Yeast Mon2p is a highly conserved protein that functions in the cytoplasm-to-vacuole transport pathway and is required for Golgi homeostasis

The late Golgi and endosomal compartments are the central protein sorting stations of the cell. They direct newly synthesized as well as endocytosed proteins to their final subcellular destinations. Transport of proteins and lipids between these specialized trafficking organelles is highly regulated. It requires sequential recruitment of cytosolic proteins that function in vesicle formation, motor recruitment and vesicle tethering (Munro, 2002). This recruitment process is largely mediated by specific lipid modifications and the activation of small G proteins of the ADP-ribosylation factor (ARF) and Rab families. The Arf proteins are distinguished from the other small GTPases by the presence of a myristoylated N-terminal amphipathic helix that facilitates interaction with membranes (Nie et al., 2003). Like other small GTPases, Arf proteins cycle between the inactive GDP-bound form, which is soluble and an active GTP-bound form, which is associated with membranes and selectively interacts with effectors (reviewed by Vetter and Wittinghofer, 2001). The stimulation of this cycle is mediated by the subsequent action of GTPase-activating proteins (GAPs) and guanine exchange factors (GEFs). The Arf GEFs represent a large and diverse protein family (Jackson and Casanova, 2000), sharing a conserved region of roughly 200 amino acids, termed the Sec7 domain. This Sec7 domain alone is sufficient for the GDP to GTP exchange activity (Chardin et al., 1996).

Mon2p/Ysl2p, a protein related to the Sec7 family, was identified in several independent genetic screens searching for mutants that are sensitive to the drugs monensin and brefeldin A (Muren et al., 2001), defective in endocytosis (Wiederkehr et al., 2001), defective in protein transport to the vacuole (Avaro et al., 2002; Bonangelino et al., 2002), or synthetic lethal with Δypt51, a mutant that perturbs vesicular transport between the late Golgi and endosomal compartments (Singer-Kruger and Ferro-Novick, 1997). Recently, Mon2p was proposed to be a GEF for the small GTPase Arl1p as the two proteins physically interact and Δmon2 and Δarl1 mutants exhibit similar endocytic transport and vacuolar protein sorting defects (Jochum et al., 2002). Although Arl1p shares structural features with Arf proteins (Pasqualato et al., 2002), its precise function in membrane trafficking remains poorly established. Arllp is associated with the Golgi apparatus (Lu et al., 2001; Rosenwald et al., 2002) and the active Arl1-GTP form directly interacts with GRIP domains of several golgins, thus mediating their recruitment onto late-Golgi membranes (Lu and Hong, 2003; Panic et al., 2003a; Panic et al., 2003b; Setty et al., 2003). Golgins are long coiled-coil proteins that maintain the structure of the Golgi complex and are necessary for vesicular tethering events (Barr and Short, 2003). Arl1p was also purified with the GARP complex, which is part of the tethering complex required for the fusion of endosome-derived vesicles at the late Golgi (Panic et al., 2003b).

Mon2p also binds to Neo1p, a member of the putative aminophospholipid translocases (APTs) (Wicky et al., 2004). APTs couple ATP hydrolysis to the translocation of phosphatidylserine (PS) and phosphatidylethanolamine (PE) to the cytosolic leaflet of biological membranes (Balasubramanian and Schroit, 2003; Natarajan et al., 2004). A higher concentration of these lipids in this leaflet is believed to induce the recruitment of specific proteins or help to deform membranes during vesicle budding. The NEO1 gene was originally isolated as a multicopy suppressor of neomycin sensitivity (Prezant et al., 1996) and performs an essential function that cannot be replaced by the other APT members. Neo1p is involved in membrane transport within the endosomal and Golgi systems (Hua and Graham, 2003; Wicky et al., 2004), a function that could be linked to that of Mon2p and Arl1p.

In this study we show that Mon2p contains six domains that are conserved in several higher eukaryotic homologs, thus forming a new family of proteins. We found that Mon2p is active without its putative Sec7 and Arl1p binding domains, suggesting that Mon2p could function independently of Arl1p. Furthermore, our results demonstrate that Mon2p localizes to the late Golgi/early endosomes, where it is involved in protein trafficking and maintenance of homeostasis. It is also required for the constitutive maturation of the hydrolase aminopeptidase I (Ape1), which relies on the cytoplasm-to-vacuole transport (Cvt) pathway and requires the GARP complex. Finally, we show a genetic and biochemical interaction of Mon2p with Dop1p, a protein that functions in cell polarity.

Yeast strains used in this study are shown in Table 1, and their construction is described below. Strains were grown in complete medium (yeast extract-peptone-dextrose, YPD) or synthetic dextrose (SD) growth medium (Sherman, 1991). Anti-Pep12p, monoclonal mouse 12CA5 anti-hemagglutinin (HA) and rabbit anti-GFP antibodies were purchased from Molecular Probes (Eugene, OR), Berkeley Antibody (Richmond, CA) and Clontech (Palo Alto, CA), respectively. The polyclonal antibodies against CPY and Ape1 were generous gifts of Randy Schekman, University of California, Berkeley, CA and Daniel Klionsky, University of Michigan, Ann Arbor, MI, respectively. The monoclonal anti-myc antibody (9E10) was purchased from the Developmental Studies Hybridoma Bank at the University of Iowa (Iowa City, IA). Chlorpromazine (C-8138) was obtained from Sigma (St Louis, MO).

