FEMS Microbiology Letters 100 (1992) 233-242 © 1992 Federation of European Microbiological Societies 0378-1097/92/$05.00 Published by Elsevier
233
FEMSLE 80061
Incompatibility of outer membrane proteins OmpA and OmpF of Escherichia coli with secretion in Bacillus subtilis" Fusions with secretable peptides Marjo S i m one n 1, Eveliina Tarkka, Ritvaleena Puohiniemi and Matti Sarvas National Public Health Institute, Helsinki, Finland
Received 18 June 1992 Accepted 29 June 1992
Key words: Bacillus subtilis; Escherichia coli; Secretion; Outer membrane protein; OmpA; OmpF
1. SUMMARY The secretion of the outer membrane proteins OmpA and OmpF of Escherichia coli has previously been found to be blocked at an early intracellular step, when these proteins were fused to a bacillar signal sequence and expressed in Bacillus subtilis. We have now fused these proteins to long secretable polypeptides, the amino-terminal portions of a-amylase or /3-1actamase. In spite of this, no secretion of the fusion proteins was detected in B. subtilis. With the exception of a small fraction of the /3-1actamase fusion, the proteins were cell-bound with uncleaved signal sequences. Protease accessibility indicated that the fusion proteins were not even partially exposed on the outer surface of the cytoplasmic membrane. Thus there was no change of the location compared to the OmpA or OmpF fused to the signal sequence
Correspondence to: M. Sarvas, Department of Molecular Bac-
teriology, National Public Health Insitute, SF-00300 Helsinki, Finland. 1 Present address: Institute of Biotechnology, University of Helsinki, SF-00380 Helsinki, Finland.
only. We conclude that, like OmpA and OmpF, the fusion proteins fold into an export-incompatible conformation in B. subtilis before the start of translocation, which we postulate to be a late post-translational event.
2. INTRODUCTION Many proteins that are normally secreted by their natural hosts are secreted poorly or not at all, when synthesized in Bacillus subtilis [1-3]. This indicates that some of the components of the bacillar secretion machinery fail to recognize or interact productively with a wide range of heterologous secretory proteins. This is a serious limitation for the utilization of the secretion capability of bacilli in biotechnical applications. However, the nature of the incompatible components and the steps of secretion affected are largely unknown. We have approached this problem using outer membrane proteins of Gram-negative bacteria as models. In their natural hosts these proteins are secreted through the cytoplasmic membrane and
234 assembled into the outer membrane [4,5]. We introduced the genes of the two major outer membrane proteins OmpA and OmpF of Escherichia coli into B. subtilis [6,7] downstream of the signal sequence of a bacillar exoenzyme, the a-amylase of B. amyloliquefaciens [8]. In spite of this, the fusion proteins were not secreted, but a block of secretion occurred at a very early step of the process [7]. In fact, there was no indication of interaction between the fusion proteins with the secretion machinery. We were therefore interested in learning how OmpA and OmpF would behave when fused to a secretable polypeptide; these studies are reported in this paper.
3. MATERIALS AND METHODS
3.1. Growth conditions For protein expression, B. subtilis strains were grown in liquid culture in two-fold concentrated L-broth containing: NaC1, 10 g 1-1; kanamycin, 10 mg l-1; and potato extract, 30 ml 1-1; overnight at 37°C [6]. E. coli strains were grown on L-agar plates or L-broth [6], containing either ampicillin (100 /xg ml-1), tetracycline (12.5 /xg ml-*) or kanamycin (10 ~g m1-1) at 37°C. 3.2. Construction of plasmids encoding fusion proteins 3.2.1. a-amylase-OmpA-fusions. The ompA fragments were recovered from plasmids pKTH158 [6] and pKTH165 by digestion with HindlII. The ompA fragment from pKTH158 encodes amino acid residues 8-228 of OmpA. The fragment from pKTH165 is similar to that from pKTH158 except that the 3' end of the ompA fragment has been further shortened with exonucleases and encodes residues 8-138. The HindIII fragments were converted to blunt-ended (filled in), and inserted into the plasmid pKTH10 [9], which was cleaved at the KpnI site of the aamylase gene (at codon 289) and blunt-ended with Sl-nuclease. The plasmids thus constructed were designated pKTH198 and pKTH201, respectively; the structures of the fusion proteins encoded are designated Amy-OmpA8-228-Amy and Amy-OmpA8-138 and are shown in Fig. 1.
To express the OmpA fragment 8-228 in E. coli, the plasmids pKTH158 (encoding the residues 8-228 of OmpA fused to the signal sequence of the a-amylase of B. amyloliquefaciens) and pBR322 [10] were ligated to each other at their BamHI sites to make the plasmid pKTH202. 3.2.2. ~-Lactamase-OmpF fusions. The Pstl fragment of the plasmid pJP33 [11], encoding amino acid residues 1-340 of OmpF, was inserted into the Pst I site of the TEM /3-1actamase (Bla) gene of E. coli (at codon 158) in the plasmid pKTH78 [8] to construct pKTH180 (Fig. 1). To synthesize the same fusion protein in E. coli, the PstI fragment of pJP33 was inserted in the PstI site of /3-1actamase of pBR322, resulting in the plasmid pKTH204. The fusion genes in the above plasmids were downstream of the promoter of a-amylase of B. amyloliquefaciens [8], except in pKTH204.
