Epithelial Coculture and l-Lactate Promote Growth of Helicobacter cinaedi under H₂-Free Aerobic Conditions

ABSTRACT

Helicobacter cinaedi is an emerging opportunistic pathogen associated with infections of diverse anatomic sites. Nevertheless, the species demonstrates fastidious axenic growth; it has been described as requiring a microaerobic atmosphere, along with a strong preference for supplemental H₂ gas. In this context, we examined the hypothesis that in vitro growth of H. cinaedi could be enhanced by coculture with human epithelial cells. When inoculated (in Ham's F12 medium) over Caco-2 monolayers, the type strain (ATCC BAA-847) gained the ability to proliferate under H₂-free aerobic conditions. Identical results were observed during coculture with several other monolayer types (LS-174T, AGS, and HeLa). Under chemically defined conditions, 40 amino acids and carboxylates were screened for their effect on the organism's atmospheric requirements. Several molecules promoted H₂-free aerobic proliferation, although it occurred most prominently with millimolar concentrations of l-lactate. The growth response of H. cinaedi to Caco-2 cells and l-lactate was confirmed with a collection of 12 human-derived clinical strains. mRNA sequencing was next performed on the type strain under various growth conditions. In addition to providing a whole-transcriptome profile of H. cinaedi, this analysis demonstrated strong constitutive expression of the l-lactate utilization locus, as well as differential transcription of terminal respiratory proteins as a function of Caco-2 coculture and l-lactate supplementation. Overall, these findings challenge traditional views of H. cinaedi as an obligate microaerophile.

IMPORTANCE H. cinaedi is an increasingly recognized pathogen in people with compromised immune systems. Atypical among other members of its bacterial class, H. cinaedi has been associated with infections of diverse anatomic sites. Growing H. cineadi in the laboratory is quite difficult, due in large part to the need for a specialized atmosphere. The suboptimal growth of H. cinaedi is an obstacle to clinical diagnosis, and it also limits investigation into the organism's biology. The current work shows that H. cinaedi has more flexible atmospheric requirements in the presence of host cells and a common host-derived molecule. This nutritional interplay raises new questions about how the organism behaves during human infections and provides insights for how to optimize its laboratory cultivation.

INTRODUCTION

Helicobacter cinaedi is an emerging pathogen of the class Epsilonproteobacteria. This spiral-shaped Gram-negative rod is associated with opportunistic infections of multiple anatomic sites, including the intestinal tract (1, 2), blood (3, 4), peripheral soft tissue (5, 6), and other body fluids (7–9). This behavior contrasts with related Helicobacter and Campylobacter species, which typically demonstrate a more limited tissue tropism. H. cinaedi was identified ∼30 years ago in men who have sex with men (1) and HIV patients (10, 11), although infections have since been described in other immunocompromised groups (12, 13), as well as immunocompetent (14–17) and postsurgical patients (18).

Despite abundant case reports, the ecology/epidemiology of H. cinaedi remains poorly defined and sparsely studied. Limited molecular evidence suggests that it can be found in the stools of a small percentage of healthy individuals (19). The organism has also been isolated from animal feces (including rodents, dogs, and primates), although its role as a veterinary and/or zoonotic pathogen is uncertain (20, 21). An environmental reservoir has not yet been identified. The majority of recent literature on H. cinaedi (for instance, many of the current references) originates from Japan, although it is unclear whether this reflects regional prevalence or an increased awareness among Japanese investigators. Knowledge of H. cinaedi biology and tools for its genetic manipulation are likewise sparse, limited to several studies of targeted deletion mutants. These deletion mutants include strains deficient in cytolethal distending toxin (22) and alkyl hydroperoxide reductase (23), both of which demonstrate attenuated virulence in a murine colitis model.

Likely contributing to this paucity of information, H. cinaedi is a challenging organism to manipulate in vitro. Strains are notable for poor, film-like growth on agar plates, a preference for biphasic culture (broth over agar), and decreased viability on subculture (24, 25). Recommended medium conditions for H. cinaedi are not standardized. Various formulations have been utilized (e.g., brucella broth/agar, brain heart infusion, tryptic soy, and CDC anaerobic media), both with and without supplemental blood/serum products (20, 26). In a recent study, Levinthal broth (Mueller-Hinton broth with horse serum) was proposed as a favorable medium for susceptibility testing of clinical isolates (20). H. cinaedi is considered microaerophilic, like many members of the class Epsilonproteobacteria, and demonstrates a strong preference for H₂ gas (24, 27, 28). The latter phenomenon is presumably due to H₂ serving as a respiratory electron donor (29), although this mechanism has not been demonstrated explicitly. The large majority of recent published work on H. cinaedi has employed H₂-enhanced microaerobic conditions (which, notably, are not available within all diagnostic/research laboratories). Even with this favorable atmosphere, the organism is characterized by slow growth and attains low optical density in liquid culture (26).

