QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 18, Issue 4, 1203-1219, April 2007

This Article Abstract Full Text (PDF) Supplemental Material All Versions of this Article: E06-11-1035v1 E06-11-1035v2 18/4/1203    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Harsay, E. Articles by Schekman, R. Search for Related Content PubMed PubMed Citation Articles by Harsay, E. Articles by Schekman, R.

Avl9p, a Member of a Novel Protein Superfamily, Functions in the Late Secretory Pathway

Edina Harsay^*,, and Randy Schekman

*Department of Molecular Biosciences, University of Kansas, Lawrence, KS 66045; and Department of Molecular and Cell Biology, Howard Hughes Medical Institute, University of California, Berkeley, CA 94720

Submitted November 22, 2006; Revised December 29, 2006; Accepted January 10, 2007
Monitoring Editor: Benjamin Glick

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The branching of exocytic transport routes in both yeast and mammalian cells has complicated studies of the late secretory pathway, and the mechanisms involved in exocytic cargo sorting and exit from the Golgi and endosomes are not well understood. Because cargo can be sorted away from a blocked route and secreted by an alternate route, mutants defective in only one route do not exhibit a strong secretory phenotype and are therefore difficult to isolate. In a genetic screen designed to isolate such mutants, we identified a novel conserved protein, Avl9p, the absence of which conferred lethality in a vps1 apl2 strain background (lacking a dynamin and an adaptor-protein complex 1 subunit). Depletion of Avl9p in this strain resulted in secretory defects as well as accumulation of Golgi-like membranes. The triple mutant also had a depolarized actin cytoskeleton and defects in polarized secretion. Overexpression of Avl9p in wild-type cells resulted in vesicle accumulation and a post-Golgi defect in secretion. Phylogenetic analysis indicated evolutionary relationships between Avl9p and regulators of membrane traffic and actin function.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Membrane vesicle–mediated intracellular transport of proteins and lipids is a fundamental process in all eukaryotic cells. Membrane transport pathways are essential for cell growth and division as well as for maintaining normal cell homeostasis of nondividing cells. The mechanisms of transport carrier formation and fusion of these carriers with their target membranes are well conserved among eukaryotes, so relatively simple organisms can serve as useful models for studying these processes. The yeast Saccharomyces cerevisiae has proven to be an especially valuable tool in the study of membrane traffic, and the majority of the components of the secretory machinery shared by all eukaryotic cells were originally identified in yeast (Novick et al., 1980; Bankaitis et al., 1986; Rothman and Stevens, 1986; Robinson et al., 1988).

The first extensive yeast genetic screen for mutants that block the exocytic pathway identified many of the components involved in endoplasmic reticulum-to-Golgi transport and vesicle fusion with the plasma membrane, but only 2 of the 23 genes identified in this screen blocked anterograde transport from the Golgi (Novick et al., 1980; Guo et al., 2000; Lee et al., 2004). As in mammalian cells, there are multiple exocytic transport routes from the yeast Golgi, so cargo in a blocked pathway can be missorted and secreted by an alternate route (Harsay and Bretscher, 1995; Harsay and Schekman, 2002). This has made it difficult to identify mutants that have a defect in exocytic transport from the Golgi, and relatively little is known about the molecular machinery involved at this transport step.

Although small (40–100 nm) vesicles are the best-characterized transport intermediates in the secretory pathway, larger tubular structures also transport cargo. In mammalian cells, tubular vesicles appear to be the most abundant class of exocytic carriers from the Golgi (Hirschberg et al., 1998; Toomre et al., 1999; Polishchuk et al., 2000; Puertollano et al., 2003). Exocytic vesicles in yeast are smaller but they are formed by some of the same processes, which include elaborate mechanisms that regulate membrane lipid composition at the trans-Golgi network (TGN). These involve lipid modifying enzymes such as phospholipase D (Chen et al., 1997; McDermott et al., 2004) and phosphatidylinositol (PtdIns) kinases and phosphatases (Simonsen et al., 2001; Roth, 2004); PtdIns transporter proteins (Bankaitis et al., 1990; Litvak et al., 2005); and aminophospholipid translocases (Chen et al., 1999; Natarajan et al., 2004). The Ras-type small GTPase Arf1 is both a regulator of, and regulated by, the membrane lipid–modifying processes and is a key control factor in transport from the Golgi and endosomes (Nie et al., 2003; D'Souza-Schorey and Chavrier, 2006).

Numerous vesicle coat complexes are involved in recruiting cargo and deforming membranes for vesicle formation. Clathrin-coated vesicles were identified first and are the best characterized transport intermediates (Pearse and Robinson, 1990; Kirchhausen, 2000). Clathrin is linked to membranes by adaptor proteins (APs), which are recruited to donor membranes by cargo proteins, Arf, and specific phosphoinositides (Ghosh and Kornfeld, 2004; Robinson, 2004). Of the various transport vesicle species formed at the TGN, the AP-1 (adaptor protein complex 1)-containing clathrin-coated vesicles may mediate transport to endosomes en route to lysosomes. More recently discovered adaptor proteins, Gga's and epsins, are also involved in recruiting cargo in this transport route and are components of at least some AP-1–containing vesicles (reviewed by Robinson, 2004). These AP-1 or another class of clathrin-coated vesicles also mediate transport from the Golgi to the cell surface, either directly from the Golgi or to an endosomal intermediate (Fölsch et al., 2001; Harsay and Schekman, 2002; Gall et al., 2002; Ang et al., 2004). However, clathrin does not appear to be involved in the major exocytic routes from the Golgi, so these pathways may rely on a mechanism of vesicle formation not involving coat proteins, or on yet-unidentified coat complexes. Possible coat proteins, FAPPs (four-phosphate adaptor proteins), that regulate exocytic transport from the TGN have been identified in mammalian cells (Godi et al., 2004; Vieira et al., 2005). The FAPP proteins bind to both Arf and PtdIns(4)P at the TGN and play a role specifically in non–clathrin-coated vesicle formation. One explanation for why coat proteins mediating cargo transport to the cell surface have been difficult to identify may be that there could be multiple, perhaps partially redundant, coat complexes with unique cargo specificities. A coat complex involved in transporting only a few cargoes from the Golgi to the plasma membrane in yeast and other fungi is the Chs5/6, or exomer, complex (Sanchatjate and Schekman, 2006; Trautwein et al., 2006; Wang et al., 2006).