3.3. Protease treatment of protoplasts Protoplasts were prepared from overnight grown cells with lysozyme treatment [7]. Protoplasts were treated with trypsin (2 mg m1-1) before or after disruption with sonication or Triton X-100. As a marker for the exposure of intracytoplasmic proteins to the protease, we analysed the glucose-6-P dehydrogenase activity [7]. 3.4. Other methods 3.4.1. SDS-PAGE. Proteins were electrophoresed in SDS-polyacrylamide gels [12], immunoblotted with specific rabbit antisera [12,13] and visualized by autoradiography using [~25I]labelled protein A or with peroxidase-conjugated anti-rabbit IgG (Bio-Rad, Richmond, CA). 3.4.2. In vitro protein synthesis. An in vitro transcription-translation assay was carried out as described [14]. The proteins were labelled with [35S]methionine (Amersham) and immunoprecipitared [15]. 4. RESULTS
4.1. Fusion constructions We chose two secreted proteins to be fused to the amino-terminus of OmpA or OmpF. One was
235
a bacillar exoprotein, the a-amylase (Amy) of B. amyloliquefaciens, the other a periplasmic protein of E. coli, the TEM /3-1actamase, encoded by amyE and bla, respectively. The fusion proteins were expressed from the multicopy secretion vector based on the promoter and signal sequence of the a-amylase of B. amyloliquefaciens [8]. Both the a-amylase and the /3-1actamase have been shown to be synthesized and secreted at a similar high rate when expressed in B. subtilis from a multicopy plasmid from this promoter and signal sequence [8,9]. Segments of ompA gene were inserted into the amyE gene of the plasmid pKTH10, resulting in the plasmids pKTH198 and pKTH201 (Fig. 1). pKTH198 encodes a tripartite fusion protein designated Amy-OmpA8-228-Amy and composed of the signal sequence and the 289 NH2-terminal residues of the mature a-amylase, followed by residues 8-228 of mature OmpA and finally the C-terminal residues 290-463 of a-amylase (Fig. 1). The fusion protein encoded by pKTH201 and designated Amy-OmpA8-138 contains the same 5' fragment of amyE, followed by ompA codons 8-138 and a translational stop codon 16 residues downstream created by an out-of-frame fusion to the Y-terminal segment of amyE. The larger OmpA8-228 fragment corresponds approximately to the amino-terminal domain of OmpA that is embedded in the outer membrane of E. coli [16]; it lacks the C-terminal domain which is located in the periplasm in E. coli. This amino-terminal fragment is exported and assembled into the outer -31 2
158
membrane in E. coli [17], but fails to be exported and accumulates intracellularly in B. subtilis like the untruncated OmpA [6,7]. OmpF protein was fused to /3-1actamase. The structural gene of the mature part of OmpF was inserted into the bla gene in the plasmid pKTH78 [8] to obtain pKTH180. This resulted in an inframe fusion consisting of the signal sequence of a-amylase, 158 codons corresponding to the amino-terminal part of mature /3-1actamase and the entire ompF gene encoding the amino acid residues 1-340 of the mature protein (Fig. 1); the fusion protein was designated Bla-OmpF.
4.2. Expression and localization of the fusion proteins in B. subtilis The constructed plasmids were used to transform B. subtilis IH6140, which is a mutant with a decreased level of exoproteases [8]. Cultures of these transformants were studied by immunoblotting using anti-OmpA, anti-OmpF, anti-a-amylase and anti-/3-1actamase rabbit sera. No proteins with antigenic specificity of either part of the appropriate fusion protein were detected in the culture supernatants of any of these strains. TCA precipitates of the culture medium of exponential phase cultures were analysed, when there was minimal amount of cellular lysis and very low levels of exoproteases. The secretion of unfused, nontruncated a-amylase and /3-1actamase expressed from plasmids based on the same secretion vector were readily detectable under these conditions [8,9].
1
340
pKTHI80 SS
Bla
OmpF
-31
298
pKTHI98
8
228
290
463
iii!iiiiiiili~iliiii~iiii~iiiil}iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii iii~ili~i~iisi~iiiiii ili~i~i~i~i~!~i~i~!iii~ii~i~iiii!i~ SS
Amy
-31
OmpA
298
8
Amy
138
pK'rH2 01
|| SS
Amy
OmpA
Fig. 1. The structures of the fusion proteins encoded by the inserts in the plasmids pKTH180, pKTH198 and pKTH201. The negative numbers refer to the amino acid residues of the signal sequence and the positive ones to those of appropriate mature proteins. SS, signal sequence of u-amylase of B. amyloliquefaciens. For the other designations see SECTION 3.2.
236
In contrast, the fusion proteins were detected in the particulate fractions of the bacteria, prepared as described [6]. Figure 2 shows that a strongly reactive protein of the expected size of the fusion protein was found in the particulate fraction of strains expressing the a-amylaseOmpA fusions (lanes C, D and G, proteins of about 80 kDa (Amy-OmpA8-228-Amy) and 50 kDa (Amy-OmpA8-138), respectively). In bacteria expressing the Bla-OmpF fusion one strongly and another more weakly labelled protein (about 56 kDa and 54 kDa) were detected. Such proteins were not present in the control strains devoid of the plasmid encoding the fusion protein (lane I). Both anti-a-amylase and anti-OmpA reacted with the fusion in IH6140(pKTH198), and both anti]3-1actamase and anti-OmpF with that of IH6140(pKTH180). However, in the case of pKTH201 the protein was not detected by antiA
15
C
D
OmpA serum, presumably due to the small size of the OmpA fragment (Fig. 2, lane A). Small amounts of the fusion proteins were also detected in the cytoplasmic fractions. The presence of the fusion proteins in the particulate fraction of the bacteria is a finding analogous to the accumulation of precursors of OmpA and OmpF with uncleaved signal peptides in B. subtilis [6,7]. Indeed, the apparent molecular masses of the two Amy-OmpA fusions and the major Bla-OmpF fusion protein seen in the immunoblot (Fig. 2) are close to the deduced molecular mass of their primary translates without cleavage of the signal sequence. Thus the deduced molecular mass of the processed AmyOmpA8-138 protein is about 48 kDa, and that of the primary translate is 51 kDa, the protein observed in Fig. 2 (lane G) is more than 50 kDa. To further confirm the nature of fusion proteins in F
E
94
G
H
94
94
67
67
i!!!i~!/iiill ~~ i~iii!~ ~ i~i~
I
~!~i !ili~
- 67
43
67 43
43
43
30
iS!L!iI! 30
anti-OmpA serum
anti-u-amylase
serum
anti-OmpF serum
Fig. 2. Immunoblotting of the particulate fractions of B. subtilis IH6140 carrying the plasmids: A, pKTH201(Amy-OmpA8-138); B, the parent plasmid of pKTH198; C, pKTH198(Amy-OmpA8-228-Amy); D, pKTH198(Amy-OmpA8-228-Amy); E, the parent plasmid of pKTH198; F, 500 ng of B. amyloliquefaciens o~-amylase (Sigma A6380); G, pKTH201(Amy-OmpA8-138); H, pKTH180(Bla-OmpF); I, the parent plasmid of pKTH180.
