|Home | About | Journals | Submit | Contact Us | Français|
Clathrin-associated adaptor protein (AP) complexes are major structural components of clathrin-coated vesicles, functioning in clathrin coat assembly and cargo selection. We have carried out a systematic biochemical and genetic characterization of AP complexes in Saccharomyces cerevisiae. Using coimmunoprecipitation, the subunit composition of two complexes, AP-1 and AP-2R, has been defined. These results allow assignment of the 13 potential AP subunits encoded in the yeast genome to three AP complexes. As assessed by in vitro binding assays and coimmunoprecipitation, only AP-1 interacts with clathrin. Individual or combined disruption of AP-1 subunit genes in cells expressing a temperature-sensitive clathrin heavy chain results in accentuated growth and α-factor pheromone maturation defects, providing further evidence that AP-1 is a clathrin adaptor complex. However, in cells expressing wild-type clathrin, the same AP subunit deletions have no effect on growth or α-factor maturation. Furthermore, gel filtration chromatography revealed normal elution patterns of clathrin-coated vesicles in cells lacking AP-1. Similarly, combined deletion of genes encoding the β subunits of the three AP complexes did not produce defects in clathrin-dependent sorting in the endocytic and vacuolar pathways or alterations in gel filtration profiles of clathrin-coated vesicles. We conclude that AP complexes are dispensable for clathrin function in S. cerevisiae under normal conditions. Our results suggest that alternative factors assume key roles in stimulating clathrin coat assembly and cargo selection during clathrin-mediated vesicle formation in yeast.
Selective protein transport between membrane organelles is mediated by transport vesicles. Formation of such vesicles depends on recruitment of evolutionarily conserved multimeric protein complexes to the cytoplasmic aspect of organelle membranes (Schekman and Orci, 1996 ). Recruited complexes assemble into membrane-associated coats that propel membrane invagination and orchestrate cargo selection, leading to generation of coated transport vesicles. A major class of coated transport vesicles is distinguished by clathrin coats. Clathrin coats at the plasma membrane and trans Golgi network (TGN) give rise to endosome-targeted vesicles, and clathrin may also participate in vesicle formation at endosomes (Schmid, 1997 ).
The major structural components of clathrin coats are two protein complexes, clathrin and clathrin adaptor proteins (APs) (Schmid, 1997 ; Hirst and Robinson, 1998 ). Clathrin is a tripod-shaped molecule, with each leg composed of a heavy chain and an associated light chain (Kirchhausen and Harrison, 1981 ; Ungewickell and Branton, 1981 ; Pishvaee and Payne, 1998 ; Musacchio et al., 1999 ). Clathrin assembles into a polyhedral lattice that forms the outer shell of the coat (Vigers et al., 1986a ,b ; Smith et al., 1998 ). The heterotetrameric APs bridge the clathrin lattice to the membrane. Purification of mammalian clathrin-coated vesicles revealed two related AP complexes, AP-1 and AP-2 (Pearse and Robinson, 1984 ; Keen, 1987 ). AP-1 localizes to the TGN and endosomes, whereas AP-2 localizes to the plasma membrane (Robinson, 1987 ; Ahle et al., 1988 ). Each complex contains two large subunits (~100 kDa; γ and β1 in AP-1, α and β2 in AP-2), one medium subunit (~50 kDa; μ1 in AP-1, μ2 in AP-2) and one small subunit (~20 kDa; ς1 in AP-1, ς2 in AP-2) (Schmid, 1997 ; Hirst and Robinson, 1998 ). The highly similar β subunits bind to clathrin and promote clathrin coat assembly (Gallusser and Kirchhausen, 1993 ). The μ and β subunits interact with sorting signals in the cytoplasmic domains of transmembrane proteins, thereby collecting appropriate vesicle cargo (Ohno et al., 1995 ; Rapaport et al., 1998 ). The AP-2 α subunit, and by analogy the AP-1 γ subunit, appear to be important in recruiting additional factors necessary for clathrin-coated vesicle formation (Benmerah et al., 1995 ; Wang et al., 1995 ; David et al., 1996 ; Wigge et al., 1997a ,b ; Chen et al., 1998 ; Owen et al., 1999 ). Through these combined activities, AP complexes are thought to play a central role in clathrin-coated vesicle formation by coupling coat assembly and cargo collection.
A more widespread role for AP complexes in protein sorting is evident from recent discoveries of mammalian heterotetrameric complexes related to AP-1 and AP-2. AP-3, which is associated with endosomes and/or the TGN, plays a role in membrane protein sorting to lysosomes and synaptic vesicle formation (Le Borgne and Hoflack, 1998 ; Odorizzi et al., 1998 ). Whether AP-3 acts with clathrin has not been resolved (Simpson et al., 1996 , 1997 ; Dell’Angelica et al., 1998 ). Initial characterization of AP-4 indicates that this complex localizes to the vicinity of the TGN but does not appear to interact with clathrin (Dell’Angelica et al., 1999a ). The function of AP-4 has not been addressed.
The complete genome sequence of Saccharomyces cerevisiae allows a systematic approach to investigate AP function. Database searches using mammalian AP subunit sequences indicate that the S. cerevisiae genome has the potential to encode three AP β subunits, three non-β large subunits, four μ subunits, and three ς subunits (Table (Table1)1) (Cowles et al., 1997a ; Panek et al., 1997 ). Based on the degree of primary sequence similarity between each yeast protein and the different mammalian AP subunits, the yeast proteins can be grouped into three potential AP complexes, leaving Apm2p unassigned (Cowles et al., 1997a ; Panek et al., 1997 ). Biochemical analyses and phenotypic characterization of strains carrying gene disruptions defined a yeast AP-3 complex involved in clathrin-independent traffic from the Golgi apparatus to vacuoles (Table (Table1)1) (Cowles et al., 1997a ; Panek et al., 1997 ; Stepp et al., 1997 ; Vowels and Payne, 1998a ). Surprisingly, deletion of several other AP subunit genes yielded no detectable phenotypes, even though the subunits exhibit substantial evolutionary conservation with their mammalian counterparts (up to 50% identity) (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). However, disruption of APS1, APM1, or APL2 specifically enhanced growth and Golgi-related protein sorting defects in cells carrying a temperature-sensitive allele of the clathrin heavy chain gene (chc1-ts). These results offer genetic evidence for an AP-1-like complex, consisting of Aps1p, Apm1p, and Apl2p, that is involved in clathrin-dependent function at the Golgi apparatus. The fractionation properties of selected AP proteins, including the putative AP-1 subunits, is consistent with organization into multimeric complexes, but the composition of such complexes has not been addressed except for AP-3 (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ).
Here we present a more comprehensive biochemical and genetic characterization of yeast AP complexes. Our results assign subunits to one of three distinct complexes, defined by the three β subunits. Only the AP-1 complex physically and genetically interacts with clathrin. Surprisingly, in cells expressing wild-type clathrin, combined deletion of genes encoding all four AP-1 subunits, or deletion of the three β-encoding genes, does not affect clathrin-dependent trafficking processes or reduce the level of clathrin-coated vesicles. These findings suggest that clathrin function in yeast does not depend on AP complexes.
Unless noted, all reagents were purchased from Sigma (St. Louis, MO).
