INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Newly synthesized proteins destined for the yeast vacuole are inserted into or transported across the endoplasmic reticulum (ER) membrane. After exit from the ER, they are delivered to the Golgi complex, and it is upon leaving the late Golgi that transport pathways to the vacuole diverge (reviewed in Bryant and Stevens, 1998; Burd et al., 1998; Conibear and Stevens, 1998; Figure 11Ai). Proteins such as the vacuolar protease carboxypeptidase Y (CPY), follow a route from the late Golgi via an endosomal intermediate, the prevacuolar compartment (PVC), to the vacuole. This pathway is referred to as the CPY pathway. Other vacuolar proteins, for example, alkaline phosphatase (ALP), follow a distinct route (the ALP pathway) to the vacuole that bypasses the PVC. A third route allows proteins transported to the cell surface from the Golgi to reach the vacuole after endocytosis. The route taken to the vacuole by proteins endocytosed from the plasma membrane is thought to converge with the route taken by vacuolar proteins following the CPY pathway, at the PVC (Davis et al., 1993; Piper et al., 1995; Rieder et al., 1996).

Genetic screens in yeast have identified >50 genes required for CPY trafficking and processing (PEP genes, Jones 1977; VPS genes, Bankaitis et al., 1986; Rothman and Stevens, 1986; Robinson et al., 1988; Raymond et al., VAC genes, Weisman et al., 1990; 1992; and VAM genes, Wada et al., 1992). Complementation analysis has revealed extensive overlap between the vps, pep, vac, and vam mutants. The VPS genes have been grouped into classes (A through F) based on the vacuolar morphology of mutant yeast cells (Banta et al., 1988; Raymond et al., 1992). Based on the assumption that the loss of any one of the proteins required at a certain transport step will result in a similar vacuolar morphology phenotype, this classification of vps mutants has been helpful in identifying proteins that act at the same step of CPY transport to the vacuole.

Previous reports have suggested that the lumenal acidic environment of endocytic organelles, generated by the proton-translocating, V-type ATPase (V-ATPase), is essential for endosomal trafficking in mammalian cells. The acidic lumenal pH of endosomes is required for ligand-receptor dissociation, binding of coat proteins to early endosomes, and for the entry of some enveloped viruses (reviewed in Stevens and Forgac, 1997). However, the role of endosomal pH and ion balance in endosomal trafficking in yeast is less clear. We have identified the first yeast Vps protein that shows homology to known ion transporters. VPS44, a previously uncloned gene identified in a screen for vps mutants, encodes the sodium/proton exchange protein Nhx1p. Loss of this ion transporter affects the transport of CPY to the vacuole, suggesting a role for ion balance in protein trafficking to the yeast vacuole.

By protein sequence homology, Nhx1p/Vps44p is the only member of the NHE family of sodium/proton exchangers in yeast. Mammalian sodium/proton exchangers of the NHE protein family are required for several key cellular processes, and function to regulate cell volume, intracellular pH, and sodium reabsorption across renal, intestinal, and other epithelial membranes (reviewed in Orlowski and Grinstein, 1997; Counillon and Pouyssegur, 2000). NHE proteins exchange one sodium ion for one proton across the membrane, and are passive exchangers, driven by ion gradients in the cell. All NHE proteins are predicted to have a similar structure, with 10-12 membrane-spanning domains and a large C-terminal domain. Experimental evidence suggests that the C-terminal domain of NHE1 is cytosolic (Shrode et al., 1998; Wakabayashi et al., 2000). However, residues within the C-terminal region of NHE3 may be extracellularly exposed (Biemesderfer et al., 1998), making it difficult to draw general conclusions concerning NHE protein topology. There is some evidence that NHE proteins may exist as homodimers (Fliegel et al., 1993; Fafournoux et al., 1994).

By amino acid sequence alignment, and construction of a phylogenetic tree, NHE proteins fall into two subfamilies (Fukuda et al., 1999). The first contains NHE proteins known to be localized to the plasma membrane, including mammalian NHE proteins 1-5. The second subfamily contains several exchangers proposed to be on the membranes of intracellular organelles, including the yeast NHE protein Nhx1p, the Arabidopsis thaliana vacuolar sodium/proton exchanger (AtNhx1), and human NHE6. The phylogenetic analysis has led to the hypothesis that the two subfamilies of NHE proteins have distinct cellular localizations: there is one subfamily of plasma membrane NHE proteins, and a second of NHE proteins localized to intracellular membranes (Fukuda et al., 1999).

The plasma membrane class of NHE proteins has been extensively studied, and much is known about their expression and regulation (Orlowski and Grinstein, 1997; Counillon and Pouyssegur, 2000). However, much less is known about the intracellularly localized subfamily of NHE proteins. Nhx1p has been proposed to function in salt tolerance, by sequestering sodium ions from the cytosol into the PVC in exchange for protons (Nass et al., 1997). Intracellular NHE proteins from other organisms may also play a role in salt tolerance, and this is supported by a recent study showing that overexpression of the A. thaliana AtNhx1 allowed plants to survive salt stress (Apse et al., 1999).

Data reported in this article reveal a novel role for an intracellular NHE protein. Nhx1p/Vps44p is required for normal membrane traffic out of the PVC in yeast. Whereas vacuolar, prevacuolar, and Golgi membrane proteins following the CPY pathway accumulate in an exaggerated PVC in cells lacking Nhx1p, traffic along the ALP pathway is unaffected in these cells.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Plasmid Construction and Site-directed Mutagenesis

The plasmids used in this study are shown in Table 1. Enzymes used in DNA manipulations were purchased from New England Biolabs (Beverly, MA), or Boehringer Mannheim Biochemicals (Indianapolis, IN). Oligonucleotides were synthesized by Keystone Laboratories (Camarillo, CA). All site-directed mutagenesis was performed by QuikChange mutagenesis (Stratagene, La Jolla, CA). The NHX1 gene was generated as a polymerase chain reaction (PCR) product from a yeast genomic DNA template by using Pfu polymerase (Stratagene) and the oligonucleotides 5'-CCGATGAGTACGGTCGACATTAGC-3' (incorporating a SalI site, shown in bold type) and 5'-GCTTATCGATAGCGGCGAGTTTCTC-3'. This 2.9-kb PCR product included 500 bp upstream of the ATG and 450 bp downstream of the stop codon of NHX1, and was cloned first into pCR-Blunt (Invitrogen, Carlsbad, CA), and then subcloned by using SalI and NotI (from the polylinker of pCR-Blunt) into pBluescript II KS⁺ (Stratagene) to give pKEB24, or pRS316 (Sikorski and Hieter, 1989) to produce pKEB37. Knockout constructs were generated by digesting pKEB24 with HpaI and StuI removing bp 45-1682 of the NHX1 ORF, and replacing this by blunt end ligation with either the TRP1 gene (the 948-bp PvuII-StuI fragment from pJJ246; Jones and Prakash, 1990) or the kanamycin resistance gene (Kan^r; the 1-kb SalI-EcoRV fragment from pFA6-kanMX2; Wach et al., 1994) generating pKEB32 and pKEB43, respectively. From pKEB37, the ATG of ORF YDR455C was mutated to ATT (confirmed by sequencing) to give pKEB36. This mutation leaves the predicted amino acid sequence of Nhx1p unchanged. A SnaBI site was inserted in-frame before the stop codon of NHX1 (changing AAC to TAC, and CAC to GTA) in pKEB36 to allow the C-terminal tagging of Nhx1p by using a triple hemagglutinin (HA) epitope fragment from pKEB35, generating pKEB38. NHX1-HA from pKEB38 was subcloned into pRS306 (pKEB39), and pRS315 (pKEB53). pKEB38 was used as a template to generate mutations GAC to AAC (D201N; pKEB44, pFP1, and pFP2), GAA to CAA (E225Q; pKEB45 and pFP1), GAT to AAT (D230N; pKEB46 and pFP2), and GAG to CAG (E355Q; pKEB47), which were confirmed by sequencing. The GEF1 gene was generated as a PCR product from a yeast genomic DNA template by using Pfu polymerase and the oligonucleotides 5'-GCGAATTCGCTTCCTCGATCCTCA-3' and 5'-CGGGATCCTGCGAGCCTTATAATG-3', incorporating EcoRI and BamHI sites (shown in bold type). This 3.3-kb PCR product included 500 bp upstream of the ATG and 470 bp downstream of the stop codon of GEF1, and was cloned first into pCR-Blunt and then subcloned by using EcoRI and BamHI into pBluescript II KS⁺ to produce pBL2. A deletion construct, pBL20, was generated by digesting pBL2 with HpaI and NarI (removing bp 302-1558 of the GEF1 ORF), and replacing this by blunt-end ligation with a DNA fragment containing LEU2. pKEB58 was generated by subcloning the 4.2-kb BamHI-SalI fragment containing PEP4 from pTS18 into the same sites of pRS315.

