Yeast Translation Elongation Factor-1A Binds Vacuole-localized Rho1p to Facilitate Membrane Integrity through F-actin Remodeling*

Background: eEF1A is a dual-function protein with independent roles in RNA translation and actin organization. Results: eEF1A interacts with Rho1p on yeast vacuolar membranes. Conclusion: Rho1p-eEF1A may be required for membrane-associated F-actin organization. Significance: eEF1A links vacuole-associated F-actin and membrane integrity to Rho1p signaling. Rho GTPases are molecular switches that modulate a variety of cellular processes, most notably those involving actin dynamics. We have previously shown that yeast vacuolar membrane fusion requires re-organization of actin filaments mediated by two Rho GTPases, Rho1p and Cdc42p. Cdc42p initiates actin polymerization to facilitate membrane tethering; Rho1p has a role in the late stages of vacuolar fusion, but its mode of action is unknown. Here, we identified eEF1A as a vacuolar Rho1p-interacting protein. eEF1A (encoded by the TEF1 and TEF2 genes in yeast) is an aminoacyl-tRNA transferase needed during protein translation. eEF1A also has a second function that is independent of translation; it binds and organizes actin filaments into ordered cable structures. Here, we report that eEF1A interacts with Rho1p via a C-terminal subdomain. This interaction occurs predominantly when both proteins are in the GDP-bound state. Therefore, eEF1A is an atypical downstream effector of Rho1p. eEF1A does not promote vacuolar fusion; however, overexpression of the Rho1p-interacting subdomain affects vacuolar morphology. Vacuoles were destabilized and prone to leakage when treated with the eEF1A inhibitor narciclasine. We propose a model whereby eEF1A binds to Rho1p-GDP on the vacuolar membrane; it is released upon Rho1p activation and then bundles actin filaments to stabilize fused vacuoles. Therefore, the Rho1p-eEF1A complex acts to spatially localize a pool of eEF1A to vacuoles where it can readily organize F-actin.

late the release of GDP, allowing GTP to bind. Once activated, Rho GTPases recruit downstream effectors, creating membrane-bound signaling complexes where actin dynamics are stimulated and membrane remodeling occurs (1,3). Rho GTPase activity is down-regulated by GTPase-activating proteins (GAPs), which stimulate GTP hydrolysis. Some Rho family proteins can also be dissociated from membrane by binding to RhoGDI (Rho guanine nucleotide dissociation inhibitor), which sequesters the isoprenyl group in a hydrophobic pocket creating soluble 1:1 complexes (5,6).
Rho GTPases are known to activate two different types of effectors that directly stimulate actin polymerization. WASP/ WAVE proteins (Las17p in Saccharomyces cerevisiae), when bound by Cdc42p, serve as the signaling scaffold for the activation of the Arp2/3 complex leading to the formation branched actin filaments (7). RhoA (Rho1p in S. cerevisiae) activates Diaphanous-related formins (Bni1p/Bnr1p in S. cerevisiae), which in turn stimulate the nucleation and elongation of nonbranching actin filaments (8).
We have used yeast vacuolar membranes to study Rho GTPase/actin-mediated modulation of membrane dynamics during vesicular transport and membrane fusion (9). Previously, we have shown the requirement for actin dynamics and turnover in membrane fusion (10), and we have linked fusiondependent actin dynamics on vacuoles with two Rho GTPases, Rho1p and Cdc42p (11)(12)(13). Cdc42p is rapidly activated during in vitro fusion reactions and stimulates the rapid polymerization of actin (11,14). Rho1p activation kinetically coincides with fusion kinetics; however, further understanding of the action of Rho1p remains unknown. Identifying interacting partners during fusion reactions may provide some insights into this question. Therefore, our aim here was to identify vacuolar Rho1p-binding partners that could potentially be involved in actin remodeling-coupled membrane fusion.
In this study, we found that translation elongation factor 1-␣ (eEF1A or Tef1p in S. cerevisiae) was present in low abundance on isolated vacuoles and associated with Rho1p. eEF1A is a highly abundant cytosolic protein encoded by two sequence-redundant genes, TEF1 and TEF2. eEF1A is a GTPase component of the eukaryotic elongation factor complex (15). In the GTP-bound active form, eEF1A binds to and delivers aminoacylated tRNA to the A-site of ribosomes for elongation of nascent polypeptides. The ribosome acts as a GTPase activator (GAP) for eEF1A when a correct codon-anticodon match occurs between the aminoacyl-tRNA and the A-site codon of the ribosome-bound mRNA. The inactive GDP-bound form of eEF1A is released from the ribosome and can be reactivated by its GEF, eEF1B␣, before binding another aminoacyl-tRNA.
Intriguingly, eEF1A is a multifunctional protein which, beyond its canonical role in translation elongation, also functions in various cellular processes, including nuclear export and F-actin remodeling (16 -18). The noncanonical function of eEF1A, to bind and bundle actin filaments, has been shown across species from yeast to mammals (19 -25). Current models propose that eEF1A is a key factor in regulating cytoskeleton organization by creating tightly packed F-actin cables.
