A Method for Determining the in VivoTopology of Yeast Polytopic Membrane Proteins Demonstrates That Gap1p Fully Integrates into the Membrane Independently of Shr3p*

The general amino acid permease (Gap1p) ofSaccharomyces cerevisiae is an integral membrane protein that contains 12 hydrophobic regions predicted to be membrane-spanning segments. A topological reporter construct, encoding an internal 53-amino acid peptide of invertase (Suc2p) containing three Asp-X-Ser/Thr glycosylation sites, was inserted in-frame into the hydrophilic NH2- and COOH-terminal domains and each of the 11 hydrophilic loops that separate the 12 hydrophobic segments of Gap1p. The resulting 13 gene sandwich fusion proteins were expressed in a gap1Δ null mutant strain; 9 of these retain amino acid transport activity and are folded and correctly targeted to the plasma membrane. The glycosylation state of each of the fusion proteins was monitored; the results indicate that all 12 hydrophobic segments of Gap1p span the membrane, and the NH2 and COOH termini are cytoplasmically oriented. These results were independently tested by isolating sealed right-side-out microsomes from sec12–1 strains expressing six different Gap1p constructs containing functional factor Xa protease cleavage sites. The pattern of factor Xa protease cleavage was found to be consistent with the presence of 12 membrane-spanning domains. Gap1p exhibited the same membrane topology in strains lacking Shr3p; therefore, Gap1p fully integrates into the ER membrane independently of this permease-specific packaging chaperone.

Amino acid uptake across the plasma membrane (PM) 1 in Saccharomyces cerevisiae is catalyzed by a gene family of structurally related transport proteins known as amino acid permeases (AAPs). The AAP family contains 18 core members that share extensive homology, and 6 other members that exhibit less sequence conservation (1). Based upon available predictive algorithms, AAPs are thought to be integral membrane proteins comprising 12 membrane-spanning domains. The general AAP (GAP1) is a low affinity, high capacity permease with broad substrate specificity, which is capable of transporting most amino acids, even D-amino acids (2,3). The majority of other AAP family members are low capacity, high affinity permeases, each exhibiting characteristic and rather narrowly defined substrate specificities (4). One of the core AAP mem-bers (SSY1) contains a unique NH 2 -terminal extension not present in the other AAPs. Ssy1p has recently been shown to function as a sensor of extracellular amino acids that serves to transduce metabolic signals that differentially regulate the activity of the general and specific AAPs to control amino acid uptake under a variety of environmental conditions (5)(6)(7)(8).
AAPs are co-translationally inserted into the membrane of the endoplasmic reticulum (ER). Subsequent to membrane insertion, these proteins must fold properly to attain native conformations in order to be transported to the PM. Protein folding is a process that often requires specific processing events (9). At an early stage in the secretory pathway, AAPs are co-transported (10) together with other secreted proteins from the ER to the Golgi apparatus via ER-derived COPII coated transport vesicles (reviewed in Refs. 9, 11, and 12). The AAP gene family members require Shr3p to be included in COPII transport vesicles (10,(13)(14)(15). Recently, several other complex polytopic PM proteins in S. cerevisiae have been found to require the assistance of accessory proteins to exit the ER (16 -18). As is observed with shr3 mutations, when these accessory proteins are mutated or deleted, their cognate cargo do not enter COPII transport vesicles and accumulate within the ER. The secretory block in these mutants is specific; other plasma membrane, secretory, and vacuolar proteins are processed and targeted correctly.
Shr3p is an integral membrane protein with four membranespanning segments and a hydrophilic cytoplasmically oriented carboxyl-terminal domain (13). Shr3p has been shown to physically associate with Gap1p, but not with other polytopic membrane proteins such as Sec61p, Gal2p, or Pma1p, in a transient complex that can be purified from N-dodecylmaltoside solubilized membranes (15). The COPII coatomer components Sec13p, Sec23p, Sec24p, and Sec31p, but not Sar1p, bind Shr3p via interactions with its carboxyl-terminal domain. As Shr3p does itself not exit the ER (10), Shr3p must dissociate and diffuse away prior to completion of vesicle formation. By facilitating the membrane association and assembly of COPII coatomer components, Shr3p is thought to function as a packaging chaperone that directs the formation of vesicle buds around AAPs, thereby ensuring their inclusion into transport vesicles (15). An analysis of whether Shr3p influences the membrane structure of AAPs is essential to fully understand the molecular mechanisms governing the exit of permeases from the ER.
