HOPS subunit vacuolar protein sorting-associated protein 41 (VPS41) is recruited to late endosomes after Ras-related protein 7 (Rab7) activation and is essential for vacuole fusion. However, whether VPS41 plays conserved roles in plants is not clear. Here, we demonstrate that in the model plant Arabidopsis (Arabidopsis thaliana), VPS41 localizes to distinct condensates in root cells in addition to its reported localization at the tonoplast. The formation of condensates does not rely on the known upstream regulators but depends on VPS41 self-interaction and is essential for vegetative growth regulation. Genetic evidence indicates that VPS41 is required for both homotypic vacuole fusion and cargo sorting from the adaptor protein complex 3, Rab5, and Golgi-independent pathways but is dispensable for the Rab7 cargo inositol transporter 1. We also show that VPS41 has HOPS-independent functions in vacuolar transport. Taken together, our findings indicate that Arabidopsis VPS41 is a unique subunit of the HOPS complex that carries out plant-specific roles in both vacuolar transport and developmental regulation.

Arabidopsis HOPS subunit VPS41 localizes at electron-dense condensates and carries out plant-specific roles in vacuolar transport and developmental regulation.

Introduction

Plant vacuoles are essential organelles and have diverse functions, including storage of secondary metabolites, protein degradation and recycling, osmoregulation, detoxification, and defense response (Eisenach et al., 2015). Newly synthesized proteins are transported to the vacuole through multiple transport pathways. Among them, the evolutionarily conserved Ras-related protein 5 (Rab5)-to-Rab7 activation cascade is the most widely studied. The MONENSIN SENSITIVITY1 (MON1)/CALCIUM CAFFEINE ZINC SENSITIVITY 1 (CCZ1) complex facilitates the Rab5-to-Rab7 transition by inactivating the Rab5 guanine nucleotide exchange factor (GEF) and activating Rab7 simultaneously and/or sequentially during endosomal maturation. Cargo proteins in this pathway include the inositol transporter 1 (INT1) and seed storage protein 12S globulin (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Uemura and Ueda, 2014; Minamino and Ueda, 2019). Apart from the canonical pathway, a Rab5 dependent but Rab7 independent pathway has also been proposed in the model plant Arabidopsis (Arabidopsis thaliana). Tonoplast localized syntaxin 22 (SYP22) is mistargeted to the cytosol in the Rab5 GEF mutant vps9a-2 but could properly reach the vacuolar membrane in the knock-out mutants of Rab7 GEF (Ebine et al., 2014; Uemura and Ueda, 2014; Minamino and Ueda, 2019). Recently, it is reported that the vacuolar H⁺-ATPase v-type proton ATPase subunit a3 (VHA-a3) is directly transported from the endoplasmic reticulum (ER) to the vacuole by bypassing the Golgi apparatus (Viotti et al., 2013). However, overexpression of a dominant-negative canonical Rab5 homolog disrupts the tonoplast targeting of VHA-a3 (Feng et al., 2017a, 2017b), indicating that both the Rab5 and Golgi-independent pathways rely on Rab5 activation. In addition to these pathways, adaptor protein complex 3 (AP-3) mediates vacuolar transport independent of Rab5 or Rab7 activation. Soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) proteins vesicle-associated membrane protein 711 (VAMP711), VAMP713, and protein S-acyltransferase 10 (PAT10) are cargos transported via the AP-3-dependent trafficking pathway (Ebine et al., 2014; Uemura and Ueda, 2014; Feng et al., 2017a, 2017b; Takemoto et al., 2018; Minamino and Ueda, 2019). All these evidence indicate that plant vacuolar trafficking systems share conserved mechanisms with other eukaryotic systems but have also developed plant unique characteristics during evolution.

Vacuole fusion is achieved by recruiting the tethering complex and SNARE proteins (Epp et al., 2011). Tethering factors physically bridge two opposing membranes, functioning in quality control before recruiting SNARE proteins for membrane fusion (Brocker et al., 2010). In yeast, the homotypic fusion and protein sorting (HOPS) complex mediates late endosome tethering and vacuole fusion (Ostrowicz et al., 2010; Plemel et al., 2011; Brocker et al., 2012). HOPS complex consists of the core subunits vacuolar protein sorting-associated protein 16 (Vps16), Vps18, Vps11, and Vps33, shared by the other tethering complex class C core vacuole/endosome tethering (CORVET) (Balderhaar and Ungermann, 2013). In addition, HOPS contains two specific Ypt7 (yeast Rab7)-binding subunits Vps39 and Vps41, whereas CORVET contains two Vps21 (yeast Rab5)-binding subunits, Vps3 and Vps8 (Epp et al., 2011; Balderhaar and Ungermann, 2013; Solinger and Spang, 2013). These different subunits target the complexes to membranes via a direct interaction with Vps21 or Ypt7 (Peplowska et al., 2007; Brocker et al., 2010; Ho and Stroupe, 2015). In both yeast and mammalian cells, HOPS complex also integrates several upstream sorting pathways for vacuolar transport, including the Rab7, AP-3, and autophagy-related pathways (Darsow et al., 2001; Cabrera et al., 2010; Wang et al., 2011; Jiang et al., 2014; Takats et al., 2014; Iaconis et al., 2020). Based on sequence alignment, single copies of the homologs of HOPS subunits are present in plants (Rojo et al., 2003). A recent study proposed that the HOPS complex in model plant Arabidopsis is mainly responsible for homotypic vacuole fusion through the interaction with the R-SNARE protein VAMP713 (Takemoto et al., 2018). Apart from that, cargo sorting in AP-3 pathway also requires proper function of the HOPS complex (Feng et al., 2017a, 2017b). However, whether HOPS subunits also regulate other vacuolar transport pathways in plants remains elusive.

HOPS subunits have been functionally characterized in Arabidopsis through the analysis of loss-of-function mutants. Knock-out of VACUOLESS1, the homolog of yeast Vps16, results in aberrant embryonic development as early as the two-cell stage (Rojo et al., 2001). Pollen transmission deficiency is observed in vps41, vps33, vps11, and vps16 mutants (Hicks et al., 2004; Hao et al., 2016; Tan et al., 2017; Brillada et al., 2018). The mutation of VPS3, VPS39, and VPS18 also leads to embryo death (Takemoto et al., 2018). These findings suggest that the HOPS subunits are essential for gametophyte and embryonic development. However, the role of HOPS subunits in postembryonic development is largely unknown.

In Arabidopsis, VPS41 is a unique subunit of the HOPS complex that controls the pollen tube–stigma interaction. Loss of VPS41 function does not interfere with pollen germination or growth in vitro but disrupts the penetration of pollen tubes into the transmitting tract (Hao et al., 2016). However, the role of VPS41 in vegetative tissues is largely unknown. A recent publication reported that the vps41 knock-down mutants have severe fragmented vacuoles, indicating that VPS41 is required for homotypic vacuole fusion (Brillada et al., 2018). In this study, we analyzed the functions of VPS41 during the vegetative stage. VPS41 localizes to subdomains of the tonoplast, and so-far uncharacterized condensates in root cells. This is different from the reported model in yeast and mammalian cells, in which the homologous VPS41 proteins are recruited to the late endosomes directly or indirectly after Rab7 activation (Cabrera et al., 2009; Balderhaar and Ungermann, 2013; Lin et al., 2014). We demonstrate that VPS41 retains the conserved roles as part of the HOPS complex for homotypic vacuole fusion but also carries out HOPS independent functions to regulate vacuolar transport. We also show that the formation of condensates depends on VPS41 self-interaction and is essential for plant vegetative growth. Our findings indicate that VPS41 is a unique subunit of the HOPS complex with plant-specific roles in vacuolar transport and developmental regulation.

Results

VPS41 is localized to subdomains of the tonoplast and distinct punctate structures

VPS41 participates in the regulation of pollen–stigma interactions (Hao et al., 2016). To explore the roles of VPS41 in plants at the vegetative stage, we examined the endomembrane dynamics of VPS41 in Arabidopsis root cells by using a previously described VPS41-GFP transgenic line. This line was generated by transforming the vps41(+/-) heterozygous mutants with VPS41 genomic fragment fused with GFP tag at the C terminal, which fully rescued the pollen transmission defects in the null-mutants with no notable developmental defects (Hao et al., 2016). In root cells, VPS41 was detected at subdomains of the tonoplast with numerous puncta in the cytoplasm (Supplemental Figure S1A, white arrowheads) and near the nuclear envelope (Supplemental Figure S1A, yellow arrowheads). Time-lapse recording revealed that the VPS41 puncta moved rapidly towards the tonoplast and quickly fused with this membrane (Supplemental Figure S1B and Supplemental Movie S1). We also examined the localization of VPS41 in different tissues. Interestingly, although the tonoplast localization of VPS41 was detected in various tissues, the punctate structures could only be detected in root meristem but were substantially reduced in root elongation zone and disappeared in leaf (Supplemental Figure S2), indicating that these special puncta are developmentally related.