Here, we identified eEF1A as a binding partner of Rho1p on yeast vacuolar membranes. The Rho1p-eEF1A interaction likely has functional implication in regulating actin remodeling coupled to vacuolar fusion because actin is enriched at the "vertex ring" of membranes during vacuolar docking, and actin remodeling is required for the terminal step leading to vacuolar membrane fusion (10,26). Therefore, we hypothesize that eEF1A is recruited to vacuoles from the cytosol by Rho1p and may mediate Rho1p signaling for actin reorganization. Here, we characterized the Rho1p-eEF1A interaction, including the binding specificity, the putative binding site, the regulation of interaction, and the functional implication of eEF1A in vacuolar fusion.

EXPERIMENTAL PROCEDURES
Biochemical Reagents and Antibodies-Proteinase inhibitor cocktail (PIC) was made as ϫ60 stock solution (ϫ60 PIC ϭ 10 g/ml leupeptin, 20 g/ml pepstatin, 25 mM o-phenanthroline, 5 mM Pefabloc SC). Mouse anti-GST antibody was purchased from Sigma; rabbit anti-MBP antibody was from New England Biolabs; mouse anti-HA antibody was from Roche Applied Science; and rabbit anti-GFP antibody was a gift from Dr. L. Berthiaume (University of Alberta). Vac8p, Nyv1p, Act1p, Rho1p, and Vma1p antibodies have been described previously (10). Alexa Fluor 680/750 goat anti-mouse and goat anti-rabbit IgG secondary antibodies were purchased from Invitrogen.
Yeast Strains and Preparation of Lysates-Yeast strains used in this study are listed in Table 1. Strains expressing C-terminal GFP-tagged eEF proteins and BNI1/BNR1 gene deletions were purchased from Open Biosystems Inc. Strains with C-terminal 3ϫHA-tagged eEF1A were constructed by integration of a PCR product generated from pFA6a-3HA-His 3 MX6 (27). Strains harboring plasmids with N-terminal GFP-tagged and 7ϫHAtagged eEF1A (full length and domains) were generated by drop and drag cloning using the plasmids pGREG536 and pGREG576, respectively, and transformed into strain BJ5459 (28). Strains were grown at 30°C in YPDG (1% yeast extract, 2% peptone, 1% dextrose, 1% galactose) or YPDG with 40 g/ml kanamycin for strains harboring plasmids. All tagged eEF1A proteins were expressed in a background that also contained the endogenous proteins. Therefore, it was not determined whether tagged proteins were fully functional. Yeast lysates were prepared by suspending cells in lysis buffer (20 mM PIPES, pH 6.8, 60 mM KCl, 1 mM MgCl 2 , 0.1 mM DTT, 0.1 mM PMSF, 1ϫ PIC, 10% (v/v) glycerol) and passing the suspension through an emulsiflex C-3 (Avestin) for 5 min at 25,000 p.s.i. The emulsion was cleared by centrifugation for 45 min at 4°C and 20,000 ϫ g, and the supernatant was collected. Typical cytosol protein concentrations were 15 mg/ml.
Recombinant Protein Production in E. coli-GST-and MBPtagged proteins were cloned by amplifying open reading frames from genomic DNA and ligating fragments into pGEX-4T1 (GE Healthcare) and pMAL-c2x (New England Biolabs), respectively. Cdc42p and Rho1p were cloned without the four C-terminal amino acids (ϪCAAX box, where A is any aliphatic amino acid and X is any amino acid). Full-length eEF1A was cloned as an MBP fusion protein because the GST fusion protein was insoluble. GST and MBP proteins were expressed in Escherichia coli Rosetta cells (Novagen) in LB media containing 100 g/ml ampicillin and 34 g/ml chloramphenicol. 1-Liter cultures were grown to an A 600 of 0.4 -0.6, cooled to room temperature, and isopropyl 1-thio-␤-D-galactopyranoside was added (0.5 mM final) to induce protein expression. After 5 h, cells were harvested by centrifugation for 10 min at 5000 ϫ g, washed once in distilled water, and suspended in 20 ml of H-Buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 0.1 mM DTT, 0.1 mM PMSF, 1ϫ PIC). Cells were lysed by running the suspension through an emulsiflex C-3 for 5 min at 15,000 p.s.i. After emulsification, Triton X-100 was added to 0.5% (v/v), and

eEF1A Associates with Vacuolar Rho1p
the emulsion was centrifuged for 45 min at 4°C and 20,000 ϫ g. GST-tagged and MBP-tagged proteins were bound to glutathione-agarose (Sigma) or amylose resin (New England Biolabs), and 15 l of packed beads were used in pulldown assays. Elution of GST-tagged proteins was in 3 volumes of 50 mM Tris-Cl, pH 8.5, 10 mM reduced glutathione; elution of MBP-tagged proteins was in 3 volumes of 50 mM Tris-Cl, pH 7.5, 200 mM NaCl, 10 mM maltose. Protein samples were buffer-exchanged into desired buffers using a Sephadex G-25 column (GE Healthcare).
