Immunization with the Haemophilus ducreyi Hemoglobin Receptor HgbA with Adjuvant Monophosphoryl Lipid A Protects Swine from a Homologous but Not a Heterologous Challenge

Abstract

Haemophilus ducreyi, the etiological agent of chancroid, has a strict requirement for heme, which it acquires from its only natural host, humans. Previously, we showed that a vaccine preparation containing the native hemoglobin receptor HgbA purified from H. ducreyi class I strain 35000HP (nHgbA_I) and administered with Freund's adjuvant provided complete protection against a homologous challenge. In the current study, we investigated whether nHgbA_I dispensed with monophosphoryl lipid A (MPL), an adjuvant approved for use in humans, offered protection against a challenge with H. ducreyi strain 35000HP expressing either class I or class II HgbA (35000HPhgbA_I and 35000HPhgbA_II, respectively). Pigs immunized with the nHgbA_I/MPL vaccine were protected against a challenge from homologous H. ducreyi strain 35000HPhgbA_I but not heterologous strain 35000HPhgbA_II, as evidenced by the isolation of only strain 35000HPhgbA_II from nHgbA_I-immunized pigs. Furthermore, histological analysis of the lesions showed striking differences between mock-immunized and nHgbA_I-immunized animals challenged with strains 35000HPhgbA_I but not those challenged with strain 35000HPhgbA_II. Mock-immunized pigs were not protected from a challenge by either strain. The enzyme-linked immunosorbent assay (ELISA) activity of the nHgbA_I/MPL antiserum was lower than the activity of antiserum from animals immunized with the nHgbA_I/Freund's vaccine; however, anti-nHgbA_I from both studies bound whole cells of 35000HPhgbA_I better than 35000HPhgbA_II and partially blocked hemoglobin binding to nHgbA_I. In conclusion, despite eliciting lower antibody ELISA activity than the nHgbA_I/Freund's, the nHgbA_I/MPL vaccine provided protection against a challenge with homologous but not heterologous H. ducreyi, suggesting that a bivalent HgbA vaccine may be needed.

Chancroid is one of the genital ulcer diseases and is transmitted through sexual contact. Lesions caused by chancroid initially appear as papules, which evolve within several days into pustules. If left untreated, chancroid pustules develop into painful, bleeding ulcers with soft, irregular borders. Chancroid is prevalent in certain developing countries but is rarely found in the United States (47, 62). Several studies have shown that chancroid serves as an important independent cofactor in the heterosexual transmission of HIV where both diseases are endemic (29, 35, 48, 51, 66). Commercial sex workers serve as the reservoir of chancroid, and control of disease in this population strikingly reduces the number of cases of chancroid in their male clients (58). Thus, one possible approach to control chancroid is to implement a limited vaccination program to control infection in this reservoir; however, no vaccine for chancroid currently exists.

Haemophilus ducreyi, the etiologic agent of chancroid, is a fastidious Gram-negative bacterium and a strict human pathogen. An interesting biologic feature of H. ducreyi is its obligate requirement for heme. Heme (for H. ducreyi) and iron are critical nutrients required for most pathogenic bacteria. Many Gram-negative bacteria obtain heme/Fe through systems that include TonB-dependent outer membrane receptors specific for heme/Fe compounds. The relatively small genome of prototypical H. ducreyi strain 35000HP encodes only three TonB-dependent receptors; in comparison, other bacterial genomes can encode more than 30 (14). Using isogenic mutants, the Spinola and Elkins laboratories surveyed the ability of H. ducreyi TonB-dependent receptor mutants to initiate infection in the human experimental model of chancroid. An H. ducreyi mutant that does not express the gene encoding the hemoglobin (Hb) receptor, hgbA, did not establish human infection (3). In contrast, an isogenic double mutant lacking the genes encoding the two other TonB-dependent receptors of H. ducreyi, tdhA and tdX, was fully virulent, indicating that HgbA is the only TonB-dependent receptor of H. ducreyi to be a virulence factor in the human experimental model of chancroid (38). Since HgbA is required for the utilization of heme from Hb by H. ducreyi (17), these data suggest that Hb is the most important source of heme in the early stages of the human experimental model of chancroid and that HgbA is a potential vaccine candidate.

HgbA is a large, 100-kDa outer membrane protein that has a complex structure similar to that of other TonB-dependent receptors whose structure has been solved (10, 12, 20, 21, 40). HgbA is believed to contain 22 transmembrane beta sheets and 11 putatively surface-exposed loops. A recent study in our laboratory using H. ducreyi mutants expressing single loop deletions in HgbA provided evidence for surface exposure of loops 4, 5, 6, and 7 (44). Moreover, deletions of loops 5 and 7 but not of the other 9 loops of HgbA abrogated the binding of human Hb to HgbA. We also found that IgG from pigs immunized with native HgbA (nHgbA) bound loops 4, 5, and 7 and that antibodies directed at loops 4 and 5 partially blocked Hb binding to HgbA in vitro. Thus, a central domain of the primary sequence of HgbA is immunogenic, required for binding Hb, and surface exposed.

H. ducreyi strains exist in two groups, designated class I and class II, based on striking primary sequence differences in certain outer membrane proteins, such as DsrA (ducreyi serum resistance A) and NcaA (necessary for collagen adhesion A), and on their lipooligosaccharide (LOS) structures (49, 50, 53, 67). In contrast, the HgbA proteins of different classes of H. ducreyi strains are more than 95% identical. Prototypical H. ducreyi strain 35000HP, a class I isolate, is the strain used for most studies, including isogenic mutant construction and the experimental human model of chancroid. 35000HP is the only H. ducreyi strain whose genome has been sequenced.

