The Class C Vps Complex Functions at Multiple Stages of the Vacuolar Transport Pathway

Authors


*Corresponding author: Scott D. Emr, semr@ucsd.edu

Abstract

The Class C Vps complex, consisting of Vps11, Vps16, Vps18, and Vps33, is required for SNARE-mediated membrane fusion at the lysosome-like yeast vacuole. However, Class C vps mutants display more severe and pleiotropic phenotypes than mutants specifically defective in endosome-to-vacuole transport, suggesting that there are additional functions for the Class C Vps complex. A SNARE double mutant which is defective for both Golgi-to-endosome and endosome-to-vacuole trafficking replicates many of the phenotypes observed in Class C vps mutants. We show that genetic interactions exist between Class C vps alleles and alleles of the Class D vps group, which are defective in the docking and fusion of Golgi-derived vesicles at the endosome. Moreover, the Class D protein Vac1 was found to physically bind to the Class C Vps complex through a direct association with Vps11. Finally, using a random mutagenic screen, a temperature-conditional allele which shares many of the phenotypes of mutants which are selectively defective in Golgi-to-endosome trafficking was isolated (vps11–3ts). Collectively, these results indicate that the Class C Vps complex plays essential roles in the processes of membrane docking and fusion at both the Golgi-to-endosome and endosome-to-vacuole stages of transport.

The specific targeting of transport vesicles to appropriate target membranes is essential for the maintenance of the unique composition of eukaryotic organelles. The molecular mechanisms that act to ensure specificity in the docking and fusion reactions have been subject to intense scrutiny in the last decade, and are now beginning to be understood. One of the first mechanisms proposed to ensure targeting specificity focused on two families of integral membrane proteins, the vesicle, or v-SNARE (for SNAP receptor) family and the target, or t-SNARE family. The SNARE hypothesis postulates that pairing between a vesicular v-SNARE and a target membrane t-SNARE is the primary molecular interaction underlying the specificity of vesicle targeting as well as lipid bilayer fusion (1). Although recent experiments from James Rothman and colleagues have found that there is some amount of specificity to SNARE–SNARE interactions (2), it is clear that SNARE pairing is not the sole determinant of specificity in vesicular transport, and that there are a host of other factors which must contribute to the specificity of vesicular transport (3–5). In particular, members of the Rab GTPase and Sec1 families are also required, and may act in conjunction with SNARE proteins to increase the specificity of vesicular docking and fusion (6–8). Beyond these conserved protein families, there have been several other unique accessory factors identified, such as the TRAPP, Exocyst, Vps52/53/54, and Class C Vps complexes, which appear to facilitate the processes of docking and fusion (9–13).

The vacuolar protein sorting pathway (VPS) mediates the transport of biosynthetic traffic from the late-Golgi to the yeast vacuole. In genetic screens, over 50 gene products have been identified which mediate this pathway; these fall into five groups which correspond to distinct steps of the VPS pathway (14,15). The Class D VPS group mediates the trafficking of transport vesicles from the late-Golgi to the prevacuolar endosome (16,17), while members of the Class B VPS group are required for the consumption of transport vesicles and late endosomes at the vacuole (3,18). The Class C VPS group consists of four genes: VPS11/PEP5/END1, VPS16, VPS18/PEP3, and VPS33. Inactivation of the vps18ts allele results in the accumulation of multivesicular bodies, and the missorting of multiple biosynthetic cargoes (19). In addition, the Class C Vps complex exhibits genetic and physical interactions with the vacuolar t-SNARE Vam3 and the vacuolar Rab GTPase Ypt7 (13,20,21). Thus, there is abundant evidence for a role for the Class C Vps complex in facilitating the docking and fusion of transport vesicles and late endosomes at the vacuolar surface.

Interestingly however, deletion of any of the Class C VPS genes results in severe and pleiotropic phenotypes, compared with other genes required for endosome-to-vacuole transport (i.e. the Class B VPS group) (15,22,23). These data raise the possibility that the Class C Vps complex plays additional roles at other transport steps. A previous report has suggested that this may in fact be correct. In Srivastava et al. (24), it was reported that Vps11 and Vps18 function in Golgi-to-endosome anterograde and retrograde transport, and in the endocytic pathway. In addition, genetic and two-hybrid evidence was presented suggesting that Vps18 may interact with the Class D VPS protein Vac1 (24). We here report the first isolation of a Vps11 temperature-conditional allele which is defective only at the Golgi-to-endosome transport step, indicating that Vps11, and by extension the Class C Vps complex, play distinct, separable roles at both the Golgi-to-endosome and endosome-to-vacuole transport steps. In addition, genetic interactions between alleles of VPS11, VPS16, and VPS18 and alleles of the Class D VPS genes VAC1, VPS21, and VPS45 are reported. Finally, we demonstrate a direct, biochemical interaction between the phosphatidyl-inositol 3-phosphate (PI3P) binding protein Vac1 and the Class C Vps complex. Vac1, Vps21 (a Rab GTPase), and Vps45 (a Sec1 homolog) are thought to function in the initial stages of docking of transport vesicles with the endosome (25,26). Thus, our data indicate that the Class C Vps complex plays a role in the early stages of docking at the endosome, which is consistent with the function of the Class C Vps complex at the vacuolar surface (13,20,21).