Pulldown Assays and Immunoprecipitation-Soluble extracts from E. coli strains expressing recombinant proteins were incubated with 15 l of glutathione or amylose resin per reaction. The resin was washed three times with H-Buffer, and 2 mg of target protein lysate was incubated for 90 min at 4°C in a total volume of 0.8 ml. The beads were then washed three times with H-Buffer, suspended in 1.2ϫ SSB (72 mM Tris-Cl, pH 6.8, 2.4% SDS, 12% glycerol, 6% ␤-mercaptoethanol, 0.012% bromphenol blue), and heated 5 min at 95°C before being analyzed by SDS-PAGE and immunoblot. For immunoprecipitation of GFP-eEF1A, 15 l of protein A-Sepharose resin (Pierce) was incubated with 30 l of rabbit anti-GFP serum. Antibodies were cross-linked to the resin in 100 mM HEPES, pH 8.8, 20 mM dimethyl pimelimidate. The resin was then pre-stripped with 100 mM glycine, pH 2.5, followed by three washes in H-Buffer. 1 mg of whole cell lysate or 50 g of vacuoles was added to 15 l of normal rabbit IgG beads (nonspecific) in 1 ml and nutated for 30 min at 4°C. The pre-cleared reactions were added to 15 l of GFP-specific beads and nutated for 2 h at 4°C. After immunoprecipitation, beads were washed three times with H-Buffer and then eluted using 40 l of cold 2ϫ SSB without ␤-mercaptoethanol. 20 l of 10% ␤-mercaptoethanol was added before boiling samples for 5 min at 95°C.
Mass Spectroscopy (MS) Analysis-1 mg of resin-bound MBP-Rho1p was used as an affinity ligand to bind Rho1p-interacting proteins from 0.5 mg of detergent-solubilized vacuoles that were preincubated in cytosol and fusion reaction buffer. Interacting proteins were eluted with 10 mM EDTA, run on SDS-PAGE, stained with Coomassie Brilliant Blue, and compared with eluates from resin-bound MBP eluates or mock reactions without vacuoles (Ϫ-vac). Unique bands were excised from the gel, and in-gel digestion was performed using an automated MassPREP station. The gel samples were reduced in 10 mM DTT, alkylated in 50 mM iodoacetamide, followed by overnight digestion at 37°C with 10 ng/ml trypsin in 50 mM ammonium bicarbonate, pH 8.0. Peptides were extracted, and nanospray liquid chromatography MS/MS was performed to determine peptide sequence using a quadrupole time-of-flight mass spectrometer coupled to a CapLC capillary HPLC and a nanoelectrospray source (Waters, Milford, MA). Peak lists were developed using a Mascot Distiller (version 1.1.1.0), and the appearance of three or more unique peptide sequences from one protein was used for its identification. At least five specific peptides of 10 or more amino acids were identified for each protein representing at least 15% coverage.
Chemical Nucleotide Exchange and Quantification of Binding-Resin-bound GST-Rho1p or MBP-Rho1p was washed three times in H-Buffer without MgCl 2 . EDTA was added to 3 mM, and GTP␥S or GDP was added to 40 M. Samples were incu-bated for 5 min at 30°C, then MgCl 2 was added to 10 mM. The resin was washed three times with H-Buffer containing 5 mM MgCl 2 before incubation with yeast lysates. eEF1A-GFP target lysates were similarly nucleotide-exchanged and then diluted 5-fold in H-Buffer with MgCl 2 . Immunoblots were scanned and quantified on an Odyssey infrared imaging system (Li-Cor). Bands of both GST-Rho1p or MBP-Rho1p and eEF1A-GFP were quantified using densitometry software in Odyssey version 1.2. eEF1A-GFP band levels were normalized to the 10% load band. These values were then divided by their corresponding GST, GST-Rho1p or MBP, MBP-Rho1p band value. Nonspecific binding of eEF1A-GFP to GST and MBP was subtracted from the specific binding values. 100% binding levels were set as the binding when no additional proteins or drugs were added.
Yeast Membranes-Yeast membranes were isolated from lysates by 100,000 ϫ g centrifugation. Membranes were further analyzed by sucrose density centrifugation using linear gradients of 5-50% sucrose in PS buffer. Yeast vacuoles were purified by flotation on Ficoll density gradients essentially as described previously (29). Vacuole fusion assays were done in fusion reaction buffer (FRB: 20 mM PIPES-KOH, pH 6.8, 200 mM sorbitol, 125 mM KCl, 5 mM MgCl 2 , 10 M coenzyme A, 1ϫ PIC), and contained 3.5 g of vacuoles from strains KTY1 and KTY2, an ATG-regenerating system (ATG-regenerating system ϭ 0.5 mM ATP, 0.5 mM MgCl 2 , 20 mM creatine phosphate, 0.2 mg/ml creatine kinase) and 0.5 mg/ml cytosol with test substances as indicated. Fusion reactions were incubated for 90 min at 27°C. Levels of activated alkaline phosphatase were assayed, which are indicative of the level of membrane fusion. Vacuole-associated actin polymerization assays were performed as described previously (12).