Previously, we showed that immunization of swine with native HgbA from class I strain 35000HP (nHgbA_I) in Freund's adjuvant provided complete protection from a homologous challenge infection with H. ducreyi strain 35000HP (1). The antibodies elicited by nHgbA_I/Freund's showed modest bactericidal activity, bound to the cell surface of both class I and class II H. ducreyi strains, and partially blocked Hb binding to nHgbA (1). nHgbA_I antisera did not recognize the surface of, nor did they show bactericidal activity against the isogenic hgbA mutant, demonstrating specificity of the humoral response to HgbA.

In the current study, we pursued two objectives. First, we investigated the effectiveness of monophosphoryl lipid A (MPL), an adjuvant approved for use in humans, to elicit an immune response to the nHgbA_I vaccine that is protective against an H. ducreyi challenge in the experimental swine model of chancroid. Second, we examined the ability of this vaccine to protect swine from a challenge infection with H. ducreyi strain 35000HP expressing either class I hgbA (hgbA_I) or class II hgbA (hgbA_II) from H. ducreyi strain DMC111 (homologous versus heterologous challenge, respectively).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

H. ducreyi strain 35000HP (human-passaged variant, class I strain) (4, 25) used in this study was obtained from Stanley Spinola, Indiana University, Indianapolis, IN. H. ducreyi class II strain DMC111 is an isolate from Bangladesh (67). The construction of isogenic hgbA_I mutant H. ducreyi strain FX547 was previously described (44). FX547 contains a complete deletion of the hgbA_I open reading frame, replaced with a chloramphenicol resistance cassette insert, and cannot grow on Hb plates (100 μg/ml) as a sole source of heme (44). For routine growth, H. ducreyi strains were maintained on chocolate agar plates containing gonococcal medium base (GCB; Difco, Detroit, MI) and 1% bovine Hb (Becton Dickinson, Sparks, MD) and supplemented with 1% GGC (0.1% glucose, 0.001% glutamine, 0.026% cysteine) and 5% Fetalplex (Gemini Bio-Products, West Sacramento, CA). Cultures were incubated at 34.5°C with 5% CO₂.

Construction of H. ducreyi strain 35000HP expressing the hgbA_II gene from strain DMC111.

A class I H. ducreyi strain expressing the class II hgbA gene (hgbA_II), 35000HPhgbA_II, was constructed in the following fashion: PCR was used to amplify hgbA_II from H. ducreyi strain DMC111 using class II-specific hgbA primers that contain restriction sites (underlined) (5′-TAACACTTAAGGAATACGTAATGAAAACGAATAAACTC-3′ and 5′-GAGAAATCATATGACAAAAAAGGGCACCGAAGT-3′) and cloned into the AflII and NdeI sites of pUNCH1411 (44) to create pUNCH1413. pUNCH1413 contains the open reading frame of class II hgbA with class I hgbA flanking sequences. pUNCH1413 DNA was digested with EcoRI and XhoI, and the hgbA fragment isolated. This DNA fragment was electroporated into H. ducreyi FX547, and cells plated on Hb agar plates. Transformants that grew on Hb as a sole source of heme were screened for the expression of HgbA by Western blotting. A single colony was chosen, isolated, and sequenced. Two mutations were detected during sequencing. The first mutation was silent and the second mutation changed amino acid 21 of the mature protein from valine to glutamic acid. This change is in the “plug” region of mature HgbA and is thought not to be surface exposed. The HgbA_II protein functions normally for growth on medium with Hb as the sole heme source in 35000HP, and H. ducreyi strain 35000HPhgbA_II is fully virulent in the swine model. For the purpose of this report, wild-type H. ducreyi strain 35000HP will be referred to as 35000HPhgbA_I. To ensure that HgbA_II was expressed at the same level as HgbA_I in the 35000HP background, outer membrane proteins were prepared from low-heme liquid cultures of strains 35000hgbA_I, 35000HPhgbA_II, and the hgbA isogenic null mutant 35000HPΔhgbA and subjected to SDS-PAGE followed by Coomassie staining. HgbA_II is expressed at a level similar to that of HgbA_I in the 35000HP background in both whole cells and outer membrane protein preparations (data not shown).

HgbA purification.

Purification of nHgbA_I and nHgbA_II from H. ducreyi strains 35000HPhgbA_I and 35000HPhgbA_II was performed as previously described (1). The purity of both nHgbA_I and nHgbA_II was greater than 95% (data not shown).

Animals.

Eight Yorkshire Cross pigs were used in two separate immunization experiments (four pigs per experiment). Pigs were obtained at 3 weeks of age and housed at ambient temperature (20° to 25°C) in individual pens at the North Carolina State University (NCSU) School of Veterinary Medicine. Pigs were given water and antibiotic-free high-protein feed ad libitum starting 6 weeks prior to the start of and continuing throughout the study. During inoculation and biopsy procedures, pigs were sedated with 2 mg of ketamine HCl (Fort Dodge Laboratories, Fort Dodge, IA) and 2 mg of xylazine (Miles Laboratories, Shawnee Mission, KS) per kg of body weight, injected intramuscularly. Pigs at challenge generally weighed between 70 and 100 lbs. The Institutional Animal Care and Use Committee (IACUC) at NCSU approved the use of animals for these experiments.

Immunization protocol.

The immunization protocol used in this study was exactly replicated from the original nHgbA_I vaccine trial so that the results from the current study would be directly comparable to the one from the previous study (1). In each of 2 experiments, 2 pigs were mock immunized with buffer (1% octylglucoside [Calbiochem, La Jolla, CA] in phosphate-buffered saline [PBS]) and adjuvant alone and 2 pigs received nHgbA_I in buffer and adjuvant. The MPL adjuvant (Sigma, St. Louis, MO) was prepared according to the manufacturer's instructions. The immunogen was prepared by mixing 250 μg of nHgbA_I in 500 μl PBS containing 1% octylglucoside with 500 μl of MPL adjuvant. Each pig received three intramuscular injections in the nuchal region in two sites 21 days apart. Pigs were bled prior to each of the three immunizations, as well as before challenge and biopsies. Pigs were euthanized immediately following biopsies in accordance with IACUC protocols.

Challenge infection.

