Vol. 176, No. 17JOURNAL OF BACTERIOLOGY, Sept. 1994, p. 5350-53560021-9193/94/$04.00+0Copyright © 1994, American Society for Microbiology

Detection and Subcellular Localization of Three Ptl Proteins Involvedin the Secretion of Pertussis Toxin from Bordetella pertussis

FREDERICK D. JOHNSON AND DRUSILLA L. BURNS*

Division of Bacterial Products, Center for Biologics Evaluation and Research,Food and Drug Administration, Bethesda, Maryland 20892

Received 3 February 1994/Accepted 20 June 1994

Theptl locus ofBordetella peltussis contains eight open reading frames which are predicted to encode proteins(PtlA to PtlH) that are essential for secretion of pertussis toxin from the bacterium and which are membersof a family of transport proteins found in other types of bacteria. We have detected PtlE, PtlF, and PtlG inimmunoblots of extracts of B. pertussis by using antibodies raised to fusion proteins consisting of maltose-binding protein and the individual Ptl proteins. These proteins have apparent molecular weights similar tothose predicted by DNA sequence analysis. Cell fractionation studies indicated that all three tl proteins areassociated with the membranes of B. pertussis, suggesting that the Ptl proteins form a gate or channel whichfacilitates transport of pertussis toxin. Cell extracts of other BordeteUa spp. were probed with antibodies to Ptlproteins for the presence of these transport proteins. Neither Bordetella parapertussis nor Bordetella bronchi-septica contained detectable levels of PtlE or PtlF. This lack of detectable Ptl protein may provide anexplanation for previous observations which indicated that introduction of the genes encoding pertussis toxinsubunits from B. pertussis into other Bordetella spp. results in production of the toxin but not secretion of thetoxin.

Export of virulence factors such as toxins, adhesins, andinvasins from pathogenic bacteria is often critical to the successof the pathogen, since exposure of these virulence factorsoptimizes their interaction with host cells. Bordetella pertussis,the causative agent of pertussis (whooping cough), secretesseveral virulence factors, one of which is pertussis toxin (PT),a toxin that may contribute to pathogenesis by impairing theimmune system of the host (11). PT is an important virulencefactor of B. pertussis, since mutant strains of B. pertussis whichdo not produce PT are less virulent in animal models than thewild-type strain (36). A strain of B. pertussis which produces PTbut which is defective in the secretion of the toxin is also lessvirulent than the wild-type strain in animal models (35),suggesting that secretion of the toxin, not simply production ofPT, is important for pathogenesis.The question arises as to how PT might be transported

across the inner and outer membranes ofB. pertussis. The toxinis a large, complex protein composed of six subunits (Si, S2,S3, S4, and S5 found in a 1:1:1:2:1 ratio) and has a molecularweight of 105,000 (21, 27). Each subunit is synthesized with asignal sequence (18, 21), suggesting that the subunits may betransported into the periplasmic space via a general exportpathway analogous to the sec system of Escherichia coli (39).Others have reported that the S1 subunit is required forefficient secretion of PT (23), suggesting that an assembledform of the molecule must be transported across at least onemembrane of the bacterium.

Recently, we defined a region of the B. pertussis chromo-some, termed the ptl locus, which is critical for efficientsecretion of PT (37). This region, which spans approximately9.5 kb, contains eight open reading frames (ORFs), termedorfA to orfH. Evidence suggests that the ptl genes are locatedwithin a single operon (37). Each of the eight ORFs is

* Corresponding author. Mailing address: FDA/CBER/DBP, HFM-434, 1401 Rockville Pk., Rockville, MD 20852-1448. Phone: (301)402-3553. Fax: (301) 402-2776.

predicted to encode a protein homologous to one of the VirBproteins produced by Agrobacterium tumefaciens (6, 37). TheVirB proteins are thought to be involved in the transport of apiece of single-stranded DNA, T-DNA, across bacterial mem-branes (13, 31, 32, 40). These homologies, together with thedemonstration that mutations in the ptl locus affect PT secre-tion (37), suggest that theptl genes encode proteins needed fortransport of PT across the bacterial membrane(s).While the existence of Ptl proteins is predicted, they have

not as yet been detected. In this report, we describe theidentification of three of the Ptl proteins, PtlE, PtlF, and PtlG,and their cellular localization. In addition, we used antibodiesto the Ptl proteins to probe extracts of related B; pertussis,Bordetella parapertussis and Bordetella bronchiseptica strainswhich are known to be defective in the secretion of PT fordetection of the presence of Ptl proteins, and we offer anexplanation for the lack of secretion of PT in these strains.

MATERLALS AND METHODS

Bacterial strains, plasmids, and growth conditions. B. per-tussis BP338, BP356, BPPTL2, BPPTL3, BPPTL4, andBPM3171 were obtained from Alison Weiss, University ofCincinnati. B. pertussis Tohama III was from the collection ofthe Laboratory of Pertussis, Center for Biologics Evaluationand Research, Food and Drug Administration, Bethesda, Md.B. pertussis 18323 (ATCC 9797), B. parapertussis 9305, and B.bronchiseptica BbSS (ATCC 31437) were from the AmericanType Culture Collection (Rockville, Md.). These strains weregrown at 37°C on plates of Bordet-Gengou medium containing1.25% glycerol and 15% defibrinated sheep blood (UniversityMicro Reference Laboratory, Inc., Linthicum, Md.) either withor without 1.5% proteose peptone. The following antibioticswere added to the medium for the indicated strains: gentami-cin, 10 ,ug/ml (strains BPPTL2, BPPTL3-2, and BPPTL4);streptomycin, 150 jig/ml (strain BPPTL3-2); and kanamycin,25 ,ug/ml (strain BPM3171). Modulated B. pertussis cells were

5350

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION AND LOCALIZATION OF Ptl PROTEINS 5351

harvested from media supplemented either with 50 mMMgSO4 or with 10 mM nicotinic acid and 20 mM MgSO4.