Microscopy-For live cell microscopy, overnight cultures were grown in minimal media and then re-inoculated into YPDG ϩ 40 g/ml kanamycin and grown 3 h to induce protein expression from the GAL1 promoter. Vacuoles were stained by the addition of 1 l of 4 mM FM4-64 (Invitrogen) to 100 l of cell culture. Isolated vacuoles were stained by the addition of 1 l of 0.5 mM FM4-64 dye to 50 l vacuoles at 0.3 mg/ml and either imaged immediately or exposed to drugs for 30 min at 30°C prior to imaging. Images were captured using a CoolSnap HQ camera and ImageJ software on a Zeiss Axioplan 2 using a ϫ100/1.40 NA objective. Images were processed in Adobe Photoshop and quantified with ImageJ.

RESULTS
Tef1p Interacts with Vacuolar Rho1p-Previously, we showed that two Rho proteins, Cdc42p and Rho1p, localize to the yeast vacuolar membrane and were needed for vacuolar membrane fusion (11). In reconstituted vacuolar fusion reactions, Cdc42p was rapidly activated, which resulted in the stimulation of membrane-associated actin polymerization (14,12). Rho1p showed much slower activation kinetics, and its role in fusion is unknown. To further define the function of Rho1p in membrane fusion, we searched for vacuole-specific Rho1p-associating proteins. Vacuoles were incubated for 40 min in fusion reaction buffer and cytosol, optimal conditions for Rho1p activation (14). Vacuoles were re-isolated from reactions and then deter-eEF1A Associates with Vacuolar Rho1p FEBRUARY 20, 2015 • VOLUME 290 • NUMBER 8 gent-solubilized and incubated with MBP-Rho1p or MBP bound to amylose beads. A prominent ϳ50-kDa protein that was bound specifically to MBP-Rho1p beads was identified as translation elongation factor 1〈, eEF1A (Fig. 1A). eEF1A is a dual-function protein that, in addition to its role in translation, binds and bundles F-actin filaments (18 -25). The eEF1A actin bundling activity is independent of its role in translation and does not require other factors.
In yeast, eEF1A is an essential protein encoded by duplicate genes, TEF1 and TEF2, that encode identical proteins. We tagged eEF1A with GFP and HA for further immunological and microscopic studies. Rho1p was detected in immunoprecipitates of eEF1A-GFP from vacuolar extracts but not yeast whole cell lysate (Fig. 1B). This suggests that these two proteins associate on vacuolar membranes and not in the cytosol. Because eEF1A is an abundant cytosolic protein, enrichment of the low abundance Rho1p-eEF1A complex via vacuolar isolation was required to detect this interaction. Pulldown assays using immobilized Rho1p also showed eEF1A binding (Fig. 1C). The GFP tag did not influence this association as similar results were observed in pulldown assays with C-terminal HA-tagged eEF1A. Similar eEF1A pulldown efficiency was observed using GST-Rho1p beads and MBP-Rho1p beads.
To determine whether the interaction between eEF1A and Rho1p was specific, pulldown assays were performed using lysates containing GFP-tagged eEF1A, eEF3, or the ␥ subunits of eEF1B, encoded by TEF2, YEF3, and TEF4 genes, respectively (30,31). These were incubated with immobilized GST-Rho1p, the additional Rho GTPase, GST-Cdc42p, or GST. eEF1A-GFP interacted specifically with GST-Rho1p and not with GST-Cdc42p, although eEF3-GFP and eEF1B␥-GFP showed no binding ( Fig. 2A). eEF1A is known to bind formins (Bni1p and Bnr1p in S. cerevisiae), and formins also interact with Rho1p and mediate downstream effects on actin cable formation (32). However, the eEF1A-Rho1p interaction was unaffected in pulldown assays using yeast lysate from a bni1⌬ or bnr1⌬ strain. 4 Because bni1⌬/bnr1⌬ double mutant strains are not viable, it cannot be completely ruled out whether Bni1p or Bnr1p act redundantly to bridge the eEF1A-Rho1p interaction in cells. To address this, we expressed MBP-eEF1A in E. coli. MBP-eEF1A purified from E. coli lysates showed direct binding to GST-Rho1p in a dose-dependent manner (Fig. 2B). These data indicate that the eEF1A-Rho1p interaction is direct and functionally unique for the signaling activity of Rho1p.