To further explore the identity of puncta in the cytoplasm, we analyzed the co-localization of VPS41 with the styryl dye FM4-64 in an uptake study. FM4-64 treatment at different time points is commonly used to track endocytosis from early endosomes to late endosomes to the tonoplast. However, no obvious co-localization was observed between FM dye-labeled vesicles and the VPS41 puncta after 30 min of dye uptake (Figure 1, A and F), suggesting that VPS41 does not localize to conventional endosomes. Since the VPS41 homologs in yeast and mammalian cells are recruited to late endosomes directly or indirectly through Rab7 activation (Cabrera et al., 2009; Balderhaar and Ungermann, 2013; Lin et al., 2014), we next examined whether VPS41 also colocalizes with the Rab7-related proteins on late endosomes in plant cells. VPS41 colocalized with Rab7 homologs RABG3f and RABG3c at subdomains of the tonoplast (Figure 1, B and C, white arrowheads). However, the VPS41 puncta only occasionally colocalized with RABG3-labeled late endosomes (Figure 1, B and C), while a large portion of the puncta did not exhibit clear colocalization with RABG3s (Figure 1, B, C, and F). We also investigated whether the VPS41 puncta colocalized with the Rab5-related proteins. In model plant Arabidopsis, the canonical Rab5 homologs RHA1 and ARA7 localize to the late endosome/prevacuolar compartment (Goh et al., 2007; Ebine et al., 2011). However, the VPS41 puncta did not significantly colocalize with RHA1 or ARA7 (Figure 1, D, E, and F). We further examined the colocalization of VPS41 with late-endosomal proteins by 3D analysis. Again, there was no significant colocalization and even lower colocalization coefficiency between VPS41 and ARA7 or RABG3s after 3D reconstruction (Supplemental Figure S3 and Supplemental Movies S2–S4).

Figure 1.

VPS41 localizes to subdomains of the tonoplast and distinct puncta, which show no significant colocalization with late endosomes. A, VPS41 shows no significant colocalization with FM4-64 dye-labeled early or late endosomes. B and C, VPS41 colocalizes with Rab7 homologs at subdomains of the tonoplast but seldom overlaps with them at the puncta. White arrowheads indicate the colocalization between VPS41 and RABG3s at subdomains of the tonoplast. D and E, VPS41 shows no clear colocalization with canonical Rab5 homologs at the puncta. F, Quantification of the colocalization efficiency in lines derived from various crosses. All images are representative images from at least three repeats. Scale bars represent 10 μm. For F, 20 seedlings for each crossed line were applied for statistical analysis with PSC plug-in in ImageJ which calculates PSCs over the same size of image. Values shown are means ± SD with all individual data points plotted.

Since the VPS41 puncta did not significantly colocalize with FM dye or late endosomal proteins, we crossed VPS41-GFP plants with wave lines (Geldner et al., 2009) harboring many other organelle markers, including the Golgi markers syntaxin of plants 32 (SYP32), membrin-12 (MEMB12), and Golgi transport 1 (GOT1) as well as the early endosome/trans-Golgi network (TGN) marker vesical transport v-SNARE 12 (VTI12) (Geldner et al., 2009). VPS41-GFP was also crossed with the retromer subunit sorting nexin 1 (SNX1), which shows TGN localization (Stierhof et al., 2013). However, no significant colocalization was observed with any of the organelle markers (Supplemental Figure S4, A–E and G). We next checked whether VPS41 colocalized with the tonoplast and provacuole marker VHA-a3 (Viotti et al., 2013). VPS41 colocalized with VHA-a3 at tonoplast similar to our previous observation. However, the VPS41 puncta did not show significant colocalization with VHA-a3 (Supplemental Figure S4, F and G). We also checked whether the VPS41 puncta colocalized with ER. Interestingly, a portion of puncta was observed to be closely associated with the ER tubule (Supplemental Figure S5, top section, white arrowheads) or at nuclear envelope (Supplemental Figure S5, middle section, white arrowheads) labeled by the ER-resident transmembrane protein calnexin (Liu et al., 2017).

Since the VPS41 puncta were frequently associated with ER, we further analyzed whether they have similar characteristics as the recently reported ER- and microtubule-associated compartment (EMAC), which is also involved in vacuolar transport (Delgadillo et al., 2020). The formation of EMAC relies on the proper function of microtubule (Delgadillo et al., 2020), therefore we analyzed whether the VPS41 puncta are affected by the most common microtubule assembly inhibitor oryzalin (Morejohn et al., 1987). Oryzalin treatment (100 μM, 1 h) reduced the number of puncta per cell but a significant number of puncta were still observed (Supplemental Figure S6, A–C). In contrast, tubulin was depolymerized under the same treatment condition (Supplemental Figure S6, D and E). We also examined the response of VPS41 puncta to latrunculin B (LatB), a well-known actin polymerization inhibitor (Gibbon et al., 1999). Interestingly, the VPS41 puncta were not disrupted but induced after LatB treatment (0.5 μM, 5 h) (Supplemental Figure S7). We also noticed that the tonoplast localization of VPS41 disappeared after oryzalin or LatB treatment (Supplemental Figures S6 and S7). These results indicate that the proper function of cytoskeleton is not essential for the formation of VPS41 puncta but is required for its localization at tonoplast.

We next tested whether the VPS41 puncta could be autophagosomes or related structures. However, no obvious co-localization was observed between VPS41 and the autophagosome marker protein ATG8a (Li et al., 2014; Supplemental Figure S8A). The distribution of VPS41 was not changed in the canonical autophagy mutants atg5-1 or atg7-2 (Supplemental Figure S8B), supporting the notion that the subcellular localization of VPS41 does not depend on the canonical autophagy pathways.

Taken together, these findings indicate that the HOPS subunit VPS41 is localized to subdomains of the tonoplast and to distinct puncta that do not share common features with the organelle markers but are partially associated with ER.

VPS41 is resistant to brefeldin A treatment but sensitive to the PI3K inhibitor wortmannin

We next performed pharmaceutical experiments to further analyze the cytological properties of VPS41. Brefeldin A (BFA) is a fungal toxin that targets ADP ribosylation factors GEFs and induces aggregate formation by ectopically accumulating coatomers (Renault et al., 2003). Strikingly, VPS41 was largely resistant to BFA (50 μM) treatment, whereas the Rab5 and Rab7 homologs formed large aggregates (so-called BFA bodies) (Figure 2, A–H and M, white arrowheads point to the VPS41 puncta that are BFA resistant).

Figure 2.

VPS41 is insensitive to brefeldin A but sensitive to wortmannin treatment. A–D, The colocalization pattern of VPS41 with either Rab5 (A, B) or Rab7 (C, D) homologs after DMSO treatment. E–H, VPS41 is resistant to brefeldin A (BFA) treatment, whereas Rab5 homologs (E, F), and Rab7 homologs (G, H) are sensitive to BFA treatment. White arrowheads indicate that the VPS41 puncta are BFA insensitive and are not colocalized with the Rab5 or Rab7 homologs. I–L, VPS41 is sensitive to wortmannin (WM) treatment but seldom forms ring-like structures. In contrast, the Rab5 homologs (I, J), and Rab7 homologs (K, L) form a large number of ring-like structures after WM treatment. White arrowheads point to the WM insensitive VPS41 puncta. M, Quantification of the percentage of cells that show BFA bodies in the lines after WM treatment. N, Quantification of the percentage of cells that show ring-like structures in the lines after WM treatment. All images are representative images from at least three repeats. Scale bars represent 10 μm. For (M) and (N), 20 seedlings were selected with 10 cells per seedling. Values shown are means ± sd with all individual data points plotted. All datasets were statistically analyzed via one-way ANOVA followed by Dunnett’s multiple comparison test to identify significantly different data points (***P < 0.001).