The release of some types of vesicles, including clathrin-coated vesicles, from donor membranes involves dynamin GTPases (Damke et al., 1994; Hinshaw, 2000) as well as various BAR domain proteins that directly modify membrane curvature (Peter et al., 2004; Ren et al., 2006). Actin and its regulators also play important roles in vesicle formation, both at the plasma membrane and the TGN (Engqvist-Goldstein and Drubin, 2003; Carreno et al., 2004; Cao et al., 2005; Egea et al., 2006; Kessels et al., 2006). Other regulators of vesicle fission, unique for TGN-to-plasma membrane transport in some cell types, are protein kinase D (PKD; Liljedahl et al., 2001; Yeaman et al., 2004) and CtBP/BARS (Weigert et al., 1999). An effector of PKD is a PtdIns 4-kinase (Hausser et al., 2005), the activity of which is critical for transport from the Golgi to the plasma membrane in both yeast and mammalian cells (Hama et al., 1999; Walch-Solimena and Novick, 1999; Bruns et al., 2002; Godi et al., 2004). CtBP/BARS and PKD function only in the transport of cargo that have basolateral sorting signals (when expressed in either polarized or nonpolarized cells) and not in the transport of apical cargo (Yeaman et al., 2004; Bonazzi et al., 2005). Clearly, although some of the participants in exocytic transport from the TGN are known, other components are yet to be identified.

The genetic screens in yeast that identified many of the components of the exocytic machinery generally relied on a strong secretory block and temperature-conditional lethality. Such screens usually identified essential genes required for the transport of all (or most) cargo and likely missed genes that are required in only one of multiple routes from the Golgi to the cell surface. We therefore conducted a screen for new secretory mutants by using a mutant strain background that has a block in one of two known exocytic transport routes, so that a remaining route becomes essential. Using this strategy, we have identified a novel conserved protein that has a function in exocytic transport from the Golgi.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Reagents and Growth of Yeast Strains
Minimal medium for growing plasmid-carrying yeast strains was CSM (complete synthetic medium) lacking a nutrient for plasmid selection, with amino acid mixes from Q-Biogene (Carlsbad, CA). All other growth media components were from Difco (Detroit, MI), and were prepared following recipes described by Sherman (2002). Rich medium was YPD (yeast extract, peptone, dextrose). All media contained 2% glucose unless otherwise stated. Although our vps1 and vps1 apl2 strains grew only slightly slower than wild-type cells at 37°C, the vps1 mutant has been reported to be temperature-sensitive in some strain backgrounds (Rothman et al., 1990), so all cultures were maintained at 24°C unless otherwise noted. Culture growth was monitored by measuring OD₆₀₀ in a Genesys 5 spectrophotometer (Spectronic Instruments, Westbury, NY).

PCR for cloning genes or for generating gene deletion fragments was performed using Pfu DNA polymerase (Stratagene, La Jolla, CA). All other DNA manipulating enzymes were from New England Biolabs (Beverly, MA). Other reagents were from Sigma-Aldrich (St. Louis, MO) unless otherwise stated.

Strains and Plasmids
Yeast Strains. The yeast strains used in this study are listed in Table 1. The strain 4513-216 is derived from strains described by Koshland et al. (1985); all other strains are derived primarily from S288C (Mortimer and Johnston, 1986). Crosses and tetrad analysis to generate yeast strains were performed as described by Sherman (2002). Yeast transformations for introducing plasmids or PCR products were by the lithium acetate method (Schiestl and Gietz, 1989). EHY767 was derived from eight sequential crosses of strains having EHY47, EHY362, 4513-216, and GPY1783-10A (Yeung et al., 1999) backgrounds. EHY769 was generated from EHY441 by integrating a PCR product containing apl2::KAN (generated using primers EH130, EH131; see below) from EUROSCARF strain Y14985 [GenBank] (Brachmann et al., 1998). To disrupt AVL9 in our sec6-4 strain background, a PCR product containing avl9::KAN from EUROSCARF strain Y12725 [GenBank] (generated using primers EH162, EH163) was integrated into EHY188 (Harsay and Schekman, 2002) to generate EHY883. Proper targeting of PCR products was confirmed by PCR analysis. The construction or origin of all other strains is described in Table 1.

Primers. The primers mentioned above for strain and plasmid construction are as follows: EH130: TAACGCTTTACAAACAGAGCATA, EH131: TGAGAAATATTTAGATGGTGAAAGGGA, EH162: TGATGCTTGTCCGTCGGGT, EH163: CTTGTGGAGGTCACCCCAGTT, EH164: GCCTAGCAAAAATAGCCGCTG, EH165: TCTATTCCATTTTTGGAAAGCCCC, EH203: CCTCGCTGCAACCTTGTTGTCA.

