membrane localisation of a udp-sugar hydrolase, in salmonella, is by an uncleaved n-terminal signal...

6
FEMS Microbiology Letters 114 (1993) 299-304 © 1993 Federation of European Microbiological Societies 0378-1097/93/$06.00 Published by Elsevier 299 FEMSLE 05726 Membrane localisation of a UDP-sugar hydrolase, in Salmonella, is by an uncleaved N-terminal signal peptide Robert A. Jones, Dennis M. Burns, Daryl J. Carruthers and Ifor R. Beacham * Faculty of Science and Technology, Griffith Uniuersity, Nathan, Brisbane, Queensland 4111, Australia (Received 7 September 1993; Accepted 29 September 1993) Abstract: Most isolates of Salmonella contain two unrelated UDP-sugar hydrolases, one of which, encoded by the ushB gene, is inner membrane-associated. Previous studies showed that this enzyme contains a typical N-terminal signal peptide; the evidence also indicated, however, that this peptide is not cleaved, and serves to anchor the UshB protein in the inner membrane. In this report, we present strong evidence that this is indeed the case by using ushB'-'blaM fusions to demonstrate that this signal peptide is capable of localising/3-1actamase to the inner membrane. We also present evidence that UshB is located on the exterior (periplasmic) side of the membrane, and hence has an 'N-terminus inside/C-terminus outside' membrane orientation, consistent with a role in the degradation of external substrates. Key words." UDP-sugar hydrolase; Salmonella; Signal peptide; Cytoplasmic membrane; Periplasm; Membrane anchor Introduction Salmonella enterica contains UDP-sugar hydro- lase isozymes, encoded by the ushA and ushB genes [1]. The former gene encodes a periplasmic enzyme which also possesses 5'-nucleotidase ac- tivity, and it is also characterised by the presence of silent alleles in some Group I Salmonella iso- lates, including 'S. typhimurium LT2' [1,2]. The product of the ushB gene encodes a membrane associated enzyme, which is not bifunctional, and * Corresponding author. Tel.: (07) 875 7445; Fax: (07) 875 7656. with no immunological or primary sequence simi- larity to the ushA gene product [3-5]. The UshB protein has previously been sug- gested to be associated with both the inner and outer membranes [5,6]. The primary sequerice of the enzyme reveals the presence of a classical signal peptide, and the remaining polypeptide does not contain any evident transmembrane do- mains; however, since a precursor form of UshB protein could not be detected by a variety of techniques, it was suggested that the signal pep- tide is uncleaved [5]. Membrane localisation was therefore proposed to involve export to the mem- brane and subsequent anchoring via this N-termi- nal hydrophobic topogenic sequence [5].

Upload: robert-a-jones

Post on 20-Jul-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

FEMS Microbiology Letters 114 (1993) 299-304 © 1993 Federation of European Microbiological Societies 0378-1097/93/$06.00 Published by Elsevier

299

FEMSLE 05726

Membrane localisation of a UDP-sugar hydrolase, in Salmonella, is by an uncleaved N-terminal signal peptide

R o b e r t A. Jones , Denn i s M. Burns , Dary l J. C a r r u t h e r s and I for R. B e a c h a m *

Faculty of Science and Technology, Griffith Uniuersity, Nathan, Brisbane, Queensland 4111, Australia

(Received 7 September 1993; Accepted 29 September 1993)

Abstract: Most isolates of Salmonella contain two unrelated UDP-sugar hydrolases, one of which, encoded by the ushB gene, is inner membrane-associated. Previous studies showed that this enzyme contains a typical N-terminal signal peptide; the evidence also indicated, however, that this peptide is not cleaved, and serves to anchor the UshB protein in the inner membrane. In this report, we present strong evidence that this is indeed the case by using ushB'-'blaM fusions to demonstrate that this signal peptide is capable of localising/3-1actamase to the inner membrane. We also present evidence that UshB is located on the exterior (periplasmic) side of the membrane, and hence has an 'N-terminus inside/C-terminus outside' membrane orientation, consistent with a role in the degradation of external substrates.

Key words." UDP-sugar hydrolase; Salmonella; Signal peptide; Cytoplasmic membrane; Periplasm; Membrane anchor

Introduction

Salmonella enterica contains UDP-sugar hydro- lase isozymes, encoded by the ushA and ushB genes [1]. The former gene encodes a periplasmic enzyme which also possesses 5'-nucleotidase ac- tivity, and it is also characterised by the presence of silent alleles in some Group I Salmonella iso- lates, including 'S. typhimurium LT2' [1,2]. The product of the ushB gene encodes a membrane associated enzyme, which is not bifunctional, and

* Corresponding author. Tel.: (07) 875 7445; Fax: (07) 875 7656.

with no immunological or primary sequence simi- larity to the ushA gene product [3-5].