The mon2Δ::ADE2 (YFP93 and YFP94) and MON2 (wild type/YFP95) were constructed by inserting the ADE2 gene into the internal NcoI/HpaI restriction sites of the MON2 gene in the wild-type diploid W303 strain and obtained after tetrad dissection, using standard yeast methods. The correct integration of the mon2Δ::ADE2 disruption cassette was confirmed by Southern analysis. To construct diploid double mutants with the mon2Δ::ADE2 allele, YFP93 was crossed with YFP71, CDK17-6B and YFP35 to give YFP72, YFP155 and YFP38, respectively. YFP94 was crossed with RSY906 to give YFP103. The resulting diploid strains were sporulated and tetrads were dissected. Spore clones containing disrupted genes were selected by testing for the presence of the corresponding auxotrophic markers. Authentic chc1-521^ts strains were identified based on the absence of growth at the restrictive temperature. The ODY318/Y23304 strain was obtained from the EUROSCARF consortium.

Deletion of MON2 in the SEY6210 strain background (AAY1120) and genomic integrations for C-terminal tagging (JEY001, JEY002, JEY003, and JEY069) were carried out by transformation of PCR-amplified constructs using the appropriate plasmids as templates (Longtine et al., 1998) and primers specific for MON2. Deletion and integrations were subsequently verified by PCR and western blot analysis (the latter only for fusion proteins). The diploid strain JEY026 was obtained by crossing JEY001 with JEY002. For the DOP1-3xHA-mRFP integration, a modified version of (Longtine) plasmid pFA6A-13Myc-HIS3-MX6 was used. Briefly, using PacI/AscI sites, the 13×MYC tag was replaced with a 3×HAmRFP (kind gift of R. Tsien, Howard Hughes Medical Institute, UCSD, La Jolla, CA) PCR product, and LEU2 (including 598 bp upstream and 46 bp downstream sequence) was substituted for the HIS3 ORF using BglII and PmeI sites.

The relevant plasmids used in this study are listed in Table 2 and they were constructed as briefly described below. A 5.89 kb SalI/SpeI fragment containing the entire MON2 open reading frame (ORF) was subcloned from a genomic library clone pDK7 into the SalI/XbaI-restricted YEplac181 (2 μ, LEU2), YEplac112 (2 μ, TRP1) and YEplac195 (2 μ, URA3) to form YEplac181-MON2 (pDK30), YEplac112-MON2 (pDK35) and YEplac195MON2 (pDK36), respectively. For the construction of GFP plasmids, the entire ORF of MON2[A-F (1-4911)], and all the 5′-DNA MON2[A-E (1-4500)], MON2[A-d (1-1800)], MON2[A-C (1-1230)], MON2[A-B (1-651)], MON2[A (1-180)] terminal fragments were fused upstream of a GFP-encoding fragment and cloned into the YEplac181 vector (2 μ, LEU2) under the control of their own promoter. MON2[A-d (1-1800)] was also subcloned into the pRS315 vector (CEN, LEU2) (pOD130). All the 3′-DNA MON2[B-F (151-4911)], MON2[d-F (2383-4911)], MON2[E-F (3443-4911)] terminal fragments were fused downstream of a GFP, under the control of the ADH promoter into the pRS415-ADH-GFP vector (CEN, LEU2).

YEplac181-DOP1 (pDK64) and pRS423-DOP1 (pDK62) were obtained by subcloning a 8.3 kb SalI/SacI fragment containing the DOP1 gene from pRS413-DOP1 (pDK15) into the SalI/SacI-restricted YEplac181 and pRS423, respectively. YEplac181ARL1 (pRB95) and YEplac112-ARL1 (pRB108) were constructed by subcloning a 1.25 kb SacI/BamHI fragment containing the ARL1 gene from pN5 into the SacI/BamHI-restricted YEplac181 and YEplac112, respectively. Finally, YEplac181-NEO1 (pRB111) and YEplac112-NEO1 (pRB113) were obtained by subcloning a SpeI (Klenow blunt-ended)/XbaI fragment containing the NEO1 gene from pA32 into the SmaI/XbaI-restricted YEplac181 and YEplac112, respectively. All cloned DNA fragments generated by PCR amplification were verified by sequencing.

Plasmids pCS198 and pCS212 have essentially been described previously (Stefan et al., 2002), except that GFP in the respective FYVE_EEA1 and PH_FAPP1 reporter constructs was replaced with DsRed using PCR amplification to introduce BglII and SpeI sites and subsequent (Klenow) blunt ligation.

YFP44 (Δmon2/chc1-521) cells harboring YEplac195-MON2 and transformed with YEplac181-GFP, YEplac181MON2[1-4911]-GFP, YEplac181mon2[1-4500]-GFP, pRS415-ADH-GFP-mon2[151-4911], pRS415-ADH-GFP-mon2[2388-4911] or pRS415-ADH-GFP-mon2[3443-4911] were first streaked out on 5-FOA-containing plates to select for the loss of YEplac195-MON2 at 22°C and then tested for growth on SD-LEU plates at the indicated temperatures. YFP93 (Δmon2) transformed with YEplac181-GFP, YEplac181MON2[1-4911]-GFP, YEplac181mon2[1-4500]-GFP, pRS415-ADH-GFP-mon2[151-4911], pRS415-ADH-GFP-mon2[2388-4911] or pRS415-ADH-GFP-mon2[3443-4911] were streaked on SDLEU plates containing 50 mM potassium phosphate (pH 7.0) and 20 μM CPZ as indicated.

