A Haemophilus ducreyi strain lacking the yfeABCD iron transport system is virulent in human volunteers

ABSTRACT

Haemophilus ducreyi causes the genital ulcer disease chancroid and painful cutaneous ulcers in children who live in the tropics. To acquire heme from the host, H. ducreyi expresses a TonB-dependent hemoglobin receptor, HgbA, which is necessary and sufficient for H. ducreyi to progress to the pustular stage of disease in a controlled human infection model. HgbA transports hemoglobin across the outer membrane; how heme is transported across the cytoplasmic membrane is unclear. In previous studies, transcripts encoding the YfeABCD heme transporter were upregulated in experimental lesions caused by H. ducreyi in human volunteers, suggesting the latter may have a role in virulence. Here we constructed a double deletion mutant, 35000HPΔyfeABΔyfeCD, which exhibited growth defects relative to its parent 35000HP in media containing human hemoglobin as an iron source. Five human volunteers were inoculated at three sites on the skin overlying the deltoid with each strain. The results of the trial showed that papules formed at 100% (95% CI, 71.5, 100) at both 35000HP and 35000HPΔyfeABΔyfeCD-inoculated sites (P = 1.0). Pustules formed at 60% (95% CI, 25.9, 94.1) at parent-inoculated sites and 53% (95% CI, 18.3, 88.4) at mutant-inoculated sites (P = 0.79). Thus, the ABC transporter encoded by yfeAB and yfeCD was dispensable for H. ducreyi virulence in humans. In the absence of YfeABCD, H. ducreyi likely utilizes other periplasmic binding proteins and ABC-transporters such as HbpA, SapABCDF, and DppBCDF to shuttle heme from the periplasm into the cytoplasm, underscoring the importance of redundancy of such systems in gram-negative pathogens.

INTRODUCTION

Haemophilus ducreyi causes chancroid, a sexually transmitted genital ulcer disease that is endemic in resource-poor areas of Africa and Asia and facilitates the transmission and acquisition of the human immunodeficiency virus (HIV-1) (1–4). Once thought to be exclusively sexually transmitted, H. ducreyi recently emerged as a major cause of painful exudative cutaneous ulcers in children who live in equatorial regions of the South Pacific and Africa (2, 5–8); the latter infections are likely due to traumatic breaks in the skin of either asymptomatically colonized individuals or subsequent contact of wounds with environmental reservoirs of the organism (8, 9). Thus, infections due to H. ducreyi remain an important problem for global health.

To study the biology of H. ducreyi, in 1992, we developed a controlled human infection model (reviewed in references (10–12)). In the model, healthy adult volunteers are inoculated via puncture wounds at multiple sites with H. ducreyi strain 35000HP (HP, human passaged) on the skin overlying the deltoid. Within 24 h of inoculation of ~1–150 CFU, papules form at inoculated sites, and either spontaneously resolve or evolve into pustules 2 to 5 days later, mimicking the initial stages of natural chancroid (11). Due to safety and practical considerations, volunteers are infected until the pustules become painful, usually 6 to 8 days after inoculation, or for a maximum of 14 days. In both natural chancroid and experimental pustules, H. ducreyi is found in an abscess and co-localizes with collagen, fibrin, macrophages, and polymorphonuclear leukocytes, which fail to ingest the organism (13, 14). Thus, the model recapitulates several essential aspects of H. ducreyi pathogenesis.

The H. ducreyi challenge model has afforded the unique opportunity to study the role of putative bacterial virulence determinants in the initial stages of the disease by performing mutant vs parent trials. In these double-blind dose-ranging trials, groups of volunteers are infected on one arm at three sites with fixed doses of the parent strain and on the other arm at three sites with escalating doses of the mutant strain (10–12). Of 36 mutants tested to date, 17 are categorized as virulent (form pustules at statistically equivalent rates at doses similar to the parent), nine as partially attenuated (form pustules at doses twofold or threefold that of the parent, but at statistically lower rates at doses equivalent to the parent) and 10 as fully attenuated (unable to form pustules even at doses 10-fold that of the parent) (12, 15, 16).

