introduction/background Bio lab paper

Proc. Natl. Acad. Sci. USA Vol. 82, pp. 7500-7504, November 1985 Biochemistry

GTP-binding membrane protein of Eslcherichia coli with sequence homology to initiation factor 2 and elongation factors Tu and G

(LepA/protein secretion/photoaffinity cross-linking/pIN-Il vector)

PAUL E. MARCH AND MASAYORI INOUYE Department of Biochemistry, State University of New York at Stony Brook, Stony Brook, NY 11794

Communicated by Russell F. Doolittle, July 11, 1985

ABSTRACT The amino acid sequence of LepA protein, which has been shown to be cotranscribed with signal peptidase I in Escherichia coli, was compared with >2000 known protein sequences. it was revealed that, of the 598 amino acid residues contained in LepA, an antino-terminal domain of 112 residues is homologous to h domain of similar size found in initiation tbctor 2, elongation factor Tu, and elongation factor G (IF2, EF-Tu, and EF-G), factors required for translation in E. coli. In this domain, 46 and 34 residues align perfectly with the corresponding regions of EF-G and EF-Tu, respectively. If functionally conserved residues within this domain (19 for EF-G and 17 for EF-Tu) are included, the overall resemblance is 58% and 46%, respectively, for EF-G and EF-Tu. A similar domain exists internally in IF2, where there is 42% overall resemblance with the domain of LepA. Immediately adjacent to this region is a small sequehce of limited similarity that exists not only in EF-G, EF-Tu, and IF2 but also in the protooncogene c-Ha-ras-1 (from human bladder) and other GTP-binding proteins. Given these homologies, GTP-photoaffinity labeling and subcellular fractionation experiments were undertaken, and it was round that LepA is indeed a membrane-bound GTP-binding protein.

Within the past few years a great deal of information has accumulated about genes required for secretion in Esche- richia coli. One gene (lep) and its gene product (signal peptidase I) were extensively characterized (1-5) and shown to be responsible for endoproteolytic removal of signal peptides from precursor proteins during the process of secretion (3). The lep gene, and flanking sequences, were cloned into pBR322 and it was shown that a 350-base-pair Ava I-Bgl II DNA fragment was required for expression of the lep gene. This fragme tt possessed promoter activity but was located =1.5 kilobases upstream from the lep coding region (1, 5). The DNA sequence of the entire upstream region demonstrated that the lep gene is indeed present in an operon containing two genes, and lep is the promoter distal gene (6). The lep promoter was defined by huclease Si mapping of the 350-base-pair Ava I-Bgl II fragment, demon- strating that lep and the newly discovered upstream gene (lepA) are encoded by the same mRNA (6). The lepA gene encodes a protein of unknown function

(LepA) containing 598 amino acid residues (6). The predicted amino acid sequence was compared with >2000 known sequences with the use of an updated version of the NEWAT amino acid sequence collection (7). In this paper we report that the amino-terminal domain of 112 amino acid residues of LepA shares regions of homology to E. coli translation factors [initiation factor 2, elongation factor Tu, and elonga- tion factor G (IF2, EF-Tu, and EF-G)]. All of these proteins and several other GTP-binding proteins (8) share an adjacent

region of similarity, which, in the case of EF-Tu, is known to interact with guanine nucleotides (8-10). GTP cross-linking and subcellular fractionation experiments revealed that LepA is a membrane-bound GTP-binding protein.

MATERIALS AND METHODS Reagents and Materials. [35S]Methionine (1435 Ci/mmol; 1

Ci = 37 GBq) and [a-32P]GTP (420 Ci/mmol) were supplied by Amersham.

Strains and Plasmids. E. coli K-12 strain JA221 (lacY hsdR AtrpES leuB6 recAF' lacIq lac' pro') (11) was employed in cell localization experiments and SB221 (lpp lacY hsdR AtrpE5 leuB6 recAIF' lacIq lac pro') (11) harboring pPEM109 was employed in GTP-photoaffinity cross-linking experiments. The construction and use of pPEM109, a plasmid that overexpresses LepA, has been. described (6).