It is not trivial to determine the in vivo membrane association and orientation of hydrophobic domains within polytopic membrane proteins. Fusion proteins, with COOH-terminal reporter sequences, have been extensively used to study the topology of Escherichia coli membrane proteins (19,20). Similar gene fusion approaches, often using the invertase gene (SUC2), have been used in yeast to study the topology of a variety of ER and PM proteins (21)(22)(23)(24). Suc2p, comprising 530 amino acids, functions as a topological reporter as it becomes efficiently glycosylated at multiple sites when translocated across the ER membrane. Unfortunately, COOH-terminal fusion approaches have proven to be unreliable in determining the topology of polytopic membrane proteins. In several instances the NH 2 -terminal membrane segments of complex polytopic proteins have been shown to require downstream COOH-terminal membrane domains to attain their correct membrane orientation (23)(24)(25). Similarly, we have observed that hybrid proteins consisting of full-length Suc2p fused at positions following the first six hydrophobic segments of Gap1p were present in both glycosylated and unglycosylated forms, an indication that these fusion proteins are capable of adopting different membrane conformations (15).
Several experimental approaches have been developed to circumvent the problems associated with COOH-terminal gene fusions. The gene sandwich approach is based on the insertion of a reporter protein, e.g. PhoA, into internal sequences of a target membrane protein (26). In the context of the full-length target protein, gene sandwich reporter constructs have provided more reliable models of membrane protein topology (26). A significant drawback to this approach is that, in most instances, the insertion of a large reporter construct inactivates the catalytic properties of the target protein (27,28). The insertion of shorter reporter sequences, e.g. factor Xa protease cleavage sites, into hydrophilic loops has also proven to be a useful alternative to COOH-terminal fusion approaches (24,29). However, this approach requires isolation of homogeneous preparations of intact membranes, many tedious control experiments, and experimental difficulties associated with protease accessibility are well documented (29,30). The insertion of N-glycosylation consensus sequences (NXS/T) into hydrophilic loops (30,31) has been successfully used to study the topology of several mammalian polytopic membrane proteins. Primarily due to the fact that glycosylation sites are rather inefficiently used in yeast, similar glycosylation scanning approaches have not been used to examine the in vivo topology of proteins in yeast.
We have adapted the glycosylation scanning method for use in yeast and have determined the in vivo topology of Gap1p in SHR3 and shr3⌬6 cells. A novel gene sandwich reporter construct, encoding an internal 53-amino acid segment from Suc2p containing three NXS/T glycosylation sites, was inserted inframe into the hydrophilic NH 2 -and COOH-terminal domains and each of the 11 hydrophilic loops that separate the 12 hydrophobic segments of Gap1p. Nine of these fusion proteins are correctly targeted to the PM and retain amino acid transport activity. By analyzing the glycosylation state of these fusion proteins, we have found that each of the 12 hydrophobic segments of Gap1p span the membrane, and both the NH 2 -and COOH-terminal domains are cytoplasmically oriented. This model was tested by isolating sealed right-side-out microsomes from sec12-1 strains expressing Gap1p constructs containing cleavable factor Xa protease sites. The proteolytic fragments derived from the Gap1p constructs corroborated the 12-membrane-spanning model. Additionally, our data clearly indicate that Gap1p fully integrates into the membrane independently of Shr3p. The Suc2p-based topological reporter cassette should be useful in studies aimed at determining the in vivo structure of other polytopic membrane proteins in yeast.