To obtain multicopy suppressors of the Δmon2 null mutation, the CDK13-1A (MATα tif3Δ::TRP1 mon2Δ::ADE2 [pSEY18-TIF3]) strain was transformed with a YEp13-based yeast genomic library (Nasmyth library) or a YEplac181-based library (D.K., unpublished). Transformants containing plasmids with suppressor activity were selected by replica plating onto 5-FOA-containing plates and re-tested by transformation into YFP44 (Δmon2/chc1-521), YFP101 (Δmon2/Δvps1) and YFP86 (Δtif3/vps54) double mutants. The specific suppressing genes of YFP44 and YFP101 strains were determined by subcloning.

For most GFP images, cells harboring GFP fusion constructs were grown in SD-LEU medium to mid-logarithmic phase at 30°C. Images (100× magnification) were obtained using a Zeiss Axiovert S1002TV (Thornwood, NY) microscope equipped with an Axiocam color digital camera and the AxioVision™ software. Figures were prepared with the use of the Adobe Photoshop 8.0 (Adobe Systems, San Jose, CA) software program.

For Mon2-GFP colocalization experiments and those using GFP-Ape1, cells were visualized using fluorescein isothiocyanate (FITC) and rhodamine filters, and images were captured with a Photometrix camera and processed with DeltaVision deconvolution software (Applied Precision, Seattle, WA). For the GFP-Ape1 experiments, cells were first labeled with the fluorescent lipophilic dye N-[3-triethylammoniumpropyl]-4-[p-diethylaminophenylhexa-trienyl] pyridinium dibromide (FM4-64; Molecular Probes, Eugene, OR) as previously described to highlight vacuoles (Vida and Emr, 1995). All observations are based on the examination of at least 100 cells, and representative fields are shown.

Early log-phase cells (approximately 60 OD₆₀₀ units) were grown in YPD, harvested and fixed in 3% glutaraldehyde, 0.1 M sodium cacodylate (pH 7.4), 5 mM CaCl₂, 5 mM MgCl₂ and 2.5% sucrose for 1 hour. Cells were first washed in 100 mM Tris-HCl (pH 7.5), 25 mM DTT, 5 mM EDTA and 1.2 M sorbitol for 10 minutes, then resuspended in 100 mM K₂HPO₄ (pH 5.9), 100 mM citrate and 1 M sorbitol. Following addition of β-glucuronidase (50 μl) and Zymolyase T100 (400 μg/ml), cells were incubated for 40 minutes at 30°C. They were then spun down, washed and resuspended in cold buffer containing 500 mM sodium cacodylate (pH 6.8) and 25 mM CaCl₂, followed by osmium thiocarbohydrazide staining. Further processing details have been described previously (Rieder et al., 1996). Structural analyses were based on analysis of more than 50 cells for each strain.

The preparation of whole cell extracts for western blot analysis was made from cultures grown in SD medium to mid-logarithmic phase. Five OD₆₀₀ units of cells were converted to spheroplasts and lysed by the addition of 200 μl of 1% SDS/8 M urea at 65°C for 10 minutes.

For subcellular fractionations, cells (∼20 OD₆₀₀ units) were grown at 26°C to mid-logarithmic phase and converted to spheroplasts (Darsow et al., 1997). Spheroplasts were resuspended in 1 ml ice-cold lysis buffer (200 mM sorbitol, 50 mM potassium acetate, 20 mM HEPES, pH 7.2, 2 mM EDTA) containing protease inhibitors, and lysed with 12 strokes in a Dounce homogenizer. Following a clearing spin at 500 g for 5 minutes, lysates were spun at 13,000 g for 10 minutes (4°C), thereafter the supernatant was treated as indicated for 15 minutes with frequent agitation. Finally, the supernatant was further fractionated at 100,000 g for 1 hour (4°C), the S100 supernatant was removed and the resulting P100 pellet was resuspended in lysis buffer. Proteins were precipitated from both fractions with trichloroacetic acid (TCA; 10% final concentration). SDS-PAGE and western analyses were used to determine the relative amount of Dop1-HA-mRFP present in each fraction.

For coimmunoprecipitations, cells (50 OD₆₀₀ units) were lysed as above, and Tween 20 was added to the 13,000 g supernatants to a final concentration of 0.5%. Following a 15-minute incubation, detergent-insoluble material was pelleted with a 10-minute 16,000 g spin. After removing and TCA-precipitating 10% of the supernatant to determine total protein content, antibodies to the protein of interest were added to the remainder and incubated at 4°C for 4 hours with 20 μl Gammabind G-sepharose beads (Amersham Biosciences, Piscataway, NJ) added 2 hours into the incubation. Protein complexes bound to the beads were recovered by washing three times with 1 ml ice-cold lysis buffer containing 0.5% Tween-20, and three times with detergent-free lysis buffer, followed by elution in boiling buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 5% β-mercaptoethanol, 10% glycerol, 0.005% Bromophenol Blue) at 100°C for 10 minutes. As above, SDS-PAGE and western analysis were used to detect proteins of interest.

For sucrose density gradient fractionation, cells transformed with the indicated plasmids were grown in 500 ml SD medium to an OD₆₀₀ of 1.2 at 30°C. Cells were subsequently lysed as described (Harsay and Bretscher, 1995). The yeast lysate was centrifuged for 20 minutes at 4°C at 13,000 g in a SS34 rotor (Sorvall). The pellet was discarded and the supernatant was centrifuged for 60 minutes at 4°C at 100,000 g in a SW41 rotor (Beckman Instruments) onto an 80-μl cushion of 60% sucrose. The membrane pellet was carefully resuspended in 2 ml buffer containing 0.8 M sorbitol, 10 mM triethanolamine (pH 7.2), 1 mM EDTA, 0.5 mM PMSF, 1 μg/ml pepstatin, 1 μg/ml leupeptin and 1 mM benzamidine. The resuspended pellet was then layered onto a 10.5-ml sucrose step gradient (1 ml 60% sucrose, 2 ml each at 42, 36, 24% sucrose and 1.5 ml of 18% sucrose) buffered with 0.8 M sorbitol, 10 mM triethanolamine (pH 7.2) and 1 mM EDTA. After centrifugation for 18 hours at 4°C at 100,000 g in an SW41 rotor, fractions (0.4 ml) were manually collected from the top and equal volumes of each fraction were processed for immunoblotting.