GTP-Photoafflnity Cross-Linking. To prepare lysates for photoaffinity cross-linking E. coli SB221 was grown to a cell density of 4 x 108 cells per ml, and then LepA production from pPEM109 was induced with isopropyl /3-D-thiogalacto- side (IPTG) for 2 hr as described (6). A cell-free extract was prepared as described (12), and a 10-,ul aliquot Was incubated with 40 A.l of reaction mixture such that the final concentra- tions were as follows: 1.2 A.M [a-32PjGTP (420 Ci/mmol), 1 MM ATP, 50mM Tris hydrochloride (pH 8.0), 100mM NaCl, 5 mM Mg9l2, and 1% Triton X-100. This mixture Was incubated for 1 hr at 40C in the dark. The reaction mixture was exposed to UV light (maximal emission, 254 nm) in shallow culture trays on ice at an intensity of 1.8 mW/cm2 for various lengths of time (0-60 min). In agreement with a previous report, we observed maximal incorporation of label after irradiation for 30 min (12). This timhe was employed in all subsequent experiments. Antiserum preparation and immunoprecipitation was car-

ried out as described for antilipoprotein serum (13) except that the source of purified LepA was from preparative gel electrophoresis.

SubcellulAr Fractionation. E. coli strain JA221 was grown in M9 minimal medium (14) to a cell density of 4 x 101 cells per ml. Then 100 A.Ci of [35S]methionine was added to 2 ml of cells and incubation was continued for 2 min. At the end of this incubation excess nonradioactive methionine was added and labeled cells were chilled to 0WC. The cells were collected by centrifugation, suspended in 1 ml of 10 mM sodium phosphate (pH 7.1), and subjected to sonication. The total envelope and soluble fractions were collected after ultracentrifugation (6). Total fractions and immunopre- cipitates offractions were analyzed by NaDodSO4/polyacryl- amide gel electrophoresis. NaDodSO4/polyacrylamide gel electrophoresis was described by Anderson et al. (15).

Abbreviations: IPTG, isopropyl ,f-D-thiogalactoside; EF, elongation factor; IF, initiation factor; SRP, signal recognition particle.

7500

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. NatL. Acad Sci USA 82 (1985) 7501

RESULTS AND DISCUSSION The entire DNA sequence of the lep operon has been determined, and it was revealed that this operon encodes two genes, the upstream lepA gene (6) and the downstream lep gene for signal peptidase 1 (2). The function of the lepA gene product (LepA) has not been determined. However, a pos- sible role ofLepA in the process ofprotein secretion has been suggested because of its coexpression with signal peptidase 1 (6). The DNA sequence ofthe lepA gene predicts that LepA consists of 598 amino acid residues. Indeed, LepA has been identified as a cytoplasmic membrane protein of an apparent molecular weight of 76,000 (6). As one of the attempts to characterize the function of

LepA, the amino acid sequence of LepA was searched against an updated version of the NEWAT amino acid sequence collection (7). To our surprise, homology was found between LepA and E. coli translation factors such as IF2, EF-Tu, and EF-G, which are known to be GTP-binding proteins and to interact with ribosomes (for a review, see ref. 16). As shown in Fig. 1, the amino-terminal domains of EF-Tu

and EF-G contain large regions that are homologous to the amino-terminal domain of LepA. These observations are most striking for EF-G, where 46 of 112 amino-terminal residues (41%) of LepA align perfectly with EF-G (boxed residues in Fig. 1). Another 19 residues are functionally conserved between the two proteins (circled residues in Fig.

1), making the overall similarity 58% in this region. EF-Tu and LepA have 34 perfect matches (30%) and another 17 functionally conserved matches (overall, 46% resemblance). In the case of IF2, the region of homology starts from the proline residue at position 392 (17). Within this internal domain 30 perfect matches (27%) and an additional 17 functionally conserved matches (overall, 42% similarity) were found within an 87 amino acid domain. The amino acid homology between EF-Tu, IF2, and EF-G

has been noted previously and it was pointed out that both EF-Tu and IF2 interact with tRNA. In addition, all three proteins bind GTP and ribosomes (17, 19). Though little is known about the residues involved in ribosomal interactions, it was proposed that the first region of homology shown in Fig. 1 may be important for ribosome binding in IF2 (residues 392-416) (17). Somewhat more is known about amino acid residues involved in GTP binding. It has been reported that x-ray crystallographic analysis shows that the y-phosphate of GTP is positioned closely to residues 80-85 in EF-Tu (20). These residues are located in one of the most highly con- served regions shown in Fig. 1. The sequence Asp-Xaa-Pro- Gly-His (the aspartic acid residue is amino acid number 80 in EF-Tu, 77 in LepA, and 88 in EF-G) is shared by all proteins compared and may be functionally important for these proteins. Other GTP-binding proteins do not necessarily share this pentapeptide, however. The amino acid sequences of several GTP-binding proteins have been compared from both prokaryotes and eukaryotes (8), including EF-G, EF-

