INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Haemophilus ducreyi is the etiologic agent of chancroid, a genital ulcer disease transmitted by sexual contact (reviewed in references 2 and 50). Chancroid has gained importance recently because it has been implicated as an independent risk factor for the heterosexual transmission of human immunodeficiency virus (HIV) in Africa (2, 18, 25, 50, 54; F. A. Plummer, M. A. Wainberg, P. Plourde, P. Jessamine, L. J. DaCosta, I. A. Wamola, and A. R. Ronald, J. Infect. Dis. 161:810-811, 1990 [Letter]).

Several facets of H. ducreyi biology are interesting. H. ducreyi is an obligate human pathogen and its growth is fastidious and slow in vitro. It is unable to synthesize heme (3) and must obtain heme compounds from its only known host, humans, presumably by use of its heme or hemoglobin receptors (14, 15, 40, 45). Free heme, hemoglobin, or catalase (16, 48) can supply the heme requirement for H. ducreyi in vitro; however, the inability of a hemoglobin receptor mutant to initiate disease in the human experimental model of H. ducreyi infection implicates hemoglobin as the most important source of heme in vivo (4a).

Serum resistance has been shown in numerous bacterial systems to be critical for the survival of certain invading bacteria and the establishment of disease, since mutations which result in the loss of serum resistance render several bacterial pathogens avirulent (6, 10, 28, 33, 42). In most systems, the serum-resistant phenotype requires the product of more than one gene. H. ducreyi is resistant to high levels of normal human serum (NHS) (up to 50%). Early studies of H. ducreyi serum resistance by Odumeru et al. led to the conclusion that truncation of lipooligosaccharide (LOS) in several strains (including strains CIP A75 and CIP A77 used in this study) is associated with avirulence and loss of serum resistance (30-32). However, a recent study by Hiltke et al. (22) came to the opposite conclusion. The impetus for the present study was our observation that one of Odumeru's serum susceptible strains (CIP A75) (31) lacked a major outer membrane protein common to serum-resistant strains. The objective of this study was to characterize the role of this protein, which we termed DsrA, in the serum-resistant phenotype of H. ducreyi.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and media. The bacterial strains used in this study are listed in Table 1. It should be noted that two different strains of CIP 542 were used in this study (Table 1); CIP 542 (CAN) (4) is avirulent, whereas CIP 542 (CDC) is virulent (49). For routine growth, H. ducreyi was maintained on chocolate agar plates obtained from the University of North Carolina Hospital Clinical Microbiology Laboratory. This medium was prepared using Mueller-Hinton base as specified by the manufacturer and contained no fetal calf serum. When 5% fetal calf serum was required for optimal growth (strains CHIA and V-1157), gonococcal medium base (GCB) (Difco) was used instead of Mueller-Hinton base and prepared as specified by the manufacturer. Both chocolate agar media contained Iso VitaleX and 1% autoclaved hemoglobin. Large-scale cultures of H. ducreyi were grown in Fernbach flasks with 1 liter of GCB broth containing 5% fetal calf serum, 1% IsoVitaleX, and 50 µg of heme per ml (14). For H. ducreyi, the antibiotics used included chloramphenicol (1 µg/ml) or streptomycin (100 µg/ml), which were incorporated into the GCB chocolate agar.

Outer membrane isolation and analysis, SDS-PAGE, LOS, and immunoblotting. Outer membranes were harvested as previously described (14). Protein concentrations were determined using the bicinchoninic acid kit from Pierce (Rockford, Ill.). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting were performed as previously described (14). The LOS of H. ducreyi was prepared using the method of Hitchcock and Brown (23). SDS-PAGE and silver staining (51) or Western blotting with monoclonal antibody (MAb) 3F11 (5) was used to analyze LOS.