237
the particulate fraction, we synthesized the proteins from the pKTH180 DNA template in a cell-free system. The synthesis was carried out in the absence of membranes and thus no cleavage of the signal peptide was expected to take place. Figure 3 shows the products of in vitro synthesis of the Bla-OmpF fusion protein; the protein synthesized had the same apparent molecular mass as the 56 kDa fusion protein found in the particulate fraction of bacteria carrying pKTH180 (Fig. 3, lane A). On the other hand, no 54-kDa protein, also seen in the particulate fraction, was found among the proteins made in vitro. Although this protein was not detected in culture supernatant, its size suggests that it could have been produced by cleavage of the signal peptide. In some membrane preparations of the strain carrying pKTH198 (like the one in Fig. 2, lane D) there was also a second, minor protein with the antigenic specificity of both OmpA and aamylase. It was about 15 kDa smaller than the primary translate and thus clearly not the product of the cleavage of the signal sequence. Although no fusion proteins were detected in the culture supernatants of exponentially growing cells, both Bla-OmpF and Amy-OmpA8-228-Amy proteins were found in small amounts in supernatants of stationary phase cultures. However, the amount was small, less than a few percent of that in the bacteria. Furthermore, in both cases the size of the proteins in the supernatant and cells was similar, suggesting release of the proteins from autolysing cells.
ton X-100) or sonication, and then subjected to immunoblotting. Figure 4 shows that in all cases the size of the major membrane associated fusion protein was not affected by trypsin treatment of intact protoplasts (compare lanes A and B in Fig. 4); in most cases the amount was slightly decreased presumably due to some breakage of protoplasts which was measured by monitoring the enzyme activity of a cytoplasmic marker protein as in [7] (data not shown). In contrast, the A
E5
C
D
u
4.3. Protease accessibility of the fusion proteins in protoplasts One indication of the cytoplasmic localization of the fusion proteins synthesized in B. subtilis was their inaccessibility to exogenous proteases in intact protoplasts and complete breakdown by similar treatment after breakage of the protoplasts [7]. To determine whether any translocation of the fusion proteins across the cytoplasmic membrane took place, we carried out a similar trypsin treatment of protoplasts of strains carrying pKTH198, pKTH201, or pKTH180. In brief, protoplasts were treated with trypsin before and after breakage either by lysis with detergent (Tri-
Fig. 3. Comparison of in vitro and in vivo synthesized BlaOmpF fusion proteins. Lane A, particulate fraction of B. subtilis IH6140(pKTH180) expressing the Bla-OmpF fusion immunoblotted with anti-OmpF and labelled with [125I]protein A. Lanes B and C, proteins synthesized in vitro from pKTH180 and immunoprecipitated with anti-OmpF or antiBla serum, respectively. D, [35S]methionine-labelled proteins synthesized in vitro from pKTH180. The asterisk indicates the position of the 56-kDa Bla-OmpF fusion protein. (In the particular experiment shown in the figure there was not a full set of molecular mass maskers, but the comparison of lane A with other lanes is unambiguous even without markers.)
238 fusion p r o t e i n s were a b s e n t if the protoplasts were b r o k e n before trypsin t r e a t m e n t (Fig. 4 lanes C a n d D). Significantly, n o p e p t i d e s smaller t h a n
1A°AB CD z~
....
~
the p r i m a r y translate were f o u n d by i m m u n o b l o t ting with any of the sera. T h e effect of the protease t r e a t m e n t on the
2AoA -
B co
94
- 43
anti -a-amylase serum
3
A
e
anti-OmpF
c
serum
anti
-a-amylase serum
D
4
A
B
C
D
anti-B- lactamase ~erum
Fig. 4. Protease accessibility of the fusion proteins in protoplasts. (1) IH6140(pKTH]98) (Amy-OmpA8-228-Amy). (2) IH6140(pKTH20D (Amy-OmpA8-138).(3, 4) IH6140 (pKTH180) (Bla-OmpF). Immunoblotting of protoplasts. Ao, a control strain IH6140 or IH6140(pKTH198) in panels 1 and 2, respectively; A, no treatment of protoplasts; B, trypsin treatment of intact protoplasts; C, trypsin treatment after sonication; D, trypsin treatment after solubilization with Triton X-100. The arrowheads show the positions of Amy-OmpA fusion proteins, lmmunoblots were detected with peroxidase-conjugated anti-rabbit IgG (BioRad, Richmond, CA) (panels 1 and 2) or with [12SI]labelledprotein A (panels 3 and 4).
239 minor 54-kDa species of Bla-OmpF fusion protein differed significantly from the above-mentioned pattern. This protein was completely removed by trypsin treatment of intact protoplasts (Fig. 4, lanes B in panels 3 and 4); this sensitivity to exogenous protease was found consistently in several experiments, some of them with negligible breakage of protoplasts. Similar results were obtained when trypsin was replaced with proteinase K (data not shown). 4.4. Export and assembly o f fusion proteins in E. coli
We considered the possibility that the aminoterminal export signals used might change the structure of O m p A and O m p F proteins in such a way that they become inherently non-exportable. To test this we constructed the plasmids pKTH202 and pKTH204, as derivatives of the E. coli plasmid pBR322 (see MATERIALSAND METHODS). They encode an OmpA8-228 fragment fused to the signal sequence of the a-amylase, or a 13-1actam a s e - O m p F fusion similar to the Bla-OmpF of pKTH180 with the exception that the signal sequence and promoter of a-amylase were replaced with those of/3-1actamase. The plasmids were transformed into E. coli strains devoid of O m p A or O m p F (Table 1). S D S - P A G E and immunoblotting with appropriate sera (anti-OmpA, anti-OmpF or anti-Bla) showed that both O m p A protein and the BlaO m p F fusion proteins were found in the cell envelope fraction of the transformants (data not shown). To study whether the proteins were exported to the outer m e m b r a n e we determined the sensitivity of the transformants to bacteriophage K3 and K20, whose receptors are the native O m p A and O m p F proteins, respectively [18,19]. Table 1 shows that the transformants had also become sensitive to the appropriate phage, while the O m p A and O m p F deficient parents were resistant showing unequivocally their export and assembly into the outer membrane. 5. D I S C U S S I O N In this study the two A m y - O m p A fusion proteins and most of the protein produced from the
Table 1 Effect of plasmids pKTH202 and pKTH204 on the sensitivity of E. coli to OmpA- and OmpF-specific phage K3 and K20, respectively
Phage K3
Phage K20
Strain number
Relevant genotype
Number of plaques ~
PK236 b PK188 h PK188 (pKTH202)
ompA + ompA::Tn5 ompA ::Tn5 [amy - ompA ]
70
PK188 b EHI313 c EH1313 [pKTH204]
ompF + ompF::Tn5 ompF::Tn5 [bla - ompF ]
30
0
44 0
40
a 50 #1 of appropriate dilution of a phage preparation was applied as a spot on a L-plate inoculated with approx, l0 s bacteria and incubated overnight at 37°C. b A nearly isogenic derivative of MC4100 [21]. c A tetracycline-sensitive derivative of PK236 (ompF::Tn5, ompC::TnlO).