Challenge infection was performed 21 days after the third immunization. H. ducreyi strains 35000HPhgbA_I and 35000HPhgbA_II were grown overnight at 34.5°C and 5% CO₂ on chocolate agar plates supplemented with 1% GGC and 5% FetalPlex. For each strain, a cell suspension at an optical density at 600 nm (OD₆₀₀) of 1.5 (approximately 1 × 10⁹ CFU/ml) was prepared in GC broth, along with a 1:10 dilution (approximately 1 × 10⁸ CFU/ml). Multi-Test skin test applicators (Lincoln Diagnostics, Decatur, IL) were used to inoculate 10 μl of each bacterial suspension at multiple sites on the pig ears. Since the efficiency of dose delivery by the Multi-Test skin test applicator has been estimated at 1:1,000, 10 μl of a 10⁹ CFU/ml bacterial suspension delivers approximately 10⁴ CFU, and 10 μl of a 10⁸ CFU/ml bacterial suspension delivers an estimated 10³ CFU dose (55). Prior to inoculation, the ears of the pigs were cleaned thoroughly with alcohol wipes and lesion sites were marked with an ethanol-resistant pen. Each pig was inoculated with H. ducreyi strain 35000HPhgbA_I on the left ear and strain 35000HPhgbA_II on the right ear. The viability of bacterial suspensions was determined prior to and after each experiment, and no decrease in CFU was observed.

Biopsy processing.

Pig ear lesions were biopsied 1 week after infection using disposable 6-mm skin biopsy punches (Acuderm, Ft. Lauderdale, FL). Biopsied tissues were either placed in 0.5 ml of GC broth for bacterial recovery or in 0.5 ml of 4% paraformaldehyde for histological studies. In order to recover H. ducreyi from biopsied lesions, tissues were thoroughly minced in GC broth using a sterile scalpel and spread on chocolate agar plates supplemented with 1% GGC, 5% FetalPlex, and 3 μg/ml vancomycin to reduce contaminants (1). Plates were incubated at 34.5°C with 5% CO₂ for up to 72 h, the presence/absence of H. ducreyi colonies was noted, and representative colonies were subcultured for confirmatory PCR.

Sections of biopsy specimens preserved in 4% paraformaldehyde were stained with hematoxylin and eosin (H&E) (Histology Laboratory, College of Veterinary Medicine, North Carolina State University). Each slide was coded and graded blindly by 2 authors according to a previously described protocol (1, 52). The appearance of normal skin received a score of 1; the presence of perivascular and interstitial mononuclear cell infiltrate received a score of 2; the presence of an intraepidermal pustule with neutrophils, fibrin, and necrotic debris received a score of 3; the presence of an epidermal pustule with keratinocyte cytopathology and mononuclear and polymorphonuclear infiltrate received a score of 4; and ulceration or epidermal necrosis and dermal erosion with confluence of immune cells was scored a 5. All slides were viewed using a Leica DM IRB inverted microscope (Leica Microsystems, Bannockburn, IL), and images were saved using QCapture software (QImaging, Surrey, BC, Canada). The scores were compared using Cohen's kappa, which indicated that the similarity between the evaluators (κ = 0.541) was in moderate agreement.

ELISA.

Enzyme-linked immunosorbent assays (ELISAs) were performed as previously described (1), except that plates were coated with 100 ng/well of nHgbA_I instead of 200 ng/well. Each sample was run in duplicate, and the assay was performed on at least 3 different days.

Hb-blocking ELISA.

An ELISA was used to determine the ability of anti-HgbA_I to block Hb binding to HgbA (1). One hundred nanograms of nHgbA_I or nHgbA_II in 0.1 M carbonate buffer (pH 9.5) was added to wells and incubated overnight at 4°C. The next morning, wells were blocked for 1 h with 2% bovine serum albumin (BSA)-PBS. Twenty micrograms of anti-nHgbA_I IgG was added to wells before a 30-min incubation at room temperature. Digoxigenin (DIG)-labeled human Hb (400 ng in 1% BSA-PBS) was added to wells, and plates were incubated at room temperature for 1 h. Three washes with PBS-0.05% Tween 20 (PBS-T) were followed by the addition of alkaline phosphatase-conjugated anti-DIG antibodies (100 μl, 1:5,000 in PBS) (Roche Diagnostics, Indianapolis, IN). The plates were incubated for 1 h at room temperature and washed thrice with PBS-T. One hundred microliters of alkaline phosphatase substrate (1-Step PNPP [p-nitrophenyl phosphate]) (Pierce, Rockford, IL) was added to wells, and the optical density at 405 nm was measured after 45 min using a 1420 VICTOR² multilabel counter (Wallac Oy). The Hb was labeled with DIG according to the manufacturer's instructions (DIG protein labeling kit; Roche Diagnostics, Indianapolis, IN). IgG was purified from pig antisera using protein A/G resin according to the manufacturer's instructions (ExAlpha Biologicals, Watertown, MA).

Whole-cell binding ELISA.

H. ducreyi strains 35000HPhgbA_I and 35000HPhgbA_II, isogenic hgbA_I mutant FX547, and H. ducreyi class II strain DMC111 were grown overnight at 34.5°C with 5% CO₂ and shaking for 15 h in GC broth supplemented with 1% GGC and 5% Fetalplex. The whole-cell binding ELISA was performed as previously described (1) with the following changes: two hundred microliters of each strain of H. ducreyi (OD₆₀₀ = 0.2) was added in triplicate to a Multiscreen HTS 96-well plate (Millipore, Billerica, MA). Fifty microliters of anti-nHgbA antisera was added to wells (final dilution 1:2,000 in 0.25% Tween 20-GC broth) and incubated for 90 min at room temperature with gentle rocking. Wells were washed 4 times with 0.1% Tween 20-GC broth using a vacuum manifold. One hundred microliters of rabbit anti-pig IgG-horseradish peroxidase (HRP) (product no. A5670; Sigma, St. Louis, MO), diluted 1:20,000 in 0.25% Tween 20-GC broth, was added to wells, and the plate incubated at room temperature for 60 min with gentle rocking. Wells were washed 4 times with 0.1% Tween 20-GC broth, and 100 μl of HRP substrate (Amersham ECL; GE Healthcare, Buckinghamshire, United Kingdom) was added to wells. Chemiluminescence was detected using the Wallac 1420 VICTOR² plate reader.

