New Findings on Interactions among the Yeast Oligosaccharyl ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2003 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 278, No. 35, Issue of August 29, pp. 33078 –33087, 2003 Printed in U.S.A.

New Findings on Interactions among the Yeast Oligosaccharyl Transferase Subunits Using a Chemical Cross-linker* Received for publication, May 21, 2003, and in revised form, June 12, 2003 Published, JBC Papers in Press, June 12, 2003, DOI 10.1074/jbc.M305337200

Aixin Yan, Eilaf Ahmed, Qi Yan‡, and William J. Lennarz§ From the Department of Biochemistry and Cell Biology and the Institute for Cell and Developmental Biology, State University of New York at Stony Brook, New York 11794-5215

At present, there is very limited knowledge about the structural organization of the yeast oligosaccharyl transferase (OT) complex and the function of each of its nine subunits. Because of the failure of the yeast two-hybrid system to reveal interactions between luminal domains of these subunits, we utilized a membrane permeable, thiocleavable cross-linking reagent dithiobis-succinimidyl propionate to biochemically study the interactions of various OT subunits. Four essential gene products, Ost1p, Wbp1p, Swp1p, and Stt3p were shown to be cross-linked to each other in a pairwise fashion. In addition, Ost1p was found to be cross-linked to all other eight OT subunits individually. This led us to propose that Ost1p may reside in the core of the OT complex and could play an important role in its assembly. Ost4p and Ost5p were found to only interact with specific components of the OT complex and may function as an additional anchor for optimal stability of Stt3p and Ost1p in the membrane, respectively. Interestingly, Ost3p and Ost6p subunits exhibited a surprisingly identical pattern of crosslinking to other subunits, which is consistent with their proposed redundant function. Based on these findings, we analyzed the distribution of the lysine residues that are likely to be involved in cross-linking of OT subunits and propose that the OT subunits interact with each other through either their transmembrane domains and/or a region proximal to it, rather than through their luminal or cytoplasmic domains.

Over the past decade it has been established that in both higher eukaryotes and yeast, the enzyme oligosaccharyl transferase (OT),1 is a complex consisting of multiple, nonidentical membrane protein subunits residing in the endoplasmic reticulum (ER) (1). In the case of the yeast Saccharomyces cerevisiae, nine different subunits have been cloned and identified. Among them, the genes encoding Ost1p, Ost2p, Stt3p, Wbp1p, and Swplp are essential for the viability of the cell; OST4 gene * This study was supported by National Institutes of Health Grant GM33185 (to W. J. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: Dept. of Molecular Biology, The Scripps Research Institute, La Jolla, CA 92037. § To whom correspondence should be addressed. Tel.: 631-632-8560; Fax: 631-632-8575; E-mail: [email protected]. 1 The abbreviations used are: OT, oligosaccharyl transferase; DSP, dithiobis-succinimidyl propionate; HA, hemagglutinin; Dol, dolichol; ER, endoplasmic reticulum, PDI, protein-disulfide isomerase; LLOS, lipid-linked oligosaccharide synthase.

is essential for growth of the cell at 37 °C, but not at 25 °C; Ost3p, Ost5p, Ost6p subunits are not essential for the viability of the cell, but are required for maximal activity of the enzyme complex in catalyzing N-glycosylation (for review, see Refs. 1 and 2). To understand how these nine subunits interact with each other, a genetic approach using the yeast two-hybrid screen was utilized, but it failed to show interactions between any of the luminal domains of the yeast OT subunits (3). Studies on the mechanism of enzyme catalysis have proposed that Wbp1p may contain a site for the binding of the lipidlinked oligosaccharide substrate (4), and Stt3p was shown to be directly involved in peptide substrate recognition and/or the catalytic glycosylation process (5). However, there is little detailed information on how these two subunits interact to catalyze the glycosylation reaction. Study of the structural interactions of membrane proteins of the ER offers a special challenge because of the hydrophobicity of the proteins and the special environment in which these proteins reside: the hydrophobic bilayer and the aqueous environment both outside of the ER membrane and within its lumen. X-ray crystallography and NMR spectroscopy are traditional means to elucidate the structural organization of proteins; however, both techniques require relatively large amounts of pure proteins, and NMR is only useful with lowmolecular-weight proteins; therefore these methods have limitations with respect to membrane complexes (6). Structural genomics and protein modeling are also limited in the case of the OT complex because of the lack of an extensive data base of known folds or structures of membrane proteins existing in part in the ER membrane and also in the aqueous environment of either the cytoplasm or the ER lumen. As an alternative approach to study the structural organizations of the OT subunits, we carried out chemical cross-linking experiments using a molecule that contains homo bifunctional amino-reactive groups with a 12 Å hydrophobic spacer arm between them to examine interactions between the potential domains of OT subunits. We took advantage of the membrane permeability and thiocleavability of the cross-linking reagent dithiobis-succinimidyl propionate (DSP) and prepared a collection of 20 yeast strains in which different OT subunits were HA- or Myc-tagged to study the inter-relationship of OT subunits. Based on examination of the interactions between any two subunits of OT, a comprehensive view of the inter-relationship of these proteins was developed. To further explore which segments are involved in the interactions of OT subunits, we investigated the distribution of lysine residues in the nine subunits of OT by which a covalent linkage between two proteins could occur and found that although different OT subunits display different distributions of lysine residues across their peptide backbones, all of them have lysine residues, which are located at the interface of the hydrophilic and trans-

33078

This paper is available on line at http://www.jbc.org

Insight into the Inter-relationship of OT Subunits membrane segments. This finding leads to our speculation that probably these lysine residues located in the peripheral segments of transmembrane domains contribute to the cross-linking reaction. We then tested this idea by monitoring the crosslinking of OT subunits with two other enzymes, Alg1p and Pdi1p, which are not included in the OT complex but act in assembly of the lipid-linked oligosaccharide substrate, and protein folding disulfide bond formation of nascent proteins, respectively (7, 8). It was found that Alg1p, an ER membraneassociated enzyme, was cross-linked to Ost1p, whereas the ER luminal enzyme, Pdi1p was not cross-linked to any of the OT subunits. Taking these results together, we concluded that the interactions among OT subunits occur through transmembrane domains and/or the region proximal to it, rather than their luminal or cytoplasmic domains. EXPERIMENTAL PROCEDURES

