|Home | About | Journals | Submit | Contact Us | Français|
Membrane fusion depends on conserved components and is responsible for organelle biogenesis and vesicular trafficking. Yeast vacuoles are dynamic structures analogous to mammalian lysosomes. We report here that yeast Env7 is a novel palmitoylated protein kinase ortholog that negatively regulates vacuolar membrane fusion. Microscopic and biochemical studies confirmed the localization of tagged Env7 at the vacuolar membrane and implicated membrane association via the palmitoylation of its N-terminal Cys13 to -15. In vitro kinase assays established Env7 as a protein kinase. Site-directed mutagenesis of the Env7 alanine-proline-glutamic acid (APE) motif Glu269 to alanine results in an unstable kinase-dead allele that is stabilized and redistributed to the detergent-resistant fraction by interruption of the proteasome system in vivo. Palmitoylation-deficient Env7C13-15S is also kinase dead and mislocalizes to the cytoplasm. Microscopy studies established that env7Δ is defective in maintaining fragmented vacuoles during hyperosmotic response and in buds. ENV7 function is not redundant with a similar role of vacuolar membrane kinase Yck3, as the two do not share a substrate, and ENV7 is not a suppressor of yck3Δ. Bayesian phylogenetic analyses strongly support ENV7 as an ortholog of the gene encoding human STK16, a Golgi apparatus protein kinase with undefined function. We propose that Env7 function in fusion/fission dynamics may be conserved within the endomembrane system.
Eukaryotic organelles strike a delicate balance between fusion and fission in time and space. The most elaborate example of that balance is the endomembrane system, a dynamic network comprised of the endoplasmic reticulum (ER), Golgi apparatus, late endosomes, lysosomes, and plasma membrane. This network is interconnected through regulated membrane fusion/fission, including vesicular trafficking and direct homotypic/heterotypic fusion. As such, membrane fusion is central to organelle biogenesis and vesicular trafficking and has been extensively studied (reviewed in references 1–7). The lysosomal vacuole of baker's yeast, Saccharomyces cerevisiae, has served as a productive system in these studies. Yeast vacuoles are functionally analogous to mammalian lysosomes (reviewed in references 8 and 9). Both import hydrolytic enzymes and are responsible for degradation and recycling of cargo, including macromolecules, organelles, and endocytosed receptors and ligands. More recently, both have been implicated in cellular homeostasis and stress survival functions (reviewed in reference 10). Moreover, yeast vacuoles are prominent landmarks that constitute about 25% of the cell volume and undergo fusion/fission in response to stimuli (11). Vacuoles fragment under hyperosmotic stress as an adaptive response to maintain osmotic balance (12–15). Vacuole fusion/fission are also regulated during the cell cycle (for reviews, see references 16 and 17). Early in G1, fusion/fission equilibrium shifts toward fission to maximize proper vacuolar inheritance during budding. As such, vacuoles offer an excellent model for organelle biogenesis and membrane fusion/fission. Through a combination of in vivo vacuolar morphology studies and in vitro homotypic vacuole fusion assays, many components of vacuole fusion have been identified (for the latest reviews, see references 1, 4, 5, 7, and 18). What remains less defined is how vacuolar fusion/fission dynamics may be regulated within the cell cycle or during orchestration of the stress response in cells. The yeast vacuolar protein kinase yeast casein kinase 3 (Yck3) has been shown to inhibit membrane fusion by phosphorylation of membrane fusion components (19–22). How such phosphorylation may be regulated in fine tuning fusion/fission dynamics or whether phosphorylation may be a general strategy for fusion inhibition is not yet clear. The vacuole proteome includes >140 proteins, the biological significance of many of which is unknown (23). Presumably, key fusion/fission regulators remain to be identified.
We recently reported that ENV7 is one of several genes identified in a genome-wide screen for defects at late endosome and vacuole interface (ENV) genes (24). ENV7 deletion leads to internal accumulation of precursor carboxypeptidase Y (pro-CPY), a vacuolar hydrolytic enzyme that transits the rough ER, Golgi apparatus, and late endosome prior to arrival at the vacuole, where it is processed to mature form via PEP4-dependent proteolytic cleavage (25, 26). Here, we report that Env7 is a palmitoylated protein kinase that negatively regulates fusion at the vacuole membrane. Its N-terminal Cys13 to -15 track is essential for Env7 vacuolar membrane anchoring and kinase activity. We also present phylogenetic data supporting ENV7 as a widely conserved ortholog of the serine/threonine kinase 16 (STK16) gene and discuss a possible role for STK16-related kinases in regulation of the endomembrane architecture.
The yeast strains used in this study are listed in Table 1. BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 MET15 ura3Δ0) and the yeast deletion strain library were gifts from Greg Payne (University of California, Los Angeles [UCLA]). Proteasome-deficient (pre1-1 pre2-1) yeast strains were gifts from Dieter H. Wolf (University of Stuttgart, Stuttgart, Germany). The SEC7-6×dsRed (I33) TRP yeast strain was generously provided by Daniel Klionsky (University of Michigan). Cells were routinely grown in rich medium (yeast extract-peptone-dextrose [YPD]; Difco Chemicals, St. Louis, MO; 1% yeast extract, 2% peptone, and 2% glucose) or synthetic minimal dropout medium (SMD) (Difco Chemicals, St. Louis, MO; 0.67% yeast nitrogen base [YNB], 2% glucose, and specific amino acids) at 30°C unless otherwise stated. For galactose induction of green fluorescent protein (GFP) fusion reporters, cells were grown in synthetic minimal medium without uracil (SM-URA) and then shifted to SM-URA containing 0.2% galactose, 1% glycerol, and 1% ethanol. The ENV7-GFP strain was purchased from Invitrogen and confirmed by colony PCR.