H. ducreyi lacks the known biosynthetic pathway for the synthesis of heme, and mutant vs parent trials show that the organism must acquire heme and iron to survive in the human host (17), which limits iron availability to the bacterium through a process called “nutritional immunity” (18, 19). In gram-negative pathogens, TonB-dependent receptors are usually responsible for transporting heme and iron across the outer membrane into the periplasm. H. ducreyi expresses 3 TonB-dependent receptors called the hemoglobin receptor (HgbA), the heme receptor (TdhA), and TdX, whose function is unknown. In mutant vs parent trials, an hgbA mutant was fully attenuated for pustule formation, while a tdhA and tdX double mutant were virulent in humans (17, 20). The hgbA mutant fails to grow in media containing human hemoglobin as a sole iron source but does grow in media containing hemin or catalase, implying that hemoglobin is the major source of iron available to H. ducreyi during experimental infection (20, 21). However, how H. ducreyi transports heme and iron from the periplasm to the cytoplasm is unclear.

Transport of heme and iron across the cytoplasmic membrane usually requires a periplasmic binding protein and an ATP-binding cassette (ABC) transporter (18). These transport systems frequently have multiple and redundant functions that help to ensure bacterial survival in the host. For example, the YfeABCD system transports both iron and manganese, the SapABCDF system transports antimicrobial peptides and iron, and the DppABCDF system transports dipeptides and iron (22–24). Pathogens usually contain multiple periplasmic binding proteins that share homology and multiple ABC transport systems that provide redundant functions (18, 23, 25). Despite this redundancy, the homologous ABC transporters encoded by yfeABCD in Yersinia pestis and by sitABCD in Salmonella enteritica serovar Typhimurium exhibit growth defects in the presence of iron chelators and in two well-described studies are required for virulence in their respective murine models of infection (22, 26). Similarly, the ABC transporter encoded by dppABCD in Mycobacterium tuberculosis is required for growth in media containing hemin or hemoglobin and bacterial survival in a macrophage cell line (24).

A YfeA homolog was identified in periplasmic fractions of H. ducreyi grown under heme-deficient conditions (27). In most gram-negative bacteria, yfeABCD is a single operon; however, in H. ducreyi, yfeAB and yfeCD are in separate operons. In three studies using dual RNA-sequencing, transcripts of hgbA, yfeAB, and yfeCD were significantly upregulated in H. ducreyi-infected pustules compared to mid-log phase bacteria used to infect the volunteers (28–30). These data suggest that YfeABCD might work in concert with HgbA and also be required for the virulence of H. ducreyi in humans.

Here we characterized the yfeAB and yfeCD operons of H. ducreyi, constructed a yfeAB yfeCD double deletion mutant, characterized its growth in media containing human hemoglobin as an iron source, and tested the mutant for virulence in human volunteers.

RESULTS

H. ducreyi yfeAB and yfeCD are members of two distinct operons

As predicted by St. Denis and colleagues, yfeA (HD 1816) and (yfeB) (HD1817) are in a putative operon that spans HD1814 to HD1817 (Fig. 1A). Similarly, yfeC (HD 1025) and yfeD (HD1024) are in a putative operon that spans HD1027 to HD1021 (Fig. 1B). To assess whether yfeAB and yfeCD are co-transcribed with their neighboring genes, we performed RT-PCR on RNA isolated from aerobically grown mid-log phase cultures of 35000HP. The primers used in the reactions produced products that spanned the gene junctions and are listed in Table S1. As controls for these experiments, we included reactions that lacked reverse transcriptase or contained genomic DNA as a template. We found that yfeAB was in an operon that contained four genes from HD1814 to HD1817, but did not extend to HD1818, confirming the predictions made by St. Denis and colleagues (27) (Fig. 1A). Similarly, yfeCD was in an operon that contained seven genes from HD1027 to HD1021 (Fig. 1B).