N-terminal amino acid sequence determination. The N-terminal amino acid sequence of DsrA was determined for strain 35000. Outer membranes were subjected to preparative SDS-PAGE and Western transfer to polyvinylidene difluoridine membranes. The blot was stained temporarily with Ponceau S protein stain to locate the DsrA protein, which in strain 35000 migrates just below the 30-kDa standard protein. Antibodies were used in Western blots to unequivocally identify the proper band for sequencing; strips of the blot were probed with rabbit anti-OpaF of gonococcal strain FA1090 and with MAb 5C9. Anti-OpaF, for unknown reasons, cross-reacts with DsrA and MAb 5C9 reacts with a previously described H. ducreyi lipoprotein (termed Hlp) with a similar molecular mass (21). The corresponding 30-kDa OpaF-reactive band from the remainder of the Ponceau S-stained blot was sequenced.

Production of anti-DsrA antibodies. The antiserum to DsrA was produced as follows. Outer membranes from H. ducreyi strain 35000 were electrophoresed on large preparative SDS-PAGE gels (12% polyacrylamide), which were briefly stained, and the corresponding 30-kDa band was excised and electroeluted using a Centrilutor apparatus (Amicon) as specified by the manufacturer. Mice were immunized a total of three times at 3-week intervals with 25 µg of gel-purified protein per immunization. Freund's complete adjuvant was used for the first immunization, and Freund's incomplete adjuvant was used for the remaining immunizations. Serum was collected 2 weeks after the final immunization.

Chemicals and reagents. All chemicals and reagents, unless otherwise noted, were from Sigma Chemical Co. (St. Louis, Mo.).

DNA manipulations. Standard recombinant DNA methods were used as described by Sambrook et al. (38) or as specified by the manufacturer.

V-A PCR. Two degenerate oligonucleotides (no. 6 and 7 in Table 2; also see Fig. 2) deduced from the N-terminal amino acid sequence specifically hybridized to a 1.1-kb EcoRI genomic fragment (data not shown). Attempts to establish a replicating plasmid containing this fragment by using size-selected DNA ligated into several plasmid vectors were unsuccessful. Therefore, a series of three separate vector-anchored PCR (V-A PCR) strategies were used to obtain sequencing templates of the dsrA structural gene, upstream flanking DNA, and downstream flanking DNA. When possible, we established clones for these PCR products (Table 1 and 2). The first V-A PCR (see Fig. 2, V-A PCR 1) used the ligation between the 1.1-kb EcoRI size-selected DNA and vector pBluescript as template and used 5' primer 6 and vector primer KS as amplimers. An approximately 700-bp fragment was amplified, and a preliminary sequence was obtained (this fragment was also cloned to establish pUNCH 1248). The N-terminal sequence originally obtained from Edman degradation matched the deduced amino acid sequence of the PCR product but was not homologous to known sequences in databases. By coincidence, the C terminus ended with a phenylalanine (the terminal phenylalanine) encoded by the TTC of the 3' EcoRI site, GAATTC (thus, there was no stop codon on this EcoRI fragment). This C-terminal domain was similar to UspA2 and YadA (see Results), suggesting the possibility of a PCR-generated artifact(s). To rule out a PCR artifact, additional PCR was performed. The primers used included 5' primers 6, 8, and 9 and 3' primers 11 and 12. The last four primers were derived from the DNA sequence obtained from the original anchored PCR product above (see Fig. 2) (data not shown). We amplified identically sized products from total H. ducreyi chromosomal DNA template and the original V-A PCR 1 product by using 3' primers from the region with homology to C-terminal YadA (primers 11 and 12) (data not shown). Furthermore, Southern hybridization of H. ducreyi chromosomal DNA probed with oligonucleotides 6, 7, 8, 9, 11, and 12 and the PCR EcoRI product generated with oligonucleotides 8 and 12 all specifically recognized the 1.1-kb band (see Fig. 2) (data not shown). Taken together, these data demonstrated that the N-terminal amino acid sequence obtained from the polyvinylidene difluoride blot of the 30-kDa protein is found on the same gene product that has C-terminal homology to virulence factors UspA2 and YadA and provided a motive to examine additional sequences.

