ABSTRACT
We have identified two extremely large open reading frames (ORFs) in Haemophilus ducreyi 35000, lspA1 andlspA2, each of which encodes a predicted protein product whose N-terminal half is approximately 43% similar to the N-terminal half of Bordetella pertussis filamentous hemagglutinin (FhaB). To the best of our knowledge, lspA1 (12,500 nucleotides [nt]) and lspA2 (14,800 nt) are among the largest prokaryotic ORFs identified to date. The predicted proteins, LspA1 and LspA2, are 86% identical overall to each other and also have limited amino acid sequence similarity at their N termini to other secreted bacterial proteins, including certain hemolysins. Southern blot analysis indicated that lspA1 and lspA2sequences were present in 15 other geographically diverse H. ducreyi strains. Reverse transcriptase PCR analysis of total RNA isolated from H. ducreyi 35000 grown in liquid medium, grown on solid agar medium, and isolated from lesions of H. ducreyi-infected rabbits indicated that lspA1 andlspA2 were transcribed both in vitro and in vivo. A 260-kDa protein present in culture supernatant from eight virulentH. ducreyi strains reacted with both polyclonal serum from rabbits infected with H. ducreyi 35000 and a monoclonal antibody predicted to bind both LspA1 and LspA2. This 260-kDa protein in H. ducreyi 35000 culture supernatant was shown to be the protein product of thelspA1 ORF based on its reactivity with a monoclonal antibody specific for LspA1. Four H. ducreyi strains, previously shown to be avirulent in the temperature-dependent rabbit model for chancroid, did not produce either LspA1 or LspA2 in vitro. This finding raised the possibility that LspA1, LspA2, or both may be involved in the ability of H. ducreyi to cause lesions in this animal model.
Chancroid, a sexually transmitted disease characterized by painful genital ulceration, is caused by the fastidious gram-negative bacterium Haemophilus ducreyi(60). Much remains unknown concerning the pathogenesis of chancroid and the factors produced by H. ducreyi which enable this bacterium to cause disease. Potential virulence factors produced by H. ducreyi include pili (11), lipooligosaccharide (LOS) (13, 21), a hemoglobin-binding outer membrane protein (18, 56), a cell-associated hemolysin (5, 37, 38, 59), a diffusible cytotoxin (16, 43), and a copper-zinc superoxide dismutase (51).
Chancroid is transmitted by direct sexual contact, with H. ducreyi presumably gaining entry into the host through microabrasions in the skin. Therefore, it is likely that early steps in the pathogenesis of chancroid involve the adherence of H. ducreyi to host cells and extracellular matrix components located below the keratinized epithelium. This hypothesis is supported by the finding that H. ducreyi causes chancroidal lesions in humans only when applied to a damaged epithelium (55). Consistent with these observations, H. ducreyi is reported to adhere to and invade several different human cell lines in vitro (2-4, 12, 31, 32, 53, 55, 58) as well as to bind extracellular matrix components (1, 11). Expression of wild-type LOS appears to be essential for maximal attachment ofH. ducreyi to human foreskin fibroblasts (HFF) and keratinocytes in vitro (21), but there is some evidence that other proteinaceous factors also may be involved (3, 20, 39). However, to date, no proteinaceous adherence factors produced by H. ducreyi have been definitively identified.
The specific attachment of bacteria to host tissues is recognized as an important step in the pathogenesis of many infectious diseases (27). One bacterial adhesin that has been the focus of intensive research is the filamentous hemagglutinin (FHA) ofBordetella pertussis. The mature form of FHA is a 220-kDa protein processed from a 367-kDa precursor encoded by the 10.8-kbfhaB gene (17, 33, 45); mature FHA, in conjunction with an accessory protein (FhaC) (66), is efficiently exported from the cell and is found in B. pertussis culture supernatant. Mutant analysis has indicated that FHA is involved in the ability of B. pertussis to colonize the mouse trachea (29, 35).
In this study, we report the identification of two very largeH. ducreyi 35000 open reading frames (ORFs),lspA1 and lspA2 (lsp stands for large supernatant protein), which are predicted to encode proteins that have significant similarity to B. pertussis FhaB. We also describe a 260-kDa protein encoded by the lspA1 gene that is secreted intoH. ducreyi 35000 culture medium. The proteins encoded by the lspA1 and lspA2 ORFs have the potential to be involved in the interaction of H. ducreyi with its human host.