Arg Asn Phe Ser Ile Ile Ala His Ile Asp His Gly Lys

Arg Asn Gly Ile Ser Aia His Ile Asp iIGly Lys

Val Asn Gly Thr Ile His Voi Asp His Gly Lys

Pro Val ( Trie aHis Vol Asp His Gly Lys

Ser hr eu|Ser Asp Arg lie lie Gln lie Cys Gly 29

r Thr hr l le ;) Phe () Thr GIy 34

Thr Leu TrAl Ala lIle ( )Leu (9 8 8 Leu Tyr lie Arg Ser Thr Lys ) 416

LepA 38 Ala Gin

EFG 46 Al[ Ala

EFTu 50

I F2 423

Glu Arg Gly lie Thr Ile Lys Ala Gln Ser Val7Tr LeujjTyr Lys

Glu Arg Gly lie Thr lie Thr Ser Ala Ala Tr Phe E Ser Al Arg Gly Ile Thr Ile Asn Thr Ser His u

64

72

70

426

LepA 69 FT1 Thr Tyr Gin Leu Asn Phe lie Asp Thr Pro Gly His Vol Asp Phe Ser Tyr Giu Voi Ser Sr LeuEIa

EFG 80 [lj Pro His Arg eAsn Ile Asp Thr Pro Gly His Val Asp Phe Gu VaiGluArg)Arg63

EFTu 77 Alo His VO, AspEIPro Gly His Asp OTyr Vol Lys Asn 3& lie Thr Gly lin IF2 436 jijAsn Gly Met Thr Phe Asp Thr Pro Gly Hi A Aio Phe TSer Met Arg Ala Gly(3 i;l

94

105

97

461

LepA 95 Cys Gil

EFG 106 Lcu

EFTu 98

IF2 462

112

123

115

479

FIG. 1. The amino acid sequence homologies of IF2, EF-G, and EF-Tu are compared to LepA. Three regions were observed that contained large blocks of homology (LepA residues 5-29, 38-64, and 69-112). Amino acid residues that are identical to the corresponding residue of LepA are boxed. Amino acid residues in which the side chain is functionally conserved compared to the LepA residue are circled (sequence information is from refs. 17, 18, and 19 for IF2, EF-Tu, and EF-G, respectively).

LepA 5

EFG 10

EFTu 1 2

IF2 392

Biochemistry: March and Inouye

7502 Biochemistry: March and Inouye

Tu, and p21, the product from the protooncogene c-Ha-ras-1, which is a known membrane-bound GTP-binding protein (21). The only resemblance found was located near the amino terminus, containing the sequence Val-Xaa-Xaa-Xaa-Yaa- Yaa-Yaa-Yaa-Asn (or aspartic acid) (where Xaa can be any residue and Yaa represents a hydrophobic residue). In every protein examined, except eukaryotic EF eEF-Tu, the valine was followed by a proline. Also in every protein except a- and P-tubulin the sequence ended with an asparagine, not an aspartic acid residue. Furthermore, the asparagine, when present, was followed by the sequence Lys-Yaa-Asp (where Yaa is a hydrophobic residue). These features are summa- rized in Fig. 2 and the comparison is extended to the corresponding regions of LepA and IF2. In LepA, EF-G, EF-Tu, and IF2, this region is found immediately after the amino-terminal domain shown int Fig. 1. The importance of this region is highlighted by the fact that

x-ray crystallographic data show that Asn-135 of EF-Tu (see boxed residue in Fig. 2) forms a hydrogen bond with the guanosine group of GDP (8). In addition, modification of Cys-137 of EF-Tu eliminates nucleotide binding (9, 10). It has been pointed out by Sacerdot et al. (17) that the predicted secondary structure in the region of IF2 and EF-Tu shown in Fig. 2 is very similar. Thus, we also attempted to examine the possible secondary structure of LepA using the rules devel- oped by Chou and Fasman (22). We found that all four proteins-LepA, IF2, EF-G, and EF-Tu-possess a high potential to form a p-sheet structure at the region between the conserved Val-Pro and Asn-Lys shown in Fig. 2 (see boxed residues). The amino acid sequence immediately upstream of this region, ending with the conserved Val-Pro, possesses a high potential to form an a-helix.