VPS41 is reported to directly bind to phosphatidylinositol 3-phosphate

(PtdIns (3) P) (Brillada et al., 2018). Since VPS41 did not colocalize with typical late endosomal proteins, we examined whether its colocalization could be affected by altered PtdIns (3) P levels. Wortmannin (WM) is a specific inhibitor of the type III PI-3 kinase that induces endosome fusion by depleting PtdIns (3) P (Wang et al., 2009). WM treatment (33 μM WM) strongly depleted VPS41 from the tonoplast and mis-sorted VPS41 to the cytoplasm (Figure 2, I–L compared with A–D). However, when treated with WM, VPS41 seldom formed typical ring-like structures produced by endosome fusion, which is quite different from the Rab5 or Rab7 homologs (Figure 2, I–L and N). A portion of VPS41 puncta was resistant to WM, and were still observed within the cytoplasm (Figure 2, I–L, white arrowheads point to the VPS41 puncta that are WM resistant). Therefore, VPS41 and the late endosomal proteins are recruited to the proper subcellular localization via different mechanisms, although both mechanisms require PtdIns (3) P.

In summary, the pharmaceutical experiments reveal that VPS41 shows different response to BFA and WM treatment compared with the late endosomal proteins.

The VPS41 puncta are electron-dense condensates

The unique characteristics of VPS41 prompted us to further investigate the ultrastructure of the protein by using immuno-electron microscopy (immuno-EM). For this purpose, we generated a polyclonal antibody that specifically recognizes the VPS41 protein (Supplemental Figure S9A). Our data demonstrated that the gold particles labeled VPS41–GFP fusion proteins were detected at the tonoplast (Figure 3A). We also observed electron-dense structures labeled by almost saturated gold particles (Figure 3, B–F). No clear membrane was observed with these structures, indicating that VPS41 may localize at condensates. The condensates were observed to be in the cytoplasm (Figure 3, B and C), or close to vacuole (Figure 3, D and F) or associated with ER (Figure 3, D–F).

Figure 3.

VPS41 localizes at electron-dense condensates. A–F, Immuno-EM analysis of root tip cells of VPS41-GFP transgenic plants. Immuno-labeling using anti-VPS41 antibodies was performed on TEM sections, followed by labeling with secondary antibodies conjugated with 10 nm gold particles. The VPS41–GFP fusion proteins were detected at tonoplast (A), or at electron-dense condensates in the cytoplasm (B, C), or close to the vacuole (D, F), or associated with ER (D–F). G–J, CLEM imaging of root tip cells in wild-type nontransgenic plants. Immunofluorescence labeling using anti-VPS41 antibodies was performed on TEM sections, followed by labeling with secondary antibodies conjugated with green fluorophore (lane 1). About 100 nm FluoSpheres were used to indicate relative localization (lane 2, magenta dots). Lane 3 is the TEM micrograph of the same section. Merged images are shown in lane 4. G and I, CLEM imaging of a cell. H and J, Enlarged images of box 1 in (G) or Box 2 in (I), respectively. White arrowheads indicate the electron-dense condensates; black arrow points to the ER. The condensates were observed to be close to vacuole (H) or associated with ER (J). All images are representative images from at least three repeats. V, vacuole; N, nucleus. From (A) to (F), scale bars are 200 nm. G and I, scale bars are 2 μm. H and J, scale bars are 200 nm.

To exclude the possibility that the condensates were formed due to GFP fusion in the transgenic lines, we next used correlative light and electron microscopy (CLEM) to confirm the ultrastructure where VPS41 localized in the wild-type nontransgenic seedlings. The VPS41 antibody labeled the puncta in the cytoplasm in wild-type root cells (Figure 3, G–J), as also observed in the stable transgenic VPS41-GFP line. CLEM analysis of the same labeled loci revealed electron-opaque condensates (Figure 3, H and J, white arrowheads). Similar to the results from immuno-EM, these condensates were also observed to be close to vacuole (H) or associated with ER (J).

Taken together, by using immuno-EM and CLEM analysis, our data suggest that VPS41 is localized at the tonoplast, and at electron-dense condensates partially associated with ER or close to vacuole.

The VPS41 condensates have liquid-like properties

Previous studies have revealed that the membrane-less condensates in both plant and nonplant systems, such as stress granules (SGs), processing bodies (PBs), Cajal bodies, etc., are formed by multivalent interactions among proteins, or between proteins and nucleoid acids (Banani et al., 2017). The physical features of these structures are different from membrane-bound organelles in that they are not stably formed within the cells but constantly change content with the surrounding environment, therefore have liquid-like properties (Banani et al., 2017). To further explore whether the VPS41 condensates also have such characteristics, we treated the VPS41-GFP transgenic seedlings with 1,6-hexanediol (1,6-HD, 10%), a compound that is known to perturb weak hydrophobic interactions to disassemble condensates that exhibit liquid-like properties (Lu et al., 2018; Strom et al., 2018; Xie et al., 2021). The derivative 2,5-hexanediol (2,5-HD) has minimal impact and thus serves as a negative control (Lin et al., 2016; Nair et al., 2019). Strikingly, the VPS41 condensates disappeared within 2 min after 1,6-HD treatment and were translocated to the cytoplasm (Supplemental Figure S10A, left and middle panel and B). What’s more interesting, part of the VPS41 protein was observed within the nucleus (Supplemental Figure S10A, middle panel, white arrowheads). In contrast, no significant change was observed with VPS41 when treated with 2,5-HD (Supplemental Figure S10A, right panel and B). None of the late endosomal proteins, including the Rab5 and Rab7 homologs, as well as the provacuole marker VHA-a3, showed similar response to 1,6-HD in the same crossed seedlings (Supplemental Figure S10, C–L). These data support the notion that the response to 1,6-HD is a unique property of VPS41 protein. Live-cell imaging revealed that the VPS41 condensates exhibited dynamic motility. They fused to form larger condensates (Supplemental Figure S11A and Supplemental Movie S5) or split into smaller condensates (Supplemental Figure S11B and Supplemental Movie S6). The above data suggest that the VPS41 condensates have liquid-like properties that are not shared by the membrane-bound organelles.

The formation of condensates does not rely on the AP-3 or Rab7 pathways

In yeast and mammalian cells, the homologous VPS41 proteins are either recruited to the late endosomes/MVBs after Rab7 activation (Cabrera et al., 2009; Lin et al., 2014) or they mediate the direct Golgi-to-vacuole transport bypass MVBs through interaction with the AP-3 subunits (Darsow et al., 2001; Cabrera et al., 2010). Therefore, we further examined whether the subcellular localization of VPS41 required upstream factors from AP-3 or Rab7 pathways in plants. When crossed with the AP-3δ mutant pat4-2 (Feraru et al., 2010; Zwiewka et al., 2011), VPS41 could still be targeted to the tonoplast and puncta, with no significant change in the number of puncta per cell (Figure 4, A, B, and D). In contrast, VPS41 was largely depleted from the tonoplast in Rab7 GEF mutant mon1-1 (Figure 4C). Interestingly, although the tonoplast localization of VPS41 requires Rab7 activation, the punctate structures were not reduced but actually increased in mon1-1 mutant (Figure 4, C and D), indicating that Rab7 activation is important for recruiting VPS41 to the tonoplast but is dispensable for its punctate localization.

Figure 4.

The formation of condensates does not depend on the upstream regulators from AP-3 or Rab7 pathways. A–C, Subcellular localization of VPS41-GFP protein in wild-type (A), AP-3δ mutant pat4-2 (B), or Rab7 GEF mutant mon1-1(C) background. D, Quantification of the number of puncta in the genotypes mentioned above. E–G, Immuno-EM analysis of VPS41-GFP protein in Rab7 GEF mutant mon1-1. In mon1-1 mutant, VPS41–GFP protein could still be detected at electron-dense condensates in the cytoplasm (E, F) or associated with ER (G). V, vacuole. All images are representative images from at least three repeats. From (A) to (C), scale bars are 10 μm. From (E) to (G), scale bars are 200 nm. For (D), 20 seedlings with 5 cells for each genotype were used for statistical analysis by particle analysis in ImageJ. Values shown are means± sd with all individual data points plotted. All datasets were statistically analyzed via one-way ANOVA followed by Dunnett’s multiple comparison test (***P < 0.001); ns means no significant difference.

We next examined the identity of the puncta through pharmaceutical experiments. Like VPS41 in wild-type background, the puncta in mon1-1 mutants also did not colocalize with FM4-64 labeled endosomes (Supplemental Figure S12A) and were BFA insensitive (Supplemental Figure S12B). WM treatment increased the cytosolic distribution of VPS41 without forming the ring-like structures, and a substantial portion of the puncta were insensitive to WM (Supplemental Figure S12C) but were sensitive to 1,6-HD (Supplemental Figure S12D). We further analyzed the ultrastructure of VPS41 in mon1-1 background by immuno-EM. The gold particles labeled VPS41 proteins were detected at electron-dense condensates (Figure 4, E–G) and were partially associated with ER (Figure 4G), similar to the results from the noncrossed plants. This result indicates that the VPS41 puncta still retain their cytological properties when Rab7s are inactivated.