Materials—The cross-linking reagent DSP was purchased from Pierce. Antibodies to HA and Myc epitopes were purchased from SantaCruz Biotechnology Inc. (Santa Cruz, CA). Protein G-agarose beads were supplied by Invitrogen (Carlsbad, CA). Procedures—Yeast microsomes were prepared as described (9). Protein assays were performed using a BCA protein assay kit purchased from Pierce. Lithium acetate transformation was utilized for all yeast transformations in this study; standard yeast media and genetic techniques were used (10 –12). Strains—The yeast wild type strain W303-1a (MATa ade2 can1 his3 leu2 trp1 ura3) was used as the origin strain to generate QYY102 (W303-1a OST2::HA-his5⫹ S. pombe), QYY104 (W303-1a SWP::HAhis5⫹ S. pombe), QYY105 (W303-1a OST5::HA-his5⫹ S. pombe), AYY7 (W303-1a OST6::HA-his5⫹ S. pombe), and AYY10 (W303-1a OST4:: myc-KanR). QYY102 (W303-1a OST2::HA-his5⫹ S. pombe), QYY105 (W303-1a OST5::HA-his5⫹ S. pombe), AYY7 (W303-1a OST6::HA-his5⫹ S. pombe), AYY10 (W303-1a OST4::myc-KanR) as well as RGY330 (W303-1a OST3::HA-his5⫹ S. pombe) were used to generate double epitope-tagged strains. The strains used in this study are summarized in Table I. Construction of Plasmids and Strains—QYY102 (W303-1a OST2:: HA-his5ⴙ S. pombe), QYY104 (W303-1a SWP::HA-his5ⴙ S. pombe), QYY105 (W303-1a OST5::HA-his5ⴙ S. pombe) were prepared using ME-3 plasmid and the same method as mentioned earlier (5, 13). AYY7 (W3031-a OST6::3HA-his5⫹ S. pombe) and AYY10 (W303-1a OST4:: 13myc::KanR) were prepared as follows: A 1019-bp DNA fragment encoding the OST6 gene was obtained by PCR amplification using genomic DNA of yeast as the template (primers were: 5⬘-GCTAGCAGCTGATGAAGTGGTGTAGCACATAC-3⬘ and 5⬘-GCTAGGGATCCCAAAAACAATTGGGTACCCTGG-3⬘). This fragment was then ligated into the PvuII/BamHI-digested plasmid pFA6a-3HA-His3MX6 (14) followed by subcloning of a 300-base pair DNA fragment encoding the downstream region of the OST6 gene into the SacI/SacII sites to generate the plasmid, pFA6a-OST6 –3HA-His3-300 bp downstream of OST6MX6. Using this plasmid as the template, the PCR product, which consisted of the OST6 gene followed by 3 HA epitope repeats, the HIS3 gene, and the 300 base pairs downstream of the OST6 gene, was transformed into the W303-1a strain, resulted in the strain AYY7 (W303-1a OST6::HA- his5⫹ S. pombe). The colonies were selected on -His plates, and the integration was confirmed by colony polymerase chain reaction (PCR), the expression of epitope-tagged protein was confirmed by SDS/PAGE and Western blot analysis. The same procedure was utilized to generate AYY10 (W303-a OST4::13myc-KanR) except HindIII/SalI sites were used to replace PvuII/BamH I sites and plasmid pFA6a-13myc-KanMX6 was used to replace plasmid pFA6a3HA-His3MX6. The colonies were consequently selected on YPAD ⫹ G418 plate. QYY102, QYY105, AYY7, RGY330, and AYY10 were used as recipient strains for preparing the constructs that contain two chromosomal epitope tags in a single strain (Table I) using the same strategy as described above. All the constructs made in this study were confirmed by colony polymerase chain reaction (PCR), and the expression of proteins was confirmed by Western blot analysis using suitable antibodies. The integration of OT complex in these strains was checked by coimmunoprecipitation followed by SDS/PAGE and then Western blot analysis using available antibodies (anti-HA, anti-Myc, and antibodies to Ost1p, Wbp1p, Swp1p, and Stt3p).

33079

The Co-immunoprecipitation Procedure Using Mild Detergent—This procedure was modified from Karaoglu et al. (15). Yeast microsomes were collected by centrifugation, and the membrane pellet was resuspended in 5% glycerol, 20 mM Tris-Cl, pH 7.4, 5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride (final concentration of protein, 10 mg/ml). The microsomal suspention was then adjusted to contain 1.5% digitonin, 0.5 M NaCl, 20 mM Tris-Cl, pH 7.4, 3.5 mM MgCl2, and 1 mM MnCl2. The mixture was centrifuged for 20 min at 55,000 revolutions/minute, and the clarified supernatant was used for immunoprecipitation. The immunoprecipitation conditions were described earlier (15). Samples were resolved on SDS/PAGE and transferred to nitrocellulose membranes. The membranes were then probed with the indicated antibodies. The Conditions for Cross-linking and Immunoprecipitation Using Harsh Detergent—Microsomes were collected as mentioned above, and the membrane pellet (final concentration of protein, 10 mg/ml) was resuspended in phosphate-buffered saline, 1 mM phenylmethylsulfonyl fluoride, and incubated on ice for 0.5 h. The cross-linking reagent DSP, which was dissolved in Me2SO, was then added to the suspension to a final concentration of 2 mM; the final concentration of Me2SO was 1% (For control analysis, the same amount of Me2SO alone was added). The cross-linking reaction was carried out at room temperature for 0.5 h followed by the addition of 5 ␮l of 1 M Tris-HCl pH 7.4 to quench the reaction. The resulting membrane proteins were then solubilized with 1 ml of LB buffer containing 10 mM Hepes, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 1% Triton X-100, 0.2% SDS, and protease inhibitors. Immunoprecipitation was carried out using mouse anti-HA antibody followed by affinity purification using protein G-agarose beads. After incubation with protein G-agarose beads for 2 h at room temperature, the immunocomplex was washed three times with 1 ml of LB buffer each time and once with 1 ml of TBS. SDS sample buffer containing 50 mM DTT was used to cleave the cross-linking reagent and elute the proteins from protein G-agarose beads. The protein samples were resolved by SDS/ PAGE followed by Western blot analysis with anti-HA, anti-Myc, antiOst1p, -Wbp1p, -Swp1p, or -Stt3p antibodies. RESULTS

