A screen for vacuole morphology defects upon gene overexpression
Overexpression of genes can result in significant imbalances in protein complexes by sequestering critical factors. We reasoned that negative regulators of protein transport should affect endosomal and vacuole biogenesis and therefore screened an overexpression library for aberrant vacuole morphology. All genes in this library are encoded by a high-copy 2-µ plasmid and under the control of the strong, inducible GAL1 promoter. Due to their cloning, all genes contain a C-terminal His6-HA-ProteinA cassette, which did not interfere with the functionality of most proteins (19), although we took this possibility into account during our analysis (see below). To identify proteins that could interfere with endosomal and vacuole biogenesis, we induced the expression of plasmid-encoded genes for 24 h at 30°C in rich and nonselective medium (YPG), stained vacuoles with the lipophilic dye FM4-64 and analyzed vacuole morphology by fluorescence microscopy. The plasmid loss was around 10% under these conditions. We then classified vacuole morphology according to previous VPS screens (13,14,16): wild-type cells have one to three normal round vacuoles, class B mutants multiple smaller vacuoles, class C mutants have strongly fragmented vacuoles, class D mutants an enlarged vacuole and class E mutants have an enlarged prevacuolar compartment (Figure 1A). Of the 6153 open reading frames (ORFs), 218 caused altered vacuole morphology. Several ORFs that caused a mild phenotype were eliminated from further analysis. Likewise, proteins with an assigned function at a distinct compartment like the endoplasmic reticulum (ER) were not included. We then scored the 218 ORFs under conditions of plasmid maintenance and deleted ORFs showing mild phenotypes under these conditions. This procedure reduced the list to 53 ORFs (Tables 1–5).
Figure 1. Representative vacuolar phenotypes. Overexpression of certain genes affects vacuole morphology. Strain Y258 carrying different overexpression plasmids (pBG1805) was grown in selective medium containing 2% galactose (SGC) to induce overexpression. Vacuoles were stained with FM4-64 as described in Materials and Methods. A) Representative phenotypes obtained in the screen are shown. Indicated phenotypes correspond to following genes: class B, YOL159C-A; class C, IVY1; class D, FRK1; class E, VPS2. B) Examples of genes that affect cell and vacuole morphology when overexpressed. C) Phenotype induced by overexpression of VPS4. Size bar = 10 µm. DIC, differential interference contrast.
Download figure to PowerPoint
Table 1. Known catalysts in vacuolar trafficking
|Locus||Common name(s)||Vacuolar phenotype (overexpression)||Vacuolar phenotype (deletion)||CPY secretion||CAN sensitivity||Localization||Function|
|ESCRT and related|
| YCL008C||VPS23||10% D, 10% B, 10% E||E||+||+||Endosome||ESCRT-I complex, MVB sorting|
| YJR102C||VPS25||25% D, 5% B (5% amorph)||E||+||+||Endosome||ESCRT-II complex, MVB sorting|
| YKL002W||VPS2, DID4||30% E||E||+||+||Endosome||ESCRT-III complex, MVB sorting|
| YPL065W||VPS28||40% D, 10% B||E||−||+||Endosome||ESCRT-I complex, MVB sorting|
| YPL084W||BRO1||15% B, 5% D||E||+||+||Endosome||MVB sorting, Doa4 recruitment|
| YPR173C||VPS4||35% E (15% amorph)||E||−||+||Endosome||AAA-ATPase, ESCRT-III disassembly|
| YDL077C||VPS39||10% B, 10% C||C||+||+||Vacuole||HOPS complex, vacuole fusion|
| YDR495C||VPS3||10% B, 45% C (10% amorph)||D||+||+||Endosome||CORVET complex|
| YLR148W||VPS18||50% B, 20% D (larger cells)||C||+||−||Vacuole||HOPS/CORVET complex|
| YCR068W||ATG15||75% B|| ||—||n.d.a||Vacuole||Intravacuolar lipase|
| YDR229W||IVY1||55% C|| ||−||+||Vacuole|| |
| YER072W||VTC1||10% B, 10% C|| ||−||+||Vacuole||VTC complex|
| YJL059W||YHC3, BTN1||25% B|| ||−||−||Vacuole||Transport of arginine into vacuole|
