Identification of Secretion Determinants of the Bordetella pertussis BrkA Autotransporter

The autotransporters comprise a functionally diverse familyof gram-negative proteins that mediate their own export acrossthe bacterial outer membrane. They consist of an amino-terminalpassenger region called the " ${alpha}$ -domain" and the structural hallmarkof the autotransporter family, a carboxy-terminal transporterregion usually referred to as the "ß-domain." Thepassenger region can be quite diverse and constitutes the effectorfunctions of these proteins, whereas the C-terminal region isconserved and is responsible for translocating the passengermoiety across the outer membrane. BrkA is the 103-kDa autotransporterprotein in Bordetella pertussis that is cleaved to yield a 73-kDaN-terminal ${alpha}$ -domain and a 30-kDa C-terminal ß-domain.We have previously shown that a recombinant form of the ß-domainof BrkA is capable of forming channels in artificial membranes.Here, we define two additional secretion determinants of BrkA.N-terminal sequencing of the 73-kDa BrkA passenger from B. pertussisand Escherichia coli revealed that BrkA has a 42-amino-acidsignal peptide. In addition, deletion analysis of BrkA identifieda 31- to 39-amino-acid region found immediately upstream ofthe ß-domain that was essential for surface expression.This 31- to 39-amino-acid linker region, together with the ß-domain,defines the minimal BrkA translocation unit. The linker regionmay also serve to anchor the BrkA passenger to the bacterialsurface.

INTRODUCTION

Top

Introduction

Autotransporters are outer membrane proteins found in gram-negativebacteria. They are multidomain proteins composed of an N-terminalsignal sequence to direct translocation across the inner membrane,a passenger region to be delivered to the cell surface, anda conserved C-terminal transporter region that is proposed tomediate secretion across the outer membrane (9). Most autotransportersare proteolytically processed close to the transporter regionto yield an ${alpha}$ domain, which comprises most of the passenger region,and the remaining fragment called the ß-domain, whichencompasses the transporter.

Several hundred autotransporters can now be identified fromthe growing list of completed bacterial genomes. They representputative virulence factors (8), possible vaccine candidates(33), and an efficient way to display heterologous proteins(18), including antigenic determinants (14), important enzymes(16), heavy metal-detoxifying agents (32), and platforms forsteroid biosynthesis (10). A detailed understanding of autotransportersecretion mechanisms not only will shed light on the biologicalproblem of traversing the outer membrane, seemingly in one step,but also will contribute to the improved engineering of autotransportersfor specific biotechnological purposes. At the moment, it isnot known whether or what parts of the autotransporter secretionprocess (9, 26, 35) apply universally to all autotransporters.

Bordetella pertussis is the gram-negative mucosal pathogen thatcauses whooping cough (17). The autotransporter protein BrkAis a B. pertussis virulence factor that confers serum resistanceand also acts as an adhesin (4). BrkA expression is controlledby a sensor kinase response regulator system called "Bvg" (37).Consistent with the model of autotransporter secretion, BrkAis expressed as a 103-kDa precursor that is processed duringsecretion to yield a 73-kDa N-terminal ${alpha}$ -domain and a 30-kDaC-terminal ß-domain (28). Following translocation,the cleaved ${alpha}$ -domain remains tightly associated with the bacterialsurface and is not detected in B. pertussis culture supernatants(24). The processed ß-domain has been isolated fromB. pertussis outer membrane fractions, and the processing sitehas been determined to occur between residues Asn⁷³¹ and Ala⁷³²(25).