Plasmid constructions were carried out using standard molecular biology techniques (Sambrook et al., 1989 ). pBKS-URA3 contains a 1.1-kb HindIII fragment of URA3 (Rose et al., 1984 ) inserted as a blunt-ended fragment into the SmaI site of pBlusescript KS+ (pBKS+; Stratagene, La Jolla, CA). pBKS-TRP1 contains a 1.0-kb SspI–StuI fragment of TRP1 (Tschumper and Carbon, 1980 ) inserted as a blunt-ended fragment into the SmaI site of pBKS+. YEp352-APL2 contains a 3.8-kb EcoRI–SnaBI fragment of APL2 (Rad et al., 1995 ) in YEp352 (Hill et al., 1986 ). PCR amplifications were carried out with either Deep Vent (New England Biolabs, Beverly, MA) or Elongase (Life Technologies, Rockville, MD). Primers are listed in Table Table2.2. All PCR products were sequenced to confirm accurate amplification.
The 5′ region of APL1 (bp 1–347; bp 1 corresponds to the A in the initiating ATG) was amplified by PCR from pAPL1–100 (Rad et al., 1995 ) using a 5′ primer (primer 5) homologous to APL1 bp 1–16 and containing an NcoI site and a 3′ primer (primer 6) homologous to APL1 bp 349–363 with an XhoI site. The resulting product was subcloned into the NcoI and XhoI sites in pGEX-KG. A 2.5-kb BglII–HindIII fragment containing the remaining 3′ coding region and downstream sequences of APL1 was transferred from pAPL1-100 to create pGEX-KG-APL1. HA-APL1 was generated by amplifying a tandem repeat of the hemagglutinin (HA) epitope from pGDA-4HA (a gift from Jennifer Vowels, University of California, Los Angeles, CA) with a 5′ primer (primer 3) that contains a BamHI site and a 3′ primer (primer 7) carrying EcoRI and NcoI sites. The tandem tag fragment was subcloned into the BamHI and EcoRI sites of pBKS+. The tandem HA fragment was then transferred to the BamHI and NcoI sites in pGEX-KG-APL1, creating pGEX-KG-HA-APL1. HA-APL1 was excised as a 2.5-kb BamHI–HindIII fragment and introduced into pRS305 (Sikorski and Hieter, 1989 ) to create pRS305-HA-APL1. A 360-bp region upstream of the APL1 ATG was amplified from YAP100-1 with primer 8, containing a NotI site and (primer 9) with a BamHI site. This fragment was subcloned into pRS305-HA-APL1 with NotI and BamHI to create pRS305-ProHA-APL1.
The 5′ region of APL2, (bp 1–464; same numbering as APL1) was amplified by PCR from YEp352-APL2 using a 5′ primer (primer 1) homologous to APL2 bp 1–20 and containing a 5′ XbaI recognition site and a 3′ primer (primer 2) homologous to APL2 bp 445–465 with a 5′ SalI site. The resulting product was subcloned into the XbaI and SalI sites in pEG-KG (Mitchell et al., 1993 ). A 2.9-kb NdeI–SalI fragment containing the remaining 3′ coding region and downstream sequences of APL2 was transferred from YEp352-APL2 to create pEG-KG-APL2. A 3.3-kb XbaI fragment from pEG-KG-APL2 containing full-length APL2 was subcloned into pGEX-KG (Guan and Dixon, 1991 ). HA-APL2 was generated by amplifying a tandem repeat of the HA epitope from pGDA-4HA with a 5′ primer (primer 3) carrying a BamHI site and a 3′ primer (primer 4) with XhoI and XbaI sites. The tandem tag fragment was subcloned into the BamHI and XhoI sites of pBKS+. The HA fragment was then transferred to the BamHI and XbaI sites in pGEX-KG-APL2, creating pGEX-KG-HA-APL2. HA-APL2 was excised as a 3.3-kb BamHI–XhoI fragment and introduced into pRS305 to create pRS305-HA-APL2. A 428-bp region upstream of the APL2 ATG was amplified from YEp352-APL2 with primer 5 containing a NotI site and primer 6 containing a BamHI site. This fragment was subcloned into pRS305-HA-APL2 with NotI and BamHI to create pRS305-ProHA-APL2.
The 3′ region of APL4, bp 1825–2496, was amplified with primer 14 (BamHI site) and primer 15 (EcoRV site). The resulting fragment was subcloned into pBS14, which carries the HA epitope coding sequence (a gift from T. Kirchhausen, Harvard University Medical School, Boston, MA), creating pAPL4-HA. A BamHI–HindIII (filled in) fragment from pAPL4-HA was subcloned into the BamHI and SacII (filled in) sites of pBKS-URA3, creating pAPL4-HA-URA3. pAPL4-HA-URA3 was cleaved with MunI to integrate APL4-HA into the chromosome copy of APL4.
The 5′ region of APL6 (bp 1–396; same numbering as APL1) was amplified by PCR from YKS5 (Panek et al., 1997 ) using a 5′ primer (primer 10) homologous to APL6 bp 1–19 and containing a NcoI site and a 3′ primer (primer 11) homologous to APL6 bp 381–396 and containing a XhoI site. The resulting product was subcloned into the NcoI and XhoI sites of pGEX-KG. A 3.1-b SacII–SacI YKS5 fragment was then introduced, creating pGEK-KG-APL6. HA-APL6 was generated by transferring the tandem HA tag from pBKS+ (see above) to the BamHI and NcoI sites in pGEX-KG-APL6, creating pGEX-KG-HA-APL6. HA-APL6 was excised as a BamHI–HindIII fragment and introduced into pRS315 (Sikorski and Hieter, 1989 ) to create pRS315-HA-APL6. A 360-bp region upstream of the APL6 ATG (bp −360 to 1) was amplified from YKS5 with a 5′ primer (primer 12) containing an XhoI site and a 3′ primer (primer 13) containing a BamHI site. This fragment was subcloned into pRS315-HA-APL6 with XhoI and BamHI to create pRS315-ProHA-APL6.
A BamHI–SmaI fragment from pBKS-URA3 was subcloned into BamHI–EcoRV sites of pAPM1 (described below), creating pAPM1-HA-URA3. pAPM1-HA-URA3 was cleaved with EcoRI to integrate the APM1-HA into the chromosome copy of APM1.
A 3′ fragment of APM4, beginning 775 bp region upstream of the stop codon was amplified with primer 16 homologous to bp 701–720 and containing a BamHI site and primer 17 homologous to bp 1476–1542 that was designed to lack the endogenous stop codon and contain a PvuII site. The PCR product was cleaved with BamHI and PvuII and subcloned into BamHI–EcoRV site of pBS14, creating pAPM4-HA. To generate pAPM4-HA-URA3, a BamHI–HindIII (filled in) fragment was cloned into pBKS-URA3 cut with SacII (filled in) and BamHI. pAPM4-HA::URA3 was cut with MunI to integrate the APM4-HA into the chromosome copy of APM4.
APM1 was amplified from genomic DNA with primer 20 homologous to 288 bp upstream of APM1 and containing a BamHI site and primer 21 homologous to APM1 bp 1405–1422 and containing a PvuII site. The product was cut with BamHI–PvuII and cloned into BamHI and EcoRV sites of pBS14, creating pAPM1. The URA3 gene was transferred from pBKS-URA plasmid as a BamHI (filled in) and EcoRI (filled in) fragment into PstI (filled in) and EcoRV (filled in) sites of pAPM1, creating papm1::URA3.
The 5′ region of APL3, 693 bp upstream of the ATG to 197 bp upstream of the ATG, was amplified with primers 22 and 23. The resulting fragment was cut with KpnI and ClaI to release a 322-bp fragment that was subcloned into pBKS-URA3, creating pBKS-URA3–5′APL3. The 3′ region of APL3, bp 2367–3075, was amplified with primer 20 (BamHI site) and primer 21 (EcoRV site). The amplified fragment was subcloned into the BamHI and EcoRV sites of pBKS-URA3–5′APL3 creating papl3::TRP1.