MATERIALS AND METHODS
Bacterial strains and culture conditions.The bacterial strains and plasmids used in this study are listed in Table1. H. ducreyi andEscherichia coli strains were grown as described previously (42, 50). To prepare H. ducreyi concentrated culture supernatants (CCS), overnight growth from two chocolate agar plates (ca. 2 × 109 CFU) was inoculated into 15 ml of Columbia broth (Difco Laboratories, Detroit, Mich.) containing 1% (vol/vol) IsoVitaleX (Becton Dickinson, Cockeysville, Md.), equine hemin (25 μg/ml) (Sigma Chemical Co., St. Louis, Mo.), and 2.5% (vol/vol) heat-inactivated fetal bovine serum in a 75 cm2tissue culture flask (Costar, Cambridge, Mass.). These flasks were incubated at 33°C in an atmosphere of 95% air–5% CO2for 48 h without agitation. The culture fluid was centrifuged at 8,000 × g for 15 min to remove the bacteria, passed through a 0.22-μm-pore-size filter, and ultracentrifuged at 125,000 × g for 1.5 h to remove membrane fragments and insoluble debris. The culture supernatant was then concentrated approximately 40-fold by ultrafiltration with a Centricon-30 filtration unit (Millipore, Inc., Bedford, Mass.) and stored at −20°C.
Bacterial strains and plasmids used in this study
Recombinant DNA techniques and genomic DNA libraries.Standard recombinant DNA techniques were performed as described previously (50). Two plasmid-based H. ducreyi 35000 genomic DNA libraries were used in these studies. The pUC19-based library has been described previously (56), and a pBR322-based library was constructed with 6- to 15-kbPstI fragments of H. ducreyi 35000 genomic DNA.
Southern blot and nucleotide sequence analyses.Southern blot analysis and probing of libraries with DNA probes were performed as described previously (16). The nucleotide sequences of the H. ducreyi 35000 lspA1 andlspA2 genes were determined by using a model 373A automated DNA sequencer (Applied Biosystems Inc., Foster City, Calif.) to analyze DNA contained in recombinant plasmids and PCR products. To determine the nucleotide sequence of the 3′ portion of the lspA2 ORF, the 8.5-kb PstI insert of pCW107 was cloned into thePstI site of pBluescript II KS(+) to create pCW114, which was then used to construct nested deletions. Both strands of the DNA constituting the lspA1 and lspA2 loci were sequenced in their entirety. DNA sequences were assembled into larger contiguous sequences and analyzed with AssemblyLign and MacVector DNA analysis software (version 6.0; Oxford Molecular Group, Campbell, Calif.).
RT-PCR analysis.Multiplex reverse transcriptase PCR (RT-PCR) analysis of H. ducreyi RNA was performed with the Titan one-tube RT-PCR system according to the manufacturer’s recommendations (Boehringer Mannheim, Indianapolis, Ind.). The Ultraspec total RNA isolation reagent (Biotecx Laboratories Inc., Houston, Tex.) was used to purify template RNA for RT-PCR analysis from (i) confluent H. ducreyi colonies scraped from one chocolate agar plate, (ii) H. ducreyi cultures grown in tissue culture flasks (as described above for the production of CCS), and (iii) lesions excised 48 h postinfection from rabbits inoculated intradermally with 106 CFU of H. ducreyi 35000 (42). The excised lesions were quartered with a sterile scalpel blade and washed with sterile phosphate-buffered saline (PBS). This PBS wash was centrifuged at 12,000 ×g to pellet bacteria and other cellular debris. The pellet was suspended in 1 ml of the Ultraspec total RNA isolation reagent, and the total RNA was isolated according to the manufacturer’s instructions. To remove contaminating DNA prior to RT-PCR analysis, purified RNA (10 μg) in water containing 5 mM MgCl2 was treated with 4 U of RQ1 RNase-free DNase (Promega Corp., Madison, Wis.) for 30 min at 37°C. The DNase was then inactivated by incubating the mixture for 5 min at 75°C. Each RT-PCR reaction mixture (50 μl total volume) contained 1× RT-PCR buffer, a 150 μM concentration of each dNTP, 100 ng of each primer, 1.5 mM MgCl2, 5 mM dithiothreitol, 8 U of RNasin (Promega), and 1 μl of RT-PCR enzyme mixture.
Three oligonucleotide primer sets (Table2) were added to each reaction mixture: (i) P8 and P9 were added to reverse transcribe and amplify a 354-nucleotide (nt) region of the H. ducreyi pal RNA transcript (54), (ii) P10 and P11 were added to reverse transcribe and amplify a 264-nt region of the lspA1 RNA transcript, and (iii) P12 and P13 were added to reverse transcribe and amplify a 320-nt region of the lspA2 RNA transcript. All RT-PCR mixtures were cycled with a PTC-100 programmable thermal controller (MJ Research Inc., Watertown, Mass.). The reverse transcription step of the protocol was performed for 30 min at 50°C and was followed by a 2-min denaturation step at 94°C. The reaction mixtures were then subjected to 30 cycles of PCR amplification consisting of a 94°C denaturation step for 1 min, a 52°C annealing step for 1 min, and a 68°C extension step for 1 min 15 s. The reaction mixtures were then subjected to a final extension step for 10 min at 68°C. Control reaction mixtures to check for DNA contamination of RNA templates were prepared and cycled exactly as described above except that these were not subjected to a 50°C reverse transcription step.