bla-ompF fusion construction in B. subtilis were
intracellular. Their size was that expected of the primary translational products, and shows that no cleavage of the signal sequence took place. We did not find significant secretion of either type of fusion protein into the culture medium, although some secretion followed by rapid and complete degradation by exoproteases cannot be excluded. The detection of the uncleaved forms in the medium in stationary phase of growth (presumably due to lysis of cells) suggests that the fusion proteins are resistant to exoproteases and strengthens the evidence against secretion. This pattern of localization is exactly the same found earlier for O m p A and O m p F produced in B. subtilis from constructions containing only the signal sequence of a-amylase in front of the omp genes [6,7]. Clearly, the fusion of a stretch of about 300 (Amy) or 150 (Bla) amino acid residues of a secretable protein did not release the block of secretion of these outer m e m b r a n e proteins in B. subtilis. Furthermore, protease treatment of intact protoplasts did not affect the cell bound A m y - O m p A fusion proteins or the major BlaO m p F protein. Some cleavage of the Bla or Amy part would have been anticipated, even if only
240 this amino-terminal part of the fusion would have been translocated, but the O m p A or O m p F parts were incapable of being inserted in the membrane. The accumulation of non-processed full-size precursors of the fusion proteins and their localization inside the protoplasts with no indication of even partial insertion into the membrane, suggest that, like O m p A and OmpF, they also fail to engage with the export machinery. This is remarkable, because both a-amylase and /3-1actamase alone were fully capable of being exported in B. subtilis [8,9]. We also demonstrated that the fusion of the signal peptide of a-amylase to OmpA8-228 or of the amino-terminal part of /3-1actamase to OmpF, although preventing export in B. subtilis, did not render them exportincompatible in E. coli. We have previously speculated that O m p A and O m p F are not secreted in B. subtilis because of rapid folding into a secretion-incompatible conformation; this would be prevented by some specific chaperones found in E. coli but absent in B. subtilis [7]. Secretion-compatible long aminoterminal extensions of the fusion proteins could be expected to fold to a conformation to be recognized by the export machinery and thus convert the export-incompatible outer m e m b r a n e proteins to an exportable form. However, if the translocation of these proteins is a post-translational event occurring late in relation to the elongation of the polypeptide chain, the fusion protein might not be exported. Thus, if the translocation does not start before the synthesis of at least part of the O m p A or O m p F sequence has taken place, these stretches could lead the folding of the entire fusion protein into a conformation incompatible with translocation and processing. Little is known about the extent of elongation of the secreted proteins in Bacillus before their translocation starts. Extensive variation has been demonstrated in E. coli, with some periplasmic proteins being nearly fully elongated before any processing of the signal sequence as an indication of the passage of the amino-terminus across the m e m b r a n e [20]. Interestingly, a small portion of the Bla-OmpF fusion protein was found to be processed. Pro-
tease accessibility in protoplasts further showed that both the/3-1actamase and O m p F parts of the processed molecule were accessible on the outer surface of the protoplasts and thus translocated. The /3-1actamase part might have slowed down the folding of the O m p F part so that the fusion molecules remained in a loose, export-compatible conformation long enough for a portion of them to be translocated post-translationally. Alternatively transiocation of the Bla part may start relatively early, at a time when some molecules are still unfolded. This portion is translocated early and thus escapes the change to a non-compatible conformation. Indeed, pulse chase experiments have indicated fast processing of /3-1actamase synthesized in B. subtilis relative to its synthesis (I. Palva, personal communication).
ACKNOWLEDGEMENTS This work was supported by the Sigrid Juselius Foundation, Academy of Finland and SITRA, the Finnish National Fund for Research and Development (contract no. 16003TA). We thank Dr. P.H. M~ikel~i for critical reading of the manuscript.
REFERENCES [1] Schein, C.H., Kashiwagi, K., Fujisawa, A. and Weissmann, C. (1986) Bio/Technology 4, 719-725. [2] Dion, M., Rapoport, G. and Doly, J. (1989) Biochimie 71, 747-755. [3] Sarvas, M., Kontinen, V., Himanen, J.-P., Saris, P., Taira, S. and Runeberg-Nyman,L. (1990) In: Proceedings of the 6th International Symposium on Genetics of Industrial Microorganism. Soci~t~ Fran~jaise de Microbiologie. [4] Nikaido, H. (1992) Mol. Microbiol. 6, 435-442. [5] Freudl, R., Schwarz, H., Stierhof, Y.-D., Gamon, K., Hindennach, I. and Henning, U. (1986) J. Biol. Chem. 261, 11355-11361. [6] Kallio, P., Simonen, M., Palva, I. and Sarvas, M. (1986) J. Gen. Microbiol. 132, 677-687. [7] Puohiniemi, R., Simonen, M., Muttilainen, S., Himanen, J.-P. and Sarvas, M. (1992) Mol. Microbiol. 6, 981-990. [8] Palva, I., Sarvas, M., Lehtovaara, P., Sibakov, M. and K~i~iri~iinen, L. (1982) Proc. Natl. Acad. Sci. USA 79, 5582-5586. [9] Palva, I., Pettersson, R.F., Kalkkinen, N., Lehtovaara, P., Sarvas, M., S6derlund, H., Takkinen, K. and K~i~iri~iinen, L. (1981) Gene 15, 43-51.
241 [10] Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. [11] Tommassen, J., van der Ley, P., van der Ende, A., Bergmans, H. and Lugtenberg, B. (1982) Mol. Gen. Genet. 185, 105-110. [12] Sarvas, M. and Nurminen, M. (1985) In: Enterobacterial Surface Antigens: Methods for Molecular Characterization (Korhonen, T.K., Dawes, E.A. and M~ikel~i, P.H., Eds.), pp. 123-137. Elsevier, Amsterdam. [13] Towbin, H., Staehelin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. [14] Fuchs, E. (1976) Eur. J. Biochem. 63, 15-22. [15] Sarvas, M., Hirth, K.P., Fuchs, E. and Simons, K. (1978) FEBS Lett. 95, 76-80.
[16] Klose, M., Maclntyre, S., Schwarz, H. and Henning, U. (1988) J. Biol. Chem. 263, 13297-13302. [17] Bremer, E., Cole, S.T., Hindennach, I., Henning, U., Beck, E., Kurz, C. and Schaller, H. (1982) Eur. J. Biochem. 122, 223-231. [18] Van Alphen, L., Havekes, L. and Lugtenberg, B. (1977) FEBS Lett. 75, 285-290. [19] Lugtenberg, B. and van Alphen, L. (1983) Biochim. Biophys. Acta 737, 51-115. [20] Josefsson, L.-G. and Randall, L.L. (1981) Cell 25, 151157. [21] Casadaban, M.J. (1976) J. Mol. Biol. 104, 541-555.