Yeast Strains

Yeast strains used in this study are shown in Table 2. Strains were constructed by standard genetic techniques and grown in rich medium (1% yeast extract, 1% peptone, 2% dextrose; YEPD) or synthetic dextrose minimal medium with appropriate supplements (SD; Kaiser et al., 1994). YEPD buffered to pH 5.0, and YEPD with 100 mM CaCl₂, were also used for the selection and growth of Vma strains (Yamashiro et al., 1990). All strains were derived from SF838-9D (Rothman and Stevens, 1986) with the exception of KEBY10, which was derived from SEY6210 (Robinson et al., 1988). Strain KEBY11 was generated by transforming SF838-9D with BglII digested pKEB39. Ura⁺ colonies were plated on 5-fluoroorotic acid-containing minimal medium to select for Ura loopouts, and colonies that contained Nhx1p-HA identified by Western blot. KEBY12 was derived from KEBY11 by transformation with BamHI and PstI digested pKJH2. Leu⁺ colonies were screened for deletion of VPS27 by colony overlay assay to detect CPY secretion (Roberts et al., 1991), and by immunofluorescence with anti-Vma2p to screen for class E Vps vacuolar morphology (Raymond et al., 1992). KEBY15 was generated by transformation of SF838-9D with XhoI/NotI digested pKEB43. Kanamycin-resistant colonies were selected on YEPD with 200 µg/ml geneticin sulfate (G418), and then screened for secretion of CPY. Insertion of the Kan^r gene into the NHX1 open reading frame (ORF) was confirmed by PCR from genomic DNA by using oligonucleotides complementary to the DNA sequence 500 bp upstream of the NHX1 ORF, and to a sequence within Kan^r. KEBY13, KEBY14, and KEBY10 were similarly derived from RPY10, AACY28, and SEY6210 (with pKEB32), respectively. KEBY26 and KEBY27 were generated from SF838-9D and RPY10, respectively, by transformation with SalI/ClaI digested pCY40 and Leu⁺ colonies were selected for lack of growth on YEPD with 100 mM CaCl₂. KEBY34 and KEBY35 were similarly derived from KEBY15 and KEBY13, respectively. KEBY29 was derived from SF838-9D by transformation with EcoRI/HindIII digested pMK10, and Ura⁺ colonies were screened for the Vma phenotype as described above. KEBY32 was derived from RPY10 by transformation with XbaI/XhoI digested pBL20. Leu⁺ colonies were screened for correct insertion of LEU2 into the GEF1 ORF by PCR, with oligonucleotides complementary to the genomic DNA sequence 600 bp upstream of GEF1, and a sequence within LEU2. BLY1 was generated by transforming NBY72 (Bryant et al., 1998) with SacI-digested pSL1417. Ura⁺ colonies were plated on 5-fluoroorotic acid-containing medium to select for Ura loopouts, and colonies that contained proteinase A identified by APNE assay (Wolf and Fink, 1975). KEBY37 was derived from BLY1 by transformation of BamHI/PstI digested pKJH2 and screened for deletion of VPS27 as described above.

Immunoblotting

Exponentially growing cells (10 OD₆₀₀ total) were harvested and treated with 10 mM dithiothreitol for 15 min at room temperature. Cells were converted to spheroplasts by incubation with 250 µg/ml Zymolyase 100T (Seikagaku America, Ijamsville, MD) in 1.2 M sorbitol, 50 mM potassium phosphate buffer, pH 7.5, 1 mM MgCl₂ for 45 min at 30°C. Spheroplasts were washed once with 1.2 M sorbitol and lysed by resuspension in 0.2 M sorbitol, 50 mM Tris, pH 7.5, 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 1 µg/ml leupeptin, 1 µg/ml pepstatin. Unbroken cells were removed by centrifugation at 500 × g for 5 min at 4°C. The cell lysate was assayed for total protein concentration by using the bicinchoninic acid protein assay (Pierce, Rockford, IL), and equal amounts of total protein loaded per lane on 8% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membranes and immunoblot analysis of Nhx1p-HA performed by using the 12CA5 anti-HA monoclonal antibody (Roche Molecular Biochemicals, Indianapolis, IN), a secondary anti-mouse horseradish peroxidase-conjugated antibody (Bio-Rad Laboratories, Hercules, CA), and chemiluminescent detection (New England Nuclear, Boston, MA).

Radiolabeling and Immunoprecipitation

Pulse-chase immunoprecipitation of radiolabeled CPY, ALP, A-ALP, and Vps10p was carried out essentially as previously described (Raymond et al., 1992; Nothwehr et al., 1995). For CPY immunoprecipitation, yeast cultures were grown overnight at 30°C in minimal medium without methionine to mid-log phase. Cells were then resuspended at 1 OD₆₀₀/ml in fresh minimal medium (without methionine) with 50 mM KPO₄, pH 5.7, and 2 mg/ml bovine serum albumin (BSA), incubated for 15 min at 30°C, and pulse-labeled with 200 µCi of Express [³⁵S] labeling mix (New England Nuclear) per 1 OD₆₀₀ for 10 min at 30°C. The chase was initiated by the addition of 100 µg/ml each of unlabeled cysteine and methionine, and after incubation for the appropriate time at 30°C, 500 µl aliquots were removed and the chase terminated by addition of 50 mM sodium azide and incubation on ice. Cells were collected by centrifugation and the supernatant (extracellular fraction) reserved. After conversion to spheroplasts using oxalyticase (Enzymogenetics, Corvallis, OR), 0.5% SDS was added to lyse the cells, and the samples incubated at 100°C for 5 min. Samples were then adjusted to 0.1% SDS, 0.1% Triton X-100, 2 mM EDTA, and 90 mM Tris, pH 8.0 (1× IP buffer). Then 100 µl of 10× IP buffer was added to each extracellular fraction and the samples incubated at 100°C for 5 min before diluting 10-fold with water. Intracellular and extracellular fractions were precleared with 50 µl of a 10% slurry of fixed Staphylococcus aureus cells (IgGSorb; The Enzyme Center, Malden, MA) for 15 min. Anti-CPY (1 µl) serum was added and incubated for 1 h on ice, followed by a second incubation for 1 h on ice after addition of 50 µl of a 10% slurry of IgG Sorb. IgG Sorb with bound immune complexes was collected by centrifugation and washed three times in 1× IP buffer. Immunoprecipitated proteins were eluted in SDS sample buffer and separated on 7% SDS polyacrylamide gels. ALP, A-ALP, and Vps10p immunoprecipitations were carried out as described above, except extracellular fractions were discarded, and spheroplasts were lysed in 2% urea, 0.25% SDS for 5 min at 100°C.