The UshB protein has previously been sug- gested to be associated with both the inner and outer membranes [5,6]. The primary sequerice of the enzyme reveals the presence of a classical signal peptide, and the remaining polypeptide does not contain any evident transmembrane do- mains; however, since a precursor form of UshB protein could not be detected by a variety of techniques, it was suggested that the signal pep- tide is uncleaved [5]. Membrane localisation was therefore proposed to involve export to the mem- brane and subsequent anchoring via this N-termi- nal hydrophobic topogenic sequence [5].

300

If this proposal is correct, then this N-terminal sequence should direct the membrane localisa- tion of other proteins, such as the normally periplasmic /3-1actamase. In this report we demonstrate that this is the case. Our results provide strong evidence that membrane localisa- tion of UshB protein is determined by an un- cleaved N-terminal topogenic sequence, and strongly suggest, together with other data, locali- sation of the remaining polypeptide in the periplasmic space.

Materials and Methods

Plasmid constructions pAGS2, which contains the ushB gene [5], was

cleaved with NsiI and then treated briefly with BAL31 nuclease; NsiI cleaves, on the upper strand, between those base pairs which corre- spond to the last and first codon positions of amino acid residues 31 and 32, respectively, of UshB protein. Following BAL31 treatment, cleavage with SalI results in a 0.45-kb fragment which was ligated with the larger PvuII-SalI fragment of pJBS633 [7]. The resulting plasmid, pJBF761, is a pJBS633 derivative containing an ushB'-'blaM fusion. The fusion junction is shown in Fig. 1.

A derivative of pJBF761, designated pJHD1, was constructed which contains a deletion of much of the region upstream of ushB. This region was derived from pAGS6 [5], which contains this dele- tion, and pJBF761 by exchange of the SacII-SalI region. In order to allow selective labelling of the fusion protein using the T7 promoter system [8], the T7 promoter was isolated as a SalI-PvuII fragment from pLJI (a derivative of pT7-5; [8,9]) and ligated with the SalI-NruI fragment of pJHD1 containing the ushB'-'blaM fusion and the origin of replication. The resulting construct, pGBH1, lacks the kanamycin-resistance determi- nant present in pJBS633 and its derivatives.

Membrane fractionation Cells were disrupted by sonication, and mem-

branes recovered by centrifugation at 140 000 x g for 18 h. Membranes were separated into inner

and outer membrane fractions by differential sol- ubility in sodium lauryl sarcosinate [10] or by density gradient centrifugation [6] as previously described [5].

Enzyme assays /3-Lactamase was measured using the chro-

mogenic substrate CENTA (Calbiochem; [11]).

Selective labeling of plasmid-encoded proteins Cells containing pGBH1 (which contains the

T7 promoter; see above), were labelled with [35S]methionine in the presence of rifampicin as described [8].

Trypsin accessibility using fractionated maxicells Maxicells [12] containing pUB1, which encodes

the native UshB protein, were fractionated and treated with different amounts of trypsin as fol- lows. A 15-ml culture of CSR603/pUB1 was grown aerobically at 37°C in 'maxicell medium' (M9(X1), 0.1 mM CaCle, 1 mM MgSO4, 0.2% glucose, 1% casein hydrolysate, 20 /~g ml-~ thi- amine, and 100 /xg m1-1 ampicillin) to late-log phase (A450 about 0.8). 10 ml of this culture was placed in a Petri dish on a magnetic stirrer in a laminar flow hood, subjected to 12 s of UV irradi- ation while being gently stirred, then incubated for 1 h at 37°C with shaking. After adding D- cycloserine to 200 /xg m1-1, this incubation was continued overnight at 37°C. The cells were then washed by centrifugation and resuspension in 5 ml M9(× 1) salts twice, followed by resuspension in 5 ml 'supplemented maxicell medium' without magnesium (M9(x1), 0.1 mM CaCI2, 0.4% glu- cose, 20/~g ml -~ thiamine, 100/xg ml -~ arginine, 50/~g m1-1 threonine, 50/xg ml -~ proline, and 1 mM leucine). Cells were then shaken at 37°C for 1 h prior to the addition of Tran35S-label (ICN Biomedicals) to 5/zCi m1-1, followed by incuba- tion for a further 1 h at 37°C with shaking. After pelleting the cells by centrifugation, they were resuspended in 100/xl 0.2 M Tris. HC1 (pH 8.0), 0.5 M sucrose, 1 mM EDTA and transferred to a microcentrifuge tube. Lysozyme was added to 200 /xg ml ~, followed by the addition of 100 /xl of ice-cold deionised water, and the suspension was incubated at 22°C for 40 min. To stabilise sphero-