In an effort to elucidate the mechanism(s) of autotransportersecretion, we have begun to characterize BrkA secretion. Wehave shown that a recombinant form of the C-terminal regionof BrkA encompassing the ß-domain has the capacityto form channels with a conductivity of 3.2 nS in planar lipidbilayer experiments (28). Such a channel would be sufficientto translocate an unfolded or partially unfolded passenger pastthe outer membrane. Here we report the characterization of additionalregions within BrkA that are necessary for secretion. We showthat BrkA has a 42-amino-acid signal peptide to traverse theinner membrane, and we define the minimal translocation unitnecessary to mediate secretion of the BrkA passenger to thebacterial surface.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and plasmids and growth conditions. The bacterial strains and plasmids used in this study are listedin Table 1. E. coli strains were cultured at 37°C on Luriabroth or Luria agar supplemented with the appropriate antibiotics.B. pertussis strains were maintained on Bordet-Gengou (BG) medium(BBL, Cockeysville, Md.) containing 15% sheep's blood (RA Media,Calgary, Alberta, Canada), as previously described (4). Whennecessary, the following antibiotics were added at the indicatedconcentrations to the media: nalidixic acid, 30 µg/ml;kanamycin, 50 µg/ml; ampicillin, 100 µg/ml; andgentamicin, 10 µg/ml.

View this table:
[in this window]
[in a new window]
TABLE 1. Strains and plasmids used in this study

Recombinant DNA techniques. All DNA manipulations were carried out by standard techniques(27). Restriction enzymes were purchased from New England Biolabs(Beverly, Mass.). The primers used in this study were purchasedeither from the University of British Columbia (UBC) NucleicAcid Protein Services Unit (Vancouver), Sigma-Genosys (The Woodlands,Tex.), or Alpha DNA (Montreal, Quebec, Canada) (Table 2). DNAsequencing was done with an ABI Prism 377 DNA sequencer (AppliedBiosystems, Foster City, Calif.) at the UBC Nucleic Acid ProteinServices Unit.

View this table:
[in this window]
[in a new window]
TABLE 2. Primers used in this study

The B. pertussis strain BBC9DO was made by introducing a secondcopy of the brkA gene (on plasmid pUW2171) into the chromosomeof strain BBC9, a pertactin mutant of B. pertussis, as describedpreviously (5).

Construct pDO6935, which constitutively expresses low levelsof BrkA in E. coli, was derived by excision of a 476-bp AatIIfragment of pRF0166. Plasmid pDO6935 was used as a templatein all subsequent PCRs described in this study. All PCRs wereperformed with Vent polymerase (New England BioLabs) with thefollowing cycles: an initial denaturation step of 2 min at 94°Cfollowed by 30 cycles of 45 s at 94°C, 30 s at 60°C,and 1 min/kb at 72°C. The last cycle was followed by anadditional 10 min at 72°C. Amplified PCR products were separatedon an agarose gel, and a band of the expected size was extractedand cloned as described below. The primers used in this studyare listed in Table 2.

Construct pDO181 was made by PCR with primer pairs DO1210F andDO1614R and DO2894F and BRK-CR. The resulting products weredigested with restriction enzyme pairs AscI and XbaI and XbaIand BamHI, respectively. In a triple-ligation reaction, theseproducts were ligated into a 5-kb AscI- to BamHI-digested fragmentof pDO6935 to yield pDO181. Construct pDO182 was generated viathe same strategy with primer sets DO1210F and DO1893R and DO2894Fand BRK-CR. Constructs pDO244 and pDO246 were made with primerpair DO1975F and BRK-CR to generate a PCR product that was subsequentlydigested with AscI and BamHI. The resulting 1.3-kb product wasthen ligated into either a 5.3-kb XbaI- to BamHI-digested fragmentof pDO181 or a 5.5-kb XbaI- to BamHI-digested fragment of pDO182to yield pDO244 and pDO246, respectively.

Constructs pGD1, pGD2, pGD3, pGD4, pGD5, pGD6, pGD7, pGD8, pGD9,pGD10, pGD10.5, pGD11, and pGD12 were made by PCR with forwardprimers BRK-2113F, BRK-2398F, BRK-2650F, BRK-2752F, BRK-2821F,BRK-2890F, BRK-3010F, BRK-3184F, BRK-3238F, BRK-3289F, BRK-3310F,BRK-3370F, and BRK-3601F, respectively. BRK-CR was used as thereverse primer in each of the reactions. The amplified productswere purified, digested with XbaI and HindIII, and ligated intoa 4.3-kb XbaI- to BamHI-digested fragment of pDO246.