Nucleotide Dependence of the eEF1A-Rho1p Interaction-We next examined the nucleotide dependence of the eEF1A-Rho1p interaction. We deemed that eEF1A is most likely a downstream effector of Rho1p because its protein sequence does not contain domains typical for GEFs and GAPs such as Dbl homology-pleckstrin homology domains for Rho GEFs (33) and BH domains for Rho GAPs (34,35). Rho effectors preferentially bind to Rho proteins in the GTP-bound state (36 -38). Therefore, if eEF1A is a bona fide Rho1p effector, we would expect preferential binding to the GTP-bound state of Rho1p. To assess this, we performed Rho1p-eEF1A binding reactions using resin-bound GST-Rho1p that was pre-loaded with GDP, GTP␥S, or no nucleotide via chemically induced nucleotide exchange. eEF1A associated with Rho1p under all conditions with varying levels of affinity (Fig. 2C). Replication of binding reactions showed a 52 Ϯ 14.1% increase in eEF1A binding to GST-Rho1p in the GDP-bound state compared with the GTP␥S-bound state (Fig. 2D, left panel). Hence, this result does not support the hypothesis that eEF1A is a typical effector of Rho1p. We also performed nucleotide exchange on eEF1A lysates that showed that the GDP-bound state of eEF1A preferentially binds to Rho1p (Fig. 2D, right panel). The lower affinity of GTP␥S-bound eEF1A for Rho1p was expected because eEF1A GTP binding and hydrolysis are needed for protein synthesis (21). Although Rho1p and eEF1A are both GTPases, the functional effect of GTP binding and hydrolysis differs significantly for each protein. The Rho1p GTPase activity has no catalytic function, whereas eEF1A GTP hydrolysis catalyzes aminoacyl-tRNA transfer to ribosomes. eEF1B␣ (TEF5 gene), the GEF for eEF1A, disrupts the actin bundling properties of eEF1A and shift its function toward aminoacyl-tRNA transferase activity (21). Therefore, we investigated the effect of eEF1B␣ on the  interaction between Rho1p and eEF1A. Preincubation of eEF1A-GFP lysates with purified GST-eEF1B␣ effectively blocked the association of eEF1A with Rho1p, as detected by MBP-Rho1p pulldown assay (Fig. 3A). This blockage was a specific effect of eEF1B␣ because preincubation with GST had no effect. Subsequent incubation of reactions with glutathione resin showed eEF1A-GFP was bound to GST-eEF1B␣ (Fig. 3B). Quantitative analysis of the binding competition assay showed that eEF1B␣ reduced the ability of eEF1A to interact with Rho1p in a dose-dependent manner by as much as 89 Ϯ 10.3% (Fig. 3C). This suggests that eEF1B␣ and Rho1p competitively bind the same region of eEF1A.
eEF1A Stably Associates with Vacuoles-We next determined whether a pool of eEF1A is localized to the vacuole. Vacuoles were purified from strains expressing eEF1A-GFP, eEF3-GFP, and eEF1B␥-GFP and analyzed by comparing protein levels on vacuoles to whole cell lysates. eEF1A-GFP was enriched on vacuoles compared with other translation elongation factors (Fig. 4A). eEF1A-GFP also showed slight enrichment compared with Vma1p and actin, which are known to have a vacuole-associated pool (2-4, 12).  Tris-Cl, pH 8.5, Ͻ 100 mM Na 2 CO 3 , pH 11.5. Carbonate extraction (Na 2 CO 3 ) strips eEF1A from the vacuolar membrane but has no effect on the integral membrane protein, Nyv1p. B, right panel, Rdi1p, which strips Rho proteins from membranes, reduces vacuolar eEF1A levels. 10 g of purified vacuoles were incubated with 10 M GST or GST-Rdi1p for 30 min at 30°C, reisolated, and analyzed by immunoblot. C, vacuoles were isolated from strains expressing eEF1A-GFP, eEF1B␥1-GFP, eEF1B␥2-GFP, and the known vacuolar protein Vma1p-GFP. The lipophilic dye FM4-64 was added as a general membrane stain, and then samples were imaged by fluorescence microscopy. Bar, 1 m.

eEF1A Associates with Vacuolar Rho1p
eEF1A is an abundant cytosolic protein that raises the concern that eEF1A found in vacuolar fractions could be due to cytosolic contamination. Purified vacuoles from the eEF1A-GFP strain were washed under increasingly stringent ionic conditions to test whether eEF1A was specifically associated with vacuolar membranes. The level of membrane-associated eEF1A-GFP was minimally affected by high salt or mild alkaline washes (200 mM KCl and 50 mM Tris-HCl, pH 8.5, respectively), whereas high pH washes (50 mM Na 2 CO 3 , pH 11.5) stripped eEF1A from the vacuolar membranes (Fig. 4B, left panel). This suggests that eEF1A stably associates with the peripheral vacuolar membrane.
Rdi1p (RhoGDI) extracts Rho proteins from membranes (11). Vacuoles incubated with Rdi1p had significantly reduced the levels of eEF1A-GFP when compared with the vacuole markers, Nyv1p and Vac8p (Fig. 4B, right panel). This suggests that Rho1p is involved in vacuolar eEF1A recruitment. However, a pool of eEF1A remained associated with vacuoles, while the level of Rho1p extracted by Rdi1p was nearly complete (Fig.  4B, right bottom panel). This suggests that although EF1A vacuolar recruitment may be Rho1p-dependent, membrane binding may be Rho1p-independent.