Oligonucleotides used in this study
Fusion protein construction and MAb production. (i) Production of a MAb reactive with both LspA1 and LspA2.A 997-bp region common to both lspA1 (nt 6165 to 7161, encoding amino acids [aa] 1428 to 1759 of LspA1) (Fig. 1A) andlspA2 (nt 6139 to 7136, encoding aa 1557 to 1888 of LspA2) (Fig. 1B) was amplified by PCR from H. ducreyi 35000 genomic DNA with primers P2 and P3 (Table 2), which contained restriction enzyme sites on the 5′ ends (EcoRI in P2 andAvaI in P3). The gel-purified PCR product was digested withEcoRI and AvaI and ligated in frame to the gene encoding glutathione-S-transferase (GST) in vector pGEX4T-2 (Pharmacia Biotech, Piscataway, N.J.) to construct pCW103. The purified 65-kDa GST-LspA fusion protein was used to immunize mice for hybridoma production (48). Monoclonal antibody (MAb) 11B7 (immunoglobulin G1 [IgG1]) was shown by Western blot analysis to recognize the 65-kDa GST-LspA fusion protein.
Partial restriction endonuclease map of theH. ducreyi 35000 chromosomal lspA1 (A) andlspA2 (B) loci and the cloned DNA inserts and DNA probes used in this study. Detailed descriptions of the cloned DNA inserts (in plasmids) are provided in Table 1. The lspA1 ORF was flanked by genes encoding homologs of H. influenzaephosphomannomutase (ORF A) and H. influenzae GMP synthase (ORF B). The lspA2 ORF was flanked by genes encoding homologs of B. pertussis fhaC (ORF C) andH. influenzae anaerobic glycerol-3-phosphate dehydrogenase subunit A (ORF D).
(ii) Production of an LspA1-specific MAb.A 459-bp region encoding an amino acid sequence specific to lspA1 (nt 5225 to 5683, encoding aa 1115 to 1267 of LspA1) (Fig. 1A) was amplified by PCR from H. ducreyi 35000 genomic DNA with primers P4 and P5 containing a BamHI site and an EcoRI site, respectively, at their 5′ ends (Table 2). The gel-purified 459-bp PCR product was digested with BamHI and EcoRI and ligated in frame into the polyhistidine (His) fusion protein vector pRSETB (Invitrogen Inc., Carlsbad, Calif.) to construct pCW125. Mice were immunized with the purified 24-kDa His-LspA1 fusion protein and used to produce MAbs as described above. MAb 40A4 (IgG1) was shown by enzyme-linked immunosorbent assay (ELISA) and Western blot analyses to bind the His-LspA1 fusion protein.
(iii) Production of an LspA2-specific MAb.The synthetic peptide KASEKYKKVENVDHKENIDE (aa 1356 to 1375 of LspA2) was synthesized by the Biopolymers Facility at the University of Texas Southwestern Medical Center and was covalently bound to keyhole limpet hemocyanin (KLH) (Sigma) with glutaraldehyde. Mice were immunized with the KLH-peptide conjugate, and their splenocytes were fused as described above. MAb 1H9 (IgG2a) was shown by ELISA and Western blot analyses to bind this peptide.
To assess the specificity of MAb 40A4 for LspA1 and that of MAb 1H9 for LspA2, a third fusion protein, which contained a portion of LspA2 that was homologous (but not identical) to the LspA1 region contained in the His-LspA1 fusion protein, was produced. To produce this GST-LspA2 fusion protein, a 534-bp region from lspA2 (nt 5120 to 5653, encoding aa 1218 to 1394 of LspA2) was amplified by PCR with primers P6 and P7 (Table 2) and ligated into pGEX4T-2 to obtain pCW141. MAb 40A4 reacted only with the His-LspA1 fusion protein, not with this new GST-LspA2 fusion protein. In contrast, MAb 1H9 reacted only with the GST-LspA2 fusion protein, confirming that MAb 1H9 is specific for LspA2. Furthermore, this result indicated that MAb 1H9, originally raised against a peptide hapten, bound its epitope when the relevant peptide was present in the context of a much larger protein (i.e., GST-LspA2).
Western blot analysis and antisera. H. ducreyiCCS (22 μl) and a control consisting of uninoculated, concentrated growth medium without bacteria (22 μl) were heated at 100°C for 5 min in sample buffer (40) and then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis through 7.5% (wt/vol) polyacrylamide separating gels. Proteins in the gel were transferred overnight at 68 V to nitrocellulose (NitroBind; Micron Separations Inc., Westboro, Mass.). Membranes were blocked in PBS containing 0.05% (vol/vol) Tween 20 (Sigma) and 3% (wt/vol) skim milk. Membranes were incubated with MAbs in the form of hybridoma culture supernatants or with polyclonal antiserum diluted in blocking buffer as described previously (56). Antiserum toH. ducreyi 35000 was produced by injecting a rabbit intradermally (42) with viable cells of H. ducreyi 35000 on three occasions separated by 1-month intervals.