The 5′ region of APL4, from −557 to + 69 bp with bp 1 corresponding to the A in the initiating ATG, was amplified with primers 18 and 19. The product was digested with HindIII (filled in) and subcloned into pBKS-TRP1 at the EcoRI site (filled in), creating pBKS-TRP-5′APL4. A 3′ APL4 BamHI–HindIII (filled in) fragment from pAPL4-HA-URA3 was subcloned into BamHI and SacII (filled in) sites of pBKS-TRP-5′APL4, creating papl4::TRP.
The C-terminal portion of APL1, bp 1858–2060, was amplified from YAP100-1 with primer 26 (NcoI site) and primer 27 (SacI site). The resulting product was cloned into pGEX-KG, creating pGEX-KG-APL1 C-term (Apl1p amino acids 620–701).
The C-terminal portion of APL2, bp 1411–1910, was amplified from YEp352-APL2 with primer 28 (NcoI site) and primer 29 (XhoI site). The resulting fragment was subcloned into pGEX-KG. A 889-bp HindIII fragment containing the remaining 3′ coding region and downtream sequences of APL2 was transferred from YEp352-APL2, creating pGEX-KG APL2 C-term (Apl2p amino acids 471–727).
The C-terminal portion of APL6, bp 1621–2120, was amplified from YKS5 with primer 24 (EcoRI site) and primer 25 (NcoI site). The resulting product was cloned into pGEX-KG. A 1.54-kb NdeI–SacI fragment containing the remaining 3′ coding region and downstream sequences of APL6 was transferred from YKS5, creating pGEX-KG-APL6 C-term (Apl6p amino acids 541–810).
Genotypes of strains used in this study are listed in Table Table3.3. Yeast mating, sporulation, and tetrad analyses were conducted as described by Sherman et al. (1974) . DNA transformations were performed by the lithium acetate procedure (Ito et al., 1983 ).
GPY1100 was generated from GPY1100a by mating type switching with plasmid-borne HO (Payne and Schekman, 1989 ). Similarly, GPY404.1 was generated from SEY6210. All mutant or plasmid-carrying strains are congenic with either GPY1100 or SEY6210. GPY1599–23D is a meiotic progeny from a cross of GPY1422 and GPY1357. GPY1627-2C is a meiotic progeny from a cross of GPY1422 and GPY1354. GPY1783-21D, 1783-10A, 1783-25A, and 1783-21C are meiotic progeny from a cross of GPY1705.1 and GPY404.1. Single-step gene replacement (Rothstein, 1991 ) was carried out with papl4::TRP1 cleaved with KpnI and SacI, papm1::URA3 cleaved with BamHI and EcoRI, and papl3::TRP1 cleaved with ClaI and SacI. All gene replacements were verified by Southern blotting or immunoblotting. With all HA constructs, immunoblotting with HA-specific antibodies detected species of the expected size, which were absent in strains lacking HA tags.
YP medium is 1% Bacto-yeast extract and 2% Bacto-peptone. YPD medium is YP with 2% dextrose. SD medium is 0.67% yeast nitrogen base (Difco, Detroit, MI) and 2% dextrose. Supplemented SD is SD with 20 μg/ml histidine, uracil, and tryptophan and 30 μg/ml leucine, adenine, and lysine. SDYE is SD with 0.2% yeast extract. Cell densities in liquid culture were measured in a 1-cm plastic cuvette using a Beckman Instruments (Palo Alto, CA) DU62 spectrophotometer. One OD500 unit is equivalent to 2.3 × 107 cells/ml.
To assess growth on agar plates, cells were grown in YPD to stationary phase, diluted to 1 × 106 cells/ml, and further diluted 1:10 or 1:100. Three microliters of each of these dilutions were spotted onto YPD plates and incubated at 24, 30, or 37°C.
Cells were grown to OD500 of 0.5–1.0. Fifty OD500 units of cells were harvested and resuspended in 1 ml of 100 mM Tris-SO4 pH 9.5, and 10 mM DTT and incubated for 10 min at 30°C. Cells were pelleted and resuspended in 1 ml of YP, 1.0 M sorbitol, and 0.5% glucose and converted to spheroplasts by addition of 16 μl of 1 mg/ml oxalyticase (Enzogenetics, Eugene, OR) and incubation for 30 min at 30°C. Spheroplasts were lysed by resuspension in 0.5 ml of ice-cold PBS, 1% Triton X-100, 1 mM EDTA, and 2× PIC (1000× PIC contains 100 mM N-tosyl-l-phenyl-alanine-chloromethyl ketone, 1 M benzamidine-HCl, 25 mM pepstatinA, 4 mM leupeptin, and 1 M 4-(2 aminoethyl)-benzene sulfonyl-fluoride). The lysate was clarified by centrifugation at 16,000 × g for 10 min at 4°C and then transferred to a fresh tube containing 25 μl of a 20% suspension of protein A-Sepharose (Pharmacia, Piscataway, NJ) and appropriate antibody. For precipitations with antibodies against Apl1p, Apl2p, or Apl6p, 125 OD500 cell equivalents/ml of lysate were used. For APM1-HA coimmunoprecipitations, 12CA5 antibody was coupled to protein A-Sepharose with dimethylpimelimidate (Harlow and Lane, 1988 ).
For metabolic labeling of α-factor, cells were grown to midlogarithmic phase in SDYE at 24°C. Cultures were shifted to 24° or 30°C for 2 h. Labeling and immunoprecipitation were performed as described previously (Seeger and Payne, 1992a ), except that labeling was for 45 min instead of 10 min. For metabolic labeling of CPY, cells were grown to midlogarithmic phase in SDYE at 30°C. Cultures were resuspended in supplemented SD and shifted to 30°C for 5 min. Labeling and immunoprecipitation was performed as described previously (Seeger and Payne, 1992b ).
HA-specific monoclonal antibody 12CA5 was a gift from G. Weinmaster (University of California School of Medicine, Los Angeles, CA); monoclonal antibodies to yeast Chc1p and polyclonal antibodies to Apm2p and Apm3p were a gift from S.K. Lemmon (Case Western Reserve University, Cleveland, OH); antibodies to carboxypeptidase Y (CPY) and Apl6p were a gift from S.D. Emr (University of California, San Diego, CA); and antibodies to aminopeptidase I (API) were a gift from D. Klionsky (University of California, Davis, CA).
To generate Apl1p antibody, the C terminus of APL1, bp 1564–2060, was amplified by PCR from pAPL1–100. The 5′ primer (primer 30) contains an NdeI site, a start codon (ATG), and six additional histidines (CAC) fused in frame with APL1. The 3′ primer (primer 31) contains a BamHI site. The resulting product was subcloned into pET3c (Studier et al., 1990 ). The APL1 C terminus was then introduced as an AflII–HindIII fragment to generate pHIS-APL1 Cterm (Apl1p amino acids 523–701). Expression of pHIS-APL1 Cterm in Escherichia coli strain BL21 (DE3) was induced with 0.1 mM isopropyl thiogalactoside and lysed as previously described (Phan et al., 1994 ). Histidine-tagged protein was purified by nickel-nitrolotriacetic acid affinity chromatography (Qiagen, Chatsworth, CA) as described by Bush et al. (1991) , with the following modifications. The bacterial cell pellet was resuspended in 30 ml of buffer A; the lysate was centrifuged at 10,000 × g for 20 min, and buffers B and C contain 0.5% Triton X-100. The elutions with 10 ml of buffer D were followed with elutions of 10 ml of buffer E (8 M urea, 0.1 M NaH2PO4, and 0.01 M Tris-HCl, pH 4.5). Two-milliliter fractions were collected, and fractions 2 and 3 from the buffer E elution were pooled and dialyzed stepwise from 8 M urea into 6 M urea, 4 M urea, 2 M urea, and finally into PBS, 10% glycerol, and 10 mM DTT. The sample was used as antigen for commercial production of antibody in rabbits (Cocalico Biologicals, Reamstown, PA).