Results

Double SNARE mutant is a phenocopy of Class C vps mutant

Given the fact that deletion of Class C VPS genes results in strains that have more severe phenotypes than strains defective in endosome-to-vacuole transport, such as a severely retarded growth rate and temperature-sensitive growth, we suspected that the Class C Vps complex may also function at other docking and fusion steps. There are two well-characterized docking and fusion steps in the VPS pathway: one at the vacuolar surface, utilizing the t-SNARE Vam3 (3,18), and one at the endosome, utilizing the t-SNARE Pep12 (16,17). Therefore, we hypothesized that the Class C Vps complex may also function in the Golgi-to-endosome step of the VPS pathway. To test this hypothesis, a double-mutant strain was constructed, deleted for both the vacuolar t-SNARE Vam3 and the endosomal t-SNARE Pep12. This strain, pep12Δvam3Δ, is a phenocopy of strains deleted for Class C VPS genes by multiple criteria. The vacuolar morphology was observed using the fluorescent dye 7-amino 4-chloromethylcoumarin (CMAC) in four strains, pep12Δ, vam3Δ, pep12Δvam3Δ, and vps11Δ. Note the diffuse, punctate CMAC staining observed in both the vps11Δ and pep12Δvam3Δ strain, typical of all Class C vps deletion mutants (Figure 1A) (23,27,28). This is in contrast to the single, enlarged vacuole of the pep12Δ strain, and the clusters of small structures in the vam3Δ strain. Growth at 37 °C was also tested (Figure 1B), and the pep12Δvam3Δ double mutant was incapable of growth at 37 °C, similar to vps11Δ and in contrast to either single mutant. Besides these, this strain shares several other phenotypes with Class C vps deletion mutants, such as sensitivity to caffeine and a severely retarded growth rate (data not shown). This is not unique to the pep12Δvam3Δ double-mutant strain – a strain deleted for the Rab GTPases which function in Golgi-to-endosome and vacuolar steps of the VPS pathway (vps21Δypt7Δ) is also a phenocopy of vps11Δ (data not shown). Although not conclusive, these results are strongly suggestive of a role for the Class C Vps complex at the endosome, as well as at vacuolar surface.

Figure 1.

pep12Δvam3Δ is a phenocopy of vps11Δ. (A) Vacuolar morphology was assessed at 30 °C by CMAC staining in the following strains: pep12Δ (CBY31), vam3Δ(TDY2), pep12Δvam3Δ (MPY11), and vps11Δ (MPY8). (B) Temperature-conditional growth was assayed by streaking the above strains to YPD plates and growing them at either 26 °C or 37 °C for 3 days. Bar, 5 µm.

Class C vps alleles with stage-specific sorting defects

If the Class C Vps complex acts at both the endosome and vacuole, it is possible that mutants might be isolated which are defective at only one of these stages in the VPS pathway. Toward this end, the open reading frame of VPS11 was mutagenized by error-prone PCR, and the resulting library was transformed using gapped-plasmid repair into a vps11Δ strain (29). Transformants were replica plated and screened for secretion of a CPY-invertase fusion at the permissive (26 °C) and nonpermissive (38 °C) temperatures. Mutant plasmids which were isolated in this manner were integrated into the VPS11 locus of wild-type yeast using pop-in/pop-out allele replacement (30).

This screen generated several interesting vps11 alleles. As shown in Figure 2(A), the processing of the vacuolar hydrolase carboxypeptidase Y (CPY) was analyzed by pulse-chase radiolabeling, followed by immunoprecipitation with CPY-specific antisera. CPY is trafficked from the Golgi to the vacuole as an inactive precursor, termed p2CPY. Upon entry to the vacuole, p2CPY is proteolytically cleaved to the active, mature form, mCPY. Disruption of the VPS pathway results in the accumulation of p2CPY. Additionally, strains which are defective in the Golgi-to-endosome stage of the VPS pathway secrete p2CPY (17), while strains which are defective later in the VPS pathway retain p2CPY intracellularly (3). CPY processing and secretion was assayed at the permissive (26 °C) and nonpermissive temperatures (38 °C) in two temperature-conditional strains, vps11–1ts and vps11–3ts. As can be seen, both strains accumulated p2CPY at the nonpermissive temperature (38 °C); however, vps11–1ts retained p2CPY intracellularly (lanes 3 and 4), while vps11–3ts secreted the majority of p2CPY extracellularly (lanes 7 and 8). This result suggests that vps11–3ts is defective at an early stage of the VPS pathway, while vps11–1ts is likely to be defective at a later stage of the VPS pathway. It is possible that these alleles are not recessive but instead act in a dominant fashion; this has been tested by expressing the original plasmids which encode the mutant alleles in a wild-type strain (SEY6210), and testing CPY maturation at the nonpermissive temperature. No CPY misprocessing was observed, indicating that vps11–1ts and vps11–3ts are likely to be recessive (data not shown.)