Given this fastidious behavior, the ability of H. cinaedi to proliferate under aerobic or H₂-free conditions has never been investigated systematically. Nevertheless, several indirect lines of evidence suggest that H. cinaedi may possess more flexible atmospheric requirements than traditionally assumed. As mentioned above, the species has been associated with infections of diverse anatomic sites (stool, blood, and tissue), which can demonstrate a broad range of in situ gas concentrations (29). Moreover, the ability of clinical laboratories to propagate H. cinaedi with typical blood culture platforms implies at least some degree of aerotolerance. Although agar subculture of blood bottles (for colonial isolation/identification of H. cinaedi) requires microaerobic conditions (30, 31, 32), the initial bottles are neither microaerobic nor H₂-supplemented. Rather, microbial growth in commercial systems is detected under a fixed aerobic or anaerobic atmosphere (33). When explicitly reported, H. cinaedi has been detected overwhelmingly in aerobic blood culture bottles (30, 34).

In this context, we sought to better define the growth dynamics of the type strain (ATCC BAA-847) under various conditions of monoculture and epithelial coculture. This work demonstrated that epithelial monolayers and l-lactate facilitate H. cinaedi growth under H₂-free aerobic conditions. These findings were confirmed for a collection of clinical H. cinaedi isolates, and the transcriptional behavior of the type strain was characterized by high-throughput RNA sequencing (RNA-seq) under diverse growth conditions, including epithelial coculture and l-lactate supplementation. In total, this study demonstrates that the atmospheric requirements of H. cinaedi are not fixed but rather are interconnected with other nutritional and environmental factors. These findings are relevant for developing improved methods of isolating/propagating H. cinaedi in the laboratory and raise important questions about how the species behaves in infected tissue and blood.

MATERIALS AND METHODS

Bacterial strains and cell lines.

H. cinaedi ATCC BAA-847^T, originally isolated from a human rectum, was acquired from the American Type Culture Collection (1, 35). Other clinical isolates of H. cinaedi were obtained from the laboratories of Naoaki Misawa (University of Miyazaki) and James Fox (Massachusetts Institute of Technology); see Table 1 for additional information. Commercially available human epithelial cell lines used in this study included Caco-2 (colorectal adenocarcinoma, enterocyte-like), LS-174T (colorectal adenocarcinoma, goblet cell like), AGS (gastric adenocarcinoma), and HeLa (cervical adenocarcinoma).

TABLE 1.

Strains of H. cinaedi employed in this study

H. cinaedi strain	Sitea	Reference
ATCC BAA-847	Rectal	35
CCUG 18818	Rectal	61
N66	Blood	21
N69	Blood	21
N70	Blood	21
N71	Blood	21
N73	Blood	21
N76	Blood	21
N77	Blood	21
S1	Blood	21
S3	Blood	21
S5	Blood	21
S8	Blood	21

^aAnatomic site of original isolation.

Bacterial monoculture.

Bacterial monoculture experiments were performed at 37°C in either (i) Ham's F12 medium (F12) (GE HyClone), a chemically defined medium previously shown to support H. cinaedi growth (24); (ii) Mueller-Hinton broth (Sigma-Aldrich), a predominantly casein-based medium; (iii) brucella broth (BD BBL), a tissue/casein-based medium; or (iv) brain heart infusion broth (Oxoid), a tissue/peptone-based medium. (Complete formulations of each are provided in Table 2.) As indicated, media were supplemented with various metabolites (see below), 10% fetal bovine serum (FBS), or normal human serum (Atlanta Biologicals). Bicarbonate-free F12 (Thermo HyClone) supplemented with HEPES (10 mM; pH 7.2) was utilized for experiments conducted under CO₂-free aerobic conditions. H₂-supplemented microaerobic conditions were achieved through the Bio-Bag type A environmental chamber (BD), with omission of the palladium catalyst (27); this technology relies upon direct evolution of H₂ (with resultant dilution of O₂) through a KBH₄-HCl reaction. H₂-free/CO₂-supplemented microaerobic conditions were achieved through GasPak EZ Campy sachets (BD), which utilize activated charcoal/ascorbic acid to generate CO₂ from O₂ (36). CO₂ supplementation (5% CO₂) was employed for aerobic conditions. For anaerobic conditions, GasPak EZ anaerobic sachets (BD) and Bio-Bag type A chambers (with palladium catalyst) were employed.