DNA sequencing and analysis. DNA sequence analysis was performed at the University of North Carolina at Chapel Hill Automated Sequencing Facility, utilizing Taq terminator chemistry. The final sequences presented for strain 35000 in Fig. 2 and for the other H. ducreyi strains in Fig. 9 were obtained from PCR products with primers 14 and 24, which flank the dsrA gene (see Fig. 2). This PCR product from strain 35000 was also used to create shuttle plasmid pUNCH 1260 (Table 1), which was used for complementation studies. Both strands of the DNA were completely sequenced and assembled using the program Sequencher 3.1 (Gene Codes). The preliminary sequence for the dsrA structural gene from 35000 obtained by vector-anchored PCR was in complete agreement with the final sequence presented (see Fig. 3). Amino acid alignments were done with the Clustal algorithm in the program GeneJockeyII (BIOSOFT, Cambridge, United Kingdom) and PAM 250 setting. Bestfit (GCG Computer Group, Madison, Wis.) was used to generate similarity and identity scores, using a gap weight of 8.

Plasmid constructions. Plasmid pUNCH 1248 was constructed by PCR. A 900-bp fragment was amplified from H. ducreyi strain 35000 using primers 14 and 16 (see Fig. 2) under the conditions described above for the first two vector-anchored PCRs. The product was ligated to pCRII as specified by the manufacturer and transformed into Escherichia coli DH5, and recombinants were identified by restriction analysis. E. coli harboring pUNCH 1248 grew poorly and therefore was propagated only on agar plates to reduce the possibility of mutation or deletion, but it gave rise to an occasional larger colony. Subclone pUNCH 1254 was constructed by isolating the EcoRI fragment of pUNCH 1248 and ligating it into EcoRI-restricted pLS88. The dsrA gene of pUNCH 1254 was mutagenized by insertion of a chloramphenicol acetyltransferase (CAT) gene cassette into the open reading frame (ORF) to construct pUNCH 1255. To perform this, the CAT cassette (a 1-kb BglII fragment from pNC40 treated with Klenow to fill in the ends) was ligated into the NdeI site of pUNCH 1254 (previously treated with Klenow to produce blunt ends). pUNCH 1256 was constructed by moving the insert from pUNCH 1255 (containing mutagenized dsrA) into plasmid pRSM1791 for subsequent mutagenesis. This was done by isolation of a SmaI-HincII fragment of pUNCH 1255, Klenow treatment, and ligation into the NotI site of pRSM1791 previously treated with Klenow. Transformation of E. coli DH5 MCR was performed and selection executed with ampicillin and chloramphenicol, yielding pUNCH 1256.

Construction and characterization of an H. ducreyi dsrA mutant. An isogenic mutant (FX517; Table 1) was constructed by allelic replacement of the wild-type locus of strain 35000 with the mutation in pUNCH 1256, using a previous system of mutagenesis described by Bozue et al. (7). In this procedure, a mutagenized copy of the dsrA locus (containing the CAT cassette) was subcloned into a plasmid expressing the lacZ gene (pUNCH 1256). H. ducreyi cells were electroporated with pUNCH 1256, and Cm^r transformants were selected (19). These transformants putatively contained the entire plasmid integrated due to a single crossover event (as exemplified by FX516 [Table 1]). Individual transformants were streaked onto chloramphenicol medium containing 5-bromo-4-chloro-3-indolyl--D-galactoside (X-Gal) (40 µg/ml). Since the product of lacZ cleavage of X-Gal is highly toxic to H. ducreyi, the cointegrates grow as tiny blue colonies (7). The loss of the lacZ sequences and neighboring wild-type allele via a resolution of the cointegrate results in only the mutant allele being retained (exemplified by FX517 [Table 1]). These H. ducreyi mutants grew as normal-sized white colonies on medium containing chloramphenicol and X-Gal, similar to other H. ducreyi mutants containing CAT cassettes (references 15 and 16 and data not shown).