Figure 2.

CPY and ALP sorting phenotypes of vps11–1ts, vps11–3ts, and vps16ts. (A) CPY sorting was assayed in vps11–1ts (MPY9), vps11–3ts (MPY10), and vps16ts (MPY5) as follows. Mid-log phase cultures were harvested and preshifted to 26 °C or 38 °C for 5 min, then pulsed with 35S cysteine and methionine for 10 min. A chase solution containing excess nonradiolabeled cysteine and methionine was added, and the cells were incubated for an additional 30 min. The assay was then stopped, and the cells were converted to spheroplasts, and intracellular (I) and extracellular (E) fractions were generated by gentle centrifugation. CPY was immunoprecipitated and analyzed by SDS-PAGE analysis followed by autoradiography. (B) ALP sorting was assayed in a similar manner. Mid-log cultures were harvested and preshifted to 26 °C or 38 °C for 5 min, pulsed for 10 min, and chased for 30 min. Following the chase, ALP was immunoprecipitated and subjected to SDS-PAGE and autoradiography.

The processing of a second vacuolar hydrolase, alkaline phosphatase (ALP), was also assayed by pulse-chase radiolabeling, followed by immunoprecipitation with ALP-specific antisera. ALP traffics to the vacuole via a different pathway than CPY, bypassing the endosome. Hence temperature-conditional vps alleles which are defective in Golgi-to-endosome transport efficiently mature ALP, while vps alleles which are defective at later stages of the VPS pathway accumulate precursor ALP (pALP) (31). As can be seen in Figure 2(B), vps11–1ts missorts ALP at the nonpermissive temperature (38 °C; lane 2), similar to the vam3ts allele (3), while vps11–3ts is capable of efficiently processing ALP to its mature form (mALP; lane 4), similar to the pep12ts and vps45ts alleles (31).

The vacuolar hydrolase processing experiments indicate that two distinct alleles of VPS11 have been isolated, one that is specifically defective early in the VPS pathway at the Golgi-to-endosome step, and one that seems to be defective later, at the endosome-to-vacuole step. As an additional test of this hypothesis, the vacuolar morphology of vps11–1ts and vps11–3ts was tested at the nonpermissive temperature. Mutants which are defective in Golgi-to-endosome transport (the Class D vps group) have a characteristic vacuolar morphology, developing a single, enlarged vacuole (16,32,33). The vps11–1ts and vps11–3ts strains were incubated at the nonpermissive temperature for 2.5 h, and the vacuolar morphology was observed using electron microscopy. The fixation and staining procedures used in this study result in electron dense vacuoles, seen as the dark circular structures in Figure 3. Under these conditions, vps11–1ts develops an aberrant vacuolar morphology, with an accumulation of small vacuole-like structures (seen as the small dark disks) and assorted aberrant compartments (seen as the multilamellar structures), similar to a previously characterized Class C vps mutant, vps18ts (19). In contrast, vps11–3ts develops the single, enlarged vacuole typical of Class D vps mutants. Therefore, two distinct alleles of VPS11 have been isolated, one which is defective at the Golgi-to-endosome step of the VPS pathway (vps11–3ts), and one which may be defective primarily at the endosome-to-vacuole step (vps11–1ts).

Figure 3.

Vacuolar morphology of vps11–1ts, vps11–3ts, and vps16ts at 38 °C. Vacuolar morphology was assessed in vps11–1ts (MPY9), vps11–3ts (MPY10), and vps16ts (MPY5) following a 2.5-h shift to 38 °C using standard electron microscopy techniques (43). Wild-type strains under these conditions are noticeably different, typically containing several vacuoles (15). Black arrowheads were added to emphasize aberrant endosome-like structures. Bar, 1 µm.