We next examined vacuolar association of eEF1A-GFP by microscopy. Vacuoles were isolated from strains expressing GFP-tagged eEF1A, eEF3, and eEF1B␥ and the vacuolar H ϩ -ATPase subunit Vma1p. The lipophilic dye FM4-64 was added as a general membrane stain prior to imaging by fluorescence microscopy. Both Vma1p-GFP and eEF1A-GFP were found to localize to the vacuolar membrane, whereas eEF3-GFP and eEF1B␥-GFP were not (Fig. 4C). Interestingly, eEF1A-GFP appears to be present on the vacuolar membrane in a single discrete puncta. We examined numerous individual vacuoles (n Ͼ50) and observed that 64% contained one eEF1A puncta, although the rest had none ( Table 2). Undetected puncta could be a result of not residing within the focal plane.
We next examined the effect of the drugs narciclasine and latrunculin B (Lat B) to determine the role of actin in vacuolar localization of eEF1A-GFP. Narciclasine is a small drug-like molecule that binds eEF1A and inhibits its actin bundling activity (39). Lat B binds actin monomers, triggering F-actin depolymerization (40). In the absence of drugs, eEF1A-GFP formed a single puncta on most vacuoles (Fig. 5A, upper panels). When vacuoles were exposed to 0.5 M narciclasine, the number of eEF1A-GFP puncta per vacuole increased (Table 2), and vacuoles were deformed from the normal large round morphology (Fig. 5A, middle panels). eEF1A-GFP was observed on vacuoles exposed to Lat B; however, its localization was more dispersed as opposed to discrete puncta (Fig. 5A, lower panels). Therefore, eEF1A-GFP vacuole localization is not dependent on actin binding activity, because there was no decrease in vacuolar association in the presence of narciclasine or Lat B. However, its submembranous distribution was disrupted and therefore actin-dependent.
The effect of actin and eEF1A ligands on vacuolar protein stability and membrane integrity was examined because distinct changes in vacuolar morphology were observed after treatment with these drugs. Isolated vacuoles were incubated for 30 min at 30°C with narciclasine and Lat B and then re-isolated by centrifugation. Narciclasine reduced the levels of the vacuolar luminal peptidase CpY and peripherally associated actin (Fig. 5B, Narc.). We presume that this reduction was due to narciclasine affecting actin-dependent membrane integrity because levels of the vacuolar outer membrane protein, Vac8p, were not affected. Incubation with Lat B also reduced actin levels (Fig. 5B, Lat B). The addition of recombinant eEF1A reversed the effects that narciclasine had on vacuolar integrity (CpY and actin levels were both increased), but it did not reverse the effects of Lat B (Fig. 5B, left panel). This suggests that eEF1A may coordinate actin to increase vacuolar membrane stability.

JOURNAL OF BIOLOGICAL CHEMISTRY 4711
Analysis of eEF1A Subdomains for Rho1p Interaction-eEF1A is composed of three distinct domains. Domain 1 (amino acids 1-219) contains the GTP-binding domain; domain 2 (amino acids 220 -319) contains the aminoacyl tRNA-binding domain; and domain 3 (amino acids 320 -436) contains the actin-binding domain (Fig. 6A) (41). The crystal structure has been solved in complex with its GEF eEF1B␣, where eEF1B␣ interacts with both domain 1 and domain 2 (42). To determine which eEF1A subdomain(s) binds to Rho1p, 7ϫHA-tagged eEF1A full length (FL) and subdomains were cloned and expressed in yeast because E. coli-expressed subdomains were insoluble. All constructs were well expressed in yeast (Fig. 6B, right panel). eEF1A FL and eEF1A Dm3 showed significant association with vacuoles isolated from these strains (Fig. 6B, left panel). eEF1A FL and eEF1A Dm3 also associated with Rho1p in a GST-Rho1p pulldown assay (Fig. 6C). Similar results were obtained when using GFP-tagged constructs indicating that the tag does not influence association. 4 These data suggest that eEF1A domain 3 is sufficient for vacuolar localization and contains the Rho1pbinding region. This was unexpected because eEF1B␣, which inhibits Rho1p binding, interacts with domains 1 and 2. Therefore, eEF1B␣ likely acts as an allosteric inhibitor of Rho1p-eEF1A association via induction of a conformational change. Such conformational changes are known to occur when GTPase bind GTP (43). This is likely because GEFs are known to alter GTPase conformation to facilitate nucleotide exchange.