Library efficiency-competent cells, E. coli DH5ot, were ob-tained from GIBCO BRL (Gaithersburg, Md.), and the plas-mid pMAL-c2 was purchased from New England Biolabs(Beverly, Mass.). Cells were grown on Luria-Bertani (LB) agar(Difco Laboratories, Detroit, Mich.) or in liquid culture con-taining SOC medium (GIBCO BRL) or LB broth supple-mented where necessary with ampicillin (100 ,ug/ml), 5-bromo-4-chloro-3-indolyl-13-D-galactopyranoside (X-Gal; 50 p.g/ml),and/or isopropyl-,-D-thiogalactopyranoside (IPTG).PCR. Reagents for DNA amplification were obtained from

GIBCO BRL. The oligonucleotide primers used were obtainedfrom Lofstrand Labs Limited (Gaithersburg, Md.). For eachPCR, the upstream primer contained three nucleotides fol-lowed by an XbaI site, the ATG initiation codon, and the next19 nucleotides of the ORF at its 5' end. The downstreamprimer (5'->3') contained three nucleotides followed by aHindlll site, the termination codon and the last 19 nucleotidesof the ORF. Two templates were used for PCRs. A plasmid(pUW1096C) obtained from Alison Weiss, University of Cin-cinnati, which contained nucleotides 4691 to 10840 of the ptllocus was used for amplification of ORFs E and F. B. pertussisTohama I DNA, kindly provided by Z. M. Li (Food and DrugAdministration), was used for amplification of orfG. EachPCR mixture contained the two oligonucleotide primers (0.25,uM), template (either 30 ng of Tohama I DNA or 167 ng ofpUW1096C), deoxynucleotides (each at 1 mM) and AmpliTaqDNA polymerase (2.5 U; Perkin-Elmer Corp., Norwalk,Conn.) in 10 mM Tris-HCl, pH 8.3, containing 50 mM KCl.The reaction mixture also contained MgCl2 and glycerol atfinal concentrations of 2.5 mM and 20%, respectively. Ampli-fied DNA was purified with the Magic PCR Preps DNApurification system (Promega, Madison, Wis.).

Production and purification of fusion proteins. The ProteinFusion and Purification System (New England Biolabs) wasused to generate and purify fusion proteins composed ofmaltose-binding protein (MBP) and proteins encoded by ptlORFs. DNA manipulations and subsequent procedures wereperformed by following the manufacturer's instructions andusing standard procedures (25). Briefly, DNA which had beenamplified by PCR and which corresponded to each ptl ORFwas inserted into pMAL-c2 vectors that had been digested withXbaI and HindlIl. The resulting plasmids were introduced intoE. coli DH5ot. Cells were grown to approximately 2 x 108 cellsper ml (A600, -0.5) when production of fusion proteinsconsisting of MBP and proteins encoded by orfE, orfF, or orfGwas induced by the addition of IPTG (0.3 or 1 mM) for 2 h.Cells were harvested, suspended in 50 ml of 10 mM Tris-HCl,pH 7.4, containing 200 mM NaCl and 1 mM EDTA (columnbuffer), and sonicated, and the resulting suspension was cen-trifuged at 9,000 x g for 30 min. The supernatant was appliedto a column containing amylose resin (20 ml) according tomanufacturer's instructions. The column was washed with 8column volumes of column buffer. The fusion protein waseluted with column buffer containing maltose (10 mM). Fusionproteins were dialyzed extensively against phosphate-bufferedsaline (PBS) at 4°C and were stored frozen.

Protein determination. Protein concentration was deter-mined by the method of Bradford (4), with ovalbumin as thestandard.

Gel electrophoresis. Sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE) was performed essentially asdescribed by Laemmli (15), with 4 to 20% gradient polyacryl-amide gels (Integrated Separation Systems, Natick, Mass.).Proteins were stained with Coomassie brilliant blue R250.

Polyclonal antibody production and monoclonal antibodies.MBP-ORF fusion protein was added to PBS containing gelatin(0.05%) to give a final concentration of 60 pg/ml. Alhydrogel[Al(OH)3] (Superfos a/s, Vedbaek, Denmark) was added tothe mixture to give a 1:1 ratio (on a weight basis) of Alhydrogelto protein (fusion protein plus gelatin). Preparations wereincubated for 1 h at room temperature with occasional stirring.

Five female NIH(S) mice were each injected intraperitone-ally with 0.5 ml of the preparation containing 30 pg of fusionprotein, MBP, or the Alhydrogel solution containing only0.05% gelatin. Two and 4 weeks after the initial injection, micewere given a second and third 0.5-ml injection. Two weekslater, blood was collected from the tail vein of each mouse.Sera were prepared, pooled, and stored at -20°C.Monoclonal antibody 3CX4 which recognizes the S1 subunit

of pertussis toxin (12) was obtained from James Kenimer,Food and Drug Administration. Monoclonal antibody M08-X3C which recognizes filamentous hemagglutinin (FHA) (17)was obtained from Michael Brennan, Food and Drug Admin-istration.Immunoblot analysis. Cell extracts and fractions were sub-

jected to SDS-PAGE and electrophoretically transferred tonitrocellulose membranes overnight at 30 V in 25 mM Tris-192mM glycine with 20% methanol. Nitrocellulose sheets wereincubated for 1 h with 10% (wt/vol) skim milk in distilledwater. Nitrocellulose strips were washed three times with PBS,pH 7.0, containing 0.05% Tween 20 (PBS-Tween) and wereincubated for 2 h with sera containing antibodies to fusionproteins (used at a 1:2,000 or 1:4,000 dilution). After threefurther washes with PBS-Tween, the strips were incubated for1 h with goat anti-mouse immunoglobulin G conjugated tohorseradish peroxidase (Bio-Rad Laboratories, Richmond,Calif.) diluted 1:1,000 with PBS-Tween. The nitrocellulosestrips were washed with PBS, and the bands were visualized byadding the color reagent (15 mg of 4-chloro-1-naphthol in 5 mlof methanol combined with 25 ml of PBS containing 0.018%hydrogen peroxide).