Synthetic Lethal Screen
We used a colony sectoring scheme based on the color of ade2 ade3 mutants to identify synthetic lethal mutants, similar to the approach described previously (Koshland et al., 1985; Bender and Pringle, 1991). Overnight cultures of EHY771 or EHY1172 were grown to stationery-phase (to minimize budded cells) in CSM, –Ura medium, and mutagenized with ethyl methanesulfonate (EMS; according to Guthrie and Fink, 1991) to a 20–95% survival rate, in several screens. Cells were plated on YPD to obtain 100 colonies per plate, and nonsectoring colonies were tested for lethality on 5-fluoroorotic acid plates (5-FOA) to test inability to lose a URA-VPS1 or URA-APL2 plasmid (pEH134 or pEH308). Mutants that were unable to grow on 5-FOA were transformed with a TRP1 APL2 or TRP1 VPS1 plasmid (pEH305, pEH314) and again tested for ability to grow on 5-FOA. Mutants that grew on 5-FOA when transformed with either of these plasmids have mutations that are lethal in combination with the vps1 apl2 double mutation. Such mutants were back-crossed to confirm 2:2 segregation of the synthetic lethal phenotype. Of 17 complementation groups, most with only one mutant member, one mutant had a consistently strong phenotype in repeated back-crosses. The complementing wild-type gene for this mutant, avl9-1 (AP-1 Vps1 lethal 9), was cloned.

Cloning AVL9
The AVL9 gene was cloned from a YCp50-based genomic library (CEN, URA3) created from genomic DNA prepared from EHY529, which lacked VPS1 (E.H., unpublished reagent). The library had the advantage that we did not isolate VPS1-containing plasmids (which would be suppressors). Strain EHY840 (avl9-1 vps1 apl2 with an APL2-TRP1 plasmid) was transformed with the library and plated on –Ura plates. Colonies were scraped and plated onto 5-fluoroanthranilic acid (5-FAA) plates, which do not allow cells with a TRP1 gene to grow (Toyn et al., 2000). Transformants that could grow on 5-FAA were able to lose the TRP1 APL2 plasmid; these colonies were next streaked onto 5-FOA plates to ensure that the cells cannot grow (and therefore require the URA3 plasmid from the library for suppression). Suppressor plasmids were isolated and sequenced at the insert junctions to identify the genomic DNA fragment responsible for suppression. This strategy yielded three library clones with overlapping 10-kb inserts from chromosome XII. The suppressing gene from this region was identified by transposon mutagenesis of the plasmids using the GPS-1 Genome Priming System Kit (New England Biolabs). After transposon mutagenesis, the suppressor plasmid was transformed into Escherichia coli, transformed colonies were pooled, and plasmid DNA was isolated to generate a library of mutagenized plasmids. This library was transformed into EHY840, and transformants were screened for inability to grow on FAA by replica-plating. Three nonsuppressing clones were isolated, and all were found to contain a unique insertion in the gene YLR114c.

Linkage of YLR114c with avl9-1 was tested to determine whether it is the corresponding wild-type gene or a low-multicopy suppressor, as follows: The YLR114c genomic region was subcloned into an integrating (YIp) URA3 plasmid (pRS306; Sikorski and Hieter, 1989), cut, and integrated into its chromosomal locus in a vps1::LEU2 apl2::KAN avl9-1 strain. This strain was crossed to an AVL9 strain with the opposite MAT type and lacking the integrant but otherwise identical, sporulated and dissected for tetrad analysis. None of the progeny exhibited the synthetic lethal phenotype; therefore, the integration was linked with the avl9-1 gene. To further confirm that YLR114c is essential in a vps1 apl2 background, a strain having a deletion of YLR114c was obtained from the EUROSCARF yeast deletion collection (Y12725 [GenBank] ) and crossed to a vps1 apl2 strain to confirm the synthetic lethal phenotype.

To determine the nature of the avl9-1 mutation, DNA of the mutant allele was generated by PCR, and two separate PCR reaction products were sequenced from both ends.

Subcellular Fractionation
Nycodenz gradient fractionation to characterize secretory vesicles was performed as described previously (Harsay and Schekman, 2002) except that rather than using a triethanolamine-acetic acid buffer with sorbitol for cell lysis and the gradients, the buffer contained 50 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 6.5, with KOH, 0.8 M sorbitol, and 1 mM EDTA. This change did not alter the gradient profiles observed for sec6-4 and other strains tested. Enzyme assays were performed as previously described (Harsay and Schekman, 2002). For assaying gradient fractions by Western blots, the blots were processed as described previously (Harsay and Schekman, 2002). Images were captured in a ChemiDoc XRS digital darkroom (Bio-Rad, Richmond, CA) and quantified using Quantity One software (Bio-Rad).

Assays for Secretory Defects
Growth Conditions. For examining the effects of depleting Avl9p, cultures were grown overnight to log-phase in CSM, –Ura, 2% galactose, 2% raffinose, and 0.3% glucose and shifted into medium containing 2% glucose for 20 h, with dilution to maintain log-phase growth. For examining the effects of overexpressing Avl9p, cells were grown in CSM, –Ura with 2% glucose to log phase, and then transferred to CSM, –Ura, 2% raffinose, and grown overnight to log-phase. Growth in raffinose medium allows more rapid expression from the GAL1 promoter after the shift to galactose. Cells were diluted to OD₆₀₀ 0.05 in CSM, –Ura, 2% galactose, and 2% raffinose and growth was monitored at 1-h intervals for up to 15 h.

