The GRV2/RME-8 protein of Arabidopsis functions in the late endocytic pathway and is required for vacuolar membrane flow

Authors

  • Rebecca A. Silady,

    1. Department of Biological Sciences, Stanford University, Stanford, CA 94305, USA,
    2. Department of Plant Biology, Carnegie Institution, 260 Panama Street, Stanford, CA 94305, USA, and
    Search for more papers by this author
    • Present address: Department of Plant Breeding and Genetics, Max Planck Institute for Plant Breeding Research, Carl-von-Linné-Weg 10, 50829 Köln, Germany.

  • David W. Ehrhardt,

    1. Department of Plant Biology, Carnegie Institution, 260 Panama Street, Stanford, CA 94305, USA, and
    Search for more papers by this author
  • Karen Jackson,

    1. Institute of Molecular Plant Sciences, University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, UK
    Search for more papers by this author
  • Christine Faulkner,

    1. Institute of Molecular Plant Sciences, University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, UK
    Search for more papers by this author
  • Karl Oparka,

    1. Institute of Molecular Plant Sciences, University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, UK
    Search for more papers by this author
  • Chris R. Somerville

    Corresponding author
    1. Department of Biological Sciences, Stanford University, Stanford, CA 94305, USA,
    2. Department of Plant Biology, Carnegie Institution, 260 Panama Street, Stanford, CA 94305, USA, and
    Search for more papers by this author

(fax 650 325 6857; e-mail crs@stanford.edu).

Summary

The gravitropism defective 2 (grv2) mutants of Arabidopsis thaliana were previously characterized as exhibiting shoot agravitropism resulting from mutations in a homolog of the Caenorhabditis elegans RECEPTOR-MEDIATED ENDOCYTOSIS-8 (RME-8) gene, which is required in C. elegans for endocytosis. A fluorescent protein fusion to the GRV2 protein localized to endosomes in transgenic plants, and vacuolar morphology was altered in grv2 mutants. A defect in vacuolar membrane dynamics provides a mechanistic explanation for the gravitropic defect, and may also account for the presence of an enlarged vacuole in early embryos, together with a nutrient requirement during seedling establishment. The GRV2-positive endosomes were sensitive to Wortmannin but not brefeldin A (BFA), consistent with GRV2 operating late in the endocytic pathway, prior to delivery of vesicles to the central vacuole. The specific enlargement of GRV2:YFP structures by Wortmannin, together with biochemical data showing that GRV2 co-fractionates with pre-vacuolar markers such as PEP12/SYP21, leads us to conclude that in plants GRV2/RME-8 functions in vesicle trafficking from the multivesicular body/pre-vacuolar compartment to the lytic vacuole.

Introduction

Shoots and hypocotyls of the gravitropism defective 2 (grv2) mutants of Arabidopsis exhibit a reduced gravity response (Silady et al., 2004). This phenotype appears to be caused by reduced amyloplast sedimentation in the endodermal cells. By contrast amyloplasts in the columella cells of the root sediment properly, and roots of grv2 seedlings respond to gravity. Hypocotyls of grv2 mutants also have a slightly reduced phototropic response (Silady et al., 2004), and a reduction in apical hook maintenance (Silady, 2006), which cannot be accounted for by altered amyloplast sedimentation.

GRV2 encodes a protein with sequence similarity to an endocytosis factor, the RME-8 protein from Caenorhabditis elegans, with orthologs in Drosophila and mammalian cells (Silady et al., 2004). RME-8 is required in C. elegans for receptor-mediated endocytosis of yolk protein into growing oocytes. RME-8 is also required for fluid-phase endocytosis of a GFP marker, which is secreted into the body cavity and non-specifically endocytosed by coelomocytes in C. elegans (Zhang et al., 2001). Similarly, mutations in the Drosophila ortholog of RME-8 blocked internalization of the membrane ligand, Boss, into neighboring Sevenless receptor tyrosine kinase-expressing cells, and, in addition, reduced the uptake of endocytosis tracers. Mutations in RME-8 also enhance the rough eye phenotype of a GTP hydrolysis-defective dynamin mutant in Drosophila (Chang et al., 2004).

The mechanistic bases for the endocytosis-defective phenotypes of the rme-8 mutants are not known. GRV2 and RME-8 are large proteins of 259 and 277 kDa, respectively, with four IWN repeats and a J-domain. RME-8 localizes to the limiting membrane of large endosomes in the macrophage-like coelomocytes of C. elegans (Zhang et al., 2001), but does not have any predicted or known membrane-spanning domains. Co-localization studies in Drosophila indicated that RME-8 co-localizes with Clc, Rab5 and Rab7, which are markers for clathrin-containing organelles, early endosomes and late endosomes, respectively (Chang et al., 2004). RME-8 interacts with Hsc70-4 via the J-domain in Drosophila and mammalian cells, and may play a role as a co-chaperone in the uncoating of clathrin-coated vesicles (Chang et al., 2004; Girard et al., 2005). Recently, the GRV2 ortholog of tobacco has been found to interact with the TGB2 movement protein (MP) of potato mop-top virus (Haupt et al., 2005). Haupt et al. showed that following the targeting of the viral genome to plasmodesmata, two of the viral MPs (TGB1 and TGB2) were recycled back to the cell interior using an endocytic pathway. The direct interaction of a viral MP with GRV2 is consistent with the view that some viruses may have hijacked components of the endocytic recycling machinery, for recycling to and from plasmodesmata (Haupt et al., 2005).