Preparation of cell extracts and fractionation of extracts.Bordetella cells grown on BG agar were suspended in PBS togive an A550 of 2, centrifuged at 12,000 x g for 10 min, andsuspended in PBS to the same optical density. The suspension(7 pl) was mixed with SDS-PAGE sample buffer (13 RI of asolution containing 2% SDS, 20% glycerol, 200 mM Tris base,100 mM dithiothreitol, 0.01% bromphenol blue). In certainexperiments, as indicated, 20 ,u of the suspension was mixedwith an equal volume of 20% trichloroacetic acid. Aftercentrifugation, the precipitate was suspended in SDS-PAGEsample buffer. All samples were boiled for 2 min and thensubjected to gel electrophoresis.

Cell fractionation was performed essentially as describedpreviously (26). Briefly, cells grown on plates containing BGmedium were harvested from the plates and suspended in PBSto an A550 of 2. This mixture (10 ml) was frozen in a dryice-ethanol bath, thawed in cold water, and sonicated (modelW375; Heat Systems Ultrasonic Inc.) on 50% pulse mode andpower setting 3 for 5 min on ice. The mixture was centrifugedfor 10 min at 2,000 x g to remove unbroken cells. Supernatant(5 ml) was centrifuged for 1 h at 107,000 x g. The resultingsupernatant, which contained cytosolic and periplasmic pro-teins, was collected. The pellet containing the membranefraction was suspended in 5 ml of PBS. Portions (500 [lI) ofboth the supernatant and pellet fractions were precipitated inan equal volume of 20% trichloroacetic acid. The precipitateswere collected after centrifugation and suspended in 20 tlA ofSDS gel sample buffer and subjected to electrophoresis.

Detergent solubilization of the membrane fraction was

VOL. 176, 1994

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

5352 JOHNSON AND BURNS

Ial--|-- f '

BPPTL3-2 BPM3171 BPPTL2 BPPTL4

FIG. 1. The ptl locus. The structure of the ptl locus is shown alongwith the locations of mutations in the ptl locus of the indicated strainsof B. pertussis.

conducted essentially as described previously (26). The pelletobtained after centrifugation of the sonicated cell extract at107,000 x g for 1 h (see above) was suspended in 2.5 ml of N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES)buffer, pH 7.4, containing 2% Triton X-100 and 7.5 mMMgCl2. The preparation was incubated at room temperaturefor 30 min and was then centrifuged at 107,000 x g for 1 h. Thepellet (Triton X-100-insoluble fraction) was suspended in 2.5ml of PBS. SM-2 Bio-beads (0.7 g, Bio-Rad) were added to theTriton X-100-soluble fractions to remove detergent (9) thatinterfered with resolution of proteins during the subsequentelectrophoresis procedure. The mixture was agitated overnightat 4°C. The sample was centrifuged for 10 min at approxi-mately 2,000 x g, and the supernatant was collected. TheTriton X-100-insoluble fraction (125 ,ul) and Triton X-100-soluble fraction (400 ,ul) were each added to equal volumes of20% trichloroacetic acid. After centrifugation, the precipitateswere solubilized in 20 ,lA of SDS gel sample buffer andsubjected to electrophoresis. Similar results were obtainedwhen the Triton X-100-soluble fraction was not treated withthe Bio-beads to remove the detergent (data not shown).

RESULTS

Identification of Ptl proteins. As shown in Fig. 1, the ptllocus contains eight ORFs. Antibodies to proteins predicted tobe encoded by orfE, orfF, and orfG were made by thefollowing method. DNA corresponding to each of the ORFswas amplified using PCR. These fragments were then eachinserted into pMAL-c2 vectors. The resulting recombinantplasmids were each transformed into E. coli, and fusionproteins consisting of the MBP and the predicted Ptl proteinwere produced. Fusion proteins were purified by affinity chro-matography with amylose resin. Each of the fusion proteinpreparations was then used to immunize mice to generateantibodies to the Ptl proteins.

Sera from mice immunized with the MBP-orfE fusionprotein were used to probe extracts of virulent B. pertussis forthe presence of PtlE. As shown in Fig. 2, immunoblot analysisof extracts of BP338 (wild-type, virulent strain) revealed thatthe MBP-orfE antiserum bound to several proteins including aprotein of apparent molecular weight of 30,000. The predictedmolecular weight of PtlE obtained by nucleotide sequenceanalysis of orfE is 31,010. In order to determine whether any ofthese proteins might be PtlE, we also examined extracts ofBP338 which had been grown in the presence of MgSO4.Growth of B. pertussis in the presence of MgSO4 (known asmodulating conditions) reversibly suppresses expression ofgenes encoding virulence factors which are under control ofproducts of the bvg (Bordetella virulence genes) locus (10, 14,38). Previously, ptl genes have been shown to be bvg regulated(38) and therefore would not be expected to be expressed in anavirulent form of the organism or in a virulent form of theorganism grown under modulating conditions. The proteinhaving an apparent molecular weight of 30,000 was not ob-