RESULTS
Identification of the H. ducreyi 35000 chromosomal loci encoding FHA homologs.The screening of an H. ducreyi 35000 genomic library with an oligonucleotide probe (P1; Table 2) for the cdtABC genes encoding the diffusible cytotoxin of H. ducreyi (16) identified plasmid pJL13-1 (Fig. 1A), which weakly hybridized this probe. A nucleotide sequence analysis revealed that the pJL13-1 insert did not have substantial identity to the P1 probe but did contain an incomplete ORF which encoded a polypeptide with 43% similarity to the FHA (FhaB) of B. pertussis (17, 33, 45). Subsequent Southern blot analysis (data not shown) suggested that H. ducreyi 35000 contained two fhaB-like genes in its chromosome. Ultimately, we determined the nucleotide sequence of 33 kb of H. ducreyi 35000 DNA (Fig. 1) through a combination of genomic library screening and the use of vector-anchored PCR (30) and identified two distinct fhaB-like ORFs, designated lspA1 and lspA2 (lsp stands for large supernatant protein, for reasons that will become apparent below). The recombinant plasmids used to obtain this sequence information are depicted in Fig. 1. Confirmation of the chromosomal arrangement of the cloned DNA fragments constituting the lspA1 andlspA2 loci (Fig. 1) was provided by a combination of PCR-based and Southern blot analyses (data not shown).
Features of the lspA1 gene and its predicted protein product.The lspA1 ORF comprised 12,459 nt (nt 1883 to 14341 in Fig. 1A). A partial ORF (ORF A in Fig. 1A) encoding a predicted protein with 76% identity to the Haemophilus influenzae phosphomannomutase (19) was located upstream from lspA1. The putative lspA1 initiation codon was approximately 550 nt downstream from the termination codon of ORF A (Fig. 1A) and was preceded by a probable ribosome binding site and putative −10 (TATTCT; nt 1563 to 1568) and −35 (TTAAGT; nt 1540 to 1545) promoter sequences. A stem-loop structure that may function as a rho-independent transcriptional terminator was identified 32 nt downstream from the TAG termination codon of lspA1. A partial ORF (ORF B in Fig. 1A) located on the opposite strand encoded a predicted protein with 88% identity to the GMP synthase of H. influenzae(19).
The protein product of the lspA1 ORF was predicted to comprise 4,152 aa and to have a molecular weight of 456,141. The lspA1 gene product contained several noteworthy features including the amino acid motif NPNG(I/M) (aa 216 to 220 in Fig. 2A), which has been associated with the secretion of several bacterial proteins, including the hemolysins of Serratia marcescens and Proteus mirabilis (52, 62), FhaB of B. pertussis(28, 66), and the HMW1A and HMW2A adhesins of nontypeableH. influenzae (9, 10). The LspA1 protein also contained a putative ATP/GTP binding domain (Walker motif A), GINTKGKT (aa 2206 to 2213) (64), as well as several different, relatively short, amino acid sequences repeated throughout different regions of the protein (data not shown).
Clustal-W alignment of selected regions of LspA1 and LspA2 with homologous proteins. The N-terminal 380 aa of LspA1 and LspA2 were aligned with the N-terminal region of FhaB (A). Internal regions of LspA1 and LspA2 were aligned with the C-terminal one-third of the H. somnus P76 protein (B). The shaded box highlights the NPNG(I/M) motif. Asterisks indicate identical amino acids; periods indicate conserved amino acid substitutions.
Features of the lspA2 gene and its predicted protein product.The lspA2 ORF comprised 14,760 nt (nt 1471 to 16230 in Fig. 1B), making lspA2 one of the largest prokaryotic ORFs ever identified. The end of an incomplete ORF (ORF C in Fig. 1B) located 46 nt upstream from the putative initiation codon for lspA2 encoded a portion of a protein with 31% identity to B. pertussis FhaC, an outer membrane protein involved in the export of FHA (66). The putative ATG initiation codon of the lspA2 ORF was preceded by a probable ribosome binding site. The lspA2 termination codon, like that of lspA1, was followed by a sequence that may form a stem-loop structure and function as a rho-independent transcriptional terminator. A partial ORF starting 333 nt downstream from this termination codon (ORF D in Fig. 1B) encoded a predicted protein that was 83% similar to H. influenzae anaerobic glycerol-3-phosphate dehydrogenase subunit A (glpA gene product) (19).