Aside from the homology presented in Figs. 1 and 2, tio other homology was found between LepA and other known proteins. Also the homology between the translation factors EF-G, EF-Tu, and IF2 is limited to the regions shown in Figs. 1 and 2 (17, 19). The homologies discussed above coupled with the obser-

vation that LepA contains determinants known to be impor- taht in GTP binding suggested that LepA may bind GTP. To investigate this possibility we attempted to photochemically cross-link GTP to an amplified protein in crude cell extracts, a procedure employed to detect GTP binding to viral p21 after expression in E. coli (12). Cell lysates were prepared from cells harboring the plasmid pPEM109 that were grown in the absence and the presence of lac inducer IPTG (see Materials and Methods). Since lepA expression is under the control of the lpp promoter as well as the lac promoter and operator in pPEM109, LepA is produced only in the presence of IPTG (6). To these lysates, [a-32P]GIP was added and the lysate was UV-irradiated to cross-link [a-32P]GTP to GTP-binding proteins. By this method, EF-Tu, one of the most abundant proteins in E. coli, was labeled with 32P and easily detected as a radioactive band upon NaDodSO4/polyacrylamide gel electrophoresis of the total lysate followed by autoradi- ography (data not shown). On the other hand, to detect

LepA, immunoprecipitation with anti-LepA serum was per- formed and analyzed by NaDodSO4/polyacrylamide gel electrophoresis. As shown ih Fig. 3, a 32P-labeled band that was immunoprecipitated with anti-LepA serum was detected almost at the same position as [35S]methionine-labeled LepA marker (lane 1) in the presence (lane 3) but not in the absence (lane 2) of IPTG. The appearance of multiple molecular weight species of LepA upon preparation of cell lysates (see Fig; 3, lanes 1 and 3) has been discussed (6). When cells are treated with a formaldehyde solution prior to lysate prepa- ration, then only one band of molecular weight 76,000 is observed (see Fig. 4).

It should be noted that in the absence ofUV irradiation, no incorporation of GTP could be observed (data not shown). Furthermore, no incorporation could be detected when lysates were immunoprecipitated with anti-OmpA serum (a major outer membrane protein of E. coli) or with anti-,8- galactosidase serum (another protein whose expression is induced by IPTG) (data not shown), indicating that there was no nonspecific binding of GTP to other proteins. These data clearly demonstrate that LepA is a GTP-binding protein.

Previous studies showed that a significant fraction of overexpressed LepA was associated with the cytoplasmic membrane (6). Since all membrane-bound GTP-binding pro- teins discovered to date are eukaryotic (23), the cellular location of LepA in cells producing normal amounts of LepA was determined. It can be seen in Fig. 4B (compare lanes 2 and 3) that LepA is found exclusively in the membrane fraction. Using the sarkosyl extraction method we found that membrane-bound LepA was localized exclusively in the cytoplasmic membrane (6). The E. coli translation factors EF-Tu and EF-G are highly

expressed proteins accounting for 105 and 104 molecules per genome, respectively (24). In addition, the codon usage of IF2 is consistent with its being a highly expressed protein (17). Interestingly, the codon usage of LepA deviates from that expected for highly expressed proteins (6). This is consistent with the fact that it is present in very low amounts, since LepA expressed from the chromosomal gene can only be detected by immunoprecipitation (see Fig. 4). It is of interest to compare the codon usage of the respective genes of these proteins within the homologous regions presented in Figs. 1 and 2. Within the sequences shown in Figs. 1 and 2, there is a total of 125 aligned amino acid residues in EF-G, EF-Tu, and IF2, whose functional groups are identical to the corresponding residues of LepA (boxed residues in Figs. 1 and 2). EF-G contains 51 such residues, EF-Tu has 39, and IF2 possesses 35. Upon inspection of the codons of these 125 amino acid residues, a total of 49 possesses silent third-base changes and the remaining 76 codons are identical to those employed in the corresponding position by LepA. In 37 of the 49 silent changes, the codon utilized by LepA is a codon that is less preferred by highly expressed proteins (24). Thus, there is divergence at the nucleotide level that is consistent with the observation that LepA is not a highly expressed protein.