233
FEMSLE 80061
Incompatibility of outer membrane proteins OmpA and OmpF of Escherichia coli with secretion in Bacillus subtilis" Fusions with secretable peptides Marjo S i m one n 1, Eveliina Tarkka, Ritvaleena Puohiniemi and Matti Sarvas National Public Health Institute, Helsinki, Finland
Received 18 June 1992 Accepted 29 June 1992
Key words: Bacillus subtilis; Escherichia coli; Secretion; Outer membrane protein; OmpA; OmpF
1. SUMMARY The secretion of the outer membrane proteins OmpA and OmpF of Escherichia coli has previously been found to be blocked at an early intracellular step, when these proteins were fused to a bacillar signal sequence and expressed in Bacillus subtilis. We have now fused these proteins to long secretable polypeptides, the amino-terminal portions of a-amylase or /3-1actamase. In spite of this, no secretion of the fusion proteins was detected in B. subtilis. With the exception of a small fraction of the /3-1actamase fusion, the proteins were cell-bound with uncleaved signal sequences. Protease accessibility indicated that the fusion proteins were not even partially exposed on the outer surface of the cytoplasmic membrane. Thus there was no change of the location compared to the OmpA or OmpF fused to the signal sequence
Correspondence to: M. Sarvas, Department of Molecular Bac-
teriology, National Public Health Insitute, SF-00300 Helsinki, Finland. 1 Present address: Institute of Biotechnology, University of Helsinki, SF-00380 Helsinki, Finland.
only. We conclude that, like OmpA and OmpF, the fusion proteins fold into an export-incompatible conformation in B. subtilis before the start of translocation, which we postulate to be a late post-translational event.
2. INTRODUCTION Many proteins that are normally secreted by their natural hosts are secreted poorly or not at all, when synthesized in Bacillus subtilis [1-3]. This indicates that some of the components of the bacillar secretion machinery fail to recognize or interact productively with a wide range of heterologous secretory proteins. This is a serious limitation for the utilization of the secretion capability of bacilli in biotechnical applications. However, the nature of the incompatible components and the steps of secretion affected are largely unknown. We have approached this problem using outer membrane proteins of Gram-negative bacteria as models. In their natural hosts these proteins are secreted through the cytoplasmic membrane and
234 assembled into the outer membrane [4,5]. We introduced the genes of the two major outer membrane proteins OmpA and OmpF of Escherichia coli into B. subtilis [6,7] downstream of the signal sequence of a bacillar exoenzyme, the a-amylase of B. amyloliquefaciens [8]. In spite of this, the fusion proteins were not secreted, but a block of secretion occurred at a very early step of the process [7]. In fact, there was no indication of interaction between the fusion proteins with the secretion machinery. We were therefore interested in learning how OmpA and OmpF would behave when fused to a secretable polypeptide; these studies are reported in this paper.
3. MATERIALS AND METHODS
3.1. Growth conditions For protein expression, B. subtilis strains were grown in liquid culture in two-fold concentrated L-broth containing: NaC1, 10 g 1-1; kanamycin, 10 mg l-1; and potato extract, 30 ml 1-1; overnight at 37°C [6]. E. coli strains were grown on L-agar plates or L-broth [6], containing either ampicillin (100 /xg ml-1), tetracycline (12.5 /xg ml-*) or kanamycin (10 ~g m1-1) at 37°C. 3.2. Construction of plasmids encoding fusion proteins 3.2.1. a-amylase-OmpA-fusions. The ompA fragments were recovered from plasmids pKTH158 [6] and pKTH165 by digestion with HindlII. The ompA fragment from pKTH158 encodes amino acid residues 8-228 of OmpA. The fragment from pKTH165 is similar to that from pKTH158 except that the 3' end of the ompA fragment has been further shortened with exonucleases and encodes residues 8-138. The HindIII fragments were converted to blunt-ended (filled in), and inserted into the plasmid pKTH10 [9], which was cleaved at the KpnI site of the aamylase gene (at codon 289) and blunt-ended with Sl-nuclease. The plasmids thus constructed were designated pKTH198 and pKTH201, respectively; the structures of the fusion proteins encoded are designated Amy-OmpA8-228-Amy and Amy-OmpA8-138 and are shown in Fig. 1.
To express the OmpA fragment 8-228 in E. coli, the plasmids pKTH158 (encoding the residues 8-228 of OmpA fused to the signal sequence of the a-amylase of B. amyloliquefaciens) and pBR322 [10] were ligated to each other at their BamHI sites to make the plasmid pKTH202. 3.2.2. ~-Lactamase-OmpF fusions. The Pstl fragment of the plasmid pJP33 [11], encoding amino acid residues 1-340 of OmpF, was inserted into the Pst I site of the TEM /3-1actamase (Bla) gene of E. coli (at codon 158) in the plasmid pKTH78 [8] to construct pKTH180 (Fig. 1). To synthesize the same fusion protein in E. coli, the PstI fragment of pJP33 was inserted in the PstI site of /3-1actamase of pBR322, resulting in the plasmid pKTH204. The fusion genes in the above plasmids were downstream of the promoter of a-amylase of B. amyloliquefaciens [8], except in pKTH204.