The predicted protein product of the lspA2 ORF comprised 4,919 aa and had a calculated molecular weight of 542,559. The LspA2 protein, like LspA1, also contained an NPNG(I/M) sequence (aa 211 to 215) and a putative ATP/GTP binding domain (Walker motif A), GINTKGKT (aa 2344 to 2352). LspA2 also contained multiple different amino acid sequences repeated throughout the protein, including several identical to those found in LspA1 (data not shown). The amino acid repeats present in LspA2 included three direct 319-aa repeats near the C-terminal portion of the protein (Fig.3) as well as a series of small 21-aa asparagine-rich repeats near the N-terminal portion of the LspA2 protein (data not shown). In general, the amino acid repeats present in LspA2 were longer and were repeated more frequently than those observed in LspA1.
Schematic representation of the predicted protein products of H. ducreyi 35000 lspA1 andlspA2. Regions of nearly complete identity between LspA1 and LspA2 are displayed as boxes with identical shading. The locations of the amino acid sequences of LspA1 and LspA2 used to produce MAbs 40A4, 11B7, and 1H9 are indicated. The location of the region that has 70% identity with the H. somnus P76 protein is also indicated (hatched box), as is the location of the three contiguous 319-aa repeats present in LspA2.
The LspA1 and LspA2 proteins were 86% identical overall, with the majority of the divergence between the two proteins located between aa 90 and 1318 of LspA1 and between aa 90 and 1447 of LspA2 (Fig. 3). The LspA1 and LspA2 proteins did not contain typical prokaryotic signal peptide sequences (41) at their amino termini. Unlike FhaB, neither LspA1 nor LspA2 contained the integrin-binding triplet amino acid sequence RGD (46, 49). However, both LspA1 and LspA2 contained several alternative integrin-binding amino acid sequences, including multiple KGD, RXXD, IDS, and LDV sequences (49), which were present throughout the N-terminal two-thirds of LspA1 and LspA2 (data not shown).
Similarity of LspA1 and LspA2 to other bacterial proteins.A BLAST search (7) of the combined, nonredundant GenBank/EMBL protein databases revealed that the primary amino acid sequences of the predicted LspA1 and LspA2 proteins were 43% similar to that of B. pertussis FhaB (17, 33, 45); this similarity was restricted to the N-terminal halves of LspA1 and LspA2. An alignment of the first 380 aa of both LspA1 and LspA2 with the first 380 aa of FhaB is shown in Fig.2A. LspA1 and LspA2 were also similar to Neisseria meningitidis PspA, a very large protein (242 kDa) listed in the GenBank database (accession no. AF030941 ) whose function has not been reported to date. The N-terminal 1,000 aa of LspA1 and LspA2 also had significant similarity to the hemolysin precursors of several pathogens, including P. mirabilis (HpmA) (62) andH. ducreyi (HhdA) (38), and with the HMW1A and HMW2A adhesins produced by nontypeable H. influenzae (9). Interestingly, both LspA1 (aa 2763 to 3024) and LspA2 (aa 2892 to 3153) contained a region with a very high degree of identity (ca. 70%) to the C-terminal one-third of the P76 protein of Haemophilus somnus (14, 15) (Fig. 2B).
Evidence for expression of lspA1 and lspA2by H. ducreyi.Because initial attempts to identify the protein products of lspA1 and lspA2 were unsuccessful (data not shown), we used RT-PCR analysis to detect the presence of the lspA1 and lspA2 transcripts under different in vitro and in vivo growth conditions. Total RNA isolated from H. ducreyi 35000 cells grown on chocolate agar plates and in liquid broth culture and from saline washes ofH. ducreyi 35000-infected rabbit lesions was subjected to multiplex RT-PCR analysis to simultaneously detect the presence of the H. ducreyi 35000 lspA1,lspA2, and pal transcripts (Fig.4). Oligonucleotide primers which would specifically reverse transcribe and amplify a 354-nt region of theH. ducreyi pal mRNA transcript (P8 and P9; Table 2) were included in each RT-PCR mixture as an internal positive control because the pal gene encodes an mRNA transcript successfully used as a control in other H. ducreyi mRNA studies (16). Primers that would specifically reverse transcribe and amplify a 264-nt region of the lspA1 transcript (P10 and P11; Table 2) and a 320-nt region of the lspA2 transcript (P12 and P13; Table 2) were also included in each RT-PCR mixture.
Agarose gel electrophoresis of H. ducreyi multiplex RT-PCR products. The predicted sizes of theH. ducreyi 35000 RT-PCR products were as follows:pal, 354 bp; lspA1, 264 bp; and lspA2, 320 bp. The templates included in each of the reaction mixtures were as follows: lane 1, no template (negative control); lane 2, 100 ng ofH. ducreyi 35000 genomic DNA (positive control); lanes 3 and 7, 2 μg of RNA isolated from H. ducreyi grown on chocolate agar plates; lanes 4 and 8, 2 μg of RNA isolated fromH. ducreyi grown for 8 h in liquid broth; lanes 5 and 9, 2 μg of RNA isolated from H. ducreyi grown for 25 h in liquid broth; lanes 6 and 10, 3 μg of RNA isolated fromH. ducreyi-infected rabbit lesions. The reaction mixtures loaded in lanes 7 to 10 were not subjected to the reverse transcription step of the RT-PCR protocol and served as controls to detect DNA contamination of the RNA templates. Lanes M contain DNA size markers.