EXPERIMENTAL PROCEDURES
Strains, Media, and Microbiological Techniques-Yeast strains are listed in Table I, and plasmids used are listed in Table II. The temperature-sensitive sec12-1 mutant CKY42 was kindly provided by C. A. Kaiser (Massachusetts Institute of Technology, Cambridge, MA). A diploid strain, constructed by crossing CKY42 with FGY58 (gap1⌬), was sporulated, and tetrad analysis indicated that the temperature sensitive and gap1⌬ phenotypes segregated independently, and with the expected ratio of 2:2. Strains FGY11 (sec12-1 gap1⌬) and FGY15 (SEC ϩ   (10). The transformations were carried out with overnight cultures grown at 22°C, and transformation plates were incubated at 20°C. Ura ϩ transformants were propagated on medium containing 5-fluoroorotic acid to obtain the unmarked shr3⌬6 deletion resulting in strains FGY14 and FGY135, respectively. The successful deletion of SHR3 was confirmed by Southern blot analysis. Standard yeast media were prepared, and yeast genetic manipulations were performed as described in Ref. 32. SPD and SCitD media, respectively containing proline and citrulline as sole nitrogen sources, were prepared as described (13). Yeast transformations were performed as described (33) using 50 g of heat-denatured calf thymus DNA. Transformants were selected on solid SC media lacking uracil. Plasmid Constructions-Plasmids were constructed using standard molecular biological procedures. Plasmids containing GAP1 gene fusion alleles with topological reporter cassettes A and B were constructed in three stages. In the first stage, BamHI restriction sites were inserted at 13 positions along the GAP1 gene at sequences corresponding to amino acids 5, 121, 154, 198, 228, 275, 307, 360, 412, 444, 491, 525, and 567, creating plasmids pFG130 through pFG142, respectively (see Fig. 1B). Synthetic oligomers comprising 32-34 nucleotides containing BamHI sites flanked by 13-14 nucleotides of complementary GAP1 sequence were annealed to single-stranded pPL247 prepared with helper phage M13K07 (34) in the dutung -E. coli host RZ1032 (35). After elongation, ligation, and transformation into a dut ϩ ung ϩ host, plasmids were screened for the presence of the BamHI restriction sites diagnostic for successful mutagenesis. In stage 2, in two separate reactions, pairs of synthetic oligomers containing a BamHI and SpeI-BamHI sites flanked on each side by 16 bases complementary to the SUC2 sequence (pair A corresponding to sequences preceding amino acid 81 and following amino acid 133; pair B corresponding to sequences preceding amino acid 357 and following amino acid 410), were annealed to singlestranded pFG6. This procedure created plasmids pFG112 and pFG113, respectively (see Fig. 2A). In stage 3, plasmids pFG130-pFG142 were restricted with BamHI, and topological reporter cassettes A (168 base pair BamHI fragment from pFG112) and B (171 base pair BamHI fragment from pFG113) were separately ligated into each plasmid, creating plasmids pFG150 -pFG162 and pFG170 -pFG182, respec-tively. The gene fusions containing factor Xa cleavage sites were also constructed by site-directed mutagenesis using single-stranded pPL257 as template DNA. Successful mutagenesis was ascertained by restriction with TaqI. These plasmids contain GAP1-fXa fusion alleles encoding proteins with a maximum of eight extra amino acids that in each case generate the tetrameric repeat IEGRIEGR, inserted into Gap1p following amino acids 41 (pFG190), 271 (pFG195), 312 (pFG196), 355 (pFG197), 481 (pFG200), and 564 (pFG202), respectively.
Protein Manipulations-Total yeast protein was obtained by the method described in Ref. 36. Samples were heated for 10 min at 37°C for the analysis of Gap1p, and for 5 min at 65°C for the analysis of other proteins. Proteins were resolved by SDS-PAGE using a modified Laemmli system (37) in which SDS is omitted from the gel and lower electrode buffer. Immunoblots were incubated with primary antibodies at the following dilutions: anti-Gap1p, 1:20000; anti-Kar2p, 1:5000; anti-Wbp1p, 1:1000; anti-Sec61p, 1:3000, and anti-hemagglutinin 1 mouse monoclonal 12CA5, ascites fluid 1:1500. Immunoblots were washed, incubated with secondary anti-rabbit or anti-mouse Ig horseradish peroxidase linked antibody and developed using chemiluminescence detection reagents (ECL-Plus Western blotting detection systems, Amersham Pharmacia Biotech). Chemiluminescent signals were quantitated using the LAS1000 system (Fuji Photo Film Co. Ltd., Japan).