All restriction enzymes and Taq DNA polymerase were purchased from New England BioLabs, Inc. (Beverly, MA). Phusion DNA polymerase was purchased from Invitrogen. Oligonucleotides were ordered from Operon (Alameda, CA). MG132 was supplied by EMD Chemicals. [γ-32P]ATP was purchased from MP Biomedicals (Solon, OH). Miniprep DNA kits, kits for gel extraction of DNA, and yeast transformation kits were supplied by Zymo Research (Irvine, CA). Antihemagglutinin (anti-HA), anti-hexokinase I, and anti-rabbit IgG–horseradish peroxidase (HRP)-linked antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-alkaline phosphatase (anti-ALP) and anti-CPY antibodies were purchased from Mitosciences (Eugene, OR). HRP-conjugated anti-mouse goat monoclonal antibody was from Invitrogen. Anti-Vps41 antibody was a gift from Christian Ungermann (University of Osnabrück, Osnabrück, Germany). Chemiluminescence reagents were purchased from Pierce, and Film Biomax Light 13X18 PK50 was supplied by VWR. All other chemical reagents were from Sigma-Aldrich (St. Louis, MO).
Yeast expression plasmids (listed in Table 2) were constructed using established methods of PCR-directed homologous recombination (32). Briefly, tagged versions of the ENV7 gene were PCR amplified from a genomic DNA using forward and reverse primers containing sequences (~45 bp) that are homologous to each end of the target vector to be cloned. The primers were ordered from Operon (Huntsville, AL) and are listed in Table 3. Cells were cotransformed with a linearized vector (containing a selectable marker) and a PCR-generated insert using a yeast transformation kit (Zymo Research, Irvine, CA), and the transformants were selected by growing on SM-URA agar plates. The constructed plasmids were isolated from yeast and transformed into Escherichia coli. Plasmids with correct inserts were transformed into yeast with the desired background for functional analyses. HA- and GFP-tagged vectors were obtained from Walter Schmidt and used for construction of plasmids for expression of ENV7-HA and GFP-tagged ENV7, respectively.
6×His-tagged Env7 was PCR generated using a pair of primers listed in Table 3, transformed into env7Δ yeast, and constitutively expressed under the phosphoglycerate kinase (PGK) promoter (PPGK), and the recombinant Env7-His was purified from BY4742 env7Δ according to the company protocol (Novagen). Briefly, yeast cells transformed with the 2μm plasmid (overexpressing C-terminally 6×His-tagged Env7) were grown to mid-log phase (optical density at 600 nm [OD600] = 1.0), harvested, and spheroplasted by Zymolyase treatment. The spheroplasts were broken open by bead beating with lysis buffer containing 1% Triton X-100 (TX-100) and subjected to high-speed centrifugation (28,000 × g for 1 h) using an Eppendorf centrifuge. The soluble fraction was resuspended with nitrilotriacetic acid (NTA)-Sepharose beads and eluted after several washes. The eluted protein was directly used for kinase assay.
A PCR product of C-terminally 6×His-tagged ENV7 was digested with NcoI and HindIII restriction enzymes and inserted into a pET24d(+) expression vector digested with the same enzymes, expressed in E. coli BL21(DE3)(pLysS) with 1 mM isopropyl-β-d-thiogalactoside (IPTG), and purified using Ni-Sepharose column chromatography according to the standard protocol as described by the company (Novagen).
PCR-based cloning was also carried out to insert DNA sequences of ENV7 into the 2μm overexpression vector pRS426. The ENV7 open reading frame (ORF) was amplified from BY4742 genomic DNA using custom-made forward and reverse primers (Operon, Huntsville, AL) that introduced restriction sites (underlined in Table 3).
Single or multiple base substitutions in the target gene were introduced by using two PCR-based site-directed mutagenesis approaches as described previously (33). All plasmid constructs and mutageneses were confirmed by DNA sequencing (Macrogen, South Korea).
For localization studies of endogenously expressed GFP-tagged Env7, yeast cells purchased from the Yeast GFP Clone Collection (Invitrogen) were grown in YPD to mid-log phase and analyzed by differential interference contrast (DIC) optics and confocal microscopy. An inducible GFP-Env7 or Env7-GFP reporter was constructed as described previously (34). Mid-log-phase yeast cells containing the appropriate reporter were harvested, washed twice with sterile H2O, and incubated in SM-Ura containing 0.2% galactose for approximately 5 h at 30°C to induce expression of the GFP reporters. The expression pattern of GFP reporters was visualized using a confocal microscope as described previously (24). The galactose-induced cells were viewed with an Olympus Fluoview 1000 confocal laser scanning system mounted on an inverted microscope (Olympus IX-81) and a 100× oil immersion UPLSAPO objective (numeric aperture [NA], 1.4; working distance [WD], 0.12 mm). The Argon ion (488-nm) and blue/red diode (405-nm/635-nm) lasers were used for image capturing. Images were equally enlarged to ×4,000 and were analyzed and processed with Photoshop CS5. For each experiment, at least 150 to 200 cells from five visual fields were scored, from which representative images were selected. Images for colocalization studies were created by overlapping the DIC and fluorescence images.
For vacuole morphology studies, log-phase cells were stained with the vital dye FM4-64 [N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl)pyridiumdibromide] (Invitrogen) as described previously (24). For hyperosmolarity studies, stained cells were subjected to hyperosmotic shock (0.4 M NaCl) for 15 and 60 min and analyzed microscopically as described above. One hundred to 150 cells were blind scored from representative microscopic fields for statistical analysis of vacuolar morphology (prominent versus multilobed vacuoles). Similarly, stained cells were also scored for bud vacuolar morphology (prominent versus multilobed vacuoles) under normal growth conditions. For vacuole acidification studies, log-phase cells were stained with Quinacrine (Invitrogen) for 5 min and analyzed microscopically as described previously (35).