1 2 3 4 5 6FIG. 2. Immunoblot analysis of cell lysates probed with anti-MBP-

orfE. Cell extracts from strains BP338 (lane 1), modulated BP338 (lane2), BPPTL3-2 (lane 3), BPM3171 (lane 4), BPPTL2 (lane 5), andBPPTLA (lane 6) were subjected to SDS-PAGE and transferred tonitrocellulose. Anti-MBP-orfE was used to probe the extracts for thepresence of PtlE as described in Materials and Methods. Arrows pointto a band having an apparent molecular weight of 30,000.

served in this extract. In order to confirm the identity of the30,000-Da species as PtlE, we examined cell extracts of strainsof B. pertussis which have mutations in the ptl locus. Thelocations of these mutations are shown in Fig. 1. StrainBPPTL3-2 was constructed by deletion of a 437-bp BamHIfragment in orfC and replacement of this fragment with a4.0-kbp BamHI cassette containing the gentamicin resistancegene and the P-incompatibility origin-of-transfer region (37)and therefore would not be expected to produce PtlE. StrainBPM3171 was generated by insertion of a TnSlac transposoninto orfC (37, 38) and thus would not be expected to producePtlE. Strains BPPTL2 and BPPTL4 were generated by homol-ogous recombination of a plasmid containing portions of theptl locus with the chromosome such that the vector wouldinterrupt the chromosome in orfG and orfH, respectively (37).These mutants would be expected to produce PtlE since themutations occur downstream fromptlE. The 30,000-Da proteinwas not observed in extracts of BPPTL3-2 or BPM3171 but wasobserved in extracts of BPPTL2 and BPPTL4, as would beexpected if the 30,000-Da protein were PtlE. Sera from controlmice which had been injected with either MBP or Alhydrogelalone did not bind to the 30,000-Da protein (data not shown).

Cell extracts from the wild-type and mutant strains of B.pertussis were also probed with sera from mice immunized withthe MBP-orfF fusion protein. As shown in Fig. 3, a proteinwith an apparent molecular weight of 31,000 was seen in cellextracts from the virulent strain BP338 but not in the modu-lated strain. This protein was seen in cell extracts of BPPTL2and BPPTL4 but not in extracts of BPPTL3-2 or BPM3171,which would be the expected observation if the 31,000-Daprotein were PtIF. Nucleotide sequence analysis of orfF pre-dicts that the mature form of PtlF would have a molecularweight of 27,541 (37).Serum from mice immunized with MBP-orfG fusion protein

was used to probe B. pertussis cell extracts in order to identifyPtlG. As shown in Fig. 4, this serum bound to a number ofproteins in cell extracts from all of the strains. The reason forthis high background binding is unknown. Nonetheless, aprotein band with an apparent molecular weight of 38,000 wasobserved in extracts from the virulent form of BP338 andBPPTL4 but not in extracts of an avirulent strain (Tohama III)or BPPTL3-2, BPM3171, or BPPTL2. (It should be noted thatthe band having an apparent molecular weight of 38,000migrated only slightly above a second band of almost identicalmolecular weight such that the two proteins appeared as a

30-

J. BACTERIOL.

-01.

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION AND LOCALIZATION OF Ptl PROTEINS 5353

PtIE PtIF

4-- 4-

PtIG

4-

31- -ll

1 2 3 4 5 6FIG. 3. Immunoblot analysis of cell lysates probed with anti-MBP-

orfF. Cell extracts from strains BP338 (lane 1), modulated BP338 (lane2), BPPTL3-2 (lane 3), BPM3171 (lane 4), BPPTL2 (lane 5), andBPPTL4 (lane 6) were subjected to SDS-PAGE and transferred tonitrocellulose. Anti-MBP-orfF was used to probe the extracts for thepresence of PtlF as described in Materials and Methods. Arrows pointto a band having an apparent molecular weight of 31,000.

doublet. The lower band was present in extracts from allstrains.) BPPTL3-2 and BPM3171 would not be expected toproduce PtlG, whereas BPPTL2 would be predicted to pro-duce a truncated form of this protein which may or may not bestable. A smaller form of the protein was not observed inextracts of BPPTL2. A second band which migrated as a46,000-Da protein was present in the same strains as was the38,000-Da protein and may represent a larger form of PtlG.Two forms of VirB10, the PtlG homolog in A. tumefaciens,have been described (33).Attempts were made to identify other predicted Ptl proteins

(PtlA, PtlB, PtlC, PtlD, and Pt1H) by similar methods. We wereunable to isolate a fusion protein consisting of MBP and orfD,perhaps because of the instability of the fusion protein. Whilewe were able to obtain fusion proteins for MBP-orfA, MBP-orfB, MBP-orfC, and MBP-orfH and were able to obtainantibodies to the fusion proteins, we did not detect thepredicted proteins in extracts of B. pertussis. A variety ofexplanations may account for these findings, including thepossibility that the fusion proteins induce very little antibodythat recognizes the Ptl proteins or that Ptl proteins are present

1 2 3 4 1 2 3 4 1 2 3 4

FIG. 5. Immunoblot analysis of cell fractions of B. pertussis 338.Cell fractions were prepared as described in Materials and Methods.Immunoblots of fractions were probed with anti-MBP-orfE, anti-MBP-orfF, and anti-MBP-orfG to detect PtlE, PtlF, and PtlG, respec-tively, as indicated. Lane 1, soluble fraction (cytoplasm and periplasm)from unmodulated cells; lane 2, soluble fraction from modulated cells;lane 3, particulate (membrane) fraction from unmodulated cells; lane4, particulate fraction from modulated cells. Arrows mark positions ofbands corresponding to PtlE, PtlF, and PtlG.

in minute quantities in the cell and are therefore difficult todetect.