Endocytosed vesicles are transported from the plasma membrane to an endosome either by fusing with an existing endosome or by maturing into an endosome. Endocytosed material is then either cycled back to the plasma membrane or transported via the pre-vacuolar compartment (PVC) to either the vacuole or the trans-Golgi network (TGN) (Lam et al., 2005). Adding to the complexity of vesicle transport, newly synthesized proteins destined for the vacuole are transported from the TGN either directly via the PVC, or are first secreted to the plasma membrane and then endocytosed and transported via the PVC to the vacuole. This results in an overlap between biosynthesis pathways and endocytosis. Indeed, the PVC contains proteins from both pathways (Neuhaus and Paris, 2005).

Additional gravitropism mutants, sgr2, sgr3 and zig (zig zag)/sgr4, also show membrane-related defects. SGR2 encodes a putative phospholipase A1, SGR3 encodes the vacuolar t-SNARE AtVAM3 and ZIG/SGR4 encodes the vacuolar v-SNARE AtVTI11 (Kato et al., 2002; Morita et al., 2002). Mutations in these genes are thought to primarily affect the tonoplast, causing alterations in vacuole morphology and dynamics (Morita et al., 2002; Yano et al., 2003), which in turn disrupts amyloplast sedimentation, resulting in the observed gravitropic defects.

In this report we describe additional phenotypic effects of grv2 mutations on embryo development and seedling establishment. We also show that GRV2 localizes to endosomes that are sensitive to the drug Wortmannin. Membranes labeled with a tonoplast GFP marker protein form aggregates in the hypocotyls of grv2 etiolated seedlings. These observations complement those in a recent study in which mutations in GVR2, called katamari2 (kam2), were independently isolated on the basis of a direct microscopic screen for altered endomembranes (Tamura et al., 2007). Together, these observations indicate a role for GRV2 in vesicle trafficking to the vacuole, and provide insights into the mechanistic basis for the various mutant phenotypes.

Results

The grv2-1 mutant exhibits a defect in embryogenesis

Embryogenesis in Arabidopsis normally follows a highly uniform pattern (Jurgens and Mayer, 1994; Mansfield and Briarty, 1991). The first cell division of grv2-1 zygotes follows the same pattern as wild type, resulting in a small apical cell and an elongated basal cell (Figure 1a,g). However, the next two rounds of cell division in the grv2-1 mutant deviate from the wild-type pattern. Rather than dividing through the center of the cell, the newly formed cell wall is shifted to one side resulting in the size of one cell being much greater than the other (Figure 1b,h). This large cell persists until the heart stage (Figure 1e). Despite the abnormal early cell divisions, grv2-1 embryos continue to develop and eventually assume a relatively normal appearance. The main defect observed at the torpedo stage was a low frequency of embryos with three cotyledons (data not shown). By the torpedo stage the enlarged cells were no longer observed (Figure 1f,l).

Figure 1.

 Differential interference microscopy of cleared embryos at sequential developmental stages. One-cell (a, g), quadrant (b, h), dermatogen (c, i), globular (d, j), heart (e, k) and torpedo (f, l) stages.
(a–f) grv2-1 mutants.
(g–h) Ler wild type. Arrows indicate the enlarged cells; asterisks mark the ends of the cell plate. Scale bars: (a–e) and (g–k), 10 μm; (f–l), 20 μm.

The grv2-1 allele, which was identified in the Landsberg erecta (Ler) background, is not fully penetrant for the embryo phenotype: an average of 77% of homozygous grv2-1 embryos exhibit the large cell. By contrast, mutant alleles in the Columbia (Col) ecotype, grv2-2, grv2-3 and grv2-4, did not exhibit this embryo phenotype (Silady, 2006). Three of five T-DNA lines with insertions in the GRV2 gene (grv2-5, grv2-8 and grv2-9) produced embryos with the enlarged cell, but at a much lower penetrance than the grv2-1 line (Table 1). The GRV2 gene complemented the embryo phenotype of the grv2-1 mutant, indicating that the defects in embryogenesis are caused by the grv2-1 mutation (Silady, 2006). Interallelic crosses between the grv2-1 allele and the grv2-2, grv2-3 and grv2-4 alleles did not result in the expected 3:1 ratio between wild-type embryos and mutant embryos in the resulting F2 populations. Rather, there was a statistically significant excess of mutant embryos (Table 2). Thus, the grv2-1 allele has a residual activity that may eventually be useful in understanding the function of the protein. The grv2-1 allele creates a premature stop codon resulting in an 85-amino-acid truncation, whereas the grv2-2, grv2-3 and grv2-4 mutants are γ-ray and fast-neutron alleles resulting in deletions and insertions (Silady, 2006).

Table 1.   Penetrance of the grv2 embryo phenotype in selfed progeny of heterozygous GRV2 T-DNA insertion lines
AlleleT-DNA lineEcotypeLocation of insertionPenetrance of grv2 embryo phenotype in self-progeny of heterozygous plantsa
  1. aP-values from two-tailed Fisher exact tests are 0.4040, 0.0007 and 1 for pair-wise comparisons between GT1669 and 905_G11, GT1669 and Salk 067162, and 905_G11 and Salk 067162, respectively. Similarly, P-values were 1.4 × 10−13, 0.33, and 0.0032 for pair-wise comparisons between grv2-1 and GT1669, grv2-1 and 905_G11, and grv2-1 and Salk 067162, respectively.