Subcellular fractionation of yeast cells was performed as described previously (36) with the following modifications. The homogenate (H) was centrifuged twice at 750 × g for 10 min to remove unbroken cells, cell debris, or nuclei. The clarified supernatant (termed the postnuclear fraction [PNF]), was further centrifuged at 18,000 × g for 50 min to obtain a membrane-rich pellet (P18) and clear cytosol (S18). The membranous fraction was then resuspended in 0.1 M sodium carbonate (pH 11.5), incubated on ice for 30 min, and centrifuged at 18,000 × g for 30 min. The protein concentration was determined by the Bradford method as described below and adjusted to 0.8 mg/ml. For subcellular localization studies of the triple-cysteine (C13-15S) mutant Env7-HA, yeast cells were lysed and subjected to differential centrifugation to yield S0.4, P13, P100, and S100 fractions as described previously (20).
Yeast vacuoles were isolated as described by Haas and Wickner (37). Interface fractions were trichloroacetic acid (TCA) precipitated, and the pellets were washed 3 times with cold acetone and resuspended in 6× SDS-PAGE sample buffer. Solubilized fractions were separated on 12% SDS-PAGE, transferred to a nitrocellulose membrane, and analyzed by Western blotting.
To confirm in vivo functionality of N- and C-terminally GFP-tagged, HA-tagged, and overexpressed untagged Env7, the constructed plasmids were introduced into wild-type and env7Δ yeast strains. Transformed cells were selected on SM-URA and were subjected to patch immunodetection as previously described (24).
For hydrophobicity/hydrophilicity analysis, the Env7 sequence was copied from the Saccharomyces Genome Database (SGD) and pasted into the analysis window hosted by the Genome Consortium for Active Teaching (GCAT) at Davidson College (http://gcat.davidson.edu/rakarnik/kyte-doolittle.htm), and a Kyte-Doolittle analysis (38) was performed on the entire sequence using an amino acid window size of 19.
For multiple-sequence alignment (MSA), the ENV7 nucleotide sequence query was used in a tBLASTx search to determine putative homologs (39). Default search settings were utilized, and entire sequences with the lowest E value (an upper limit of 1e−10) from each species were chosen for further analysis. Thirty-one nucleotide sequences (including ENV7) were aligned using the MAFFT multiple-sequence translation alignment program (40). The E-INS-I alignment algorithm and a BLOSUM80 scoring matrix were employed, and a gap open penalty of 2 was used with no offset value. The program output alignment was screened and edited manually.
For phylogenetic analyses, the alignment was submitted to MrBayes, with the Zea mays sequence as the outgroup (41). The substitution model utilized was general time reversible (GTR) with a gamma rate variation and four gamma categories. One million iterations were generated with a burn-in length of 250,000. Branch lengths were unconstrained. The results were visualized with Geneious and Dendroscope (42, 43). The tree topology was assessed for congruence with a maximum-likelihood tree constructed from ATP synthase homologs using the online program Icong (44, 45).
The palmitoylation assay was carried out according to a biotinylation assay method, as described previously (46) with modifications. Briefly, P18 membranes were resuspended in ABE buffer (50 mM Tris-HCl, pH 7.4, containing 1 mM EDTA, 1% Triton X-100, and protease inhibitors) and immunoprecipitated overnight using 10 μl anti-HA rabbit monoclonal antibodies conjugated to Sepharose beads (Cell Signaling Technology, Danvers, MA). The beads were washed with ABE buffer containing 0.1% Triton X-100 and incubated with 50 mM N-ethylmaleimide (NEM) in ABE buffer for 1.5 h. They were washed and incubated with either 1 M Tris, pH 7.5 (control), or 1 M hydroxylamine, pH 7.5, for 2.5 h at room temperature. The beads were then incubated overnight with 300 μM biotin and resuspended in sample buffer. An aliquot was separated by SDS-PAGE and analyzed by Western blotting using streptavidin HRP (for biotinylation) and anti-HA antibody (for protein recovery).
For kinase assays of Env7-HA, the whole membranous fractions (P18) were used. The membranes were solubilized by resuspending them in lysis buffer containing 1% Triton X-100 and 500 mM NaCl, incubated on ice for 30 min, and centrifuged at 18,000 × g for 30 min to obtain solubilized Env7-HA, which was then immunoprecipitated overnight at 4°C as described above. The Sepharose beads were washed 3 times in lysis buffer (containing 1% Triton X-100, 500 mM NaCl), once in 1 M NaCl, and 3 times in kinase buffer (0.05 M Tris-HCl, pH 7.4, containing 10 mM MgCl2 and 2 mM dithiothreitol [DTT]) and used in kinase assays. Kinase activity was assayed in a total reaction volume of 20 μl that contained 10 μl of immunoprecipitated beads (Env7-HA), 200 μM unlabeled ATP, 1 mg/ml bovine serum albumin (BSA), and 10 μCi [γ-32P]ATP with or without the addition of 2.5 μg exogenous substrate. For some experiments, the detergent-solubilized residual membranes were also used for kinase assay. The kinase reaction was carried out by incubating the reaction mixture at 30°C for 30 min and stopped by the addition of 5 μl of 6× SDS-PAGE sample buffer. The solubilized proteins were heated at 100°C for 10 min and resolved by 12% SDS-PAGE. The gels were fixed (10% acetic acid, 40% methanol) and stained with Coomassie brilliant blue R250 (CBB), and the dried gels were subjected to autoradiography BioMax film with maximum resolution (Kodak, Rochester, NY).
The Vps41 phosphorylation assay was based on upshift of phosphorylated Vps41 migration in SDS-PAGE and was performed using the P13 membrane fraction and anti-vps41 antibody as described previously (20, 22, 47).
Western blots were carried out using a semidry method with appropriate primary and secondary antibodies, as previously described (24).
Protein concentrations were determined by the Bradford method (48) using the Quick Start Bradford protein assay kit (Bio-Rad, Hercules, CA). Protein concentrations were calculated from a linear standard curve of BSA (plotted from a graph of absorbance versus protein concentrations ranging between 0 and 1 mg/ml) using Microsoft Excel.