Complementation of FX517 and other dsrA mutants in trans. To rule out the possibility that the serum susceptibility of dsrA mutant FX517 is due to a cryptic mutation elsewhere on the chromosome or to polar downstream effects, we complemented FX517 in the trans configuration. Briefly, we PCR amplified the dsrA gene and surrounding locus using primers 14 and 24 (see Fig. 2), treated the PCR product with Klenow, and restricted it with HindIII (which restricts just downstream of dsrA [see Fig. 2]). After gel purification, the PCR product was ligated into SmaI-HindIII-restricted pLSKS (55). The ligation was ethanol precipitated, and H. ducreyi strain FX517 was electroporated. Streptomycin-resistant colonies were screened for production of DsrA by Western blotting and confirmed by restriction analysis. One experimental transformant, pUNCH 1260dsrA, and one vector transformant were selected for further study. pUNCH 1260 and the vector pLSKS (negative control) were separately purified from H. ducreyi strain FX517 and were each electroporated into the three naturally occurring dsrA mutants [CIP A75, CIP A77, and CIP 542 (CAN) (Table 1)].

Serum susceptibility. The resistance of H. ducreyi to NHS serum was measured as previously described (8, 31) with the following modifications. An 18- to 24-h culture of H. ducreyi from chocolate agar plates was scraped into GCB broth to an optical density at 600 nm of 0.2. A 10⁴ to 10⁵ dilution was made (approximately 1 to 3,000 CFU/ml, depending on the strain), and aliquots were mixed with pooled fresh NHS (fNHS) or heat-inactivated NHS (NHS) (56°C for 30 min) to a final concentration of 25 or 50% NHS. After incubation for 45 min at 35°C under 5% CO₂, 100-µl aliquots were plated onto chocolate agar plates and viable counts were measured after 48 h. Data are expressed as percent survival in fNHS compared to survival in NHS [(CFU in fHNS/CFU in NHS) × 100]. Strains containing pUNCH 1260 or pLSKS were propagated and plated on chocolate agar containing streptomycin at 100 µg/ml. Strains CHIA and V-1157 grew poorly on Mueller-Hinton chocolate agar without fetal calf serum and were grown on GCB chocolate agar containing 5% fetal calf serum.

Nucleotide sequence accession numbers. Relevant DNA and amino acid sequences have been submitted to GenBank (accession no. AF187001 through AF187009).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of a 30-kDa protein involved in serum resistance. During the course of studies characterizing the H. ducreyi interaction with polymorphonuclear leukocytes, a series of Western blots was performed using various antibodies to the Opa proteins from gonococci. Polyclonal antiserum to OpaF of gonococcal strain FA1090 reacted with a protein that varied between 28 and 35 kDa in a panel of strains (data not shown), which we later designated DsrA. One strain, CIP A75, did not react. CIP A75 was of interest because it had previously been shown to be avirulent in a rabbit model of infection, to be serum susceptible, to exhibit reduced adherence to HEp-2 cells, and to express a truncated LOS (31, 47).

View larger version (32K): [in this window] [in a new window]   FIG. 1.   Distribution of the DsrA protein and serum resistance of selected H. ducreyi strains. (A) Total cellular proteins from H. ducreyi strains were subjected to SDS-PAGE and Western blotting using anti-DsrA mouse sera. Bound antibody was detected with alkaline phosphatase-conjugated secondary antibody and BCIP/NBT substrate. An additional 20 geographically diverse H. ducreyi strains also expressed a 28- to 35-kDa protein which reacted with this serum (data not shown). The names of strains are indicated above each lane. Shown to the left of the gel are molecular mass standards. (B) Bactericidal killing of selected H. ducreyi strains in 50% NHS. H. ducreyi was mixed with either NHS or fNHS for 45 min, and viable counts were determined. The percent survival in fNHS compared to NHS (designated 100%) is shown on the y axis. The data are compiled from separate experiments done on at least three different days.

Molecular studies. Through a series of experiments involving Western blotting, immunoprecipitation, and finally N-terminal amino acid sequencing, we determined that the DsrA protein was not the same as the previously described 28-kDa lipoprotein Hlp (reference 22 and data not shown). The N-terminal amino acid sequence of the immunoreactive 30-kDa DsrA protein of strain 35000 was found to be QQPPKFAGVS SLYSYEYDYG KGKKTKSNEG. No significant homologies were initially detected when this peptide sequence was searched against GenBank, including gonococcal Opa proteins.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Restriction map of the dsrA region and PCR products. The dsrA ORF is boxed. The restriction sites are indicated. The numbered arrows indicate the direction and position of the dsrA oligonucleotides used for PCR. KS and T7 (promoter) refer to the vector primers used in the vector-anchored PCR reactions. Since oligonucleotides 6 and 7 were deduced from the amino acid sequence of the mature, processed protein, their position was placed just slightly inside the ORF. P, promoter. The jagged lines represent approximately 2 kb of sequence not shown downstream of the dsrA locus. The products of V-A PCR 1 and standard PCR were used to generate plasmids pUNCH 1248 and pUNCH 1260, respectively.