The Inter-relationship of Yeast OT Subunits Could Be Studied in a Pairwise Manner Using Cross-linking Reagent—As outlined in Fig. 1, when the microsomes prepared from a series of yeast strains (Table I) were treated with DSP, it was expected that two proteins would be covalently linked together if they contain lysine residues located within 12 Å of each other. To identify the cross-linked proteins, the complex was solubilized using a harsh detergent treatment to disrupt the physical interactions between proteins while still maintaining the previously generated disulfide covalent linkage. Since one of the OT subunits was HA-tagged, its covalent binding partner would be co-precipitated in the subsequent immunoprecipitation procedure using monoclonal HA antibody. As shown in Fig. 1, in addition to the cross-linked protein with the epitope-tagged protein, obviously, another pair of proteins could also be covalently coupled provided their lysine residues are located within 12 Å. However, they could not be recovered in the immunoprecipitation procedure because of the absence of the epitope tag. After further purification of the co-precipitated complex by immobilizing it on protein G-agarose beads followed by washing, elution, and finally cleaving the disulfide bond, the bound partner could be identified if it reacted to one of the OT subunit antibodies. For those potential binding partners whose antibodies were not available (Ost2p, Ost3p, Ost4p, Ost5p, and Ost6p), we attached two different epitope tags (HA and Myc) to two different subunits in a single strain so that the cross-linked partner could be identified by immunoblotting with anti-Myc antibody (Fig. 1). Thus, the inter-relationship of any two OT subunits could be studied by using different combinations of strains. Based on this pairwise study, a comprehensive view on the inter-relationship of yeast OT subunits could be developed. To obtain the original expression level of the proteins, the chromosomal copy of the corresponding genes were epitope-tagged. The growth rates of these generated strains

33080

Insight into the Inter-relationship of OT Subunits

FIG. 1. Schematic representation of the experimental strategy. Two alternatives for immunoprecipitation are shown: 1) harsh detergent conditions, which will disrupt the physical interactions among OT subunits but retain the disulfide bond linkages between subunits intact; and 2) gentle, non-denaturing conditions, which may retain non-covalent interactions among OT subunits. Dashed line represents non-covalent interactions. TABLE I List of strains used in this study

a b

Strains

Genotype

Source

W3031-a QYY101 QYY102 QYY103 QYY104 QYY105 RGY330 QYY697 AYY1 AYY2 AYY3 AYY4 AYY5 AYY6 AYY7 AYY10 AYY11 AYY12 AYY14 AYY15 ALGIHA

MATa ade2 can1 his3 leu2 trp1 ura3 W303–1a OST1::HA-his5⫹ S. pombe W303–1a OST2::HA-his5⫹ S. pombe W303–1a WBP::HA-his5⫹ S. pombe W303–1a SWP::HA-his5⫹ S. pombe W303–1a OST5::HA-his5⫹ S. pombe W303–1a OST3::HA-his5⫹ S. pombe W303–1a STT3::HA-his5⫹ S. pombe RGY330 OST4::13myc-KanR QYY105 OST4::13myc-KanR QYY105 OST6::13myc-TRP1 RGY330 OST5::13myc-TRP1 RGY330 OST6::13myc-TRP1 AYY7 OST4::13myc-KanR W3031-a OST6::3HA-his5⫹ S. pombe W303-a OST4::13myc-KanR RGY330 OST2::13myc-TRP1 AYY7 OST2::13myc-TRP1 QYY105 OST2::13myc-TRP1 QYY102 OST4::13myc-KanR SEY6210 ALG1::3HA-KanR

Lab stock Qi Yan, et al. (13) This study G.T. Li, et al.a This study This study Karaoglu, D. et al. (15) Qi Yan, et al. (5) This study This study This study This study This study This study This study This study This study This study This study This study X.D. Gao, et al.b

G. T. Li, Q. Yan, H. O. Oen, and W. J. Lennarz, submitted for publication. X. D. Gao, A. Wishikawa, and N. Dean, unpublished data.

were the same as the wild-type strain, which indicated that the epitope tags did not disrupt OT complex formation (data not shown).

Four Essential Gene Products, namely Ost1p, Wbp1p, Swp1p, and Stt3p Were Found to Be Cross-linked to Each Other—As shown in Fig. 2A, the microsomes of the strains


33081

FIG. 2. Cross-linking reveals pairwise attachment of five essential gene products. A, microsomes of strain WBP1HA, SWP1HA, Stt3HA, respectively were treated as outlined in Fig. 1 and then immunoblotted with anti-Ost1p antibody; B, with anti-Swp1p antibody; C, with anti-Wbp1p antibody; D, with anti-Stt3p antibody; E, the microsome of strain Ost2pHA were treated as outlined in Fig. 1 and then immunoblotted with the indicated antibodies. g, gentle, non-denaturing detergent condition (1.5% digitonin); h, harsh detergent condition, (1% Triton X-100 and 0.2% SDS).