In gel filtration chromatography experiments, approximately 100 OD₆₀₀ units of cells were grown, converted to spheroplasts and lysed as described above (for subcellular fractionations), except that PBS (pH 7.2) was used as the lysis buffer. The S100 supernatants were run over a Sephacryl S-300 16/60 column (Pharmacia) in PBS. Fractions of 1.4 ml were eluted at a flow rate of 0.4 ml/minute and, after TCA precipitation, one-tenth of each fraction was analyzed by SDS-PAGE and western blot. Sizing standards for the column were blue dextran, ferritin, catalase and thyroglobulin.

Cell labeling and immunoprecipitations were performed as described previously (Gaynor et al., 1994). Briefly, mid-log phase (OD₆₀₀∼0.6) cultures were concentrated to 3 OD₆₀₀ units/ml and labeled with 3 μl Tran ³⁵S label per OD₆₀₀ (PerkinElmer Life and Analytical Sciences, Boston, MA) for 10 minutes in SD-URA medium. Cells were chased with 10 mM methionine, 4 mM cysteine and 0.4% yeast extract for the indicated times. Proteins were then precipitated in 10% TCA. Resulting protein pellets were washed twice with ice-cold acetone, dried and processed for immunoprecipitation as described previously (Gaynor et al., 1994). Labeling and immunoprecipitation of CPY was performed as described (Wilcox and Fuller, 1991). Immunoprecipitated proteins were resolved on SDS-PAGE gels and analyzed by autoradiography.

Iterative database searches using profiles were performed on the non-redundant database Swiss-Prot/TrEMBL for detection of Mon2p homologues. The default parameters of the pftools and the PSI-BLAST packages were used for the construction of profiles. The graphical representation of Mon2p was adapted from the PROSITE domain visualizer (http://www.expasy.org/tools/scanprosite/psviewdoc.html). The alignment was shaded in conservation mode by using GeneDoc (version 2.6) according to the amino acid property. The secondary structure elements were predicted using the PHD program (Rost, 1996).

We originally isolated a mon2 mutant in a synthetic lethal screen with a mutant of TIF3, which encodes a translation initiation factor (Coppolecchia et al., 1993). This genetic interaction helped us to establish a direct connection between protein synthesis and membrane trafficking (Deloche et al., 2004). Mon2p is a large protein of 186 kDa, with weak homology to the large Sec7 GEFs. Initial sequence analysis revealed that this homology is not restricted to the catalytic Sec7 domain but applies to other domains of unknown functions. To define the structural properties of these different domains, we performed iterative PSI-BLAST (Altschul et al., 1997) and generalized profile searches (Bucher et al., 1996) on the Swiss-Prot/TrEMBL non-redundant database (Bucher et al., 1996). We initiated our search with six conserved regions (A to F) that are clearly evident when aligning Mon2p with the various fungal homologues (Fig. 1A). Only sequences that matched a profile with a significant score (E-value<0.01) were used for subsequent iteration cycles. After several cycles, our different profiles converged on two sets of sequences, shown in the different multiple sequence alignments (see supplementary material Figs S1-S3). Profiles directed against domains B, C and D revealed that these three domains of Mon2p are also present in the large Sec7 GEF proteins and closely correspond to the conserved upstream and downstream regions of the Sec7 catalytic domain, as recently defined (Mouratou et al., 2005) (see Fig. 1B). Among these three domains, the B and C domains display important sequence conservation (41% and 40% similarity, respectively). A secondary structure prediction found ten α-helices in the C domain present in a region that was shown to interact with Arl1p (Jochum et al., 2002). As previously noted (Jochum et al., 2002), the number of these predicted α-helices suggests a possible structural similarity between the C domain and the Sec7 domains of GEFs (Cherfils et al., 1998). However, it is unclear whether the C domain is capable of nucleotide exchange as Mon2p does not bind to the GDP-bound form of Arl1p with higher affinity (Jochum et al., 2002). Moreover, Mon2p is not essential for stabilization of Arl1p on membranes (Panic et al., 2003b). Finally, a 100% conserved Y-D motif was detected in a predicted loop between the last two α-helices (Fig. 1A and supplementary material Fig. S2B), a region that was previously described as being present in all members of the large Sec7 GEFs (Jackson and Casanova, 2000). Despite this homology, we found that Mon2p differs from the Sec7 GEF family in the following important aspects. First, Mon2p does not possess the typical Sec7 catalytic domain, and second, it contains three extra domains (A, E and F). These three domains are found associated with domain B, C, and D in a subgroup of well conserved proteins present in all eukaryotic kingdoms (Fig. 1A).

The identification of six conserved sequence regions suggests that Mon2p and its homologues function in a similar cellular process, and might interact with other highly conserved proteins. The alignments of the A and F domains show a sequence similarity of 42% and 40%, respectively, with a predominance of charged and hydrophobic residues (supplementary material Fig. S1). The type of conserved amino acids, the small size, and the predicted secondary structure for these two domains suggest that they could be involved in protein binding activity. The E domain is remotely, if at all, conserved and is absent from all identified homologues.