  2. n: Number of embryos screened.

grv2-1EMS alleleLerStop codon in exon 20 (Silady et al., 2004)19%, n = 315
grv2-5GT1669 (Sundaresan et al., 1995)LerExon 1 (Silady et al., 2004)3%, n = 458
grv2-6519_H08 (Sessions et al., 2002)ColExon 15 (Silady, 2006)0%, n > 90
grv2-7712_G08 (Sessions et al., 2002)ColExon 7 (Silady, 2006)0%, n > 50
grv2-8905_G11 (Sessions et al., 2002)ColIntron 10 (Silady, 2006)6%, n = 17
grv2-9Salk 067162 (Alonso et al., 2003)ColIntron 13 (Silady et al., 2004)9%, n = 159
Table 2.   The grv2-1 embryo phenotype is dominant to other alleles in F1 individuals from interallelic crosses
CrossPhenotypically mutant embryosaPhenotypically wild-type embryosχ2 test (P-value)b
  1. aThe presence of the enlarged cell was the scored phenotype.

  2. bThe observed number of embryos with the grv2 phenotype compared with the number expected from a 3:1 ratio.

grv2-1/GRV2+ self61 (19%)254<0.05
grv2-1/grv2-2 self140 (40%)207<0.005
grv2-1/grv2-3 self120 (42%)163<0.005
grv2-1/grv2-4 self176 (37%)301<0.005
grv2-3/grv2-1 self121 (42%)170<0.005

Viability of the gvr2 mutants

When germinated and grown in continuous light on agar plates without sucrose 84–98% of grv2 mutant seedlings arrested, compared with 7–36% of wild-type seedlings (Figure 2a,c, Table 3). By contrast, when grown on agar plates with 1% sucrose only 4% of the mutant seedlings arrested compared with 0% of wild-type seedlings.

Figure 2.

 Seedlings of Ler wild type (c, d) and the grv2-1 mutant (a, b) grown on MS plates with (b, d) and without (a, c) sucrose. Seedlings were grown in continuous light for 10 days. Scale bars: 2 mm.

Table 3.   Percentages of wild-type and grv2 mutant seedlings that arrested when grown with and without sucrose
Allele or ecotypeArrest
Without sucroseWith sucrose
Ler36%, n = 1070%, n = 117
grv2-198%, n = 1804%, = 175
Col 7%, n = 300%, n = 31
grv2-284%, n = 380%, n = 31
grv2-395%, n = 380%, n = 39
grv2-491%, n = 340%, n = 39

The mutant seedlings arrested in the absence of sucrose approximately 3 days after germination. The proportion of arrested seedlings and the morphology of arrested seedlings were indistinguishable among the various grv2 alleles. At the time of arrest the root had elongated to an average length of 1.11 ± 0.1 mm and the cotyledons were green, but the first true leaves had not appeared (Figure 2a). Seedlings that were transferred from plates without sucrose to plates containing sucrose within 6 days of germination recovered and developed into normal seedlings. Seedlings grown on 1% sorbitol also arrested, whereas seedlings grown on 1% glucose did not, indicating that the sucrose is necessary as a nutrient source rather than as an osmoticum.

Subcellular localization of GRV2

To detect GRV2 protein in biochemical studies, an antibody was raised to a 67-kDa region of the 7th exon of GRV2. The antibody recognized a protein of the expected size of 277 kDa on Western blots of protein extracts from wild-type leaves and flowers, but did not recognize a protein of the expected size in protein extracts from grv2-1, grv2-2, grv2-3 or grv2-4 alleles (Figures 3a and S1). There were no additional smaller bands in grv2-1, grv2-2, grv2-3 or grv2-4 alleles in either the soluble or microsomal fractions, compared with wild type, indicating that residual truncated protein was not detectable. GRV2 protein was present in the microsomal fraction (Figure 3b), and could be solubilized from that fraction by a denaturing agent (7 m urea) or a detergent (3% Triton X-100), but not by high pH (0.1 m Na2CO3, pH 11.5) or by salt (2 m NaCl; Figure 3d). This indicates that GRV2 is either a membrane spanning protein itself (as it is not released by high pH or salt), or that it engages in hydrophobic interactions with other membrane spanning protein(s).

Figure 3.

 Western blots of GRV2 in microsomal preparations and fractions from sucrose gradients.
(a) Microsomal fractions (100 000-g pellet) from Ler and Col wild types and grv2-1, grv2-2, grv2-3 and grv2-4 alleles probed with anti-GRV2 antibody.
(b) Pellets and supernatants from the 10 000-g and the 100 000-g centrifugation of protein extracts from wild-type plants probed with GRV2 and SEC12 antibodies.
(c) Microsomal fractions (100 000 g pellet) from Col wild-type and grv2-1::GRV2YFP transgenic plants probed with GRV2 and YFP antibodies.
(d) Microsomal fractions (100 000 g pellet) solubilized in extraction buffer plus an additive or in buffer alone probed with GRV2 and SEC12 antibodies; P, pellet; S, supernatant.
(e) Western blots of sucrose gradient fractions of extracts from Ler probed with the antibodies shown on the right.
(f) Western blot of sucrose gradient fractions of extracts of Col::GRV2YFP plants probed with GRV2 antibody.

To determine the subcellular localization of GRV2, the cellular membranes were fractionated on a sucrose gradient. GRV2 protein was present in the same fractions as PEP12, a t-SNARE localizing to the PVC (Figure 3e). In addition, GRV2 protein was found in the same fractions as SEC12, an endoplasmic reticulum protein, α-TIP, a tonoplast intrinsic protein, and VPS45, a t-SNARE localized to the TGN (Figure 3e; Bar-Peled and Raikhel, 1997; Bassham and Raikhel, 1998; Jauh et al., 1999). These data indicate that GRV2 does not localize to one subcellular compartment, but rather may be distributed over a number of interconnected endomembrane compartments.