| YMR297W||PRC1||20% B|| ||+||−||Vacuole||Vacuolar CPY|
| YBR164C||ARL1||30% B (15% amorph)|| ||+||+||Cytosol||Soluble GTPase involved in trafficking|
| YIL088C||AVT7||20% B (10% amorph)|| ||+||−||Vacuole/PM||Vacuolar amino acid transporter|
| YIL173W||VTH1||35% B, 5% C (10% amorph)|| ||−||n.d.a||Endosome||Putative cargo receptor (Vps 10-homolog)|
| YKR014C||YPT52||20% C, 10% D (25% amorph)|| ||−||−||Endosome||Rab GTPase, endocytosis|
| YOR036W||PEP12, VPS6||30% B, 5% D||D||+||+||Endosome||Endosomal t-SNARE|
| YOR212W||STE4||20% B, 15% D (40% amorph)|| ||−||−||PM||Activation of mating signaling pathway|
Table 2. Protein modification
|Locus||Common name(s)||Vacuolar phenotype (overexpression)||CPY secretion||CAN sensitivity||Localization||Function|
| YDR075W||PPH3||20% B (10% amorph)||—||+||Cytosol/nucleus||Protein phosphatase complex|
| YKL126W||YPK1||10% B||+||−||Cytosol/PM||Ser/Thr protein kinase|
| YPL141C||FRK1||95% D||+||n.d.a||Cytoplasm||Putative kinase|
| YDR054C||CDC34||15% B, 20% E||+||−||Cytosol/nucleus||Ubiquitin-conjugating enzyme (E2)|
| YDR143C||SAN1||20% B, 10% D (10% amorph)||+||n.d.a||Cytosol/nucleus||Ubiquitin-protein ligase|
| YMR275C||BUL1||35% B, 15% D||n.d.b||n.d.b||PM||Ubiquitin-binding component of Rsp5|
| YOL003C||PFA4||25% B, 10% D||−||−||Cytosol/nucleus||Palmitoyltransferase|
Table 3. Other known catalysts
|Locus||Common name(s)||Vacuolar phenotype (overexpression)||CPY secretion||CAN sensitivity||Localization||Function|
|YBR159W||IFA38||30% B, 40% C||−||+||ER||Microsomal β-keto-reductase|
|YDR169C||STB3||40% B, 5% E||—||−||Cytosol/nucleus||RRPE-binding protein|
|YDR525W-A||SNA2||25% B, 10% D (10% amorph)||−||+||Cytosol|| |
|YGR109C||CLB6||35% B (10% amorph)||—||n.d.a|| ||B-type cyclin|
|YHR061C||GIC1||40% B, 5% D (40% amorph)||−||n.d.a||Bud neck||Involved in budding and cellular polarization|
|YLR227C||ADY4||30% B||+||n.d.a||Spindle||Component of meiotic outer plaque|
|YLR443W||ECM7||60% B (10% amorph)||+||−|| ||Integral membrane protein involved in calcium uptake|
|YNL233W||BNI4||90% B, 5% D (100% amorph)||+||−||Bud neck||Targeting of Glc7 and chitin synthase III|
|YOR166C||SWT1||25% B||−||−||Cytosol/Nucleus||RNA endoribonuclease|
Table 4. Putative proteins of unknown function
|Locus||Vacuolar phenotype||CPY secretion||CAN sensitivity||Cps 1 trafficking||Ste3 sorting||AP-3 transport||Localization|
|YBR071W||20% B (25% amorph)||—||+||−||+||−||Bud neck/cytosol|
|YDR514C||50% B, 10% D||−||+||−||−||−||Mitochondrion|
|YER128W||30% B, 20% D||+||+||+||−||−||Endosome, cytosol|
|YGR204C-A||15% B, 20% C||+||−||−||+||−|| |
|YHR112C||40% B (10% amorph)||+||−||−||+||−||Nucleus/cytosol|
|YIL089W||20% B, 5% D||+||+||+||−||−||ER/vacuole|
|YKL222C||25% B, 15% C||+||−||−||+||−||Nucleus/spindle|
|YKR041W||30% B, 10% E (35% amorph)||+||+||−||−||−||Nucleus/spindle|
|YOR223W||10% B, 10% D||−||n.d.a||−||−||−||ER/vacuole|
Table 5. Orphan ORFs
|Locus||Vacuolar phenotype||CPY secretion||CAN sensitivity|
|YBL094C||15% B, 5% D||—||+|
|YDL162C||35% B (5% amorph)||−||−|
|YFL030C-A||25% D, 5% B (5% amorph)||−||−|
|YGL052W||30% D, 10% B||−||−|
|YOR309C||30% B, 30% B||—||−|
|YOR314W||20% B, 10% D||−||−|
To validate our screen, we compared our candidates to known factors that caused vacuole fragmentation upon overproduction. The CORVET subunit Vps3 is believed to interfere with the biogenesis of the vacuolar HOPS tethering complex if present in excess (20). Vps3 was indeed among the isolated candidates. Likewise, Ivy1 was identified in our screen. This protein interacts with the HOPS subunit Vps33 and the Rab Ypt7, and is induced during stationary phase (21). The re-identification of Vps3 and Ivy1 indicates that (i) our screen was indeed able to identify endosomal factors and (ii) that the C-terminal tags generally did not inactivate the proteins. In addition to these factors, we could identify several genes that interfered with cell (and vacuole) morphology, in good agreement with a previous study (19) (Figure 1B).