Bgl2p Accumulation Assay. Cells from 10 ml of culture (grown to OD₆₀₀ 0.3–0.7) were collected by centrifugation and resuspended in ice-cold 10 mM NaN₃, 10 mM KF. After 10 min on ice, cells were incubated in prespheroplasting buffer (0.1 M Tris-H₂SO₄, pH 9.4, 50 mM -mercaptoethanol (ME), 10 mM NaN₃, 10 mM KF) for 15 min on ice, washed in 1.4 M sorbitol, 50 mM KPi, pH 7, and 10 mM NaN₃ and resuspended in the same buffer with 170 µg/ml Zymolyase 100T (United States Biological, Swampscott, MA) for 30 min at 30°C with occasional gentle agitation. Cells were collected by centrifugation at 5000 x g for 10 min, pellets were resuspended in 200 µl of Laemmli sample buffer and heated for 5 min at 95°C, and 5 µl of sample was resolved by SDS-PAGE (Laemmli, 1970). Bgl2p and PGK were detected by immunoblot as previously described (Harsay and Schekman, 2002) except detection was with a ChemiDoc XRS imager (Bio-Rad).

Pulse-Chase Analysis. Cultures were grown in CSM media as described above and then shifted into CSM, –Met, –Cys at a density of 4 OD/ml. After 15 min, [³⁵S] Easytag Express Protein Labeling Mix (1200 Ci/mmol, 14.3 mCi/ml, from Perkin Elmer-Cetus, Norwalk, CT) was added at 50 µCi/OD₆₀₀ cells. After a 5-min labeling period, cold amino acid mix (50 mM methionine, 10 mM cysteine, 4% yeast extract, 2% glucose) was added for the indicated chase periods. One OD₆₀₀ unit per immunoprecipitation reaction was removed and added to ice-cold 5x energy poison buffer (100 mM NaN₃, 100 mM KF, 0.5 M KPi, pH 7.5). After 10 min on ice, cells were converted to spheroplasts by the addition of 3x spheroplast buffer (0.3 mg/ml Zymolyase 100T, 2.8 M sorbitol, 50 mM ME) and incubation for 20 min at 36°C. Spheroplasts were chilled and gently centrifuged at 4°C for 10 min, 5000 x g, and pellets ("internal fraction") were separated from supernatants ("external fraction," containing medium and cell wall material). Supernatants were centrifuged again at 14,000 x g for 5 min to ensure complete removal of cells, diluted to 1 ml per original OD₆₀₀ unit with H₂O, and trichloroacetic acid–precipitated using sodium deoxycholate as a carrier (Rexach et al., 1994). Bgl2p or CPY was immunoprecipitated according to published procedures (Stirling et al., 1992) and subjected to SDS-PAGE on 10% gels, which were analyzed by phosphorimaging on a Cyclone phosphorimager (Perkin Elmer-Cetus).

Invertase Secretion Assay. Cultures were grown in CSM media as described above. Cells from 10 ml of culture were collected by centrifugation, resuspended in 5 ml YPD medium with 5% glucose, and incubated with shaking for 1 h. Cells were then washed in H₂O, shifted to YPD medium with 0.1% glucose, and incubated with shaking for 90 min to derepress secretory invertase expression. They were then collected by centrifugation, resuspended in ice-cold energy poison (10 mM NaN₃, 10 mM KF) for 10 min, and diluted to 10 ml with H₂O for washing and to measure OD₆₀₀. Cells were again centrifuged and resuspended in 1 ml 0.2 M NaOAc, pH 4.9. The samples were split in two, and half the cells were permeabilized by the addition of 12 µl 20% Triton X-100 and freeze-thaw. This generated a "total" (permeabilized) fraction and an "outside/wall" fraction from unpermeabilized cells (Bankaitis et al., 1990), which were assayed for invertase activity essentially as described (Goldstein and Lampen, 1975). Cell samples for each reaction originated from 0.02 OD₆₀₀ cells (5–10 µl of permeabilized or whole cell sample per 150 µl reaction).

Electron Microscopy
Supplies and reagents for electron microscopy (EM) were obtained from EMS (Hatfield, PA). For examining the effects of depleting Avl9p, we grew cells in CSM, –Ura medium with 0.3% glucose, 2% galactose, and 2% raffinose to log phase and then shifted them to CSM, –Ura medium with 2% glucose for 15 h (to OD₆₀₀ 0.3). They were then grown for 5 h in YPD with 2% glucose to obtain optimal morphology by EM. Cells were concentrated by vacuum filtration on a 0.2-µm filter to 7.5 ml, and 2.5 ml fixative was added to a final concentration of 2% formaldehyde, 2% glutaraldehyde, in 40 mM potassium phosphate buffer, pH 6.7. Cells were fixed with gentle agitation for 10 min and then collected by centrifugation and resuspended in fixative for continued fixation with gentle agitation for 45 min. Cells were then washed in 40 mM potassium phosphate, pH 6.7, postfixed in 4% KMnO₄ (Mallinckrodt, St. Louis, MO), stained en bloc with 0.5% aqueous uranyl acetate overnight, and processed for thin-section EM essentially as described by Wright (2000). Dehydration was in a graded series of ethanol and embedding was into Spurr's resin. Thin sections were stained with lead citrate (Reynolds, 1963) and stabilized by applying a thin layer of carbon with a Baltec vacuum evaporator (Boeckeler Instruments, Tuscon, AZ). Images were acquired on Kodak 4489 or SO-163 film (Eastman-Kodak, Rochester, NY) using a JEOL 1200 EX transmission electron microscope (TEM; Peabody, MA) operated at 80 or 100 kV. Negatives were scanned and images were adjusted for brightness and contrast using Photoshop CS (Adobe, San Jose, CA).