TABLE 2.

Liquid culture media employed in this study

Medium	Component(s)	Concna
Ham's F12 mediumb	l-Arginine and l-glutamine	1 mM
	l-Proline	0.3 mM
	l-Cysteine	0.2 mM
	l-Isoleucine, l-methionine, l-phenylalanine, and l-tyrosine	0.03 mM
	l-Tryptophan	0.01 mM
	l-Lysine	0.12 mM
	Glycine, l-alanine, l-asparagine, l-aspartic acid, l-glutamic acid, l-histidine, l-leucine, l-serine, l-threonine, and l-valine	0.1 mM
	Biotin	3E−05 mM
	Choline chloride	0.1 mM
	Ca pantothenate	0.001 mM
	Folic acid	0.003 mM
	Niacinamide	3E−04 mM
	Pyridoxine HCl	2.9E−04 mM
	Riboflavin	9.8E−05 mM
	Thiamine HCl	8.9E−04 mM
	Vitamin B12	0.001 mM
	meso-Inositol	0.1 mM
	CaCl2	0.3 mM
	CuSO4	1E−05 mM
	FeSO4 and ZnSO4	0.003 mM
	MgCl2	0.6 mM
	KCl	3 mM
	NaCl	131 mM
	Na2HPO4	1 mM
	d-Glucose	10 mM
	Na hypoxanthine	0.03 mM
	Linoleic acid	4E−04 mM
	Lipoic acid	0.001 mM
	Phenol red	0.003 mM
	Putrescine 2HCl	0.001 mM
	Sodium pyruvate	1 mM
	Thymidine	0.003 mM
Mueller-Hinton broth	Casein hydrolysate	17.5 g/liter
	Beef infusion solids	2 g/liter
	Starch	1.5 g/liter
Brucella broth	Casein, pancreatic digest	10 g/liter
	Animal tissue, peptic digest	10 g/liter
	Dextrose	1 g/liter
	Yeast extract	2 g/liter
	NaCl	5 g/liter
	Sodium bisulfite	0.1 g/liter
Brain heart infusion broth	Brain infusion solids	12.5 g/liter
	Beef heart infusion solids	5 g/liter
	Proteose peptone	10 g/liter
	Dextrose	2 g/liter
	NaCl	5 g/liter
	Disodium phosphate	2.5 g/liter

^aThe concentration for each component of Ham's F12 medium is shown (e.g., 1 mM for l-arginine and 1 mM for l-glutamine).

^bNote that two different buffering systems were used for Ham's F12 medium, depending on the experimental conditions (see Materials and Methods), either bicarbonate (14 mM) or HEPES (10 mM).

Aerobic growth promotion screen.

Amino acid and carboxylate additives were acquired from Sigma-Aldrich; the stock solution pH was adjusted (7.0 to 7.2) prior to bacterial supplementation. Following initial culture in F12 under H₂-supplemented microaerobic conditions, bacteria were subcultured (optical density at 600 nm [OD₆₀₀] of 0.01) into F12 medium supplemented with each supplement(s). Growth was assessed by optical density after 24 h of incubation under H₂-free/CO₂-supplemented microaerobic conditions. Each culture was then rediluted to an OD₆₀₀ of 0.01 in the same supplemented F12 and incubated for an additional 24 h under CO₂-supplemented aerobic conditions. OD₆₀₀ was then measured again to evaluate promotion of aerobic growth.

Bacterial/monolayer coculture.

Epithelial cells were initially propagated (37°C, 5% CO₂) in Eagle's minimal essential medium (Corning) with 10% fetal bovine serum, l-glutamine (2 mM), and penicillin-streptomycin. When freshly confluent, immediately prior to bacterial inoculation, the monolayers were washed three times in F12 without serum or antibiotics. Prior to coculture, H. cinaedi (both the type strain and clinical isolates) were propagated in F12 under H₂-enhanced microaerobic conditions. The bacteria were then directly subcultured to an OD₆₀₀ of 0.01 onto the washed monolayers in F12 under the atmosphere of interest (10 ml per 25-cm² tissue culture flask) at an initial bacterial cell/epithelial cell ratio of ∼100:1. Bacterial growth was assessed by OD₆₀₀ measurement at 24 h, as neither dissociation nor overgrowth of the monolayer was evident at this time point for all cell types (at longer times, overgrowth/dissociation of monolayer was occasionally observed).

Viability staining.