During the course of our analysis, we noticed that several transmembrane proteins of the ER and Golgi affected vacuole morphology, presumably due to spill-over from the resident compartment and subsequent sorting into the endocytic pathway. We therefore did not include most membrane proteins with multiple transmembrane domains in our further analysis, although we are aware that we might miss out on potential regulators. The remaining transmembrane proteins in our list may have similar side effects on trafficking.
To further validate our results, we employed two distinct assays. VPS proteins were initially identified in screens that scored for the secretion of the vacuolar hydrolase carboxypeptidase Y (CPY). We thus measured CPY secretion upon overexpression, taking the BY wild-type strain as a control. In contrast to the corresponding wild-type employed in the VPS screen (13), BY wild-type cells secrete some CPY (Figure S1). Consequently, we were able to observe both suppression of secretion and its enhancement due to gene overexpression (Tables 1–5 and Figure S1). The reduction in CPY secretion is possibly due to interference with general secretion or more efficient recycling of the cargo receptor Vps10. The exact reason for this phenomenon is not yet known. As a second assay, we employed sensitivity to the toxic arginine homolog canavanine, which is transported into cells via the arginine permease Can1. Can1 normally cycles between endosomes and the plasma membrane and therefore leaves wild-type cells resistant to canavanine, whereas ESCRT deletions, for instance, result in increased sensitivity due to the stabilization of the permease at the plasma membrane (22). Indeed, several overexpressors-induced canavanine sensitivity (Figure S2). Below, we have grouped the different genes according to their known function and evaluated the implications in the context of the literature.
Identification of endosomal proteins that affect vacuole biogenesis
Our main interest of the screen was the identification of endosomal and vacuolar proteins with a putative regulatory function in trafficking. Indeed, we could identify three proteins of the HOPS and CORVET complex, several ESCRT subunits (presented below) and 10 endosomal and vacuolar proteins, which reproducibly affected vacuole morphology upon overexpression (Table 1). Besides Vps3 and Ivy1, we isolated the HOPS subunit Vps39 and Vps18. In previous studies, we observed that overproduction of chromosomally integrated VPS39 interfered with vacuole morphology, while overproduction of chromosomally integrated VPS18 did not (23,24), indicating that this effect is dose dependent. Interestingly, other HOPS and CORVET subunits were not identified, suggesting a more dynamic association of Vps39, Vps18 and Vps3 with their complexes, as previously demonstrated for Vps3 (20).
Seven of the characterized endosomal proteins were probably identified due to their general negative impact on the protein sorting machinery. This includes Atg15, a transmembrane protein with lipase activity found in the vacuole lumen (25); Avt7, a multispan amino acid transporter of the vacuole (26); Btn1, an arginine transporter that has been implicated in acidification of the yeast vacuole (27); Pep12, the endosomal syntaxin homolog involved in transport between Golgi and endosomes (28); Vtc1, a subunit of the Vtc complex of polyphosphate phosphatases implicated in vacuole fusion and microautophagy (29,30); Vth1, a Vps10-like transporter (31) and Prc1 (CPY) (32), which is transported to the vacuole via the Vps10 cargo receptor (33). CPY is known to accumulate in the culture supernatant, if provided in excess.