The three predicted PCR products were successfully amplified whenH. ducreyi 35000 genomic DNA was used as the template (Fig. 4, lane 2), indicating that these three primer sets were able to simultaneously amplify the specified regions of pal,lspA1, and lspA2. When these same three primer sets were used to reverse transcribe and amplify RNA isolated fromH. ducreyi 35000 grown in vitro or RNA isolated directly from 35000-infected rabbit lesions, all three predicted PCR products again were obtained (Fig. 4, lanes 3 to 6). The RT-PCR amplification products obtained in these studies were not produced as a result of DNA contamination present in the RNA templates because no products were obtained when the reaction mixtures were not subjected to a 50°C reverse transcription step (Fig. 4, lanes 7 to 10). These results indicated that the H. ducreyi 35000lspA1 and lspA2 genes were transcribed both in vitro and during the development of lesions in rabbits infected with this strain.
Identification of a 260-kDa protein in H. ducreyiCCS reactive with antiserum from H. ducreyi35000-infected rabbits. B. pertussis FHA and the HMW1A and HMW2A proteins of nontypeable H. influenzae are known to be secreted into culture medium during growth (8, 33). These observations prompted us to use Western blot analysis to examine H. ducreyi CCS for the presence of LspA1 and LspA2. Antiserum from a rabbit infected intradermally withH. ducreyi 35000 recognized a 260-kDa protein in CCS from H. ducreyi 35000, R018, 181, CA173, WPB506, BG411, 041, and CIP542 (Fig. 5A, lanes 2 to 9, respectively). This 260-kDa protein was not present in concentrated, uninoculated growth medium (Fig. 5A, lane 1), indicating that the protein was likely produced by H. ducreyi in vivo during lesion formation in the temperature-dependent rabbit model.
Western blot analysis of CCS from 12 H. ducreyi strains. CCS from each strain was probed with polyclonalH. ducreyi 35000-infected rabbit serum (A), MAb 11B7 (B), MAb 40A4 (C), and MAb 1H9 (D). Lanes: 1, uninoculated medium control; 2, 35000; 3, R018; 4, 181; 5, CA173; 6, WPB506; 7, BG411; 8, 041; 9, CIP542; 10, A77; 11, 6V; 12, E1673; 13, 78226. Rabbit antibodies were detected with a horseradish peroxidase-conjugated secondary antibody; MAbs were detected with radioiodinated secondary antibodies. The immunoreactive doublets apparent in panel A are the result of the tendency of LspA1 to give rise to multiple forms.
It is important to note that this polyclonal rabbit H. ducreyi 35000 antiserum did not react with CCS from H. ducreyi A77, 6V, E1673, and 78226 (Fig. 5A, lanes 10 to 13, respectively), a finding which suggested that these four strains did not secrete the 260-kDa CCS protein. Interestingly, these four strains have been shown to be essentially avirulent in the temperature-dependent rabbit model for experimental chancroid (6).
Reactivity of H. ducreyi CCS with LspA1- and LspA2-specific MAbs.Prior to the discovery of the existence of the lspA2 gene, MAb 11B7 was produced by immunizing mice with a fusion protein (expressed by pCW103; Fig. 1 and 3) containing an amino acid sequence that later proved to be present in both LspA1 and LspA2 (aa 1428 to 1759 of LspA1 and aa 1557 to 1888 of LspA2). MAb 11B7 was thus predicted to be reactive with both LspA1 and LspA2. It was found that the MAb bound a 260-kDa protein present in CCS fromH. ducreyi 35000, R018, 181, CA173, WPB506, BG411, 041, and CIP542 (Fig. 5B, lanes 2 to 9, respectively) but did not react with CCS from strains A77, 6V, E1673, and 78226 (Fig. 5B, lanes 10 to 13, respectively). These results proved that the 260-kDa antigen reactive with the polyclonal H. ducreyi 35000 antiserum (Fig.5A) was likely the mature or processed form of either LspA1 or LspA2 or both. To determine which of these proteins was actually present inH. ducreyi 35000 CCS, we produced the LspA1-specific MAb 40A4 and the LspA2-specific MAb 1H9. The specificity of these MAbs was confirmed by Western blot analysis using LspA1- and LspA2-derived fusion proteins (see Materials and Methods).