To assess the functions of the distinct domains of Mon2p, a series of N- and C-terminal deletions was generated by PCR and fused to a GFP reporter under the control of its own promoter in a 2 μ plasmid or a constitutive ADH promoter in a centromeric plasmid (Fig. 2A). The different constructs were transformed into the Δmon2 strain, and the level of protein expression was verified by western blot analysis. Each fusion protein migrates with the expected molecular mass, although the levels of expression are not uniform (Fig. 2B).

We next determined the minimal construct capable of suppressing the growth defect of the Δmon2 mutant. For this purpose, we used the YFP44 (Δmon2/chc1-521) strain, which does not grow at 30°C. As a positive control, we first showed that the expression of the entire Mon2[A-F]p-GFP protein restores growth of the YFP44 mutant at 30°C (Fig. 2A). Subsequently, we examined the activity of the newly defined conserved A and F domains in the Mon2 protein family. The overproduction of a Mon2p construct lacking the A domain showed a normal growth complementation profile. By contrast, the deletion of only the F domain (MON2[A-E]) led to reduced restoration of cell growth (Fig. 2A), and larger deletions of the C-terminal part of Mon2p resulted in a complete loss of activity. In agreement with these results, we showed that overexpression of the C-terminal fragment containing only the E and F domains is sufficient to complement the growth defect of YFP44 cells at 25°C (data not shown), and partially at 30°C (Fig. 2A).

To confirm the presence of an active domain in the C-terminal region of Mon2p, we next used chlorpromazine (CPZ) as an agent to block the growth of the Δmon2 mutant. CPZ is a permeable cationic amphipathic molecule that changes the lateral organization of cellular membranes (Jutila et al., 2001) and interacts with negatively charged lipids (Chen et al., 2003). We previously observed that CPZ modifies internal membrane structures, making Golgi mutants such as Δvps1 and Δvps54 highly sensitive to CPZ (unpublished data). By analogy, we reasoned that a CPZ treatment would also be lethal to the Δmon2 mutant. Only Δmon2 cells overexpressing Mon2[A-F]p-GFP and Mon2[B-F]p-GFP are resistant to 20 μM CPZ (Fig. 2C). Cells overexpressing Mon2[A-E]p-GFP remain sensitive, further supporting the conclusion that the F domain is essential for cell growth under membrane stress. Finally, in contrast to the growth complementation study in the YFP44 strain, the Mon2[E-F]p-GFP and Mon2[d-F]p-GFP are not sufficient to completely suppress the CPZ sensitivity and alleviate membrane transport defects, as judged by the presence of fragmented vacuoles (data not shown).

Sequence analyses revealed that the N-terminal part of Mon2p has significant homology to the large Sec7 GEFs (yeast Sec7, BIG1 and BIG2) (Jochum et al., 2002). BIG1 and BIG2 were originally isolated as parts of large, cytosolic, macromolecular complexes (Morinaga et al., 1997; Togawa et al., 1999). However, these two proteins have been shown to interact with Golgi membranes via their N termini containing the B and almost the entire C domains (Mansour et al., 1999; Yamaji et al., 2000). This prompted us to reexamine the localization of Mon2p as its previous localization was restricted to endocytic elements (Jochum et al., 2002). We first analyzed the subcellular localization of Mon2p by equilibrium sedimentation on a sucrose gradient. A whole-cell extract from wild-type cells harboring a functional MON2-GFP allele was first subjected to differential centrifugation.

Then, the enriched Mon2p-GFP high-speed pellet (P100) was loaded on a sucrose gradient and centrifuged at 100,000 g for 18 hours. Fractions were collected and analyzed by western blotting. Mon2p-GFP was detected in fractions containing Kex2p and Pep12p, markers of the late Golgi/early endosome and the late endosomal compartment, respectively (Fig. 3A). Second, we examined the cellular localization of Mon2p by fluorescence microscopy in wild-type cells. We detected Mon2p-GFP on multiple punctate structures, where it partially colocalized with two specific markers of the late Golgi: a Sec7p-DsRed chimera and a PH_(FAPP1)-DsRed fusion that localizes to membranes enriched in phosphatidylinositol 4-phosphate (Fig. 3B). By contrast, no overlap was observed between Mon2p-GFP and DsRed-FYVE, a phosphatidylinositol 3-phosphate probe localizing to late endosomes. Altogether, our results indicate that Mon2p predominantly resides in the late Golgi.

We next defined the minimal domains required for Mon2p to interact with late-Golgi/early endosomal membranes. Various N-terminal fragments still displayed a punctate pattern (Fig. 4A). In fact, our fluorescence experiments indicate that the A and B domains are sufficient to bind membranes, although an increase in the number of dots is observed, revealing perhaps a lower binding specificity (see below). By contrast, Mon2[d-F]p-GFP does not appear to associate with membranes, as judged by entirely cytosolic fluorescence, corroborating the finding that the N-terminus of Mon2p possesses the specific amino acid sequences that interact with membranes. Interestingly, we observed that N-terminal fragments of Mon2p accumulate in a polarized manner in nascent buds (neck and cortical anchor) (Fig. 4A) and possibly colocalize with cytoskeletal elements, suggesting that Mon2p plays a function in polarized membrane transport (see below).