Live cell imaging of GRV2

A YFP protein fusion to GRV2 was made by inserting YFP into the MluI site 153-bp upstream from the stop codon of the genomic clone that was previously used to complement grv2 mutations (Silady et al., 2004). This is the same location that was used to make an RME-8 GFP fusion that complemented the rme-8 mutation in C. elegans (Zhang et al., 2001). The fusion protein was recognized by both the GRV2 antibody and an antibody against YFP (Figure 3c). However, the GRV2::YFP construct did not complement either the embryonic or the gravitropic phenotypes of the grv2-1 mutant in 60 independent transformed lines, or the gravitropic phenotype of the grv2-4 mutant (Silady, 2006).

The GRV2::YFP protein co-localized with endogenous GRV2 in sucrose gradient fractions (Figure 3f), indicating that the two proteins have similar localization in fractionated cellular material. Confocal images of hypocotyl cells from transgenic plants expressing GRV2::YFP under the control of the GRV2 promoter showed the signal in a morphologically heterogeneous population of endosomes (Figure 4). It was not possible to distinguish between TGN, PVC or endosomes based on GRV2::YFP localization alone. However, GRV2::YFP was clearly not localized to either the plasma membrane or the endoplasmic reticulum, and showed only occasional and weak co-localization with endosomes labeled with endocytosed FM4-64 (Figure 4). Although GRV2 protein was detected in sucrose gradient fractions that contained SEC12, an endoplasmic reticulum-localized protein, these fractions also contain α-TIP, a tonoplast-localized protein, and VPS45, a TGN-localized protein. We did not detect GRV2::YFP in the endoplasmic reticulum or the tonoplast in live cell imaging; therefore, it is likely that the presence of GRV2 in these sucrose gradient fractions results from its localization to the TGN or an unknown membrane. The GRV2::YFP endosomes also showed strong motility in the cytoplasm (Figure S2).

Figure 4.

 Confocal sections of etiolated hypocotyl cells of transgenic plants expressing GRV2::YFP or treated with FM4-64.
(a, c) GRV2::YFP in hypocotyl cells, (b) FM4-64 stained hypocotyl cell, (d) merged images of panels (b) and (c) showing GRV2::YFP (green) and FM4-64 (red) together. Images in panels (b–d) were taken approximately 20 min after the addition of FM4-64. The scale bar represents 10 μm in panels b–d, and 50 μm in panel a.

GRV2::YFP structures are sensitive to Wortmannin, but not BFA

To further identify the nature of the GRV2-containing endosomes, roots of Arabidopsis seedlings were treated with BFA. This drug blocks the early stages of endocytosis, leading to the accumulation of recently endocytosed membranes and the TGN into so-called BFA compartments (Geldner, 2004; Samaj et al., 2005). Addition of BFA to transgenic tobacco lines expressing the trans-Golgi marker sialyl transferase (ST)-GFP resulted in the formation of distinct BFA compartments. However, this drug had no effect on the appearance or distribution of GRV2 endosomes (data not shown). Tse et al. (2004) showed that in tobacco BY2 cells the PVC and the multivesicular body (MVB) are, in fact, the same structure, playing a role in endocytic trafficking to the lytic vacuole, and in clathrin-mediated protein trafficking from the trans-Golgi to the vacuole. In contrast to BFA, the drug Wortmannin, which acts on the late endocytic pathway, and interferes with membrane trafficking between MVB/PVC and the lytic vacuole, induced the GRV2-positive endosomes to balloon up into discrete rounded structures (Tse et al., 2004; Figure 5a,b).

Figure 5.

 Confocal sections of control (a) and Wortmannin-treated (b) root tips of transgenic seedlings expressing GRV2::YFP (green). Propidium iodide staining is shown in red and the scale bars are 20 μm.

Tonoplast morphology in grv2 mutants

A GFP::δ-TIP fusion that localizes to the tonoplast was transformed into wild-type and grv2-1 plants. When visualized in etiolated 3–5-day-old wild-type seedlings (n > 50) the tonoplast in the small apical hook cells appears to be a highly folded continuous membrane (Figure 6a,c). This morphology is consistent with the highly dynamic nature of the vacuole. By contrast, cells in the same region in the grv2-1 mutant (n > 50) appeared to have multiple smaller vacuoles that formed smooth spheres, one or two of which were much larger than the others (Figure 6b,d). In addition, grv2-1 mutant cells, in both the apical hook region and the elongated region below the apical hook, contained aggregates of GFP::δ-TIP that were not commonly observed in the wild type (Figure 6e,f). In elongated hypocotyl cells there were an average of 2.14 aggregates per 100 000 μm3 (n = 28) in the grv2-1 mutant, compared with 0.02 aggregates per 100 000 μm3 in the wild type (n = 26). These aggregates appeared to be composed of several closely-packed endosomes.

Figure 6.

 Confocal sections and three-dimensional reconstructions of etiolated seedlings transformed with GFP::δ-TIP.
(a, b) Confocal sections of apical hook cells.
(c, d) Three-dimensional reconstructions of apical hook cells.
(e, f) Hypocotyls cells.
(a, c, e) Ler wild type.
(b, d, f) grv2-1 mutant. The scale bar for (a–d) is 10 μm, and for (e, f) the scale bar is 20 μm.