DNA sequence and deduced amino acid sequence of the H. ducreyi dsrA locus from strain 35000. The DNA sequence of the dsrA locus, including 100 bp of sequence upstream of the ATG start and 126 bp of sequence downstream of the TAA termination codon, are presented in Fig. 3. The sequence was obtained from PCR products amplified using primers 14 and 24, whose sequence are derived from DNA outside of the dsrA ORF (Fig. 2). Putative promoter elements similar to 35 (TGATAA) and 10 (TATATT) E. coli consensus sequences were identified beginning at nucleotide (nt) 13 (TTGACA) and nt 35 (TAGAAT), respectively, and were separated by 16 nt. A putative ribosome-binding site (TAATGAGG) was found beginning 13 nt upstream of the dsrA start codon. Beginning at nt 913 and ending at nt 946 was an inverted hairpin loop containing 13 matched nucleotides, consistent with a transcription terminator. The gene located immediately downstream of dsrA but in the opposite orientation was an ORF with homology to hypothetical protein HI0107 of the genome sequence of H. influenzae. The G+C content of the 1 kb of DNA sequence presented was 34.5%, consistent with the AT-rich nature of Haemophilus DNA.

View larger version (39K): [in this window] [in a new window]   FIG. 3.   DNA sequence and deduced amino acid sequence of the dsrA locus. The putative 35 and 10 promoter sequences are indicated and underlined. A putative ribosome-binding site is labeled RBS and underlined. A total of 21 amino acids comprising the signal peptide are underlined. The stop codon TAA is indicated by an asterisk. The opposing arrows show a potential stem-loop transcription terminator.

View larger version (33K): [in this window] [in a new window]   FIG. 4.   Comparison of the amino acid sequence of DsrA with related proteins. The amino acid sequences of the C-terminal domains of the DsrA protein of H. ducreyi, the UspA2 protein of M. catarrahalis, and the YadA protein of Y. enterocolitica are compared. Shaded, boxed residues indicate identical or conserved residues.

Construction and characterization of an H. ducreyi dsrA mutant. An isogenic mutant (FX517; Table 1) was constructed by allelic replacement of the wild-type dsrA locus of strain 35000. Initial attempts to obtain a double crossover with a CAT cassette in the cloned dsrA gene were unsuccessful when pUNCH 1255 was used (data not shown). Therefore, we used a recently described two-stage method to obtain mutants (7) (see Materials and Methods). Using this procedure, several chloramphenicol-resistant cointegrates were obtained. After each cointegrate was streaked onto X-Gal chocolate plates, several mutants were obtained for each cointegrate, none of which expressed dsrA (data not shown). One mutant, FX517, was selected for further study. Outer membranes were prepared from the parent and mutant strain FX517 and subjected to SDS-PAGE and Coomassie staining or SDS-PAGE and Western blotting (Fig. 5A and B, respectively). DsrA, an abundant outer membrane protein in strain 35000, was absent in the mutant. No reactivity was obtained from FX517 when anti-DsrA antiserum (Fig. 5B) or anti-OpaF (data not shown) was used.

View larger version (71K): [in this window] [in a new window]   FIG. 5.   SDS-PAGE and Western blotting of parent strain 35000 and dsrA mutant FX517. (A) Outer membranes were prepared, solubilized at 37 or 100°C, and subjected to SDS-PAGE and Coomassie blue staining. (B) For the Western blot, outer membranes were solubilized at 100°C, transferred to nitrocellulose, and probed with anti-DsrA mouse serum. Bound antibody was detected with alkaline phosphatase-conjugated secondary antibody and BCIP/NBT substrate. The asterisks indicate the positions of the DsrA protein. STD, molecular weight standards.