For vacuolar morphology studies, 150 to 200 cells were blind scored from random fields in at least two separate experiments, and their mean values and standard deviations were calculated using standard statistical tools (Excel). P values were calculated using a t test. P values of <0.05 were considered statistically significant.
For further understanding of the role of ENV7 in vacuolar events, we probed the cellular localization of its product. In a systematic study, GFP-tagged Ypl236c (Env7) has been reported to be localized to vacuoles (29). We confirmed the localization of endogenously expressed Env7-GFP to the vacuole membrane in the same strain (Fig. 1A). We also constructed inducible plasmids expressing N- or C-terminally GFP-tagged Env7 and assessed their localization in the env7Δ strain (Fig. 1B to toE).E). Ectopically expressed GFP-tagged Env7 localizes to vacuolar membranes independently of the tag orientation. Over 80% of scored cells showed vacuolar localization of the fusion protein; the majority of the cells also had fusion protein localization in punctate structures suggestive of the Golgi apparatus. Coexpression of GFP-tagged Env7 with the Golgi apparatus marker Sec7-dsRed confirmed the Golgi apparatus localization of overexpressed Env7 (Fig. 1F). Off-target localization was directly proportional to the induction time and inducer concentration (data not shown).
In order to biochemically explore Env7, we constructed a 2μm vector expressing HA-tagged ENV7 from a constitutive promoter (Env7-3×HA) (Fig. 2A), introduced it into wild-type and env7Δ strains, and analyzed the expression product by Western blotting using anti-HA antibody. The predicted native Env7 is a 364-amino-acid (aa) protein, and the HA-tagged recombinant Env7 was detected as a stable 45-kDa protein, which corresponds to the predicted size of the 407-aa tagged protein (Fig. 2B).
We next examined whether the HA-tagged Env7 is a membrane-associated protein. PNF, cytosolic (S18), and membrane (P18) fractions were subjected to Western blotting (Fig. 2C). HA-tagged Env7 was detected in all fractions except S18, suggesting its exclusive membrane localization. To confirm vacuolar localization, we isolated vacuoles from yeast expressing Env7-HA using Ficoll gradient flotation and analyzed fractions by Western blotting using monoclonal antibodies to HA and CPY, a vacuole-resident protease. Env7-HA was exclusively localized to a vacuolar-marker-enriched fraction (Fig. 2D). Thus, both microscopic and biochemical approaches confirmed the localization of tagged Env7 to the vacuolar membrane.
It was necessary to confirm the functionality of the tagged Env7 proteins prior to further studies. We have previously reported that deletion of ENV7 results in internal accumulation of precursor CPY, while reintroduction of the gene complements this phenotype, as assessed by patch immunoblots using a monoclonal antibody specific for pro-CPY (24). We used the same approach to confirm that C- and N-terminally tagged Env7 (Fig. 3A and andB),B), HA-tagged Env7 (Fig. 3C), and Env7 expressed from a 2μm plasmid (Fig. 3D) are functional.
Conserved regions of protein sequences often signify functional importance. Following a tBLASTx query using ENV7 nucleotide sequence, 30 nucleotide sequences with the lowest E values were aligned using the MAFFT program; a representative subset of entries is presented in Fig. 4. ENV7 putative homologs were not present in prokaryotes. The alignment revealed two highly conserved motifs. Env7 contains a consensus sequence for S-palmitoylation in an N-terminal string of three cysteine residues, Cys13 to -15. Palmitoylation is a common lipidation that increases the affinity of a soluble protein for the cytoplasmic faces of membranes and affects protein localization and function (49). Since Env7 does not exhibit putative signal sequence or transmembrane domains when subjected to Kyte-Doolittle analysis (data not shown), palmitoylation could confer its vacuolar membrane association. This hypothesis is in agreement with a systematic study that identified Ypl236c/Env7 among palmitoylated yeast proteins (50).
Env7 also contains an extensive eukaryotic kinase (ePK) domain that is most consistent with STK16-related kinases, a divergent kinase family (51). The kinase domain contains the second conserved motif, the alanine-proline-glutamic acid (APE) motif (52). The motif is known to stabilize the N lobe of many protein kinases, with the glutamic acid residue the most conserved element (52). Env7 contains a proline residue in lieu of alanine within the motif. Variations at the APE motif alanine are present in a number of putative STK16-related kinases, while the glutamic acid of the motif is invariant.
These bioinformatics results informed our investigations into Env7 membrane association and molecular function.