In a similar manner, a temperature-conditional allele of VPS16 has been isolated via PCR-based mutagenesis that appears to be defective at both Golgi-to-endosome and endosome-to-vacuole stages of the VPS pathway. As can be seen in Figure 2(A), vps16ts secretes p2CPY outside the cell at the nonpermissive temperature, similar to vps11–3ts and pep12ts, consistent with a defect early in the Golgi-to-endosome stage (lanes 11 and 12.) However, this strain also misprocesses ALP at the nonpermissive temperature (Figure 2B, lane 6), and has a dramatically aberrant vacuolar morphology, lacking any recognizable vacuoles and accumulating many multilamellar compartments (Figure 3.) This is similar to vps11–1ts, vps18ts, and vam3ts, and is consistent with a defect late in the VPS pathway. Thus, this allele is the first Class C vps allele that has been isolated which appears to be strongly defective at both the Golgi-to-endosome and endosome-to-vacuole stages of the VPS pathway. The vacuolar hydrolase sorting characteristics and vacuolar morphologies of four Class C vps alleles are summarized in Table 1. It is interesting that such dramatically different alleles can be isolated from this group; we speculate that this is due to the multifunctional nature of the Class C Vps complex.

Table 1. : Relevant phenotypes of temperature-conditional Class C vps alleles
Class C
vps allele
Accumulated
p2CPY
ALP
processing
Vacuolar
morphology
  1. The phenotypes of four temperature-conditional Class C vps strains are summarized above. The ‘Accumulated p2CPY’ column indicates whether the indicated strain retains p2CPY internally or secretes it externally after a short shift to the nonpermissive temperature. The ‘ALP processing’ column indicates whether the strain is capable of processing precursor ALP to mature ALP at the nonpermissive temperature. The vacuolar morphology is summarized in the last column; ‘fragmented/no vac.’ indicates that few to no recognizable vacuolar structures are to be found upon EM analysis

vps11–1tsinternaldefectivefragmented/no vac.
vps11–3tssecretedno defectsingle, enlarged
vps16tssecreteddefectivefragmented/no vac.
vps18tsinternaldefectivefragmented/no vac.

The mutations of the vps11–1ts and vps11–3ts alleles have been mapped and sequenced. Interestingly, both alleles have three point mutations each, which lie either in or immediately preceding the C-terminal cysteine-rich RING domain, indicating that this domain is critical for the proper functioning of Vps11 (vps11–1ts: I877S, I878S, C931S; vps11–3ts: S863P, D958G, L966P). Vps18 also has a C-terminal cysteine-rich RING domain; vps18ts similarly consists of a single point mutation which changes a putative zinc-coordinating cysteine to a serine residue (23). The mutations of vps16ts have not been mapped and sequenced.

Genetic interactions between Class C and D vps alleles

Alleles of the Class C vps group display synthetic genetic interactions with alleles of genes involved in docking and fusion at the vacuole (13). If the Class C Vps complex also functions at the endosome, Class C vps alleles may display synthetic genetic interactions with alleles of the Class D vps group which mediate docking and fusion of Golgi-derived vesicles with the endosome. To test this hypothesis, four Class C vps alleles, vps11–1ts, vps11–3ts, vps16ts, and vps18ts, were combined with three Class D vps alleles, vac1ts, vps45ts, and vps21 S21L. As shown in Figure 4, combinations of these alleles resulted in CPY missorting at the permissive temperature, indicating a synthetic genetic interaction. In Figure 4(A), vac1ts was strongly synthetic with vps11–3ts (lane 5), vps16ts (lane 7), and vps18ts (lane 9) and in Figure 4(B), vps45ts was strongly synthetic with vps11–1ts (lane 3), vps11–3ts (lane 5), and vps16ts (lane 7). The vps21 S21L allele is weakly dominant negative when expressed in a wild-type background; this allele exhibits synthetic genetic interactions with alleles of vac1 (25,26). In Figure 4(C), expression of this allele in vps11–1ts (lane 3), vps11–3ts (lane 5), vps16ts (lane 7), and vps18ts (lane 9) resulted in missorting of CPY at the permissive temperature (26 °C), despite the fact that endogenous wild-type Vps21 is present in these strains. It should be noted that despite the fact that vps18ts and vps11–1ts seem to be primarily defective at late stages of the VPS pathway (13,19), these alleles are likely to also have subtle defects at the Golgi-to-endosome step. Thus, the observation of genetic interactions between these alleles and the Class D vps alleles is not unexpected. As controls for the above synthetic genetic interactions, the Class C vps alleles were combined with an unrelated VPS temperature-conditional allele, vps4ts. Additionally, the vac1ts allele was combined with the temperature-conditional allele vam3ts and the vps21 S21L allele was expressed in a vam3ts strain. None of these double mutants resulted in a significant amount of CPY missorting at the permissive temperature (data not shown), suggesting that the synthetic genetic interactions observed between the Class C vps alleles and the Class D vps alleles are due to direct, functional interactions.

Figure 4.