Fig 1.

yfeAB and yfeCD are members of distinct operons. (A) RT-PCR of gene junctions between HD1814, HD1815, HD1816 (yfeA), HD1817 (yfeB), and HD1818 indicated that there are four genes in the operon that contain yfeAB. (B) RT-PCR of the gene junctions between HD1027, HD1026, HD1025 (yfeC), HD 1024 (yfeD), HD1023, HS1022, and HD1021 indicated that there are seven genes in the operon that contains yfeCD. RT+, reactions containing reverse transcriptase; RT−, reactions lacking reverse transcriptase; gDNA, genomic DNA control. The presence of dnaE transcripts (right panels) was used as quality control for the RNA in each experiment. The gels are representative of three independent experiments.

Deletion of yfeAB and yfeCD restricts H. ducreyi growth in media containing human hemoglobin as a sole iron source in aerobic conditions

We have shown that of the three TonB-dependent outer membrane heme receptors expressed by H. ducreyi, only the hemoglobin receptor (HgbA) is required for the virulence of 35000HP in human volunteers; therefore, hemoglobin is likely the sole source of heme H. ducreyi utilizes in vivo (17, 20). To investigate whether the deletion of yfeABCD affects the growth of H. ducreyi in media containing hemoglobin, we generated an unmarked, in-frame deletion mutant of yfeAB and yfeCD using recombineering methodology (16, 31). Whole-genome sequencing showed that 35000HPΔyfeABΔyfeCD contained the intended deletions but did not contain any other mutations.

While H. ducreyi can grow anaerobically in the absence of heme, heme is absolutely required for the growth of the organism under aerobic conditions (32, 33). To deplete the organisms of heme, we grew both strains anaerobically on agar plate media lacking heme, inoculated the strains into broth, added human hemoglobin at final concentrations of 5 or 50 µg/mL, and monitored growth by determining CFU and OD₆₆₀ at various timepoints under aerobic conditions exactly as described previously (34). Since heme depletion results in a 12-h lag phase for 35000HP, we monitored growth from 12 to 20 h (Fig. 2). To analyze the data, we performed a mixed model repeated measures analysis with the Tukey adjustment. As the two growth conditions were examined simultaneously, the initial model included an interaction of the four conditions (strain and hemoglobin concentration) with time; in three independent experiments, for both the CFU and OD₆₆₀ data, there was no indication of a significant interaction (P = 1.0). Next, we ran the model without the concentration interaction and examined the adjusted P values for the parent versus the mutant for each hemoglobin concentration separately over the 12-h to 20-h time period. For the CFU data, the adjusted P values were 0.052 for the high hemoglobin concentration and 0.002 for the low hemoglobin concentration with the parent growing better than the mutant in both cases. For the OD₆₆₀ data, the adjusted P-values were 0.025 for the high hemoglobin concentration and 0.022 for the low hemoglobin concentration with the parent having higher OD₆₆₀ in both cases. Taken together, the data suggested that relative to 35000HP, 35000HPΔyfeABΔyfeCD was impaired for growth in media containing 5 µg/mL human hemoglobin and also exhibited a growth defect in higher hemoglobin concentrations.

Fig 2.

Growth kinetics of 35000HP (circles) and 35000HPΔyfeABΔyfeCD (squares) in brain heart infusion broth supplemented with human hemoglobin at final concentrations of 5 (open symbols) and 50 (closed symbols) µg/mL. Growth was measured by monitoring optical density (OD₆₆₀) (A) or colony forming units (CFU) (B) over time in three independent experiments. For clarity, only the upper components of error bars representing SDs are shown.