We next examined whether GFP-eEF1A subdomains showed any distinct localization. GFP-eEF1A FL was localized throughout the cell, similar to GFP alone; this was also observed for GFP-eEF1A Dm1 and Dm2 constructs (Fig. 7B). However, GFP-eEF1A Dm3 formed a distinct single puncta that consistently localized near vacuoles but did not overlap with the staining pattern of the vacuolar dye FM4-64 (Fig. 7B, lower panels). Membrane isolation and display by density gradient centrifugation showed that GFP-eEF1A Dm3 localized to membranes with similar density as vacuoles (Fig. 7C). This supports the conclusion that domain 3 is sufficient for localization to a subset of vacuolar membranes or perhaps a pre-vacuolar compartment. Full-length eEF1A shows dispersed cellular localization; therefore, other eEF1A domains may regulate membrane localization. Furthermore, domain 3 harbors the Rho interacting domain that supports the findings that eEF1A vacuolar localization is linked to Rho1p interaction (see Fig. 4B). We also observed that expression of GFP-eEF1A Dm3 affected vacuolar morphology. Cells expressing domain 3 had highly fragmented vacuoles that were not observed in cells expressing domain 1 or 2 (Fig. 7B, FM4-64 column). This suggests that eEF1A may play a role in controlling vacuolar fragmentation, which is needed when vacuoles are partitioned into daughter cells (44,45).
Functional Effects of eEF1A on Vacuoles-Vacuoles are partitioned into daughter cells via an inheritance reaction that involves partial vacuolar fragmentation in the mother cell, projection of the vacuolar particles into the daughter cell, and then fusion to reform the vacuole (44). The fusion reaction that reforms yeast vacuoles is a multistep process that involves three stages as follows: priming, docking, and fusion (9). In the last stage, Rho1p-stimulated nucleation of nonbranching actin filaments could facilitate the generation of short range forces that propel vesicles to complete membrane fusion. To determine whether vacuole-associated eEF1A also plays a role in promoting vacuolar membrane fusion, purified MBP-eEF1A or MBP was added to standard fusion reactions. In the absence of cytosol, vacuolar fusion was slightly enhanced by MBP-eEF1A, but not MBP, in a dose-dependent manner (Fig. 8A). This enhancement was reduced when cytosol was included in reactions, presumably because of the presence of eEF1A in the cytosol (Fig.  8B). Inhibition of eEF1A by preincubation with narciclasine negated the eEF1A stimulatory effect (Fig. 8, A and B, ؉1 M  narc). These results suggest that eEF1A has a minimal contribution to vacuolar fusion.
To further explore the role of eEF1A, we also examined its effect on vacuole-mediated actin polymerization. We have previously shown that primed vacuoles stimulate the formation of F-actin in a reaction that is initiated by Cdc42p (26). Vacuoles primed in the presence of narciclasine still maintained normal

eEF1A Associates with Vacuolar Rho1p
levels of initial F-actin nucleating activity; however, polymerization activity was not sustained (Fig. 8, C and D, 1 M). In contrast, vacuoles primed in the presence of eEF1A showed reduced levels of F-actin nucleating activity but attained higher levels of polymerization activity over the entire course of the reaction period (Fig. 8, C and D, 1 M eEF1A). Incubation with equimolar amounts of narciclasine and eEF1A resulted in reduced polymerization in an effect that resembled narciclasine alone. 4 These results suggest that narciclasine targets the actin polymerization activity of eEF1A, particularly filament elongation and not nucleation. This role of eEF1A is not directly involved in vacuolar membrane fusion.

DISCUSSION
We identified two Rho GTPases, Rho1p and Cdc42p, on the vacuolar membrane and are interested in further defining their signaling role at this site (11). Their activation occurs sequentially once vacuolar membrane fusion is initiated, with initial activation of Cdc42p followed by Rho1p (14). An essential actin polymerization step in membrane fusion is associated with Cdc42p activation (12), whereas the downstream effects of Rho1p signaling have not been defined. Here, we show that translation elongation factor 1A (eEF1A) is an interacting partner of Rho1p on yeast vacuoles or pre-vacuolar compartments. This interaction was specific to eEF1A and not other members of the translation elongation complex. Although a direct eEF1A-Rho1p interaction could be detected with recombinant proteins (Fig. 2), the levels of association were quite low compared with that detected with native proteins on isolated vacuoles (Fig. 1). Therefore a lipid environment may be needed to facilitate this interaction.
Our initial hypothesis was that eEF1A was a downstream effector of Rho1p that would stimulate the actin bundling activity of eEF1A upon Rho1p activation. However, for this to occur, Rho1p activation would have to release eEF1A because it is tightly packed within F-actin structures that exclude the binding of other factors (25,41,46). This explains why Rho1p-GDP showed the highest affinity for eEF1A, which we describe as an atypical downstream effector interaction (Fig. 2C). The Rho1p-GDP-eEF1A complex could provide a membrane proximal reserve of eEF1A. This complex need only form through a low affinity interaction because of the high abundance of eEF1A. Rho1p activation triggers a reduction in affinity for eEF1A that is released to bind and organize newly forming actin filaments in proximity to the membrane. We show that inhibition of eEF1A function with narciclasine has little effect on membrane fusion; however, membrane integrity and actin polymerization activity are reduced (Figs. 5 and 8). This suggests that the association with actin-organizing factors is crucial for maintaining membrane integrity.