To generate Apl2p antibody, pGEX-KG-APL2 C-term was expressed in BL21 (DE3) strain as described above. Cell lysis and fusion protein affinity purification with glutathione-Sepharose (Pharmacia) were carried out as recommended by the manufacturer. Purified fusion protein was used as antigen for commercial production of antibody in rabbits (Cocalico Biologicals).
Antibodies against Apl1p were affinity purified using GST-APL1 fusion protein coupled to cyanogen bromide-Sepharose 4B (Pharmacia) according to the method of Harlow and Lane (1988) .
Immunoblotting was carried out according to the method of Burnette (1981) with secondary antibodies coupled to alkaline phosphatase (ALP; Bio-Rad, Richmond, CA) or coupled to horseradish peroxidase (HRP; Bio-Rad). Antibodies were visualized using color development for ALP (Bio-Rad) or epichemiluminescence (New England Nuclear, Boston, MA) for HRP (Pharmacia).
pGEX-KG-APL1 C-term, pGEX-KG-APL2 C-term, and pGEX-KG-APL6 C-term were expressed in BL21 (DE3), and GST fusion proteins were affinity purified with glutathione-Sepharose. For preparation of yeast extract, wild-type strain TVY 614 was grown to midlogarithmic phase in YPD. Eight hundred fifty OD500 units of cells were converted to spheroplasts and resuspended at 85 OD500/ml in ice-cold 20 mM HEPES, pH 7.2, 0.1 M KCl, 2 mM MgCl2, 1 mM DTT, 1% Trition X-100, and 2× PIC. Cells were further lysed by 20 strokes of a Dounce homogenizer. After centrifugation at 27,000 × g for 30 min, the supernatant was applied to 250 μl of a 50% suspension of glutathione-Sepharose carrying GST fusion proteins and incubated for 2 h at 4°C with rotation. Fusion proteins and associated proteins were eluted by three consecutive treatments with 125 μl of reduced glutathione buffer (20 mM reduced glutathione, 100 mM Tris-HCl, pH 9.0, 200 mM NaCl, 5 mM DTT, and 0.1% Triton X-100).
Clathrin-coated vesicles were enriched by differential centrifugation and gel filtration chromatography of the high-speed pellet fraction (100,000 × g for 60 min) as previously described (Chu et al., 1996 ). Fractions were precipitated by addition of 10% trichloroacetic acid and subjected to SDS-PAGE followed by immunoblotting with monoclonal antibodies to detect Chc1p and polyclonal antibodies to detect Kex2p.
Specific genetic interactions with chc1-ts have led to the proposal that Aps1p, Apm1p, and Apl2p (β1) are associated in an AP-1 complex that functions with clathrin at the Golgi apparatus (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). To investigate the physical association of these AP subunits and identify the non-β large subunit of the presumptive AP-1 complex, selected subunits were immunoprecipitated under nondenaturing conditions and probed for associated AP proteins by immunoblotting. We were particularly interested in monitoring Apl4p, because this protein is most similar in sequence to the γ large subunit of mammalian AP-1. For this purpose a strain was constructed in which a functional version of Apl4p tagged with the influenza HA epitope was integrated at the chromosomal APL4 locus. Whole-cell extracts were prepared by lysis with 1% Triton X-100 to provide a population of AP complexes representative of both soluble and membrane-associated pools. Extract was incubated with antibodies specific for Apl2p (β1), and the resulting immunoprecipitate was analyzed by SDS-PAGE and immunoblotting. Both Aps1p and Apl4p-HA were coprecipitated with Apl2p (β1) (Figure (Figure1A,1A, compare lane 3 with the total extract in lane 1). In contrast, Aps2p was not precipitated, providing evidence for the specificity of coprecipitation (Figure (Figure1A,1A, lane 3). These experiments were carried out with antigen in excess, resulting in immunoprecipitation of ~10% of the AP subunits. Under conditions of antibody excess, essentially all of the Aps1p present in the extract was coprecipitated with Apl2p (Yeung, unpublished results). The specificity of small subunit interaction with Apl2p (β1) was sufficiently stringent that even in cells completely lacking the Aps1p subunit because of deletion of APS1, Aps2p was not detected in precipitates of Apl2p (Yeung, unpublished results).
To monitor interactions with Apm1p, a strain expressing functional Apm1p-HA was lysed, and Apm1p was immunoprecipitated with HA-specific monoclonal antibody. Immunoblotting of the resulting precipitates revealed association with Apl2p (β1) and Aps1p (Figure (Figure1B,1B, lane 3) but not Apl1p (β) or Aps2p (Figure (Figure1B,1B, lane 6). Detection of Apl2p (β1) and Aps1p was dependent on precipitation of Apm1p, because neither protein was precipitated by HA antibodies when extracts were used from cells expressing Apm1p without the HA tag (Figure (Figure1B,1B, lanes 2 and 5). These results suggest that the AP-1 complex consists of Aps1p (ς1), Apm1p (μ1), Apl4p (γ), and Apl2p (β1).
A similar coimmunoprecipitation strategy was applied to characterize the complex containing the Apl1p (β) subunit. Aps2p, but not Aps1p or Apl4p-HA, was coprecipitated with Apl1p (β) (Figure (Figure1A,1A, lane 2). To determine whether Apl3p is the non-β large subunit associated with Apl1p (β), a variation of the coprecipitation strategy was adopted. The approach was based on analyses of mammalian and yeast AP complexes, which demonstrate that loss of single AP subunits can dramatically reduce the ability of the remaining subunits to form a stable complex (Panek et al., 1997 ; Dell’Angelica et al., 1999b ). Accordingly, we monitored the effect of deleting APL3 (apl3Δ) on association of Aps2p with immunoprecipitated Apl1p (β). The absence of Apl3p eliminated Aps2p interaction with Apl1p (β) observed in wild-type cells (Figure (Figure1C,1C, compare lane 3 with lane 1). The effect of apl3Δ was specific for the Apl1p (β)-Aps2p association, because the AP-1 complex was unaffected by the deletion as assessed by coprecipitation of Aps1p (ς1) with Apl2p (β1) (Figure (Figure1C,1C, lanes 2 and 4). Apm4p was assigned to the Apl1p (β) complex based on coprecipitation of Apl1p (β) and Aps2p with an HA-tagged version of Apm4p (Figure (Figure1D,1D, lane 1). The specificity of these interactions was evident from the absence of AP-1 subunits (Apl2p and Aps1p) in the immunoprecipitate of Apm4p-HA and the corresponding absence of Apl1p (β) and Aps2p in immunoprecipitates of Apm1p-HA (Figure (Figure1D,1D, lane 2, also see B). The lower-molecular-weight band in the Figure Figure1D,1D, lane 2, upper panel, is most likely a degradation product of Apl2p (β1) (see Figure Figure1E,1E, lane 2). These results group Aps2p (ς2R), Apm4p (μ2R), Apl3p (αR), and Apl1p (β2R) in a distinct complex that we term AP-2R, because of the prevailing sequence similarity of the yeast subunits with mammalian AP-2 subunits. The “R” is attached to signify “related” because, unlike the established connection between the mammalian AP-2 complex and clathrin-mediated endocytosis, there is no physical or genetic evidence linking AP-2R to clathrin or endocytosis (see below).