Determination of Topology Reporter Cassette Glycosylation-Plasmids encoding topological reporter cassette gene fusions were transformed into FGY58 (SHR3) and FGY60 (shr3⌬6). Ura ϩ transformants were selected on SC (minus uracil) agar plates. Overnight cultures grown in liquid SC (minus uracil) were harvested, washed once in water, and resuspended to an OD 600 of 0.5 in SPD (plus adenine and lysine), and total yeast protein was prepared after cells were grown for an additional 4 h at 30°C to OD 600 of 1.2. Duplicate protein samples (25

FIG. 2. Topological reporter cassettes A and B, derived from
internal regions of SUC2, exhibit differential effects on Gap1p fusion protein stability. A, schematic presentation of the 1.8-kilobase pair SalI-XmnI DNA fragment in pFG6 containing the coding sequence of mature Suc2p. Single-stranded pFG6 was used as template for the construction of topological reporter cassettes A and B indicated as black boxes (see "Experimental Procedures"). These cassettes, flanked by BamHI and SpeI-BamHI endonuclease sites, are cloned into plasmids pFG112 and pFG113, respectively. Asterisks indicate potential glycosylation sites within Suc2p. The plus symbols (ϩ) indicate positions of positively charged amino acids (arginine or lysine), and the minus symbols (Ϫ) indicate negatively charged amino acids (glutamate or aspartate). B, native Gap1p (pPL247), cassette A gene fusion proteins Gap1p-5A (pFG150) and Gap1p-275A (pFG155), and cassette B gene fusion proteins Gap1p-5B (pFG170) and Gap1p-275B (pFG175) were expressed in FGY58. Extracts of total cell protein were prepared from transformants grown in SPD (plus lysine and adenine), and analyzed as described under "Experimental Procedures." Briefly, protein preparations were solubilized in SDS-PAGE sample buffer, treated with endo H where indicated, resolved in 10% polyacrylamide gels, and immunoblotted with polyclonal anti-Gap1p antibody. The positions of molecular mass (kDa) markers are indicated.
l, equivalent to OD 600 of 0.2 cell suspension) were diluted with an equal volume of 100 mM sodium citrate, pH 5.5, and heated for 10 min at 37°C. Three milliunits of endoglycosidase H (endo H; Roche Molecular Biochemicals) was added to half of the samples, and all samples were incubated overnight at 37°C (38). Proteins were resolved by 10% SDS-PAGE and immunoblotted.
Yeast Microsome Preparation-Strains FGY135 (shr3⌬), FGY11 (sec12-1), and FGY14 (shr3⌬ sec12-1) transformed with plasmids encoding gene fusion proteins containing factor Xa cleavage sites were grown in SC (minus uracil) to an OD 600 of 2. Cells from 50 ml of culture were harvested, washed once in water, resuspended in 5 ml of water and incubated at 34°C for 10 min. Temperature-shifted cells were used to inoculate 200 ml of SPD (plus lysine); the starting OD 600 was 0.5. The cultures were incubated shaking for 2 h at 34°C, after which cells were harvested and resuspended in spheroplasting buffer (1.2 M sorbitol, 50 mM potassium acetate, 20 mM Tris-HCl, pH 7.5, 0.5 mM dithiothreitol, 10 mM sodium azide) at an OD 600 of 50. Cells were spheroplasted by incubating cells at 30°C for 30 min in the presence of 1.0 mg/ml zymolyase-100T (ICN Biomedicals, Inc., Aurora, OH). Spheroplasts were collected by centrifugation at 2000 ϫ g for 5 min at 4°C, resuspended at an OD 600 of 20 in lysis buffer (100 mM sorbitol, 50 mM potassium acetate, 20 mM Tris-HCl, pH 7.5, 0.5 mM dithiothreitol) containing protease inhibitors; 1 mM phenylmethylsulfonyl fluoride (Sigma) and 10 kallikrein inhibitor units/ml aprotinin (Bayer Leverkusen). Spheroplasts were lysed with 20 strokes of a Teflon pestle in a 5-ml tissue grinder. Unbroken cells were removed by centrifugation at 1000 ϫ g for 5 min at 4°C. The cleared microsome suspension was centrifuged at 100,000 ϫ g for 30 min at 4°C. Pelleted microsomes were washed once with storage buffer (250 mM sorbitol, 50 mM potassium acetate, 20 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol) and resuspended in storage buffer at 1 mg/ml protein. Microsome preparations were stored at Ϫ80°C.