The yfeABCD iron transport system is dispensable for virulence in humans

Since yfeAB and yfeCD transcripts are significantly upregulated in experimental pustules compared to the inocula used to infect human volunteers (28–30), and because 35000HPΔyfeABΔyfeCD was growth impaired in media containing human hemoglobin compared to 35000HP, we compared the virulence of 35000HPΔyfeABΔyfeCD to 35000HP in human volunteers.

As required by our Investigational New Drug (IND) protocol, the strains were grown aerobically in protease peptone media that contained bovine heme as an iron source. In this media, 35000HP and 35000HPΔyfeABΔyfeCD displayed similar growth kinetics (Fig. 3). Per protocol, we attempted to infect three volunteers with estimated delivered doses (EDDs) of 45, 90, and 180 CFU of the mutant at three sites on one arm and 90 CFU of the parent at three sites on the opposite upper arm. One volunteer withdrew prior to inoculation, and two were inoculated with EDDs of 26, 52, and 105 CFU of the mutant and 114 CFU of the parent (Table 1). Papules formed at all mutant and parent-inoculated sites. However, pustules formed at 3 of 6 mutant-inoculated and 1 of 6 parent-inoculated sites.

Fig 3.

35000HP (circles) and 35000HPΔyfeABΔyfeCD (squares) were grown aerobically in human challenge media containing 50 µg/mL of bovine hemin for 10 h. Growth was measured by optical density (OD₆₆₀) in three independent experiments.

TABLE 1.

Response to inoculation with live H. ducreyi

Volunteera (gender)	Observation period (days)	Strainb	EDD, (CFU)c	No. of initial papules	No. of initial pustules	No. of pustules at endpoint
506 (F)	7	P	114	3	0	0
		M	26–105	3	3	3
507 (F)	7	P	114	3	1	1
		M	26–105	3	0	0
509 (M)	6	P	122	3	3	3
		M	36–144	3	3	3
511 (F)	7	P	122	3	2	2
		M	36–144	3	1	1
513 (M)	7	P	122	3	3	3
		M	36–144	3	1	1

^a Volunteers 506 and 507 were inoculated in the first iteration; 509, 511, and 513 in the second iteration. F, female; M, male.

^b P, parent strain 35000HP; M, mutant strain 35000HP∆yfeAB∆yfeCD.

^c EDD, estimated delivered dose; 26–105, one dose each of 26, 52, and 105 CFU; 36–144, one dose each of 36, 72, and 144 CFU.

Since the mutant formed pustules at doses similar to the parent, the data suggested that the mutant was virulent. Per protocol, we attempted to repeat the experiment with one replacement volunteer plus three additional volunteers. One of the four volunteers withdrew on the day of inoculation, and three were inoculated with EDDs of 36, 72, and 144 CFU of the mutant and 122 CFU of the parent (Table 1). Papules formed at all mutant and parent-inoculated sites; pustules formed at 5 of 9 mutant-inoculated sites and 8 of 9 parent-inoculated sites.

The cumulative results of the trial showed that papules formed at 100% (95% CI, 71.5, 100) at both 35000HP and 35000HPΔyfeABΔyfeCD-inoculated sites (P = 1.0) (Table 1). The mean area of the papules 24 h after inoculation was 24.1 ± 41.2 mm² at parent-inoculated sites and 39.5 ± 50.5 mm² at mutant-inoculated sites (P = 0.30). Pustules formed at 60% (95% CI, 25.9, 94.1) at parent-inoculated sites and 53% (95% CI, 18.3, 88.4) at mutant-inoculated sites (P = 0.79). Thus, yfeAB and yfeCD were dispensable for virulence in the model.