Immunofluorescence Microscopy

Indirect immunofluorescence microscopy was performed essentially as previously described (Roberts et al., 1991). Cells were grown in YEPD at 30°C to 1 OD₆₀₀/ml, and then fixed by the addition of 3% formaldehyde for 1 h, followed by a 16-h incubation at room temperature in 4% paraformaldehyde, 50 mM KPO₄, pH 6.5. Cells were converted to spheroplasts using Zymolyase 100T and permeabilized by treatment with 5% SDS for 5 min for visualization of ALP or Vph1p, or 1% SDS for 2 min for detection by using all other antibodies (unless otherwise described in the figure legend). Cells were allowed to adhere to poly(L-lysine) multiwell-coated slides. Nonspecific antibody binding was blocked by incubation of the cells in PBS with 5 mg/ml BSA and 1% normal goat serum. All antibodies were diluted in PBS with 5 mg/ml BSA, which was also used for all washes. Antibodies against Vph1p, ALP, Pep12p, Vps10p, and HA were preabsorbed to yeast proteins (to remove nonspecific binding) by incubation with vph1, pho8, pep12, vps10, or wild-type SF838-9D cells, respectively. Antibody incubations were performed at room temperature: 2 h for primary antibodies (3 h for anti-ALP), and 1 h for secondary and tertiary antibodies. Affinity-purified polyclonal antibodies against, Vma2p, and Vps10p have been described previously (Yamashiro et al., 1990; Cooper and Stevens, 1996) and were used at dilutions of 1/500 and 1/400, respectively. Polyclonal serum against Vph1p (Hill and Stevens, 1994) was used at a 1/100 dilution. To produce the polyclonal anti-HA antibody, E. coli cells containing pKH105 were used to express a dihydrofolate reductase-12xHA fusion protein with a 6xHis tag. The antigen was purified over a nickel-agarose column, and used to inject rabbits as previously described (Vaitukaitis, 1981). The anti-HA serum was affinity-purified as described in Stevens et al. (1982), and used at a dilution of 1/200. Monoclonal antibodies against ALP (1D3-A10), and Pep12p (24-2C3G4) are available from Molecular Probes (Eugene, OR), and were used at concentrations of 1/3 for anti-ALP tissue culture supernatant, and 1/1000 for purified anti-Pep12p. The purified anti-HA monoclonal antibody (HA.11) was purchased from Babco (Berkeley, CA), and used at 1/500. Biotin-conjugated goat anti-mouse, biotin-conjugated goat anti-rabbit, and streptavidin-conjugated fluorescein isothiocyanate (FITC) antibodies were obtained from Jackson Immunoresearch (West Grove, PA). Alexa (A594)-conjugated goat anti-rabbit and Alexa (A594)-conjugated goat anti-mouse antibodies were obtained from Molecular Probes. Images were generated by using a Bio-Rad MRC 1024 confocal microscope, or an Axioplan 2 fluorescence microscope (Carl Zeiss, Thornwood, NY) fitted with an Orca 100 digital camera (Hamamatsu, Bridgewater, NJ).

FM4-64 Staining

FM4-64 [N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenyl-hexatrienyl, pyridinium dibromide; Molecular Probes] staining was performed as described in Vida and Emr (1995), with a 15-min incubation of the cells in YEPD with 40 µM FM4-64 at 30°C, followed by a 30-min chase in YEPD without the dye at 30°C. Cells were viewed with a 100× oil immersion lens on an Axioplan 2 fluorescence microscope and images captured with an Orca 100 digital camera.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Identification of VPS44 as NHX1

VPS44 is a previously uncloned gene identified in a genetic screen for mutants that affect protein trafficking to the yeast vacuole (vps mutants; Rothman and Stevens, 1986; Rothman et al., 1989; Raymond et al., 1992). More recently, we have used a transposon-based mutagenic procedure to screen for yeast mutants that have a Golgi retention deficient (Grd) phenotype, and thus mislocalize late Golgi proteins to the vacuole (Nothwehr et al., 1996; Voos and Stevens, 1998). Mutants identified in this screen were tested by complementation analysis with all vps and grd mutant strains. One mutant failed to complement a vps44 mutant yeast strain for secretion of CPY, and sequence analysis showed that the transposon had inserted into the genome within the ORFs YDR456W and YDR455C. The 2-kb YDR456W ORF encodes the sodium/proton exchanger Nhx1p (Nass et al., 1997). YDR455C, which overlaps NHX1, is present on the opposite strand (Figure 1), and is a small yeast ORF of unknown function (Cherry et al., 1997). The transposon-based screen therefore suggests that the VPS44 gene is either NHX1 or YDR455C.

View larger version (11K): [in this window] [in a new window]   Figure 1.   Deletion of the NHX1 gene. The NHX1 ORF is located on S. cerevisiae chromosome IV. YDR455C, a small yeast ORF of unknown function, overlaps NHX1 on the other DNA strand. Replacement of NHX1 with Kanr by homologous recombination also disrupts YDR455C.

We have replaced bp 45-1682 of NHX1 in the yeast genome with the gene conferring kanamycin resistance (Kan^r), to produce the mutant yeast strain nhx1, as shown in Figure 1. This deletion strain also has a disruption of the 5' terminus of the YDR455C ORF. The nhx1 cells were assayed for secretion of CPY by pulse-chase immunoprecipitation. Figure 2 shows that after a 10-min pulse of radiolabel and no chase, three intracellular forms of CPY were seen in wild-type cells. These three forms represent the newly synthesized ER form of CPY (p1), the slightly larger Golgi-modified form (p2), and the proteolytically cleaved, mature vacuolar form (m). After a 40-min chase, 96 ± 2% of newly synthesized CPY was intracellular and mature in wild-type cells, consistent with correct localization of CPY to the vacuole. In contrast, nhx1 cells secreted 34 ± 7% of newly synthesized CPY into the extracellular medium in the Golgi-modified p2 form, suggesting that CPY trafficking to the vacuole was disrupted. Transformation of nhx1 cells with centromere-based (CEN) plasmids containing either NHX1 and YDR455C, or NHX1 alone (NHX1-HA; the ATG of the YDR455C ORF was mutated to ATT) reduced the CPY secretion to wild-type levels. Thus, NHX1 expressed from its endogenous promoter on a CEN plasmid was able to complement the CPY secretion phenotype of nhx1 cells. The secretion of CPY by nhx1 cells was therefore due to the loss of Nhx1p rather than the loss of a protein encoded by the YDR455C ORF. As well as a disruption of YDR455C, the NHX1-HA construct also has an epitope tag sequence introduced into NHX1 before the stop codon. NHX1-HA was able to fully complement the CPY secretion phenotype of nhx1 cells (Figure 2), showing that the addition of an HA tag to the extreme C terminus of Nhx1p did not affect the function of the protein in CPY trafficking. Examination of CPY secretion levels by immunoprecipitation also revealed that NHX1-HA was able to fully complement a vps44 mutant strain for CPY secretion (Figure 2), adding further evidence that NHX1 and VPS44 are identical.

View larger version (29K): [in this window] [in a new window]   Figure 2.   Secretion of CPY in wild-type (WT), nhx1::Kanr, and vps44 mutant strains, and complementation by NHX1 and NHX1-HA. Cells were labeled with [35S]methionine and cysteine for 10 min, and then chased for 0 or 40 min at 30°C. CPY was immunoprecipitated from intracellular (I) and extracellular (E) fractions for each time point. The positions of ER and Golgi precursor forms of CPY (p1 and p2, respectively), and mature vacuolar CPY (m) are indicated. nhx1 cells were transformed with NHX1 and YDR455C on a CEN plasmid (NHX1+YDR455C, pKEB37), or NHX1 alone with an HA epitope tag (NHX1-HA, pKEB38). vps44 cells were transformed with a CEN plasmid containing NHX1 alone with an HA-epitope tag (NHX1-HA, pKEB53). WT and nhx1 strains have a wild-type PEP4 gene, encoding proteinase A (required for proteolytic processing). vps44 (vpl27-1) has the pep4-3 mutation, and was transformed with PEP4 on a CEN plasmid (pTS18) for this experiment. The percentage of CPY secreted into the extracellular medium after a 40-min chase was calculated by phosphoimager analysis, and the numbers shown are averages over several separate experiments.

Nhx1p Is Required for Traffic Out of the PVC

The classification of VPS genes into six groups (A-F) based on vacuolar morphology allows the identification of genes that may act at distinct steps in the trafficking pathway to the vacuole (Raymond et al., 1992). Although VPS44 (NHX1) was originally designated a class A VPS gene (on the basis that vps44 mutant cells exhibited wild-type vacuolar morphology; Raymond et al., 1992), we reexamined nhx1 cells by using a variety of organelle markers that are now available. We have used Pep12p as a marker for the prevacuole (Becherer et al., 1996; Gerrard et al., 2000b). In wild-type cells, Pep12p is localized to small punctate structures throughout the cytoplasm (Figure 3 top row). Vps10p/Pep1p, the CPY receptor, binds CPY in the late Golgi and is transported to the PVC where CPY is released. From the PVC, CPY travels to the vacuole, whereas Vps10p is recycled back to the late Golgi where it captures more CPY (Marcusson et al., 1994; Cereghino et al., 1995; Cooper and Stevens, 1996). In wild-type cells Vps10p is localized by immunofluorescence to punctate structures typical of the Golgi in yeast (Figure 3, top row). Vph1p, the 100-kDa transmembrane subunit of the V-ATPase follows the CPY pathway to the vacuole and is localized to the vacuole membrane in wild-type cells (Piper et al., 1997; Figure 3, top row). Unlike their distribution in wild-type cells, Pep12p, Vps10p, and Vph1p were all accumulated in an aberrant structure next to the vacuole in nhx1 cells (Figure 3, middle row).