Class C vps mutants exhibit genetic interactions with Class D vps mutants. For all of the following experiments, CPY sorting was assessed as follows: 2.5 OD600 units of culture were harvested at mid-log phase, and pulsed at 26 °C for 10 min with 35S cysteine and methionine, followed by a 30-min chase with an excess of nonradiolabeled cysteine and methionine. Following the chase, CPY was immunoprecipitated and subjected to SDS-PAGE and autoradiography. (A) Class C vps mutants vps11–1ts, vps11–3ts, vps16ts, and vps18ts display genetic interactions with vac1ts. Wild-type VAC1 (p414.19) or vac1ts (pCB174) were transformed into vac1Δ (CBY35), vps11–1tsvac1Δ (MPY12), vps11–3tsvac1Δ (MPY13), vps16tsvac1Δ (MPY14), vps18tsvac1Δ (MPY15). (B) As in A, vps45ts displays genetic interactions with Class C vps mutants. Wild-type VPS45 (pVPS45-10) or vps45ts (pVPS45-37) were transformed into vps45Δ (CCY12), vps11–1tsvps45Δ (MPY18), vps11–3tsvps45Δ (MPY19), vps16tsvps45Δ (MPY20), or vps18tsvps45Δ (MPY21). (C) The weakly dominant allele vps21 S21L displays strong synthetic genetic interactions with Class C vps mutants. Wild-type VPS21 (pMP38) or mutant vps21 S21L (pMP37) were transformed into SEY6210, vps11–1ts (MPY9), vps11–3ts (MPY10), vps16ts (MPY5), or vps18ts (SRY18T-1).

Vac1, Vps21, and Vps45 have all been shown to participate in docking and fusion at the endosome (17,25,26). The data presented in Figure 4 strongly suggest a role for the Class C Vps complex in the docking and fusion of Golgi-derived vesicles at the endosome.

Vac1 binds to Vps11 and the Class C Vps complex

Given the synthetic interaction between the Class D vps alleles and alleles of members of the Class C Vps complex, it was hypothesized that there was a direct physical interaction between the Class C Vps complex, and the endosomal proteins which mediate docking and fusion. To test this in a preliminary fashion, the two-hybrid technique was used to test for binding between these proteins. A strong two-hybrid interaction was observed between Vps11 and Vac1, but not between Vps18 and Vac1 (data not shown). Two-hybrid interactions are susceptible to false-positives, thus the binding of Vac1 to Vps11 was directly tested by biochemical means.

An in vitro binding reaction was developed to further test for an interaction between Vac1 and Vps11, using bacterially expressed GST-Vac1 immobilized on glutathione-Sepharose beads to precipitate proteins from a detergent-solubilized yeast lysate. After washing, bound proteins were subject to SDS-PAGE and Western blotting. As can be seen in Figure 5(A), GST-Vac1 is capable of precipitating epitope-tagged versions of both Vps11 and Vps18. Interestingly, deletion of VPS18 did not abrogate binding of Vps11-HA to GST-Vac1, but deletion of VPS11 did abrogate the interaction of Vps18-HA with GST-Vac1 (Figure 5B). This result, in combination with the two-hybrid data, suggests that Vac1 may interact directly with Vps11.

Figure 5.

In vitro GST-Vac1 binding assays. (A) Bacterially expressed GST or GST-Vac1 was immobilized on glutathione-Sepharose beads, and washed extensively, resulting in approximately 10 µg of purified GST or GST-Vac1 associated with the beads. Yeast lysates were prepared from the Vps11-HA (MPY1), Vps18-HA (MPY2) strains by glass-bead lysis in the presence of 1% Triton X-100. Cleared yeast lysates (2 mg of total protein/sample) were added to the immobilized GST or GST-Vac1 beads, and incubated at 4 °C for 1 h. Following washing, bound proteins were analyzed by SDS-PAGE followed by Western blotting using the anti-HA antibody. To demonstrate consistent loading, 10% of the lysate was run next to the GST and GST-Vac1 lanes (Load). (B) As in A, lysates from Vps11-HA vps18Δ (MPY17) and Vps18-HA vps11Δ (MPY16) were incubated with GST or GST-Vac1 beads under the same conditions and analyzed in the same manner.

Discussion

Previously, the Class C Vps complex has been demonstrated to interact genetically and physically with the vacuolar t-SNARE Vam3 (13,19), as well as with the vacuolar Rab GTPase Ypt7 (20,21). These studies established a role for the Class C Vps complex in the endosome-to-vacuole step of the VPS pathway, and more specifically in the docking and fusion of transport vesicles and late endosomes with the vacuole. In the present study, we demonstrate that the Class C Vps complex may play a similar role in the Golgi-to-endosome step of the VPS pathway.