Three volunteers (509, 511, and 513) developed pustules at both mutant and parent-inoculated sites (Table 1). One mutant (N = 3) and one parent (N = 3) pustule were biopsied from each of these participants. Two volunteers developed pustules only at mutant (506) or parent (507) sites; 506 agreed to one biopsy and 507 declined the biopsy. All biopsies were divided in half and semi-quantitatively cultured or fixed in formalin and stained with hematoxylin-eosin and anti-CD3 antibodies as described (35). Both mutant and parent specimens contained pustules that eroded through the epidermis. There was a dense monocytic perivascular and interstitial infiltrate in the dermis; the dermal infiltrate consisted primarily of perivascular CD3⁺ cells. These findings were typical of experimental pustules (36).

Of three biopsy specimens cultured from parent sites, two yielded H. ducreyi. Of four biopsy specimens cultured from mutant sites, four yielded H. ducreyi. The number of viable H. ducreyi recovered from parent sites was 3.5 × 10⁶ ± 4.7x10⁶ (mean ± SD) CFU/g tissue and from mutant sites was 2.7 × 10⁵ ± 2.7x10⁵ CFU/g tissue (P = 0.6).

To ensure that no cross-contamination had occurred between the two mutant and two parent inocula or the mutant and parent-inoculated sites, we tested a minimum of 30 colonies isolated from each of the inocula and colonies isolated from daily swab surface cultures and biopsies for the presence of yfeAB, yfeCD, and dnaE sequences by colony hybridization. The dnaE probe hybridized to all the colonies tested from the parent (N = 72) and mutant (N = 72) inocula, while the yfeAB and yfeCD probes hybridized only to colonies isolated from the parent inocula. At least one positive surface culture for H. ducreyi was obtained during follow-up visits from 33% of the parent-inoculated and 33% of the mutant-inoculated sites. The yfeAB and yfeCD probes hybridized to all colonies isolated from surface cultures of parent sites (N = 134) and none of the mutant sites (N = 129); the dnaE probe hybridized to all colonies isolated from both parent and mutant sites (N = 263). The dnaE probe hybridized to all colonies recovered from biopsy specimens of both parent (N = 71) and mutant (N = 108) sites; the yfeAB and yfeCD probes hybridized to only colonies derived from parent sites. Thus, there was no evidence of cross-contamination between mutant and parent inocula or inoculation sites.

DISCUSSION

Expression of the TonB-dependent receptor HgbA is required for the virulence of 35000HP in humans, suggesting that hemoglobin is likely the major source of heme and iron utilized by H. ducreyi during human infection (17, 20). Although HgbA is necessary to transport heme across the outer membrane into the periplasm, how heme is transported across the cytoplasmic membrane into the cell is unclear. Since we found that transcripts corresponding to the yfeABCD transport system were upregulated in experimental lesions, and since 35000HPΔyfeABΔyfeCD had growth defects when grown aerobically in media containing human hemoglobin as an iron source, we compared the virulence of 35000HPΔyfeABΔyfeCD to that of 35000HP in human inoculation experiments. Surprisingly, we found that the mutant formed pustules at a rate that was similar to the parent and was virulent.

There are no data available on the concentration of hemoglobin in experimental lesions caused by H. ducreyi in humans, but the requirement for hemoglobin uptake for the virulence of H. ducreyi in humans appears to be absolute (17). The fact that 35000HPΔyfeABΔyfeCD exhibited growth defects in aerobic conditions in media containing as much as 50 µg/mL of human hemoglobin suggests that the hemoglobin levels encountered by H. ducreyi in vivo may be higher than the concentrations we tested in vitro. However, abscesses in humans are generally anaerobic (37); genes differentially expressed by H. ducreyi at the pustular stage of disease significantly overlap with those differentially expressed during anaerobic growth (38); and H. ducreyi does not require heme for growth under anaerobic conditions. The fact that HgbA is required for virulence in vivo and heme is required for aerobic growth in vitro implies that at some point during infection, likely at the papular stage of disease, the organism encounters an aerobic environment. Alternatively, the concentrations of hemoglobin that are sufficient to maintain bacterial growth during shifting or mixed aerobic/anaerobic conditions in vivo may not be relevant to what we tested in aerobic broth cultures here.