plasts, MgC12 was added to 20 mM just prior to centrifugation for 10 min. The supernatant (200 /zl, periplasmic fraction) was then carefully re- moved, and 60-/zl aliquots placed into three mi- crocentrifuge tubes. The spheroplast pellet (cyto- plasm-plus-membrane fraction) was resuspended in 200/zl 0.1 M Tris- HC1 pH 8.0, 0.25 M sucrose (EDTA was not added in case this allowed access of trypsin to the cytoplasm (see ref. [13]) in a later step; MgC12 was not added at this stage, either, since it has been reported [13] that Mg 2+ may make membrane-bound proteins resistant to trypsin), and 60-~I aliquots were placed into three microcentrifuge tubes. Trypsin was then added to two periplasmic fraction aliquots, and also two cytoplasm-plus-membrane fraction aliquots, to fi- nal concentrations of 5 /zg ml- 1 and 20/zg ml - 1. The remaining aliquots of both fractions, to which no trypsin was added, served as controls. Tris buffer was added where necessary to ensure all samples had the same volumes (66 pA). Following incubation of all tubes on ice for 30 min, MgC12 was added to the cytoplasm-plus-membrane sam- ples to a final concentration of 20 mM, then the spheroplasts were pelleted by centrifugation and resuspended in 66 ~1 0.1 M Tris . HC1 (pH 8.0), 0.25 M sucrose. 66 ~1 sample buffer (0.2 M Tris . HC1 (pH 6.8), 2% SDS, 20% glycerol, 10% /3-mercaptoethanol) was then added to all sam- pies. Prior to loading equivalent amounts of all samples, including an unfractionated cells sam- ple, on a 12% SDS-polyacrylamide gel, they were denatured by heating at 100°C for 4 min. Follow- ing electrophoresis, the gel was soaked in 7% acetic acid for 30 min, treated with Amplify (Amersham) for 20 min, vacuum-dried, and au- toradiographed.

Results and Discussion

An in-frame u s h B ' - ' b l a M fusion was con- structed, as described in Materials and Methods. This fusion, present in pJBF761, contains the UshB signal peptide plus a further eleven UshB amino acid residues prior to the Bla mature se- quence (Fig. 1). This ensures that if proteolytic cleavage of the signal peptide can potentially

301

ATG AAA AAA ACA GGC TAC TTT TTA CTG GCG GTG ATA CTG ATT GTC GCC Me~ Lys Lys Thr Gly Tyr Phe Leu Leu Ala Val lie Leu lie Val Ala

GCG GCG GGC GTC GGT TAC TGG AAA TTT TCC GGC , ~ \ C CCT CTG CGT CAC... Aia A!a GIy Val GIy Tyr Trp Lys Phe Set Gly Asrt Pro Leu Af~ His. ,

? ? -sla~-

Fig. 1. The N-terminal nucleotide and derived amino acid sequences of ushB'-'blaM. The blaM moiety is underlined, and the remaining sequences are ushB. The end of the signal peptide is indicated, in terms of apparent processing sites (see

[5]), by arrows.

occur, it will not be prevented by the absence of amino acid residues required for cleavage; Le. by truncation of the cleavage window [14].

In-frame fusions were obtained by screening on media containing ampicillin, 200 p.g m1-1, using a high inoculum of cells. However, single cells containing pJBF761 give rise to colonies on 500/~g ml-1 of ampicillin; this level of resistance is indicative of a location of the BlaM moiety which is external to the inner membrane [7].