Bactericidal assay.

In order to determine the bactericidal activity of the nHgbA_I antisera against H. ducreyi strains, an immune bactericidal assay was performed as previously described (1, 18). Percent killing was determined by dividing the number of colonies that survived in fresh normal human serum (NHS) by the number of colonies that survived in heat-treated NHS and multiplying by 100. There was no decrease in CFU for the heated NHS control compared to the count in the GC broth control.

PCR amplification of loop 4 of the hgbA gene.

To confirm strain identity, H. ducreyi colonies obtained from biopsy specimens were subjected to PCR using primers specific for the conserved flanking regions of loop 4 hgbA_I and hgbA_II. The upstream and downstream hgbA primers (5′-CTAACCCTTCTGGGCTATAC-3′ and 5′-GCTAGGTAAATACACACGGC-3′, respectively) generated a PCR product of approximately 350 bp from either strain. The PCR steps used for amplification are as follows: an initial 5-min 94°C denaturing step was followed by 30 cycles of 1 min of denaturation at 94°C, annealing for 30 s at 45°C, and extension for 30 s at 72°C. A final polishing step for 5 min at 72°C was used prior to completion. PureTaq Ready-To-Go PCR beads (GE Healthcare, Buckinghamshire, United Kingdom) were used according to the manufacturer's instructions. Ten picomoles of PCR primers were used, along with 1 μl of cells (OD₆₀₀ of 0.2 in 50 μl of water) as template. PCR products were purified using a Qiagen PCR purification kit (Qiagen, Valencia, CA) and sequenced using the upstream PCR primer at the UNC Genome Analysis Facility.

Whole-cell immunoprecipitation.

H. ducreyi strains were grown overnight in heme-limiting conditions to induce the expression of HgbA at 34.5°C in 50-ml broth cultures of GC broth, 5% Fetalplex, 1% GGC in the presence of 5% CO₂ (17). Whole-cell immunoprecipitation was performed as previously described with the following modifications (26, 44, 60): H. ducreyi cultures were centrifuged and the pellets suspended to an OD₆₀₀ of 1.0 (approximately 5 × 10⁸ CFU/ml) in GC broth. Ten microliters of serum was added to 1 ml of the bacterial suspension and rocked at room temperature for 20 min. Bacterial cells were centrifuged, and the supernatant discarded to remove unbound antibody and serum components. The washed cell pellet was resuspended in 100 μl PBS, and 1 ml of 2% Zwittergent 3-14 (ZW 3-14) (Calbiochem) in TEN buffer (50 mM Tris-HCl, 5 mM EDTA, 150 mM NaCl, pH 8.0) was added to solubilize proteins. After incubation at 37°C with agitation for 1 h, the tube was centrifuged for 10 min at 14,000 rpm to remove insoluble debris. The supernatant containing ZW 3-14-soluble total cellular proteins and antigen/antibody complexes was moved to a new tube containing 25 μl of a 50% slurry of protein A/G agarose beads (ExAlpha Biologicals, Shirley, MA). The tubes were incubated for 2 h or overnight to allow binding of antigen/antibody complexes to the protein A/G beads. The tubes were centrifuged and washed thrice using 0.5% ZW 3-14 in TEN. The agarose pellet was then resuspended in 1 ml TEN, moved to a fresh tube, and centrifuged, and the supernatant discarded. Forty microliters of Laemmli sample buffer lacking any reducing agents was added to the washed agarose, the tubes boiled for 5 min at 95°C, and 20 μl of the mixture subjected to SDS-PAGE and Coomassie staining.

Western blotting.

Samples were subjected to SDS-PAGE and transferred onto nitrocellulose, which was blocked in 0.5% Tween 20-PBS for 1 h. Rabbit anti-N-terminal HgbA peptide antiserum diluted at 1:500 was used as the primary antibody (17). The sequence of the peptide used to generate the anti-N-terminal antiserum is identical in both classes of HgbA proteins and should therefore bind HgbA_I and HgbA_II equally. Alkaline phosphatase-conjugated goat anti-rabbit IgG (Sigma, St. Louis, MO) diluted at 1:20,000 was used as the secondary antibody. Protein bands were visualized using Lumi-Phos (Thermo Scientific, Rockford, IL).

Statistics.

Statistical analyses were performed using the SigmaStat program (version 3.5; Systat Software, CA). A Mann-Whitney rank sum test was performed for the nonparametric data obtained in the ELISAs and for histology scoring. Fisher's exact test was used to compare bacterial recovery at the animal and lesion levels (Table 1).

TABLE 1.

Recovery of H. ducreyi 35000HPhgbA_I and 35000HPhgbA_II from biopsy specimens

HgbA class	Expt	Mock immunization			HgbA immunization
		Pig	No. of H. ducreyi-positive specimens/total no. of specimens		Pig	No. of H. ducreyi-positive specimens/total no. of specimens
			104 CFU inoculum	103 CFU inoculum		104 CFU inoculum	103 CFU inoculum
I	1	1	2/3	3/3	3	0/4	NDa
		2	3/3	2/3	4	0/4	ND
	2	5	4/4	2/4	7	0/4	0/4
		6	4/4	2/4	8	0/4	0/4
		Total	13/14b	9/14d		0/16b	0/8d
II	1	1	2/3	3/3	3	3/4	ND
		2	3/3	2/3	4	1/4	ND
	2	5	4/4	4/4	7	3/4	4/4
		6	1/4	1/4	8	3/4	1/4
		Total	10/14c	10/14e		10/16c	5/8e

^aND, not determined.