During growth curve analysis of H. cinaedi BAA-847^T, bacterial viability was assessed using the BacLight bacterial viability kit (Molecular Probes). This method was employed due to the strain's poor colony-forming ability, precluding reliable CFU-based viability determination. Propidium iodide (final concentration, 30 nM) and SYTO-9 (final concentration, 5 nM) were added directly to liquid cultures and visualized under fluorescence excitation at a magnification of ×1,000 (Meiji Techno MT6000 microscope; Chroma 19002 and 19004 filters). Green- and red-staining bacteria were interpreted as viable and nonviable, respectively (500 organisms counted per replicate).

mRNA sequencing.

H. cinaedi BAA-847^T was diluted to an initial OD₆₀₀ of 0.01 for all media/atmospheric conditions, with the exception of H₂-free/CO₂-supplemented microaerobic growth in F12 medium, for which an initial OD₆₀₀ of 0.025 was used (given the organism's poor growth under these conditions). Following 24 h at 37°C, 2 volumes of RNA fixative (RNAprotect bacterial reagent; Qiagen) were added directly to the cultures. Bacteria were pelleted and preincubated (10 min, 65°C) with Max bacterial RNA enhancement reagent (Ambion). Total RNA was subject to TRIzol extraction; RNA quality was confirmed by 260/280 absorbance (Nanodrop) and 23S/16S rRNA ratios (Agilent Bioanalyzer). rRNA was subsequently depleted through the Gram-negative Ribo-Zero rRNA removal kit (Epicenter), and a cDNA library was constructed for Illumina sequencing through the TruSeq Stranded mRNA library prep kit (Illumina). The library was subjected to 50-bp paired-end sequencing on the Illumina HiSeq 2000 instrument at the Vanderbilt University genomics core facility (Vanderbilt Technologies for Advanced Genomics). Reads were aligned to the organism's genome (35) and quantified (total read counts and RPKM [reads per kilobase transcript per million reads] values) with the EDGE-pro software package (37), accessed online via the Orione Galaxy Server (38). For each biologic replicate, 18.9 to 25.8 million aligned reads were obtained. RNA-seq data were visualized as heatmaps using the R environment for statistical computing v3.3.0 (39) and the gplots analysis package (40).

Statistical analysis.

Bacterial optical density measurements were compared across culture conditions using an unpaired two-tailed Student's t test. In RNA-seq experiments, the relative abundance of each transcript was compared through DESeq2 (41). This R software package utilizes total read counts (generated here by EDGE-pro) to calculate an adjusted P value, a parameter that considers these values in light of the genomic size. Covariance analysis of RPKM values was conducted in R, with Pearson coefficients and correlation P values calculated through the HMisc package (42). Hierarchical clustering and image generation was also performed with the latter package (heatmap.2 algorithm).

Accession number(s).

For all RNA-seq experiments, raw reads (paired FASTQ files) and processed data (total assigned reads and RPKM values) have been uploaded to ArrayExpress, with accession number E-MTAB-5062. Statistical metadata for transcriptional coexpression analysis are available via figshare (https://figshare.com/articles/H_cinaedi_Pearson_Matrix_csv/3749190).

RESULTS

H. cinaedi monoculture.

In this study, we sought to define how the in vitro growth of H. cinaedi is impacted by variations in media, atmosphere, and the presence of cocultured epithelial monolayers. As a reference point for these experiments, liquid-phase growth curves were first determined for the type strain (ATCC BAA-847) under a known favorable atmosphere, H₂-enhanced microaerobic atmosphere. Optical density was assessed for monocultured bacteria in three complex media (Mueller-Hinton broth [MHB], brucella broth [BB], and brain heart infusion broth [BHI]) and one chemically defined medium (Ham's F12). In these experiments, bacteria initially proliferated with an ∼4-h doubling time (Fig. 1A), reaching a maximum density of OD₆₀₀ of 0.06 for F12 and 0.25 to 0.35 for the complex broth media. With all media, bacterial viability decreased markedly once maximum density was achieved (as determined by SYTO-9/propidium iodine staining; see the legend to Fig. 1 for additional details). Based on these curves, subsequent experiments were conducted over 24 to 48 h with an initial OD₆₀₀ of 0.01 (or ∼3.5 × 10⁷ cells/ml, per counting chamber calculation).

FIG 1.