that had Wbp1p, Swp1p, and Stt3p HA epitope-tagged were treated with DSP followed by immunoprecipitation using mouse anti-HA antibody under harsh detergent solubilization conditions. After cleaving the disulfide bond, Western blot analysis revealed that Ost1p had been cross-linked to and then co-precipitated with Wbp1p, Swp1p, and Stt3p, respectively (lanes 3, 6, and 9). In contrast, in the absence of the cross-linking reagent (lanes 2, 5, and 8), no co-precipitation of Ost1p was observed. As expected, immunoblots of the samples that were subjected to co-immunoprecipitation under gentle, non-denaturing conditions (lanes 1, 4, and 7) also showed the corresponding co-precipitation of Ost1p indicating that these three OT subunits individually remained associated with Ost1p under non-denaturing conditions. We further examined these biochemical interactions reversibly by utilizing the strain that had Ost1p HA-tagged to undertake the same experimental procedures and immunoblotting with anti-Wbp1p, Stt3p, and Swp1p antibody. Western blot analysis shown in Fig. 2 confirmed the biochemical interactions between Ost1p with Wbp1p, Swp1p, or Stt3p (Fig. 2B, lanes 1, 2, and 3; Fig. 2C, lanes 1, 2, and 3 and Fig. 2D, lanes 4, 5, and 6). Similarly, as shown in Fig. 2, B–D, pairwise biochemical interactions between any of the two proteins among these four essential gene products were demonstrated. However, using the same approach, another essential gene product Ost2p, was found to be cross-linked to Ost1p, Wbp1p, and Stt3p, but not to Swp1p (Fig. 2E), indicating Ost2p and Swp1p are not in close proximity to each other although both are located close to Ost1p, Wbp1p, and Stt3p. We noted that in all cases the bands of expected proteins were more intense under non-denaturing conditions without cross-linking than the corresponding ones after cross-linking followed by harsh detergent treatment. This

finding was not unlikely because the chemical cross-linker would not completely couple all the pairs of proteins in a half-hour incubation. Ost1p Directly Interacts with All Other Eight Subunits— A variety of epitope-tagged strains, OST2HA, OST3HA, OST4myc, OST5HA, OST6HA, STT3HA, WBP1HA, and SWP1HA, were used in this cross-linking study. As shown in Fig. 3, Western blot analysis using anti-Ost1p antibody revealed that Ost1p could directly interact with all other eight subunits of OT. Ost1p was the only OT subunit that could be cross-linked to all of the other eight subunits. Since previous studies using techniques of non-denaturing immunoprecipitation and 35S labeling (15), as well as the result of blue native electrophoresis (16), revealed that the yeast OT is composed of nine subunits in approximately equimolar amounts, our result led us to propose that Ost1p may reside in the core of the OT complex probably playing an important role in its assembly. This proposal may explain the observations in a number of earlier studies that different subunit compositions of the OT complexes have been proposed using different purification procedures, but in all cases Ost1p was always shown to be one of the components of the complex by different groups (16 –20). Ost4p and Ost5p Are Found Only to Interact with Specific Components of OT Complex—Ost4p and Ost5p are two small, hydrophobic, non-essential gene products in the OT complex. As shown in Fig. 3, lanes 6 and 8, Ost4p and Ost5p were found to be in close proximity with Ost1p. However, unlike some of the essential gene products in the OT complex, which contain long luminal tails (Wbp1p, Stt3p, and Swp1p) and were found could be cross-linked to Ost1p (see above), Ost4p and Ost5p contain short luminal tails with most of their peptide backbones residing within the ER membrane. Since it was demonstrated that the active site of OT is 30 – 40 Å away from the ER

33082


FIG. 3. Ost1p was cross-linked to all other 8 subunits of OT. The microsomes of indicated strains were subjected to the procedures outlined in Fig. 1 under harsh detergent conditions (1% Triton X-100 and 0.2% SDS) and, following SDS/ PAGE, were immunoblotted with polyclonal anti-Ost1p antibody. As indicated, in all cases cross-linking via DSP revealed that all eight OT subunits had been individually covalently linked to Ost1p.

membrane in the ER lumen (21), we speculate that the close contact between Ost1p with Ost4p and Ost5p may not contribute to the catalytic mechanism of OT; instead, these interactions may indicate their roles on assembly of the complex. Indeed, in contrast to the interaction pattern of the essential gene products that interact with each other as mentioned above, Ost4p and Ost5p are found only to interact with specific components of those essential gene products (Table II). The exact function of these small, hydrophobic proteins is not clear. However, taking together our cross-linking study and the results of a genetic screen on OST5 gene (22) as well as a biochemical study on Ost4p, which demonstrates that Ost4p specifically functions to stabilize the subcomplex of Stt3p-Ost4pOst3p (23), we suggest that Ost4p and Ost5p may function as additional anchors in the membrane required for optimal stability of the essential gene products, Stt3p and Ost1p, respectively. Despite their common accessory role, Ost4p displays multiple interactions with other OT subunits compared with Ost5p (Table II). This finding may fulfill the genetic observation that deletion of the OST4 gene renders the inviability of the cell at 37 °C (24) whereas the OST5 gene is not essential for the viability of the yeast cell, and its depletion only results in a reduction of OT activity to about half of wild-type levels (22). Ost3p and Ost6p Showed an Identical Pattern of Cross-linking with Other OT Subunits—Based on the overall cross-linking study of pairs of OT subunits, a comprehensive view on their inter-relationships is summarized in Table II. To our surprise, Ost3p and Ost6p showed an identical pattern of crosslinking to other subunits. Interestingly, these two proteins could not be co-immunoprecipitated by each other under gentle, non-denaturing detergent conditions (1.5% digitonin) (Fig. 4, lane 1), but they could be cross-linked to each other by DSP (Fig. 4, lane 3), indicating that they do not exist in the same complex but nevertheless they are close to each other. Furthermore, the surprising similarity of their interaction pattern with other OT subunits in this study implied that they may have some similar or redundant functions in the cell. This proposal is consistent with a genetic study that revealed that the single null mutant of either ost3 or ost6 exhibits a low level of underglycosylation whereas the double mutants of ost3 and ost6 shows severe underglycosylation, as well as a biophysical analysis, which demonstrated that these two proteins are strikingly similar with respect to their hydropath plot (16). OT Subunits Are Not Equal Concerning Their Ability to Remain Associated with Other OT Complex Components under Non-denaturing Immunoprecipitation Conditions—We tested all of the nine yeast OT subunits for their ability to co-immunoprecipitate the other OT subunits under gentle, non-denaturing detergent conditions without cross-linking. We found that HA-tagged Ost5p did not remain associated with Swp1p in