In summary, our data suggest that the formation of condensates does not rely on the AP-3 or Rab7 pathways whereas the tonoplast localization of VPS41 requires Rab7 activation.

VPS41 is essential for both the homotypic vacuole fusion and vacuolar transport

The above-mentioned data indicate that VPS41 exhibits plant unique characteristics different from the homologous proteins in yeast and mammalian cells. We next explored the cytological functions of this unique protein. Since VPS41 is a conserved subunit of HOPS complex, we analyzed whether VPS41 regulates vacuole fusion and/or vacuolar transport as reported in other eukaryotic systems. As a pollen defect appears to be a barrier for the generation of homozygous vps41 mutants (Hao et al., 2016), we designed a complementation strategy to rescue VPS41 activity specifically in vps41 pollen. A construct harboring the Lat52 promoter fused with the VPS41 coding sequence fully complemented the male transmission deficiency in vps41. The rescued homozygous mutants were viable, although strong developmental defects were observed at the postembryonic stage (Supplemental Figure S13A). Reverse transcription-quantitative PCR (RT-qPCR) analysis revealed that the transcript of VPS41 in the pollen-rescued mutants was reduced to about 5% of that in wild-type control plants (Supplemental Figure S13B). Severe vacuole fragmentation was observed in the pollen-rescued vps41 mutants (Supplemental Figure S13C), consistent with the recent publication which reported the essential role of HOPS subunits in homotypic vacuole fusion (Brillada et al., 2018; Takemoto et al., 2018).

When we crossed the pollen-rescued vps41 mutants with transgenic plants harboring labeled cargo proteins in different vacuolar sorting pathways, the AP-3 pathway cargo VAMP711 was depleted from the tonoplast. Part of the protein was also mis-sorted to the plasma membrane (Figure 5, A and E). The Rab5 pathway cargo SYP22 and the Golgi-independent pathway cargo VHA-a3 were reduced but not diminished in tonoplast localization and substantially increased their distribution in the cytoplasm (Figure 5, B, C, F, and G). In contrast, INT1 from the Rab7 pathway was properly transported to the severely fragmented tonoplasts (Figure 5, D and H). The subcellular localization of different cargos in vps41 pollen rescued mutants was further examined by co-staining with the FM4-64 dye. While cargos from Rab5, AP-3, and Golgi-independent pathways showed no obvious colocalization with the FM dye-stained tonoplast, INT1 from Rab7 pathway still largely overlapped with FM4-64 dye at the tonoplasts (Supplemental Figure S14). These results indicate that VPS41 is required for vacuolar sorting in the AP-3, Rab5, and Golgi-independent transport pathways but is dispensable for the Rab7 pathway cargo INT1. This result is surprising, as in both yeast and mammalian systems, the homologous VPS41 proteins are essential for tonoplast/lysosomal localization of Rab7 cargos (Wang et al., 2011; Brocker et al., 2012; Lin et al., 2014), indicating that VPS41 in plants has evolved distinct functions in vacuolar transport.

Figure 5.

Cargos from different vacuolar sorting pathways in vps41 pollen rescued mutants. A–D, Cargo proteins from the AP-3 (A), Rab5 (B), Golgi-independent (C), and Rab7 (D) pathways are properly targeted to the tonoplast in the wild-type background. E, Cargo protein VAMP711 from the AP-3 pathway completely loses its tonoplast localization in the vps41 pollen rescued mutants, and a portion of the protein is redirected to PM. F and G, Cargo protein SYP22 from the Rab5 pathway (F) and VHA-a3 from the Golgi-independent pathway (G) substantially reduces the localization at tonoplast and are mistargeted to the cytosol in the vps41 pollen rescued mutants. H, Cargo protein INT1 from the Rab7 pathway is still properly transported to the highly fragmented tonoplasts in the vps41 pollen rescued mutants. I and J, Severe PSV fragmentation is observed in seeds of the vps41 pollen rescued mutants. K, The processing of 12S globulin is impaired in seeds of the vps41 pollen rescued mutants. Protein samples were extracted from mature seeds of wild-type plants or vps41 pollen rescued mutants, and probed by immunoblot. Molecular weight markers are in kilodalton (kD). All images are representative images from at least three repeats. Scale bars are 10 μm.

To further analyze whether VPS41 affects protein storage vacuole (PSV) formation, autofluorescence of PSVs was examined in mature seeds of the pollen rescued vps41 mutants. The homozygous mutant seeds displayed severe PSV fragmentation compared with the Col control (Figure 5, I and J). The processing of 12S globulin seed storage protein was also severely impaired in the mutant (Figure 5K), indicating that cargo sorting to the PSV also requires proper function of VPS41. Interestingly, previous studies have shown that the transport of 12S globulin to PSV depends on the sequential activation of Rab5 and Rab7 GTPases (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Minamino and Ueda, 2019). Therefore, both INT1 and 12S globulin are considered as the cargos of Rab7 pathway (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Minamino and Ueda, 2019). Our result indicates that, despite sharing similar upstream regulators, the two cargos delivered to the lytic vacuole and PSV are not the same in terms of VPS41 dependency.

Taken together, these findings indicate that VPS41 is indispensable for homotypic vacuole fusion and cargo sorting via the Rab5, AP-3, and Golgi-independent pathways. PSV cargo also requires the proper function of VPS41, whereas the Rab7 pathway cargo INT1 is transported to the tonoplast largely independent of VPS41 activity.

VPS41 carries out HOPS independent functions in vacuolar transport

What we have shown about the subcellular localization of VPS41 and its roles in vacuolar transport are quite different from the results from yeast and mammalian cells. Therefore, we further explored whether the properties are unique to VPS41 protein itself or shared with the other HOPS subunits.

In mammalian cells, the C-terminal really interesting new genes (RING) domain of homologous Vps41 is essential for its integration into the HOPS complex via a direct interaction with Vps18 (Hunter et al., 2017). Previous study has generated the transgenic line containing the truncated genomic fragments of VPS41 without the RING domain (VPS41ΔRING) in the vps41 null-mutant background (Hao et al., 2016; Figure 6A). We also generated a polyclonal antibody that specifically recognizes VPS18 protein (Supplemental Figure S9B). Our result showed that the full-length VPS41 protein interacted with VPS18, as revealed in an in vivo co-immunoprecipitation (Co-IP) experiment. In contrast, the VPS41ΔRING truncated protein failed to coprecipitate with VPS18 (Figure 6, A and B), confirming the notion that the RING domain is also the major determinant of this interaction in Arabidopsis. We also examined whether the VPS41ΔRING truncated protein interacted with the other HOPS subunits through mass spectrometry (MS) analysis. The VPS41 full-length protein co-precipitated with all the other HOPS subunits as expected. In contrast, the interactions of VPS41ΔRING protein with the other HOPS subunits were reduced to almost background levels (Supplemental Table S1), indicating that the presence of the RING domain is essential for recruiting VPS41 to the HOPS complex.

Figure 6.

VPS41 carries out HOPS dependent and independent functions. A, Schematic map of full-length VPS41 protein and the truncated proteins. B, Immunoprecipitation and immunoblot analysis showing that full-length VPS41 protein interacts with VPS18. However, the VPS41ΔRING truncation abolishes this interaction. Protein extract from 35s::GFP seedlings is used as the negative control. Molecular weight markers are in kD. C, Quantification of the number of VPS41 puncta in VPS41-GFP and VPS41ΔRING-GFP seedlings. D, VPS41ΔRING truncated mutants display severe defects in homotypic vacuole fusion. White arrowheads point to the highly fragmented vacuoles labeled by FM4-64. E–G, Cargo proteins from the Rab5 (E), Golgi-independent (F), and Rab7 (G) pathways are properly targeted to the highly fragmented vacuoles in the VPS41ΔRING truncated mutants. H, VAMP711 from the AP-3 pathway could still be transported to the tonoplast, but part of the protein is redistributed to PM in the VPS41ΔRING truncated mutants. White arrowheads point to the mis-sorted VAMP711 protein at PM. All images are representative images from at least three repeats. Scale bars are 10 μm. For C, 20 seedlings with 10 cells for each genotype were used for statistical analysis by particle analysis in ImageJ. Values shown are means ± sd with all individual data points plotted. All datasets were statistically analyzed via two-tailed unpaired t test (***P < 0.001).