In vitro growth of H. cinaedi ATCC BAA-847^T. (A) Bacterial growth curves. Optical density (OD₆₀₀) was assessed for up to 96 h for bacteria grown in one chemically defined medium (Ham's F12 medium [F12]) and three complex media (brucella broth [BB], brain heart infusion broth [BHI], and Mueller-Hinton broth [MH]), each under H₂-supplemented microaerobic conditions. A starting OD₆₀₀ of 0.005 was employed for each replicate; mean OD₆₀₀ values ± standard deviations (error bars) (n = 4) are plotted. For later time points, bacterial viability was assessed by SYTO-9/propidium iodide staining and fluorescence microscopy. Although maximum achievable density varied between medium types, bacterial viability decreased markedly in each case once this maximum density was reached. The following mean viabilities were observed for each medium and time point as follows: for F12, 91.5% at 16 h, 56% at 48 h, and 18.5% at 72 h; for BB, 98.4% at 48 h, 54.1% at 72 h, and unable to quantify due to ubiquitous bacterial fragmentation at 96 h; for BHI, 99.1% at 48 h, 55.7% at 72 h, and unable to quantify as before at 96 h; for MH, 98.5% at 48 h, 95.2% at 72 h, and unable to quantify as before at 96 h. Ninety-six-hour analysis was omitted for F12 due to nonviability at 72 h. (B) Variations in F12 medium. Bacteria were grown in F12 medium with variations in key culture parameters. The variations in culture parameters include either monophasic (Mono) or biphasic (Bi) F12, with (Y) or without (N) 10% supplemental FBS, and in one of three atmospheres. The atmosphere or gas conditions were as follows: 1, CO₂-supplemented aerobic; 2, H₂-free/CO₂-supplemented microaerobic; 3, H₂-supplemented microaerobic. The initial inoculum is indicated by the arrow labeled Inoc. In each case (n ≥ 4), an initial inoculation density of OD₆₀₀ of 0.01 was employed (as indicated), with 24- and 48-h optical density depicted here. Values that are significantly different (P < 0.01) at both 24 and 48 h are indicated by a bar and asterisk. Note the prominent error bars (indicated by the number sign) for serum-supplemented aerobic conditions, indicative of growth in certain replicates but static density in others. (C) Variations in complex media. Culture parameters were likewise varied for MH, BB, and BHI, including 10% FBS supplementation and atmosphere (n ≥ 4) (same abbreviations as in panel B). All conditions represent monophasic growth. For the sake of clarity, designations of statistical significance are not included, although growth in atmosphere 3 > growth in atmosphere 2 > growth in atmosphere 1 for all media. Prominent error bars (indicated by the number sign) again highlight growth in certain replicates—but static density in others—for aerobic, serum-supplemented BB. (D) Epithelial coculture. H. cinaedi was inoculated directly onto Caco-2 epithelial cells and incubated under CO₂-supplemented aerobic (O₂ + CO₂) or H₂-supplemented microaerobic (mO₂ + H₂) conditions. Bacteria were likewise inoculated over LS-174T, AGS, and HeLa cells and incubated under CO₂-supplemented aerobic conditions. Bacterial densities at 24 h are summarized here (n ≥ 4). For each monolayer type, epithelial coculture permitted aerobic proliferation of the bacteria. No aerobic growth was observed for monolayer-free conditions (X). *, P < 0.001.

For each of the above media, culture parameters were varied to determine their effect on H. cinaedi proliferation. Three atmospheres were compared: H₂-supplemented microaerobic, H₂-free/CO₂-supplemented microaerobic, and CO₂-supplemented aerobic. In each case, H. cinaedi was first propagated under H₂-supplemented microaerobic conditions, followed by subculture in the atmosphere of interest and assessment of growth. For F12 medium, monophasic growth (liquid alone) in each atmosphere was compared to biphasic growth (liquid over agar) and serum-supplemented growth (F12 plus 10% FBS). As depicted in Fig. 1B, biphasic F12 medium led to higher bacterial density under H₂-supplemented microaerobic conditions (relative to monophasic), although it was not associated with increased growth under H₂-free/CO₂-supplemented microaerobic conditions or with any growth under aerobic conditions. In contrast, the presence of serum enhanced bacterial proliferation in F12 under all three atmospheres, including (most notably) aerobic growth. The effect of serum under aerobic conditions was inconsistent, however, as aerobic growth was present in certain serum-supplemented replicates but absent in others.

Similar experiments were conducted in complex media (MHB, BB, and BHI) under the same three atmospheres, both with and without 10% FBS (Fig. 1C). As expected, H₂ supplementation uniformly increased bacterial growth under microaerobic conditions. Likewise, serum supplementation of complex media enhanced microaerobic growth, and once more, it facilitated occasional H. cinaedi proliferation under aerobic conditions. Nevertheless, the effect was again inconsistent: aerobic growth was limited to serum-supplemented BB (not BHI/MHB) and was variable from one replicate to the next. Together, the above findings indicate that H. cinaedi will occasionally grow aerobically in the presence of serum, although other environmental or stochastic factors likely influence this ability. The behavior, in fact, is reminiscent of the isolation of H. cinaedi from aerobic blood culture bottles; the latter is a well-described phenomenon in clinical settings, albeit one that occurs with noteworthy interlaboratory variability (30, 34).