We also identified three proteins related to GTPase-dependent reactions. The Arf-like Arl1 protein (34,35) is involved at the endosome–TGN interface and required for the GGA-dependent transport to the vacuole. Arl1 overproduction results in increased canavanine sensitivity (Figure S2), consistent with a defect in Can1 recycling at the endosome. Furthermore, we identified Ste4, which affects both cell and vacuole morphology upon overproduction (Figure 1B). Ste4 is the β-subunit of the heterotrimeric G-protein complex, which activates together with the γ-subunit Ste18 the mating signaling pathway (36). We assume that the excess of Ste4 displaces its α-subunit Gpa1, which may subsequently traffic to endosomes and perturb its biogenesis by binding and activating Vps34 (37). Finally, Ypt52 was identified, which is homologous to Rab5 GTPase in yeast, Vps21/Ypt51 and Ypt53 (38). Its function in yeast is still uncertain, even though it binds like Vps21 to the CORVET subunit Vps8 and accumulates like many endosomal proteins in enlarged class E endosomes (39) (our unpublished observations). We assume that the overexpressed Rab is not prenylated due to its C-terminal tags, and thus generate an inactive, soluble Rab, which could interfere with endosomal function.
Proteins involved in posttranslational modifications
Within our screen, we expected to find kinases and phosphatases that may target proteins involved in endosomal biogenesis (Table 2). Indeed, two kinases (Ypk1 and Frk1) and one phosphatase (Pph3) were identified. Among these, Pph3 is an interesting candidate as its overexpression results in a strong sensitivity to canavanine and suppression of CPY secretion. Pph3 is part of a phosphatase complex with multiple functions in cell division, DNA repair and Tor signaling (40–43).
Overexpression of the E2-enzyme Cdc34, the Rsp5 interacting protein Bul1 and the ubiquitin ligase San1 also resulted in alterations in vacuole morphology, as did the overexpression of the acyltransferase Pfa4, a polytropic membrane protein. The relevance of these factors is not yet clear as only CDC34 and SAN1 overexpression resulted in a weak CPY secretion phenotype. Table 3 lists additional factors identified in our screen, which we do not take into account as endosomal regulators, because they have assigned functions elsewhere in the cell. For instance, BNI4 overproduction results in completely amorphous cells (Table 3 and Figure 1B), consistent with earlier observations (19). Therefore, the CPY secretion phenotype in these cells is probably due to cell lysis.
Several ESCRT proteins affect vacuole biogenesis
One striking observation was the identification of several ESCRT proteins that affected vacuole morphology upon overproduction (Table 1). We did not observe any preference for one of the four complexes, and found Vps23 and 28 (ESCRT-I), Vps25 (ESCRT-II), Vps2 (ESCRT-III) and Vps4. Some of the overexpression phenotypes may be due to the C-terminal tagging, although we would expect a class E-like phenotype under these conditions. This was only the case for VPS4 and VPS2 overexpression (Table 1). In addition, Vps4 overproduction resulted in FM4-64 positive dots that were present throughout the cytoplasm (Figure 1C). These dots might represent endosomal intermediates, which would indicate that the endocytic pathway is slowed down under these conditions. Furthermore, two hallmarks of ESCRT deletions, sensitivity to canavanine (Figure S2) and missorting of Cps1 (not shown) were also observed for ESCRT overexpressors. Almost all of these subunits also promoted CPY secretion (Table 1 and Figure S1). These observations are similar to the findings on Ist1, a negative regulator of Vps4 function, which also resulted in additional ESCRT-specific phenotypes like missorting of green fluorescent protein (GFP)-Cps1 upon overexpression (17) (M. Babst, personal communication). In addition to the described ESCRT subunits, 12 other genes caused canavanine sensitivity upon overexpression. Among these, three previously uncharacterized ORFs showed strong defects comparable to ESCRT overexpression: YDR541C, YIL089W and YER128W (Table 4). Deletion of these genes did not cause a GFP-Cps1 trafficking defect, which would be expected for negative regulators of ESCRT function (see below).