Serum resistance phenotype of the dsrA mutant. The serum susceptibility of the naturally occurring dsrA mutants and the role of the related YadA and UspA2 proteins in mediating serum resistance prompted us to test FX517 for serum sensitivity. Studies of serum killing of parent strain 35000 and dsrA mutant FX517 were performed using 25 and 50% pooled NHS (Fig. 6). FX517 was highly susceptible to NHS and demonstrated 0 and 2% survival in 50 and 25% NHS, respectively. In contrast, parent strain 35000 was relatively serum resistant, exhibiting 79 and 50% survival in 50 and 25% NHS (P = 0.002 and 0.004 for 50 and 25% NHS, respectively, using Student's paired t test). Thus, DsrA was essential for expression of a serum-resistant phenotype.

View larger version (19K): [in this window] [in a new window]   FIG. 6.   Bactericidal killing of parent strain 35000 and dsrA mutant FX517. Bactericidal killing was performed as in Fig. 1, except that two serum concentrations were used. The data presented in Fig. 1 and 6 for strain 35000 with 50% sera are the same data. The data presented for FX517 were obtained in parallel experiments with strain 35000.

Complementation of dsrA mutants. It was possible that a cryptic mutation (other than the dsrA mutation) had occurred during the construction of FX517, which could account for its serum susceptible phenotype. Furthermore, we wished to determine whether the serum susceptibility of FX517 and the three naturally occurring dsrA mutants could be converted to serum resistance if they expressed dsrA. Each dsrA mutant [isogenic mutant FX517 or naturally occurring mutants CIP A75, CIP A77, and CIP 542 (CAN)] was electroporated with pUNCH 1260 (dsrA) or pLSKS (vector control) plasmids. These shuttle plasmids are able to replicate in H. ducreyi. Strains containing pUNCH 1260, but not pLSKS, expressed dsrA (Fig. 7A). Expression of dsrA from plasmid pUNCH 1260 suggested that the tentatively identified promoter (Fig. 3) was driving expression of the cloned dsrA gene, since very little additional upstream DNA was present and the insert was in the opposite direction with respect to the lacZ promoter in pLSKS.

View larger version (44K): [in this window] [in a new window]   FIG. 7.   Complementation of dsrA mutants and bactericidal killing. (A) Total cellular proteins from the indicated H. ducreyi strains were subjected to SDS-PAGE (12% polyacrylamide) and Western blotting with anti-DsrA antisera. Bound antibody was detected with horseradish peroxidase-conjugated secondary antibody followed by chemiluminescence. Plasmids: A, pUNCH 1260 (contains the entire dsrA ORF from strain 35000 and its putative native promoter as illustrated in Fig. 2); B, pLSKS (vector without insert). (B) Bactericidal killing of the complemented dsrA mutants. Bactericidal killing was performed as in the experiment in Fig. 1 (50% serum), except that the medium used contained streptomycin. %, percent survival. The data are compiled from separate experiments done on at least three different days.

LOS expression by H. ducreyi. In some bacterial systems, mutants with mutations in LOS are more serum susceptible than are wild-type strains. Indeed, the serum susceptible phenotype of H. ducreyi strains CIP A75 and CIP A77 was attributed to their LOS truncation. It was possible that the lack of dsrA expression in dsrA mutants [FX517, CIP A75, CIP A77, and CIP 542 (CAN)] resulted in the truncation of LOS directly or indirectly. Alternatively, repair of dsrA expression in LOS dsrA apparent double mutants (CIP A75 and CIP A77) might affect LOS expression and subsequent serum susceptibility. To address these possibilities, LOS were analyzed by SDS-PAGE and silver staining (Fig. 8) and in Western blots (data not shown). We compared 35000 and FX517 LOS (without plasmids) in several silver-stained gels and found that the migration patterns were similar. Furthermore, Western blots of 35000 and FX517 LOS with anti-LOS MAb 3F11 (5) were similar (data not shown).