Next, we used the complementation lines with VPS41 full-length protein or VPS41ΔRING truncated protein to compare the function of VPS41 as part of the HOPS complex and its function independent of HOPS. Microscopic observation revealed that the VPS41ΔRING protein reduced but not diminished the tonoplast localization (Supplemental Figure S15). Strikingly, the number of puncta was not decreased but increased (Figure 6C and Supplemental Figure S15), indicating that the punctate localization of VPS41 does not require the association with the HOPS complex. Nonetheless, severe vacuole fragmentation was observed in this truncated mutant as confirmed by the FM4-64 dye-stained tonoplasts (Figure 6D, white arrowheads), suggesting that the integration of VPS41 into HOPS complex is essential for homotypic vacuole fusion.

We also analyzed whether the VPS41ΔRING truncated mutant interfered with the vacuolar sorting process. Surprisingly, SYP22 from Rab5 pathway, INT1 from Rab7 pathway, and VHA-a3 from Golgi-independent pathway were all properly transported to the highly fragmentated vacuoles (Figure 6, E–G) in the VPS41ΔRING truncated mutant. The AP-3 pathway cargo VAMP711 could still be targeted to the vacuoles. However, a small portion of VAMP711 was redistributed to the PM (Figure 6H, white arrowheads). Since the vps41 pollen rescued mutant disrupted all the vacuolar and PSV sorting pathways except the Rab7 pathway cargo INT1 (Figure 5) whereas only mild vacuolar transport defects were observed in VPS41ΔRING truncated mutant, we conclude that VPS41 may carry out HOPS independent functions in vacuolar transport.

Taken together, our data suggest that VPS41 preserves the conserved roles as part of the HOPS complex for homotypic vacuole fusion but also carries out HOPS independent functions in vacuolar transport.

The formation of condensates depends on VPS41 self-interaction and is essential for plant vegetative growth regulation

Since the VPS41 condensates exist when Rab7s are inactivated and when VPS41 is detached from the HOPS complex, we further explored how the condensates formed and the biological importance of the unique structures. It was reported before that the C-Terminal Clathrin Heavy-Chain Repeat (CHCR) domain regulates the self-interaction of the homologous Vps41 proteins in yeast and mammalian cells (Darsow et al., 2001; Asensio et al., 2013). The transgenic line harboring the truncated genomic fragments of VPS41 without the CHCR domain (VPS41ΔCHCR) in vps41 null-mutant background was generated in previous study (Hao et al., 2016; Figure 6A). Interestingly, the number of condensates was significantly decreased in the VPS41ΔCHCR truncated mutants (Figure 7, A and B), indicating that the formation of condensates depends on the CHCR domain. We also analyzed the association of VPS41ΔCHCR truncated protein with the other HOPS subunits by MS. The VPS41ΔCHCR protein still interacted with the other HOPS subunits, which is different from the VPS41ΔRING truncated protein (Supplemental Table S1).

Figure 7.

The formation of condensates depends on VPS41 self-interaction and is essential for plant survival. A, Representative images of the subcellular localization of VPS41-GFP and VPS41ΔCHCR-GFP fusion proteins. B, Quantification of the number of VPS41 puncta in the genotypes mentioned in (A). C, Yeast-two-hybrid experiment proves that VPS41 self-interacts with the full-length protein but not with the VPS41ΔCHCR truncated protein. The pBGKT7-53 (BD-53) and the empty prey or bait vectors are used as the positive and negative controls, respectively. D, Schematic map of the 2in1 vector used in the BiFC experiment. E, BiFC experiment shows that the full-length VPS41 protein interacts with itself at the puncta (white arrowheads) whereas the VPS41ΔCHCR truncated protein disrupts the self-interaction. The interaction of nYFP-VPS41 with cYFP is used as the negative control. F, Growth phenotypes of VPS41-GFP, VPS41ΔRING-GFP, and VPS41ΔCHCR-GFP transgenic lines at the seedling stage. G, Growth phenotypes of VPS41-GFP and VPS41ΔRING-GFP transgenic lines at the reproductive stage. All images are representative images from at least three repeats. In (A) and (E), scale bars are 10 μm. In (F) and (G), scale bars are 1 cm. For (B), 20 seedlings with 10 cells for each genotype were used for statistical analysis by particle analysis in ImageJ. Values shown are means ± sd with all individual data points plotted. All datasets were statistically analyzed via two-tailed unpaired t test (***P < 0.001).

We further examined whether VPS41 also self-interacts in Arabidopsis and if it does, whether it also depends on the CHCR domain as in other eukaryotic systems. To test the hypothesis, we first carried out yeast-two-hybrid analysis. Although using VPS41 as bait stimulated self-activation, co-transformation of VPS41 full-length protein as both prey and bait induced much stronger yeast growth when cells were diluted to lower concentration (Figure 7C). In contrast, no yeast growth was observed in the combination of VPS41ΔCHCR truncated protein with VPS41 full-length protein (Figure 7C), indicating that the CHCR domain is essential for VPS41 self-interaction.

We also performed a bimolecular fluorescence complementation (BiFC) assay in Nicotiana benthamiana leaf epidermis cells. We used the 2in1 cloning system, which contains an internal expression control (RFP) within a single vector to ensure a 1:1 expression ratio of the nYFP and cYFP halves and enables ratiometric analysis of the interaction partners (Grefen and Blatt, 2012; Karnik et al., 2013; Figure 7D). YFP signals were detected in cells coexpressing nYFP-VPS41 with cYFP-VPS41, whereas no signals were detected in cells coexpressing nYFP-VPS41ΔCHCR with cYFP-VPS41ΔCHCR (Figure 7E). Interestingly, the reconstituted YFP signal was mainly observed at the punctate structures (Figure 7E, white arrowheads), similar to the puncta in VPS41-GFP stable transgenic lines, indicating that VPS41 may self-interact at the puncta. The interaction of nYFP-VPS41 with cYFP was used as the negative control and no YFP signal was detected as expected (Figure 7E).

We next explored the biological importance of VPS41 self-interaction by analyzing the growth phenotypes of the VPS41ΔCHCR truncated mutants. The homozygous VPS41ΔCHCR mutants displayed severe defects at vegetative growth stage similar to the vps41 pollen rescued mutants and died within 2 weeks (Figure 7F). In contrast, the VPS41ΔRING protein fully rescued the null-mutants without notable growth defects at both vegetative and reproductive growth stage (Figure 7, F and G), indicating that VPS41 also has HOPS independent functions in developmental regulation. Since only the VPS41ΔCHCR but not the VPS41ΔRING mutants disrupted the condensates, the self-interaction of VPS41 protein and the localization at the condensates are essential for plant survival.

In summary, our data suggest that the formation of condensates depends on VPS41 self-interaction and is important for plant vegetative growth.

Discussion

In this study, we demonstrate that an evolutional conserved HOPS subunit VPS41 carries out plant-specific functions in vacuolar transport. VPS41 localizes to the tonoplast and thus far uncharacterized condensates. The formation of condensates requires VPS41 self-interaction but does not rely on the known upstream factors from AP-3 or Rab7 pathways. We further show that VPS41 preserves the conserved roles as part of the HOPS complex for homotypic vacuole fusion but also carries out HOPS independent functions in vacuolar transport and vegetative growth regulation. Finally, the genetic evidence indicates that the condensates exist independently of the HOPS complex and are essential for plant survival.

In the model plant A. thaliana, four vacuolar sorting pathways have been reported. Among them, the Rab5-to-Rab7 activation cascades regulate the vacuolar transport from late endosomes (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014) while the AP-3 adaptor complex mediates the cargo transport directly from Golgi to vacuole (Ebine et al., 2014; Feng et al., 2017a, 2017b; Takemoto et al., 2018). These two pathways are conserved among the eukaryotic systems (Darsow et al., 2001; Nordmann et al., 2010). In addition, two plant-specific pathways are reported, namely the Rab5 dependent and the Golgi-independent sorting pathways (Viotti et al., 2013; Ebine et al., 2014). In this study, through genetic and cytological experiments, we proved that the HOPS subunit VPS41 is essential for cargos transport from AP-3, Rab5, and Golgi-independent pathways as well as the PSV cargo 12S globulin but is dispensable for Rab7 pathway cargo INT1. This is different from the model in yeast and mammalian cells, in which the homologous Vps41 proteins are required for vacuolar/lysosomal transport of Rab7 cargos (Cabrera et al., 2009; Wang et al., 2011; Lin et al., 2014). Results from previous studies in Arabidopsis also indicate that both the PSV cargo 12S globulin and the inositol transporter INT1 are transported through Rab5 and Rab7 dependent pathway involving MON1/CCZ1a/CCZ1b complex (Cui et al., 2014; Ebine et al., 2014; Singh et al., 2014; Minamino and Ueda, 2019). However, our data revealed that INT1 could be properly transported to the highly fragmentated tonoplast in the vps41 pollen-rescued mutant while the processing of PSV cargo 12S globulin was substantially impaired in the same mutant. This indicates that the two cargos delivered to the lytic vacuole and PSV are not the same in terms of the VPS41 dependency, although they both require sequential activation of Rab5 and Rab7 GTPases. Even more interestingly, although the VPS41ΔRING protein fails to interact with the other HOPS subunits, all the tested cargos could be properly transported to the fragmented vacuoles except minor defects with the AP-3 pathway cargo VAMP711. All the pieces of evidence indicate that VPS41 carries out plant unique functions in vacuolar transport.