The vacuoles in root cells were also examined. Aggregates were observed in the vascular cells in the elongation zone in the mutant (Figure 7), but were absent in the other cell types scored: root tip cells and epidermal cells at the beginning of the elongation zone (data not shown). Although the frequency of aggregates in vascular cells was not quantified because of the reduced optical accessibility of this tissue, they appear similar to those that were observed in the hypocotyl.

Figure 7.

 Three-dimensional reconstructions of medial confocal sections of GFP::δ-TIP (green) and FM4-64 (red) in elongated root cells. Overlap of the signals from the red and green channels is shown as yellow.
(a) Ler wild type.
(b) grv2-1 mutant. Abbreviations: c, cortex; e, epidermis; v, vasculature. The scale bar is 20 μm.

To determine if the aggregates were membranous or were aggregates of protein, the localization of the lipophilic dye FM4-64 was observed. In FM4-64 pulse-chase experiments, FM4-64 localized after 18 h to the tonoplast in both the mutant and the wild type (Figure 8). In the grv2-1 mutant, FM4-64 also localized to the aggregates, suggesting that they were not just aggregates of the GFP::δ-TIP protein, but rather that they were membranes, either endosomes or tonoplast, containing GFP::δ-TIP. This indicates that the aggregates contain both proteins in transit to the tonoplast and endocytosed molecules originating from the plasma membrane. The localization of FM4-64 to the aggregates allowed for their visualization in plants homozygous for an additional allele, grv2-4, for which a GFP::δ-TIP transgenic line was not generated (data not shown). These results indicate that the presence of GFP::δ-TIP does not cause the aggregates in the mutant background, and that the defect is not specific to the grv2-1 mutant, but is a more general defect in grv2 mutants. The two defects associated with membranes labeled with a vacuolar marker, the irregular shape of the vacuole and the aggregates of endosomes indicate that grv2 mutants may be defective in the fusion of tonoplast and vesicular membranes.

Figure 8.

 Confocal sections through epidermal cells of GFP::δ-TIP (green) and FM4-64 (red) in etiolated hypocotyl cells, 18 h after a 10-min exposure to FM4-64.
(a–c) grv2-1 mutants.
(d–f) Ler wild type.
(c, f) Overlap between merged green and red channels is shown as yellow. Solid arrows indicate tonoplast. Open arrows indicate aggregates. The scale bar is 10 μm.

Measurement of endocytosis

In order to assess a possible defect in endocytosis, the rate of internalization of an endocytosis marker, FM4-64, was examined in grv2-1 and in grv2-4. Etiolated seedlings were perfused with FM4-64 and confocal microscopy was utilized to create a time series, with frames acquired every 10 sec, of medial sections of epidermal cells in the apical hook region. Individual frames were analyzed to determine when the initial internalization of FM4-64 into endosomes occurred. In 4-day-old etiolated seedlings of wild-type Ler, endosomes were first observed on average at 124 sec after the addition of the dye (n = 19; Figure S3). In grv2-1, endosomes were first observed on average at 85 sec after the addition of the dye (n = 22; Figure S4). In Col wild type (n = 33) and in the grv2-4 mutant (n = 21), endosomes were first observed on average at 202 and 191 sec after addition of the dye, respectively. Both wild-type and mutant alleles showed considerable variability (Figure 9). A Mann–Whitney rank sum test indicated that the grv2-1 mutant allele was significantly different from Ler wild type (P ≤ 0.0013). However, the grv2-4 mutant allele was not significantly different from the Col wild type (P ≤ 0.23). Thus, we infer that there is not a consistent difference in the rate of endocytosis between the mutant and the wild type.

Figure 9.

 Time course of internalization of FM4-64 by wild types and grv2 mutants. Seedlings were perfused with 10 mm FM4-64 in 6 μm DMSO, immediately placed on the confocal microscope, and a medial optical section was acquired. The time until the first endosome was observed was measured. In order to ensure unbiased measurement, the images were scored without prior knowledge of the genotype of the sample or of the time after the addition of dye at which the first time point was imaged.

Discussion

Based on the sequence similarity of GRV2 to endocytosis factors in C. elegans, Drosophila and mammalian cells, we hypothesized that GRV2 functions in vesicle budding, transit or fusion, either at the plasma membrane or subsequently in the endomembrane system. To test this hypothesis we first determined if GRV2 localized to a subcellular region that would be conducive to such a role, such as the plasma membrane, endosomes, TGN or MVB/PVC. Then we determined if grv2 mutants had defects in endocytosis or vesicle trafficking.

GRV2 localization to endosomes, as indicated by confocal microscopy, and more specifically, to sucrose gradient fractions that contained the MVB/PVC and TGN, is appropriate for a role in vesicle trafficking. GRV2::YFP showed only occasional weak co-localization with FM4-64-labeled early endosomes, indicating that it more likely to localize to late endosomes. To test this hypothesis, we treated seedlings with either BFA or Wortmannin. The former disrupts the morphology of early endosomes, and the latter disrupts late endosomes. BFA was without effect on the GRV2 endosomes, consistent with our observations that GRV2 does not localize to early endosomes. However, Wortmannin did cause a disruption of GRV2 endosomes. In tobacco BY2 cells, Wortmannin caused the membrane of the MVB/PVC to enlarge, identical to our observations of GRV2 structures, with a concomitant loss of internal vesicles within the MVB/PVC. The appearance of the GRV2 structures following Wortmannin treatment, together with our data showing co-fractionation of GRV2 with the MVB/PVC-localized PEP12/SYP21 (Foresti et al., 2006), suggests that the MVB/PVC is one of the main locations of GRV2 in plants. As GRV2 lacks predicted membrane-spanning domains, and can be removed by treatment with urea, it does not appear to be an integral membrane protein.