SDS-PAGE and immunoblot analysis. For detection of expressed BrkA via sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE) or immunoblotting, E. coli cultureswere grown to an optical density at 600 nm (OD₆₀₀) of 0.7 andpelleted. Trypsin accessibility experiments were performed accordingto a previously described protocol (18) with slight modifications.In brief, cell pellets were resuspended in 0.2 ml of phosphate-bufferedsaline (PBS) to an OD₆₀₀ of $~$ 10. To 0.1 ml of cells, 2 µlof 10-mg/ml trypsin was added to yield a final trypsin concentrationof 200 µg/ml. Cells were incubated in the presence ofprotease for 10 min at 37°C, pelleted by centrifugation,and washed three times with PBS containing 10% fetal calf serumto stop digestion and once in PBS alone. As a control, cellpellets were simultaneously processed in the same manner inthe absence of trypsin. Washed pellets were finally resuspendedin sample buffer and immediately boiled for 5 min prior to SDS-PAGE.

For immunoblot analysis, samples were resolved by SDS-PAGE (4,15) and transferred to Immobilon-P membranes (Millipore, Etobicoke,Ontario, Canada) as described previously (24). Blots were probedwith heat-inactivated rabbit anti-BrkA antiserum and horseradishperoxidase-conjugated goat anti-rabbit secondary antibody (ICNBiomedicals, Costa Mesa, Calif.) diluted 1/50,000 and 1/10,000,respectively (24). Renaissance chemiluminescent reagent (NENLife Science Products, Boston, Mass.) was used to develop immunoblots.The rabbit anti-BrkA antiserum is specific for residues Met¹to Glu⁶⁹³ of BrkA (24). Molecular masses were determined withKaleidoscope prestained markers purchased from Bio-Rad (Hercules,Calif.).

N-terminal sequencing. Whole-cell lysates of strains BBC9DO (a pertactin [prn] mutantwith two copies of brkA), and BBC9BrkA (a prn brkA double mutant)(4) were resolved by SDS-PAGE and transferred to an Immobilon-Pmembrane (Millipore). A unique band migrating at approximately73 kDa in the BBC9DO lane was excised from the membrane andsequenced by Edman degradation by the UBC Nucleic Acid and ProteinServices core facility.

Immunofluorescence analysis. E. coli cells were grown to an OD₆₀₀ of 0.7, pelleted by centrifugation,and resuspended in PBS. Resuspended cells were immobilized ona glass slide that had been previously treated with 0.1% poly-L-lysine(Sigma). Slides were washed three times with PBS to remove unboundbacteria and subsequently probed with a 1/200 dilution of heat-inactivatedrabbit anti-BrkA antiserum (24) and a 1/100 dilution of fluoresceinisothiocyanate (FITC)-conjugated goat anti-rabbit antibody (JacksonImmunoResearch Laboratories, West Grove, Pa.), respectively.Slides were washed three times with PBS containing 1% bovineserum albumin between each step to remove unbound material.Bacteria were visualized under epifluorescence with a ZeissAxioscop-2 microscope. Phase-contrast and fluorescent imageswere captured digitally.