To confirm the identity of the domains that bind to membranes, we determined the Mon2p-GFP fragments that are enriched on membrane fractions separated by sucrose density gradient (Fig. 3A). As expected, the overexpressed Mon2[A-F]p-GFP was detected as a single peak in fractions containing Kex2p and Pep12p (Fig. 4B). We subsequently analyzed the distributions of the different GFP fusion constructs on similar sucrose gradients (Fig. 4B). We confirmed that the Mon2[A-C]p-GFP fragment constitutes the core that associates with membranes as it is detected in the same fractions as Mon2[A-F]p-GFP. Surprisingly, in contrast to our fluorescence experiments, we failed to detect Mon2[A-B]p-GFP associated with membrane fractions (data not shown). These seemingly contradictory data are probably because Mon2p is stabilized on membranes by multiple protein and lipid interactions. Perhaps the short Mon2[A-B]p fragment binds membrane fractions with a lower affinity and is released during the harsh cell lysis procedure. In this context, it is worth noting that the lysis of yeast cells often leads to the solubilization of certain peripheral membrane proteins such as clathrins. We also showed that the distribution of the GFP-Mon2p protein lacking the A domain (Mon2[B-F]p-GFP) is not affected on gradients, supporting the notion that this domain is not essential for the localization of Mon2p. Taken together, we have demonstrated that Mon2p utilizes conserved domains present in other large Sec7 proteins to bind to late-Golgi/early endosomal membranes.

In addition, we tested whether Arl1p is needed to recruit Mon2p to membranes, as the N-terminal part of Mon2p is capable of binding to Arl1p (Jochum et al., 2002). The distribution of Mon2[A-F]p-GFP in a Δarl1 strain was not affected, as it still cofractionated with endosomal and Golgi membranes (Fig. 4B). This result indicates that Arl1p is not needed for interaction of Mon2p with membranes.

The localization of Mon2p on the late Golgi coupled to the fact that Arl1p binds to the GARP complex (Panic et al., 2003b) suggested that Mon2p could function in protein recycling to the Golgi apparatus. Kex2p is a late-Golgi endoprotease that functions in the proteolytic maturation of the α-factor mating pheromone (Fuller et al., 1988). The Golgi localization of Kex2p depends on the active retrieval pathways from a post-Golgi compartment (Wilsbach and Payne, 1993). Cells with missorted Kex2p secrete highly glycosylated precursor α-factor (Payne and Schekman, 1989). Thus, to test whether Mon2p is required for the sorting of Kex2p, we analyzed the forms of the secreted α-factor in a Δmon2 mutant where the open reading frame was replaced by the ADE2 marker. No trace of incompletely matured forms of α-factor was detected in the Δmon2 mutant in a pulse-chase labeling and immunoprecipitation experiment (data not shown). We also showed that the transport of the carboxypeptidase Y (CPY) to the vacuole, which is dependent on the recycling of its sorting receptor, Vps10p, is not affected (Fig. 5A).

We next examined the distribution of proteins (Snc1p, Tlg2p, and Chs3p), which are known to cycle through the late Golgi and early endosome in normal growth conditions at 30°C. We found that all these proteins were properly sorted as judged in vivo by fluorescence microscopy (data not shown). However, at high temperature, where Δmon2 cells display a severe growth defect, we found different internal distributions of Snc1p and Tlg2p, indicating a drastic membrane trafficking defect (supplementary material Fig. S3). These observations support a role of Mon2p in membrane trafficking between the late Golgi and early endosomes, despite the fact that none of the critical steps in vesicular transport, i.e. budding, docking or fusion, is completely blocked in the Δmon2 mutant.

Recently, it was demonstrated that the late-Golgi GARP complex is required for the formation of Cvt vesicles (Reggiori et al., 2003). The Cvt pathway mediates the constitutive and specific transport of the hydrolases Ape1 and α-mannosidase (Ams1) from the cytoplasm to the vacuole. Precursor forms of Ape1 and Ams1 are first enwrapped by a double membrane, leading to the formation of Cvt vesicles. Subsequently, these vesicles fuse with the vacuole, leading to digestion of the inner membranes and the cleavage of the two precursor proteins to their mature forms. To determine whether Mon2p is involved in this process, we analyzed the processing of prApe1 in the Δmon2 mutant. Wild-type and Δmon2 cells were subjected to pulse-chase labeling, and Ape1 was subsequently immunoprecipitated from cells and analyzed by SDS-PAGE and fluorography. We found that the processing of prApe1 is almost completely defective in the Δmon2 mutant (Fig. 5B). This defect is not due to reduced proteolytic activity of the vacuole as the maturation of CPY was not affected under the same conditions. Nevertheless, to visually confirm a transport defect, we next determined the localization of a GFP-prApe1 fusion by fluorescence microscopy. In wild-type cells expressing GFP-prApe1 protein, fluorescence accumulates in the vacuole lumen (Fig. 5C). By contrast, GFP-prApe1 was observed on a few dots in close proximity to the highly fragmented vacuolar membranes of Δmon2 cells (Fig. 5C). This localization is similar to that observed in mutants blocking either the formation or completion of Cvt vesicles, suggesting that Mon2p function is critical for one of these two events. When the Δmon2 strain was treated with rapamycin (to mimic starvation) prior to pulse-chase labeling, normal maturation of prApe1 was observed, indicating that the non-selective autophagy pathway is not affected in the mutant (data not shown).

We next tried to identify the domains that are required for the delivery of prApe1 into the vacuole. The Δmon2 mutant was transformed with the various truncated forms of MON2 and the maturation of prApe1 was subsequently analyzed. The F domain must be present in truncations to restore Cvt pathway function to near-wild-type levels in the deletion strain (Table 3). This result suggests that the C-terminal region of Mon2p has a key function in endomembrane transport, which could also be linked to its requirement in cell growth.