^bP < 0.001, Fisher's exact test.

^cP = 0.709, Fisher's exact test.

^dP = 0.006, Fisher's exact test.

^eP = 1.000, Fisher's exact test.

RESULTS

Construction of H. ducreyi strain 35000HP expressing the hgbA gene from class II strain DMC111.

In preliminary swine challenge experiments using H. ducreyi strains other than 35000HP, all tested strains from either class were less infectious than strain 35000HP (data not shown). Furthermore, H. ducreyi wild-type strain 35000HP grows faster than the class II strains. Therefore, because of these and other phenotypic differences between classes of H. ducreyi strains, we engineered an isogenic 35000HP strain expressing the class II hgbA gene from strain DMC111 (35000HPhgbA_II) to determine if the nHgbA_I vaccine would be protective against a class II HgbA-expressing strain. In preliminary pig virulence studies, lesions produced by strain 35000HPhgbA_II were indistinguishable from lesions caused by the 35000HP parent strain and the HgbA_II protein was expressed at the same level as HgbA_I in the 35000HP background (data not shown).

The nHgbA_I/MPL vaccine reduced lesion severity after a homologous challenge.

Pigs were immunized and challenged as described in Materials and Methods using a protocol identical to that of the previous HgbA vaccine study (1). The following parameters were analyzed: pig ear lesions were examined at both the macroscopic and microscopic levels; recovery of H. ducreyi was determined; and antisera from nHgbA_I-immunized animals were subjected to several in vitro immunological assays, including different types of ELISAs, to measure antibody activity to nHgbA_I, binding to whole cells of H. ducreyi, and the ability of IgGs from immunized animals to block Hb binding to both nHgbA_I and nHgbA_II. The bactericidal activity of anti-nHgbA_I antisera was measured using a classic bactericidal assay.

Figure 1A shows representative photographs of lesions from pigs challenged with both strains of H. ducreyi taken immediately prior to biopsy. In animals immunized with nHgbA_I, inoculation with homologous strain 35000HPhgbA_I produced small lesions (Fig. 1A1), and in some cases, lesions were undetectable, save for puncture wounds from the applicator device. In contrast, infection of mock-immunized animals with both strains of H. ducreyi (Fig. 1A2 and 4) produced raised, visible lesions, much larger than those found in nHgbA_I-immunized animals challenged with strain 35000HPhgbA_I, indicating that both challenge strains are virulent in the experimental swine model of chancroid. The lesions of nHgbA_I-immunized animals challenged with heterologous strain 35000HPhgbA_II (Fig. 1A3) appeared similar to those from mock-immunized pigs.

FIG. 1.

nHgbA_I/MPL vaccine reduces lesion severity of experimental chancroid at the macroscopic and microscopic levels. (A) Photographs of pig ear lesions. Pigs were either mock immunized with MPL adjuvant only (2 and 4) or immunized with nHgbA_I/MPL (1 and 3) and challenged with either H. ducreyi strain 35000HPhgbA_I (1 and 2) or 35000HPhgbA_II (3 and 4). Photos were taken 1 week after challenge and immediately prior to biopsy. (B) H&E-stained biopsy sections from pig ears 1 week after infection. Pigs were either mock immunized with MPL adjuvant only (6 and 8) or immunized with nHgbA_I/MPL (5 and 7) and challenged with either H. ducreyi strain 35000HPhgbA_I (5 and 6) or 35000HPhgbA_II (7 and 8) (magnification, ×50).

FIG. 4.

nHgbA_I/MPL antiserum immunoprecipitates only HgbA_I from whole cells of H. ducreyi. The ability of anti-nHgbA_I antisera to immunoprecipitate HgbA from different H. ducreyi strains is shown in a Coomassie blue-stained 10% SDS-PAGE gel (A) and a Western blot probed with an antiserum to full-length rHgbA_I (19) (B). The H. ducreyi strains used in these immunoprecipitation assays were 35000HPhgbA_I (1), 35000HPhgbA_II (2), and DMC111 (3). Antisera used in these experiments were either pooled anti-nHgbA_I/Freund's (Fp) purified from antisera obtained in a previous study (1) or pooled anti-nHgbA_I/MPL (Mp). Molecular size markers are shown in kilodaltons.

Figure 1B shows representative histological sections of lesion biopsy specimens stained with H&E. In nHgbA_I-immunized animals challenged with homologous strain 35000HPhgbA_I (Fig. 1B5), a low level of inflammatory cell infiltrate was present in the lesions, and the basement membrane was intact, resembling healthy sections (52). In mock-immunized animals challenged with either strain (Fig. 1B6 and 8), there was a dense infiltrate of neutrophils and the epidermal-dermal border was completely destroyed, which left the dermis exposed, similar to previous results (1) and comparable to natural and experimental infection in humans (34, 56). In the nHgbA_I-immunized heterologous challenge group (Fig. 1B7), the lesions appeared to be similar to those of mock-immunized animals.

We used a previously described grading system based on a 1-to-5 range (for lesion scores, see Materials and Methods) to judge the severity of lesions in each group of animals (52). The mean lesion grade of mock-immunized animals challenged with the homologous strain was 4.0 ± 0.76, whereas the nHgbA_I-immunized group challenged with the same strain had a score of 2.6 ± 0.78 (P = 0.002). The mean lesion grade for the control group challenged with the heterologous strain was 4.1 ± 1.3, compared to a score of 3.7 ± 1.3 for nHgbA_I-immunized animals (P = 0.442).

Viable H. ducreyi cells were not recovered from lesions of nHgbA_I-immunized pigs challenged with a homologous H. ducreyi strain.

The recovery of viable H. ducreyi cells from biopsy specimens is shown in Table 1. At both challenge inocula (10⁴ and 10³ CFU), no H. ducreyi cells were isolated from nHgbA_I-immunized animals challenged with homologous strain 35000HPhgbA_I, whereas all mock-immunized animals yielded viable H. ducreyi (P < 0.001). In heterologous challenge experiments, all pigs, whether nHgbA_I or mock immunized, yielded strain 35000HPhgbA_II (P = 1.000).