The HOPS complex is an evolutionary conserved protein complex. In yeast and mammalian cells, the major function of this multi-subunit complex is to tether the donor membrane from upstream vesicles with the target membrane at vacuoles or lysosomes (Balderhaar and Ungermann, 2013; Solinger and Spang, 2013). Recent studies suggest that the HOPS complex and VPS41 subunit in Arabidopsis is mainly responsible for homotypic vacuole fusion (Takemoto et al., 2018). In this study, we demonstrated that the HOPS subunit VPS41 carries out plant-specific functions in vacuolar transport. As part of the HOPS complex, VPS41 is indispensable for homotypic vacuole fusion as the VPS41ΔRING truncated mutant failed to be integrated into the HOPS complex, resulting in severe vacuole fragmentation. In addition, we provide several evidences that VPS41 also carries out HOPS independent functions. First, the number of condensates is not reduced but increased in VPS41ΔRING mutant. Second, cargos from Rab5, Rab7, and Golgi-independent pathways are properly targeted to the tonoplast in VPS41ΔRING mutant with only minor defects in the AP-3 pathway cargo VAMP711. Third, the VPS41ΔRING truncated protein fully rescued the null-mutants although the vps41 pollen rescued mutants are seedling lethal. All the pieces of evidence suggest that VPS41 interacts with the other HOPS subunits for homotypic vacuole fusion but also carries out HOPS independent functions in both vacuolar transport and developmental regulation.

In yeast and mammalian cells, the homologous Vps41 proteins are recruited to late endosomes and vacuoles/lysosomes directly or indirectly after Ypt7/Rab7 activation (Cabrera et al., 2009; Wang et al., 2011). In this study, we unexpectedly find that HOPS subunit VPS41 in Arabidopsis localizes at electron-dense condensates in addition to its known localization at tonoplast. Our results are different from the recent publication by Brillada et al. (2018). While we have also observed that a very small portion of VPS41 puncta colocalize with the late endosomal proteins similar to their findings, we highlight the differences between the two sets of proteins. Quantitative analysis from 3D imaging reveals that the majority of VPS41 puncta do not colocalize with the late endosomal proteins RHA1, ARA7, or RABG3s. The conclusion is further supported by pharmaceutical experiments. VPS41 is insensitive to BFA and seldom forms ring-like structures after WM treatment, which is very different from the responses of late endosomal proteins. The identity of the VPS41 puncta is further validated by immuno-EM and CLEM analysis in both VPS41-GFP transgenic lines and wild-type nontransgenic seedlings. We are surprised to see that the VPS41 antibody labels the electron-dense condensates that are partially associated with ER and vacuoles. These condensates are different from the recently reported EMAC because their formation is only slightly affected by the microtubule assembly inhibitor oryzalin and even induced by the actin inhibitor LatB. We also show that the condensates are very sensitive to 1,6-HD treatment and frequently undergo fusion and split, indicating that they have liquid-like property. The presence of the condensates does not rely on upstream regulators from Rab7 or AP-3 pathways and they exist even when VPS41 is detached from the HOPS complex. Domain deletion experiments highlight the importance of the structures as the VPS41ΔCHCR truncated mutant disrupts the formation of condensates and causes seedling lethality. Interestingly, the VPS41 condensates can only be observed in root meristem but disappeared in root elongation zone or leaves, indicating that the special structures are developmentally related. Even more interestingly, the VPS41 condensates are frequently detected close to the nuclear envelope and are translocated to the nucleus after 1,6-HD treatment. Future work will further investigate the biological function of the unique condensates through interactome analysis.

The presence of VPS41 at the electron-dense condensates and their association with ER and vacuoles seems to be paradox. However, recent publications from mammalian cells have revealed the close relationship between the condensates and membrane-bound organelles (Milovanovic et al., 2018; Koppers et al., 2020; Lee et al., 2020; Zhao and Zhang, 2020). On the one hand, membrane-bound organelles can provide the platform to support the formation of condensates. In vitro experiments demonstrated that condensates formed by multivalent interactions among proteins or between proteins and nucleotides occur at much lower concentration on artificial membrane than in liquid culture, therefore is more in line with the real situation within the cells (Ditlev, 2021). In neuron cells, the clustering of synaptic vesicles (SVs) is mediated by phase separation of synapsin 1 on SVs, which ensures spatial confinement of SVs at synapses (Milovanovic et al., 2018). The formation of PBs and SGs also occurs close to ER in response to stress stimulation and is regulated by ER-associated proteins (Lee et al., 2020). On the other hand, partition of vesicles into oligomers or condensates induces strong physical force to drive membrane curvature and assist cargo sorting, thereby regulating many basic cell biology processes, such as autophagy initiation (Fujioka et al., 2020; Agudo-Canalejo et al., 2021), COPII vesicles budding at ER–Golgi intermediate compartment (Johnson et al., 2015), and sorting of cargo proteins across membrane (Schmidt and Gorlich, 2015; Ouyang et al., 2020). The close association of VPS41 condensates with ER indicates that these special structures may be important regulator for ER membrane dynamics as protein oligomerization was recently reported to induce ER membrane curvature or membrane scission (Bhaskara et al., 2019; Mochida et al., 2020). Alternatively, the VPS41 condensates may function as the tether because they were detected to be associated with both ER and vacuole under immuno-EM and CLEM. More experiments need to be performed to test the above-mentioned hypothesis.

In summary, our work sheds light on the unique functions of the conserved HOPS subunit VPS41. The plant endomembrane system has evolved many plant-specific properties via multiple genomic duplications, with genes evolving distinct functions via functional differentiation. VPS41 is an excellent example of a highly conserved endomembrane protein with plant-specific characteristics. Future work will focus on exploring how the cytological features of VPS41 correspond with its physiological importance.

Materials and methods

Plant materials and growth conditions

The Arabidopsis (A. thaliana) transgenic lines VPS41-GFP, VPS41ΔCHCR-GFP, and VPS41ΔRING-GFP were described before (Hao et al., 2016). Briefly, these lines were generated by transforming the vps41 (+/-) heterozygous mutants with VPS41 full-length genomic fragment containing the 5′ regulatory sequence, intron, and exon, or genomic fragment with ΔCHCR or ΔRING truncation. The homozygous lines were selected in the T2 generation (Hao et al., 2016). The transgenic lines YFP-SYP22 (Toyooka et al., 2009) and VHA-a3-GFP (Dettmer et al., 2006) were reported before. The mon1-1 (SALK_075382) mutant was obtained from the Arabidopsis Biological Resource Center. The RFP-ARA7 and VENUS-VPS18 transgenic lines were kindly provided by Dr. Takashi Ueda, Tokyo University in Japan. The transgenic lines VHA-a3-RFP and INT1-GFP were kindly provided by Dr. Yan Zhang from Shandong Agricultural University; the calnexin-RFP (Groves et al., 2019) transgenic line was shared by Dr. Caiji Gao from Huanan Normal University; the mCherry-TUB6 (Wang et al., 2017a, 2017b) was provided by Dr. Zhaosheng Kong from Shanxi Agricultural University in China. Transgenic lines mCherry-RHA1, mCherry-RABG3c, mCherry-RABG3f, mCherry-SYP32, mCherry-GOT1, mCherry-MEMB12, YFP-VAMP711, mCherry-VAMP711, and mCherry-VTI12 were obtained from the WAVE collection (Geldner et al., 2009). Transgenic lines RFP-SYP22 and INT1-RFP were generated by transforming the wild-type Columbia-0 (Col-0) plants with pUBQ::RFP-SYP22 and pUBQ::INT1-RFP constructs, respectively. The homozygous lines were selected in the T2 generation.