Apm2p is unusual because it is significantly larger (~70 kDa) than the other three μ subunits (~50 kDa), and it is less conserved with mammalian μ chains (Cowles et al., 1997a ; Panek et al., 1997 ). Association of Apm2p with AP β subunits was investigated by coimmunoprecipitation using HA antibody to precipitate HA-tagged versions of Apl1p (β2R), Apl2p (β1), or Apl6p (β3). Using extracts from wild-type cells we were unable to detect reproducible association with any of the β subunits, but occasionally Apm2p appeared to be coprecipitated with Apl2p (β1). To increase the sensitivity of the assay, the immunoprecipitation was repeated with extracts from cells expressing Apm2p at elevated levels from a multicopy plasmid. Apm2p in these extracts was specifically coprecipitated with Apl2p (β1) (Figure (Figure1E).1E). Conversely, Apl2p (β1) and Aps1p were preferentially precipitated with Apm2p antibodies (Yeung, unpublished results). Because of the unique properties of Apm2p, we considered the possibility that the protein was peripherally associated with intact AP-1 (containing Apm1p) rather than an integral part of a separate Apl2p-containing complex. As an approach to distinguish between these alternatives, we investigated whether overexpressed Apm2p could be coprecipitated with Apm1p-containing AP-1 complexes. For this purpose, the multicopy APM2 plasmid was introduced into a strain expressing Apm1p-HA, and AP-1 was immunoprecipitated from extracts of these cells with HA antibody. Although AP-1 subunit Aps1p (ς1) was coprecipitated with Apm1p-HA, no associated Apm2p was detected (Figure (Figure1F,1F, lane 2). In contrast, a parallel precipitation of Apl2p (β1)-HA coprecipitated Apm2p and Aps1p (Figure (Figure1F,1F, lane 1). This finding suggests that Apm2p is able to interact with β1, potentially as part of an alternative AP-1-like complex lacking Apm1p.
In mammalian cells, β subunits of AP-1, AP-2, and AP-3 interact with clathrin in vitro (Gallusser and Kirchhausen, 1993 ; Dell’Angelica et al., 1998 ). However, in yeast, deletions of AP-1 β or ς subunits, but not cognate AP-2R or AP-3 subunits, display genetic interactions with chc1-ts, raising the possibility that only yeast AP-1 interacts with clathrin (Phan et al., 1994 ; Rad et al., 1995 ; Panek et al., 1997 ). To address this possibility we examined physical interactions of yeast AP complexes with clathrin using in vitro binding assays and coimmunoprecipitation.
For in vitro binding experiments, N-terminal truncations of β subunits were fused to GST and expressed in E. coli. These truncated versions were selected because clathrin-binding sites in mammalian β subunits are located toward the C termini (Kirchhausen, 1990 ; Shih et al., 1995 ; Dell’Angelica et al., 1998 ), and initial attempts to express GST fused to full-length yeast β subunits resulted in insoluble proteins. Each GST-yeast β fusion was bound to glutathione-Sepharose and then incubated with extract from a wild-type yeast strain. Bound proteins were eluted with reduced glutathione and were analyzed by SDS-PAGE followed by immunoblotting with clathrin heavy chain antibodies or staining with Coomassie brilliant blue. As shown in Figure Figure2A,2A, GST-Apl2p (β1) bound clathrin heavy chain (Figure (Figure2A,2A, lane 2). Specificity of binding was apparent from Coomassie blue staining of the bound fraction, which revealed the clathrin heavy chain to be the single major protein larger than the 58.5-kDa fusion protein when compared with the starting extract (Figure (Figure2B,2B, lanes 1 and 2). Neither GST-Apl1p (β2R) nor GST-Apl6p (β3) was found to bind clathrin (Figure (Figure2,2, A and B, lanes 3 and 4). No other major high-molecular-weight species larger than the fusion proteins were detected in the bound fractions by Coomassie blue staining (Figure (Figure2B,2B, lanes 3 and 4). The identity of bands migrating faster than the fusions have not been addressed but could represent degradation products from the fusions. These results suggest that only Apl2p (β1) interacts with clathrin.
As an alternative approach to assess clathrin binding by AP complexes, native immunoprecipitations of each AP complex were probed for associated clathrin heavy chain. AP complexes were immunoprecipitated from extracts of cells expressing HA-tagged β subunits with polyclonal antibodies directed against the β subunits. These antibodies are known to recognize the native AP complexes (Figure (Figure1;1; Yeung, unpublished results). Immunoblotting with HA-specific antibodies indicated that approximately equal amounts of each AP complex were precipitated, but clathrin was associated only with AP-1 (Figure (Figure2C).2C). Thus, clathrin interacts selectively with AP-1 by both coimmunoprecipitations and GST fusion binding assays. Together with results from studies of genetic interactions between AP subunit deletions and chc1-ts (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ) our findings argue that AP-1 is the sole clathrin-associated adaptor of the three yeast AP complexes.
Earlier studies failed to detect growth or protein trafficking defects in strains carrying deletions of AP-1 β, μ, or ς subunits or a combination of β and ς subunits (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). However, the same AP-1 subunit deletions accentuate growth and protein trafficking defects in chc1-ts cells. The effects are specific to AP-1 subunit deletions; AP-2R and AP-3 mutations do not interact with chc1-ts (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ; Panek et al., 1997 ). To address the possibility that the subtle effects of AP-1 single and double subunit deletions are attributable to residual activity of partial complexes, we examined growth, protein sorting, and clathrin-coated vesicle formation in cells (ap1-null) lacking the four AP-1 subunits, Apl4p (γ), Apl2p (β1), Apm1p (μ1), and Aps1p (ς1).
Growth was monitored by incubating serial dilutions of cells on agar plates at 24, 30, or 37°C. Wild-type and ap1-null strains grew at the same rate at all three temperatures, indicating that the absence of AP-1 does not perturb growth (Figure (Figure3,3, rows 1 and 7). We also compared the effects of AP-1 single subunit deletions with the AP-1-null combination in a congenic set of chc1-ts strains. In agreement with previous findings (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ), apl2 (β1), apm1, or aps1 reduced the ability of chc1-ts cells to grow at 37°C but not at lower temperatures (Figure (Figure3,3, rows 2, 3, 5, and 6). Deletion of APL4 (γ) in the chc1-ts strain caused a similar defect (Figure (Figure3,3, row 4). Although not readily apparent in Figure Figure3,3, limited growth of apl4 (γ) chc1-ts and aps1 chc1-ts cells was observed at the highest cell densities at 37°C. No growth of apl2 (β1) chc1-ts and apm1 chc1-ts cells was observed at 37°C, suggesting that loss of β or μ AP-1 subunits is slightly more deleterious to growth of chc1-ts cells than loss of γ or ς subunits. The growth properties of the ap1-null chc1-ts strain mirrored those of the apl2 (β1) chc1-ts and apm1 chc1-ts strains (Figure (Figure3,3, rows 3, 5, and 8).