Cells for examining the effects of the avl9 mutation in a sec6-4 background were grown as described in the figure legend and prepared as above. For examining the effects of overexpressing Avl9p, cells were grown in CSM, –Ura with 2% glucose to log phase, then transferred to CSM, –Ura, 2% raffinose and grown for two doublings (7 h). Cells were then shifted into CSM, –Ura, 2% raffinose, and 2% galactose, grown for 11 h (while maintaining the culture in log-phase), and prepared as above for thin-section EM.

Light Microscopy
For experiments examining the effects of Avl9p depletion, cultures were grown as above for EM. Fixative (2.5 ml) was added to 10 ml culture, for a final concentration of 4% formaldehyde, 40 mM phosphate buffer, pH 6.7. Cells were fixed for 1 h and then overnight in fresh fixative. We followed published procedures to stain cells with caclofluor (Chant and Pringle, 1995). To visualize polymerized actin, fixed cells were washed with PBS, pH 7.4, permeabilized in 0.1% Triton X-100 for 15 min, washed four times in PBS, and stained in 0.1 U/µl Alexafluor-568 phalloidin (Molecular Probes, Eugene, OR) for 2 h. Washed cells were mounted in Prolong Gold Antifade (Molecular Probes) and observed with a 100x/1.4 PlanApochormat lens on an Axioplan 2ie microscope equipped with a motorized stage (Carl Zeiss MicroImaging, Thornwood, NY). Images were captured with a charge-coupled device camera (Orca ER, Hamamatsu Photonics, Bridgewater, NJ) and Openlab imaging software (Openlab; Improvision, Lexington, MA). Z-series were acquired in 0.2-µm steps and assembled and deconvolved using Volocity 2.6 software (Improvision). 2D projections were exported as TIFF files and adjusted for brightness and contrast in Photoshop CS.

Sequence Analysis
We identified orthologues of yeast Avl9p by BLAST searches (Altschul et al., 1997) of the nr protein database using the NCBI BLAST server with default parameters. We also searched for protein families distantly related to Avl9p by PSI-BLAST using as query the most conserved regions of Avl9 proteins and consensi of these regions (200–250 residues). Members of potentially related protein families were then further analyzed by ClustalW multiple sequence alignments (Thompson et al., 1994) and by the MEME/MAST system (Multiple Em for Motif Elicitation/Motif Alignment and Search Tool; Bailey and Elkan, 1994; Bailey and Gribskov, 1998), available at the San Diego Supercomputing Center (http://meme.sdsc.edu/meme/intro.html). Regions identified as similar by MEME/MAST were then used in new PSI-BLAST and MAST searches to identify more potentially related proteins, and these were then added to new MEME/MAST analyses. This process was repeated to generate a collection of potentially related protein sequences from representatives of diverse phyla. Five Avl9 homology (AH) regions of roughly 50–100 residues each were defined by multiple MEME/MAST analyses using varying parameters. The combined AH regions as well as combined uDENN, DENN, and dDENN regions (as identified in the SMART database, Letunic et al., 2004) from DENN-domain–containing proteins were aligned with ClustalW using Lasergene 7 software (DNASTAR, Madison, WI). To optimize the alignments, we also generated separate alignments containing subsets of proteins using T-Coffee (Notredame et al., 2000) and Muscle (Edgar, 2004) and combined these with ClustalW alignments using T-Coffee. The aligned sequences were imported into MacVector 9 software (Accelrys, San Diego, CA) for editing of gaps by eye. The Entrez accession numbers for the sequences included in the alignment are as follows: DENN domain proteins: NP_079174, XP_397549, NP_079177, AAH36655; Avl9 proteins: XP_624093, NP_498416, XP_638363, NP_055875, XP_327309, CAA61692, CAB16239, EAR89363; ANR1 proteins: XP_394349, AAH40291, XP_783119, AAC02577, XP_641200, EAR85516; ANR2 proteins: EAL24019, XP_785244, P36090, XP_322253; MesA proteins: XP_646288, CAA22879, XP_657872, EAR95260; ANR3/Fam45 proteins: XP_782995, AAH22271; XP_647590; XP_001120151.

Phylogenetic reconstruction of the evolutionary relationships between Avl9 and related proteins was performed using the Phylip 3.66 software package, obtained from the author (Felsenstein, 2006). An inferred phylogenetic tree was generated by testing different trees and parameters in Phylip's maximum likelihood program, PROML, using as input the alignment obtained above along with user-defined trees generated by parsimony and distance matrix methods. The maximum parsimony user tree was generated using the PROTPARS program from Phylip, and a distance matrix tree was generated using PROTDIST with FITCH from Phylip, with the following parameters: Henikoff/Tillier PMB model of amino acid substitutions (Veerassamy et al., 2003), distribution of substitution rates among positions with coefficient of variation 0.45 (shape parameter 4.9), Fitch-Margoliash criterion, multiple randomized input orders, and the global rearrangement option. The same distance matrix tree was generated by using the above PROTDIST matrix with a weighed neighbor-joining method, Weighbor (Bruno et al., 2000). Estimated branch lengths were supplied to the most likely tested tree using PROML with a discrete approximation to distributed substitution rates (four rate categories, 4.9, PMB model of amino acid change). For bootstrap analysis (Felsenstein, 1985), pseudoreplicates of the alignment were generated using SEQBOOT from Phylip, and bootstrap trees were generated using PROTPARS (1000 replicates) and PROTDIST with FITCH (100 replicates). Consensus trees and bootstrap values were obtained using CONSENSE, and the tree was drawn using DRAWGRAM (Phylip). To test possible long-branch artifacts, additional analyses were performed with the exclusion of the ANR3 family proteins as well as with different DENN domains.