Recently, the laboratory of Elizabeth Jones has shown that the Class C Vps complex functions at multiple stages of the VPS pathway, as well as functioning in endocytic transport. Briefly, temperature-conditional alleles of VPS18 and VPS11 were isolated that rapidly block the maturation of the vacuolar hydrolases CPY and ALP after shift to the nonpermissive temperature. In addition, these alleles stabilize the amino acid permease Gap1, suggesting that Vps11 and Vps18 function in endocytic transport. Similarly, these alleles result in the secretion of pro-alpha factor, indicating a potential role in the recycling of the protease Kex2 from the endosome to the late-Golgi. Finally, and most importantly, genetic and two-hybrid interactions were reported between VPS18 and VAC1. This study indicated that Vps11 and Vps18 function at both the Golgi-to-endosome and endosome-to-vacuole stages of the VPS pathway. However, it did not clearly demonstrate that the Class C Vps complex has distinct, separable roles at both the Golgi-to-endosome and endosome-to-vacuole steps. In addition, no direct physical interaction between Vac1 and the Class C Vps complex was demonstrated.

In contrast, we here present direct evidence that VPS11, and by extension the Class C Vps complex, has distinct roles at both the Golgi-to-endosome and endosome-to-vacuole steps of the VPS pathway. Specifically, we have isolated two different mutant alleles of VPS11, one which replicates many of the phenotypes of VPS mutants which are defective in Golgi-to-endosome transport (vps11–3ts), and one which exhibits defects late in the VPS pathway (vps11–1ts; Figures 2 and 3). Additionally, an allele of VPS16 has been isolated that displays phenotypes which are consistent with defects at both the Golgi-to-endosome and endosome-to-vacuole stages of the VPS pathway (Figures 2 and 3.) Further, these alleles, as well as the vps18ts allele, interact genetically with mutants defective in the processes of docking and fusion at the prevacuolar endosome, namely vac1ts, vps45ts, and vps21 S21L (Figure 4). Finally, we present direct, biochemical evidence that Vps11 interacts with Vac1, a protein known to function in the docking and fusion of Golgi-derived vesicles with the prevacuolar endosome (Figure 5). It is not presently clear whether the Class C Vps complex member and Sec1 homolog Vps33 is functioning at this stage; further experiments will be needed to determine if the Golgi-to-endosome stage requires the action of two Sec1 homologs, Vps45 and Vps33. Together, these results indicate that, in addition to its function at the vacuole, the Class C Vps complex also plays an essential role at the endosome, probably in the early stages of docking of transport vesicles with the endosome.

It is of particular interest that the Class C Vps complex functions in two distinct transport steps. We speculate that stage-specific factors may recruit and/or activate the Class C Vps complex. Previous studies have proposed that Vac1 acts to ensure that both the activated Rab GTPase Vps21 and the signaling lipid PI3P are present before docking and fusion can proceed (25,26). This hypothesis is congruent with the results of our present study, and we speculate that Vac1 may act as the stage-specific transport factor at the endosome which recruits and/or activates the Class C Vps complex. If the function of the Class C Vps complex at the endosome is similar to that observed at the vacuolar surface, we predict that following recruitment/activation by Vac1, the Class C Vps complex facilitates ternary SNARE complex formation. However, due to technical limitations, we have not been able to demonstrate this conclusively.

In addition, we speculate that the Class C Vps complex functions closely with the endosomal Rab GTPase Vps21, given the genetic interaction between the vps21 S21L and Class C vps alleles (Figure 4). At the vacuolar surface, the Class C Vps complex associates with the Class B VPS protein Vps39, which acts as a GEF for the vacuolar Rab GTPase Ypt7 (20); the Class C Vps complex has also been shown to be an effector of the GTP-bound form of Ypt7 (21), suggesting that the Class C Vps complex plays multiple roles with respect to Ypt7. It will be interesting to determine if the GEF for the Rab GTPase Vps21, Vps9, is similarly associated with the endosomal Class C Vps complex, and if Vps21 interacts with the endosomal Class C Vps complex in a nucleotide-dependent manner (34). Factors which function in the docking and fusion processes in the Golgi-to-vacuole transport pathway are summarized in Figure 6.

Figure 6.

Stage-specific and general factors required for Golgi-to-vacuole transport. At the top of the figure stage-specific factors are listed, while the Class C Vps complex, which is required for both Golgi-to-endosome and endosome-to-vacuole transport is below. GEF, guanine nucleotide exchange factor.

The Class C Vps complex is one of several protein complexes identified which seem to facilitate the docking and fusion of transport vesicles with acceptor compartments. Examples of these are the TRAPP, Exocyst, and Vps52/53/54 complexes (9–12). Although no sequence conservation is observed between these complexes, they seem to have certain similarities, in that they all: (i) are multisubunit protein complexes, (ii) seem to facilitate the early stages of docking of incoming transport vesicles with acceptor compartments, in a capacity that has been termed ‘tethering’, and (iii) typically function closely with Rab GTPases (reviewed in (35)). Recently, it has been reported that there are two variants of the TRAPP complex, TRAPP I and II, which function in ER-to-Golgi and intra-Golgi transport, respectively (36). Thus, the Class C Vps complex is a second example of one of the ‘tethering’ complexes which functions at multiple transport steps.