the co-immunoprecipitation experiment using mouse HA antibody (Fig. 5A, lane 10). Moreover, Ost2p did not remain associated with Stt3p, Ost4p, and Ost6p under the same conditions (Fig. 5B, lanes 1, 4, and 7). These observations indicated that by using this technique, no direct physical interactions can be detected between the pairs of OT subunits mentioned above, namely, Ost5p and Swp1p, Ost2p, and Stt3p, Ost4p, Ost6p, respectively. Among them, the physical interaction between Ost2p and Ost6p as well as between Ost2p and Stt3p was not observed under non-denaturing detergent conditions for coimmunoprecipitation. However, each pair was shown to be cross-linked together by DSP (Fig. 5B, lanes 3 and 6), which indicated that these two pairs of proteins are within 12 Å to each other, whereas the physical contact between their polypeptide chains, if any, is insufficient to result in co-immunoprecipitation. In the case of Ost5p and Swp1p, and of Ost2p and Ost4p, our results revealed that they are neither in close proximity nor do they physically interact. Selectivity of the Cross-linker—As demonstrated in Table II, some OT subunits exhibit multiple interactions with other components; we next asked if these multiple interactions were artifacts caused by the lack of selectivity of the cross-linker we used. To investigate this idea, we chose two ER proteins, an ER luminal protein, protein-disulfide isomerase (PDI) (8), and an ER membrane protein Alg1p, an enzyme encoding a ␤-1,4mannosyltransferase that adds the first mannosyl residue to GlcNAc2-PP-Dol in the assembly of full-length lipid-linked oligosaccharides (7), to examine their cross-linking with OT subunits. The results revealed that no cross-linking was observed between OT subunits Ost1p, Ost2p, Ost3p, Stt3p, Wbp1p, Swp1p, with Pdi1p (Fig. 6A), an abundant ER luminal protein. However, in the case of Alg1p, Ost1p among the OT subunits tested (Ost1p, Stt3p, Wbp1p, and Swp1p) was found to be cross-linked to the Alg1p (Fig. 6B). It is noteworthy that unlike OT subunits, Alg1p did not co-immunoprecipitate with Ost1p under non-denaturing conditions (Fig. 6B, lane 1), which suggests Alg1p does not exist in the OT complex, but could be chemically covalently bound to Ost1p by DSP, and hence is likely to be within 12 Å of OT in the ER membrane. These observations clearly indicate that the cross-linker can only react with two proteins that are located within a complex or in close proximity. This observation also implies that probably there is some common characteristic of the ER membrane proteins studied by us that distinguishes them from ER luminal proteins and relates to the cross-linking reaction occurring among them. The Portion of Transmembrane Domains and/or the Peripheral Segments of the Transmembrane Domains of OT Are Responsible for the Interaction among Its Subunits—Only the amino group in the side chain of lysine residues on the protein


33083

TABLE II Summary of interactions observed among OT subunits OSt1p

Ost1p Ost2p Ost3p Ost4p Ost5p Ost6p Stt3p Wbp1p Swp1p

⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹

Ost2p

Ost3p

Ost4p

Ost5p

Ost6p

Stt3p

Wbp1p

Swp1p

⫹

⫹ ⫹

⫹ ⫺b ⫹

⫹ ⫺ ⫺ ⫺

⫹ ⫹ ⫹ ⫹ ⫺

⫹ ⫹ ⫹ ⫹ ⫺ ⫹

⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫹

⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹

a

⫺ ⫹ ⫺ ⫺ ⫹ ⫹ ⫺

⫹ ⫺ ⫹ ⫹ ⫹ ⫺

⫺ ⫹ ⫹ ⫺ ⫺

⫺ ⫺ ⫹ ⫺

⫹ ⫹ ⫺

⫹ ⫹

⫹

⫹, detected interactions with DSP. b ⫺, not detected interactions with DSP. a

FIG. 4. Ost3p and Ost6p do not exist in the same complex (lane 1) but could be cross-linked by DSP (lane 3). The microsome of strain OST3HA OST6myc were subject to the procedure outlined in Fig. 1 and then immunoblotted with anti-Myc antibody.

backbone can react with cross-linker if the distance between them is 12 Å or less. Therefore, it was of interest to investigate the lysine residue distribution across the amino acid sequences of OT subunits. Based on the information of PSORT II prediction program as well as experimental data, the topology, and the lysine residues in approximate proportion to their positions in the polypeptide chains is shown in Fig. 7A. It was demonstrated that except for Ost4p, a miniprotein with 36 amino acid residues, which contains only one lysine residue, all other eight subunits contain multiple lysine residues. Some of them (Ost1p, Wbp1p, Stt3p) contain abundant lysine residues in their C-terminal luminal domains, which is not unexpected because of the hydrophilic environment of the ER lumen. Additionally, although different OT subunits display different distribution patterns of lysine residues, a common phenomenon is that each subunit of the OT complex, except for Ost4p and Wbp1p, has lysine residues that are located at the interface of the hydrophilic and transmembrane segments, one to two amino acids away from the hydrophobic transmembrane domain (Fig. 7B). In the case of Ost4p and Wbp1p, their corresponding lysine residues are slightly further from the ER membrane (Lys35 relative to the transmembrane segment of Ost4p from 9 to 25 amino acid residues; Lys421,422 relative to the transmembrane segment of Wbp1p from 398 to 414 amino acid residues) (Fig. 7B). These lysine residues probably serve as stop-transfer signals to define the membrane orientation of transmembrane proteins. This observation initiates the question once again about the segments through which the crosslinking reaction occurs. Based on topology prediction only two components, Ost3p and Ost6p, are shown to have relatively long cytoplasmic domains whereas the others contain much shorter corresponding segments, and it was proposed previously that the active site of OT exists 30 – 40 Å away from the

FIG. 5. OT subunits are not equal in their ability to remain associated with other OT complex components under non-denaturing immunoprecipitation conditions. A, Swp1p did not co-precipitate with Ost5HAp under non-denaturing condition treatment. The microsomes of strain OST5HA were subjected to the procedures outlined in Fig. 1 and then immunoblotted with anti-Ost1p, Wbp1p, Stt3p, or Swp1p antibody. B, Ost2HAp did not remain associated with Stt3p, Ost4p and Ost6p under non-denaturing condition treatment. Microsomes of the indicated strains were subjected to the procedures outlined in Fig. 1 and then immunoblotted with anti-Stt3p or anti-Myc antibody.