The finding that constitutive transport of prApe1 is defective in the Δmon2 mutant strengthens the idea that Mon2p functions in a similar capacity to the GARP complex and is needed to recycle materials to the late Golgi. Therefore, we reasoned that mutants blocking membrane traffic through the late Golgi should display a synthetic growth defect when combined with the Δmon2 null allele. The Δmon2 mutant was crossed with mutants blocking vesicular trafficking to and from the late Golgi such as Δvps1, Δvps26 (retromer component), Δvps52, Δvps54 (members of the GARP complex), Δrcy1, Δvps45 and Δarf1. The resulting diploid strains were sporulated and tetrads were dissected. Synthetic lethality was concluded based on the inability of dissected spores containing both deletions to grow at 30°C. As expected, all the double mutants were synthetic lethal at 30°C or at 33°C for the Δmon2/Δarf1 double mutant (data not shown).

Membrane transport defects at the late Golgi often lead to a marked distortion of organelles of the late secretory pathway and appearance of aberrant membrane structures such as Berkeley bodies. To investigate endomembrane integrity in the Δmon2 mutant, we grew cells at 26°C or at 38°C for 30 minutes to accentuate membrane alterations and performed an electron microscopy analysis. As previously observed (Jochum et al., 2002), Δmon2 cells contain a large number of fragmented vacuoles (Fig. 6). In addition, accumulation of large membrane inclusions and what appear to be lipid droplets are seen throughout the cytoplasm at both temperatures (Fig. 6B-D,F,H), very similar to that observed in a pik1^ts strain (pik1-83) at the restrictive temperature (Fig. 6E,G). More surprisingly, multinucleated cells were also observed. These findings support a role for Mon2p in the regulation of membrane exchange at the Golgi (and perhaps on endosomes) and may play a role in nuclear segregation during cell division.

Our structural protein analysis along with previous studies (Jochum et al., 2002; Wicky et al., 2004) indicates that Mon2p interacts with multiple protein partners. This led us to investigate the nature of protein complexes associated with Mon2p. Cells expressing Mon2p-HA protein were lysed under native conditions and protein complexes resolved by gel filtration chromatography. Mon2p-HA was detected in two distinct high molecular mass complexes of ∼550 and 900 kDa (Fig. 7A). The large size of these complexes suggests that Mon2p may oligomerize in vivo, potentially acting as a scaffold for recruitment of its protein partners. To address this possibility, we first constructed a diploid strain harboring both a Mon2p-myc and a Mon2p-HA chimera. Detergent-solubilized whole cell native extracts from a Mon2p-HA control strain and the doubly tagged diploid were then subjected to centrifugation at 13,000 g. Proteins were immunoprecipitated from the soluble fractions (S13) using anti-HA antibodies. In subsequent western blotting, Mon2p-myc was found to be specifically immunoprecipitated only in the presence of Mon2p-HA (Fig. 7B).

To further elucidate the cellular function of Mon2p, we next searched for new partners of Mon2p. To do this, we performed a multicopy suppressor analysis of a Δmon2 mutant by using a yeast genomic library cloned in a multicopy vector (2 μ, LEU2). The candidate genes capable of suppressing the Δmon2 growth defect were isolated and subsequently identified by comparing their DNA sequences to that of the yeast genome database. Three distinct genes, ARL1, NEO1 and DOP1 were able to restore the growth of Δmon2/chc1-521 at 30°C (Fig. 8). The isolation of Arl1p and Neo1p is consistent with the finding that these two proteins are genetically and biochemically linked to Mon2p (Jochum et al., 2002; Wicky et al., 2004). The DOP1 gene is essential for cell growth and is implicated in cell polarity, as the overexpression of the N-terminal region of Dop1p in wild-type cells leads to an abnormal budding pattern (Pascon and Miller, 2000).

Based on computer sequence analyses, Dop1p is a large protein of 195 kDa that does not contain any putative transmembrane domain(s). However, Dop1p was detected mostly in the high-speed pellet fraction (P100) by differential centrifugation (data not shown), indicating that Dop1p is peripherally associated with light membranes. To substantiate this finding, we tested the conditions required to release Dop1p-HA-mRFP from the P100 membrane pellet into the supernatant. Lysates were subjected to various treatments prior to centrifugation at 100,000 g for separation into pellet and supernatant fractions. Most Dop1p-HA-mRFP was solubilized by 1 M NaCl, 0.1 M Na₂CO₃ (pH 11) and 2 M urea (Fig. 9A). We next determined whether Dop1p resides in close proximity to Mon2p membranes in a double immunofluorescence experiment. The Mon2p-GFP fusion protein was coexpressed with a Dop1p-HA-mRFP chimera. Dop1p-HA-mRFP displays a very similar punctate fluorescence to that of Mon2p-GFP (Fig. 9B). The merged image reveals a partial colocalization of the two proteins, with approximately 50% overlapping signal (n=50). Next, we tested for in vitro interaction of Dop1p with Mon2p by coimmunoprecipitation. Detergent-solubilized whole cell native extracts from wild-type cells expressing Mon2p-GFP either alone or together with Dop1p-HA-mRFP were processed as previously described (Fig. 8B). Mon2p-GFP was found to be specifically immunoprecipitated only in the presence of Dop1p-HA-mRFP (Fig. 9C). Collectively, these results indicate that Dop1p is a peripheral membrane protein, which binds to Mon2p on Golgi and/or endosomal elements, and that its function(s) might be at least partly linked to Mon2p.

In the present study, we show that Mon2p belongs to a new protein family, with nine homologues identified. The presence of six distinct domains suggests that this group of proteins shares similar functions via conserved protein interactions. The B, C and D domains, whose functions have not yet been determined, are present in the large Sec7 GEFs. As already noted (Jochum et al., 2002), the C domain has structural similarity to the catalytic Sec7 domain with a high content of α-helices. The remaining A, E and F domains were only detected in this new group of conserved proteins and thus, we propose that they represent the signature of this family.