RESULTS AND DISCUSSION

Top

Results and Discussion

Identification of the BrkA signal peptide. It was previously reported that sequence analysis of the 1,010-amino-acidprotein BrkA did not identify a conventional signal peptide(4). More recent analysis with SignalP V2.0 (22) has predicteda signal peptide of 44 amino acids by the neural network predictionmethod and a cleavage site at 43 amino acids by the hidden Markovmodel method (23). To experimentally determine the BrkA signalpeptide, N-terminal sequencing was performed with the 73-kDamoiety of BrkA. The amount of BrkA seen in whole-cell lysatesof B. pertussis represents a small fraction of the total amountof cellular protein. Furthermore, at 73 kDa, BrkA migrates toa similar position on SDS-PAGE gels as the 69-kDa protein pertactin,a protein with which it shares sequence identity (4). To circumventthese issues, we introduced a second copy of the brkA gene intothe chromosome of strain BBC9, a pertactin mutant of B. pertussis,to create strain BBC9DO. Western blot analysis of this strainwith antibodies to pertactin and BrkA confirmed the lack ofexpression of pertactin and the increased expression of BrkArelative to that in wild-type strains (data not shown). Whole-celllysates of strain BBC9DO were resolved by SDS-PAGE and transferredto an Immobilon-P membrane, and a unique band migrating at approximately73 kDa was excised and sequenced by Edman degradation. Six cyclesof Edman degradation revealed an N-terminal sequence of QAPQA.This sequence corresponds to amino acids 43 through 47 of BrkA.Similar results were obtained with a recombinant brkA constructexpressed in E. coli (data not shown). Thus, both in B. pertussisand in E. coli, BrkA is processed between residues Ala⁴² andGln⁴³. A signal peptide of this length is not unusual for autotransporters(9).

Expression of BrkA in E. coli. We chose to study BrkA secretion in E. coli, since it has beenused as a host to study secretion of a variety of autotransporters(11-13, 18, 21, 29, 31, 34, 35), thus allowing comparisons tobe made between different autotransporters and because mutationalanalysis of BrkA is greatly facilitated in E. coli. PlasmidpDO6935 was derived from pRF1066 (4), which carries the entirebrk locus encoding the divergently transcribed brkA and brkBgenes (Table 1). pDO6935 was generated by excision of a 476-bpAatII fragment from pRF1066, resulting in a deletion of the5' region of the brkB gene. pDO6935 was transformed into E.coli strain UT5600, which is deficient for the outer membraneproteases OmpT and OmpP (7). UT5600 has been used in the pastto study secretion of the Neisseria immunoglobulin A (IgA) protease(11, 34, 35), the E. coli AIDA-1 adhesin (18, 19), and the ShigellaVirG (IcsA) autotransporters (31). BrkA expression was assessedby a previously described anti-BrkA polyclonal antiserum (24)that specifically recognizes both denatured and native formsof the 73-kDa BrkA ${alpha}$ -domain. Immunoblots of whole-cell lysatesresolved by SDS-PAGE show that BrkA was expressed to yield twomajor species migrating at approximately 103 and 73 kDa. The103-kDa product corresponds to the unprocessed full-length precursorand the species migrating at 73 kDa corresponds to the cleaved ${alpha}$ -domain (Fig. 1A and B, lane 1). Although BrkA is Bvg regulatedin B. pertussis, the promoter region responsible for drivingBrkA expression from pDO6935 in E. coli is not known. We previouslyreported that the overexpression of full-length BrkA in E. coliis toxic (24); however, in the absence of IPTG induction, thelevels of BrkA expression in E. coli with this construct aresimilar to those seen in B. pertussis (4, 24).

View larger version (34K):
[in this window]
[in a new window]
FIG. 1. BrkA expression in E. coli strain UT5600. (A) BrkA domain organization: signal peptide (SP [residues 1 to 42]), passenger or ${alpha}$ -domain (residues 43 to 731), and ß-domain (residues 732 to 1010). (B) Western immunoblot of E. coli UT5600 whole-cell lysates resolved by SDS-PAGE (11% polyacrylamide), probed with anti BrkA antiserum, and detected with goat anti-rabbit antiserum conjugated to horseradish peroxidase. Lanes: 1 and 2, pDO6935 (wild-type copy of brkA gene); 3 and 4, pBluescript (vector control). Specific BrkA bands are indicated. U, unprocessed 103-kDa precursor protein; *, 73-kDa processed passenger moiety. Cells were processed in the presence (+) or absence (-) of trypsin as described in Materials and Methods. (C) Surface expression of BrkA in E. coli UT5600 detected via indirect immunofluorescence. The top panels show phase-contrast images, and the bottom panels show epifluorescence images.