H. cinaedi coculture with epithelial cells.

The preceding observations led us to hypothesize that one or more host-derived factors present in serum were responsible for promoting aerobiosis in H. cinaedi. Therefore, the organism's growth was next evaluated during coculture with epithelial monolayers. The type strain was inoculated in serum-free F12 medium over freshly confluent Caco-2 intestinal cells. Compared to monocultured bacteria, the cocultured bacteria not only grew to higher densities under microaerobic conditions, but they consistently proliferated in a CO₂-supplemented aerobic environment (Fig. 1D). The ability of epithelial cells to promote H₂-free aerobic growth of H. cinaedi ATCC BAA-847^T was not limited to Caco-2 monolayers. The same phenomenon occurred with several other gastrointestinal and nongastrointestinal cell lines (LS-174T, AGS, and HeLa [Fig. 1D]), as well as polarized Caco-2 cells from mature Transwell cultures (more than 2 weeks [data not shown]). Caco-2 cells likewise stimulated H₂-free aerobic growth when separated from H. cinaedi by a Transwell barrier (see Fig. S1 in the supplemental material). Spent Caco-2 supernatant promoted mild growth enhancement, while boiling the spent media did not abolish this effect (Fig. S2). From these observations, we suspected that a diffusible metabolite—produced in common by the monolayers—was responsible (at least in part) for the altered atmospheric requirements.

l-Lactate supplementation.

To identify potential growth-promoting metabolites, 40 amino acids and short-chain carboxylates were next screened (10 mM each) for their effect on the H. cinaedi type strain under sequential microaerobic and aerobic atmospheres (see Materials and Methods for procedural specifics). Microaerobic and aerobic growth enhancement at 24 h was noted for several supplements, including pyruvate, propionate, acetate, and serine (confirmed in additional experimental replicates), although the effect of l-lactate (but not d-lactate) was most prominent (Fig. 2). Note that unsupplemented F12 medium contains a low concentration of pyruvate (1 mM) and serine (0.1 mM) but none of the other growth-promoting molecules. The molecules promoting aerobic growth were also tested in pairwise combinations (5 mM each), although they demonstrated no synergistic effects above 10 mM l-lactate alone. l-Lactate concentration was varied in subsequent trials, employing the same microaerobic-to-aerobic experimental strategy. In these experiments, growth promotion was observed at 24/48 h in both atmospheres with as little as 1 mM supplementation (Fig. 3A). Once aerobic growth was achieved, it continued in l-lactate supplemented F12 for multiple subcultures and was lost upon the molecule's removal (see Fig. S3 in the supplemental material). These results demonstrate that, in the proper context, the organism can proliferate without H₂ under ambient oxygen levels.

FIG 2.

Growth promotion screen. Microaerobic and aerobic growth promotion was evaluated at 24 h in F12 medium supplemented with a 10 mM concentration of the indicated molecule or 5 mM each when two molecules are listed in combination (l-lactate [Lac], pyruvate [Pyr], serine [Ser], and propionate [Prop]). Every supplement was initially screened through two independent replicates (the average of which is plotted here). For conditions with increased growth, additional replicates were performed to confirm statistical significance compared to unsupplemented culture (n ≥ 4; P < 0.001 for both white and gray bars, except acetate [white bars only]).

FIG 3.

l-Lactate promotion of H. cinaedi growth. (A) Variable l-lactate concentration. The concentration of supplemental l-lactate (in F12 medium) was varied from 100 μM to 25 mM. For both H₂-free/CO₂-supplemented microaerobic and CO₂-supplemented aerobic conditions, growth was assessed by OD₆₀₀ measurement at 24 and 48 h (n = 4). With both atmospheres and time points, statistically significant increases in bacterial density (relative to unsupplemented) were noted for 1 mM l-lactate, with more-prominent changes at 10 and 25 mM (*, P < 0.001). (B) Effects of preceding culture conditions on aerobic growth. With 10 mM l-lactate supplementation, the stepwise transition from H₂-free/CO₂-supplemented microaerobic to CO₂-supplemented aerobic conditions was systemically altered in four manners (as depicted in the figure and described in the text). For each daily subculture, bacteria were rediluted to an OD₆₀₀ of 0.01 in monophasic F12. Final, aerobic OD₆₀₀ values for each scenario at 24 h are depicted (gray bars), along with the corresponding lactate-free controls (white bars). Consistent aerobic growth promotion was observed for all scenarios (n = 4), except when bacteria were transitioned directly from lactate-free media under H₂-supplemented microaerobic conditions to lactate-containing media under CO₂-supplemented aerobic conditions. Here, aerobic bacterial growth was observed only in certain replicates, as indicated by prominent error bars indicated by the number sign.