Analysis of bacterial recovery at the lesion level revealed striking differences between nHgbA_I- and mock-immunized animals infected with the homologous H. ducreyi strain. In the control group, 13 of 14 lesions yielded H. ducreyi strain 35000HPhgbA_I at an inoculum of 10⁴ CFU. In the nHgbA_I-immunized group, 0 of 16 lesions yielded H. ducreyi (P < 0.001). At the lower dilution (10³ CFU), 9 of 14 lesions in mock-immunized animals yielded H. ducreyi strain 35000HPhgbA_I, whereas 0 of 8 lesions yielded H. ducreyi strain 35000HPhgbA_I in nHgbA_I-immunized animals (P = 0.006). Thus, the nHgbA_I vaccine provided protection against a bacterial challenge of at least 10 times the minimum infectious dose. In heterologous challenge experiments, 10 of 14 lesions in the control group yielded H. ducreyi strain 35000HPhgbA_II, compared to 10 of 16 lesions from the nHgbA_I-immunized group at a 10⁴ CFU inoculum (P = 0.709). Therefore, based on macroscopic examination, histological sections, and recovery of H. ducreyi cells, there was homologous protection against strain 35000HPhgbA_I but no protection against the genetically engineered heterologous strain 35000HPhgbA_II by the nHgbA_I vaccine.

The activity of the antiserum from nHgbA_I/MPL-immunized animals was lower than the activity of antiserum of nHgbA_I/Freund's-immunized pigs.

We analyzed the humoral immune response of nHgbA_I-immunized animals in a number of in vitro assays. An ELISA was performed to compare the antibody activity developed in response to our current MPL adjuvant HgbA vaccine to that of antiserum generated from the previous study using Freund's adjuvant (Fig. 2). The Freund's antiserum activity was approximately 7-fold higher than the MPL antiserum activity prior to the second immunization (P = 0.029) and approximately 2 to 3 times higher than the MPL activity after the third immunization (prior to infection) (P = 0.057). Of note, one of the four pigs immunized with nHgbA_I/MPL (pig 4M) showed ELISA activity twice as high as that found for the other three MPL-immunized pigs after the third immunization, indicating that different animals can respond differently to the vaccine preparation. Preimmune and mock-immune pig sera were free of anti-nHgbA_I antibodies (data not shown).

FIG. 2.

The activity of the nHgbA_I/MPL antisera is lower than that of the antisera from nHgbA_I/Freund's-immunized animals. Sera were collected preimmunization (prebleed), before the second immunization, and prior to challenge infection (three weeks after the third immunization). Data are expressed as OD₄₀₅ readings and given as median ± variance. Solid lines represent data from each pig immunized with nHgbA_I/Freund's adjuvant (1F, 4F, 5F, and 6F) (1), whereas dotted lines represent data from pigs immunized with nHgbA_I/MPL adjuvant (3M, 4M, 7M, and 8M). ELISA data for week 6 (prior to 3rd immunization) are not shown due to an incomplete data set for nHgbA_I/Freund's antiserum samples. P values comparing anti-nHgbA_I antibodies using Freund's and MPL adjuvants after one and after three immunizations (just prior to challenge) were 0.029 and 0.057, respectively, and were determined using the Mann-Whitney rank sum test (n = 3).

Two different assays were employed to determine whether antisera from nHgbA_I/MPL-immunized pigs recognized surface-exposed epitopes of HgbA in the context of whole cells of H. ducreyi. The first technique, a whole-cell binding ELISA, measured the relative amounts of antibodies bound to the surface of different H. ducreyi strains expressing either class I (wild-type strain 35000HPhgbA_I) or class II (strain 35000HPhgbA_II) HgbA or expressing class II HgbA in the native class II H. ducreyi background (wild-type strain DMC111), as detected by a secondary antibody conjugate (Fig. 3) and compared to an H. ducreyi strain that does not express HgbA (strain 35000HPΔhgbA). Antisera from all four nHgbA_I/MPL-immunized pigs and pooled anti-nHgbA_I/Freund's antisera showed greater reactivity with strain 35000HPhgbA_I than with strains 35000HPhgbA_II, 35000HPΔhgbA, or DMC111 (Fig. 3). As was observed in the solid-phase ELISA, there was also a statistically significant difference between the reactivity of pooled nHgbA_I/Freund's antiserum and all four nHgbA_I/MPL antisera with strain 35000HPhgbA_I (P values range from < 0.001 to 0.005) (Fig. 3). Antisera taken from mock-immunized pigs and prebleeds from nHgbA_I/MPL-immunized pigs reacted poorly with all strains of H. ducreyi tested (data not shown).

FIG. 3.

nHgbA_I/MPL antisera bind HgbA_I but not HgbA_II in the context of whole H. ducreyi cells. Antisera from individual pigs (pigs 3, 4, 7, and 8), obtained after three immunizations with nHgbA_I/MPL, and pooled antiserum from pigs immunized with nHgbA_I/Freund's (Fp; obtained from a previous study) (1) were tested for binding to whole cells of H. ducreyi strains 35000HPhgbA_I, 35000HPhgbA_II, 35000HPΔhgbA, and DMC111. The antisera used in these assays were from preinfection bleeds. Experiments were performed in triplicate on at least three different days. Data are expressed as relative light units and given as median ± variance. A P value of 0.005 was found for the difference between the binding of nHgbA_I/MPL and nHgbA_I/Freund's antisera to whole cells of H. ducreyi strain 35000HPhgbA_I (Mann-Whitney rank sum test).