Cleavage with Factor Xa Protease and Proteinase K-Fifteen microliters of microsomal preparations were diluted with 5 l of fXa buffer (250 mM sorbitol, 100 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA). Samples were mock-digested or digested with 1.1 g of factor Xa protease (1.1 mg/ml; Promega, Madison, WI) in the presence or absence of 0.2% Nonidet P-40 on ice for 2 h. Sixty microliters of thawed microsomes (60 g of protein), prepared from strain FGY11 (sec12-1) were diluted with 30 l of fXa buffer. A 6-l aliquot of a 4 mg/ml solution of proteinase K (Sigma) prepared in fXa buffer was added (final concentration ϭ 250 g/ml). The proteinase K treatment was carried out on ice in the absence or presence of 0.2% Nonidet P-40.

RESULTS
The Gap1p Sequence Contains 12 Potential Membrane-spanning Domains-Gap1p comprises 602 amino acids. Hydropathic profile analysis indicates that Gap1p has 12 hydrophobic regions (Fig. 1A), numbered I-XII, each of which are of sufficient length and hydrophobicity to function as membranespanning domains (39). Using single-stranded mutagenesis, BamHI restriction sites were individually introduced into GAP1 sequences encoding the amino-and the carboxyl-terminal domains, and into each of the hydrophilic loops (L1-L11) that separate the hydrophobic domains (see "Experimental Procedures"). The BamHI sites were inserted in such a manner as not to disrupt the reading frame of the Gap1p coding sequence, and enabled the subsequent construction of in-frame gene sandwich fusions with Suc2p based topological reporter cassettes A or B (Fig. 1B, open circles). Additionally, tandomly repeated in-frame factor Xa protease sites were inserted into the Gap1p sequence at positions within the NH 2 -and COOHterminal domains, and into hydrophilic loops L5, L6, L7, and L10 (Fig. 1B, open squares). The factor Xa protease recognizes the tetrapeptide motif IEGR and specifically cleaves the protein sequence COOH-terminal of the arginine residue (40). The recognition motif was tandomly (IEGRIEGR) inserted to increase the probability of cleavage (29). The ability of factor Xa protease to cleave the fXa fusion constructs was initially tested in Nonidet P-40-solubilized membrane preparations; in each case, the fusion constructs were efficiently cleaved. Factor Xa sites inserted into the other hydrophilic loops of Gap1p were refractory to added protease; consequently, these fusion constructs were not further analyzed.
We examined whether the modified Gap1p fusion proteins retained the capacity to transport amino acids. Gap1p is the only AAP capable of transporting citrulline at rates sufficient to support growth on media containing citrulline as sole nitrogen source (3). Strain FGY58 (gap1⌬::LEU2), lacking its chromosomal copy of GAP1, was individually transformed with plasmids pFG150 -pFG162 (encoding cassette A gene fusions), pFG170 -pFG182 (encoding cassette B gene fusions), and pFG190, pFG195, pFG197, pFG200, and pFG202 (encoding factor Xa insertions). Plasmids capable of supporting growth on media containing citrulline as sole nitrogen source were judged to carry functional Gap1p alleles (Fig. 1B, plus signs within  mutant symbols). The fusion constructs are designated by the amino acid to which they are fused, functional constructs begin with a capital letter, e.g. Gap1p-198A (cassette A fused at position 198), and non-functional constructs begin with a lowercase letter, e.g. gap1p-312fXa (factor Xa cleavage sites fused at position 312).
Construction of Topological suc2 Reporter Cassettes A and B-Two regions within Suc2p, amino acids 81-133 and 357-410 (the amino acid coordinates are derived from the sequence of the mature secreted form of Suc2p), were selected to construct topological reporter cassettes for the in vivo analysis of the membrane structure of Gap1p. These regions, each comprising approximately 50 amino acids, contain three NXS/T sequons for asparagine-linked glycosylation that are known to be glycosylated in the secreted form of invertase (41). It has FIG. 3. Membrane topology of Gap1p is independent of Shr3p. Cassette A gene fusion proteins (pFG150-pFG162) were expressed in FGY58 (SHR3) and FGY60 (shr3⌬). Extracts of total cell protein were prepared from transformants grown in SPD (plus lysine and adenine), and analyzed as described under "Experimental Procedures." Protein preparations were treated with endo H as indicated, and resolved by SDS-PAGE in 10% polyacrylamide gels, immunoblotted with polyclonal anti-Gap1p antibody. The topological orientation, as indicated (cyt, cytosolic; lum, lumenal), of each reporter cassette was identical in both strains. been shown that the distance between the lumenal end of a transmembrane segment and a potential glycosylation acceptor site influences whether the site is efficiently glycosylated (42); the reporter cassettes were therefore constructed to place the first and last NXS/T acceptor sites at least 12 amino acids away from the ends. Both cassettes contain equal numbers of negatively and positively charged amino acids, and thus are not expected to influence the orientation of adjacent membranespanning segments. Plasmids pFG112 (cassette A) and pFG113 (cassette B) contain the sequences encoding these regions flanked by BamHI sites (Fig. 2A).