View larger version (74K): [in this window] [in a new window]   Figure 3.   nhx1 cells have a similar morphological phenotype to vps27 cells. Immunofluorescence was performed as described in MATERIALS AND METHODS in wild-type (WT; SF838-9D), nhx1 (KEBY15), and vps27 cells (KEBY12), by using monoclonal antibodies to ALP or Pep12p, and polyclonal antibodies to Vps10p, Vph1p, or Vma2p. Staining for Vps10p, Pep12p, and ALP was enhanced by using secondary anti-mouse or anti-rabbit biotin-conjugated antibodies followed by a tertiary streptavidin-FITC conjugated antibody. Vph1p and Vma2p were detected by using a secondary Alexa-conjugated anti-rabbit antibody.

VPS27 is characterized as a class E VPS gene, and cells lacking Vps27p have a large, aberrant prevacuole structure next to the vacuole (the class E compartment; Raymond et al., 1992; Piper et al., 1995). Proteins that normally reside in the PVC, those that normally cycle via the PVC back to the late Golgi, and proteins that travel via the PVC to the vacuole accumulate in the class E compartment of vps27 cells (Piper et al., 1995). Significantly, the aberrant structure observed in cells lacking Nhx1p closely resembles the class E compartment seen in vps27 cells (compare Figure 3, middle and bottom rows). In vps27 cells Vps10p and Vph1p were seen almost exclusively in the class E compartment. However, unlike vps27, nhx1 cells appeared to possess a somewhat weaker morphological phenotype because some Vps10p and Vph1p was seen in the Golgi and vacuole, respectively.

We have also followed the immunolocalization of Vma2p/Vat2p, a peripheral membrane subunit of the V-ATPase in wild-type, nhx1, and vps27 cells. In wild-type cells, Vma2p is localized to the vacuole membrane by immunofluorescence using an anti-Vma2p antibody (Figure 3, top row). Vma2p was predominantly in the class E compartment in vps27 cells, with little staining of the vacuole membrane (Figure 3, bottom row). In nhx1 cells, however, Vma2p clearly stained the vacuole and little could be seen in a class E-like structure (Figure 3, middle row). This result is consistent with nhx1 cells having a somewhat weaker PVC transport block than the vps27 cells. In addition, because Vma2p was used as a vacuolar marker for the original classification of vps mutants, this explains why vps44 was previously designated as a class A VPS gene.

ALP follows a distinct pathway separate from CPY to the vacuole (Cowles et al., 1997; Piper et al., 1997). In wild-type cells, ALP stained the vacuole membrane by immunofluorescence with an anti-ALP antibody (Figure 3, top row). ALP was also localized to the vacuole membrane in nhx1 and vps27 cells (Figure 3, middle and bottom rows), suggesting that loss of either Nhx1p or Vps27p has no effect on the trafficking of ALP in these cells.

To determine whether PVC, late Golgi, and vacuolar proteins accumulated in the same aberrant structure in nhx1 cells, we performed colocalization immunofluorescence experiments. As shown in Figure 4, Pep12p, Vps10p, and Vph1p all colocalized in the class E-like compartment of nhx1.

View larger version (92K): [in this window] [in a new window]   Figure 4.   Colocalization of Pep12p, Vph1p, and Vps10p in nhx1 cells. Immunofluorescence was performed as described in MATERIALS AND METHODS, except cells were permeablized with 5% SDS for 5 min. The strain used was KEBY15 (nhx1). Pep12p was visualized by using a monoclonal antibody against Pep12p followed by anti-mouse biotin and streptavidin-FITC. The same cells were also stained for Vps10p, or Vph1p by using polyclonal antibodies and a secondary Alexa-anti-rabbit antibody. Confocal micrographs were taken simultaneously of the red and green fluorescence channels and overlapped to produce the merged image.

In addition to proteins that follow the CPY and ALP trafficking pathways to the vacuole, we have also examined endocytic protein trafficking by using the lipophilic fluorescent dye FM4-64, and an a-factor receptor-green fluorescent protein chimera (Ste3p-GFP). FM4-64 incorporates into the plasma membrane, from where it is endocytosed and transported via the endocytic pathway to the vacuole (Vida and Emr, 1995). Ste3p is a seven transmembrane receptor that is endocytosed from the plasma membrane in both a constitutive and ligand-dependent manner, and delivered to the lumen of the vacuole in wild-type cells, where it is degraded (Davis et al., 1993). In wild-type cells lacking vacuolar proteases (pep4-3), Ste3p and Ste3p-GFP can be visualized in the vacuolar lumen (Piper et al., 1995; Urbanowski and Piper, 1999; Gerrard et al., 2000a; Figure 5). Cells expressing Ste3p-GFP were incubated for 15 min with FM4-64 dye, followed by a 30-min chase in fresh medium. The dye was seen in the limiting vacuole membrane of wild-type cells (Figure 5). However, in vps27 cells, both Ste3p-GFP and FM4-64 accumulated in the class E compartment. Some FM4-64 staining of the vacuole membrane was also visible (Vida and Emr, 1995; Figure 5). nhx1 cells also showed FM4-64 accumulation in the class E-like compartment, with some vacuolar staining (Figure 5). Our data indicate that nhx1 cells, similar to vps27 cells, are defective for the transport of endocytosed proteins to the vacuole.

View larger version (81K): [in this window] [in a new window]   Figure 5.   FM4-64 and Ste3p-GFP accumulate in the class E compartment of nhx1 cells. Wild-type (WT; SF838-9D), nhx1 (KEBY15), and vps27 cells (SGY73) were transformed with pJLU34. Cells were labeled in 40 µM FM4-64 for 15 min at 30°C and then chased in fresh medium for 30 min at 30°C. FM4-64 and Ste3p-GFP were photographed under the red and green fluorescence channels, respectively, and a merged image of these two channels is also shown. Differential interference contrast (DIC) images of the same cells were collected to visualize the vacuoles.

In the experiments presented above, we have shown that Pep12p, Vps10p, Vph1p, Ste3p, and FM4-64 all accumulate in a large aberrant structure next to the vacuole in nhx1 cells. This phenotype is similar to that seen for vps27 cells, and leads us to classify NHX1/VPS44 as a class E VPS gene. The weaker morphological phenotype of nhx1 compared with vps27 is consistent with the observed levels of CPY secretion. nhx1 cells secrete 34 ± 7% of newly synthesized CPY (Figure 2), compared with vps27 cells that secrete 53 ± 1% (Bowers and Stevens, unpublished results; Piper et al., 1995).

Previous experiments have shown that the class E compartment contains active proteases (Raymond et al., 1992; Piper et al., 1995; Bryant and Stevens, 1997). Thus, in a vps27 strain, Vps10p accumulates in the class E compartment and is proteolytically cleaved to a lower molecular mass form more rapidly than in wild-type cells (Piper et al., 1995). To investigate the proteolytic cleavage of Vps10p in nhx1 cells, we performed pulse-chase immunoprecipitations of radiolabeled Vps10p (Figure 6, top). In wild-type cells, Vps10p was cleaved very slowly, with only 10% of Vps10p seen in the lower molecular weight form after 30 min. However, in nhx1 cells, 25% of newly synthesized Vps10p was proteolytically cleaved to the smaller form after 30 min, and in vps27 cells Vps10p was cleaved rapidly with 45% in the smaller form after 30 min.

View larger version (51K): [in this window] [in a new window]   Figure 6.   Pulse-chase immunoprecipitations of Vps10p, A-ALP, and ALP. Wild-type (WT; RPY10), nhx1 (KEBY13), and vps27 cells (AACY5) were pulsed for 10 min at 30°C with [35S]cysteine and methionine, and chased for 0, 30, 60, or 120 min with excess unlabeled cysteine and methionine. Vps10p immunoprecipitates were separated on a 7% SDS polyacrylamide gel. The positions of mature Vps10p (m) and the smaller proteolytically cleaved form (*) are shown. A-ALP immunoprecipitations were carried out under the same conditions as used for Vps10p. The following strains were used, all of which had a deletion of the PHO8 gene, which encodes ALP, and were transformed with pSN55 (A-ALP on a CEN-based plasmid): BLY1 (WT), KEBY14 (nhx1), and KEBY37 (vps27). The positions of the precursor form of A-ALP (p) and the proteolytically cleaved form (m) are indicated. ALP immunoprecipitations were carried out as for Vps10p, but by using 0-, 5-, and 15-min chase times. The positions of precursor ALP (p) and mature ALP (m) are shown.