Cells with mutations in clathrin subunits secrete a highly glycosylated precursor form of the α-factor mating pheromone (Payne and Schekman, 1989 ; Seeger and Payne, 1992b ; Chu et al., 1996 , 1999 ; Huang et al., 1997 ). This defect is attributed to mislocalization of the Golgi membrane protein Kex2p, which normally initiates proteolytic maturation of the pheromone precursor in the TGN (Fuller et al., 1988 ). In the absence of clathrin function, Kex2p is mislocalized to the plasma membrane, and the resulting depletion of TGN Kex2p allows some fully glycosylated precursor to avoid proteolytic maturation (Payne and Schekman, 1989 ; Seeger and Payne, 1992b ). Thus, the level of secreted highly glycosylated α-factor precursor provides a convenient measure of Kex2p localization. Previous studies indicated that effects of single and double AP subunit deletions on α-factor maturation generally parallel effects on growth; in both cases defects are apparent only when AP-1 subunits are deleted in combination with chc1-ts cells (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). However, α-factor maturation is a more sensitive assay than growth, because effects of AP-1 subunit deletions on maturation in chc1-tscells can be detected at a temperature (24°C) at which growth is unaffected. Accordingly we assessed α-factor maturation in cells carrying various combinations of AP-1 subunit deletions and chc1-ts.
Cells were labeled with [35S]methionine and cysteine at 24 or 30°C, α-factor was immunoprecipitated from the medium, and the immunoprecipitate was analyzed by SDS-PAGE. Maturation was complete in wild-type cells at either temperature (Figure (Figure4,4, lanes 8 and 17) and virtually complete in chc1-ts cells at 24°C (Figure (Figure4,4, lane 9). No maturation defect was detected in the ap-1-null mutant at either temperature (Figure (Figure4,4, lanes 7 and 16) or in cells with single AP-1 subunit deletions (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ; Phan and Yeung, unpublished results). However, even at the permissive temperature for chc1-ts, elimination of individual AP-1 subunits in chc1-ts cells resulted in secretion of precursor α-factor (Figure (Figure4,4, lanes 1–4 and 9). Deletion of the APS1 or APL4 (γ) resulted in slight maturation defects (Figure (Figure4,4, compare lanes 1 and 4 with lane 9). Deletion of APM1 had a greater effect, and deletion of APL2 (β1) produced the most severe defect (Figure (Figure4,4, lanes 2 and 3). Analysis of Kex2p in the apl2 (β1) chc1-ts strain at 24°C confirmed that the α-factor maturation defect was accompanied by Kex2p mislocalization (Phan, unpublished results). We considered two interpretations of the observation that the apl2 (β1) chc1-ts strain exhibited the most pronounced defect. Either β1 is the most important subunit for AP-1 function (at least in α-factor maturation), or the absence of β1 results in a partial complex with inhibitory activity. To distinguish between these possibilities, we examined apl2Δ (β1) apl4Δ chc1-ts, and ap-1-null chc1-ts strains. If the β1-deficient partial complex is inhibitory, then elimination of the other subunits should alleviate inhibition and result in minor defects comparable with the effects of aps1Δ or apl4Δ. However, the double and quadruple AP-1 subunit deletion combinations in chc1-ts cells caused severe α-factor maturation defects (Figure (Figure4,4, lanes 5 and 6), supporting the interpretation that β1 is particularly important for Kex2p localization. Similar results were obtained with cells incubated at 30°C, a temperature at which clathrin heavy chain expressed from the chc1-ts allele is partially defective (Figure (Figure4,4, lane 18). However, at 30°C, accentuation of α-factor maturation defects by aps1Δ and apl4Δ was more apparent (Figure (Figure4,4, lanes 10 and 13 compared with lane 18). These data are generally consistent with results from the growth assays and support the conclusion that AP-1 is not required for growth or α-factor maturation (and Kex2p localization) in cells expressing wild-type clathrin. In cells with compromised clathrin function, roles for AP-1 in growth and α-factor maturation can be detected, and the β subunit appears to be especially important.
The innocuous effects of AP-1 subunit deletions on clathrin-dependent processes suggest that clathrin-coated vesicle formation does not rely on AP-1. To address the role of AP-1 in clathrin coat assembly, we determined whether clathrin-coated vesicles could be identified in extracts of ap-1-null cells. Extracts from ap-1-null cells or wild-type cells were fractionated by differential centrifugation, and the high-speed pellet fraction was subjected to gel filtration chromatography through Sephacryl S-1000. Fractions from the S-1000 column were analyzed by SDS-PAGE and immunblotting for clathrin heavy chain and the likely clathrin-coated vesicle cargo protein, Kex2p (Payne and Schekman, 1989 ; Seeger and Payne, 1992a ). The peaks of clathrin heavy chain and Kex2p from wild-type cells occurred in fraction 42 (Figure (Figure5),5), corresponding to the profile expected from previous analyses of yeast clathrin-coated vesicles (Phan et al., 1994 ; Chu et al., 1996 ). Material from ap-1-null cells yielded essentially the same elution peaks of clathrin heavy chain and Kex2p, indicating that elimination of AP-1 does not affect clathrin coated vesicles (Figure (Figure5).5). The levels of Kex2p in nonpeak fractions often varies in different preparations from the same strain, suggesting that the minor differences in Kex2p distribution in Figure Figure55 are not significant. Together, our analyses of ap-1-null cells argue that AP-1 is not required for clathrin-coated vesicle formation or function.
Although AP-1 specifically displays physical and genetic interactions with clathrin, limited functional redundancy between AP-1 and AP-2R and/or AP-3 could account for the absence of defects in cells expressing wild-type clathrin in combination with AP-1 subunit deletions. Our analysis of AP-1 function identifies the β subunit as a particularly important subunit, suggesting that deletion of the β subunit is an effective strategy to abolish the activity of an AP complex. Characterization of cells lacking AP-3 subunits also supports this approach (Cowles et al., 1997a ; Panek et al., 1997 ; Stepp et al., 1997 ). We therefore generated a strain carrying deletions of all three AP β subunits (referred to as 3βΔ) and carried out phenotypic analyses to assess functional redundancy between AP complexes. Growth of the 3βΔ strain was equivalent to wild type at 24, 30, and 37°C (Yeung, unpublished results). Maturation of α-factor was compared in wild-type, an apl1 (β2R) apl2 (β1) double mutant, the 3βΔ strain, and a chc1-ts strain. At 30°C, no defects in α-factor maturation were observed except the expected mild maturation defect in chc1-ts cells (Figure (Figure6).6).
Trafficking through the endocytic pathway was evaluated by measuring turnover of the α-factor mating pheromone receptor Ste3p. Ste3p is normally consititutively internalized and transported to the vacuole where it is degraded (Davis et al., 1993 ). In cells with defects in the endocytic pathway, either at the internalization step or at subsequent steps, delivery of Ste3p to the vacuole is delayed or blocked, thereby enhancing Ste3p stability (for examples see Davis et al., 1993 ; Tan et al., 1993 ). To determine the rate of Ste3p degradation, wild-type and 3βΔ cells were subjected to a pulse–chase regimen followed by lysis and immunoprecipitation of Ste3p. No change in the kinetics of Ste3p turnover was apparent in the 3βΔ strain compared with wild type (Figure (Figure7A,7A, lanes 1–8). Phosphorimage quantitation of the data in Figure Figure77 yielded a t½ for Ste3p degradation of 18 min for wild-type cells and 20 min for 3βΔ cells. As a control, the same procedure was applied to chc1-ts cells labeled at 24°C and then shifted to the nonpermissive temperature (37°C) upon initiation of the chase period. Imposition of the endocytic defect in the chc1-ts cells resulted in a 2.5-fold decrease (t½ = 46 min) in the rate of Ste3p turnover (Figure (Figure7A,7A, lanes 9–12).