Glu Met Asp Leu Glu Val|Val Pro] - - - Val LeuI Asn Gin Ala Asn Lys Tyr LysIVal ProjArg - lie Ala Phe Val IAsn

Leu Gly Arg Gin Val Gly IVal Prol Tyr Ile Ile Val Phe LeuIAsn

Lys

Lys

Lys

I IelAspi Leu Pro

Met IAspi Arg Met

Cys IAsp

His Ala Lys Ala Ala GIniVal Prol Val Val Val Ala Val IAsn LysIIleIAsp

Val Lys Asp Ser Asp AspIVol ProjMet - Val Leu Vol GlyIAsn !is CYsL

Met Val

Lys Pro

Leu Ala

FIG. 2. Comparison of the amino acid sequence homologies of EF-G, EF-Tu, IF2, and p21 to LepA. The region shown contains common features found in other GTP-binding proteins. The conserved amino acid residues discussed in the text are boxed.

LepA 121

EFG 128

EFTu 121

1F2 485

p21 103

136

146

140

503

121

Proc. NatL Acad Sci. USA 82 (1985)

I-

Proc. Natl. Acad ScL USA 82 (1985) 7503

1 2 3.. ...... ] d

....

... ,

.:R ;si

.. ..t

.'s,

:.

LepAIm

EF-Tu I

A 1 2

i.:

92- , *

66-

>.lir. #, 41 -

31- mpov

B 1 2 3

- - - - 92-

- - --- 66 -

FIG. 3. Photoaffinity cross-linking of [a-32P]GTP to lysates prepared from cells overproducing LepA. Lane 1, [35S]methionine- labeled LepA marker. Lane 2, immunoprecipitation with LepA antiserum of lysate irradiated for 30 min in the presence of [a- 32P]GTP. This lysate was prepared from cells that were not incubated with IPTG and thus do not overproduce LepA. Lane 3, as in lane 2, but the lysate was prepared from cells overproducing LepA in the presence of IPTG. Electrophoresis was performed on a 10%o NaDodSO4/polyacrylamide gel. After electrophoresis the gel was dried and exposed to x-ray film (Kodak XAR-5 film) for 48 hr. The autoradiogram is shown.

The data presented in this report demonstrate that LepA is a GTP-binding protein localized in the cytoplasmic mem- brane of E. coli. In eukaryotes, membrane-bound GTP- binding proteins are thought to play a variety of roles usually centered upon the transduction ofintracellular signals (23). In light of the fact that this protein is cotranscribed with signal peptidase I, it is intriguing to speculate that it may play a role in protein secretion. The involvement of a membrane-bound GTP-binding protein in the protein secretion process would be surprising, and it may have important implications in the eukaryotic process as well. In protein secretion in eukaryotes it has been demonstrated that elongation of the nascent polypeptide chain is arrested by the signal recognition par- ticle (SRP) and elongation is continued only when the SRP-receptor (or docking protein) displaces the SRP at the secretion site on the membrane (25-27). It is possible that SRP mediates translation arrest by blocking elongation factor binding sites on the ribosome, and the displacement of SRP by the SRP-receptor may require the hydrolysis of a high- energy phosphate bond supplied by GTP. In addition, the endoplasmic reticulum of eukaryotic cells possesses ribo- phorins, which are responsible for binding ribosomes to the membrane (28). It is possible that GTP may be involved in mediating the binding of ribosomes to ribophorins. This speculation, if correct, leads to the prediction that the SRP-receptor or ribophorins may be GTP-binding proteins. Further work is necessary to investigate the possibility that LepA could be involved in bacterial protein secretion and, more precisely, that a specific LepA-ribosome interaction may occur as a step in the secretion mechanism.

FIG. 4. Subcellular fractionation and localization of LepA pro- duced from the chromosome. (A) Fractions employed in immuno- precipitation experiments. Lane 1, total membrane fraction. Lane 2, total soluble fraction. (B) Immunoprecipitations and LepA marker. Lane 1, LepA marker (prepared by immunoprecipitation from cells overexpressing LepA). Lane 2, immunoprecipitation from the total soluble fraction. Lane 3, immunoprecipitation from the total mem- brane fraction. The position of migration of molecular weight markers and their molecular weights (x 10-3) are indicated. The position of migration of LepA is indicated by the arrowhead. Cells were treated with 0.8% formaldehyde prior to fractionation and LepA marker preparation. This was done to prevent the multiple banding pattern observed in Fig. 3 (lanes 1 and 3). Gel electropho- resis and autoradiography were as in Fig. 3, except the exposure time for A was 16 hr and for B was 8 days.