Localization of PtIE, PtlF, and PtlG in B. pertussis. B.pertussis cells were fractionated into soluble and particulatefractions representative of the cytosolic or periplasmic com-partments and membranes, respectively. As shown in Fig. 5,PtlE, PtlF, and PtlG were found in the particulate fraction.

Differential solubilization of bacterial membrane proteins inTriton X-100 has been used as an indicator of whether aprotein is located in the inner or the outer membrane, sinceTriton X-100 is believed to preferentially solubilize innermembrane proteins of B. pertussis (3, 7, 26). Other methodssuch as isopycnic centrifugation do not separate B. pertussismembranes into distinguishable fractions corresponding toinner and outer membranes (7) and therefore would not beuseful in this regard. The total membrane fraction was treatedwith 2% Triton X-100, and the detergent soluble proteins wereseparated from insoluble proteins by centrifugation. PtlF andPtlG were insoluble in Triton X-100 (Fig. 6). PtlE was partiallysoluble in the detergent.

PtIE

46-38- -.

1 2 3 4 5 6FIG. 4. Immunoblot analysis of cell lysates probed with anti-MBP-

orfG. Cell extracts from strains BP338 (lane 1), Tohama III (lane 2),BPPTL3-2 (lane 3), BPM3171 (lane 4), BPPTL2 (lane 5), and BPPTL4(lane 6) were subjected to SDS-PAGE and transferred to nitrocellu-lose. Anti-MBP-orfG was used to probe the extracts for the presenceof PtlG as described in Materials and Methods. Arrows point to bandshaving apparent molecular weights of 46,000 and 38,000.

1 2 3 4

PtIF

1 2 3 4

PtIG

1 2 3 4

FIG. 6. Detergent solubility of Ptl proteins. Membrane fractions ofBP338 were extracted with 2% Triton X-100 as described in Materialsand Methods. Immunoblots of the resulting fractions were probed withanti-MBP-orfE, anti-MBP-orfF, and anti-MBP-orfG as indicated.Lane 1, Triton X-100-insoluble fraction from unmodulated cells; lane2, Triton X-100-soluble fraction from modulated cells; lane 3, TritonX-100-soluble fraction from unmodulated cells; lane 4, Triton X-100-soluble fraction from modulated cells. Arrows point to bands corre-sponding to PtlE, PtlF, and PtlG.

VOL. 176, 1994

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

5354 JOHNSON AND BURNS

A

PtIF -.

B

} FHA

- Si

1 2 3 1 2 3

FIG. 7. Immunoblot analysis of extracts of B. pertussis BP356probed with anti-MBP-orfF. Cell extracts (20 ,ul) from BP338 (lane 1),modulated BP338 (lane 2), and BP356 (lane 3) were subjected toelectrophoresis and transferred to nitrocellulose. Anti-MBP-orfF was

used to probe the extracts for the presence of PtlF (A). Monoclonalantibodies M08-X3C and 3CX4 were used to probe the extracts forFHA and the S1 subunit of PT, respectively (B).

Examination of related B. pertussis strains and other Borde-tella spp. for the production of Ptl proteins. B. pertussis BP356contains a TnS insertion within the gene coding for subunit S3of PT (34). Others have reported that this strain made PTsubunits S1, S2, and S4, and S5, but the subunits were notsecreted, and they concluded, therefore, that S3 was necessaryfor secretion of PT (19, 22). Since the Ptl genes are locateddirectly downstream from the Tn5 element, we wanted todetermine whether the transposon affected production of Ptlproteins in this strain. Using antibodies which recognize PtlEand PtlF, we probed this strain for the presence of these Ptlproteins. BP356 did not contain detectable amounts of PtIF(Fig. 7) or PtlE (data not shown), although this strain producednormal amounts of FHA, another bvg-regulated protein. FHAis readily degraded and therefore appears as one major band aswell as several breakdown products. BP356 also containsdetectable levels of the S1 subunit of PT (Fig. 7).

Other Bordetella spp. have been reported to contain regionshomologous to theptl genes (2). Extracts of B. parapertussis, B.bronchiseptica, and B. pertussis 18323, a pertussis strain whichhas been reported to be an intermediate in the evolution fromB. pertussis to B. parapertussis and B. bronchiseptica (1), wereprobed for protein which binds to antibodies which recognizePtlF. While B. pertussis 18323 produced PtlF, B. parapertussis9305 and B. bronchiseptica Bb55 did not produce a protein thatreacted with PtlF antibodies (Fig. 8). A few faint bands havingapparent molecular weights different from that of PtlF wereobserved in extracts from B. bronchiseptica; however, all ofthese bands were also present in extracts from the modulatedform of the organism. Also, while B. pertussis 18323 producedPtlE, extracts from B. parapertussis 9305 and B. bronchisepticaBb55 did not contain a protein that reacted with antibodiesthat recognize PtlE (data not shown).

DISCUSSION

The existence of the Ptl proteins is predicted from nucle-otide sequence analysis of a region of the B. pertussis chromo-some which is necessary for efficient secretion of PT from thebacterium (6, 37). Using antibodies raised to fusion proteinscontaining the predicted primary sequence of PtlE, PtlF, andPtlG, we have detected these Ptl proteins in extracts of B.pertussis; this constitutes the first identification of any of the Ptlproteins.