View larger version (33K): [in this window] [in a new window]   FIG. 8.   Analysis of H. ducreyi LOS by SDS-PAGE. Crude LOS was prepared as described in the text and subjected to SDS-PAGE and silver staining. None, no plasmid present; A, pUNCH 1260 (contains the entire dsrA ORF from strain 35000 and its putative native promoter as illustrated in Fig. 2); B, pLSKS (vector without insert).

Structural analysis of dsrA in other H. ducreyi strains. Western blotting of a variety of H. ducreyi strains (Fig. 1A) strongly suggested that DsrA varied in molecular mass and/or amino acid sequence(s) among the strains. Furthermore, we desired to understand whether mutations had occurred in the naturally occurring dsrA mutants or whether the possibility of phase variation could account for their inability to express dsrA. PCR was used to amplify a 1.2-kb fragment containing dsrA from eight additional strains, including the dsrA mutants (Fig. 2, primers 14 and 24). The deduced amino acid sequence indicated that, overall, the DsrA protein was quite similar between strains (Fig. 9). Two regions with modest variability were observed and designated variable regions 1 and 2 (VR1 and VR2). VR1 included, roughly, amino acids 90 to 100 (depending on the strain), and a few substitutions and insertions were noted. VR2 contained either one, two, or three identical copies of the heptamer repeat sequence NTHNINK and spanned amino acids 174 to 195 in the various strains. It is likely that the different number of repeat sequences was the predominant factor accounting for the variable migration seen in SDS-PAGE and Western blotting. Except for the DsrA protein in mutant strain CIP 542 (CAN), which contained a stop codon (see below), the sequences of all other eight DsrA proteins were identical after VR2, the same region which was similar to UspA2 and YadA. We conclude that DsrA is highly conserved in sequence, despite its variable mobility in gels.

View larger version (123111K): [in this window] [in a new window]   FIG. 9.   Comparison of the deduced amino acid sequences of dsrA from nine H. ducreyi strains. VR1 and VR2 are indicated. Shaded, boxed residues indicate identical or conserved residues.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We identified and characterized a protein, DsrA, whose expression is required for serum resistance in H. ducreyi. The following evidence supports this conclusion. (i) The H. ducreyi dsrA isogenic mutant FX517 is serum susceptible compared to its parent. (ii) Three naturally occurring serum-susceptible strains, CIP A75, CIP A77, and CIP 542 (CAN), failed to express dsrA or expressed reduced amounts of dsrA. (iii) When each mutant was complemented in trans with dsrA from strain 35000, each expressed abundant amounts of dsrA and became serum resistant.

In data not presented here, introduction of the dsrA shuttle plasmid pUNCH 1260 into E. coli resulted in modest expression of dsrA compared to H. ducreyi. It did not render E. coli serum resistant (data not shown). It is possible that in E. coli the DsrA protein is not assembled or exported to the outer membrane properly or that additional H. ducreyi components, absent in E. coli, are required for expression of serum resistance. Other H. ducreyi proteins required for full expression of serum resistance include the simultaneous expression of both LspA proteins (53; C. Ward and E. Hansen, Fifth Int. Symp. Haemophilus ducreyi Pathog. Chancroid) and the major outer membrane protein (26). Further studies are needed in this area.

Previously, Odumeru et al. concluded that truncation of LOS was the reason for the serum susceptibility of certain avirulent strains including CIP A75 and CIP A77 (31, 32). Our results and those obtained just recently by others (22; R. Munson, Jr., personal communication) are in direct contradiction to those of Odumeru et al. In the present study, we found that expression of dsrA from plasmid pUNCH 1260 conferred serum resistance to LOS mutants CIP A75 and CIP A77 without affecting their LOS composition. Furthermore, isogenic dsrA mutant FX517 was serum susceptible and its LOS profile was indistinguishable from that of parent strain 35000. In a recent study by Hiltke et al. (22), serum susceptibility in strain 35000 was unaffected by mutations in LOS. In addition, Munson and colleagues have recently identified the LOS mutation of strain CIP A77 as galactosyltransferase (Munson, personal communication). Upon introduction of a shuttle plasmid expressing galactosyltransferase back into CIP A77, full-length LOS was synthesized; however, the strain remained serum susceptible. These results suggest that truncations in LOS have little or no effect on the serum susceptibility of H. ducreyi.