The precise role of the Class C Vps complex at the endosome and the vacuole remains to be determined. Specifically, the role the Class C Vps complex plays in endosomal ternary SNARE complex formation will need to be investigated. Homologs of Class C Vps complex components have been identified in several organisms, including C. elegans, D. melanogaster, and humans. Future studies on the Class C Vps complex in these organisms should help to shed additional light on the precise function of this conserved, essential transport complex.

Materials and Methods

Strains and media

Strains used in these studies are listed in Table 2. Strains were grown in standard yeast extract/peptone/dextrose (YPD), yeast extract/peptone/fructose (YPF), or synthetic medium (YNB) supplemented with appropriate amino acids (30). Standard bacterial medium supplemented with 100 µg/mL ampicillin was used to grow Escherichia coli (E. coli) strains and to maintain plasmids (37).

Table 2. :S. cerevisiae strains used in this study
StrainGenotypeReference or source
SEY6210MATαleu2–3, 112 ura3–52 his3-Δ200 trp1-Δ901 lys2–801 suc2-Δ9(14)
BHY10SEY6210; leu2–3,112::pBHY11 (CPY-Inv LEU2)(32)
CBY31SEY6210 pep12Δ::HIS3(17)
TDY2SEY6210 vam3Δ::LEU2(3)
MPY11TDY2 pep12Δ::HIS3This study
MPY8SEY6210 vps11Δ::HIS3This study
MPY7BHY10 vps11Δ::HIS3This study
MPY9SEY6210 vps11–1tsThis study
MPY10SEY6210 vps11–3tsThis study
MPY5SEY6210 vps16ts(13)
SRY18T-1SEY6210 vps18ts(23)
CBY35SEY6210 vac1Δ::URA3(17)
MPY12MPY9 vac1Δ::HIS3This study
MPY13MPY10 vac1Δ::HIS3This study
MPY14MPY5 vac1Δ::HIS3This study
MPY15SRY18T-1 vac1Δ::HIS3This study
MPY18MPY9 vps45Δ::HIS3This study
MPY19MPY10 vps45Δ::HIS3This study
MPY20MPY5 vps45Δ::HIS3This study
MPY21SRY18T-1 vps45Δ::HIS3This study
MPY1SEY6210 Vps11:HA:HIS3This study
MPY2SEY6210 Vps18:HA:HIS3(13)
MPY17MPY1 vps18Δ::HIS3This study
MPY16MPY2 vps11Δ::HIS3This study

Plasmids and DNA manipulation

Recombinant DNA manipulations were performed by standard methods (38). Transformation of S. cerevisiae was done according to the lithium acetate method (39). Gene disruptions for MPY7, MPY8, and MPY11-20, were achieved via amplification of HIS3, with the addition of flanking sequence homologous to the target gene. Disruptants were screened by histidine prototrophy, and the resulting strains verified by genomic PCR, Western blotting where possible, and CPY sorting assays. The chromosomal loci of VPS11 and VPS18 were tagged with the HA epitope using the method of Longtine et al. to create MPY1 and MPY2 (40). To create pMP56, the wild-type ORF of VPS11 was amplified from genomic DNA using flanking oligos approximately 500 bases upstream from the start site and approximately 300 bases downstream from the stop codon. The resulting PCR product was subcloned via the Topo-TA kit (Invitrogen, Carlsbad, CA, USA), and then subsequently subcloned into the yeast shuttle vector pRS416 using BamHI sites added by the oligos. The GST-Vac1 plasmid was created by PCR amplification of the wild-type ORF of VAC1 using oligos complementary to the start site and approximately 300 bases downstream of the stop codon, and subcloning the resulting product via the Topo-TA kit. This fragment was subsequently subcloned in frame into pGEX-KG (Amersham-Pharmacia Biotech, Piscataway, NJ, USA) using the XhoI and HindIII sites added by the PCR oligos, to make pCB192. The creation of the wild-type and mutant Class D vps alleles utilized in Figure 4 are described in: VAC1 (pCB174 and p414.19; (17)), VPS45 (pVPS45-10 and pVPS45-37; (33)), VPS21 (pMP37 and pMP38; (25)).