PtIF--_

1 2 3 4 5 6FIG. 8. Immunoblot analysis of extracts of B. pentussis, B. paraper-

tussis, and B. bronchiseptica probed with anti-MBP-orfF. Cell extracts(20 ,ul) from B. pertussis BP338 (lane 1), modulated B. pertussis BP338(lane 2), B. pertussis 18323 (lane 3), B. parapertussis 9305 (lane 4), B.bronchiseptica Bb55 (lane 5), and modulated B. bronchiseptica Bb55(lane 6) were subjected to SDS-PAGE and transferred to nitrocellu-lose. Anti-MBP-orfF was used to probe the extracts for the presence ofPtlF as described in Materials and Methods.

PtlE has an apparent molecular weight on SDS-polyacryl-amide gels of 30,000, which is in good agreement with amolecular weight of 31,010 predicted from the DNA sequence(37). PtlE cofractionates with membranes of B. pertussis ex-tracts. This protein is partially soluble in Triton X-100, sug-gesting that PtlE has an inner membrane location but mighthave some association with components of the outer mem-brane. An inner membrane location for this protein is consis-tent with its amino acid sequence, which predicts that theprotein has no signal sequence which would target the proteinto the outer membrane. PtlE exhibits 32% identity with theVirB8 protein of A. tumefaciens over a stretch of 180 aminoacids (37). VirB8 has been reported to be an inner membraneprotein that is likely associated with outer membrane compo-nents (28, 29).

PtlF has an apparent molecular weight on SDS-polyacryl-amide gels of 31,000. The predicted molecular weight of theprotein encoded by the entire orfF is 29,471 (37); however, thepredicted protein has an N-terminal sequence, MMAARMMAAGLAATALSAHA, which has the characteristics of asignal sequence (5). The mature form of PtlF would thereforebe predicted to have a molecular weight of 27,541. PtlFcofractionates with the membranes of B. pertussis extracts andis insoluble in Triton X-100, suggesting that it is an outermembrane protein or associated with components of the outermembrane. PtlF exhibits 27% identity with VirB9 of A. tume-faciens over a stretch of 252 amino acids (37). VirB9 also hasa signal sequence and is found in both inner and outermembrane fractions (28). Others have postulated that VirB9 isactually an inner membrane protein that is associated withcomponents of the outer membrane, causing it to fractionate inboth inner and outer membrane fractions (28).

PtlG, which migrates with an apparent molecular weight of38,000, exhibits 34% identity with VirBlO over a 270-amino-acid stretch (37). PtlG cofractionates with the membranes of B.pertussis extracts and is insoluble in Triton X-100. VirBlO wasfound in the inner membrane fraction of A. tumefaciens (33)but is thought to be tightly associated with outer membranecomponents, causing it to fractionate with the outer membraneif certain fractionation methods, including differential solubil-ity in Sarkosyl, are used (28). Since PtlG has no obvious signal

J. BAC7ERIOL.

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

DETECTION AND LOCALIZATION OF Ptl PROTEINS 5355

sequence, it may also be an inner membrane protein thatassociates tightly with components of the outer membrane.We have used antibodies to Ptl proteins to examine a strain

of B. pertussis which has previously been shown to be defectivein the secretion of PT subunits (19, 22). We found that BP356,which contains a TnS insertion in the structural gene forsubunit S3 of PT (19, 22), does not produce detectable levels ofPtlE or PtlF. Thus, the presence of the TnS insertion in the S3gene appears to affect expression of at least certain of the ptlgenes located directly downstream of the insertion, in additionto its effect on production of the S3 subunit that has beenreported (19, 22). Previously, others concluded that S3 is likelyessential for export of PT, since other PT subunits are pro-

duced by BP356 but they are not secreted (19, 22). However,this conclusion may need to be reexamined given our findingthat production of at least certain of the Ptl proteins is also

affected by the transposon. Further investigation is needed todetermine why the transposon affects expression of the ptlgenes, since it has been suggested that ptx genes which encodethe PT subunits and ptl genes are parts of different operons

(37).The Bordetella genus contains four species, B. pertussis, B.

parapertussis, B. bronchiseptica, and Bordetella avium. While B.pertussis is the only one of these species that produces PT (2),all strains of B. parapertussis and the majority of B. bronchisep-tica strains contain the ptx genes (2, 20). In addition to the ptxgenes, these closely related species also contain regions homol-ogous to the entire ptl region (2). Arico and Rappuoli (2)sequenced the ptx region and portions of the ptl region(includingptU4,ptlB, and part ofptlC) from B. parapertussis andB. bronchiseptica and found that these regions have 98.6 and96% identity, respectively, to the ptx-ptl regions of B. pertussis.At the amino acid level, the DNA sequence predicts that thesePtl proteins from B. parapertussis and B. bronchiseptica wouldhave an average of 99 and 95% identity, respectively, to thepredicted amino acid sequence of the corresponding Ptl pro-teins from B. pertussis. Using antibodies that recognize the B.pertussis Ptl proteins, we examined B. parapertussis and B.bronchiseptica extracts for the presence of PtlE and PtlF.Neither species produced detectable levels of these proteins.While it is possible that the lack of detection of these Ptlproteins in B. parapertussis and B. bronchiseptica is due to lackof cross-reactivity of the antibodies, this explanation seemsunlikely given the high level of homology that has beendescribed between the ptx-ptl genes of Bordetella spp. (2). TheB. parapertussis strain that we used in our study (strain 9305) isknown to contain the entireptl region (2). Therefore, while theptl genes are present, they are likely not expressed, a findingsimilar to that observed for theptx genes ofB. parapertussis andB. bronchiseptica.