ER membrane in the ER lumen (21). Therefore, we speculate that either those lysine residues that are located close to the transmembrane domain or the lysine residues residing further in the ER lumen are responsible for the cross-linking reactions. If those distal lysine residues in the ER lumen contribute to the cross-linking reaction among OT subunits, it seemed unlikely that Pdip, an abundant ER luminal protein, which acts sequentially with OT in the process of protein folding and assembly in the ER (8) and carries many lysine residues scattered across its peptide backbone, could not be cross-linked with a collection of OT subunits, even those OT subunits that contain long luminal tails (Ost1p, Wbp1p, Stt3p) (Fig. 6A). Thus, it is highly possible that those lysine residues that are proximal or within the transmembrane domains of OT subunits are involved in cross-linking among them. Indeed, this idea was further supported by the cross-linking of OT subunit

33084


FIG. 6. Selectivity of the cross-linker. A, no interactions were detected between OT subunits and the ER luminal protein Pdip. The microsomes of strain Ost1pHA (lane 1), Ost2pHA (lane 2), Ost3pHA (lane 3), Wbp1pHA (lane 4), Swp1pHA (lane 5), and Stt3pHA (lane 6), respectively were subjected to the procedures outlined in Fig. 1 in the presence of DSP under harsh detergent conditions (1% Triton X-100 and 0.2% SDS), precipitated with mouse anti-HA antibody, treated with 50 mM dithiothreitol, and following SDS/PAGE, immunoblotted with polyclonal anti-Pdip antibody. Lane 7 shows authentic Pdip expressed in an E. coli lysate and Western blot with anti-Pdip antibody. B, Ost1p interacts with Alg1p. The microsomes of strain Alg1pHA were subjected to the procedures outlined in Fig. 1 in the presence of DSP under both gentle detergent conditions and harsh detergent conditions, immunoprecipitated with mouse anti-HA antibody, treated with 50 mM dithiothreitol, and following SDS/PAGE, immunoblotted with polyclonal antiOst1p antibody.

Ost1p with another ER membrane protein, Alg1p, with regard to their relationship as follows: several OT subunits (Ost1p, Wbp1p, Swp1p) were found containing the consensus sequence, which is distinctive among dolichol binding enzymes in their membrane-spanning domains and has been proposed to bind to the assembled lipid-linked oligosaccharides (25, 26), and Alg1p was recently found to exist as a homo-oligomer and perform dual functions in sugar transfer and in the recognition of dolichol phosphate-derived substrates.2 Therefore, Alg1p could serve as a good model to investigate our hypothesis if it ideally interacts with one or more OT subunits that are involved in recognition and the binding of lipid-linked oligosaccharide donors. As expected, our result (as shown in Fig. 6B) strongly supports our proposal that the cross-linking of these ER membrane-anchored OT subunits by DSP occurs through those lysine residues that are located proximal to the transmembrane domains of OT subunits. DISCUSSION

Previous genetic and biochemical studies have suggested that the nine OT subunits of yeast exist in three subcomplexes: 2

X. D. Gao, A. Wishikawa, and N. Dean, unpublished data.

Ost1p-Ost5p, Stt3p-Ost4p-Ost3p, and Ost2p-Wbp1p-Swp1p (15). However, a number of observations that are not consistent with this model have been reported. By using an anti-Ost6p affinity matrix, a novel “trimeric” complex could be isolated that consisted of Ost1p, Wbp1p, and Ost6p as main components (16). A high-throughput screen for protein-protein interactions of nearly all 6,000 predicted proteins in S. cerevisiae revealed that using Ost1p as a bait protein, Ost6p, Stt3p, and Wbp1p were found to interact closely with Ost1p (27). Finally, the split ubiquitin system, which is a recently developed approach to study membrane protein interactions in vivo (28) demonstrated that Ost1p directly interacts with Wbp1p. These different groupings of yeast OT subunits obviously were based on a variety of different experimental approaches by different researchers. In this article, we utilized a single biochemical method to study subunit interaction of potential pairs of all OT subunits. The cross-linking reagent DSP used in this study has been successfully used in a variety of applications for intracellular and intramembrane conjugation of proteins (29 –33) and has exhibited the expected selectivity with respect to the distance among proteins in our study. The results using this approach provide a new and more comprehensive understanding of interrelationship between the components of the OT complex. Furthermore, compared with genetic means, this biochemical method is more direct for identification of protein-protein interactions. Among the four essential gene products that were shown to be in close proximity to each other, namely, Ost1p, Wbp1p, Stt3p and Swp1p, physical interaction between Wbp1p and Swp1p has been observed both genetically and biochemically by te Heesen et al. (26) and this interaction has been further confirmed by the cross-linking approach used in this study. However, the close inter-relationship between Ost1p and Stt3p was only indicated by a photolabeling study. A photoreactive pentapeptide, 125I-bh-Asn-Bpa-Thr-Ala-Bpa-Am, was found to react with three subunits of OT, Ost1p, Stt3p, and Ost3p. Since the photoreactive group can only react with carbon atoms within the vicinity of 3.1 Å (34), this result implied that these three subunits of OT must be in close proximity in terms of their tertiary structures. However, among them, Ost3p and Stt3p have previously been suggested to be within a subcomplex of Stt3p-Ost4p-Ost3p (15), whereas Ost1p was not detected in the same subcomplex using the same approach. Our results showing that Stt3p is in close proximity to Ost1p and Ost3p (Figs. 2 and 3, and Table II), strongly support the earlier photolabeling experiment and offer direct evidence that these three subunits are in close proximity. As for another essential gene product, Ost2p, it was previously proposed by a genetic approach that Ost2p, Wbp1p, and Swp1p form a subcomplex (35). However, a biochemical interaction between Ost2p and Swp1p was not observed in our experiments presumably because the distance between lysine groups on these two proteins are more than 12 Å from each other. Rather, another subunit, Wbp1p, bridges them via interaction with each of them. This proposal is in agreement with the result of a genetic study on mammalian homologues of Wbp1p, Swp1p, and Ost2p (Ost48, Ribophorin II, and DAD I, respectively) (36). This study revealed that interactions exist between Ost48 and ribophorin II and between Ost48 and DAD I, respectively, but no direct interaction between Ribophorin II and DAD I was observed (36). Unlike the essential gene products in OT complex, which were found to be close to each other, two non-essential OT gene products, Ost4p and Ost5p, interacted only with specific members of OT complex, implying their accessory role to specific