3.3. Protease treatment of protoplasts Protoplasts were prepared from overnight grown cells with lysozyme treatment [7]. Protoplasts were treated with trypsin (2 mg m1-1) before or after disruption with sonication or Triton X-100. As a marker for the exposure of intracytoplasmic proteins to the protease, we analysed the glucose-6-P dehydrogenase activity [7]. 3.4. Other methods 3.4.1. SDS-PAGE. Proteins were electrophoresed in SDS-polyacrylamide gels [12], immunoblotted with specific rabbit antisera [12,13] and visualized by autoradiography using [~25I]labelled protein A or with peroxidase-conjugated anti-rabbit IgG (Bio-Rad, Richmond, CA). 3.4.2. In vitro protein synthesis. An in vitro transcription-translation assay was carried out as described [14]. The proteins were labelled with [35S]methionine (Amersham) and immunoprecipitared [15]. 4. RESULTS
4.1. Fusion constructions We chose two secreted proteins to be fused to the amino-terminus of OmpA or OmpF. One was
235
a bacillar exoprotein, the a-amylase (Amy) of B. amyloliquefaciens, the other a periplasmic protein of E. coli, the TEM /3-1actamase, encoded by amyE and bla, respectively. The fusion proteins were expressed from the multicopy secretion vector based on the promoter and signal sequence of the a-amylase of B. amyloliquefaciens [8]. Both the a-amylase and the /3-1actamase have been shown to be synthesized and secreted at a similar high rate when expressed in B. subtilis from a multicopy plasmid from this promoter and signal sequence [8,9]. Segments of ompA gene were inserted into the amyE gene of the plasmid pKTH10, resulting in the plasmids pKTH198 and pKTH201 (Fig. 1). pKTH198 encodes a tripartite fusion protein designated Amy-OmpA8-228-Amy and composed of the signal sequence and the 289 NH2-terminal residues of the mature a-amylase, followed by residues 8-228 of mature OmpA and finally the C-terminal residues 290-463 of a-amylase (Fig. 1). The fusion protein encoded by pKTH201 and designated Amy-OmpA8-138 contains the same 5' fragment of amyE, followed by ompA codons 8-138 and a translational stop codon 16 residues downstream created by an out-of-frame fusion to the Y-terminal segment of amyE. The larger OmpA8-228 fragment corresponds approximately to the amino-terminal domain of OmpA that is embedded in the outer membrane of E. coli [16]; it lacks the C-terminal domain which is located in the periplasm in E. coli. This amino-terminal fragment is exported and assembled into the outer -31 2
158
membrane in E. coli [17], but fails to be exported and accumulates intracellularly in B. subtilis like the untruncated OmpA [6,7]. OmpF protein was fused to /3-1actamase. The structural gene of the mature part of OmpF was inserted into the bla gene in the plasmid pKTH78 [8] to obtain pKTH180. This resulted in an inframe fusion consisting of the signal sequence of a-amylase, 158 codons corresponding to the amino-terminal part of mature /3-1actamase and the entire ompF gene encoding the amino acid residues 1-340 of the mature protein (Fig. 1); the fusion protein was designated Bla-OmpF.
4.2. Expression and localization of the fusion proteins in B. subtilis The constructed plasmids were used to transform B. subtilis IH6140, which is a mutant with a decreased level of exoproteases [8]. Cultures of these transformants were studied by immunoblotting using anti-OmpA, anti-OmpF, anti-a-amylase and anti-/3-1actamase rabbit sera. No proteins with antigenic specificity of either part of the appropriate fusion protein were detected in the culture supernatants of any of these strains. TCA precipitates of the culture medium of exponential phase cultures were analysed, when there was minimal amount of cellular lysis and very low levels of exoproteases. The secretion of unfused, nontruncated a-amylase and /3-1actamase expressed from plasmids based on the same secretion vector were readily detectable under these conditions [8,9].
1
340
pKTHI80 SS
Bla
OmpF
-31
298
pKTHI98
8
228
290
463
iii!iiiiiiili~iliiii~iiii~iiiil}iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii iii~ili~i~iisi~iiiiii ili~i~i~i~i~!~i~i~!iii~ii~i~iiii!i~ SS
Amy
-31
OmpA
298
8
Amy
138
pK'rH2 01
|| SS
Amy
OmpA
Fig. 1. The structures of the fusion proteins encoded by the inserts in the plasmids pKTH180, pKTH198 and pKTH201. The negative numbers refer to the amino acid residues of the signal sequence and the positive ones to those of appropriate mature proteins. SS, signal sequence of u-amylase of B. amyloliquefaciens. For the other designations see SECTION 3.2.
236
In contrast, the fusion proteins were detected in the particulate fractions of the bacteria, prepared as described [6]. Figure 2 shows that a strongly reactive protein of the expected size of the fusion protein was found in the particulate fraction of strains expressing the a-amylaseOmpA fusions (lanes C, D and G, proteins of about 80 kDa (Amy-OmpA8-228-Amy) and 50 kDa (Amy-OmpA8-138), respectively). In bacteria expressing the Bla-OmpF fusion one strongly and another more weakly labelled protein (about 56 kDa and 54 kDa) were detected. Such proteins were not present in the control strains devoid of the plasmid encoding the fusion protein (lane I). Both anti-a-amylase and anti-OmpA reacted with the fusion in IH6140(pKTH198), and both anti]3-1actamase and anti-OmpF with that of IH6140(pKTH180). However, in the case of pKTH201 the protein was not detected by antiA
15
C
D
OmpA serum, presumably due to the small size of the OmpA fragment (Fig. 2, lane A). Small amounts of the fusion proteins were also detected in the cytoplasmic fractions. The presence of the fusion proteins in the particulate fraction of the bacteria is a finding analogous to the accumulation of precursors of OmpA and OmpF with uncleaved signal peptides in B. subtilis [6,7]. Indeed, the apparent molecular masses of the two Amy-OmpA fusions and the major Bla-OmpF fusion protein seen in the immunoblot (Fig. 2) are close to the deduced molecular mass of their primary translates without cleavage of the signal sequence. Thus the deduced molecular mass of the processed AmyOmpA8-138 protein is about 48 kDa, and that of the primary translate is 51 kDa, the protein observed in Fig. 2 (lane G) is more than 50 kDa. To further confirm the nature of fusion proteins in F
E
94
G
H
94
94
67
67
i!!!i~!/iiill ~~ i~iii!~ ~ i~i~
I
~!~i !ili~
- 67
43
67 43
43
43
30
iS!L!iI! 30
anti-OmpA serum
anti-u-amylase
serum
anti-OmpF serum
Fig. 2. Immunoblotting of the particulate fractions of B. subtilis IH6140 carrying the plasmids: A, pKTH201(Amy-OmpA8-138); B, the parent plasmid of pKTH198; C, pKTH198(Amy-OmpA8-228-Amy); D, pKTH198(Amy-OmpA8-228-Amy); E, the parent plasmid of pKTH198; F, 500 ng of B. amyloliquefaciens o~-amylase (Sigma A6380); G, pKTH201(Amy-OmpA8-138); H, pKTH180(Bla-OmpF); I, the parent plasmid of pKTH180.