In contrast to B. parapertussis and B. bronchiseptica strains,B. pertussis 18323, a pertussis strain that has been described as

an intermediate in the evolution of B. parapertussis and B.bronchiseptica from B. pertussis (1), does produce detectablelevels of PtlE and PtlF. Since B. pertussis 18323 produces PT(1), this organism would need the Ptl proteins for secretion ofthe toxin. In contrast, B. parapertussis and B. bronchiseptica donot produce PT (2) and therefore would not have a need forthe Ptl proteins if these transport proteins were used exclu-sively for secretion of PT and not for export of other proteinscommon to the Bordetella spp.

Previously, others introduced the ptx genes from B. pertussisinto strains of B. parapertussis and B. bronchiseptica, includingB. bronchiseptica Bb55, which we have used in this study, in anattempt to produce PT in other Bordetella spp. They attemptedto produce PT in other bordetellae, since B. pertussis is a slowly

growing organism that is not ideal for production of theamounts of PT needed in the manufacture of vaccine compo-nents for acellular pertussis vaccines (16, 30). However, theirresults indicated that while PT can be produced in theseorganisms, the toxin is not secreted and would therefore bedifficult to purify. Our lack of ability to detect PtlE and PtlF inB. parapernussis and B. bronchiseptica suggests that, at aminimum, both the ptx and ptl genes from B. pertussis need tobe introduced into these Bordetella spp. before secretion of PTcould be achieved.

In this report, we have shown that PtlE, PtlF, and PtlG canbe detected in B. pertussis but not in other Bordetella spp. Allthree Ptl proteins appear to be localized in the bacterialmembrane. A more precise localization within the bacterialenvelope (i.e., inner versus outer membrane) is difficult tomake because of technical problems associated with separationof the inner and outer membranes of B. pertussis. The mem-brane localization of Ptl proteins suggests that these proteinsmight form a gate or channel used in the transport process.The VirB proteins of A. tumefaciens have been postulated toform a channel that spans both the inner and the outermembranes of the bacterium, which may open and close inresponse to the transportable substrate (28). Transport of theT-DNA has been proposed to occur by a one-step mechanismin which the T complex crosses the membrane barriers througha channel that spans both membranes. PT might be trans-ported across the two bacterial membranes by a single-stepmechanism analogous to that proposed for T-DNA. Alterna-tively, transport of PT might occur by a two-step mechanism inwhich the individual subunits are first transported across theinner membrane, perhaps in an unfolded or partially foldedstate. The molecule might then assemble in the periplasmicspace and would then be transported via the Ptl system acrossthe final membrane barrier. Such a two-step process has beenproposed for secretion of pullulanase from E. coli (24) andsecretion of enterotoxin from Vibrio cholerae (8).

REFERENCES1. Arico, B., R. Gross, J. Smida, and R. Rappuoli. 1987. Evolutionary

relationships in the genus Bordetella. Mol. Microbiol. 1:301-308.2. Arico, B., and R Rappuoli. 1987. Bordetella parapertussis and

Bordetella bronchiseptica contain transcriptionally silent pertussistoxin genes. J. Bacteriol. 169:2847-2853.

3. Armstrong, S. K., and C. D. Parker. 1986. Heat-modifiableenvelope proteins of Bordetella pertussis. Infect. Immun. 54:109-117.

4. Bradford, M. M. 1976. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72:248-254.

5. Briggs, M. S., and L. M. Gierasch. 1986. Molecular mechanisms ofprotein secretion: the role of the signal sequence. Adv. ProteinChem. 38:109-180.

6. Covacci, A., and R Rappuoli. 1993. Pertussis toxin export requiresaccessory genes located downstream from the pertussis toxinoperon. Mol. Microbiol. 8:429-434.

7. Ezzel, J. W., W. J. Dobrogosz, W. E. Kloos, and C. R Manclark1981. Phase shift markers in Bordetella: alteratiotis in envelopeproteins. J. Infect. Dis. 143:562-569.

8. Hirst, T. R, and J. Holmgren. 1987. Transient entry of enterotoxinsubunits into the periplasm occurs during their secretion fromVibrio cholerae. J. Bacteriol. 169:1037-1045.

9. Holloway, P. W. 1973. A simple procedure for removal of TritonX-100 from protein samples. Anal. Biochem. 53:304-308.

10. Idigbe, E. O., R Parton, and A. C. Wardlaw. 1981. Rapidity ofantigenic modulation of Bordetella pertussis in modified Horni-brook medium. J. Med. Microbiol. 14:409-418.

11. Kaslow, H. R, and D. L. Burns. 1992. Pertussis toxin and targeteukaryotic cells: binding, entry, and activation. FASEB J. 6:2684-2690.

VOL. 176, 1994

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

5356 JOHNSON AND BURNS

12. Kenimer, J. G., K. J. Kim, P. G. Probst, C. R Manclark, D. G.Burstyn, and J. L. Cowell. 1989. Monoclonal antibodies to pertus-sis toxin: utilization as probes of toxin function. Hybridoma8:37-51.

13. Kuldau, G. A., G. de Vos, J. Owen, G. McCaffrey, and P.Zambryski. 1990. The virB operon of Agrobacterium tumefacienspTiC58 encodes 11 open reading frames. Mol. Gen. Genet.221:256-266.

14. Lacey, B. W. 1960. Antigenic modulation of Bordetella pertussis. J.Hyg. 58:57-93.

15. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

16. Lee, C. K., A. Roberts, and S. Perrin. 1989. Expression of pertussistoxin in Bordetella bronchiseptica and Bordetella parapertussis car-

rying recombinant plasmids. Infect. Immun. 57:1413-1418.17. Leininger, E., J. Kenimer, and M. J. Brennan. 1993. Inhibition of

Bordetella pertussis filamentous hemagglutinin mediated cell ad-herence with monoclonal antibodies. FEMS Microbiol. Lett. 106:31-38.