Protein secondary structure predictions were performed with the SAM (Karplus et al., 1998, 2003) and PSIPRED (Jones, 1999; McGuffin et al., 2000) algorithms. These two methods produced very similar predictions, but where they differed, PROFSec was used as tie breaker (Rost and Sander, 2000). All three prediction methods are freely available on the Web.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
A Synthetic Lethal Screening Strategy for Identifying New Exocytic Mutants
Defects in cargo transport in two exocytic pathways in yeast can be assayed by density-gradient fractionation analysis of the secretory vesicles that accumulate in sec mutants, such as sec6-4, that have a temperature-conditional block in vesicle fusion with the plasma membrane. (Wild-type yeast cells are not suitable for such analysis because they have few exocytic vesicles.) Such fractionation analysis has identified two post-Golgi vesicle species with different densities and unique cargo (Harsay and Bretscher, 1995). A lighter vesicle species transports surface proteins such as the cell wall protein Bgl2p and an abundant plasma membrane ATPase, Pma1p, whereas a dense population of vesicles transports invertase and other enzymes that are secreted into the periplasm or growth medium. Both vesicle species transport an exoglucanase, but these exoglucanases are encoded by different genes.

To identify new genes involved in the post-Golgi exocytic pathway, we screened for mutants that are lethal in combination with a partial secretory block. We conducted our initial synthetic lethal screen with a strain having a vps1 mutation, with the rationale that in a mutant lacking the VPS1 gene (encoding a dynamin homolog thought to be involved in vesicle formation at the TGN; Rothman et al., 1990; Bensen et al., 2000), invertase was missorted from the pathway mediated by high-density secretory vesicles to a pathway mediated by light-density vesicles (Figure 1A; Harsay and Schekman, 2002). Therefore, the vps1 mutant appears blocked in at least one exocytic transport route, and mutations that block a remaining route should cause lethality. However, this synthetic lethal screen was unsuccessful in identifying novel secretory mutants. Most likely, there are multiple alternate routes to the plasma membrane rather than just two, so just one mutation in addition to vps1 will rarely result in a lethal secretory defect. Alternatively, the vps1 mutation does not completely block the transport process in which Vps1p functions. Furthermore, Vps1p has recently been shown to be involved in processes other than secretion (Hoepfner et al., 2001; Peters et al., 2004). Therefore, the screen was modified so that the screen strain background contained both vps1 and another mutation in a gene involved in late secretory transport.

View larger version (21K): [in this window] [in a new window]   Figure 1. The vps1 and apl2 mutations have unique effects on invertase sorting into vesicles accumulated in a sec6-4 mutant background. We used a Nycodenz density gradient fractionation assay in which secretory vesicles accumulated by a sec6-4 mutant reproducibly peak in fraction 8 or 9 (light-density vesicles) or fractions 16–18 (high-density vesicles). (A) Nycodenz gradient fractionation of secretory vesicles accumulated in a sec6-4 vps1 mutant. All cargo is sorted into light-density vesicles, indicating a defect in the dense-vesicle transport pathway, as described previously (Harsay and Schekman, 2002). (B) The apl2 sec6-4 mutant accumulates cargo primarily at a density intermediate between that of vesicles in the light-vesicle and dense-vesicle pathways. (C) The gga1 gga2 mutations have a relatively small effect on exocytic cargo transport in a sec6-4 strain background. Invertase, which is a dense-vesicle cargo, is shown. The gradient density profile shown was highly reproducible for all gradients.

The difference between the fractionation results for the apl2 and vps1 mutants suggests that transport blocks in the two mutants occur at different compartments, most likely the Golgi and endosomes. We found that the gga1 gga2 mutant had only a mild effect on sorting invertase into dense vesicles (Figure 1C), so mutations in these genes are less likely to be useful in our screens for exocytic mutants. We therefore constructed a new screen strain having a deletion of both the VPS1 and APL2 genes. Details of a synthetic lethal screen using this strain background are described in Materials and Methods.

Our new screen was successful in yielding mutants with the desired synthetic phenotype: mutants that grow well in an otherwise wild-type background or when just VPS1 or just APL2 is absent, but which cannot survive in a vps1 apl2 background (Figure 2). We have cloned the corresponding wild-type copy for one of our mutants, AVL9, by a suppressor screen of the synthetic lethal phenotype (Supplementary Figure 1; see Materials and Methods). The gene was an uncharacterized ORF, YLR114c, coding for an 86-kDa protein. It is nonessential, and when deleted, the mutant has a synthetic lethal phenotype in the apl2 vps1 background, similar to what we found for the isolated mutant (in which a conserved Gly residue at position 52 is replaced by Asp).

View larger version (32K): [in this window] [in a new window]   Figure 2. The avl9-1 mutant is synthetically lethal in an apl2 vps1 strain background. The lethality of an avl9-1 apl2 vps1 strain carrying a URA3-APL2 plasmid (or URA3-VPS1 plasmid, not shown) on 5-FOA is rescued by introducing a TRP1 plasmid containing either the VPS1 or APL2 genes.