Vps10p has signals within its cytoplasmic domain for retrieval from the prevacuole to the Golgi (Marcusson et al., 1994; Cereghino et al., 1995; Cooper and Stevens, 1996). Dipeptidyl aminopeptidase A/Ste13p also cycles between the late Golgi and the PVC, but unlike Vps10p is retained in the Golgi by both retrieval and retention mechanisms (Bryant and Stevens, 1997). We have investigated the proteolytic cleavage of A-ALP, a chimeric protein formed from the cytoplasmic domain (and hence including the trafficking signals) of dipeptidyl aminopeptidase A, and the membrane and lumenal domains of ALP (Nothwehr et al., 1993). As shown in Figure 6 (middle), in wild-type cells, newly synthesized A-ALP is not significantly cleaved after 120 min. However, in nhx1 cells A-ALP is cleaved more rapidly, with 50% of the newly synthesized protein seen as the lower molecular weight form after 60 min. This aberrant processing of A-ALP in nhx1 cells is not as rapid as in vps27 cells, which show 70% of A-ALP in the smaller form after 60 min.

Unlike Vps10p and A-ALP, which travel through a prevacuolar endosome compartment, ALP follows a distinct pathway to the vacuole that bypasses the PVC (Bryant and Stevens, 1997; Piper et al., 1997). In wild-type cells, ALP is processed by vacuolar proteases to its mature form rapidly, with >50% of the protein in the mature form after a 10-min pulse of radiolabel (Klionsky and Emr, 1989; Piper et al., 1997; Figure 6, bottom). A similar ALP processing rate to that in wild-type cells was observed in nhx1 and vps27 cells (Figure 6, bottom; Bryant et al., 1998). Together with the immunofluorescence data presented above (Figure 3), these results indicate that Nhx1p plays no role in the trafficking of ALP to the vacuole.

The biochemical data for the proteolytic cleavage of Vps10p and A-ALP in nhx1 cells are consistent with the morphology studies presented above, and show that cells lacking Nhx1p have an aberrant PVC structure next to the vacuole that is proteolytically active. Thus, by both biochemical and morphological criteria NHX1 is a class E vps gene. However, the trafficking phenotype of nhx1 cells is not as severe as that seen in vps27 cells.

The Function of Nhx1p in Protein Trafficking Is Distinct from that of the V-ATPase and the Chloride Channel Gef1p

Nhx1p, an endosomal sodium/proton exchanger, is proposed to sequester sodium ions inside the PVC in exchange for protons that are released into the cytosol (Nass et al., 1997; Nass and Rao, 1998, 1999). It has been proposed that the proton gradient created by the V-ATPase across the PVC membrane is used by Nhx1p to transport sodium ions into this compartment (Nass and Rao, 1998; Gaxiola et al., 1999; Quintero et al., 2000). It is reasonable to assume therefore that Nhx1p function depends on the proton pumping activity of the V-ATPase. In addition, the chloride channel Gef1p (which has been shown to partially colocalize with Nhx1p; Gaxiola et al., 1999) is proposed to transport chloride ions into the PVC and late Golgi (Figure 7A). Transport of chloride anions would serve to neutralize the membrane potential across the PVC membrane generated by the V-ATPase, allowing further acidification of the compartment (Gaxiola et al., 1998, 1999; Schwappach et al., 1998). We have studied the protein trafficking phenotype caused by loss of the V-ATPase or Gef1p, compared with that seen after loss of Nhx1p.

View larger version (58K): [in this window] [in a new window]   Figure 7.   Analysis of the trafficking defect of vma and gef1 cells. (A) Proposed ion transporters of the PVC. The V-ATPase actively pumps protons into the PVC, while Nhx1p passively exchanges sodium ions for protons driven by ion gradients across the membrane. Gef1p is a putative voltage-gated ion channel thought to transport chloride ions into the PVC. (B) CPY processing and secretion. Wild-type (RPY10), nhx1 (KEBY13), vma2 (KEBY27), and gef1 cells (KEBY32) were labeled with [35S]methionine and cysteine for 10 min and then chased for 0 or 40 min at 30°C. CPY was immunoprecipitated as described in Figure 1, and in the MATERIALS AND METHODS. The amount of CPY secreted after a 40-min chase was quantified by using phosphoimager analysis, and is shown underneath the gel. The % CPY secreted for each strain is the average value obtained over several independent experiments (7 for WT and nhx1, 3 for vma2, and 2 for gef1). (C) Proteolytic cleavage of Vps10p. Wild-type (RPY10), nhx1 (KEBY13), vma2 (KEBY27), vma3 (KEBY29 with pTS18), gef1 cells (KEBY32), and nhx1 vma2 cells (KEBY35) were metabolically labeled as described above and chased for 0, 30, 60, or 120 min. Vps10p immunoprecipitates were separated on a SDS polyacrylamide gel, which was exposed to a phosphoimager screen and the band intensities quantified. The amount of the lower, cleaved form of Vps10p (*) was calculated as a percentage of total Vps10p at each time point. Each point on the graph represents the mean of several independent experiments (n = number of experiments), and the error bars represent SDs from the mean.

The V-ATPase is a protein complex consisting of 13 subunits, with a peripheral membrane V₁ subcomplex required for the hydrolysis of ATP, and an integral membrane V₀ subcomplex involved in proton translocation (reviewed in Graham and Stevens, 1999; Forgac, 2000). Deletion of any subunit of the complex (except Vph1p, which has a second isoform in yeast, Stv1p; Manolson et al., 1994) results in loss of proton translocation. Gef1p is a putative voltage-gated chloride channel of the CLC family, and possibly functions as a homodimer (reviewed in Jentsch et al., 1999). We have constructed yeast strains lacking Vma2p (a V₁ subunit of the V-ATPase), Vma3p (a V₀ subunit), or Gef1p, for phenotypic comparison with nhx1 cells.

As shown in Figure 7B, wild-type and gef1 cells secreted 4 ± 2% and 10 ± 0.5% of their newly synthesized CPY, respectively. This is consistent with a previous study, which demonstrated that gef1 mutant cells do not secrete significant amounts of CPY (Li et al., 1999). In contrast to gef1, vma2 cells secreted 22 ± 2% of newly synthesized CPY, which is more than wild-type, but less than the 34 ± 7% secreted by nhx1 cells (Figure 7B). Figure 7B also shows that although ~20% of CPY was secreted in the vma2 strain, the majority of the intracellular CPY was mature after 40 min. Cells lacking the V₀ subunit Vma3p also secreted 22 ± 3% of newly synthesized CPY. Therefore, although cells lacking the V-ATPase show some secretion of CPY, they secrete significantly less CPY than cells lacking Nhx1p (Figure 7B).

We have also investigated the proteolytic cleavage of Vps10p in vma2, vma3, gef1, and nhx1. As shown in Figures 6 and 7C, Vps10p was more rapidly cleaved to a smaller product in nhx1 than in wild-type cells. However, in vma2, vma3, and gef1 cells, the kinetics of Vps10p proteolytic cleavage was indistinguishable from wild-type (Figure 7C). These data suggest that although vma2 and vma3 strains secrete some CPY, they have a trafficking phenotype distinct from that observed for nhx1. This was supported by immunofluorescence experiments that localized Pep12p and Vph1p in vma2, vma3, or gef1 cells. As described above, nhx1 cells exhibit an aberrant PVC compartment next to the vacuole where Pep12p accumulates, distinct from its punctate localization in wild-type cells (Figures 3, 4, and 8). Pep12p did not accumulate in a class E-like compartment in vma2, vma3, or gef1 cells and instead exhibited a more wild-type, dispersed PVC staining (Figure 8; Bowers and Stevens, unpublished results). Vph1p also accumulated in the class E-like compartment of nhx1 cells (Figures 3, 4, and 8). However, no similar compartment was seen in vma2 and gef1 cells, and Vph1p antibodies stained the vacuole membrane (Figure 8; Bowers and Stevens, unpublished results). In vma3 cells, Vph1p is localized to the ER because the V₀ V-ATPase subcomplex is unable to assemble correctly (Jackson and Stevens, 1997).