MAb 40A4, but not MAb 1H9, reacted with the 260-kDa protein present inH. ducreyi 35000 CCS (Fig. 5C and D, respectively, lanes 2), suggesting that this band comprised only the protein product of the lspA1 ORF. However, we cannot rule out the possibility that the LspA2 protein was present in H. ducreyi 35000 CCS at a concentration beneath the limit of detection of our Western blot system. MAb 40A4 was also observed to bind a 260-kDa protein present in CCS from H. ducreyiR018, CA173, and CIP542 (Fig. 5C, lanes 3, 5, and 9), suggesting that these three strains also secreted the LspA1 protein. The LspA2-specific MAb 1H9 did not react with CCS from any of the other H. ducreyi strains tested (Fig. 5D, lanes 3 to 13).
Detection of lspA1 and lspA2 in otherH. ducreyi strains.To examine the prevalence oflspA1 and lspA2 among different H. ducreyi strains, we performed Southern blot analysis of genomic DNA isolated from strain 35000 and 15 other H. ducreyistrains from diverse geographic areas. A 1.0-kblspA1-specific DNA probe hybridized either a 5.5- or 5.1-kbAflII fragment from each strain (Fig.6A). Similarly, a 1.2-kblspA2-specific probe hybridized to 2.5-kb AflII fragments from all 16 strains (Fig. 6B). These results indicated that all 16 H. ducreyi strains contained bothlspA1 and lspA2 DNA sequences.
Detection of lspA1 and lspA2sequences among H. ducreyi strains by Southern blot analysis. Genomic DNA from each strain was digested withAflII, electrophoresed through 0.7% agarose gels, transferred to nitrocellulose, and probed with DNA fragments specific for H. ducreyi 35000 lspA1 (A) orlspA2 (B). Lanes: 1, 35000; 2, R018; 3, 181; 4, CA173; 5, WPB506; 6, BG411; 7, 041; 8, CIP542; 9, 1145; 10, 1151; 11, Cha-I; 12, Hd12; 13, A77; 14, 6V; 15, E1673, and 16, 78226. ThelspA1-specific probe consisted of the 1.0-kb HpaI restriction fragment of pDad-5 (Fig. 1A), and thelspA2-specific probe consisted of the 1.2-kbHincII restriction fragment of pCW118 (Fig. 1B). DNA size markers (in kilobases) are listed on the side of each panel.
DISCUSSION
We were surprised to discover that H. ducreyicontained not one but two enormous ORFs, lspA1 (12,459 nt) and lspA2 (14,760 nt), that encoded proteins which at the primary amino acid sequence level most closely resembledB. pertussis FhaB. Determination of the complete nucleotide sequences of the lspA1 andlspA2 ORFs proved to be a difficult task because of both the extensive sequence identity shared by these genes and the presence of three contiguous 1-kb DNA repeats in the 3′ portion oflspA2. To the best of our knowledge, lspA1 andlspA2 represent two of the largest prokaryotic ORFs in the combined nonredundant GenBank/EMBL nucleotide sequence database.
It must be noted that bacterial gene duplications are not uncommon events, particularly with respect to genes known to encode adhesins. The HMW1A and HMW2A protein adhesins of nontypeable H. influenzae are 71% identical and have homology with both LspA1 and LspA2 (10). Furthermore, both E. coli andH. influenzae biogroup aegyptius are known to contain duplications of the gene complexes encoding pili (34, 44). The fact that DNA sequences from both the lspA1and lspA2 ORFs were detected in all H. ducreyi strains examined in this study suggests that both genes have been conserved and may play some role in the development of chancroidal ulcers. However, the locations of the lspA1 andlspA2 genes in the H. ducreyi 35000 chromosome (26) have not been determined.
The identification of the H. ducreyi lspA1 andlspA2 genes further expands the list of extremely large prokaryotic ORFs, many of which encode proteins with important binding functions. Other notable members of this group include B. pertussis fhaB (10,774 nt) (17, 45),Porphyromonas gingivalis hagA (7,887 nt) (25), and N. meningitidis pspA (6,822 nt; GenBank accession no.AF030941 ). FHA, the most extensively characterized of the very large bacterial proteins listed above, is an important adherence factor produced by virulent-phase B. pertussis(33). FHA is considered to be a major virulence factor ofB. pertussis and is also a component of some of the new acellular pertussis vaccines (22). P. gingivalis hagA is known to encode a hemagglutinin, whereas the precise function of the predicted protein product of the N. meningitidis pspA gene has not been reported. Direct evidence for the involvement of these very large proteins in the expression of virulence by their respective pathogens has been obtained through mutant analysis only for FHA (29, 35).