Depletion of Avl9p Results in a Secretory Phenotype and Accumulation of Golgi-like Structures
To observe the phenotype of the avl9 vps1 apl2 triple mutant, we attempted to generate a temperature-conditional allele of the AVL9 gene. This was unsuccessful, so we instead created a plasmid construct containing AVL9 under the control of the GAL1 promoter, enabling expression of the gene in galactose-containing growth media and shut-off of expression in glucose-containing media. Triple-deletion strains carrying either this plasmid or a plasmid with AVL9 under the control of its native promoter were grown to midlog phase in 2% galactose, 2% raffinose, and 0.3% glucose-containing minimal medium (medium for optimum growth; see below) and then shifted to 2% glucose-containing medium to shut off expression from the GAL1 promoter. Growth was maintained for up to 20 h. We then assayed for secretion of secretory cargo in each of the two exocytic pathways: Bgl2p for the light-vesicle pathway and invertase for the dense-vesicle pathway.

Bgl2p in wild-type cells is almost entirely in the cell wall, because very little internal Bgl2p is detected when the cell wall is gently removed by enzymatic digestion (Figure 3A). A similar result to wild type was observed for the avl9 mutant, and only slightly more than wild-type level of internal Bgl2p was observed for the vps1 apl2 double mutant. However, when galactose-dependent Avl9p was depleted in this mutant by growing cells in 2% glucose medium for 20 h, clear accumulation of Bgl2p was apparent. To assay the kinetics of Bgl2p secretion upon Avl9p depletion, we performed pulse-chase analysis of metabolically labeled Bgl2p (Figure 3B). A slight defect in Bgl2p secretion was observed after 10 h of Avl9p depletion in 2% glucose medium, with more severe defects noted after prolonged growth in 2% glucose. No obvious defect was detected for the avl9 or vps1 apl2 mutants. We also analyzed CPY secretion and processing in the same samples for avl9 and wild type and found no CPY transport defect (not shown).

View larger version (25K): [in this window] [in a new window]   Figure 3. The avl9 apl2 vps1 mutant has defects in the exocytic pathway. (A) Western blot showing the accumulation of internal Bgl2p. The avl9 apl2 vps1 strain carrying a plasmid with AVL9 under the control of the GAL1 promoter was grown in medium containing 2% galactose, 0.3% glucose for optimal growth and then shifted to medium with 2% glucose for 20 h to deplete Avl9p. Bgl2p is primarily in the cell wall at steady state, so internal accumulation can be detected by removing the cell wall. PGK is a cytoplasmic protein used as loading control. (B) Metabolic labeling and pulse-chase analysis shows a kinetic defect in Bgl2p transport that becomes more pronounced with increased time in glucose to deplete Avl9p. Cells were pulse-labeled for 5 min, and chase was 60 min in each experiment. Cell walls were removed, and cells (I, inside) were separated from the cell wall and media fraction (O, outside). Bgl2p was immunoprecipitated and detected by SDS-PAGE and phosphorimaging. (C) Invertase secretion was assayed as described in Materials and Methods. External and total invertase levels were determined by a colorimetric enzyme assay to calculate percent secretion. The means of three experiments for each strain are shown. Error bars, SEM.

We prepared cells for thin-section EM to determine whether depleting Avl9p resulted in the accumulation or abnormal morphology of secretory organelles. Upon Avl9p depletion, the apl2 vps1 mutant accumulated abundant structures that resembled aberrant Golgi membranes seen in mutants with blocks in exit from the Golgi (Figure 4, A and B; Novick et al., 1980; Walch-Solimena and Novick, 1999). Many cells also accumulated fenestrated membranes that resembled membranes seen in two temperature-sensitive mutants that block transport from the Golgi, sec7 and sec14, after short shifts to restrictive temperature (Figure 4, C and D; Rambourg et al., 1993, 1996). These results suggest that the secretory defects observed upon depletion of Avl9p are due to defective transport from the Golgi. This defect is observed only in the vps1 apl2 background with Avl9p depleted; an avl9 strain looks essentially wild type (Figure 4E). Very little abnormal membrane accumulation is observed in the vps1 apl2 double mutant (Figure 4F).

View larger version (168K): [in this window] [in a new window]   Figure 4. The avl9 apl2 vps1 mutant accumulates Golgi-like structures. (A–D) An avl9 apl2 vps1 strain carrying a plasmid with AVL9 under the control of the GAL1 promoter was grown as in Figure 3 to deplete Avl9p. Cells were prepared for thin-section EM after a 20-h shift to glucose. (E) An avl9 strain, prepared as above. (F) The avl9 apl2 vps1 mutant carrying a plasmid with AVL9 under the control of its native promoter, prepared as above. Bars, 500 nm.

The avl9 Mutation Perturbs the Generation of Secretory Vesicles

View larger version (25K): [in this window] [in a new window]   Figure 5. The avl9 sec6-4 mutant accumulates cargo primarily in intermediate-density membranes, similar to what was observed for apl2 sec6-4. Density gradient fractionation was performed as for Figure 1.

View larger version (145K): [in this window] [in a new window]   Figure 6. The avl9 mutation perturbs vesicle formation in a sec6-4 strain background at restrictive temperature. Cells were grown to log-phase at 24°C and shifted to 36°C for 45 min before fixing for thin-section EM. (A–C) avl9 sec6-4 mutant accumulates Golgi-like membranes and vesicles that are smaller than vesicles accumulated in a sec6-4 mutant (D). Bars, 500 nm.