View larger version (35K): [in this window] [in a new window]   Figure 8.   Unlike nhx1 or nhx1 vma2 cells, vma2 cells do not show a class E Vps morphological phenotype. Immunofluorescence was performed as described in Figure 4, with wild-type (WT: SF8389D), nhx1 (KEBY15), vma2 (KEBY26), and nhx1 vma2 (KEBY34) cells and anti-Pep12p and anti-Vph1p antibodies. Images were captured by using a fluorescence microscope fitted with a digital camera.

The data presented here demonstrate that inactivation of the V-ATPase by the loss of a subunit, or loss of the chloride channel Gef1p does not have the same effect on protein trafficking in the CPY pathway as loss of the sodium/proton exchanger Nhx1p. gef1 cells showed no detectable trafficking defects of CPY, Vps10p, Vph1p, or Pep12p in our experiments, whereas vma strains showed some secretion of CPY, and no defect in Vps10p cleavage, or Pep12p and Vph1p localization. Thus, loss of Nhx1p creates a much more severe effect on protein trafficking to the vacuole than loss of the V-ATPase. There are at least two interpretations of these results. One idea is that Nhx1p requires a proton gradient for function, and can still function in the absence of the V-ATPase (suggesting that the PVC is acidified in these cells by an alternative mechanism). The second hypothesis is that Nhx1p is only required if the V-ATPase is present, to reduce the proton gradient or raise the pH within the PVC (Figure 7A). To attempt to distinguish between these hypotheses, we constructed a double mutant nhx1 vma2 yeast strain. If Nhx1p is needed in the presence of the V-ATPase for correct protein trafficking, then we might expect the double mutant to phenocopy the vma2 single delete cells.

nhx1 vma2 cells had the growth phenotypes associated with cells lacking the V-ATPase (they failed to grow at neutral pH, or when 100 mM CaCl₂ was added to the growth medium). In addition, they secreted 37% of newly synthesized CPY and were thus indistinguishable from the nhx1 cells with respect to CPY sorting. Interestingly, nhx1 vma2 cells showed the class E phenotype of nhx1 cells, as measured by increased kinetics of Vps10p proteolytic cleavage (Figure 7C), and by immunofluorescence of Vph1p or Pep12p (Figure 8).

Three Acidic Residues Are Required for the Function of Nhx1p in Protein Trafficking

Our results show that Nhx1p is essential for correct protein trafficking in the yeast CPY pathway. However, the trafficking phenotype of yeast strains with an inactive V-ATPase is not as severe as the class E phenotype of nhx1 cells. These results were surprising because Nhx1p has been proposed to use the proton gradient generated by the V-ATPase to sequester sodium ions inside the PVC, and would thus rely on this proton gradient for its exchange function. We have investigated whether Nhx1p exchange activity might be important for vacuolar protein transport, or whether the presence of the protein is important, perhaps as a scaffold for other factors.

To study the effect of Nhx1p activity on protein trafficking, we sought to make mutant cells that correctly expressed and localized a form of Nhx1p predicted to lack ion exchange activity. Glutamic acid 262 of NHE1 has previously been shown to be essential for the sodium/proton exchange activity of the mammalian protein (Fafournoux et al., 1994). In addition, studies of sodium/proton exchangers unrelated to the NHE family have shown that specific acidic residues within transmembrane domains are important for exchange activity (Inoue et al., 1995; Dibrov et al., 1998). Interestingly, multiple sequence alignment of NHE protein sequences from a wide range of organisms reveals that there are four highly conserved acidic residues within putative transmembrane domains, including E262 of NHE1, which aligns with E225 of Nhx1p (Figure 9A). We have substituted these four conserved acidic amino acids for uncharged, polar residues in Nhx1p by site-directed mutagenesis: aspartic acid (D) 201 was changed to asparagine (N), glutamic acid (E) 225 to glutamine (Q), D230 to N, and E355 to Q. Each of these mutations was made in NHX1-HA (see above) carried on a CEN-based plasmid. These plasmids were then transformed into nhx1 yeast to test for protein expression. Western blotting showed that all four of these mutant Nhx1p-HA proteins were full length and expressed at levels indistinguishable from wild-type Nhx1p-HA (Figure 9B), indicating that these mutations do not affect Nhx1p stability.

View larger version (65K): [in this window] [in a new window]   Figure 9.   Mutation of conserved acidic residues in Nhx1p. (A) Partial sequence alignment of NHE protein sequences. NHE proteins with homology to Nhx1p have been found in many organisms. Those shown here are baker's yeast (S. cerevisiae) Nhx1p; fission yeast (S. pombe) NHE; thale cress (A. thaliana) NHX1; rice (Oryza sativa) NHX1; fruit fly (Drosophila melanogaster) NHE1; human (Homo sapiens) NHE proteins 1, 2, 3, 5, and 6; and rat (Rattus norvegicus) NHE4. The alignment was generated by using complete amino acid sequences, though only predicted transmembrane domains 5, 6, and 9 are shown. Acidic residues conserved throughout the NHE family (D201, E225, and D230 in Nhx1p) are in bold type. (B) NHX1-HA point mutant constructs make full-length proteins, expressed to wild-type levels. Whole-cell lysates were prepared from nhx1 cells (KEBY10) transformed with pRS316 (empty plasmid), or pRS316 with wild-type or mutant forms of NHX1-HA (pKEB38, 44, 45, 46, 47, pFP1, or pFP2). Fifteen micrograms of total protein was loaded per lane. Although E225Q appears to express slightly less Nhx1p-HA than the other strains, this was not significant over several experiments. In addition, this strain has the same point mutation as the D201N E225Q mutant, which shows wild-type expression levels in this experiment. (C) Localization of Nhx1p-HA. Rabbit polyclonal antibodies against the HA epitope were used to immunolocalize Nhx1p-HA in wild-type cells (WT; KEBY11 with pKEB38), or vps27 cells (vps27; KEBY12 with pKEB38). Cells were costained by using a monoclonal antibody to Pep12p. (D) Localization of Nhx1p-HA point mutants. vps27 cells were transformed with pRS316 containing wild-type or mutant forms of NHX1-HA (pKEB38, 44, 45, 46, or 47). Nhx1p-HA proteins were immunolocalized with rabbit polyclonal antibodies against the HA epitope. The images in C and D are confocal micrographs.

As shown in Figure 9C, Nhx1p-HA in wild-type cells is localized to a small punctate structure typical of the PVC (Pep12p staining is shown for comparison). To produce these images, cells containing one copy of NHX1-HA integrated into the genome were transformed with NHX1-HA on a CEN plasmid, thus producing 2-3 copies of NHX1-HA per cell. This level of expression allowed us to see the PVC localization of the protein. Overexpression of Nhx1p in our strains (from a 2 µ plasmid with the endogenous NHX1 promoter) led to retention of Nhx1p-HA in the ER (Bowers and Stevens, unpublished results). Also shown in Figure 9C is the localization of Nhx1p-HA to the class E compartment of vps27 cells, where it colocalizes with Pep12p. The vps27 mutant background allows concentration of Nhx1p in the aberrant PVC of these cells, where it can be easily visualized by immunofluorescence. We thus transformed vps27 cells with wild-type or mutant forms of NHX1-HA. As shown in Figure 9D, the D201N, E225Q, D230N, and E355Q point mutant forms of Nhx1p-HA all correctly localized to the class E compartment.

As shown in Figure 10A, wild-type cells secreted ~5% of newly synthesized CPY into the extracellular medium, as did nhx1 cells with the complementing NHX1-HA plasmid. nhx1 cells expressing the E355Q mutant of Nhx1p also secreted around 5% CPY, consistent with this mutation having no effect on the function of Nhx1p in vacuolar trafficking. In contrast, the D201N, E225Q, and D230N mutant proteins were unable to complement the CPY secretion of nhx1 cells. nhx1 cells expressing these mutant proteins did not secrete as much newly synthesized CPY as nhx1 cells with empty plasmid, but secreted significantly more than wild-type cells (Figure 10A). These results suggest that Nhx1p with either a D201N, E225Q, or D230N mutation is unable to function normally in protein trafficking of CPY to the yeast vacuole. To see whether the effect of the point mutations was additive, we constructed double mutants of D201N and E225Q, and D201N and D230N. These double mutants were expressed to wild-type levels in nhx1 cells (Figure 9B), and had CPY secretion levels that were not significantly different from those seen for the D201N, E225Q, or D230N single mutants (Figure 10A).