In order to determine whether the UshB signal peptide directs BlaM to the membrane, u s h B ' -

'blaM was placed under the control of the T7 promoter in order to allow selective labelling of the fusion protein with [35S]methionine. Cells containing pGBH1 (see Materials and Methods) were pulse-labelled with [35S]methionine and then fractionated into total membrane and periplasm- plus-cytoplasm fractions; the total membranes were further fractionated into inner membrane and outer membrane fractions using differential sarkosyl solubility. /3-Lactamase, specified by the bla gene on pT7-3, was also labelled, and served as a control, periplasmic, protein. The results (Fig. 2) clearly demonstrate that, by comparison of the fractionated samples, the fusion protein is entirely membrane associated (cf. lanes 9 and 10) and predominantly in the inner membrane (lane 7), with a small proportion (< 15%) in the outer membrane fraction (lane 8; visible upon overex- posure of the autoradiogram), and none de- tectable in the periplasm-plus-cytoplasm fraction (lane 9 and lane 2). By contrast, native /3-1acta- mase is located predominantly in the periplasm- plus-cytoplasm fraction (lane 5 and lane 1) with some in the inner membrane fraction (lane 4;

302

U (.9 + + pT7-3 pGBH I 13_ cl ~_ cL F-- ~ I--

0 ~E 0 no LU OC

t "T" U 13.. U ~" 13.. i~ rn + I-- + I-- I--

1 2 3 4 5 6 7 8 9 1 0 1 1

~)~,,~i!i ! i l ~ ; ~ ~ i ~

Fig. 2. Fractionation of cells containing pGBH1 or pT7-3. The U s h B ' - ' B I a M fusion protein encoded by pGBHI (lanes 2 and 7 11), and the Bla protein encoded by pT7-3 (lanes 1 and 3-6), were selectively labelled with [35S]metbionine, and cell membranes prepared and fractioned by differential solubility in sarkosyl (see Materials and Methods): Periplasm plus cyto- plasm (PP + C: lanes 1, 2, 5, 9), outer membrane (OM: lanes 3, 8), inner membrane (IM: lanes 4, 7), total protein (TOT PROT: lanes 6, 11), total membranes (TOT MEMB: lane 10). Following SDS-polyacrylamide gel electrophoresis, the gel was dried and fluorographed. Sample Ioadings were as follows in terms of percentage of the original cells: 14.5 (lanes 3, 4, 7, 8); 12.0 (lanes 1, 2, and 9); 6.0 (lane 5); 3.0 (lane 6); 3.6 (lane 10); 2.3 (lane 11). The arrow indicates the position of the U s h B ' - ' B l a M protein, and slightly below, the position of the

mature (native) Bla protein.

approximately 50% of the labelled protein, the upper band, is unprocessed /3-1actamase which is normally membrane bound; the reason for the

presence of mature /3-1actamase is unclear) and none detectable in the outer membrane (lane 3).

Similarly, when cells containing pJBF761 are fractionated, > 90% of the /3-1actamase activity, specified by the ushB'-'blaM fusion, is found in the total membrane fraction (data not shown). Fractionation of these membranes into inner and outer membrane fractions, using sucrose gradient separation ([6]; data not shown), showed that 8-15% of the recovered /3-1actamase activity was associated with the outer membrane, as com- pared with up to 2% of NADH oxidase; the remaining /3-1actamase was inner membrane-as- sociated (61-63%) or was unfractionated (i.e. in the 'M band'; 6).

These experiments demonstrate the membrane tocalisation of UshB'-'BlaM, and that this is determined by the N-terminal signal peptide of UshB. To determine whether, in the case of native UshB protein, the remaining hydrophilic moiety is located exterior to the inner membrane, trypsin accessibility of radioactively labelled UshB was determined. Maxicells containing pUB1, which encodes the native UshB and Bla proteins, were fractionated into periplasmic and cyto- plasm-plus-membrane (spheroplast) fractions, then equivalent aliquots treated with 0, 5, and 20 /zg m1-1 trypsin. The results (Fig. 3) show that the hydrophilic moiety of UshB is indeed located on the periplasmic side of the inner membrane,

PP CM W 1 2 3 4 5 6 7

p p CM W 1 2 3 4 5 6 7

Bla - -~ ................. ~ ~ - - - U s h B

d p l • ~'-- dp2

B l a - - ~ . . . . . . . . . .