GFP::δ-TIP, a tonoplast intrinsic protein, accumulates in membranous aggregates in the grv2 mutant, but not in the wild type. The aggregates presumably arise from the vacuole, MVB/PVC membranes or vesicles that transport membrane components from the MVB/PVC to the tonoplast. Recently, grv2 mutants were also identified in a screen for seedlings with abnormal endomembrane structures (Tamura et al., 2007). In accordance with our results, all five alleles characterized by Tamura et al. (2007) exhibited similar aggregates of a GFP endomembrane marker. Mutations in class E Vacuolar Protein Sorting genes, components of the yeast ESCRT system, result in the accumulation of endosomal membranes and cargo in large aberrant structures adjacent to the vacuole (Babst, 2005). These structures may be analogous to the aggregates observed in grv2 mutants. Transport to the vacuole is not completely abolished, as sufficient GFP::δ-TIP does reach the tonoplast for visualization.

The altered morphology of the vacuole in the mutant indicates that the tonoplast has different properties from the wild type. The wild-type tonoplast is highly folded, whereas the mutant tonoplast is extended into a simpler, more spherical shape. This difference in morphology could be linked to the observed aggregates. Similar results were reported by Tamura et al. (2007), who noted that grv2 mutants lacked transvacuolar strands. If vesicles derived from the MVB/PVC in the mutant are less able to fuse with the tonoplast, the composition of lipids and proteins in the tonoplast may be altered, thus changing the properties of the membrane. Simply limiting the area of the vacuolar membrane may contribute to a less complex morphology. Alternatively, the osmotic potential of the vacuole may be altered, leading to differences in vacuole expansion.

Although we observed a small but statistically significant difference in endocytosis of FM4-64 in the grv2-1 mutant, the difference was not observed in the grv2-4 null allele. Therefore, we could not find any direct evidence that the GRV2 protein is normally involved in plasma membrane endocytosis, consistent with our view that GRV2 operates later in the endocytic pathway, prior to delivery of vesicles to the tonoplast.

Embryo development

The large cells in some of the grv2 mutant embryos may be the result of a defect in vacuole partitioning early in embryogenesis. During cell division the vacuole normally breaks down into several smaller vacuoles, which subsequently partition into two daughter cells (Segui-Simarro and Staehelin, 2006). A defect in this process may result in incorrect partitioning of the vacuole. One daughter cell would receive the main large vacuole, and the other daughter cell would receive a few smaller vacuoles. The presence of a vacuole that cannot be remodeled properly during cell division may cause a misalignment of the phragmoplast, resulting in daughter cells of unequal sizes.

Another hypothesis for the origin of the large cell comes from studies in BY-2 cells, a tobacco cell culture line. Overexpression of DRP2A, an Arabidopsis dynamin homolog, results in 5–10% of the cells expanding in cell size. Hong et al. (2003) hypothesized that this phenotype is caused by excessive endocytosis of the plasma membrane, which in turn results in an expansion of the vacuole. The large cells do not undergo further divisions and become terminal (Hong et al., 2003). In Drosophila, dynamin interacts genetically with the GRV2 homolog, RME-8, and functions as a regulator of endocytosis (Chang et al., 2004). Thus, it is possible that the observed increased rate of endocytosis from the plasma membrane to the vacuole in the grv2-1 mutant results in large cells through a similar mechanism.

grv2-1 is the only allele with a highly penetrant early embryonic phenotype. Three of the additional eight alleles, grv2-5, grv2-8 and grv2-9, also exhibit this phenotype, although at much lower penetrance. The grv2 alleles characterized by Tamura et al. (2007) also exhibit embryo developmental defects. However, the defects described are late in embryogenesis at the transition from the torpedo to the walking stick stage. All of the alleles described are reported to exhibit normal early embryogenesis from one-cell to early torpedo stages (Tamura et al., 2007). However, our examination of one of the T-DNA insertion lines they characterized, grv2-9/kam2-3, showed that a small fraction of the embryos did contain an enlarged cell at the embryo apex. In addition to this variability in embryo defects, there is variability in the severity of the gravitropic and phototropic defects of the four main alleles (Silady et al., 2004). This suggests that some alleles could have low levels of functional protein or, in the case of grv2-1, have truncated protein with recessive deleterious effects. The interallelic cross data support this hypothesis. Embryos with either grv2-1/GRV2+ or grv2-2, 3 or 4/grv2-2, 3 or 4 genotypes are wild type, but some embryos with a grv2-1/grv2-2, 3 or 4 genotype exhibit the mutant phenotype. It is possible that the grv2-1 mutant has residual truncated protein that results in an embryo phenotype that is complemented by wild-type protein, but not by the absence of GRV2 protein in the grv2-2, grv2-3 and grv2-4 alleles. However, in all four alleles, no GRV2 protein could be detected on Western blots. Because of differences in the phenotypes of the various alleles it seems likely that very low levels of protein are present. Alternatively, the protein extraction method might not be optimized for a truncated form of GRV2, which may be more sensitive to degradation.

Seedling arrest

grv2 seedlings, grown without sucrose, arrest 3 days after germination. This phenotype was not observed in the grv2 mutant alleles characterized by Tamura et al. (2007), as their growth conditions included 1% sucrose. However, they did observe that grv2 mutants mis-sort storage proteins in maturing seeds, by secreting them into the extracellular space, rather than transporting them to the vacuole for processing to the mature form, as in wild type (Tamura et al., 2007). Reduced levels of available storage proteins could result in the observed seedling arrest of grv2 mutants when grown without sucrose.