Since expression of HgbA is required while the YfeABCD transport system is dispensable for H. ducreyi virulence in humans, the organism must use alternative heme transport systems in the absence of YfeABCD. Transport of heme from the outer to the inner membrane is typically done by a periplasmic binding protein, which delivers heme to ABC-transporter. In addition to YfeA, the 35000HP chromosome contains predicted homologs of the periplasmic binding protein SapA, whose primary function is to import antimicrobial peptides for degradation, and its transporter SapBCDF (39, 40). In addition to its role in resistance to antimicrobial peptides, SapA is also responsible for heme binding and utilization in H. influenzae (23). HbpA of H. influenzae is another heme-binding lipoprotein; antisera to HbpA weakly cross-react to a band with an apparent molecular weight of 51 kDa in lysates of H. ducreyi (25). A homology search of the 35000HP chromosome showed that HD215 encodes a predicted protein of 61 kDa that is highly homologous (67% identity, 79% similarity) to HbpA. HbpA shares homology with a family of periplasmic binding proteins that include SapA and DppA (23). Although 35000HP lacks a DppA homolog, it does contain homologs of the DppBCDF transporter (HD312-316). In the absence of YfeABCD, either SapA or HbpA could have bound heme and used either SapBCDF or DppBCDF to transport heme across the cytoplasmic membrane, allowing the organism to escape nutritional immunity in vivo.

The primary limitations of the human challenge model are the artificial route of inoculation and the restriction of the duration of infection to the pustular stage of the disease. We have been discouraged by the FDA from testing mutants that contain antibiotic resistance cassettes and do not have FDA approval for competition experiments, which usually use antibiotic-resistant and antibiotic-sensitive strains. With the caveats that we cannot determine whether YfeABCD contributes to later stages of disease such as ulcer or bubo formation, or whether the mutant is less fit than the parent strain, our study underscores the idea that pathogenic bacteria usually contain redundant systems to counteract host sequestration of iron and heme during in vivo growth.

MATERIALS AND METHODS

Bacterial strains and culture conditions

H. ducreyi strain 35000HP is a human passaged variant of strain 35000 (41). 35000HPΔyfeABΔyfeCD was derived from stocks of 35000HP after minimal passage. H. ducreyi was routinely grown on chocolate agar plates supplemented with 1% IsoVitaleX in the presence of 5% CO₂. For the human challenge trials, H. ducreyi were grown in a proteose peptone broth-based medium with 1% IsoVitaleX, 5% heat-inactivated fetal calf serum, and 50 µg/mL hemin (42). All cultures were grown at 33°C.

RNA isolation and RT-PCR

To determine the operon organization of yfeAB and yfeCD, RNA was isolated from mid-log phase aerobic broth cultures of 35000HP exactly as previously described (16). The QuantiTect SYBR green master mix kit (Qiagen) and custom primers (Table S1) were used for reverse transcription (RT)-PCR. 1 ng of RNA was used per reaction. A reaction lacking reverse transcriptase served as the negative control for each primer set; purified 35000HP gDNA served as a positive control for the primer pairs and amplification of dnaE served as a control for RNA integrity.