We determined whether actin distribution is perturbed in avl9 mutants by staining cells with fluorescently labeled phalloidin to visualize polymerized actin (Figure 7, A–D). In wild-type cells, actin structures are highly polarized, with patches primarily in small- and medium-sized buds, and cables in mother cells (Adams and Pringle, 1984). This polarized actin distribution was largely lacking in the triple mutant after depleting Avl9p (Figure 7D). A large-scale analysis of yeast gene deletion mutants indicated that depolarized actin is not a general phenotype of sick mutants (Karpova et al., 1998), so this phenotype is likely to be related to Avl9p function. Two other mutants that perturb traffic in the TGN/endosomal system, grd20 and vps54, likewise have depolarized actin distribution (Spelbrink and Nothwehr, 1999; Fiedler et al., 2002). Because secretory traffic helps establish actin polarity (Zhang et al., 2001; Wedlich-Soldner et al., 2003; Aronov and Gerst, 2004), it is not clear whether this phenotype reflects primarily a defect in actin function or a defect in secretion or both. In either case, polarized secretion and growth is clearly perturbed, as indicated by often misshapen cells (Figure 7, D and G).

View larger version (62K): [in this window] [in a new window]   Figure 7. Polarized secretion is defective in avl9 mutants. Cells were stained with Alexafluor-568 phalloidin to label polymerized actin (A–D) or with calcofluor to label chitin (E–I). (A) wt; (B) apl2 vps1: (C) avl9 diploid; (D) avl9 apl2 vps1 after depleting Avl9p, as described in Materials and Methods; (E) apl2 vps1; (F, G) avl9 apl2 vps1 after depleting Avl9p; (H) wt diploid; (I) avl9 diploid. All cells are at the same magnification except the DIC (differential interference contrast microscopy) inset in D. All cells are haploid unless otherwise noted. Arrows indicate abnormal budding pattern in haploid cells.

Overexpression of Avl9p Is Toxic and Results in a Post-Golgi Secretory Defect
In the process of generating the strain for regulatable expression of AVL9, we found that maximum expression of AVL9 from the strong GAL1 promoter resulted in a severe growth defect, indicating that overexpression of the protein is toxic, in either vps1 apl2 cells (not shown) or wild-type cells (Figure 8). We compared growth in 2% galactose with 0, 0.1, 0.2, 0.3, or 0.4% glucose added to reduce AVL9 expression and found that optimum growth was obtained when 0.3% glucose was added to galactose-containing medium.

View larger version (43K): [in this window] [in a new window]   Figure 8. The overexpression of Avl9p is toxic. Strains EHY1252 (carrying a URA3 CEN plasmid vector backbone) and EHY1253 (containing a URA3 CEN plasmid with GAL1p::AVL9) were streaked on CSM, –Ura plates with either 2% galactose or 2% glucose and grown for 3 days (galactose) or 2 days (glucose).

View larger version (117K): [in this window] [in a new window]   Figure 9. The overexpression of Avl9p results in perturbation of membrane organelles. The strains described in Figure 8 were grown as described in Materials and Methods and processed for thin-section EM soon after growth rate slowed down in galactose medium. (A) Wild-type cells grown under these conditions have a dense cytoplasm with small punctate structures. (B) Cells overexpressing Avl9p accumulate heterogeneous vesicles and expanded ER membranes. The images are at the same magnification. Bar, 500 nm.

View larger version (40K): [in this window] [in a new window]   Figure 10. Overexpression of Avl9p results in a post-Golgi secretory defect. (A) The strains described in Figure 8 were grown as described in Materials and Methods, and growth after shift to galactose-containing medium was monitored. A slow-down in growth-rate was detectable by 9 h in galactose for cells with GAL1p::AVL9. (B) Comparison of the hourly growth rate (fold increase in OD600 after 1 h of growth) for cells containing a URA3 CEN plasmid with GAL1p:: AVL9 or an empty URA3 CEN vector. For 3–9 h, n = 6; for 9–15 h, n = 4. Error bars, SEM. (C) Internal Bgl2p was assayed at the indicated times after shift to galactose, as described for Figure 3. (D) Pulse-chase analysis of CPY transport after 14 h in galactose-containing medium, performed as described in Materials and Methods.

Our initial PSI-BLAST searches did not reveal clear paralogues of Avl9 proteins, and motif/domain databases did not indicate high-confidence homology to any previously characterized motifs. Therefore, in order to gain clues about Avl9p function, we performed PSI-BLAST searches using as queries the most conserved regions of Avl9 proteins from representative diverse species. We also performed MEME analyses and MAST searches, as detailed in Materials and Methods, to identify similar regions in potential homologues. This strategy was successful in identifying distantly related protein families that are difficult to align because conserved regions are separated by varying lengths of nonconserved regions. We identified four families of Avl9 paralogues, as well as a domain than is present in diverse proteins and is evolutionarily related to Avl9p and its paralogues. Five similar regions in these proteins were defined using MEME and MAST analyses (Figure 11). A criterion for defining these regions, termed AH (for Avl9 homology), was that they had to be readily identifiable in all or most proteins from diverse phyla for a given protein family. The AH regions have greatly varying distances of separation between, but generally not among, family members. AH3, followed by AH4, is the most conserved region between all Avl9 superfamily proteins, whereas AH5 is generally the most divergent region, although it is predicted to contain almost entirely alpha helices in all families. Although the AH regions were identified by the above methods and did not include predicted structure as a criterion, the regions appear to have similar secondary structures, as predicted by the SAM (Karplus et al., 1998, 2003) and PSIPRED (Jones, 1999