33085

FIG. 7. Lysine residue distribution of the proteins studied. A, the topology of the proteins studied in this article and the distribution of lysine residues in approximate proportion to their positions in the polypeptide chains. Except for Swp1p, predictions of topology and transmembrane domains are based on PSORT II program, the topology of Swp1p is based on the work of te Heesen et al. (26) and the topology of Ost4p is based on the work of Kim et al. (23) and all the others are based on PSORT II prediction program. Red line represents a lysine residue. B, diagram of transmembrane domains and the location of lysine residues that could be involved in cross-linking reactions. Hatched bar represents the transmembrane domain.

33086


FIG. 8. Proposed cluster composition including the Sec61 pore channel complex, OT complex, and complex of lipid-linked oligosaccharide synthases (LLOS). As discussed, the speculation is based on following results: Sss1p in Sec61 channel binds to Wbp1p in the OT complex, Ost1p in OT complex binds to Alg1p in the LLOS and Ost1p directly interacts with Wbp1p. See Romisch et al. (44), Stagljar et al. (28), Gao et al. (X. D. Gao, A. Nishikawa, and N. Dean, submitted for publication) and our the current study. The masses of the components are not to scale.

components of OT. It is interesting to note that the presence of small hydrophobic proteins is not uncommon in multisubunit membrane proteins complexes, such as mammalian signal peptidase complex SP12 (37), the translocation complexes of the ER Sbh1p (38) and of the mitochondria Isp6p (39) in yeast, and the bacterial cytochrome c oxidase SUIV (40). These small subunits are not essential for catalytic function of the respective protein complex but are required for complex assembly and optimal enzyme activity. Perhaps such small subunits serve as accessory elements to contribute to the assembly or stability of many membrane protein complexes. It was believed that immunoprecipitation of an epitope tag, which was attached to a component of a protein complex could recover all other members in the complex with non-denaturing detergent treatment. However, in the case of OT complex, our experiment clearly establishes that these subunits are not equal in their ability to associate with each other. Our findings suggest that those proteins that are in the core of the enzyme complex interact tightly or play a significant role in assembling the complete protein complex. In contrast, other proteins that are not in the core and serve only as accessory components are so weakly associated with other proteins that their binding cannot be detected even under non-denaturing detergent conditions. In this study, we utilized a recently developed program, PSORT II, to predict the topology and potential transmembrane domains of proteins we studied. It is noticeable that different programs may result in different proposals on the topology of a protein, but the transmembrane domains as well as the relative distance between specific amino acid residues (lysine residues in this study) and ER membrane do not change using different prediction programs. In the case of OT subunits, the topology of Ost1p and Wbp1p are experimentally established and are consistent with the prediction of PSORT II program (41– 43), whereas no definitive experimental studies are available for the other proteins studied. For this reason, we utilized this single prediction program to analyze all the proteins in this study. As mentioned above, since the transmembrane domains as well as the relative distance between lysine residues and the ER membrane are not changed with different programs, the uncertainty on the topology will not influence the proposals we put forward based on the PSORTII program. Studying protein-protein interactions is a useful tool to un-