To determine whether BrkA is translocated to the surface ofE. coli, trypsin accessibility and immunofluorescence experimentswere performed with whole cells. When cells were incubated withtrypsin, a marked decrease in the 73-kDa moiety was observed,and two products of approximately 40 and 45 kDa were detectedby Western immunoblotting (Fig. 1B, lane 2). The cleavage sitesproducing the 40- and 45-kDa species are unknown, and over time,both species are lost. The intensity of the 103-kDa productremained constant following trypsin digestion, suggesting thatthe 103-kDa band represents an intracellular form of the proteininaccessible to trypsin. Concomitant with this result, BrkAwas detected on the surface of E. coli (Fig. 1C) and appearedevenly distributed, as shown by indirect immunofluorescencestaining. Secreted BrkA could not be detected in concentratedculture supernatants, suggesting that the cleaved passengerremains noncovalently associated with the bacterium (data notshown). Taken together, these data indicate that BrkA is exportedto the surface of E. coli strain UT5600 and is processed (independentlyof proteases OmpT or OmpP) in a manner similar to that observedin B. pertussis (24).

Identification of the minimal BrkA translocation unit necessary for surface expression. The natural cleavage of three well-characterized autotransporters,IgA protease (11), VirG/IcsA (6), and AIDA-1 (30), results inß-domains of 45, 37, and 48 kDa, respectively. Byusing a series of protease susceptibility assays and experimentswith heterologous proteins fused to N-terminally-truncated ß-domains,minimal regions necessary to display passenger proteins havebeen identified for these autotransporters. They have in common,a membrane-embedded ß-core of $~$ 25 to 30 kDa found atthe extreme C terminus, preceded by a so-called "linker" region(11, 19, 31). In these autotransporters, the linker region hasbeen shown to be necessary for the translocation of the passengerdomain to the bacterial surface. The linker region togetherwith the outer membrane-embedded ß-core make up whathas been coined the "translocation unit" (19).

Having demonstrated that BrkA is targeted to the outer membraneof E. coli, we next developed a deletion-based strategy to definethe boundaries of the minimal translocation unit of BrkA. N-terminalsequencing of proteins from outer membrane preparations of B.pertussis has localized the processing of BrkA to between Asn⁷³¹and Ala⁷³² (25), resulting in a ß-domain of 30 kDa(28). At 30 kDa, the BrkA ß-domain is smaller thanthe ß-domains for IgA protease, VirG/IcsA, and AIDA-1,but it approaches the size of the outer membrane-embedded ß-coresidentified for these proteins (11, 19, 31). We constructed aseries of brkA deletion mutants by using PCR mutagenesis. Asshown in Fig. 2A, mutant proteins consisted of the first 228amino acids of BrkA (Met¹ to Gly²²⁸) fused in frame to processivedeletions of the C-terminal region of the BrkA ${alpha}$ -domain leadinginto the BrkA ß-domain. BrkA (Met¹ to Gly²²⁸) waschosen as a passenger, since heterologous passengers such ascholera toxin B subunit (12) may be inefficiently translocateddue to structural limitations (i.e., disulfide bond formation).In addition, it has been suggested that the extended signalsequences observed in many autotransporters may play a rolein secretion (9). Therefore, the inclusion of the native BrkAsignal sequence within the passenger avoids any influence thata nonnative signal sequence may have on secretion. All deletionstrains were derivatives of pDO6935, thereby ensuring a commonpromoter for the wild-type and mutant constructs (Table 1).