As described in the preceding section, the reporter cassettes were inserted into the GAP1 sequence at the BamHI sites of plasmids pFG130 -pFG142. Each of the resulting 26 gene fusion constructs were tested for Gap1p activity in a growthbased assay. Fusion proteins with either cassette A or B inserted into 9 of the 13 positions enabled strain FGY58 to grow on citrulline-based media (Fig. 1B); however, the strains expressing the functional cassette A gene fusions formed larger colonies. We compared the steady-state levels and glycosylation state of wild-type Gap1p, Gap1p-5A, Gap1p-275A, Gap1p-5B, and Gap1p-275B proteins expressed in strain FGY58 (Fig.  2B). Compared with extracts prepared from the wild-type Gap1p-expressing strain (Fig. 2B, lanes 1 and 2), extracts obtained from strains expressing Gap1p-5A and Gap1p-275A (Fig. 2B, lanes 3-6) contained similar amounts of fusion proteins. In contrast, the extracts isolated from Gap1p-5B-and Gap1p-275B-expressing strains contained significantly lower levels of fusion proteins (Fig. 2B, lanes 7-10). Strains expressing the remaining 11 cassette B reporter constructs invariably contained less fusion proteins than strains expressing the corresponding cassette A hybrids (data not shown). Despite differences in protein levels, Gap1p-275A and Gap1p-275B were similarly glycosylated as evidenced by their increased electrophoretic mobility after incubation with endo H (Fig. 2B, compare lanes 5 with 6, and lanes 9 with 10). These results indicate that the loop containing the reporter cassettes are oriented in the ER lumen. The observed 4-5-kDa decrease in molecular mass suggests that all three NXS/T sites were glycosylated. Treatment with endo H did not affect the mobility of wild-type Gap1p, Gap1p-5A, and Gap1p-5B proteins, indicating that these proteins are not glycosylated.
Analysis of Cassette A Fusion Protein Topology-Due to the consistently low levels of cassette B gene fusions in protein extracts, we present our analysis of Gap1p topology obtained using the reporter cassette A containing gene fusions. It should be noted that cassette B constructs were analyzed in parallel and identical patterns of glycosylation were obtained. Preparations of total cell protein were isolated from strains FGY58 (SHR3) and FGY60 (shr3⌬6) expressing the 13 cassette A reporter constructs. Proteins were fractionated by SDS-PAGE before and after treatment with endo H (Fig. 3). The extracts contained different amounts of fusion proteins; the observed differences are likely a consequence of differential stability, a phenomenon that has been described for other gene sandwich fusions expressed in E. coli (27,28). In all cases the unglycosylated fusion proteins migrated as bands of the predicted mass. Each of the 13 fusion proteins exhibited an identical pattern of endo H sensitivity, regardless of the SHR3 genotype of the strain in which it was produced (Fig. 3, compare the  upper and lower panels), indicating that the membrane structure of Gap1p is not dependent upon the presence of Shr3p.
Integrity and Sidedness of Microsomal Membrane Preparations-The 12 membrane-spanning model of Gap1p was tested by introducing factor Xa protease sites into six diagnostic positions of Gap1p (Fig. 1B). Gap1p in SHR3 cells is predominantly localized to the plasma membrane, whereas Gap1p in shr3⌬6 cells is almost exclusively localized to the ER (13). In order to be able to compare the membrane topology of factor Xa fusion proteins in SHR3 and shr3⌬6 cells, we took advantage of the well characterized temperature-sensitive sec12-1 mutation (44,45). At restrictive temperatures (Ն34°C), the transport of AAPs from the ER is blocked in strains carrying the sec12-1 mutation. Gap1p expression can be controlled by the nitrogen source supplied in the medium; in comparison to cells grown in synthetic complete (SC) media containing ammonia, cells grown in minimal synthetic proline medium (SPD) contain 15-fold higher levels of Gap1p (15).