It is important to emphasize that the above-described experiments utilized the same stepwise transition from microaerobic to aerobic conditions as described previously: preliminary growth, l-lactate-free F12, H₂-supplemented microaerobic atmosphere; experimental day 1, l-lactate-supplemented F12, H₂-free/CO₂-supplemented microaerobic atmosphere; experimental day 2, l-lactate-supplemented F12, CO₂-supplemented aerobic atmosphere. We therefore investigated whether the specific parameters of this transition affected the ability of l-lactate to promote aerobic growth. Variations were made to the above protocol, including (i) H₂-supplemented versus H₂-free/CO₂-supplemented microaerobic atmosphere on experimental day 1 and (ii) l-lactate supplementation during initial microaerobic conditions (experimental day 1) versus l-lactate addition concurrent to the aerobic transition (experimental day 2). For each possible transition (denoted I to IV), the final levels of 24-h aerobic growth after day 2 are summarized in Fig. 3B. As long as the initial microaerobic culture on experimental day 1 included either CO₂ or l-lactate supplementation, aerobic growth was observed in all experimental replicates. The organism demonstrated variable behavior, however, when l-lactate was added at the same time as the transition from an H₂-supplemented microaerobic atmosphere. Here, aerobic growth was observed in certain replicates but not in others (evident by the large error bars). The latter data were from experiments with monophasic media, although an identical pattern occurred in biphasic F12, with proliferation in only some replicates (see Fig. S4 in the supplemental material). Overall, these findings indicate that, while l-lactate can support H₂-free aerobic growth, the preceding environment influences the transition to this growth regimen.

Finally, several additional experiments were conducted to further define the environmental requirements of the phenomenon. Most notably, aerobic proliferation with l-lactate was dependent on the presence of CO₂, as no growth was observed under 20% O₂–80% N₂, even in HEPES-buffered F12 (which does not require ambient CO₂ to maintain neutral pH; Fig. 4). In this light, CO₂ might serve as an anaplerotic carbon source under aerobic conditions. In contrast, aerobic proliferation of H. cinaedi was not observed on the surfaces of l-lactate-supplemented F12 agar plates, regardless of the presence of CO₂ (not shown). The molecule also failed to promote growth under anaerobic conditions, for which optical density remained static or increased only slightly in both supplemented and unsupplemented F12 (see Fig. S5 in the supplemental material).

FIG 4.

CO₂ requirement for aerobic growth. H. cinaedi ATCC BAA-847^T was propagated in F12 medium buffered with either bicarbonate or HEPES. Following preculture in the respective medium with (+) 10 mM l-lactate under H₂-free/CO₂-supplemented microaerobic conditions, the bacteria were subcultured to an OD₆₀₀ of 0.01 and propagated in the same medium under various atmospheres (n = 3). In the absence of CO₂, neither the bicarbonate–buffered culture nor the HEPES-buffered culture was able to proliferate aerobically at 24 h in the presence of l-lactate (*, P < 0.001 relative to each of the other two conditions [white and gray bars]).

Aerobic growth of clinical isolates.

As the above-described experiments were conducted with the H. cinaedi type strain, we sought to confirm the phenomena with additional clinical isolates. Twelve human-derived strains were screened for their ability to proliferate aerobically in the presence of Caco-2 cells and l-lactate. Following initial culture in F12 medium under H₂-supplemented microaerobic conditions, all clinical strains proliferated when inoculated directly onto Caco-2 monolayers under CO₂-supplemented aerobic conditions, while no aerobic growth was observed during monoculture in F12 (Fig. 5A). For experiments with l-lactate, the same four microaerobic-to-aerobic transitions previously used for studies of the type strain were repeated. Summarized in Fig. 5B, all 12 strains continued to proliferate aerobically when precultured with l-lactate (10 mM) under H₂-free/CO₂-supplemented microaerobic conditions. In comparison, only certain strains grew aerobically if l-lactate presupplementation was omitted or if the aerobic transition occurred directly from an H₂-supplemented microaerobic environment (see the legend to Fig. 5B for specific details). These observations further emphasize that the preceding environmental conditions impact the transition of H. cinaedi to an H₂-independent aerotolerant state.

FIG 5.