Generation of mutant vps11 alleles

The temperature-conditional mutants of VPS11 were constructed by PCR-based mutagenesis (29). Primers complementary to chromosomal sequences upstream and downstream of the VPS11 ORF were used to amplify a 3.8-kb fragment under dATP limiting conditions (60 µm; dCTP, dGTP, and dTTP were present at 200 µm). A gapped plasmid was generated by digesting pMP56 with NruI and SphI and purifying the vector-containing fragment by gel purification. The mutagenized product and gapped plasmid were cotransformed into BHY10 vps11Δ::HIS3 (MPY7), and transformants in which homologous recombination resulted in integration of mutagenized PCR product were selected by uracil prototrophy. Transformants were replica plated onto YPF, grown overnight at 26 °C, and tested by colorimetric invertase assay (41) for temperature-conditional vps phenotype (CPY-invertase secretion) at 26 °C and after 2.5 h shift to 38 °C. Putative temperature-conditional strains were picked, and retested, and plasmid linkage of the missorting phenotype was confirmed by retransformation of isolated plasmids into SEY6210 vps11Δ::HIS3 (MPY8). Plasmids which were isolated in this manner were integrated into the VPS11 locus of the wild-type strain (SEY6210) via pop-in/pop-out allele replacement (30), resulting in vps11–1ts (MPY9) and vps11–3ts (MPY10).

CPY and ALP sorting assays

Yeast cultures were radiolabeled by previously published procedures (27,32). Immunoprecipitations of CPY and ALP were done as described previously (31,42).

Microscopy

Vacuolar morphology was assessed by 7-amino 4-chloromethylcoumarin (CMAC; Molecular Probes, Eugene, OR, USA). Staining protocol was as follows: 1 mL of mid-log phase culture was subjected to gentle centrifugation, followed by resuspension in 50 µL of media plus 1 µL of 10 mm CMAC. Cells were incubated at 30 °C for 10 min, and then washed with 1 mL of media. Cells were gently pelleted and resuspended in 50 µL media. CMAC staining was observed on a Deltavision deconvoluting microscope (Applied Precision, Issaquah, WA, USA) Electron microscopy was performed as described in (43).

In vitro GST-Vac1 binding analyses

The plasmid encoding GST-Vac1, pCB192, was transformed into the JM101 E. coli strain, and a 2-liter culture was grown to mid-log phase. GST-Vac1 protein synthesis was induced by 200 µm isopropyl B-d-thiogalactopyranoside, and the culture was grown for an additional hour at room temperature. Cells were harvested by centrifugation and frozen at − 70 °C. The cell pellet was thawed and resuspended in 20 mL of lysis buffer (10 mm Tris 8.0, 150 mm NaCl, 1 mm EDTA, 5 mm DTT, 1.5% sarkosyl, and protease inhibitors). The cell suspension was sonicated on ice for five 15-s bursts, and the resulting lysate was cleared by centrifugation at 10 000 × g for 10 min The cleared lysate was incubated with glutathione-coupled Sepharose (Amersham-Pharmacia Biotech, Piscataway, NJ, USA) for 1 h at 4 °C with rocking. The glutathione-Sepharose beads were then washed three times in phosphate-buffered saline (PBS; 140 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4, 1.8 mm KH2PO4), then twice in PBS plus 1.0% Triton X-100, resulting in approximately 10 µg bound protein/sample. The above was repeated to generate glutathione-Sepharose beads with approximately 10 µg of GST bound as a negative control. Yeast lysates were made in the following manner: 400 OD600 units of each strain were grown to mid-log phase and harvested by centrifugation. Cell pellets were resuspended in 1 mL PBS plus 1% Triton X-100 and protease inhibitors, and transferred to 15 mL Corex tubes (VWR Scientific Products Corporation, Bridgeport, CI, USA). Approximately 0.5 g of acid-washed glass beads were added, and the tubes were then vortexed 5 times in 25-s bursts. Lysates were subjected to 10 000 × g centrifugation for 10 min, and the supernatant was harvested. Total protein levels were determined via Lowry assay; no significant differences were found between the strains utilized in this study. Cleared yeast lysates were split in two (2 mg total protein/sample), and each half was incubated with either GST or GST-Vac1 immobilized glutathione-Sepharose beads for 1 h at 4 °C. These beads were then washed three times with PBS plus 0.5% Triton X-100 and two times with PBS. Fluid was aspirated, and the beads resuspended in SDS sample buffer and heated to 65 °C for 3 min. Samples were subjected to SDS-PAGE electrophoresis and Western blotting using the anti-HA antibody.

Acknowledgments

We thank Chris Burd for his contribution of Vac1 reagents, and Andrew Wurmser, Markus Babst, and, of course, David Katzmann for discussion and critical reading of this manuscript. In addition, we thank Tammie McQuistan for assisting with the electron microscopy analysis (Immuelectron microscopy Core B of the Program Project Grant CA58689 headed by M. Farquhar). MRP is supported by the UCSD Medical Scientist Training Program, which is funded by the National Institute of Health, and is a member of the Biomedical Sciences graduate program at UCSD. This work is supported by National Institutes of Health grant CA58689 (to SDE). SDE is also supported as an Investigator of the Howard Hughes Medical Institute.

Ancillary