Conclusions

The localization of GRV2 in endosomes that are sensitive to Wortmannin, but not BFA, and the pronounced vacuolar defects in the grv2 mutant suggest a role for GRV2 in the delivery of vesicles from the MVB/PVC to the tonoplast. Disruption of this pathway has been strongly implicated in other shoot gravitropic mutants (Kato et al., 2002; Morita et al., 2002), consistent with a generalized role for the tonoplast in gravitropism. At present, it remains to be shown whether the loss of gravitropic responses in these mutants is caused by a general defect in vacuolar morphology that interferes with correct amyloplast sedimentation, or whether the loss of shoot gravitropism occurs from a direct effect on membrane trafficking between the MVB/PVC and the central vacuole. Regardless, the GRV2 protein provides a new marker for the late endocytic pathway in plant cells, and may provide a tool with which to dissect this pathway further.

Experimental procedures

Plant material and growth conditions

The grv2-1 allele was identified in an EMS-mutagenized Ler population (Gillmor et al., 2002). grv2-2 and grv2-3 were identified in a fast-neutron mutagenized Col population (Yamauchi et al., 1997). grv2-4 was identified in a γ-ray mutagenized Col population (Yamauchi et al., 1997). grv2-5 is a T-DNA insertion line in a Ler background (Sundaresan et al., 1995). grv2-6, grv2-7, grv2-8 and grv2-9 are T-DNA insertion lines in Col backgrounds (Alonso et al., 2003; Sessions et al., 2002) identified from knowledge of the GRV2 gene sequence. All of the lines are available from the Arabidopsis stock center at Ohio State University.

Unless stated otherwise, plants were grown on the surface of agar medium in vertically oriented Petri plates containing MS medium with 0.8% agar, and either with or without 1.5% sucrose, glucose or sorbitol in continuous light at 22°C. Plants grown on soil were exposed to constant light in a greenhouse at 22°C at a light intensity of approximately 70 μm m−2 s−1.

Antibody preparation

DNA encoding amino acids 443–1057 from exon 7 was cloned into NotI and SacI sites of the pET-28a vector with a C-terminal 6XHIS tag, and was expressed in BL21-CodonPlus Escherichia coli cells. Cells were lysed in 500 mm NaCl, 100 mm Tris, 20 mm imidazole, pH 7.9, and were centrifuged at 10 000 g for 20 min. The insoluble fraction was resuspended in bind buffer (6 m guanidine HCl, 100 mm Tris, pH 7.9) and applied to a nickel column. The column was washed with six column volumes of bind buffer followed by six volumes of wash buffer (6 m guanidine HCl, 100 mm Tris, 1 m NaCl, pH 7.9). The protein fragment was eluted with elution buffer (6 m guanidine HCl, 100 mm Tris, 1 m NaCl, 100 mm imidazole, pH 7.9). Imidazole was removed from the protein elute by dialysis against water.

Immune rabbit sera was affinity purified against the His-tagged purified protein fragment using an AminoLink® Plus Immobilization Kit (Pierce, http://www.piercenet.com). The protein fragment was solubilized in modified coupling buffer (6 m guanidine, 150 mm NaCl, 100 mm phosphate, pH 7.2) and applied to the AminoLink column. The column was then washed with normal coupling buffer (150 mm NaCl, 100 mm phosphate, pH 7.2). Antibodies were purified on the column using the standard protocol and eluted with 0.1 m glycine-HCl, pH 2.5.

Protein extraction

Flowers and leaves were ground with a mortar and pestle with ice-cold extraction buffer 1:1 (w/v; 100 mm Tricine, pH 7.9, 50 mm NaCl, 2 mm EDTA and 8% sucrose) plus 5 mg ml−1 polyvinylpolypyrrolidone (PVPP). Freshly prepared 2 mm DTT and plant protease inhibitor cocktail (Sigma P9599, 1 ml per 15 g; Sigma-Aldrich, http://www.sigmaaldrich.com) were added. Samples were centrifuged at 10 000 g for 10 min at 4°C. The supernatant was centrifuged at 100 000 g for 1 h. The pellet was resuspended in either 1X or 0.5X original volume of extraction buffer containing fresh 2 mm DTT with either 7 m urea, 3% Triton X-100, 0.1 m Na2CO3, pH 11.5, 2 m NaCl or without additive. Samples were incubated on ice for 4 h followed by centrifugation at 100 000 g for 1 h. Supernatant and pellet fractions were analyzed by Western blot.

Sucrose gradient fractionation

Microsomal fractions (100 000-g pellet) from plant protein extracts (2 g fresh weight) in 2 ml extraction buffer were applied to step gradients 15%, 24%, 33%, 40% and 56% (w/v) sucrose in modified extraction buffer (100 mm Tricine, pH 7.9, 50 mm NaCl, 2 mm EDTA). Protein extracts from Col GRV2:YFP were supplemented with wild-type Col plant protein extract (1:3, GRV2:YFP to Col) to equalize band intensity. The gradients were centrifuged for 4 h at 100 000 g at 4°C in a SW40 rotor (http://www.beckman.com). Fractions (0.5 ml) were pipetted from the top of the gradients. Aliquots were analyzed with a refractometer to determine the sucrose concentration of each fraction. Fractions were concentrated to 150 μl with Microcon® Centrifugal Filter Devices (Millipore, http://www.millipore.com) with either a 30 000 or 100 000 kDa cut-off. A 10-μl volume of each concentrated fraction in 10 μl of loading buffer (50 mm Tris-HCl, 2 mm EDTA, 1% SDS, 8% glycerol, 0.025% bromophenol blue, 1% 2-mercaptoethanol, pH 8) was analyzed by Western blot.