18. Locht, C., and J. M. Keith. 1986. Pertussis toxin gene: nucleotidesequence and genetic organization. Science 232:1258-1264.

19. Marchitto, K. S., J. J. Munoz, and J. M. Keith. 1987. Detection ofsubunits of pertussis toxin in TnS-induced Bordetella mutantsdeficient in toxin biological activity. Infect. Immun. 55:1309-1313.

20. Marchitto, K. S., S. G. Smith, C. Locht, and J. M. Keith. 1987.Nucleotide sequence homology to pertussis toxin gene in Borde-tella bronchiseptica and Bordetella parapertussis. Infect. Immun.55:497-501.

21. Nicosia, A., M. Perugini, C. Franzini, M. C. Casagli, M. G. Born,G. Antoni, M. Almoni, P. Neri, G. Ratti, and R Rappuoli. 1986.Cloning and sequencing of the pertussis toxin genes: operonstructure and gene duplication. Proc. Natl. Acad. Sci. USA83:4631-4635.

22. Nicosia, A., and R Rappuoli. 1987. Promoter of the pertussis toxinoperon and production of pertussis toxin. J. Bacteriol. 169-.2843-2846.

23. Pizza, M., M. Bugnoli, R Manetti, A. Covacci, and R Rappuoli.1990. The subunit S1 is important for pertussis toxin secretion. J.Biol. Chem. 265:17759-17763.

24. Pugsley, A. P., I. Poquet, and M. G. Kornacker. 1991. Two distinctsteps in pullulanase secretion by Escherichia coli K12. Mol.Microbiol. 5:865-873.

25. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring Harbor Labo-ratory Press, Cold Spring Harbor, N.Y.

26. Schneider, D. R, and C. D. Parker. 1982. Effect of pyridines on

phenotypic properties of Bordetella pertussis. Infect. Immun. 38:548-553.

27. Tamura, M., K. Nogimori, S. Murai, M. Yajima, K. Ito, T. Katada,M. Ui, and S. Ishii. 1982. Subunit structure of islet-activatingprotein, pertussis toxin, in conformity with the A-B model. Bio-chemistry 21:5516-5522.

28. Thorstenson, Y. R, G. A. Kuldau, and P. C. Zambryski. 1993.Subcellular localization of seven VirB proteins of Agrobacteriumtumefaciens: implications for the formation of a T-DNA transportstructure. J. Bacteriol. 175:5233-5241.

29. Thorstenson, Y. R., and P. C. Zambryski. 1994. The essentialvirulence protein VirB8 localizes to the inner membrane ofAgrobacterium tumefaciens. J. Bacteriol. 176:1711-1717.

30. Walker, M. J., M. Rohde, J. Wehland, and K. Timmis. 1991.Construction of minitransposons for constitutive and inducibleexpression of pertussis toxin in bvg-negative Bordetella bronchisep-tica. Infect. Immun. 59:4238-4298.

31. Ward, J. E., D. E. Akiyoshi, D. Regier, A. Datta, M. P. Gordon, andE. W. Nester. 1988. Characterization of the virB operon from anAgrobacterium tumefaciens Ti plasmid. J. Biol. Chem. 263:5804-5814.

32. Ward, J. E., D. E. Akiyoshi, D. Regier, A. Datta, M. P. Gordon, andE. W. Nester. 1990. Correction: characterization of the virB operonfrom Agrobacterium tumefaciens Ti plasmid. J. Biol. Chem. 265:4768.

33. Ward, J. E., E. M. Dale, E. W. Nester, and A. N. Binns. 1990.Identification of a VirBlO protein aggregate in the inner mem-brane of Agrobacterium tumefaciens. J. Bacteriol. 172:5200-5210.

34. Weiss, A. A., and S. Falkow. 1983. Transposon insertion andsubsequent donor formation promoted by TnSOJ in Bordetellapertussis. J. Bacteriol. 153:304-309.

35. Weiss, A. A., and M. Goodwin. 1989. Lethal infection by Bordetellapertussis mutants in the infant mouse model. Infect. Immun.57:3757-3764.

36. Weiss, A. A., E. L Hewlett, G. A. Myers, and S. Falkow. 1984.Pertussis toxin and extracytoplasmic adenylate cyclase as virulencefactors in Bordetella pertussis. J. Infect. Dis. 150:219-222.

37. Weiss, A. A., F. D. Johnson, and D. L. Burns. 1993. Molecularcharacterization of an operon required for pertussis toxin secre-tion. Proc. Nati. Acad. Sci. USA 90:2970-2974.

38. Weiss, A. A., A. R Melton, K. E. Walker, C. Andraos-Selim, andJ. J. Meidl. 1989. Use of the promoter fusion transposon TnS lacto identify mutations in Bordetella pertussis vir-regulated genes.Infect. Immun. 57:2674-2682.

39. Wickner, W., A. J. M. Driessen, and F.-U. Hartl. 1991. Theenzymology of protein translocation across the Eschenchia coliplasma membrane. Annu. Rev. Biochem. 60:101-124.

40. Zambryski, P. C. 1992. Chronicles from the Agrobacterum-plantcell DNA transfer story. Annu. Rev. Plant Physiol. Plant. Mol.Biol. 43:465-490.

J. BAC7ERIOL.

on February 25, 2021 by guest

http://jb.asm.org/

Dow

nloaded from