The second method used to examine the interaction between anti-nHgbA_I and surface-exposed epitopes on H. ducreyi whole cells was an immunoprecipitation assay. Immunoprecipitates were analyzed by Coomassie blue staining (Fig. 4A) and Western blotting (Fig. 4B). Pooled anti-nHgbA_I/MPL antisera precipitated nHgbA_I from H. ducreyi strain 35000HPhgbA_I but with less reactivity than with pooled antisera from pigs immunized with nHgbA_I/Freund's (Fig. 4A1 and B1). Neither antibody preparation precipitated nHgbA_II in a 35000HP background (Fig. 4A2 and B2). Only pooled anti-nHgbA_I/Freund's was able to immunoprecipitate nHgbA_II from the wild-type strain DMC111 (Fig. 4A3 and B3).

Antiserum elicited by the nHgbA_I/MPL vaccine partially inhibited the binding of DIG-Hb to nHgbA.

We examined the ability of purified anti-nHgbA_I IgG to block the binding of Hb to immobilized nHgbA_I or nHgbA_II in an ELISA format (1). Pooled IgG purified from serum of nHgbA_I/Freund's-immunized pigs (1) blocked 59% of DIG-Hb/nHgbA_I binding, while pooled IgG purified from nHgbA_I/MPL-immunized pigs blocked 48% of binding (P = 0.180) (Fig. 5). There was a significant difference between IgGs purified from nHgbA_I/Freund's-immunized (38.1% inhibition) and nHgbA_I/MPL-immunized pigs (20.6% inhibition) (P = 0.002) (Fig. 5) in blocking Hb/nHgbA_II interactions. IgG purified from pooled sera of mock-immunized pigs failed to interfere with the binding of Hb to either nHgbA_I or nHgbA_II (data not shown).

FIG. 5.

IgGs from nHgbA_I/MPL-immunized animals partially inhibit Hb binding to nHgbA. The ability of pooled IgGs from pigs immunized with either nHgbA_I/MPL or nHgbA_I/Freund's to inhibit the binding of Hb to immobilized nHgbA_I or nHgbA_II was measured. P values comparing the ability of anti-nHgbA_I IgG to block Hb binding to nHgbA_I and nHgbA_II were determined using the Mann-Whitney rank sum test. Pooled nHgbA_I/Freund's IgG was purified from antisera obtained in a previous study (1) (n = 3).

Antisera from nHgbA_I/MPL pigs did not exhibit bactericidal activity.

We performed immune bactericidal assays using purified IgG from nHgbA_I/MPL-immunized pigs as previously described (1, 18). We were unable to demonstrate bactericidal killing in the presence of human complement using either H. ducreyi strain as a target (data not shown).

DISCUSSION

Administering the nHgbA_I vaccine with an adjuvant approved for use in humans was as effective as using Freund's adjuvant.

This study is similar to a previous vaccine study conducted in our laboratory, which demonstrated complete protection using purified nHgbA_I as immunogen (1). In the previous study, we used Freund's adjuvant, while MPL was used as adjuvant in the present study. Monophosphoryl lipid A (MPL) is a detoxified lipopolysaccharide analog isolated from Salmonella enterica serovar Minnesota R595. The structure, mechanism of action, and immunologic responses to MPL have been the subjects of intense investigation for many decades (11), and the safety of MPL has been extensively documented (5, 65). MPL has been successfully utilized as an adjuvant in numerous human vaccines, including vaccines for human papillomavirus 16/18 (HPV) (27, 28, 46), hepatitis B virus (HBV) (36, 61), and herpes simplex virus (HSV) (57). In these trials, MPL was used in combination with alum in the GSK proprietary AS04 adjuvant system (23). MPL has also been examined as an adjuvant in vaccines against malaria (24, 30, 59), leishmania (9), anthrax (32), Epstein-Barr virus (EBV) (54), and HIV (39, 43) and against allergies to house dust mites (6), ragweed pollen (7), and grass pollen (16). All of these studies showed that MPL is a successful and safe adjuvant.

The nHgbA_I vaccine administered with MPL protected only against a homologous H. ducreyi strain.

The current study advances the findings from the original nHgbA_I vaccine study in that we examined the ability of this vaccine to protect against strain 35000HP expressing a (heterologous) class II HgbA protein (HgbA_II). In alignments of HgbA from 10 strains of H. ducreyi, HgbA was highly conserved, showing ≥95% identity between strain classes (67). Based on this high degree of identity, we anticipated that nHgbA_I would provide cross-protection against strains expressing HgbA from either class of H. ducreyi strains. Surprisingly, this is not what we observed. Immunization with the nHgbA_I/MPL vaccine offered full protection to pigs when challenged with H. ducreyi strain 35000HPhgbA_I (homologous), but not from a challenge with H. ducreyi strain 35000HPhgbA_II (heterologous). Since these are isogenic strains, except for the HgbA protein (HgbA_I versus HgbA_II), two conclusions can be drawn from the results: (i) slight dissimilarities in the amino acid sequence between these 2 proteins are responsible for the difference in protection, and (ii) the observed protection cannot be the result of a protective response elicited by minor antigens contaminating the vaccine preparation.

In another of our previous studies, we found that loop 4 of HgbA is surface exposed and the largest, most immunodominant and variable loop among the 11 putatively exposed loops of HgbA (44). Including the highly conserved membrane-spanning segments, loops 4 of HgbA proteins are 91% identical between strains (16 of 186 amino acids are different). These results suggest that at least part of the protective response against HgbA_I lies in exposed epitopes from loop 4. Such divergence between strains in protective epitopes of surface-exposed outer membrane proteins has been shown in the development of gonococcal and meningococcal protein vaccines (15, 42, 64) and resulted in the requirement for multivalent vaccines in the latter.