The palmitoylation consensus sequence and absence of putative transmembrane domains prompted us to further investigate Env7 membrane association. A P18 membrane fraction enriched in Env7-HA was extracted with various reagents known to disrupt membrane association, including high-molar salt (1 M NaCl), high pH (sodium carbonate, pH 11.5), and detergent (1% Triton X-100). Samples were separated into soluble (S100) and membranous (P100) fractions and analyzed by Western blotting as before. Env7-HA was almost completely extracted with 1% TX-100 but remained associated with membranes when subjected to high-salt and high-pH treatments, consistent with membrane association via lipidation (Fig. 5A). In order to determine if Env7 is anchored to the membrane by palmitoylation, we treated the membranes with 1 M hydroxylamine, pH 7.4, which selectively removes thioester-linked palmitates (53), prior to processing for Western blots (Fig. 5B). Hydroxylamine treatment consistently solubilized >30% of membrane-associated Env7-HA as quantitated by densitometry of low-exposure autoradiographs. In the absence of predicted transmembrane domains, this implies lipid anchoring via palmitoylation is a major mechanism for Env7 membrane association. As more extensive treatments with hydroxylamine did not improve membrane dissociation, partial dissociation of Env7-HA may be attributed to poor access of hydroxylamine to buried thioester bonds. In order to confirm that Env7 is palmitoylated, we performed an established assay for palmitoylation that utilizes a maleimide derivative (biotin-BMCC) for labeling of palmitoylated cysteines (54). In repeated experiments, biotinylated Env7-HA was detectable following hydroxylamine treatment, confirming that Env7 is a palmitoylated protein (Fig. 5C, left). We then examined whether the triple-cysteine palmitoylation consensus sequence is a site for palmitoylation. We performed site-directed mutagenesis of these cysteine residues to serine (C13-15S) and analyzed the palmitoylation of the mutant protein. Env7C13-15S was defective in palmitoylation, implicating one or more of the three cysteines in the lipid modification (Fig. 5C, right). Upon subcellular fractionation, the mutant protein was found exclusively localized to the soluble fraction, indicating that Cys13 to -15 are required for Env7 membrane association (Fig. 5D). Vacuolar membrane alkaline phosphatase and cytosolic hexokinase I were used as fraction markers. A degradation product of ALP was detectable in the cytosolic fraction, as reported by others (55, 56). A GFP-tagged triple-cysteine mutant of Env7, GFP-Env7C13-15S, mislocalizes to the cytoplasm (Fig. 5E). Env7C13-15S failed to complement the p2CPY accumulation phenotype of env7Δ, as assessed by patch immunoblot assays (Fig. 3E), confirming that lipidation is required, not only for Env7 membrane association, but also for its cellular function.
Multiple-sequence alignments of Env7 indicate an extensively conserved protein kinase domain. Since we had established expression of full-length, stable, and functional Env7-HA (Fig. 2 and and3C),3C), we took advantage of the tagged protein to directly assay kinase activity in vitro. P18 membranes were solubilized with 1% Triton X-100, immunoprecipitated with anti-HA Sepharose beads, and then used in kinase assays; extracted Env7-HA was subjected to Western blotting and autoradiography. Our results show that immunoprecipitated Env7-HA is autophosphorylated in vitro (Fig. 6A). Similarly, we constructed a yeast vector expressing His-tagged Env7, and purified Env7-His was directly subjected to kinase assays (Fig. 6B). Autoradiography confirmed the predicted ~42-kDa phosphorylated band. Thus, HA- and His-tagged Env7 proteins are autophosphorylated in vitro.
In order to examine if Env7 is capable of phosphorylating exogenous substrates, we screened commonly used substrates, including histones (H1, H2B, and H3), myelin basic protein (MBP), and β-casein. Env7-HA mediated both autophosphorylation and substrate phosphorylation (Fig. 6C). Kinases require divalent cations, which make a complex with ATP before transfer of the terminal phosphate to the substrate. With most typical protein kinases (TPKs), Mg2+ and Mn2+ can be used interchangeably for this requirement. In order to determine the ion specificity of Env7, we performed kinase assays in the presence of 10 mM MnCl2 or 10 mM MgCl2. Env7 autophosphorylation and substrate phosphorylation were severalfold higher in the presence of Mg2+ than in the presence of Mn2+, indicating that Env7 is an Mg2+-dependent kinase (Fig. 6D).
We also constructed the plasmid pET24d(+)-Env7-HIS for bacterial expression, and purified Env7-His was directly assayed for kinase activity (Fig. 6E). The recombinant protein exhibited both autophosphorylation and substrate phosphorylation, indicating that kinase activity is inherent to Env7.
Our findings show that Env7 is a membrane-associated palmitoylated protein kinase. In order to explore whether palmitoylation of the protein is required for its kinase activity, we assessed the in vitro kinase activity of the triple-cysteine mutant as described for Fig. 7. Since the mutant does not localize to vacuoles, both crude membrane (P18) and soluble (S18) fractions were assessed. Env7C13-15S was defective in phosphorylation in both fractions, suggesting that the cysteine track is essential for Env7 kinase activity (Fig. 7A).
To determine whether E269 of the Env7 APE motif is critical for its kinase activity, we generated a series of Env7E269-HA (E269A, E269D, and E269S) mutants. Similarly, we also mutagenized lysine residues 69 and 70, which are well conserved within the ATP binding domain and are often mutagenized for the generation of kinase-dead TPK alleles. P18 membranes were prepared from strains expressing wild-type or mutant Env7-HA and analyzed for kinase activity using histone 1 (H1) as an exogenous substrate. A combination of autoradiography and Western blot analysis showed that while the steady-state level of the E269A mutant Env7-HA was not detectable, the remaining kinase domain mutants of Env7 were stable and kinase active (Fig. 7B).
In order to exclude any possibility of ORF errors during PCR generation of the mutant, the entire env7E269-HA insert was sequenced and did not reveal additional base changes (data not shown). We predicted that the mutant protein might be degraded by a proteasomal system. Mutant and misfolded proteins may be cleared via this protein quality control system (57). In order to test this hypothesis, we undertook a two-pronged approach of chemical inhibition and genetic intervention of the proteasomal system. In the first approach, we utilized an erg6Δ mutant of the BY4742 strain for its membrane permeability to drugs. We transformed the erg6Δ strain with wild-type or mutant ENV7-HA. Transformants were treated with either dimethyl sulfoxide (DMSO) (control) or MG-132 (50 μM), a potent proteasomal inhibitor, and assayed for the steady-state levels of HA-tagged protein by Western blotting (Fig. 8A). As assessed by ImageJ densitometry of repeated experiments, while mutant protein levels were <1% of wild-type levels in a BY4742 background and <10% of wild-type levels in DMSO-treated cells, Env7E269A-HA levels were stabilized to wild-type levels in samples treated with the proteasome inhibitor. To further implicate the proteasomal system, we also employed a genetic approach. We deleted ENV7 in a pre1-1 pre2-1 double-mutant background with deficient chymotrypsin-like proteasomal activity in vitro (58, 59), transformed the strain with a wild-type or mutant Env7-HA-encoding plasmid, and analyzed the steady-state levels of HA-tagged proteins as before (Fig. 8B). In repeated experiments, Env7E269A-HA levels were stabilized to at least 30% of wild-type levels. Proteasomal and vacuolar degradation are the two main cellular protein degradation mechanisms. In order to exclude the possibility of vacuolar degradation, we assessed the stability of wild-type or E269A mutant Env7-HA in a pep4Δ strain defective in vacuolar protease function (25) and saw no effect on Env7E269A stabilization (Fig. 8C). We next asked whether the partially stabilized mutant protein had kinase activity. Due to the altered enrichment of Env7E269A-HA in a detergent-resistant fraction (Fig. 8D), we used whole residual membranes for kinase assays. The mutant protein was defective in both autophosphorylation and substrate phosphorylation, as assessed by autoradiography and Western blotting (Fig. 8E). Since Env7E269A levels were consistently ~1/3 of the wild-type protein levels, we were concerned that the lower levels of mutant protein might have retained kinase activity beyond the detection resolution of our approach. We therefore serially diluted wild-type Env7 and analyzed samples alongside Env7E269A in kinase assays for confirmation of the kinase-dead allele (Fig. 8F). Similar results were observed in the proteasome-defective pre1-1 pre2-1 background (Fig. 8G). These findings suggest that kinase-dead env7E269A is misfolded, mislocalized, and cleared by the proteasomal system.