View larger version (91K): [in this window] [in a new window]   Figure 10.   Phenotype of nhx1 cells expressing Nhx1p-HA point mutants. The strains used in these experiments were wild-type (WT; SEY6210 with empty plasmid), nhx1 (KEBY10 with empty plasmid), nhx1 with NHX1-HA on a CEN plasmid (NHX1-HA; KEBY10 with pKEB38), and nhx1 with NHX1-HA point mutants on CEN plasmids (D201N, E225Q, D230N, E355Q, D201N E225Q, and D201N D230N; KEBY10 with pKEB44, pKEB45, pKEB46, pKEB47, pFP1, and pFP2, respectively). (A) CPY secretion. Cells were labeled in [35S]cysteine and methionine for 10 min and incubated in excess nonradioactive cysteine and methionine for 40 min. CPY was immunoprecipitated as described in Figure 2 and the MATERIALS AND METHODS. The gel was exposed to a phosphoimager screen, and the amount of CPY secreted into the extracellular medium was quantified as the percentage of total CPY in the sample. Each bar on the graph represents the mean percentage of CPY secreted over five separate experiments for WT, nhx1, and NHX1-HA; three separate experiments for D201N, E225Q, D230N, and E355Q; and two separate experiments for D201N E225Q and D201N D230N. The numbers above each bar are the mean averages of CPY secreted over all experiments, and the error bars represent SDs from this mean. (B) FM4-64 staining. Cells were stained with FM4-64 as described in Figure 5. Fluorescence images are shown, as well as differential interference contrast (DIC) images for the same cells.

As a further test of the function of Nhx1p point mutants in membrane trafficking, we investigated the morphological phenotype of these cells. nhx1 cells expressing the point mutant forms of Nhx1p-HA were analyzed by fluorescence microscopy using the dye FM4-64. As described in Figure 5, FM4-64 accumulates in the class E-like compartment of nhx1 cells. Consistent with the CPY secretion data (Figure 10A), nhx1 cells expressing the D201N, E225Q, and D230N mutant forms of Nhx1p-HA accumulated FM4-64 in a large, aberrant structure next to the vacuole (Figure 10B). In cells expressing either wild-type Nhx1p-HA or the E355Q mutant FM4-64 was seen exclusively on the vacuole membrane. As expected from our previous data (Figures 3, 4, and 5), Pep12p and Vph1p also accumulated in the class E-like compartment of cells expressing the D201N, E225Q, or D230N mutant forms of Nhx1p-HA, but showed wild-type staining patterns in cells expressing wild-type Nhx1p-HA or the E355Q mutant (Bowers and Stevens, unpublished results). Thus, we have created point mutant forms of Nhx1p that are expressed and localized normally but fail to function in the trafficking of proteins to the vacuole.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We report the identification of the yeast vacuolar protein sorting gene VPS44 as NHX1. The NHX1 gene encodes a sodium/proton exchanger, Nhx1p, of the NHE family. Nhx1p is localized to the PVC, and our data indicate that Nhx1p is required for protein trafficking out of this compartment.

Three Acidic Resides Are Essential for the Function of Nhx1p in Protein Trafficking

Of all the NHE proteins identified so far, Nhx1p is most closely related (50% identical) to a putative protein from the yeast Schizosaccharomyces pombe. NHE proteins are also found in plants, insects, amphibians, crustaceans, and mammals. The yeast Nhx1p sequence is 21-24% identical to mammalian NHE proteins 1-5, and 31% identical (60% similar) to human NHE6. The highest sequence identities are found within the putative transmembrane domains, suggesting an important role for these regions in the function of NHE proteins.

We have identified four acidic residues within putative transmembrane domains of NHE proteins that are conserved across all species (Figure 9A). Mutation of these residues in Nhx1p to uncharged, polar residues by site-directed mutagenesis results in proteins that are expressed to wild-type levels and localized correctly to the PVC (Figure 9). However, three of these mutant proteins are unable to function in protein transport out of the PVC, as measured by the secretion of CPY and the accumulation of the endocytic dye FM4-64 in an aberrant prevacuole structure (Figure 10). Although we have not measured sodium/proton exchange activity directly, there is some evidence that the acidic resides we mutated in Nhx1p may be important for ion exchange. By sequence alignment with other NHE proteins, one of these residues (E225) is equivalent to E262 of human NHE1. Interestingly, when E262 of NHE1 is mutagenized to I, the protein is expressed correctly and able to dimerize, but shows no detectable sodium/proton exchange activity (Fafournoux et al., 1994). Our results are also consistent with the idea that acidic residues in transmembrane domains 5 and 6 of the NHE proteins constitute ion binding sites of these exchangers (Counillon and Pouyssegur, 2000). Our results suggest that Nhx1p does not act as a scaffold for other proteins at the PVC, but rather that the ion exchange activity of Nhx1p may be essential for correct vacuolar trafficking in yeast. However, further experiments are needed to link the protein trafficking phenotypes of the nhx1 cells to the sodium/proton exchange activity of Nhx1p.

Nhx1p Is Required for Trafficking out of the PVC

Our results show that yeast cells lacking the sodium/proton exchanger, Nhx1p, secrete ~35% of newly synthesized CPY. nhx1 cells form an aberrant PVC structure next to the vacuole in which late Golgi, PVC, and vacuolar proteins accumulate. In addition, we have shown that proteins trapped in this aberrant PVC are subject to abnormal proteolytic cleavage, consistent with the PVC containing active proteases (Raymond et al., 1992; Piper et al., 1995). Compared with vps27 cells, nhx1 cells have a similar class E compartment next to the vacuole, but less of each marker protein appears trapped in this compartment. For example, although Vph1p accumulates in the aberrant PVC of nhx1 cells, some is also seen on the vacuole membrane (Figure 3). The morphological phenotypes are consistent with the lower levels of CPY secretion, and slower kinetics of Vps10p and A-ALP cleavage in nhx1 cells compared with vps27 cells (Figures 2 and 6). Thus, our results show that the phenotype of nhx1 cells is similar, though somewhat weaker, to that seen for vps27 cells. VPS27 belongs to the class E subset of VPS genes (Raymond et al., 1992; Piper et al., 1995). Thus, Nhx1p is likely to act at the same step in trafficking to the vacuole as the other class E Vps proteins (Figure 11A). Class E Vps proteins are thought to control protein trafficking out of the PVC (to the late Golgi, and to the vacuole; Piper et al., 1995; Rieder et al., 1996; Finken-Eigen et al., 1997). In addition, recent reports suggest that class E Vps proteins are essential for the formation of multivesicular bodies, a process that is also dependent on the phosphatidylinositol 3-phosphate 5-kinase Fab1p (Babst et al., 1997, 1998).

View larger version (28K): [in this window] [in a new window]   Figure 11.   Function of Nhx1p in yeast protein trafficking through the PVC. (Ai) Protein trafficking pathways from the late Golgi to the vacuole in wild-type cells. Proteins can reach the vacuole via several different routes. Proteins such as CPY follow a pathway to the vacuole that includes transport through an endosomal PVC. The ALP pathway allows some proteins to reach the vacuole via an alternate route, bypassing the PVC. Proteins can also reach the vacuole following endocytosis from the plasma membrane. The endocytic and CPY pathways converge at or before the PVC. (Aii) Nhx1p functions to mediate exit from the PVC. In nhx1 cells, exit from the PVC to the Golgi and to the vacuole is inhibited (represented by dashed arrows). (B) Model for the function of Nhx1p in protein trafficking. A specific environment inside the PVC is essential for protein trafficking. This environment is dependent upon the activity of Nhx1p. The lumenal ion concentration and pH are monitored by a transmembrane protein or sensor, which transmits this information to the cytosolic face of the membrane. The binding of cytosolic proteins to the PVC membrane is dependent on the correct intralumenal environment as determined by the sensor. Candidates for these cytosolic factors include the class E proteins/complexes that control exit from the PVC (vesicular budding into the cytosol) and multivesicular body (MVB) formation (vesicular budding into the PVC lumen).