237
the particulate fraction, we synthesized the proteins from the pKTH180 DNA template in a cell-free system. The synthesis was carried out in the absence of membranes and thus no cleavage of the signal peptide was expected to take place. Figure 3 shows the products of in vitro synthesis of the Bla-OmpF fusion protein; the protein synthesized had the same apparent molecular mass as the 56 kDa fusion protein found in the particulate fraction of bacteria carrying pKTH180 (Fig. 3, lane A). On the other hand, no 54-kDa protein, also seen in the particulate fraction, was found among the proteins made in vitro. Although this protein was not detected in culture supernatant, its size suggests that it could have been produced by cleavage of the signal peptide. In some membrane preparations of the strain carrying pKTH198 (like the one in Fig. 2, lane D) there was also a second, minor protein with the antigenic specificity of both OmpA and aamylase. It was about 15 kDa smaller than the primary translate and thus clearly not the product of the cleavage of the signal sequence. Although no fusion proteins were detected in the culture supernatants of exponentially growing cells, both Bla-OmpF and Amy-OmpA8-228-Amy proteins were found in small amounts in supernatants of stationary phase cultures. However, the amount was small, less than a few percent of that in the bacteria. Furthermore, in both cases the size of the proteins in the supernatant and cells was similar, suggesting release of the proteins from autolysing cells.
ton X-100) or sonication, and then subjected to immunoblotting. Figure 4 shows that in all cases the size of the major membrane associated fusion protein was not affected by trypsin treatment of intact protoplasts (compare lanes A and B in Fig. 4); in most cases the amount was slightly decreased presumably due to some breakage of protoplasts which was measured by monitoring the enzyme activity of a cytoplasmic marker protein as in [7] (data not shown). In contrast, the A
E5
C
D
u
4.3. Protease accessibility of the fusion proteins in protoplasts One indication of the cytoplasmic localization of the fusion proteins synthesized in B. subtilis was their inaccessibility to exogenous proteases in intact protoplasts and complete breakdown by similar treatment after breakage of the protoplasts [7]. To determine whether any translocation of the fusion proteins across the cytoplasmic membrane took place, we carried out a similar trypsin treatment of protoplasts of strains carrying pKTH198, pKTH201, or pKTH180. In brief, protoplasts were treated with trypsin before and after breakage either by lysis with detergent (Tri-
Fig. 3. Comparison of in vitro and in vivo synthesized BlaOmpF fusion proteins. Lane A, particulate fraction of B. subtilis IH6140(pKTH180) expressing the Bla-OmpF fusion immunoblotted with anti-OmpF and labelled with [125I]protein A. Lanes B and C, proteins synthesized in vitro from pKTH180 and immunoprecipitated with anti-OmpF or antiBla serum, respectively. D, [35S]methionine-labelled proteins synthesized in vitro from pKTH180. The asterisk indicates the position of the 56-kDa Bla-OmpF fusion protein. (In the particular experiment shown in the figure there was not a full set of molecular mass maskers, but the comparison of lane A with other lanes is unambiguous even without markers.)
238 fusion p r o t e i n s were a b s e n t if the protoplasts were b r o k e n before trypsin t r e a t m e n t (Fig. 4 lanes C a n d D). Significantly, n o p e p t i d e s smaller t h a n
1A°AB CD z~
....
~
the p r i m a r y translate were f o u n d by i m m u n o b l o t ting with any of the sera. T h e effect of the protease t r e a t m e n t on the
2AoA -
B co
94
- 43
anti -a-amylase serum
3
A
e
anti-OmpF
c
serum
anti
-a-amylase serum
D
4
A
B
C
D
anti-B- lactamase ~erum
Fig. 4. Protease accessibility of the fusion proteins in protoplasts. (1) IH6140(pKTH]98) (Amy-OmpA8-228-Amy). (2) IH6140(pKTH20D (Amy-OmpA8-138).(3, 4) IH6140 (pKTH180) (Bla-OmpF). Immunoblotting of protoplasts. Ao, a control strain IH6140 or IH6140(pKTH198) in panels 1 and 2, respectively; A, no treatment of protoplasts; B, trypsin treatment of intact protoplasts; C, trypsin treatment after sonication; D, trypsin treatment after solubilization with Triton X-100. The arrowheads show the positions of Amy-OmpA fusion proteins, lmmunoblots were detected with peroxidase-conjugated anti-rabbit IgG (BioRad, Richmond, CA) (panels 1 and 2) or with [12SI]labelledprotein A (panels 3 and 4).
239 minor 54-kDa species of Bla-OmpF fusion protein differed significantly from the above-mentioned pattern. This protein was completely removed by trypsin treatment of intact protoplasts (Fig. 4, lanes B in panels 3 and 4); this sensitivity to exogenous protease was found consistently in several experiments, some of them with negligible breakage of protoplasts. Similar results were obtained when trypsin was replaced with proteinase K (data not shown). 4.4. Export and assembly o f fusion proteins in E. coli
We considered the possibility that the aminoterminal export signals used might change the structure of O m p A and O m p F proteins in such a way that they become inherently non-exportable. To test this we constructed the plasmids pKTH202 and pKTH204, as derivatives of the E. coli plasmid pBR322 (see MATERIALSAND METHODS). They encode an OmpA8-228 fragment fused to the signal sequence of the a-amylase, or a 13-1actam a s e - O m p F fusion similar to the Bla-OmpF of pKTH180 with the exception that the signal sequence and promoter of a-amylase were replaced with those of/3-1actamase. The plasmids were transformed into E. coli strains devoid of O m p A or O m p F (Table 1). S D S - P A G E and immunoblotting with appropriate sera (anti-OmpA, anti-OmpF or anti-Bla) showed that both O m p A protein and the BlaO m p F fusion proteins were found in the cell envelope fraction of the transformants (data not shown). To study whether the proteins were exported to the outer m e m b r a n e we determined the sensitivity of the transformants to bacteriophage K3 and K20, whose receptors are the native O m p A and O m p F proteins, respectively [18,19]. Table 1 shows that the transformants had also become sensitive to the appropriate phage, while the O m p A and O m p F deficient parents were resistant showing unequivocally their export and assembly into the outer membrane. 5. D I S C U S S I O N In this study the two A m y - O m p A fusion proteins and most of the protein produced from the
Table 1 Effect of plasmids pKTH202 and pKTH204 on the sensitivity of E. coli to OmpA- and OmpF-specific phage K3 and K20, respectively
Phage K3
Phage K20
Strain number
Relevant genotype
Number of plaques ~
PK236 b PK188 h PK188 (pKTH202)
ompA + ompA::Tn5 ompA ::Tn5 [amy - ompA ]
70
PK188 b EHI313 c EH1313 [pKTH204]
ompF + ompF::Tn5 ompF::Tn5 [bla - ompF ]
30
0
44 0
40
a 50 #1 of appropriate dilution of a phage preparation was applied as a spot on a L-plate inoculated with approx, l0 s bacteria and incubated overnight at 37°C. b A nearly isogenic derivative of MC4100 [21]. c A tetracycline-sensitive derivative of PK236 (ompF::Tn5, ompC::TnlO).