The N-terminal region of Mon2p is necessary and sufficient to associate with the late-Golgi/endosomal fractions. This membrane association does not require the first 50 amino acid residues representing the A domain, but is dependent on the B and C domains. Our results suggest that the B domain is the key region that interacts with membrane fractions, whereas the C domain might increase the affinity of Mon2p for membranes. Interestingly, the Arl1p binding domains have been shown to reside in the B and C domains (Jochum et al., 2002), which are conserved in the large Sec7 GEFs (see supplementary material Fig. S2). This raises the possibility that Arl1p also binds to the large Sec7 GEFs. Further, we have shown that the activity of Mon2p is dependent on the F domain. The presence of highly conserved hydrophobic and charged residues (supplementary material Fig. S1B) indicates that the F domain could interact with a specific protein, which is probably present and conserved in most eukaryotic species. Altogether, we imagine a model in which Neo1p and Arl1p are required to maintain Mon2p in a specific lipid and protein environment, allowing Mon2p to interact with other cellular components.

Previously, Mon2p was reported to bind Arl1p and Neo1p and to play a role in endocytosis and vacuolar biogenesis (Jochum et al., 2002). Here, we extend the role of Mon2p to the late-Golgi compartment. Our data show that Mon2p predominantly colocalizes with late-Golgi markers, which is consistent with the partial localization of Arl1p and Neo1p on late-Golgi structures (Lu et al., 2001; Wicky et al., 2004) and the existence of a genetic link between Mon2p and several late-Golgi proteins. Furthermore, the presence of fragmented vacuoles, as already observed by others and an accumulation of aberrant membrane structures resembling Berkeley bodies are typical of membrane transport defects owing to compromised late-Golgi function.

Recently, Arl1p was proposed to be involved in a retrograde pathway from endosomal compartments to the late Golgi by interacting with Golgins and the GARP complex (Panic et al., 2003b). Our data show that the transport of prApe1 into the vacuole, which requires the GARP complex (Reggiori et al., 2003), also depends on the functional C-terminal domains of Mon2p. This finding further supports the conclusion that the F domain is essential for the activity of Mon2p, and that the requirement for the N-terminal late-Golgi binding domains can be bypassed under overexpression conditions. Surprisingly, we were unable to detect a direct interaction between Mon2p and components of the GARP complex (Vps53p and Vps54p; data not shown). This result suggests that Mon2p is not part of the docking/tethering complex implicated in retrograde transport to the late Golgi. Consistent with this, we only observed a protein trafficking defect between the late Golgi and endosomal compartments in the absence of Mon2p at 37°C, whereas another type of protein transport from the late Golgi such the ALP delivery to the vacuole, bypassing the endosomes, is already delayed in the Δmon2 mutant at 30°C (Bonangelino et al., 2002). Altogether, our results suggest that Mon2p interacts with multiple partners in regulating late-Golgi homeostasis, and that the protein-recycling defect in the Δmon2 mutant is probably an indirect consequence of the alteration of Golgi and/or endosomal structures.

Golgins regulate the structure and dynamics of the Golgi apparatus by associating membranes with cytoskeletal elements (Barr and Short, 2003). The yeast actin cytoskeleton is organized into morphologically distinct structures, which provide the structural basis for the polarized transport of membranes during cell division (Pruyne and Bretscher, 2000). Interestingly, actin organization is adversely affected in a Δmon2 mutant (Singer-Kruger and Ferro-Novick, 1997). In this respect, we found that N-terminal fragments of Mon2p accumulate at the polarized sites of nascent buds and a synthetic lethality between Δmon2 and the Δarc18 mutant (data not shown). Arc18p is a subunit of the Arp2/3 complex, which initiates the nucleation of actin filaments in cortical patches (Winter et al., 1999). In agreement with a putative function of Mon2p in polarized growth, we found that the highest conserved Mon2p homolog is a protein encoded by Ashbya gossypii YNL297C (40% identity). A. gossypii is a filamentous fungus that grows by hyphal extension, which represents an extreme example of polarized growth. Even more interestingly, we found that Mon2p genetically and physically interacts with Dop1p. Dop1p is an essential protein that shares conserved functions in cell polarity with DopA, its Aspergillus nidulans homolog (Pascon and Miller, 2000). DopA also plays a role in directed nuclear movement in A. nidulans cells (Pascon and Miller, 2000). Similarly, we often observed the appearance of multinucleated cells in the Δmon2 mutant, which could imply a function for Mon2p in nuclear dynamics during cell division.

In summary, we have identified a new group of conserved proteins, which probably share similar functions in the coordination of membrane trafficking between the late Golgi and early endosomes, as well as in polarized growth.

We are extremely grateful for continuous support from Costa Georgopoulos and Patrick Linder. We thank R. Boeck for the construction of pRB95, pRB108, pRB111 and pRB113 plasmids, and I. Iost, M.-C. Daugeron, H. Pelham and K. Nasmyth for providing plasmids and libraries. We also thank Anjon Audhya, William Parrish, Christopher Stefan and Simon Rudge for key reagents and helpful discussions, Ingrid Niesman for EM processing (Immunoelectron Microscopy Core B of Program Project grant CA58689 headed by M. Farquhar) and Alexander Rusnak for technical assistance. This work was supported by grants from the Swiss National Science Foundation and the Canton of Geneva to C.G. (FN-31-65403) and P.L. (FN-31-43321) and NIH training grant #5T32CA67754-08 (to J.A.E). S.D.E. is an investigator of the Howard Hughes Medical Institute.