View larger version (40K):
[in this window]
[in a new window]
FIG. 2. Expression of BrkA deletion constructs in E. coli UT5600. (A) Diagram illustrating positions of BrkA in-frame deletions. Deleted regions are indicated by dotted lines, and deletion boundaries correspond to the wild-type BrkA amino acid designation. The BrkA domain structure is described in Fig. 1. Construction of mutations is described in the Materials and Methods. Plasmids are described in Table 1. (A) pGD1. (B) pGD2. (C) pGD3. (D) pGD4. (E) pGD5. (F) pGD6. (G) pGD7. (H) pGD8. (I) pGD9. (J) pGD10. (K) pGD10.5. (L) pGD11. (M) pGD12. E. coli UT5600 bacteria were transformed with BrkA deletion constructs (plasmids A to M) and grown to an OD₆₀₀ of approximately 0.7. Bacteria were harvested, and BrkA surface expression was assessed by immunoblotting or indirect immunofluorescence. (B) Immunoblotting following resolution of whole-cell lysates by SDS-PAGE. The band migrating within the region denoted as "U" in each lane corresponds to the unprocessed, precursor form of BrkA, and the band denoted with an asterisk (*) corresponds to the processed passenger domain of BrkA. Cells were processed in the presence (+) or absence (-) of trypsin as described in Materials and Methods. Molecular mass markers (in kilodaltons) are indicated to the left of the panel. (C) BrkA expression in E. coli strain UT5600 detected by indirect immunofluorescence. The top panels show phase-contrast images, and the bottom panels show epifluorescence images.

An attempt was made to target our deletions to regions thatwould not disrupt the core structure of the protein. Secondarystructural analysis with PsiPred (20) predicts that BrkA ispredominantly composed of ß-strands (data not shown).In addition, the closest relative to BrkA in the database isthe B. pertussis autotransporter pertactin (4). The structureof the pertactin passenger domain has been solved and was shownto be a monomer folded into a single domain that is almost entirelymade up of a right-handed cylindrical ß-helix (3).Given that BrkA and pertactin passenger domains share 27% sequenceidentity and 39% sequence similarity, we refined our secondarystructural prediction by overlaying the pertactin structuralcoordinates (1DABA) onto a BrkA-pertactin primary amino acidsequence alignment. The best alignment was between Arg¹⁷⁵ toPro⁵⁷² in pertactin and Val³⁰¹ to Gln⁷⁰⁷ in BrkA. Using thisanalysis, we systematically targeted N-terminal deletions tointervening regions with the predicted ß-strands.

The effects of each deletion on BrkA expression and processingwere assessed by immunoblotting of whole-cell lysates resolvedby SDS-PAGE. As shown in Fig. 2B, each mutant form of BrkA wasexpressed, indicating that the specific deletions did not renderthe individual mutant protein products markedly unstable. Indeletion mutants A through J, products corresponding to boththe unprocessed precursor (region designated as "U") and thecleaved passenger (asterisk) were detected (Fig. 2B). In contrast,only the unprocessed precursor could be detected in deletionmutants K, L, and M. Given our previous observation that thecleaved passenger domain represents a major fraction of thesurface-expressed wild-type BrkA (Fig. 1), these data suggestedthat BrkA deletion mutants A through J were being exported tothe bacterial surface, but mutants K, L, and M were not. Insupport of this observation, trypsin accessibility assays andindirect immunofluorescence experiments were performed. As expected,exposure of whole cells to trypsin digestion resulted in thecomplete absence of the band corresponding to the processedpassenger domain (Fig. 2B, lanes A to J), whereas a significantfraction of the unprocessed precursor remained stable (Fig.2B, lanes A to M). Consistent with these data, surface expressionof the passenger region was detected via indirect immunofluorescencein mutants A through J, but not in mutants K, L, and M (Fig.2C). The absence of immunofluorescence in mutants K, L, andM supports the tenet that the unprocessed, trypsin-resistantfraction of BrkA represents an intracellular form of BrkA, andnot simply a trypsin-resistant surface molecule. It should benoted that a deletion ( ${Delta}$ Ala¹³⁶-Pro²⁵⁵) within the BrkA passenger(Met¹ to Gly²²⁸) construct used for mutants A to M did not affectsurface expression of BrkA (data not shown). Collectively, thesedata show that the region spanning residues Ala¹³⁶ to Glu⁶⁹³of BrkA is not required for surface localization of passengerproteins in E. coli strain UT5600. Furthermore, since the processedform of the passenger is also evident in deletion constructsA to J (Fig. 2) as well as construct ${Delta}$ Ala¹³⁶-Pro²⁵⁵, it arguesagainst BrkA having autoproteolytic activity.