The only other known palmitoylated protein kinase localized to the yeast vacuolar membrane is Yck3 (20, 22, 60). Yck3 has been reported to inhibit fusion of vacuoles during budding and sustained hyperosmotic stress (22). Additionally, several palmitoylated vacuolar membrane proteins have been implicated in vacuolar biogenesis, including the SNARE Ykt6 and the fusion factor Vac8 (61, 62). We speculated that Env7 may also have a role in vacuole fusion/fission dynamics. In order to test our hypothesis, we examined the vacuolar fusion/fission equilibria of wild-type and env7Δ strains under various conditions and included a yck3Δ strain in our studies. Our original studies of env7Δ had not revealed any vacuolar morphology defects under normal growth conditions (24).
To assess vacuolar fusion/fission dynamics under hyperosmotic stress, wild-type, env7Δ, and yck3Δ strains were stained with FM4-64 for 45 min and then treated with either YPD (control) or YPD plus 0.4 M NaCl. Samples were microscopically analyzed for prominent versus multilobed (>5) vacuolar morphology at 15 and 60 min following initial high-salt stress. Representative DIC and confocal microscopy images are presented in Fig. 9A, and the results of scored cells from two independent experiments are summarized in Fig. 9B. Our results confirm that env7Δ vacuolar fusion/fission dynamics is equivalent to that of the wild type in the absence of high salt. As all three strains exhibited statistically equivalent increases in vacuole fragmentation within l5 min of high-salt treatment, initial sensing of and response to hyperosmotic stress are not defective in the env7Δ strain. However, after sustained hyperosmotic stress for 60 min, while wild-type samples maintained vacuolar fragmentation, env7Δ vacuoles did not. The percentage of multilobed vacuoles for the env7Δ strain was equivalent to that of the yck3Δ strain, which has an established defect in maintaining vacuolar fragmentation (22) (P < 0.0001; confidence interval, 95%). Several vacuolar membrane proteins affect proper acidification of the vacuole, which in turn affects vacuolar biogenesis and function. This includes regulation of phosphatidylinositol (3,5)P2 by the Fab1-Fig4-Vac14 complex during hyperosmotic response (63, 64) and vacuole biogenesis (65). We used quinacrine, an acidotropic vital dye, to assess vacuolar acidification in the env7Δ mutant. Quinacrine crosses membranes at nearly neutral pH and is protonated and retained exclusively in acidic compartments (66). Both the wild-type and env7Δ strains exhibited strong quinacrine fluorescence in vacuoles while the vac14Δ mutant was poorly stained, as expected (Fig. 9C). Thus, ENV7 is not required for proper acidification of vacuoles, suggesting that its fusion inhibition function is independent of the acidification state of the vacuole.
The above-mentioned findings suggest that ENV7 has a negative regulatory role in vacuolar fusion. If this hypothesis is true, overexpression of ENV7 should increase the percentage of multilobed vacuoles under normal growth conditions. We transformed wild-type cells with either an empty multicopy 2μm plasmid (pRS416) or one encoding ENV7 from its endogenous regulatory sequences (pSMG470) and scored vacuole morphology as described above (Fig. 9D). ENV7 functional expression from the 2μm plasmid was confirmed by its ability to complement pro-CPY accumulation in the env7Δ mutant (Fig. 3D). As expected, overexpression of ENV7 results in almost a 2-fold increase in the percentage of multilobed vacuoles under normal growth conditions, as has been reported for YCK3 overexpression (22). Increased vacuole fragmentation was also seen following excessive overexpression of GFP-tagged Env7 in the env7Δ background (data not shown). Interestingly, overexpression of ENV7 did not result in a statistically significant difference in fusion/fission dynamics during short-term or sustained osmoresponse. This suggests either a saturable inhibitory mechanism for Env7 or a posttranslational regulatory mechanism acting on Env7 during hyperosmotic response.
To explore a role for Env7 in vacuole biogenesis beyond hyperosmotic response, we analyzed the bud vacuolar morphology (Fig. 9E and andF).F). Results from multiple independent experiments were scored in a blind fashion and indicated that the wild-type strain has a mixed population of prominent and multilobed vacuoles in buds, whereas the env7Δ and yck3Δ strains have nearly 100% prominent bud vacuoles. These results indicate that ENV7 and YCK3 negatively regulate vacuole fusion in buds and are consistent with the previously reported role of YCK3 in vacuolar inheritance (22).