The mature 220-kDa FHA protein is derived from a 367-kDa precursor by both N- and C-terminal processing (28, 47) and promotes the specific attachment of B. pertussis to the ciliated epithelial cells of the upper respiratory tract (45, 61). Similar to FHA, H. ducreyi LspA1 is present in culture supernatant. The mechanism by which LspA1 is released from theH. ducreyi cell remains to be determined, although the presence of the NPNG(I/M) motif near the N terminus of this protein suggests the participation of an export system similar to that involved in the secretion of FHA (28, 47), certain hemolysins (65), and HMW1A/HMW2A (57). The discrepancy between the size of the LspA1 protein detected in culture supernatant (260 kDa) and the size of the predicted protein product of the lspA1 ORF (456 kDa) raises the possibility that LspA1 undergoes both N- and C-terminal processing in a manner similar to that for FhaB (28, 47). The secreted form of FHA has been proposed to act as a bridge between B. pertussis and eukaryotic cells based on the observation that B. pertussis FHA mutants pretreated with purified FHA exhibit enhanced adherence to eukaryotic cells (61, 63). However, in contrast to FHA, which is abundant in B. pertussisculture supernatant, LspA1 is found in very small amounts inH. ducreyi culture supernatant and is difficult to detect in H. ducreyi cell envelope preparations (data not shown). It should be noted that both the H. ducreyicell-associated hemolysin (5, 37) and the H. ducreyi soluble cytotoxin (16) also appear to be expressed in minute quantities in vitro.
We are confident in reporting that, at least for H. ducreyi 35000, the LspA1 protein was released into culture supernatant under the in vitro conditions employed in this study. Interestingly, we were unable to detect the LspA2 protein inH. ducreyi 35000 CCS despite the fact that RT-PCR analysis revealed the presence of lspA2 transcripts in both in vitro- and in vivo-grown H. ducreyi 35000 (Fig. 4). Whether this inability to detect the LspA2 protein in H. ducreyi 35000 CCS was the result of a lack of expression of this protein or was caused by extremely low-level expression of LspA2 cannot be determined at this time. The presence of the lspA2transcript in the absence of detectable LspA2 protein also raises the possibility that some type of posttranscriptional regulation may occur. However, the fact that the lspA2 gene was preserved as a unusually long ORF in strain 35000 suggests that, even though we have been unable to detect LspA2 under the in vitro conditions examined in this study, expression of the LspA2 protein may be important to the survival and growth of H. ducreyi during the development of chancroidal ulcers.
Definitive confirmation of the identity of the 260-kDa protein(s) present in CCS from the other H. ducreyi strains (Fig.5A), which did not bind MAb 40A4 (Fig. 5C) or MAb 1H9 (Fig. 5D), will necessarily have to await determination of the nucleotide sequences of the lspA1 and lspA2 ORFs of these other strains. For the same reason, we cannot at this time determine whether the three other strains (strains R018, CA173, and CIP542 in Fig. 5C) whose CCS contained 260-kDa proteins reactive with the LspA1-specific MAb 40A4 might also elaborate LspA2 proteins that are unreactive with MAb 1H9.
The identification of the lspA1 and lspA2 ORFs inH. ducreyi and the discovery of the LspA1 protein inH. ducreyi 35000 CCS are particularly exciting since no adhesins have been definitively identified in H. ducreyi. Two recent reports (3, 21) suggest that LOS may be involved in H. ducreyi adherence to both keratinocytes and to HFF cells. However, it is likely that proteinaceous adhesins are expressed by H. ducreyibecause proteinase K treatment has been demonstrated to significantly reduce the attachment of these bacteria to HFF cells (3). It is tempting to speculate that the LspA1 protein may be involved in the attachment to host cells because three H. ducreyistrains (A77, E1673, and 6V) of the four examined in this study that did not produce LspA1 were previously shown to be deficient in attachment to HFF cells (6). Moreover, all four of these strains were essentially avirulent in the temperature-dependent rabbit model for chancroid (6). Whether a causal relationship exists between LspA1 expression and either the virulence or adherence of H. ducreyi remains to be determined.
An additional interesting finding from this study was the identification of a 265-aa sequence present in both LspA1 and LspA2 that was 70% identical to the C-terminal one-third of a 76-kDa protein encoded by a DNA region associated with serum resistance in virulentH. somnus strains (14, 15). The presence of this amino acid sequence in LspA1 raises the intriguing possibility that LspA1 could be multifunctional and involved in both adherence and complement resistance. Future efforts in this laboratory will be directed toward the determination of the biological function(s) of LspA1 and LspA2.
ACKNOWLEDGMENTS
This study was supported by U.S. Public Health Service grant AI32011 to E.J.H. C.K.W. was the recipient of a National Research Service Award (F32-AI09845).
We thank Marla Stevens for production of the H. ducreyi35000-infected rabbit serum, Beth Bauer for assistance with rabbit infections for RT-PCR analysis, and Sharon Thomas for expert technical assistance. We also thank Michelle Alfa for providing many of theH. ducreyi strains used in this study.
FOOTNOTES
- Received 30 April 1998.
- Accepted 14 September 1998.
- Copyright © 1998 American Society for Microbiology