Three distinct trafficking pathways to the vacuole were examined in 3βΔ cells. The first pathway is the well-characterized route from the TGN to the vacuole followed by the vacuolar hydrolase CPY (Bryant and Stevens, 1998 ). CPY is synthesized as an inactive precursor that is core-glycosylated in the endoplasmic reticulum to yield p1CPY (67 kDa). Transport via the secretory pathway to and through the Golgi apparatus results in further glycosylation to the p2 form (69 kDa). At the TGN p2CPY is sorted into vesicles targeted to a prevacuolar endosome compartment. From endosomes p2CPY is delivered to the vacuole, where proteolytic activation produces the mature form, mCPY (61 kDa). Sorting and transport from the TGN through endosomes to the vacuole (referred to here as the CPY pathway) requires the activity of a large number of proteins including clathrin and the products of the vacuolar protein sorting genes (VPS) (Bryant and Stevens, 1998 ). Defects in the pathway are manifested as secretion or intracellular accumulation of p2CPY, readily detected by pulse–chase immunoprecipitation analysis of CPY. To assess CPY sorting and delivery to the vacuole in 3βΔ cells, mutant and wild-type cells were subjected to a pulse–chase regimen, and then CPY was immunoprecipitated from intracellular and extracellular fractions. This protocol revealed no difference in the kinetics of conversion of p1 to p2 to mCPY, or in the amount of secreted p2CPY, indicating normal CPY pathway function in 3βΔ cells (Figure (Figure8A).8A). To investigate the possibility that AP complexes might provide cargo-selective function in this pathway, we also examined two other soluble vacuolar proteins that follow this route, proteinase B (PrB) and proteinase A (PrA) (Bryant and Stevens, 1998 ). Vacuolar delivery of both proteins was unaffected in the mutant cells, as judged by maturation kinetics and levels of secretion (Yeung, unpublished results). The second pathway connects the Golgi apparatus to the vacuole by a route that bypasses prevacuolar endosomes (Bryant and Stevens, 1998 ). This pathway relies on AP-3 and is independent of clathrin and those Vps proteins involved in transport to and from endosomes (Cowles et al., 1997a ,b ; Piper et al., 1997 ; Stepp et al., 1997 ; Bryant and Stevens, 1998 ; Vowels and Payne, 1998a ). Similar to the CPY pathway, integrity of the AP-3-dependent pathway can be evaluated through pulse–chase immunoprecipitation of an appropriate cargo protein such as the vacuolar membrane protein ALP. Because ALP is a membrane protein, it is not necessary to monitor secretion; so whole cell lysates were used for immunoprecipitation. By analysis of wild-type, apl6Δ (β3), and 3βΔ strains, we found the extent of the ALP maturation defect in 3βΔ cells to be no greater than that in cells lacking only the AP-3 β subunit (Figure (Figure8B).8B). Residual ALP maturation in the β3-deficient cells is due to missorting to the CPY pathway (Cowles et al., 1997a ; Stepp et al., 1997 ; Vowels and Payne, 1998a ), and the same is likely to be the case in the 3βΔ cells. The similar extent of ALP processing in 3βΔ- and β3-deficient cells indicates that eliminating all three AP β subunits does not enhance sorting defects attributable to the absence of AP-3 β alone. The third pathway delivers the cytoplasmic protein API to the vacuole by a process related to autophagy (Klionsky, 1998 ). This cytoplasmic-to-vacuole (Cvt) pathway involves formation of double-membrane vesicles, which selectively sequester cytoplasmic API. Cvt vesicles fuse directly with the vacuole leading to proteolytic maturation of API. Pulse–chase immunoprecipitation demonstrated no defect in API maturation in the 3βΔ cells (Figure (Figure8C).8C). Together, these analyses indicate that multiple transport pathways to the vacuole are unperturbed by the absence of AP complexes.
Finally, we applied the clathrin-coated vesicle isolation procedure to 3βΔ cells. As anticipated from the absence of clathrin-dependent sorting defects, gel filtration chromatography yielded matching profiles of clathrin heavy chain and Kex2p in wild-type and mutant cells, offering no evidence of defects in clathrin-coated vesicle formation (Figure (Figure9).9).
We have carried out biochemical and genetic characterization of yeast AP complexes. Of the 13 potential AP subunits identified in the yeast genome, four have been previously assigned to AP-3 (Cowles et al., 1997a ; Panek et al., 1997 ). The results reported here indicate that eight of the remaining subunits make up two distinct AP complexes, AP-1 and AP-2R. The extra medium subunit can associate with β1 when overexpressed, raising the possibility of an alternative form of AP-1. These findings argue that yeast express three principal AP complexes. Only β1-containing complexes exhibit physical and genetic interactions with clathrin, yet elimination of all four subunits of the major AP-1 form does not affect growth, clathrin-dependent maturation of α-factor precursor, or assembly of clathrin coats. Cells lacking all three β subunits were subjected to a wide survey of protein trafficking pathways. Except for anticipated defects in AP-3-dependent transport to the vacuole, mutant cells sustained normal levels of pheromone receptor endocytosis, α-factor maturation, vacuolar protein sorting, and clathrin-coated vesicles. We conclude that AP complexes are not obligatory for clathrin-coated vesicle formation and clathrin-mediated protein sorting events in yeast.
Sequence comparisons between yeast and mammalian AP complex subunits indicate that these proteins have been conserved during evolution (Cowles et al., 1997a ; Panek et al., 1997 ). In view of this conservation, as high as 50% amino acid identity, it is surprising that subunit deletions cause trafficking defects solely in the case of AP-3. In earlier studies, which involved single or double subunit deletions, the innocuous consequences of AP-1 and AP-2R mutations could theoretically be attributed to activity of incomplete AP complexes (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). Consistent with this possibility, comparisons of synthetic interactions between single AP-1 subunit deletions and chc1-ts indicate that the extent of α-factor maturation defects varies depending on the subunit that is eliminated (Figure (Figure4).4). We therefore sought to inactivate AP-1 completely by generating a strain lacking the four AP-1 subunits (β1, γ, μ1, and ς1). No defects were detected in this strain. Because Apm2p can associate with β1, at least when overexpressed, a residual contribution of this subunit to clathrin-dependent processes in the ap1-null strain might be envisioned. However, even when Apm2p is overexpressed, it cannot functionally replace μ1 in apm1Δ chc1-ts cells (Stepp et al., 1995 ). Nor does deletion of APM2 accentuate defects in apm1Δ chc1-ts cells (Stepp et al., 1995 ). These observations, combined with the absence of other AP-1 subunits in the ap1-null strain, makes it improbable that Apm2p substitutes in any significant way for AP-1. Given the likelyhood that deletion of β1, γ, μ1, and ς1 abolishes AP-1 activity, the lack of phenotypes in the ap1-null strain indicates that AP-1 is not necessary for normal clathrin function.
The relationship of Apm2p to AP-1 remains to be established. A requirement for overexpression to detect reproducible association of Apm2p with β1 suggests that either the Apm2p-containing AP-1-like complex is much less abundant than AP-1, or Apm2p does not normally associate with β1. Two-hybrid interactions between Apm2p and Apl4p(γ) have been observed (Huang and Lemmon, personal communication), favoring the idea that Apm2p is part of an AP-1-like complex. However, the absence of phenotypes associated with disruption of APM2 in either wild-type, chc1-ts, or AP-1 subunit deletion strains (Stepp et al., 1995 ) leaves the significance of these associations uncertain.