Strains FGY11 (sec12-1 SHR3), FGY14 (sec12-1 shr3⌬6), and FGY135 (SEC12 shr3⌬6) were transformed with plasmids encoding the six factor Xa fusion proteins. The resulting 18 strains were pregrown at 25°C (permissive temperature) in SC, harvested, and incubated in water for 10 min at 34°C to initiate the secretory block imposed by the sec12-1 mutation. The temperature-shifted cells were used to inoculate SPD, and the cultures were incubated an additional 2 h at 34°C. Microsomal membranes were isolated as described under "Experimental Procedures." The integrity and sidedness of microsomal preparations was determined by examining whether the ER lumenal protein Kar2p, and the ER membrane proteins Wbp1p and Sec61p were accessible to degradation by proteinase K in the absence or presence of 0.2% Nonidet P-40. At the times indicated, samples were prepared for SDS-PAGE and analyzed by immuno-blotting with anti-Kar2p, anti-Wbp1p, and anti-Sec61p antiserum (Fig. 4). We found that the lumenal protein Kar2p (46) was protected from proteinase K digestion, even during extended incubations (8 h). Kar2p was rapidly degraded when membranes were permeabilized with 0.2% Nonidet P-40. Similarly, Wbp1p, a type I membrane protein with a large lumenal domain (47), was only degraded in the presence of detergent. In contrast, the polytopic membrane protein Sec61p was rapidly degraded even in the absence of detergent, indicating that cytoplasmically localized COOH-terminal domain recognized by the anti-Sec61p antibodies was accessible to the protease (24). These results demonstrate that the microsomal membranes predominantly comprised sealed membrane preparations oriented with the cytoplasmic side facing out.
Analysis of Factor Xa Fusion Protein Topology-Microsomes prepared from strains FGY11 (sec12-1 SHR3), FGY14 (sec12-1 shr3⌬6), and FGY135 (SEC12 shr3⌬6) expressing the fXa fusion proteins were incubated with factor Xa for 2 h in the absence or presence of 0.2% Nonidet P-40. The reaction products were resolved by SDS-PAGE and analyzed by immunoblotting with antiserum directed against the NH 2 -terminal domain of Gap1p (Fig. 5). The patterns of factor Xa cleavage were identical in all three strains, confirming that the membrane structure of Gap1p is not dependent upon the presence of Shr3p. In the absence of detergent, factor Xa cleavage products were generated from microsomes containing Gap1p-41fXa, gap1p-312fXa, Gap1p-481fXa, and Gap1p-564fXa proteins, consistent with the cytoplasmic localization of the NH 2 and COOH termini, and hydrophilic loops 6 and 10. The Gap1p-271fXa and Gap1p-355fXa proteins were not cleaved without permeabilizing the membranes with Nonidet P-40. Native Gap1p does not contain factor Xa recognition sites and was not cleaved under the conditions described. The data indicate that these fXa sites, and hence hydrophilic loops 5 and 7, are oriented in the ER lumen. DISCUSSION The topological reporter cassettes described here form the basis of a general method to determine the membrane association and orientation of hydrophobic domains within polytopic yeast membrane proteins. In designing the reporter constructs, we attempted to satisfy the following criteria. We sought to establish a method for the in vivo analysis of yeast proteins, and to avoid using heterologous expression or in vitro translation systems. To enable the topological analysis to be carried out in the context of the complete protein sequence, i.e. in the FIG. 6. The proposed membrane topology of Gap1p. Twelve membrane-spanning domains have been identified. Transmembrane domains I-XII are represented by shaded rectangles. The NH 2 and COOH termini, and the even-numbered hydrophilic loops, were found to be cytoplasmically oriented. The odd-numbered loops were found to be accessible to lumenal ER-processing components, and thus will be oriented toward the extracellular milieu when Gap1p is localized at the plasma membrane.