Our data indicate that the ß-domain of BrkA is itselfinsufficient to translocate a passenger to the cell surface.The minimal translocation unit for BrkA thus consists of theß-core plus a preceding linker region, the N-terminalboundary of which maps within Glu⁶⁹³ to Ser⁷⁰¹. Historically,the ß-domain has been defined as the C-terminal outermembrane resident fragment derived from proteolytic processingof the autotransporter protein. Although the ß-domainsof IgA protease (11), VirG/IcsA (31), and AIDA-1 (19) are largerthan the ß-domain of BrkA, the sizes of their minimaltranslocation units are remarkably similar. Indeed, a comparisonof experimentally defined linkers in four diverse autotransportersincluding BrkA reveals a structurally conserved architecturethat can be viewed as a signature for autotransporters. It consistsof a 21- to 30-amino-acid ${alpha}$ -helical region that precedes a 255-to 294-amino-acid transporter domain, a region rich in ßstructure (Fig. 3). It has been proposed that the linker regionis involved in forming a hairpin-like structure that leads secretionof the passenger domain through the channel formed by the ß-core(9). The common features of the translocation unit suggest thatit, rather than the ß-domain, is a more appropriateoperational definition for the transporter domain. The regionupstream of the translocation unit would thus constitute thepassenger moiety regardless of the positioning of the proteolyticprocessing sites (Fig. 3).

View larger version (30K):
[in this window]
[in a new window]
FIG. 3. Comparison of the translocation units of different autotransporters. The C-terminal regions of four autotransporters are shown (not drawn to scale). See the text for explanation. The N-terminal boundaries noted for each translocation unit have been defined experimentally in references 11 (IgA protease), 31 (VirG/IcsA), and 19 (AIDA-1), as well as in this paper (BrkA).

IgA protease, VirG/IcsA, and AIDA-1 are naturally cleaved wellupstream of the predicted ${alpha}$ -helical region (Fig. 3) and eithercan be released naturally (6, 26) or can be induced to be releasedfollowing heat treatment (1). Unlike these proteins, BrkA issteadfastly associated with the bacterial outer membrane bothin B. pertussis and in E. coli and cannot be released by heattreatment (G. Huang and R. Fernandez, unpublished data). Cleavageof BrkA occurs within the predicted ${alpha}$ -helical region. Thus, itis possible that the linker region also acts as the anchor (11)for BrkA, since none of the deletion mutant proteins spanningAla¹³⁶ to Glu⁶⁹³ was detected in immunoblots of concentratedculture supernatants (data not shown).

In summary, we have shown that the B. pertussis autotransporterBrkA can be surface expressed in E. coli, enabling dissectionof autotransporter secretion mechanisms in a host more amenableto genetic manipulation. Adding to our previous studies on theBrkA ß-domain (28), which we demonstrated has thecapacity to form a channel, we have identified two additionalsecretion determinants of BrkA: a 42-amino-acid signal peptideand a 30- to 39-amino-acid region preceding the ß-domainthat, together with the ß-domain, defines the BrkAtranslocation unit (Fig. 3). The data presented provide furtherexperimental support for the importance of the predicted ${alpha}$ -helicalregion in autotransporter secretion of both native and heterologouspassengers (12, 19, 31).

ACKNOWLEDGMENTS

We thank V. deLorenzo and L. Fernandez for the gift of UT5600and UT2300. D.C.O. was a recipient of a University of BritishColumbia graduate student fellowship.

This work was funded by a grant from the Natural Sciences andEngineering Research Council of Canada.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Immunology, University of British Columbia, #300-6174 University Blvd., Vancouver, British Columbia, Canada V6T 1Z3. Phone: 604-822-6824. Fax: 604-822-6041. E-mail: rachelf{at}interchange.ubc.ca.

REFERENCES

Top