H. ducreyi requires hemoglobin to establish infection in the human model of experimental infection, since a mutant unable to utilize hemoglobin is unable to form pustules or to be recovered (4a). Since chancroid ulcers bleed readily (27, 36), hemoglobin is present to supply the heme requirement of H. ducreyi. However, in addition to hemoglobin, blood contains the potent bactericidal antibody-complement system capable of lysing gram-negative bacteria. Therefore, we speculate that H. ducreyi has evolved, or acquired through horizontal transfer, dsrA to resist the bactericidal activity of serum during its requisite hemoglobin acquisition.

In the present study, we describe a new member of a family of proteins that is present in a variety of gram-negative bacteria, since several bacterial genomes contain genetic information to encode protein members with similar sequences. Partial or complete ORFs with significant homology to the conserved C terminus of DsrA were found in the genome-sequencing projects from Actinobacillus actinomycetemcomitans, Pasteurella multocida, and H. influenzae Rd, all three of which are members of the family Pasteurellaecea (data not shown). However, Western blotting of a variety of 13 typeable and nontypeable H. influenzae strains failed to detect expression of DsrA, and several frameshifts exist in the H. influenzae Rd sequence (17), predicted to encode a DsrA-related protein (data not shown).

Similar to uspA2 mutants, dsrA mutants are serum susceptible. In M. catarrhalis, serum resistance is strongly correlated with disease isolates rather than strains isolated from normal carriers (24); likewise, in the present study of H. ducreyi, serum resistance was correlated with virulent strains and dsrA expression. Furthermore, active immunization with purified UspA proteins (9) or passive administration of a MAb which recognizes the UspA proteins (20) resulted in protection in animal studies. In the Y. enterocolitica system, passive immunization with anti-YadA immunoglobulin G resulted in homologous but not heterologous protection (52). It remains to be tested whether DsrA will also serve as an effective vaccine.

DsrA is widely distributed and immunologically conserved in H. ducreyi. The amino acid sequences of DsrA from nine strains indicate a high degree of conservation, suggesting that it serves an important function. The apparent differences in molecular mass observed by SDS-PAGE and in Western blots could be explained by different numbers of the repeated sequence NTHNINK in VR2 and partially by the minor variability in VR1. The function of the repeat sequence(s) present in VR2 is currently under investigation.

The lesions in the three naturally occurring dsrA mutants were identified. Interestingly, CIP A75 and CIP A77 contained identical deduced DsrA amino acid sequences, an identical mutant LOS phenotype on silver-stained gels, and identical dsrA nucleotide sequences, including the identical 5-bp promoter region deletion. In data not presented here, CIP A75 and CIP A77 total protein profiles analyzed by SDS-PAGE on large gels and Coomassie staining were indistinguishable and suggest that these strains might actually be representatives of the same isolate. We studied two strains of CIP 542, a serum-susceptible, avirulent strain from Canada [CIP 542 (CAN)] and a serum-resistant, virulent strain from the Centers for Disease Control and Prevention [CIP 542 (CDC)]. CIP 542 from the American Type Culture Collection [CIP 542 (ATCC)] is designated the type strain for H. ducreyi. We tested CIP 542 (ATCC) for DsrA in Western blots to see if it expressed a DsrA phenotype similar to CIP 542 (CAN) or CIP 542 (CDC). To our surprise, we found that CIP 542 (ATCC) made abundant DsrA but that the DsrA from it migrated more slowly than did the DsrA from CIP 542 (CDC) (data not shown). Thus, we have identified three different CIP 542 strains based on their DsrA expression phenotype. These results suggest that these strains are different and that they should be used with caution, and they suggest a possible reason for incongruous results previously reported with CIP 542.

The absence or reduced synthesis of the DsrA protein in avirulent, serum-susceptible strains suggests that it is required for virulence. However, additional virulence studies are required with isogenic parent-mutant pairs such as 35000 and FX517 in the appropriate animal and human models of H. ducreyi infection. Taken together, these data strongly suggest a role for DsrA in the pathogenesis of chancroid and as a potential virulence factor and vaccine candidate.