The colocalization of VPS41 with late endosomal proteins or other organelle markers was analyzed in plants in the F3 generations from crosses between VPS41-GFP and RFP-ARA7, mCherry-RHA1, mCherry-RABG3c, mCherry-RABG3f, mCherry-SYP32, mCherry-GOT1, mCherry-MEMB12, mCherry-VTI12, SNX1-RFP, VHA-a3-RFP, or calnexin-RFP.

Changes in the distribution of VPS41 in potential upstream mutants were observed in homozygous F3 plants from crosses between VPS41-GFP and pat4-2 or mon1-1. Changes in the distribution of cargo proteins were observed in homozygous F3 plants from crosses between vps41 pollen rescued mutants and YFP-VAMP711, YFP-SYP22, VHA-a3-GFP or INT1-GFP, or from crosses between VPS41ΔRING-GFP truncated mutants and mCherry-VAMP711, RFP-SYP22, VHA-a3-RFP, or INT1-RFP.

The seeds were stratified for 2 d in the dark at 4°C. The seedlings were germinated and grown vertically in square plates containing 0.5× MS medium and 0.8% Phytoagar (w/v) with 1% sucrose (w/v) (pH 5.6) at 22°C under a long-day photoperiod (16-h light/8-h dark). All seedlings were in the Col-0 ecotype background.

Vector construction and Arabidopsis transformation

To generate the pUBQ::RFP-SYP22 and pUBQ::INT1-RFP constructs, the coding sequences of SYP22 and INT1 were amplified from cDNA and introduced into pDONR207, followed by the destination vectors pUBQ::RFP-DEST and pUBQ:: DEST-RFP (French et al., 2008), respectively. Primer sequences are listed in Supplemental Table S2. All constructs were introduced into Arabidopsis ecotype Col-0 via Agrobacterium-mediated transformation using the floral-dip method.

Pharmaceutical experiments

Stock solutions of 50 mM Brefeldin A (Sigma-Aldrich St. Louis, MO, USA; Cat # B6542) or 33 mM wortmannin (Sigma-Aldrich; Cat # W1628) or 100 mM oryzalin (Sigma-Aldrich; Cat # 36182) or 1 mM Latrunculin B (Abcam; Cat # ab144291) were dissolved in dimethyl sulfoxide (DMSO). About 5 mM FM4-64 dye (Invitrogen, Waltham, MA, USA; Cat # T13320) was prepared in deionized water. For short-term treatment, seedlings of different genotypes were grown on 0.5 MS solid medium for 5–6 d, and 5–6 seedlings per well were transferred to 24-well plates containing 0.5 MS liquid medium (0.5 MS, 1% sucrose [w/v], pH 5.6) and chemicals. All stock solutions were diluted in liquid 0.5 MS medium at the indicated concentrations and treatments were performed for the indicated times, with equal volumes of solvent added as the control. For the treatment with 1,6-hexanediol （aladdin; Cat # H103708) or 2,5-hexanediol (aladdin; Cat # H157400), a drop of 10% (w/v) chemical was added directly on the glass slide where the seedlings are placed on. The samples were observed under confocal microscopy immediately after adding the compound.

Confocal microscopy and statistical analysis of images

All confocal images were captured on a Leica TCS SP8 confocal laser-scanning microscope using the lightning setting provided by the manufacturer (https://www. leica-microsystems.com/products/confocal-microscopes/p/leica-tcs-sp8/). All images were obtained under laser format with 1.0 pinhole and 100 gain value. The lasers were set up to 10%–20% of the maximum intensity. GFP signals were excited at 488 nm and emission was detected at 505–545 nm. YFP signals were excited at 514 nm and emission was detected at 520–545 nm. RFP and FM4-64 signals were excited at 552 nm and emission was detected at 570–620 nm.

For the statistical analysis of the condensates in VPS41-GFP and the truncation lines or in seedlings after different treatments, the ones with two datasets were analyzed by two-tailed unpaired t test while the ones with more than two datasets were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test. The 2D and 3D colocalization analysis was performed with the Pearson and Spearman Correlation coefficients (PSCs) and scatter J plug-in in ImageJ, respectively, as previously described (French et al., 2008), which was further analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test. The analysis of BFA bodies and ring-like structures was performed by counting the structures in 20 seedlings with 10 cells per seedling. Statistical analysis was carried out by one-way ANOVA followed by Dunnett’s multiple comparison test.

Co-IP assays and MS

Co-IP was carried out as previously described (Heard et al., 2015) with minor modifications. Seven-day-old seedlings were ground in a prechilled mortar and pestle on ice with extraction buffer (150 mM Na-HEPES, pH 7.5, 10 mM EDTA, 10 mM EGTA, 17.5% [w/v] sucrose, 7.5 mM KCl, 0.01% [v/v] Igepal CA-630, 10 mM dithiothreitol, 1% [v/v] protease inhibitors [Sigma, St. Louis, MO, USA; Cat # 11836170001], 0.5% [v/v] polyvinylpolypyrrolidone), with 2 mL buffer per 1 g of tissue (fresh weight). The homogenate was filtered through two layers of Miracloth (Millipore, Burlington, MA, USA; Cat # 475855-1R), and the flow-through was centrifuged at 4°C, 6,000 g for 15 min. For the immunoprecipitation assay, 20 µL GFP-Trap Sepharose beads (ChromoTek, Munich, Germany; Cat # gta-20) were added to the supernatant, followed by incubation for 6 h at 4°C. The slurry was washed five times with prechilled extraction buffer (no polyvinylpolypyrrolidone or protease inhibitors). The slurry was collected after the last wash and proteins eluted with 5× SDS-PAGE loading buffer for immunoblot analysis or liquid chromatography-tandem MS (LC-MS/MS). The LC-MS/MS assay was conducted by Protein World Biotech Ltd. (Beijing, China).

Antibodies

To generate the anti-VPS41 and anti-VPS18 antibodies, genes encoding polypeptides corresponding to 384–789 aa of VPS41 and 560–946 aa of VPS18 were separately cloned into the pGEX-4T-1 vector (GE Healthcare, Chicago, IL, USA). Protein expression and purification were carried out following the manufacturer’s guidelines. The purified VPS41 and VPS18 proteins were sent to the antibody production company GenScript for immunization and antibody purification. For each antibody, two rabbits were immunized. The antibodies were affinity-purified using HiTrap Columns coupled with antigens. The anti-GFP antibody was purchased from Sigma (Sigma; Cat # SAB2702197). The anti-Rabbit lgG, HRP conjugated secondary antibody (Abbkine, Hubei, China; Cat # A21010) and the anti-Mouse lgG, HRP conjugated secondary antibody (Abbkine; Cat # A21020) were purchased from Abbkine.

For the immunoblot analysis, the anti-VPS18 and anti-VPS41 antibodies were used at 1:1,000 dilution and the anti-GFP antibody was used at 1:5,000 dilution. The anti-Rabbit and anti-Mouse lgG, HRP conjugated secondary antibodies were used at 1:5,000 dilution.

Immuno-electron microscopy

Plant samples are prepared using the approach of high-pressure freezing and frozen substitution as previous report (Li et al., 2017) with minor modifications. Briefly, root tips of Arabidopsis seedlings (5–6 d after germination) were frozen with a high-pressure freezer (Leica EM PACT2), followed by dehydration and contrasting in acetone containing 0.4% (w/v) uranyl acetate for 24 h at −85°C. The samples were infiltrated in acetone for 3 h before placing sequentially in 33% (v/v), 66% (v/v), and 100% (v/v) HM20 resin diluted with acetone, each for 3 h. After that, the samples were embedded in 100% HM20 at −50°C overnight and incubated for another 4 h with fresh 100% HM20 from −50°C to −35°C. Finally, the HM20 polymerization was completed under UV illumination at −35°C for 48 h.

For immuno-EM, the sample blocks were cut into 100 nm ultrathin sections and preincubated in 3% (w/v) BSA for 15 min at room temperature. The grids were treated by 2% (w/v) uranyl acetate for 90 s in darkness and followed by incubation with the primary antibody (anti-VPS41, 1:20) for 4 h at 4°C. After three times wash with 1% (w/v) BSA, secondary antibody conjugated with 10 nm gold particles (Electron Microscopy Sciences, Hatfield, PA, USA; Cat # 25109, 1:30) was applied for 1 h at room temperature. Then the sections were washed for three times with 1% (w/v) BSA and one time with ddH2O. Images were collected with Hitachi H-7700 electron microscope operating at 80 kV.