We are grateful to Dr. R. Doolittle for his analysis of the LepA structure and Dr. C. A. Lunn for critical reading of the manuscript. We thank the National Institute of General Medical Sciences (Grant GM19043) and the American Cancer Society (Grant NP3871) for support of this research. P.E.M. is an American Cancer Society Postdoctoral Fellow.

1. Date, T. & Wickner, W. (1981) Proc. Natl. Acad. Sci. USA 78, 6106-6110.

2. Wolfe, P. B., Wickner, W. & Goodman, J. M. (1983) J. Biol. Chem. 258, 12073-12080.

3. Zwizinski, C. & Wickner, W. (1980) J. Biol. Chem. 255, 7973-7977.

4. Silver, P. & Wickner, W. (1983) J. Bacteriol. 154, 569-572. 5. Date, T. (1983) J. Bacteriol. 154, 76-83. 6. March, P. E. & Inouye, M. (1985) J. Biol. Chem. 260,

7206-7213. 7. Doolittle, R. F. (1981) Science 214, 149-159. 8. Leberman, R. & Egner, U. (1984) EMBO J. 3, 339-341. 9. Miller, D. L., Hachman, J. & Weissbach, H. (1971) Arch.

Biochem. Biophys. 144, 115-121. 10. Wade, M., Larsen, R. A. & Miller, D. L. (1975) FEBS Lett.

53, 37-39. 11. Nakamura, K., Masui, Y. & Inouye, M. (1982) J. Mol. Appl.

Genet. 1, 289-299. 12. Stein, R. B., Robinson, P. S. & Scolnick, E. M. (1984) J.

Virol. 50, 343-351. 13. Inouye, S., Takeishi, K., Lee, N., DeMartini, M., Hirashima,

A. & Inouye, M. (1976) J. Bacteriol. 127, 555-563. 14. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold

Spring Harbor Laboratory, Cold Spring Harbor, NY). 15. Anderson, C. W., Baum, P. R. & Gesteland, R. F. (1973) J.

Virol. 12, 241-252. 16. Maitra, U., Stringer, E. A. & Chaudhuri, A. (1982) Annu. Rev.

Biochem. 51, 869-900.

Biochemistry: March and Inouye

7504 Biochemistry: March and Inouye

17. Sacerdot, C., Dessen, P., Hershey, J. W. B., Plumbridge, J. A. & Grunberg-Manago, M. (1984) Proc. Nail. Acad. Sci. USA 81, 7787-7791.

18. Yokota, T., Sugisaki, H., Takanami, M. & Kaziro, Y. (1980) Gene 12, 25-31.

19. Zengel, J. M., Archer, R. H. & Lindahl, L. (1984) Nucleic Acids Res. 12, 2181-2192.

20. Duisterwinkel, F. J., Kraal, B., DeGraaf, J. M., Talens, A., Bosch, L., Swart, G. W. M., Parmeggiani, A., LaCour, T. F. M., Nyborg, J. & Clark, B. F. C. (1984) EMBO J. 3, 113-120.

21. Papageorge, A., Lowy, D. & Scolnick, E. M. (1982) J. Virol. 44, 509-519.

Proc. NaML Acad Sci. USA 82 (1985)

22. Chou, P. Y. & Fasman, G. D. (1978) Annu. Rev. Biochem. 47, 251-276.

23. Gilman, A. G. (1984) Cell 36, 577-579. 24. Gouy, M. & Goutier, C. (1982) Nucleic Acids Res. 10,

7055-7074. 25. Meyer, D. I., Krause, E. & Dobberstein, B. (1982) Nature

(London) 297, 647-650. 26. Gilmore, R., Blobel, G. & Walter, P. (1982) J. Cell Biol. 96,

463-469. 27. Walter, P. & Blobel, G. (1981) J. Cell Biol. 91, 557-561. 28. Kreibich, G., Czako-Graham, M., Grebenau, R., Mok, W.,

Rodrigues-Boulan, E. & Sabatini, D. D. (1978) J. Supramol. Struct. 8, 279-302.