Because preliminary experiments showed that wild-type class II strains were either weakly or noninfectious in the swine model (data not shown), an H. ducreyi class I strain expressing a class II HgbA protein was constructed in the current study in order to determine the possibility of heterologous protection. However, the use of an isogenic class II construct has certain limitations. Class II strains typically grow more slowly than class I strains, which may explain why they were less infectious in the swine model. Moreover, class II strains have a truncated LOS compared to that of class I strains (50, 53, 67), and it is possible that HgbA_II may be more exposed in a truncated LOS background. In data not presented here, anti-HgbA monoclonal antibody 1.51 binds poorly to whole cells of wild-type 35000HPhgbA_I but binds a 35000HP gmhA mutant (an isogenic mutant with a short LOS) (8) and two of four class II strains containing truncated LOS, including strain DMC111 (data not shown). These data suggest that the presence of a longer LOS molecule is responsible for the lack of monoclonal antibody 1.51 binding to HgbA on the surface of H. ducreyi. As shown by the results in Fig. 4, polyclonal anti-nHgbA_I antibodies bound better to strain DMC111 than to strain 35000HPhgbA_II, consistent with this notion. Thus, if additional HgbA epitopes are exposed in the context of a class II LOS, then it is possible that the nHgbA_I vaccine might offer some cross-protection for wild-type class II H. ducreyi strains.

Despite activity lower than that of antiserum from anti-nHgbA_I/Freund's immunization, antiserum from animals immunized with the nHgbA_I/MPL vaccine partially blocked DIG-Hb binding to HgbA.

The ELISA activity of antiserum from HgbA_I/MPL-immunized pigs to nHgbA_I was lower than the activity of antiserum from our earlier study using Freund's adjuvant after the first and third immunizations. These results are consistent with the results of two other immunization studies using other protein antigens in rabbits, which showed higher ELISA activity obtained using Freund's adjuvant (33, 41). We attempted to compare the antibody response of nHgbA_I/MPL antiserum to immobilized nHgbA_II in an ELISA format to determine if this might correlate with protection. However, the two HgbA proteins did not appear to bind the ELISA plates equally or may not have been equally in native conformation, either of which may have skewed the results. Variability in binding and folding of the different classes of HgbA proteins also prevented us from making valid conclusions about the results obtained in the ELISA measuring blocking of Hb binding (Fig. 5), even though some comparisons were statistically significant. Further work is necessary to elucidate the causes of the different binding capabilities of classes of HgbA proteins.

Instead of making comparisons of antiserum binding to purified protein, we compared the binding of nHgbA_I/MPL antisera to whole cells of the isogenic pair of 35000HP strains. As predicted, the homologous strain bound more antibodies than the heterologous strain in both whole-cell ELISA and the immunoprecipitation assay (Fig. 3 and 4). Similar to the activity measured by ELISA, the whole-cell antibody binding activity from the previous nHgbA_I/Freund's immunization study was higher than the activity from the present nHgbA_I/MPL study.

The ability of antibodies to block Hb binding to immobilized nHgbA_I and nHgbA_II was measured using an ELISA-based assay. Antibodies from both MPL and Freund's adjuvant groups modestly blocked Hb binding to either class of HgbA protein (Fig. 5). Despite significantly higher ELISA activity with Freund's antiserum, there was no significant difference in the ability of either IgG pool to inhibit Hb binding to homologous HgbA_I. However, Freund's pooled IgG significantly inhibited the binding of Hb to heterologous HgbA_II compared to the inhibition by MPL pooled IgG.

The lack of bactericidal and opsonophagocytic killing of H. ducreyi is well documented in the literature (2, 22, 31, 37, 45, 63, 68). In this study, bactericidal activity was not observed using antisera from nHgbA_I/MPL-immunized pigs. This differed from the results of a previous study that showed modest bactericidal activity using antiserum from nHgbA_I/Freund's-immunized pigs (1). The lack of bactericidal activity using nHgbA_I/MPL antisera may be due to lower antibody titers, lower antibody affinity, or inability to fix complement.

We previously studied opsonophagocytosis by nHgbA_I/Freund's pooled antisera. In a luminol-enhanced chemiluminescence assay, human neutrophils were not significantly stimulated in the presence of nHgbA_I/Freund's antiserum and complement, whereas controls stimulated neutrophils (data not shown). Furthermore, human neutrophils failed to kill H. ducreyi cells opsonized with nHgbA_I/Freund's antiserum and complement, while control bacteria were killed (data not shown). Since nHgbA_I/Freund's antiserum, which has a higher activity than nHgbA_I/MPL antiserum, did not exhibit any bactericidal activity, we did not test the opsonic activity of nHgbA_I/MPL antiserum obtained in the current study.

Future studies.

Anti-nHgbA_I antibodies bound better to class I-specific epitopes and failed to protect against a heterologous challenge, which suggests that the protective epitopes lie in the divergent central domain of HgbA. Studies are under way to determine whether a vaccine containing the central domain of HgbA (loops 4 through 7) is protective for class I strains.

The mechanism of protection of the HgbA vaccine (cellular versus humoral) is not known. However, the results of a previous study using passively transferred serum from thrice-infected pigs suggested that the humoral response was sufficient to protect from an H. ducreyi challenge (13). Studies are under way in our laboratory to determine whether protection using passively transferred serum from nHgbA_I-immunized pigs is sufficient to provide protection.

Based on our results, it remains unclear whether a single HgbA protein will protect against both classes of H. ducreyi, and it is possible that a bivalent vaccine might be needed. While anti-nHgbA_I antibodies did not bind well to class II HgbA in the genetically engineered 35000HP background, they bound moderately well to class II HgbA in the wild-type class II (DMC111) background, possibly due to differences in LOS or other antigens. Thus, based on our previous (1), present, and a third study under way (unpublished data), binding to whole cells and blocking Hb binding to HgbA are the two in vitro assays that correlate with the protection of the nHgbA_I vaccine.

The immune response to nHgbA_I/MPL was relatively modest, similar to the results of other studies using different antigens. Newer vaccines using MPL have overcome this problem by including alum in the vaccine mixtures with MPL (termed AS04), and we will examine this in our system. Despite these limitations, we have proof of concept that nHgbA_I in conjunction with an adjuvant approved for use in humans can provide complete homologous protection in a highly relevant animal model, paving the way toward phase I clinical trials.