~-- UshB

d p l - - " dp2

Fig. 3. Trypsin accessibility of UshB protein in spheroplasts. Maxicells containing pUB1, which encodes the native UshB and Bla proteins, were fractionated and treated with trypsin as described in Materials and Methods. Lanes 1-3 are equivalent a[iquots of the periplasmic fraction (PP) treated with 0, 5, and 20 /xg ml 1 trypsin, respectively. Lanes 4 -6 are equivalent aliquots of the cytoplasm-plus-membrane (spheroplast) fraction (CM) treated with 0, 5, and 20 p~g ml z trypsin, respectively. Lane 7 is unfractionated cells (W) without trypsin treatment. Bla, /3-1actamase protein; UshB, UshB protein; dp l and alp2 are degradation products of Bla and UshB, respectively (see text). The autoradiograph on the left side is a shorter exposure than that on the

right side.

since UshB protein in spheroplasts is cleaved by trypsin. As the concentration of trypsin was in- creased, the intensity of UshB decreased while the intensity of a degradation product (desig- nated dp2) increased (lanes 4-6). The conclusion that dp2 (lanes 5, 6) is, in fact, a degradation product of UshB is based on the observation that no band was present in the corresponding posi- tion in lane 4 (no trypsin treatment) even upon overexposure of the gel (data not shown). The Bla protein in the periplasmic fraction is, as ex- pected, also cleaved by trypsin: as the concentra- tion of trypsin was increased, the intensity of Bla decreased while the intensity of a degradation product (designated dpl) increased. Dpl was identified as a degradation product of Bla on the same basis that dp2 was identified as a degrada- tion product of UshB (see above).

In conclusion, the localisation of the UshB'- 'BlaM fusion protein in the inner membrane demonstrates that the N-terminal signal peptide of UshB serves as a topogenic sequence to both initiate export and to act as a membrane anchor, as previous evidence indicating that it is not cleaved, suggested [5]. Further, the ampicillin re- sistance strongly suggests that the remaining, hy- drophilic, portion of the UshB'- 'BlaM polypep- tide resides exterior to the cell membrane, in the periplasm (see Fig. 4). Since there is no evidence for hydrophobic transmembrane segments, in the UshB primary sequence (aside from the N-termi- nal sequence; [5]), and since membrane localisa- tion is fully explained by the N-terminal mem-

OM

~ C Periplasm

"I" IM N

Fig. 4. Proposed mode of localisation of UshB protein. OM, outer membrane; IM, inner membrane. N and C refer to the amino terminal and carboxyl terminal ends, respectively, of the UshB (or UshB'- 'BIaM) proteins; the dark rectangle

represents the signal peptide within the IM.

303

brane anchor, then it may be reasonably assumed that this is also the case for the UshB protein. This conclusion is, however, compromised by the possible location of about 10% of UshB'- 'BlaM in the outer membrane (see above), but is sup- ported by the trypsin accessibility of UshB pro- tein in spheroplasts (see Fig. 3). Exterior location of the majority of the polypeptide is also consis- tent with the role of this UDP-sugar hydrolase in the degradation of external substrates [3], as in the case of many soluble periplasmic enzymes [15] including the periplasmic UDP-sugar hydro- lase encoded by ushA (see Introduction; [16,17]).

UshB protein therefore constitutes an example of a 'class II' integral membrane protein, with an N-terminus inside/C-terminus outside mem- brane orientation ([18]; Fig. 4); this is consistent with positive charges at the N-terminal end of the UshB signal peptide ([19]; Fig. 1). Examples of such membrane proteins, although numerous in eukaryotes, are relatively rare in bacteria. The TonB protein of E. coil is also a membrane protein of this type [20], although its function, in signal transduction [21,22], is quite distinct from the enzymatic function of UshB.

The possible outer membrane localisation of UshB remains an open question. Although exper- iments to date, including those reported here, have failed to disprove an association, in vivo, of a small proportion of UshB in the outer mem- brane, an artifactual association following cell fractionation remains possible.

References

1 Edwards, C.J., Innes, D.J., Burns, D.M. and Beacham, I.R. (1993) UDP-sugar hydrolase isozymes in Salmonella enterica and Escherichia coli: Silent alleles of ushA in related strains of Group I Salmonella isolates, and of ushB in wild-type and KI2 strains of E. coli, indicate recent and early silencing events, respectively. FEMS Microbiol. Lett. 114, 293-298.

2 Burns, D.M. and Beacham, I.R. (1986) Identification and sequence analysis of a silent gene (ushA °) in Salmonella typhimurium. J. Mol. Biol. 192, 163-175.

3 Glaser, L., Melo, A. and Paul, R. (1967) Uridine diphos- phate sugar hydrolase. Purification of enzyme and protein inhibitor. J. Biol. Chem. 242, 1944-1954.

4 Beacham, I.R. and Wilson, M.S. (1982) Studies on the UDP-sugar hydrolases from Escherichia coli and

304

Salmonella typhimurium. Arch. Biochem. Biophys. 218, 603-608.