Functional redundancy between AP complexes could obscure a role for AP-1 in clathrin-dependent transport steps in cells expressing wild-type clathrin. However, our studies provide both biochemical and genetic evidence against this idea. In vitro binding assays with GST fusions to the three β subunits showed clathrin binding only to β1. Additionally clathrin was coimmunoprecipitated with AP-1 but not AP-2R or AP-3. These findings suggest that of the three AP complexes, only AP-1 is capable of associating with clathrin. As a genetic test for functional substitution of AP-1 by AP-2R and/or AP-3, genes encoding all three β subunits were deleted. We selected β subunits as targets to disrupt AP function based on our analysis of synthetic interactions between AP-1 subunit deletions and chc1-ts, which demonstrate that deletion of β1 is equivalent to deletion of all four AP-1 subunits. In agreement with the importance of β subunits in AP function, mutation of the AP-3 β subunit is effective in blocking the AP-3 pathway (Cowles et al., 1997a ; Stepp et al., 1997 ). However, despite disruption of all three β subunits, we were unable to detect defects in clathrin-dependent trafficking pathways. The concordance of results from both biochemical and genetic approaches prompts us to discount the idea of redundant function between the three AP complexes.
Could there be another, uncharacterized AP complex capable of substituting for AP-1? Analysis of the yeast genome sequence suggests that this possibility is remote. When mammalian or yeast AP subunits are used to search the yeast genome sequence, the most highly related sequences constitute the known set of 13 AP subunits (Cowles et al., 1997a ; Panek et al., 1997 ). Beyond this group, sequence matches are of limited length and marginal statistical significance. Thus, the 13 AP-related proteins probably represent the complete contingent of AP subunits in yeast.
The prevailing paradigm for clathrin coat formation, established primarily through studies of mammalian clathrin, assigns key roles for AP complexes in assembly of the clathrin lattice at appropriate membranes and in cargo collection (Schmid, 1997 ; Hirst and Robinson, 1998 ). In contrast to our results, AP subunit mutations in filamentous fungi, nematodes, fruit flies, and mice have readily discernable phenotypes, supporting the central importance of AP complexes in clathrin-mediated protein transport in these organisms (Lee et al., 1994 ; Keon et al., 1995 ; González-Gaitán and Jäckle, 1997 ; Zizioli et al., 1999 ). If AP complexes are unnecessary for clathrin function in yeast, then it is likely that other factors subserve clathrin assembly and cargo selection functions. Among the expanding list of proteins associated with clathrin coats, there are a number of candidates that could provide appropriate activities. Mammalian neuronal AP180 binds clathrin and stimulates lattice assembly in vitro (McMahon, 1999 ). Two recently identified yeast homologues of AP180 also interact with clathrin and could be assembly factors (Wendland and Emr, 1998 ). However, deletion of both yeast AP180-encoding genes together with apl2 (β1) has no deleterious effects on growth, pheromone receptor endocytosis, or α-factor maturation (Yeung, Payne, and Wendland, unpublished results). Similar results have been obtained in analyses of cells lacking the six AP large subunits and the two yeast AP180s (Huang et al., 1999 ). Other newly discovered clathrin-interacting proteins such as Epsin (Chen et al., 1998 ) and its yeast homologues Ent1p and Ent2p (Wendland et al., 1999 ) interact with clathrin and may promote coat assembly. Further genetic analysis of these proteins will be needed to assess their role in clathrin coat assembly. Precedents for cargo collection by proteins other than AP complexes have been established through studies of mammalian cells. In the case of β-adrenergic receptor endocytosis, nonvisual arrestins bind both the receptor and clathrin heavy chain, thereby functioning as adaptors to direct receptors into clathrin-coated vesicles (Goodman et al., 1996 , 1997 ). Although a clear homologue of arrestin has not been identified in yeast, analogous adaptors could exist. Alternatively, there may not be a need for a unique adaptor protein. For example, a peptide containing the endocytosis targeting signal from the low-density lipoprotein receptor interacts with the N-terminal globular domain of clathrin heavy chain (Kibbey et al., 1998 ), suggesting that clathrin might act directly to collect certain cargo. These examples suggest a diversification of clathrin assembly and cargo collection activities even in mammalian cells, where the importance of AP complexes is well established. Perhaps under the optimal growth conditions used in laboratory experiments, alternatives to AP-1 assume a more significant role in clathrin-mediated transport in yeast.
Our results clarify structural and functional distinctions between yeast AP complexes and offer additional insights into the relationship between yeast and mammalian APs. Previously, the synthetic growth and α-factor maturation defects caused by combination of chc1-ts with AP-1 subunit deletions were interpreted as evidence for AP-1 association with clathrin (Phan et al., 1994 ; Rad et al., 1995 ; Stepp et al., 1995 ). The specific physical interaction of clathrin with AP-1 and β1 in vitro now provides more direct evidence that AP-1 is a clathrin-associated complex. Thus, yeast AP-1 mimics mammalian AP-1 in both the primary sequence of subunits and physical interaction of the β subunit with clathrin. Although the consequences of subunit deletion appear to be substantially more severe in animal cells (Zizioli et al., 1999 ), the genetic and physical interactions between yeast AP-1 and clathrin suggest that the similarity between mammalian and yeast AP-1 extends to a functional level. As proposed above, the more subtle functional contribution of yeast AP-1 may be attributable to the artificial nature of laboratory growth conditions. In contrast to AP-1, yeast AP-3 does not bind clathrin in our assays. This finding is consistent with genetic experiments indicating that AP-3 acts in a clathrin-independent pathway for membrane protein transport from the Golgi apparatus to vacuoles (Vowels and Payne, 1998a ). Although mammalian AP-3 resembles its yeast cognate by acting in membrane protein sorting to lysosomes, the relationship to clathrin is less clear. Mammalian β3 interacts with clathrin in vitro, and AP-3 can be colocalized with clathrin coats in vivo (Dell’Angelica et al., 1998 ). However, AP-3 does not copurify with clathrin-coated vesicles (Simpson et al., 1996 , 1997 ). Resolution of these apparent discrepancies should establish the extent of similarity between yeast and mammalian AP-3 complexes. The third yeast AP complex that we defined, AP-2R, displays the highest primary sequence similarity to mammalian AP-2. However, in other ways AP-2R is clearly distinct from AP-2. Unlike mammalian AP-2, which shares a highly similar clathrin-binding β subunit with AP-1 (84% identity; Kirchhausen et al., 1989 ), AP-2R contains a β subunit that is only 24% identical to yeast β1 and does not appear to bind to clathrin. Furthermore, mammalian AP-2 associates with endocytic clathrin-coated vesicles, whereas a role for AP-2R in endocytosis has not been detected, nor have we observed synthetic interactions between AP-2R subunit deletions and chc1-ts. Identification of a role for yeast AP-2R awaits additional experiments.
In summary, the first comprehensive description of AP complexes in a single organism is now emerging from studies of S. cerevisiae. Three major, functionally distinct complexes have been described: AP-1, and perhaps an alternative form with a different medium subunit, acts in a clathrin-dependent protein sorting pathway from the TGN; AP-2R probably acts in a clathrin-independent pathway, but the identity of this pathway has not been uncovered; and AP-3 acts in clathrin-independent traffic of membrane proteins from the Golgi apparatus to vacuoles. Elimination of AP function results in AP-3 pathway defects but otherwise appears to be insignificant for clathrin-dependent events. Our results imply the existence of factors other than AP complexes, which play central roles in clathrin coat assembly and cargo selection.
We thank Sandra Lemmon, Lucy Robinson, Scott Emr, Gerry Weinmaster, and Daniel Klionsky for plasmids and antibodies. Members of the Payne laboratory, especially Diana Chu, are acknowledged for their helpful advice and discussions. We are grateful to Jenna Hutton, Eric Bensen, and Alex van der Bliek for insightful comments on the manuscript. This work was supported by National Institutes of Health grant GM-39040 to G.P.