Correlative light and electron microscopy

Wild-type Col-0 Arabidopsis seedlings were grown on 0.5 MS agar for 5 d. Root tips were dissected and high-pressure-frozen with a HPM100 high-pressure freezing apparatus. After high-pressure freezing, freeze-substitution of samples was performed in an EM AFS2 freeze-substitution device, followed by embedding in LR White resin (Electron Microscopy Sciences; Cat #14381). Freeze-substitution and LR White embedding were performed as described previously (Sobol et al., 2010). Acetone supplemented with 1% (v/v) ddH2O was used as dehydration buffer. High-pressure-frozen root tips were freeze-substituted in acetone containing 0.1% (w/v) uranyl acetate at −90°C for 24 h followed by three exchanges of dehydration buffer every 24 h. The temperature was then elevated to −40°C at a rate of 2°C/h and the samples were kept at −40°C for 10 h. The temperature was then increased to 0°C at a rate of 10°C/h, followed by adding 96% (v/v) ethanol (with 4% [v/v] ddH2O) to dehydration buffer (1:1 [v:v]) for 10 min. After that, the substituted samples were washed three times at 0°C for 10 min in 96% (v/v) ethanol. The samples were infiltrated with LR White resin/96% (v/v) ethanol mixtures in a graduated volume proportion (1:2, 1:1, and 2:1) for 30 min each. The samples were then placed in pure LR White for 50 h with exchanges of the resin after following intervals: 30 min, five times after 1 h, four times after 10, 3.5, and 1 h. Finally, samples were polymerized in a 60°C oven for 24 h anaerobically.

The 300-nm semi-thin sections were cut and collected using formvar-coated nickel slot grids. Immunofluorescent labeling was performed on the grids as reported previously (Wang et al., 2017a, 2017b). Anti-VPS41 antibodies were used as primary antibodies, while Alexa Fluor 488 antibodies (Invitrogen; Cat #A-11008) were used as secondary antibodies. The grids were labeled using 0.1 µm, red fluorescent (580/605) FluoSpheres (Invitrogen; Cat # F8801). Briefly, the grids were incubated on FluoSpheres (1:4,000 in PBS) for 10 min, followed by two incubations in ddH2O, 10 min each time. Fluorescent images were acquired under a Leica TCS SP8 confocal laser-scanning microscope using the default setting provided by the manufacturer. After fluorescent imaging, sample-loaded grids were subjected to transmission electron microscopy (TEM) imaging using a Hitachi H-7650 electron microscope at 80 kV.

BiFC assays

To test protein–protein interactions with the 2in1 BiFC cloning system, the coding sequences of VPS41 or VPS41ΔCHCR were amplified by PCR and cloned into the Pdonr221-P3P2 or Pdonr221-P1P4 entry vectors. The recombinant entry vectors were ligated into the destination vector pBiFCt-2in1-NN (Grefen and Blatt, 2012) via LR reaction. All recombinant constructs were transformed into Agrobacterium tumefaciens strain GV3101 and infiltrated into N. benthamiana leaves together with the anti-silence strain p19. The plants were recovered for 2 d at 26°C and confocal images were captured under a Leica SP8 microscope. All images were obtained under the laser format with 1.0 pinhole and 100 gain value. The lasers were set up to 10% of the maximum intensity. YFP signals were excited at 514 nm and emission was detected at 520–545 nm. RFP signals were excited at 552 nm and emission was detected at 570–620 nm.

Yeast-two-hybrid assays

To test protein–protein interactions in the yeast-two-hybrid system, the coding sequences of VPS41 full-length protein and VPS41ΔCHCR truncated protein were amplified from plasmids and ligated into both the pGADT7 prey vector and pBGKT7 bait vector. Yeast-two-hybrid assays were performed following the instructions in the yeast hand book. The pBGKT7-53 and the empty prey or bait vectors were used as the positive and negative control, respectively.

RNA extraction and RT-qPCR analysis

Five-day-old seedlings of Col-0 and pLat52::VPS41-GFP/vps41(-/-) pollen rescued lines were ground with liquid nitrogen and the total RNA was extracted by TRIzol reagent (Invitrogen; Cat # 15596-026) and RNA extraction kit (BioTeKe, Beijing, China; Cat # RP1202). cDNA was synthesized with reverse transcriptase (YEASEN, Shanghai, China; Cat # 11123ES60) followed by real-time qPCR using the SYBR green mix (YEASEN; Cat # 11184ES08). Primers for PCR analysis are listed in Supplemental Table S2.

Primers

Primers used for vector construction, cloning, and genotyping in this study are listed in Supplemental Table S2.

Accession numbers

The sequence data from this article can be found in The Arabidopsis Information Resource (https://www.arabidopsis.org/) or GenBank (http://www.ncbi.nlm.nih.gov/genbank/) databases under the following accession numbers: VPS41, AT1G08190; ARA7, AT4G19640; RHA1, AT5G45130; RABG3c, AT3G16100; RABG3f, AT3G18820; MON1, AT2G28390; AP3δ, AT1G48760; AP3β, AT3G55480; VPS18, AT1G12470; VPS39, AT4G36630; VPS16, AT2G38020; VPS11, AT2G05170; VPS33, AT3G54860; GOT1, AT3G03180; MEMB12, AT5G50440; SYP32, AT3G24350; VTI12, AT1G26670; SNX1, AT5G06140; VAMP711, AT4G32150; SYP22, AT5G46860; VHA-a3, AT4G39080; INT1, AT2G43330; and CALNEXIN, AT5G61790.

Supplemental data

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

Supplemental Figure S1. VPS41 localizes to sub-domains of the tonoplast and the punctate structures, which rapidly fuse with the tonoplast.

Supplemental Figure S2. The VPS41 puncta are developmentally related.

Supplemental Figure S3. Colocalization of VPS41 with Rab5 or Rab7 homologs by 3D analysis.

Supplemental Figure S4. VPS41 does not colocalize with the Golgi, TGN, or provacuole marker proteins.

Supplemental Figure S5. The VPS41 puncta are partially associated with ER.

Supplemental Figure S6. The formation of VPS41 puncta is slightly affected by oryzalin treatment.

Supplemental Figure S7. The VPS41 puncta are induced by latrunculin B (LatB) treatment.

Supplemental Figure S8. The subcellular localization of VPS41 does not depend on the canonical autophagy pathways.

Supplemental Figure S9. The VPS41 and VPS18 antibodies specifically detect VPS41 and VPS18 in vivo.

Supplemental Figure S10. VPS41 but not the late endosomal protein or provacuole is sensitive to 1,6-hexanediol (1,6-HD) treatment.

Supplemental Figure S11. Time-lapse images of the VPS41 puncta.

Supplemental Figure S12. The VPS41 puncta largely retain the cytological properties in mon1-1 background when Rab7s are inactivated.

Supplemental Figure S13. The vps41 pollen-rescued mutants display severe defects in development and homotypic vacuole fusion.

Supplemental Figure S14. Cargos from different vacuolar sorting pathways in vps41 pollen rescued mutants and their colocalization pattern with FM4-64 dye.

Supplemental Figure S15. The VPS41ΔRING truncated mutants increase the punctate structures.

Supplemental Table S1. Immunoprecipitation and MS spectrometry detection of the interaction proteins of VPS41 in different genotypes.

Supplemental Table S2. Primers used for vector construction, cloning, and genotyping in this study.

Supplemental Movie S1. Time-lapse video shows that the VPS41 puncta move rapidly towards vacuole and fuse with the tonoplast.

Supplemental Movie S2. 3D analysis reveals that the VPS41 puncta (green) do not show obvious colocalization with RABG3c (magenta).

Supplemental Movie S3. 3D analysis reveals that the VPS41 puncta (green) do not show obvious colocalization with RABG3f (magenta).

Supplemental Movie S4. 3D analysis reveals that the VPS41 puncta (green) do not show obvious colocalization with ARA7 (magenta).

Supplemental Movie S5. Time-lapse video shows that two VPS41 puncta move towards each other and fuse into one punctum.

Supplemental Movie S6. Time-lapse video shows that a VPS41 punctum is split into two puncta.

 

R.L. and D.J. designed most of the experiments. L.J. also contributed to the experimental design. D.J. performed most of the experiments and analyzed the data. Y.H. performed the CLEM experiments. X.Z., Z.C., L.P., and S.Z. also helped with the experiments. R.L. and D.J. wrote most of the manuscript. Y.H. wrote the part of CLEM experiment. All the authors approved the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Ruixi Li (lirx@sustech.edu.cn).