5 Garrett, A.R., Johnson, L.A. and Beacham, I.R. (1989) Isolation, molecular characterization and expression of the ushB gene of Salmonella typhimurium which encodes a membrane-bound UDP-sugar hydrolase. Mol. Microbiol. 3, 177-186.

6 0 s b o r n , M.J,, Gander, J.E., Parisi, E. and Carson, J. (1972) Mechanism of assembly of the outer membrane of Salmonella typhimurium. Isolation and characterisation of cytoplasmic and outer membrane. J. Biol. Chem. 217, 3962-3972.

7 Broome-Smith, J.K. and Spratt, B.G. (1986) A vector for the construction of translational fusions to TEM /3-1acta- mase and the analysis of protein export signals and mem- brane protein topology. Gent 49, 341-349.

8 Tabor, S. (1990) Expression using the T7 RNA poly- merase/promoter system. In: Current Protocols in Molec- ular Biology (Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Smith, J.A., Seidman, J.G. and Struhl, K., Eds.), pp. 16.1.1-16.1.11. John Wiley and Sons, New York, NY.

9 Johnson, L.A., Beacham, I.R., MacRae, I.C. and Free, M.L. (1992) Degradation of triglycerides by a pseu- domonad isolated from milk: Molecular analysis of a li- pase-encoding gene and its expression in Escherichia coli. Appl. Environ. Microbiol. 58, 1776-1779.

10 Filip, C., Fletcher, G., Wulff, J.L. and Earhart, C.F. (1973) Solubilisation of the cytoplasmic membrane of Escherichia coli by the ionic detergent sodium-lauryl sarcosinate. J. Bacteriol. 115, 717-722.

11 Jones, R.N., Novick, W.W., Barry, A.I. and Thornsberry, C. (1982) In vitro evaluation of CENTA, a new /3-1acta- mase-susceptible chromogenic cephalosporin reagent. J. Clin. Microbiol. 15, 954-958.

12 Burns, D.M., Abraham, L.J. and Beacham, I.R. (1983) Characterisation of the ush gene of Escherichia coli and its protein products. Gene 25, 343-353

13 Tommassen, J. and Lugtenberg, B. (1984) Amino terminus of outer membrane PhoE protein: localization by use of a bla-phoE hybrid gene. J. Bacteriol. 157, 327-329

14 Von Heijne, G. (1984) How signal sequences maintain cleavage specificity. J. Mol. Biol. 173, 243-251.

15 Ferguson, S.J. (1992) The periplasm. In: Prokaryotic Struc- ture and Function: A New Perspective (Mohan, S., Dow, C. and Cole, J.A., Eds.), pp. 311-339. Cambridge Univer- sity Press, Cambridge.

16 Beacham, I.R., Kahana, R., Levy, L. and Yagil, E. (1973) Mutants of Escherichia coli K12 cryptic or deficient in 5'-nucleotidase (uridine diphosphate-sugar hydrolase) and 3'-nucleotidase (cyclic phosphodiesterase) activity. J. Bac- teriol. 116, 957-964.

17 Yagil, E. and Beacham, I.R. (1975) Uptake of adenosine 5'-monophosphate by E. coli. J. Bacteriol. 121,401-405.

18 Von Heijne, G. and Gavel, Y. (1988) Topogenic signals in integral membrane proteins. Eur. J. Biochem. 174, 671- 678.

19 Von Heijne, G. (1986) Towards a comparative anatomy of N-terminal topogenic sequences. J. Mol. Biol. 189, 239- 242.

20 Postle, K. and Skare, J.T. (1988) Escherichia coli TonB protein is exported from the cytoplasm without proteolytic cleavage of its amino terminus. J. Biol. Chem. 263. 11000- 11007.

21 Hannavy, K., Barr, G.C., Dorman, C.J., Adamson, J., Mazengera, L.R., Gallagher, M.P., Evans, J.S., Levine, B.A., Trayer, I.P. and Higgins, C.F. (1990) TonB protein of Salmonella typhimurium. A model for signal transduc- tion between membranes. J. Mol. Biol. 216, 897-910

22 Brewer, S., Tolley, M., Trayer, I.P., Barr, G.C., Dorman, C.J., Hannavy, K., Higgins, C.F., Evans, J.S., Levine, B.A. and Wormald, M.R. (1990) Structure and function of X-pro dipeptide repeats in the TonB proteins of Salmonella typhimurium and Escherichia coli. J. Mol. Biol. 216, 883-895