Our evidence also suggests that an indirect effect of eEF1A is the stabilization of the vacuolar membranes because we showed that the inhibition of eEF1A with narciclasine results in leaky vacuoles (Fig. 5B). This likely results in a block in the membrane fusion process. Interestingly, cells expressing eEF1A domain 3 contain fragmented vacuoles and thus seem to promote the formation of inheritance-competent vacuolar structures (44,45). This would make for an elegant Rho GTPase cross-talk mechanism for vacuolar inheritance. Rho1p activation in the mother cell would stimulate inheritance structure formation facilitated by a block in fusion, whereas Cdc42p acti-  Fig. 6A). Cells were stained with FM4-64 to label vacuoles and analyzed by fluorescence microscopy. GFP-eEF1A FL, Dm1, and Dm2 were distributed throughout the cytosol, whereas GFP-eEF1A Dm3 was concentrated in discrete peri-vacuolar puncta. Bar, 5 m. C, lysates from cells expressing GFP-eEF1A FL or GFP-eEF1A Dm3 were separated into cytosol supernatant (sup) and membrane pellet (pel) fractions by high speed centrifugation (right panel). Membranes are also displayed on sucrose gradients (left panel). Anti-GFP immunoblot shows eEF1A Dm3 predominantly associates with membranes and co-migrates with the vacuolar marker, Vma1p, on sucrose gradients.

eEF1A Associates with Vacuolar Rho1p
FEBRUARY 20, 2015 • VOLUME 290 • NUMBER 8 vation in the daughter cell would stimulate the fusion of vacuoles after inheritance. eEF1A may be an important spatial marker due to its unique sub-localization. In cells, GFP-tagged eEF1A showed no distinct localization; however, expression of a GFP truncation of eEF1A that contains only domain 3 resulted in a single discrete puncta that localized to the vacuole or a pre-vacuolar compartment (Fig. 7B). A single discrete eEF1A domain was also observed on isolated vacuolar membranes. This was dispersed in the absence of F-actin or increased in number in the presence of the eEF1A inhibitor narciclasine (Fig. 5). The formation of vacuolar micro-domains has recently been well documented; their induction was shown to occur in response to nutrient deprivation (47). This work also showed the unique formation of multiple patterns of protein and lipid subdomains on the vacuole. One could envision that under starvation conditions a cellular response to limit protein synthesis involves reduction of eEF1A activity in translation and increased activity in actinmediated processes. This could be part of the response mechanism that helps maintain cell viability until more favorable growth conditions arise.
Although Rho1p interaction studies showed that full-length eEF1A was required for most efficient binding, the subdomain that contains the actin-bundling region also showed Rho1p binding (Fig. 6). This makes sense mechanistically if Rho1p is to interact with eEF1A upon actin filament disassembly, a property that is linked to calcium flux (23). A rise in calcium triggers a reduction in F-actin that could stimulate final membrane fusion via removal of an actin barrier. Indeed, eEF1A actin bundling activity is regulated by calcium/calmodulin levels (23,46). Calcium, which is transiently released from vacuoles to stimulate membrane fusion (48,49), may act on eEF1A releasing it from F-actin to be subsequently captured by Rho1p. But calcium is transiently released, and its removal would once again stimulate the actin coordination by eEF1A, thus stabilizing membranes and halting fusion. Hence, eEF1A can be viewed as a molecular target of calcium that modulates an actin barrier to allow fusion in conjunction with calcium flux.
Actin organization has been shown to regulate vesicle membrane fusion in many cell types that undergo regulated secretion. Recent examples include the release of immune mediators from immune cells (50,51) and endothelial cells (52), the transport of GLUT4 to the cell surface of adipocytes (53,54), and neuronal synaptic transmission and endocrine release (55,56). Actin has been documented as providing cellular location of membrane docking and fusion via actin anchorage; however, it is unclear whether actin facilitates or prevents fusion as both have been shown (57). The intracellular fusion of lysosomes and phagosomes has been particularly well dissected into multiple actin-regulated stages. The first requires low calcium and actin polymerization to associate the fusing compartments and then high calcium to condense compartments for fusion (58). It was recently shown that transient F-actin polymerization delays lysosome-phagosome fusion (59). The molecular regulators . MBP-eEF1A boosted fusion slightly in the absence of cytosol, which can be inhibited by the eEF1A drug narciclasine (narc). Narciclasine alone showed no effect on fusion. C, vacuole-associated actin polymerization activity was analyzed after vacuoles were incubated with MBP-eEF1A or narciclasine for 30 min. D, actin polymerization activity was calculated from three experiments as shown in C. Treatment with narciclasine resulted in a statistically significant decrease in activity, whereas the eEF1A stimulated activity; however, this was not statistically significant (**, p Ͻ 0.01; NS, p Ͼ 0.05, n ϭ 3). rxn, reaction.

eEF1A Associates with Vacuolar Rho1p
that control actin remodeling in these systems has not been fully defined, so it will be interesting to see whether eEF1A is also involved in these systems.