presence of all membrane-spanning domains, the cassettes were used to create gene sandwich fusion proteins. Due to the ease of the analysis, and the demonstrated success of glycosylation scanning approaches in mammalian systems (30,31), glycosylation was chosen as the topological read-out. However, it is known that glycosylation sites are rather inefficiently used in yeast, and it is not understood why some potential glycosylation sites are not used. Recent work demonstrates an inverse relationship between protein folding reactions and glycosylation, i.e. regions of proteins containing glycosylation sequons that fold quickly are less likely to be glycosylated (48). To minimize complications we chose two regions of Suc2p that are known to be effectively used (41). The cassettes have a net neutral charge, each contains equal numbers of negatively and positively charged amino acids, to minimize influencing the orientation of adjacent membrane-spanning segments (49,50). Finally, the desired reporter constructs were designed to be as small as possible to minimize potential folding artifacts, protein instability, and hopefully to enable the enzymatic activity of target proteins to be maintained. The desire for compactness was balanced against the demand for a robust signal strength; three glycosylation sequons were included in each cassette with sufficient flanking sequence to facilitate access by glycosyl transferase.
We have used the Suc2p-based topological reporter cassettes to determine the membrane structure of Gap1p. Although the results presented here have focused on cassette A-containing fusions, due to the higher steady-state levels of fusion proteins (Fig. 2), identical topological results were obtained with both cassettes at all positions. We have not investigated the underlying reasons for the lower levels of cassette B-containing fusions. The insertion of factor Xa protease cleavage sites was used to test the validity of the topological model obtained using the reporter cassette. The data obtained from these two independent approaches show that the 12 hydrophobic domains of Gap1p traverse the membrane in a zigzag fashion connected by hydrophilic loops, with the hydrophilic NH 2 and COOH termini oriented in the cytoplasm (Fig. 6). This model is based on our finding that gene fusions in the NH 2 -and COOH-terminal regions, and within the even-numbered hydrophilic loops were not glycosylated (Fig. 3) and were accessible to factor Xa protease even in the absence of detergent (Fig. 5). Conversely, gene fusions within the odd-numbered loops were extensively glycosylated (Fig. 3) and inaccessible to protease in the absence of detergent (Fig. 5), indicating that these regions of Gap1p are localized to the lumen of the ER. No structural ambiguities were detected, and it should be noted that the three data sets (cassette A, cassette B, and fXa) are internally consistent, we never observed conflicting data regarding the orientation of membrane-spanning segments.
Given the high level of sequence homology and similarity of hydrophobic profiles, the other members of the AAP gene family are likely to have similar membrane topologies. Gene sandwich approaches using alkaline phosphatase have been used to study the membrane topology of three E. coli permeases that share significant homology with the yeast AAPs. According to these studies, each of the 12 hydrophobic regions function as membrane-spanning domains (27,28,51). In contrast to these results and those presented here, earlier investigations analyzing the topology of the yeast arginine permease (CAN1) did not result in a clear consensus regarding the number of hydrophobic segments that span the membrane, both 10 and 11 membrane-spanning domains were postulated (22,52).
Finally, the data clearly demonstrate that the fusion proteins exhibited an identical pattern of endo H and protease sensitivity, and migrated the same regardless of the SHR3 genotype of the strain in which they were produced (Figs. 3 and 5). Since no differences were observed, our data suggest that Shr3p does not participate in the co-translational insertion of AAPs into the ER membrane. These results are consistent with our previous findings that suggested that AAPs are correctly folded in shr3 null mutant cells (15). Apparently AAPs integrate into the ER membrane and fold properly independent of Shr3p function. Thus Shr3p must function at a subsequent step in the secretory pathway to ensure AAPs are efficiently packaged in ER-derived COPII transport vesicles. We were surprised by the observation that fusion proteins Gap1p-198A, Gap1p-275A, Gap1p-360A, and Gap1p-444A expressed in FGY60 (shr3⌬6) appeared to receive outer chain glycolytic linkages, suggesting that these proteins exit the ER independently of Shr3p (Fig. 3, lower panel). The fact that Gap1p-525A did not exit the ER raises the interesting possibility that core glycosylation, or the Suc2p moiety, within internal hydrophilic loops may enable Gap1p to be packaged for transport without the assistance of Shr3p. The model of Gap1p topology (Fig. 6) forms the basis for further dissection of the molecular mechanisms governing Gap1p function and its intracellular transport through the secretory and endocytic pathways.