The parallels between Env7 and Yck3 cellular roles, along with the fact that both are palmitoylated kinases at the vacuolar membrane, prompted us to explore whether ENV7 and YCK3 encode redundant functions. We overexpressed ENV7 in a yck3Δ background as described above and analyzed the vacuolar morphology. The results from two independent experiments are presented in Fig. 9G and show that ENV7 overexpression does not suppress the vacuolar fragmentation phenotype of yck3Δ. Thus, ENV7 does not encode a redundant function with YCK3, nor can its overexpression bypass YCK3 function. One of the two known cellular substrates for Yck3 is Vps41, which is a component of the HOPS (homotypic fusion and vacuole protein sorting) tethering complex. Vps41 phosphorylation can be detected as a migration upshift in Western blots in the presence of ATP (20, 22, 47). Our results show that Vps41 phosphorylation is indistinguishable from that of the wild type in the absence of Env7, while it is completely blocked in the yck3Δ strain (Fig. 9H). Thus, a combination of genetic and biochemical approaches confirms that the Env7 role in negative regulation of membrane fusion is distinct from that of Yck3.
The highly conserved regions in Env7 prompted us to further explore homologous relationships using an extensive phylogenetics approach. To generate a phylogram, MAFFT MSA of the 31 putative eukaryotic homologs was subjected to Bayesian analysis (Fig. 10). This type of analysis utilizes a priori knowledge and Markov chain Monte Carlo probability sampling methods to obtain the best possible tree topology with the given multiple-sequence alignment (40). The majority of clades on the phylogram are well supported (the thick horizontal lines signify ≥95% clade credibility). Furthermore, the tree topology is consistent with accepted evolutionary relationships exhibited in the Tree of Life Web Project (67, 68). The phylogram is also congruent with a similar tree constructed from highly conserved ATP synthase homologs in the same set of 31 species (P = 1.14e−11) (68). These consistencies support the accuracy of our phylogram and the conservation of ENV7 throughout Eukarya. Our phylogram establishes ENV7 as an ortholog of mammalian STK16 and STK16-related kinase genes that belong to the numb-associated kinase (NAK) family. The latter represents a structurally divergent kinase family characterized by two adjacent glycines at the center of a highly conserved glycine-rich loop (51, 69, 70). STK16-related kinases have the additional characteristic of a modified DFG motif where the central phenylalanine is replaced with leucine (51). Both of these modified structural motifs are present in Env7, and the two share several additional features summarized in Table 4.
We recently uncovered ENV7 in a genome-wide screen for functions at the late endosome and lysosomal vacuole interface. Here, we establish for the first time that ENV7 encodes a vacuolar membrane protein kinase that negatively regulates membrane fusion in vivo and is the yeast ortholog of human STK16. We also establish that Env7 associates with the vacuolar membrane through palmitoylation of one or more cysteines at its Cys13 to -15 palmitoylation consensus sequence and that the consensus sequence is necessary for Env7 kinase activity and complementation of the env phenotype. Taken together, our results suggest that Env7 vacuolar membrane association is essential to its function.
Env7 is required for maintaining vacuole fragmentation during hyperosmotic response and budding. Hence, Env7 negatively regulates vacuolar membrane fusion dynamics under two distinct signal transduction stimuli: osmotic stress and cell cycle progression. As such, Env7 most likely regulates the mechanics of membrane fusion and not the initial sensing of stimuli. The HOG pathway has been implicated in initial sensing of osmotic stress, and Env7 may be a downstream effector of this pathway. Interestingly, HOG1 was another gene uncovered during the env screen (24). As vacuole fission dynamics remain unchanged upon ENV7 deletion, and as ENV7 overexpression leads to increased fragmentation, Env7 appears to act as an inhibitor of membrane fusion and remodeling. The absence of defects in vacuole fission and acidification suggests that Env7 functions independently of the phosphoinositol-regulated vacuole fission events orchestrated by the vacuolar membrane complex of lipid kinase Fab1, lipid phosphatase Fig4, and the scaffolding protein Vac14 (35, 63, 79–87).
The most likely molecular target of Env7-mediated fusion inhibition is the fusion machinery. Membrane fusion is a well-conserved process proposed to be comprised of priming, tethering, docking, and fusion/bilayer mixing (4, 5, 7, 88). Like Env7, Yck3 is implicated in negative regulation of vacuolar membrane fusion and is the only other known resident protein kinase of the vacuolar membrane (22). More specifically, Yck3 modulates the HOPS complex through phosphorylation of its Vps41 component; it also phosphorylates Vam3, the SNARE involved in docking of transport vesicles at the vacuolar membrane (4, 7, 12, 19, 89–91). Our results indicate that Env7 function is distinct from that of Yck3. Env7 may phosphorylate another component(s) of the fusion machinery. Alternatively, Env7 may act directly as a membrane “antifusion” factor independently of its kinase activity. We are currently raising antibodies to Env7 and will be using them to probe for Env7 interactors and substrates in vivo.
Env7E269A is an unstable kinase-dead allele; it is degraded by the proteasome system and accumulates in a detergent-insoluble state when stabilized. Since E269D and E269S mutants of Env7, which retain the potential for salt bridge formation, are stable and enzymatically active, our biochemical results suggest denaturation of Env7E269A in the absence of such stabilization. Salt bridge formation between the glutamic acid of the APE motif and a distant lysine has been implicated in stabilization of the N loops of several kinases (92, 93). The fact that denatured and mislocalized Env7 is a substrate for the proteasomal system supports a role for proteasomal degradation in the regulation of Env7 levels and, hence, vacuolar dynamics. Our preliminary results indicate that Env7 overexpression affects fitness more drastically in proteasome-defective mutants (E. Calle, S. Manandhar, and E. Gharakhanian, unpublished results). Autophosphorylation of Env7 may lead to its stabilization when fusion inhibition is necessitated, such as during sustained hyperosmotic response or budding, while dephosphorylation may lead to proteasomal degradation and downregulation of fusion inhibition. In fact, the ubiquitin-proteasome system is proposed to be generally involved in regulation of vacuole membrane fusion (94), and physical interactions between Env7 and the ubiquitin chain assembly factor (E4) Ufd2 have been reported in a high-throughput mass spectrometric study (95). Posttranslational regulation of STK16 in mammalian cells has also been suggested based on noncorresponding STK16 mRNA and protein levels (71). As the only reversible form of lipidation, palmitoylation may provide an additional level of regulation. A number of vacuole membrane proteins involved in membrane fusion/fission are palmitoylated, including the casein kinase Yck3 (60, 96), the SNARE Ykt6 (56, 97, 98), microautophagy-related Ego3/Meh1 (99, 100), and the fusion factor Vac8 (101–103). As such, palmitoylation/depalmitoylation may provide coordinated regulation of proteins involved in vacuole membrane fusion.