derstand the functions of subunits as well as the mechanism of enzyme catalysis. It is noteworthy that recently some of the OT subunits were shown to interact with other proteins and therefore were proposed to participate in processes other than the N-glycosylation per se. Stt3p was found to interact with Kre5p and Kre9p in a synthetic lethal screen and was proposed to be involved in Pkc1-regulated ␤-1,6-glucan cell wall synthesis.3 Wbp1p was found to interact with one of the Sec61p proteinaceous channel components, Sss1p, by use of a split-ubiquitin system which indicated that at least some OT subunits are in close proximity to the Sec61 channel in the ER membrane (44). In addition, in the current study Ost1p was found to be crosslinked to Alg1p, which raises the possibility of interaction of the lipid-linked oligosaccharide assembly pathway and the Nglycosylation process. In fact, enzymes involved in protein translocation followed by N-glycosylation, protein folding, and formation of disulfide bonds in the ER may be tightly coupled (2). Thus it is possible that OT is not only in proximity to the Sss1p component of the Sec61 complex, but also to the enzymes of the lipid-linked oligosaccharide synthesis pathway. Perhaps a cluster exists around each Sec61 pore that contains OT complexes and complex of the lipid-linked oligosaccharide synthase (LLOS). This hypothesis is shown diagrammatically in Fig. 8. Hopefully, in the near future, this hypothesis can be tested by more extensive studies. Acknowledgments—We thank Dr. Markus Aebi for generous gifts of anti-Wbp1p and anti-Swp1p antibodies, Dr. Reid Gilmore for antiOst1p antibody, and Dr. Satoshi Yoshida for anti-Stt3p antibody. We give special thanks to Dr. Xiaodong Gao for a generous gift of strain ALG1HA and helpful discussion. We also appreciate Geng Tian in our laboratory for providing polyclonal anti-Pdi1p antibody. We are grateful to Dr. Manasi Chavan, Dr. Tadashi Suzuki, Dr. Guangtao Li, and Dr. Hangil Park for stimulating discussion and help. We appreciate Eddie Lanier for manuscript revision. REFERENCES 1. Knauer, R., and Lehle, L. (1999) Biochim. Biophys. Acta 1426, 259 –273 2. Dempski, R. E., Jr., and Imperiali, B. (2002) Curr. Opin. Chem. Biol. 6, 844 – 850 3. Park, H., and Lennarz, W. J., (2000) Glycobiology 10, 737–747 4. Pathak, R., Hendrickson, T. L., and Imperiali, B. (1995) Biochemistry 34, 4179 – 4185 5. Yan, Q., and Lennarz, W. J., (2002) J. Biol. Chem. 277, 47692– 47700 6. Back, W. J., Sanz, M. A., De Jong, L., De Koning, L. J., Nijtmans, L. G. J., De Koster, C. G., Grivell, L. A., Van Der Spek, H., and Muijsers, A. O. (2002) Protein Sci. 11, 2471–2478 7. Albright, C. F., and Robbins, P. W. (1990) J. Biol. Chem. 265, 7042–7049 8. Luz, J. M., and Lennarz, W. J. (1996) EXS 77, 97–117 9. Karaoglu, D., Kelleher, D. J., and Gilmore, R. (1995) J. Cell Biol. 130, 567–577 10. Elble, R. (1992) Biochemistry 13, 18 –20 11. Rose, M., Winston, F., and Hieter, P. (1990) Methods in Yeast Genetics, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 12. Sherman, F. (1991) Methods Enzymol. 194, 3–21 13. Yan, Q., Prestwich, G. D., and Lennarz, W. J. (1999) J. Biol. Chem., 274, 5021–5025 14. Longtine, M. S., McKenzie, A., 3rd, Demarini, D. J., Shh, N. G., Wach, A., Brachat, A., Philippsen, P., and Pringle, J. R. (1998) Yeast, 14, 953–961 15. Karaoglu, D., Kelleher, D. J., and Gilmore, R. (1997) J. Biol. Chem., 272, 32513–32520 16. Knauer, R., and Lehle, L. (1999) J. Biol. Chem. 274, 17249 –17256 17. Knauer, R., and Lehle, L., (1994) FEBS Lett. 344, 83– 86 18. Kelleher, D. J., and Gilmore, R. (1994) J. Biol. Chem. 269, 12908 –12917 19. Pathak, R., Parker, C. S., and Imperiali, B. (1995) FEBS Lett., 362, 229 –234 20. Zufferey, R., Knauer, R., Burda, P., Stagljar, I., te Heesen, S., Lehle, L., and Aebi, M. (1995) EMBO J., 14, 4949 – 4960 21. Nilsson, I. M., and von Heijne, G. (1993) J. Biol. Chem. 268, 5798 –5801 22. Reiss, G., te Heesen, S., Gilmore, R., Zufferey, R., and Aebi, M. (1997) EMBO J. 16, 1164 –1172 23. Kim, H., Yan, Q., Heijne, G., Caputo, G., and Lennarz, W. J. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 7460 –7464 24. Chi, J. H., Roos, J., and Dean, N. (1996) J. Biol. Chem., 271, 3132–3140 25. Albright, C. F., Orlean, P., and Robbins, P. W. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 7366 –7369 26. te Heesen, S., Knauer, R., Lehle, L., and Aebi, M. (1993) EMBO J. 12, 279 –284 27. Uetz P., Giot L., Cagney G., Mansfield T. A., Judson R. S., Knight J. R.,

3 M. Chanvan, T. Suzuki, M. Rekowicz, and W. J. Lennarz, submitted for publication.


28. 29. 30. 31. 32. 33. 34. 35. 36. 37.

Lockshon D., Narayan V., Srinivasan M., Pochart P., Qureshi-Emili A., Li Y., Godwin B., Conover D., Kalbfleisch T., Vijayadamodar G., Yang M., Johnston M., Fields S., and Rothberg J. M. (2000) Nature 403, 623– 627 Stagljar, I., Korostensky, C., Johnsson, N., and te Heesen, S. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 5187–5192 Hordern, J. S., Leonard, J. D., and Scraba, D. G. (1979) Virology 97, 131–140 Joshi, S., and Burrows, R. (1990) J. Biol. Chem. 265, 14518 –14525 Baskin, L. S., and Yang, C. S. (1982) Biochim. Biophys. Acta. 684, 263–271 Schweizer, E., Angst, W., and Lutz, H. V. (1982) Biochemistry 21, 6807– 6818 Hamada, H., and Tsuro, T. (1987) Anal. Biochem. 160, 483– 488 Dorman, G., and Pretwich, G. D. (1994) Biochemistry 33, 5661–5673 Silberstein, S., Collins, P. G., Kelleher, D. J., and R. Gilmore (1995) J. Cell Biology 131, 371–383 Fu, J., Ren, M., and Kreibich, G. (1997) J. Biol. Chem. 272, 29687–29692 Kalies, K. U., and Hartmann, E. (1996) J. Biol. Chem. 271, 3925–3929

33087

38. Finke, K., Plath, K., Panzner, S., Prehn, S., Rapoport, T., Hartmann, E., and Sommer, T. (1996) EMBO J. 15, 1482–1494 39. Kassenbrock, C., Cao, W., and Douglas, M. (1993) EMBO J. 12, 3023–3024 40. Iwata, S., Ostermeier, C., Luwig, B., and Michel, H. (1995) Nature 376, 660 – 669 41. Silberstein, S., Collins, P. G., Kelleher, D. J., Rapiejko, P. J., and Gilmore, R. (1995) J. Cell Biol. 128, 525–536 42. Silberstein, S., Kelleher, D. J., and Gilmore, R. (1992) J. Biol. Chem. 267, 23658 –23663 43. Kim, H., Park, H., Montalvo, L., and Lennarz, W. J. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 1516 –1520 44. Romisch, K. B., Scheper, W., Collie, N., Thaminy, S., and Stagljar, I. (2002) Abstract for 42nd American Soceity for Cell Biology Annual Meeting (Dec. 14 –18, 2002, San Francisco), Supplement to Molecular Biology of the Cell 13, 233a, No. 1312