bla-ompF fusion construction in B. subtilis were
intracellular. Their size was that expected of the primary translational products, and shows that no cleavage of the signal sequence took place. We did not find significant secretion of either type of fusion protein into the culture medium, although some secretion followed by rapid and complete degradation by exoproteases cannot be excluded. The detection of the uncleaved forms in the medium in stationary phase of growth (presumably due to lysis of cells) suggests that the fusion proteins are resistant to exoproteases and strengthens the evidence against secretion. This pattern of localization is exactly the same found earlier for O m p A and O m p F produced in B. subtilis from constructions containing only the signal sequence of a-amylase in front of the omp genes [6,7]. Clearly, the fusion of a stretch of about 300 (Amy) or 150 (Bla) amino acid residues of a secretable protein did not release the block of secretion of these outer m e m b r a n e proteins in B. subtilis. Furthermore, protease treatment of intact protoplasts did not affect the cell bound A m y - O m p A fusion proteins or the major BlaO m p F protein. Some cleavage of the Bla or Amy part would have been anticipated, even if only
240 this amino-terminal part of the fusion would have been translocated, but the O m p A or O m p F parts were incapable of being inserted in the membrane. The accumulation of non-processed full-size precursors of the fusion proteins and their localization inside the protoplasts with no indication of even partial insertion into the membrane, suggest that, like O m p A and OmpF, they also fail to engage with the export machinery. This is remarkable, because both a-amylase and /3-1actamase alone were fully capable of being exported in B. subtilis [8,9]. We also demonstrated that the fusion of the signal peptide of a-amylase to OmpA8-228 or of the amino-terminal part of /3-1actamase to OmpF, although preventing export in B. subtilis, did not render them exportincompatible in E. coli. We have previously speculated that O m p A and O m p F are not secreted in B. subtilis because of rapid folding into a secretion-incompatible conformation; this would be prevented by some specific chaperones found in E. coli but absent in B. subtilis [7]. Secretion-compatible long aminoterminal extensions of the fusion proteins could be expected to fold to a conformation to be recognized by the export machinery and thus convert the export-incompatible outer m e m b r a n e proteins to an exportable form. However, if the translocation of these proteins is a post-translational event occurring late in relation to the elongation of the polypeptide chain, the fusion protein might not be exported. Thus, if the translocation does not start before the synthesis of at least part of the O m p A or O m p F sequence has taken place, these stretches could lead the folding of the entire fusion protein into a conformation incompatible with translocation and processing. Little is known about the extent of elongation of the secreted proteins in Bacillus before their translocation starts. Extensive variation has been demonstrated in E. coli, with some periplasmic proteins being nearly fully elongated before any processing of the signal sequence as an indication of the passage of the amino-terminus across the m e m b r a n e [20]. Interestingly, a small portion of the Bla-OmpF fusion protein was found to be processed. Pro-
tease accessibility in protoplasts further showed that both the/3-1actamase and O m p F parts of the processed molecule were accessible on the outer surface of the protoplasts and thus translocated. The /3-1actamase part might have slowed down the folding of the O m p F part so that the fusion molecules remained in a loose, export-compatible conformation long enough for a portion of them to be translocated post-translationally. Alternatively transiocation of the Bla part may start relatively early, at a time when some molecules are still unfolded. This portion is translocated early and thus escapes the change to a non-compatible conformation. Indeed, pulse chase experiments have indicated fast processing of /3-1actamase synthesized in B. subtilis relative to its synthesis (I. Palva, personal communication).
ACKNOWLEDGEMENTS This work was supported by the Sigrid Juselius Foundation, Academy of Finland and SITRA, the Finnish National Fund for Research and Development (contract no. 16003TA). We thank Dr. P.H. M~ikel~i for critical reading of the manuscript.
REFERENCES [1] Schein, C.H., Kashiwagi, K., Fujisawa, A. and Weissmann, C. (1986) Bio/Technology 4, 719-725. [2] Dion, M., Rapoport, G. and Doly, J. (1989) Biochimie 71, 747-755. [3] Sarvas, M., Kontinen, V., Himanen, J.-P., Saris, P., Taira, S. and Runeberg-Nyman,L. (1990) In: Proceedings of the 6th International Symposium on Genetics of Industrial Microorganism. Soci~t~ Fran~jaise de Microbiologie. [4] Nikaido, H. (1992) Mol. Microbiol. 6, 435-442. [5] Freudl, R., Schwarz, H., Stierhof, Y.-D., Gamon, K., Hindennach, I. and Henning, U. (1986) J. Biol. Chem. 261, 11355-11361. [6] Kallio, P., Simonen, M., Palva, I. and Sarvas, M. (1986) J. Gen. Microbiol. 132, 677-687. [7] Puohiniemi, R., Simonen, M., Muttilainen, S., Himanen, J.-P. and Sarvas, M. (1992) Mol. Microbiol. 6, 981-990. [8] Palva, I., Sarvas, M., Lehtovaara, P., Sibakov, M. and K~i~iri~iinen, L. (1982) Proc. Natl. Acad. Sci. USA 79, 5582-5586. [9] Palva, I., Pettersson, R.F., Kalkkinen, N., Lehtovaara, P., Sarvas, M., S6derlund, H., Takkinen, K. and K~i~iri~iinen, L. (1981) Gene 15, 43-51.
241 [10] Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. [11] Tommassen, J., van der Ley, P., van der Ende, A., Bergmans, H. and Lugtenberg, B. (1982) Mol. Gen. Genet. 185, 105-110. [12] Sarvas, M. and Nurminen, M. (1985) In: Enterobacterial Surface Antigens: Methods for Molecular Characterization (Korhonen, T.K., Dawes, E.A. and M~ikel~i, P.H., Eds.), pp. 123-137. Elsevier, Amsterdam. [13] Towbin, H., Staehelin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. [14] Fuchs, E. (1976) Eur. J. Biochem. 63, 15-22. [15] Sarvas, M., Hirth, K.P., Fuchs, E. and Simons, K. (1978) FEBS Lett. 95, 76-80.
[16] Klose, M., Maclntyre, S., Schwarz, H. and Henning, U. (1988) J. Biol. Chem. 263, 13297-13302. [17] Bremer, E., Cole, S.T., Hindennach, I., Henning, U., Beck, E., Kurz, C. and Schaller, H. (1982) Eur. J. Biochem. 122, 223-231. [18] Van Alphen, L., Havekes, L. and Lugtenberg, B. (1977) FEBS Lett. 75, 285-290. [19] Lugtenberg, B. and van Alphen, L. (1983) Biochim. Biophys. Acta 737, 51-115. [20] Josefsson, L.-G. and Randall, L.L. (1981) Cell 25, 151157. [21] Casadaban, M.J. (1976) J. Mol. Biol. 104, 541-555.