Thirteen other class E VPS genes have been identified to date: REN1 (VPS2), VPS4 (END13), VPS20, VPS22, VPS23 (STP22), VPS24, VPS25, VPS27, VPS28, VPS31, SNF7 (VPS32), VPS36 (VAC3), and VPS37 (Raymond et al., 1992; Davis et al., 1993; Tu et al., 1993; Piper et al., 1995; Nothwehr et al., 1996; Rieder et al., 1996; Babst et al., 1997, 1998, 2000; Finken-Eigen et al., 1997; Luo and Chang, 1997; Li et al., 1999). Although several of the class E VPS genes have yet to be cloned, those that have been identified encode cytosolic or peripheral membrane proteins. This has led to the hypothesis that the class E proteins associate transiently with the PVC membrane and function (perhaps as complexes) in trafficking out of the PVC and in multivesicular body formation (Babst et al., 1998). To support this idea, recent studies have shown that both Vps24p and Snf7p transiently associate with the PVC membrane. Vps24p and Snf7p require the activity of an AAA-type ATPase, Vps4p, for dissociation from the membrane (Babst et al., 1998). Vps23p and Vps28p have also been demonstrated to be part of a cytosolic protein complex (Babst et al., 2000).

We hypothesize that the sodium/proton exchange activity of Nhx1p is important for its role in protein trafficking. There is evidence from studies of mammalian cells that the internal environment of endosomes is critical for the binding of cytosolic factors to the membrane. The binding of a subset of COP1 coat proteins to early endosome membranes is dependent on the pH within the organelle (Aniento et al., 1996; Gu et al., 1997). This has led to the hypothesis that an unknown transmembrane protein acts as a pH sensor, transmitting information about the internal environment across the membrane to the cytosolic face (Gu and Gruenberg, 1999).

Given that nhx1 cells show a class E Vps phenotype, we suggest that sodium/proton exchange regulates the environment inside the PVC, and that this internal environment determines the binding of other, cytosolic proteins to the PVC membrane (Figure 11B). These cytosolic proteins may control vesicle budding from the PVC, multivesicular body formation (budding of membrane into the PVC), and/or the sorting of membrane proteins at this compartment. Some amount of membrane association of these cytosolic factors to the PVC membrane, even in the absence of Nhx1p may account for the less severe class E phenotype of nhx1 cells compared with deletion of other class E genes. Possible candidates for association with the PVC in a Nhx1p-dependent manner include proteins that appear to act at the same transport step such as the class E Vps proteins (or complexes of proteins).

We show that Nhx1p plays a role in endosomal protein transport in yeast. Is this role conserved in higher organisms? Of the six NHE proteins identified in mammalian cells, NHE6 shows higher sequence homology than the others to Nhx1p. NHE6 is also the only mammalian NHE to fall into the subset of intracellular NHE proteins identified by phylogenetic analysis (Fukuda et al., 1999). However, NHE6 has been localized to the mitochondrial inner membrane, suggesting that it is unlikely to be the Nhx1p orthologue (Numata et al., 1998). The identity of a mammalian Nhx1p orthologue localized to endosomes therefore remains unclear, but it is possible that the plasma membrane NHE proteins undergo endocytosis and remain active in endosomal compartments (D'Souza et al., 1998). Several other species, such as fruit fly, rice, and thale cress have intracellular NHE proteins with homology to Nhx1p (Figure 9). It will be interesting to see whether these NHE proteins also prove to be involved in intracellular protein trafficking. Of the other yeast class E Vps proteins, Vps23p, Vps27p, and Vps4p are known to have mammalian orthologues (TSG101, Hrs, and SKD1, respectively), suggesting that their functions are conserved (Komada et al., 1997; Li et al., 1999; Scheuring et al., 1999). This is supported by experimental evidence that suggests Tsg101, Hrs, and SKD1 play roles in transport through the endosomal system (Komada et al., 1997; Komada and Soriano, 1999; Babst et al., 2000; Bishop and Woodman, 2000; Yoshimori et al., 2000). Thus, it seems that the functions of the yeast class E Vps proteins are conserved in higher organisms.

Nhx1p Can Function in the Absence of the V-ATPase or the Chloride Channel Gef1p

Previous studies have proposed that Nhx1p sequesters sodium ions into the PVC in exchange for protons, by using the proton gradient generated by the V-ATPase (Nass et al., 1997; Nass and Rao, 1998, 1999; Gaxiola et al., 1999; Quintero et al., 2000; Figure 7A). In addition, a putative voltage-gated chloride channel of the CLC family (Gef1p) is proposed to transport chloride ions into the PVC (Gaxiola et al., 1999; Figure 7A). Gef1p is thought to neutralize the membrane potential generated by ion transporters such as the V-ATPase and Nhx1p, to allow further acidification of the PVC (Gaxiola et al., 1998, 1999; Schwappach et al., 1998). Thus, activity of Nhx1p (which presumably requires a proton gradient for function) may depend on the activities of the V-ATPase and Gef1p.

Our data indicate that the chloride channel Gef1p is not required for protein transport to the yeast vacuole because cells lacking Gef1p do not secrete significant amounts of CPY, or show abnormal processing of Vps10p. Furthermore, cells lacking Gef1p showed no characteristics of the vps mutant classes by immunofluorescent localization of Pep12p, Vph1p, Vma2p, or Vps10p, or by uptake of the endocytic tracer FM4-64 (Levi, Bowers, and Stevens, unpublished results). Cells lacking Gef1p therefore do not have the same phenotype as cells lacking Nhx1p.

We find that loss of V-ATPase activity by the deletion of a gene encoding either a V₁ subunit (Vma2p), or a V₀ subunit (Vma3p), results in less secretion of CPY than loss of Nhx1p. In addition, cells lacking the V-ATPase (unlike cells lacking Nhx1p) do not show abnormal cleavage of Vps10p, and do not accumulate the aberrant PVC structure typical of class E vps mutants (Figures 7C and 8). We have also constructed a yeast strain lacking both the V-ATPase and Nhx1p (vma2 nhx1). This double deletion strain shows a CPY secretion phenotype, Vps10p proteolytic cleavage, and morphological phenotype identical to that seen for nhx1 cells (Figures 7C and 8; Bowers and Stevens, unpublished results). Our results establish that loss of V-ATPase function produces a less severe vacuolar trafficking phenotype than the loss of Nhx1p. Because loss of Nhx1p either in the presence (nhx1 cells) or absence (nhx1 vma2 cells) of the V-ATPase produces a similar protein trafficking phenotype, this suggests that Nhx1p can function even in the absence of the V-ATPase. The nhx1 vma2 results also argue against the lack of Vps10p cleavage in the vma2 cells being due to lower protease activity in the PVC of cells lacking an active V-ATPase.

Experiments in mammalian cells have shown that the acidification of endosomes is mediated by the V-ATPase, and is required for protein trafficking in the endocytic system (Johnson et al., 1993; Clague et al., 1994; van Weert et al., 1995; Aniento et al., 1996). Studies on yeast protein trafficking in the absence of V-ATPase function, however, have led to conflicting results (Nelson and Nelson, 1990; Yamashiro et al., 1990; Klionsky et al., 1992a,b; Yaver et al., 1993; Morano and Klionsky, 1994). Our results, comparing cells lacking Nhx1p to cells lacking the V-ATPase, indicate that a sodium/proton exchanger can function in the absence of the V-ATPase. This leads us to hypothesize that yeast are able to acidify the PVC even in cells lacking the V-ATPase. What then functions to acidify the PVC of these yeast cells?

Cells devoid of a functional V-ATPase are unable to grow in medium buffered to pH 7.5. However, they are able to survive at pH 5.0, suggesting that in low pH medium the essential function of the V-ATPase is bypassed. The cellular basis for this phenotype is unclear, although some have suggested that acidification of an intracellular compartment is essential for yeast cell viability (Nelson et al., 2000). It has been suggested that fluid phase endocytosis might deliver protons from the external medium to endosomes, allowing cell survival (Nelson and Nelson, 1990; Munn and Riezman, 1994). However, a recent study by Plant et al. (1999) argues against this idea because blocking endocytosis in vma mutants does not prevent acidification of the vacuole in low pH medium. It is also possible that mislocalization of the plasma membrane P-type ATPase, Pma1p, in cells lacking the V-ATPase may allow growth at low pH (Nelson et al., 2000). It has also been suggested that transport of ammonium ions may allow equilibration of the pH across the plasma membrane and intracellular membranes under acidic growth conditions (Plant et al., 1999). Our results are consistent with the idea that under low pH growth conditions, the PVC of yeast cells lacking the V-ATPase still maintains a proton gradient with respect to the cytosol. This proton gradient, established even in the absence of the V-ATPase, would allow Nhx1p to exchange sodium ions for protons across the PVC membrane, a function critical for normal vacuolar protein sorting.