New players in trafficking to the vacuole
In addition to proteins with known function, the screen revealed several proteins of unknown function, which might have a role in trafficking to the vacuole (Table 4) and some ambiguous ORFs (Table 5). To unravel novel players in trafficking to the vacuole, we focused on the uncharacterized proteins (Table 4). We first analyzed, which trafficking routes to the vacuole were affected by overproduction. We looked for defects in localization of Ste3-GFP and GFP-CPS as model cargoes of the endocytic pathway, and GNS (GFP-Snc1-TMD-Nyv1)(44) as a cargo of the AP-3 pathway (Figure 2). This analysis enabled us to arrange some of the uncharacterized genes into groups that share certain phenotypes and therefore might act in one trafficking pathway as discussed below. We also analyzed the deletion mutants of these genes in the same trafficking assays, but could not detect clear mislocalization of the marker proteins (Figure S3A). However, we detected a weak mislocalization of Ste3-GFP to the ER in some deletion mutants (Figure S3A). The deletion mutants also showed no increased sensitivity to canavanine or abnormal CPY secretion (not shown).
Figure 2. Phenotypic characteristics caused by overexpression of uncharacterized genes. Eleven uncharacterized genes identified in the screen (Table 4) were analyzed for trafficking defects to the yeast vacuole. A) Wild-type and indicated deletion strains were grown in synthetic medium to retain empty pRS426 plasmids and CEN plasmids containing trafficking markers (pCU1775, GFP-CPS; pCU2944, GNS; pCU3078, STE3-GFP) as control. Overexpression was induced by addition of 2% galactose and incubation for 24 h followed by observation under the microscope. B–D) Cells carrying indicated marker plasmids and 2-µ plasmids for overexpression of indicated genes were grown as described in (A). Size bar = 5 µm. DIC, differential interference contrast.
Download figure to PowerPoint
The overexpression of YBR071W, YGR204C-A, YHR112C and YKL222C resulted in an accumulation of the endocytic cargo Ste3 in the cell, without affecting sorting of the biosynthetic cargo Cps1 (Figure 2B), which indicates that these proteins might have a role at the early endosome or during endocytosis.
The genes YDR541C, YER128W and YIL089W share an accumulation of GFP-CPS at prevacuolar compartments and on the vacuolar membrane, as well as increased canavanine sensitivity when overexpressed (Figures 2C and S1). In addition, overexpression of these genes causes CPY secretion (Figure S1). Interestingly, overexpression of all of these genes triggers a partial class D-like vacuole morphology, which was also observed for some ESCRT subunits (Table 1), supporting the idea that these proteins affect endosomal maturation or ESCRT activity.
YDR514C, YKR041W, YOL159C-A and YOR223W showed no missorting phenotype of either model cargo (Figure 2D). Therefore, it remains unclear if these proteins have a specific role in trafficking.
None of the overexpressors affected the AP-3 pathway (Figures 2 and S3A), which can be explained by the fact that even deletions of essential AP-3 coat proteins have wild-type vacuoles (10).
In addition to the trafficking assays described above, we tried to investigate effects on the Cvt pathway by following the clipping of Ape1 to its mature form due to fusion of Cvt vesicles with the vacuole (45). While we were able to detect a weak clipping defect for the deletion of YKR041W (Figure S3B), we were unable to carry out the assay under overexpression conditions due to the increased level of proteinA-tagged protein that interfered with the western blot detection of Ape1 (not shown).
To further investigate the role of the newly identified candidates, we localized these proteins by C-terminal GFP-tagging (Figure 3A–E). We observed an unexpected variety of localizations, which makes a common assignment of their interfering function in vacuole biogenesis challenging. Yhr112c and Yol159c-a localize to the nucleus and to the cytoplasm. Ykr041w and Ykl222c also localize to the nucleus and the mitotic spindle (Figure 3A). Ydr514c, however, clearly localized to mitochondria (Figure 3B). Yil089w and Yor223w were found at the ER and partially in the vacuolar lumen (Figure 3C). These two proteins have predicted transmembrane domains (http://www.yeastgenome.org). We noticed though that even strong overexpression of YIL089W from the GPD promoter does not lead to mislocalization of this protein to other compartments (Figure 3F). Ybr071w was found in the cytosol and at the bud neck (Figure 3D). This agrees with the high percentage of amorphous cells with elongated buds observed, when this gene was overproduced (not shown). Overexpression probably interferes with septin ring formation or actomyosin ring contraction. GFP-tagged Ydr541c and Ygr204c-a could not be detected if expressed from a plasmid but overexpressed Ydr541c was located in the cytoplasm (Figure 3F).