Construction and characterization of 35000HPΔyfeABΔyfeCD

35000HPΔyfeABΔyfeCD was constructed as described previously (16) using the primers in Table S1. In brief, three DNA fragments [a spectinomycin cassette flanked by flippase recognition target (FRT) sites from pRSM2832], ~ 500 bp of sequence upstream of the target operon (e.g., either yfeAB or yfeCD) that included the start codon, and ~500 of sequence downstream of the target operon that included the last seven codons of the operon were PCR amplified with Phusion polymerase (Thermo Fisher) using primers designed by the NEBuilder program (New England Biolabs; Table S1). Fragments were assembled and inserted into BamHI-digested pRSM2072 using HiFi Assembly Master Mix (NEB). The resulting plasmids—pRSM2072-∆yfeAB, and pRSM2072-∆yfeCD—were transformed into E. coli NEB-10β, which lacks methylases. pRSM2072-∆yfeAB was electroporated into H. ducreyi 35000HP. A spectinomycin-resistant colony was selected and electroporated with pRSM2975, which encodes a tetracycline-inducible flippase. Following induction with anhydrotetracycline, clones that were spectinomycin sensitive were selected. The resulting, unmarked, in-frame deletion mutant, 35000HP∆yfeAB was subjected to whole genome sequencing; Sanger sequencing was used to confirm the deletion and resolve low-quality SNP calls. Once we confirmed that 35000HP∆yfeAB only contained the deletion of interest, 35000HP∆yfeAB was transformed with pRSM2072-∆yfeCD and subjected to the same procedures described above to obtain 35000HPΔyfeABΔyfeCD, which was also subjected to whole genome and Sanger sequencing and confirmed to contain only the two deletions of interest.

Human inoculation experiments

Mutant vs parent comparison trials are double-blinded dose-ranging studies with a minimum of two stages (11, 12). Seven healthy adult volunteers [4 men and 3 women; 1 Asian person, 1 Black person, 1 Middle Eastern person, 4 White persons; mean age ± standard deviation (SD) 38.4 ± 14.1 years] between 26 and 61 years of age enrolled in the study. Two volunteers withdrew consent prior to inoculation, and five participated (Table 1).

Stocks of 35000HP and 35000HPΔyfeABΔyfeCD were prepared according to FDA guidelines under BB-IND #13064. Methods for preparation of the bacteria, inoculation, determination of the EDD, surface cultures, biopsies, clinical observations, and antibiotic treatment of the volunteers were performed exactly as described (11). Clinical endpoints included the resolution of infection at all sites, the development of a painful pustule at any site, or 14 days of observation (11). As the outcomes of infected sites within a subject are not independent, comparisons of papule and pustule formation rates were performed using a logistic regression model with generalized estimating equations (GEE) (43). The GEE sandwich estimate for the standard errors was used to calculate 95% confidence intervals (95% CI) for the rates.

Colonies recovered from the inocula, surface cultures, and biopsies were tested for the presence of yfeAB, yfeCD, and dnaE by using probes generated by specific primers (Table S1) for these genes/operons exactly as described previously (16).

Growth of H. ducreyi strains in the presence of hemoglobin as a sole iron source

To measure the growth dependence of 35000HPΔyfeABΔyfeCD and 35000HP on hemoglobin as a sole iron source, we followed methods exactly as outlined by Stevens and coworkers that we had utilized previously (34, 44). To eliminate intracellular heme, cells were grown in a BD Gas Pak EZ at 33°C for 36 to 48 h on plates containing a GC agar base supplemented with glucose, L-cysteine, and L-glutamine with no heme sources. These cells were then grown aerobically at 33°C in brain heart infusion broth supplemented with glucose, L-cysteine, and L-glutamine, 5% heat-inactivated fetal calf serum, and human hemoglobin at final concentrations of 5 or 50 µg/mL. H. ducreyi has tight intracellular junctions that cause the organism to clump; whether it is more accurate to measure growth by CFU or turbidity measurements is unclear. Therefore, we monitored growth by both optical density readings and removing aliquots from the culture, vigorously vortexing the sample to break up clumps, and determining the CFU over time in three independent experiments. A mixed-effects analysis of variance (ANOVA) was performed on the log-transformed CFU data with strain, hemoglobin concentrations, and time as fixed effects and a random effect for the experiment.

ETHICS STATEMENT

The human challenge trial was approved by the Institutional Review Board of Indiana University-Purdue University Indianapolis. Written informed consent was obtained from all participants for HIV-1 serology and enrollment in the study.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/iai.00058-24.

Table S1. iai.00058-24-s0001.xlsx.

Oligonucleotides used in this study.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.