Bayesian phylogenetic studies have established ENV7 as an ortholog of a group of atypical protein kinases, including human STK16. STK16 is also known as myristoylated and palmitoylated serine-threonine kinase (MPSK1) (51), kinase related to cerevisiae and thaliana (KRCT) (76, 104), protein kinase expressed in day 12 fetal liver (PKL12) (71, 77), embryo-derived protein kinase (77), and transforming growth factor-beta (TGF-β)-stimulated factor 1 (TSF-1) (75). STK16-related kinases belong to the structurally divergent NAK family (105), for which an atypical activation loop architecture has been proposed (51). This structural divergence is due to a unique activation loop diglycine repeat shared among STK16-related kinases, including Env7. Our mixed results with the mutagenesis of Env7 residues, which have yielded kinase-dead alleles in TPKs, further support the atypical nature of STK16-related kinases. Consistent with the observed instability of Env7E269A, reported kinase-deficient mammalian STK16 mutants at the ATP binding site lysine (K49M) and APE motif glutamic acid (E202A) have notably lower steady-state levels than their wild-type counterparts (32, 71). Atypical kinases are considered to have diverged early in evolution to form a distinct phyletic group (92). Hence, common structural features of TPKs may be altered or absent in atypical kinases. We are currently probing other conserved residues specifically among STK16-related kinases for generation of stable kinase-dead alleles.
Both ENV7 and STK16 localize to late endomembrane organelles. Our results establish Env7 localization to the lysosomal vacuole membrane; excessive overexpression of GFP-tagged Env7 led to additional localization to the Golgi apparatus. Mammalian STK16 is localized to the Golgi apparatus in fixed murine epithelial cells and fibroblasts (71), and GFP-tagged STK16 localizes to the Golgi apparatus in transformed simian epithelial cell lines (77). Endomembrane localization of these orthologs is consistent with our observation that while STK16-related kinases are conserved across a wide array of eukaryotes, they are conspicuously absent in prokaryotes, an evolutionary pattern seen for orthologs functioning within the endomembrane system (106). While localization of mammalian STK16 to the Golgi apparatus has been established, its cellular function remains elusive. Two studies have suggested a possible role for it in general secretory events. Overexpression of STK16 during mammary gland development in a transgenic mouse model results in increased end bud morphogenesis (104), and overexpression of STK16 in murine cell lines results in enhanced VEGF production and secretion (71). A role for STK16 in DNA binding and transcriptional activation has also been suggested (71, 75). Here, we establish that its orthologous Env7 is a conserved protein kinase negatively regulating fusion at the lysosomal vacuole. Within the endomembrane system, fusion/fission dynamics are central to trafficking and organelle biogenesis beyond the vacuole. Mammalian Golgi apparatuses are organized into stacked cisternae that undergo regulated fusion/fission during the cell cycle (for a review, see reference107). Early in mitosis, Golgi cisternae fragment, and the vesiculated Golgi fragments are inhibited from fusing by phosphorylation of several target proteins. Once Golgi vesicles are distributed to daughter cells, fusion is resumed in late mitosis. This is analogous to vacuolar inheritance events. The Golgi apparatus localization of mammalian STK16 would be consistent with a conserved role for STK16-related kinases in negatively regulating fusion at that organelle. In fact, STK16 overexpression and drug-induced Golgi apparatus disorganization lead to mislocalization of the protein throughout the cell (71), and overexpressed STK16 is found in dispersed and disorganized Golgi structures in simian cells (77). These studies indicate that the Golgi apparatus architecture and STK16 function may be connected. While S. cerevisiae Golgi apparatuses are constitutively disorganized, stacked cisternae are the most common structural form of the Golgi apparatus across eukaryotes, including most fungi (51, 97). Interestingly, the fission yeast Schizosaccharomyces pombe has a stacked Golgi apparatus (108, 109), and its uncharacterized PPK13 ortholog of the ENV7 protein has been reported to be localized to the Golgi apparatus and ER in a systematic protein localization study (110). Our report establishes the budding yeast ortholog of STK16-related kinases as an inhibitor of vacuolar fusion and supports a conserved role for them in regulating endomembrane fusion/fission equilibria.
This project is funded by NSF-RUI research grant 0843569 to E.G. and NSF-MRI grant DBI0722757 for confocal microscopy. S.M. was supported by the NSF-RUI grant.
We are grateful to Greg Payne (UCLA) and his laboratory members for continuous support and reagents and to Simon Malcomber (CSULB) for guidance with the phylogenetic analyses. We thank Walter Schmidt (University of Georgia) for plasmids, Dieter Wolf (University of Stuttgart, Stuttgart, Germany) and Daniel Klionsky (University of Michigan) for yeast strains, Christian Ungermann (University of Osnabrück, Osnabrück, Germany) for the gift of anti-Vps41 antibodies, and Stephane Lefrancois (University of Montreal, Montreal, Canada) for advice on palmitoylation experiments. We also thank Maribeth Seranilla for construction of the 2μm ENV7-carrying plasmid and assistance with the ENV7-GFP strain experiments.
Published ahead of print 19 November 2012