Figure 3. Localization of uncharacterized proteins identified in the screen and effect of untagged protein overexpression. A–E) Localization of uncharacterized proteins identified in the overexpression screen. Expression of C-terminally tagged proteins from a CEN plasmid under control of the NOP1pr in wild-type strain. Cells were grown in synthetic medium to retain plasmids. F) Localization of Ydr541c, Yer128w and Yil089w by GFP-tagging in the genome. Growth in YPD followed by washing in synthetic medium to reduce background in the GFP channel. G) Yeast strains overexpressing YDR541C, YER128W or YIL089W from the strong GPD promoter were transformed with plasmid pCU3078 to analyze Ste3 trafficking or Cps1 was N-terminally tagged with PHO5pr-GFP to analyze biosynthetic transport. Size bar = 5 µm. DIC, differential interference contrast.
Download figure to PowerPoint
Interestingly, the protein Yer128w was detected in the cytoplasm and in dots proximal to the vacuole, even if expressed from the endogenous promoter (Figure 3A,F; see below).
As overexpression of YER128W, YDR541C and YIL089W showed the strongest influence on trafficking of GFP-CPS (Figure 2C and Table 4), we asked whether missorting phenotypes could be reproduced for overexpression of the untagged proteins under control of the strong GPD promoter. For YDR541C, we could reproduce the defects, whereas overproduction of the other two proteins was without effect (Figure 3G). At present, we cannot yet distinguish if this wild-type phenotype is due to the reduced overproduction or the removal of the epitope tags.
The novel endosomal protein Vfa1/Yer128w interacts with Vps4
We then decided to explore the role of Yer128w in more detail. To address whether Yer128w localizes to endosomes, we performed colocalization studies with the endosomal Rab GTPase Vps21, the late endosomal/vacuolar Rab Ypt7 and Mnn9 as a resident protein of the Golgi (23,39). Yer128w clearly colocalizes with Vps21-positive structures and partially to Ypt7-positive structures, but not with Mnn9 (Figure 4A). Furthermore, previous high-throughput analyses identified Yer128w as a Vps4 interactor (46,47). We thus tested this physical interaction directly using a pull-down approach with purified Vps4 and observed a specific interaction (Figure 4B). We therefore named YER128W VFA1 for Vps Four-Associated 1.
Figure 4. The endosomal protein Vfa1/Yer128w interacts with Vps4. A) Yeast with chromosomally GFP-tagged Vfa1 (YER128W) carrying plasmids encoding red fluorescent protein (RFP)-VPS21 or RFP-YPT7 as early and late endosomal markers or integrated MNN9-TOMATO as Golgi marker were grown in synthetic medium and analyzed by fluorescence microscopy. Colocalization of Vps21 and Yer128w/Vfa1 are indicated with arrows. B) Localization of Vfa1 depends on Vps4. Wild-type or vps4 mutant cells were transformed with plasmids encoding NOP1pr-VFA1-eGFP. Growth in synthetic medium, FM4-64 staining as described in Materials and Methods. C) Vfa1 interacts with Vps4. Recombinant His6-Vps4 or buffer was added to Ni-NTA beads. Yeast lysate of a strain with C-terminal TAP-tagged VFA1 was used for Ni-NTA pull down. Retained proteins were analyzed by western blot with a PAP or anti-His antibody. A and C) Size bar = 5 µm.
Download figure to PowerPoint
Another Vps4-interacting protein, Ist1, also localizes to the class E compartment, when VPS4 was deleted (17). When we analyzed the Vfa1 in vps4 mutant cells, we could not detect any dot-like localization, although the class E compartment was clearly visible (Figure 4C). This indicates that Vfa1 requires Vps4 for its recruitment to endosomes.