Western blot analysis

Protein extracts were electrophoresed in 5% SDS-PAGE (Bio-Rad; http://www.bio-rad.com) and were electrophoretically transferred to nitrocellulose. The transfer of large proteins was confirmed by the complete transfer of a 250-kDa standard. The affinity purified GRV2 antibody was applied at a 1:200 dilution, SEC12 at 1:2000, α-TIP at 1:250, PEP12 at 1:500 and VPS45 at 1:200. A peroxidase-coupled secondary antibody against rabbit (1:10 000) was used, except for α-TIP, for which a secondary antibody against chicken (1:10 000) was used.

GRV2:YFP protein fusion

The YFP coding sequence was amplified by PCR from pRSET(B)-Citrine (Griesbeck et al. 2001, JBC, 276, 29188–29194) using the primers gcacgcgttggcagagcgggcgctggtggcGTGAGCAAGGGCGAGGAGCTGTTG and gcacgcgtcaccagctccggcgccagcgcctgcCTTGTACAGCTCGTCCATGCCGA. The PCR fragment was digested with MluI and cloned into the MluI site pRAS03 (Silady, 2006), and was then transformed into DH5α-E ElectroMax competent cells (Invitrogen, http://www.invitrogen.com). Recombinant plasmids pRAS30 and pRAS43 were identified by PCR and confirmed by sequencing.

grv2-1, grv22, grv2-3 and grv2-4 mutants and Col wild type were transformed with both pRAS30 and pRAS43. The presence of the original mutation and the wild-type copy of GRV2 were assayed in grv2-1::pRAS30, grv2-1::pRAS43, grv2-4::pRAS30, and grv2-4::pRAS43 transformants. The transformants in the grv2-1 background were assayed with a dCAPs marker using 7-7F/R primers (5′-GAGCAATGTGCTCCTTCTGTTGC-3′ and 5′-GTAGGTATGGGCTCAGTTCTCCCAT-3′, in which the underlined base pair is changed from an A in the genomic sequence to facilitate cleavage) to amplify a 191-bp fragment. The fragments were digested with Van911 restriction enzyme, which cuts wild-type GRV2, but not grv2-1, into 164- and 27-bp fragments. The transformants in the grv2-4 background were assayed with a CAPs marker using UKN27R/13R primers (5′-CACGCCACATCACAAGTA-3′ and 5′-AGCTGGCATGGTTTCTGATTCC-3′) to amplify a 405-bp fragment. The fragments were digested with Cac81 restriction enzyme, which cuts wild-type GRV2, but not grv2-4, into 348 and 57-bp fragments.

Inhibitor treatment

Five-day-old GRV2::YFP seedlings were incubated in either 50 μm BFA in 0.5% MS media with 1% (w/v) sucrose or 33 μm wortmannin in 0.5% MS with 1% (w/v) sucrose for 2 h. Control seedlings were incubated in 0.5% MS with 1% (w/v) sucrose for 2 h. Seedlings were mounted in 10 μg ml−1 propidium iodide in 0.5% MS with 1% (w/v) sucrose for imaging.

Microscopy

Ovules were dissected from siliques at different stages of development and cleared overnight at 4°C with Hoyer’s solution (Gillmor et al., 2002). Embryos were visualized with Nomarski optics at 40× magnification on a Leica compound microscope (Leica, http://www.leica.com).

The GFP vacuole marker used was a δ-TIP (AT3G16240) GFP protein fusion under the control of potato ubiquitin promoter UBI3 (GenBank accession L22576; Garbarino and Belknap, 1994). δ-TIP was fused to the C-terminal end of GFP in the pEGAD vector (Cutler et al., 2000).

FM4-64 (10 μm; Molecular Probes, http://probes.invitrogen.com) dissolved in 6 μm DMSO was added to 4-day-old etiolated seedlings. Images were obtained in 10-sec intervals for up to 5 min. For the pulse-chase experiment FM4-64 was added for 10 min, and then the seedlings were washed three times with water. Samples were viewed 18–20-h later.

A Bio-Rad MRC 1024 confocal microscope with a 60× water immersion lens or a Leica TCS SP2 confocal microscope with a 63× water immersion lens was used for visualization. The samples were excited with two lasers (Ar/Kr and He/Cd) at the following wavelengths: 488 nm (GFP and YFP); 568 nm (FM4-64). On the Leica TCS SP2, YFP was excited with the Ar/Kr laser at 514 nm. Images were processed with image j 1.34s (http://rsb.info.nih.gov/ij), volocity 2.5 (Improvision, http://www.improvision.com), adobe photoshop 7.0 and adobe illustrator 10.0 (Adobe Systems, http://www.adobe.com). Aggregates were identified by volumetric segmentation and filtering using the volocity program. The minimum intensity threshold for identifying an aggregate was set to the maximum signal intensity in a given slice of the tonoplast. The minimum size was set to 5 μm3.

Acknowledgements

We gratefully acknowledge assistance and advice from Natasha Raikhel and members of the Raikhel laboratory. We thank T. Hamann for comments on the manuscript.

This work was supported in part by a grant from the US Department of Energy (DE-FG02-03ER20133).

Ancillary