Growth of clinical H. cinaedi strains. (A) Caco-2 coculture. Using the same experimental approach employed for the type strain (Fig. 1D), 12 additional human-derived strains of H. cinaedi were screened for aerobic growth promotion by Caco-2 cells. For all strains, aerobic growth was observed at 24 h for cocultured bacteria [F12 (+) Caco-2], but not corresponding monoculture controls [F12 (−) Caco-2]. Each bar represents a single replicate. (B) Aerobic growth promotion by l-lactate. The 12 clinical strains were likewise cultured with 10 mM l-lactate (in F12) and screened for growth in a CO₂-supplemented aerobic atmosphere. As with the H. cinaedi type strain, several stepwise transitions of media and atmosphere were evaluated (with or without l-lactate preculture, with or without H₂ preculture). These are indicated as transitions I to IV, as defined for the type strain in Fig. 3B. For each strain/transition, 24-h OD₆₀₀ values are depicted for the final aerobic cultures. Two l-lactate-free negative controls are likewise included for each strain: aerobic growth in F12 following preculture in H₂-supplemented microaerobic (Neg. Control 1) and H₂-free/CO₂-supplemented microaerobic (Neg. Control 2) conditions. All data points represent the averages of two independent replicates. When first cultured with l-lactate under an H₂-free/CO₂-supplemented microaerobic atmosphere (transition I), all 12 strains consistently proliferated following the aerobic transition. Only a portion of the strains continued to proliferate aerobically, however, if the preceding microaerobic environment failed to include l-lactate (transitions II and IV) or was supplemented with H₂ (transitions III and IV).

Differential transcription analysis.

To investigate the organism's physiology during microaerobic and aerobic growth, H. cinaedi was next subjected to whole-transcriptome profiling. High-throughput RNA sequencing (RNA-seq) was performed on the type strain for various conditions of bacterial monoculture and epithelial coculture (unless noted, monophasic medium is implied): (condition 1) F12, H₂-supplemented microaerobic (n = 4); (condition 2) biphasic F12, H₂-supplemented microaerobic (n = 4); (condition 3) biphasic F12, H₂-free/CO₂-supplemented microaerobic (n = 4); (condition 4) F12 plus Caco-2 cells, H₂-supplemented microaerobic (n = 5); (condition 5) F12 plus Caco-2 cells, aerobic (n = 5); (condition 6) F12 plus 10 mM l-lactate, H₂-supplemented microaerobic (n = 4); (condition 7) F12 plus 10 mM l-lactate, aerobic (n = 4); and (condition 8) MHB, H₂-supplemented microaerobic (n = 4). Expression levels for protein encoding open reading frames (ORFs) were calculated with the EDGE-pro software package for prokaryotic transcriptomics (37).

The distribution of RPKM values (reads per kilobase transcript per million reads), averaged across all conditions/replicates, is summarized as a histogram in Fig. 6. Several genes of interest are denoted explicitly in Fig. 6 and discussed in the following paragraphs. RPKM values for each gene—averaged across all culture conditions—were ranked by top abundance. The 25 highest values are summarized in Table S1 in the supplemental material, with a complete ordered list provided in Data Set S1 in the supplemental material. Many of the most abundant genes served clear housekeeping (ATP synthase, GroEL, and EF-Tu) or antioxidative (alkyl hydroperoxide reductase, peroxiredoxin, and superoxide dismutase) roles. For several other transcripts, however, prominent expression was unanticipated, and their specific physiologic roles remain to be determined. These transcripts include the hcp component of a predicted type VI secretion locus (ORF 1435), as well as two proteins with predicted signal peptides but lacking any homologues of known function (ORFs 542 and 1720).

FIG 6.

Distribution of transcript levels across the H. cinaedi genome. For each predicted gene in H. cinaedi ATCC BAA-847^T, RPKM values (reads per kilobase transcript per million reads mapped) were calculated. The distribution of these values—averaged across the 8 culture conditions and 34 biologic replicates—is plotted here for the 2,321 predicted protein-encoding ORFs. Values have been converted to a log₁₀ scale, with a bin width of 0.1. RPKM values for several genes discussed in the text are indicated explicitly. These include genes encoding components of respiratory nitrate reductase under both Caco-2/l-lactate supplemented (NO₃⁻ Red., Supp.) and unsupplemented (NO₃⁻ Red., Unsupp.) conditions (further summarized in Table 3), components of the lactate utilization operon (lctP and lldE [Table 4]), and the organism's most abundant protein-encoding transcripts (see Table S1 in the supplemental material).

In Fig. 7A, relative transcription of each gene is illustrated as a heatmap according to the experimental condition and its position in the genome. From these data, expression levels were compared on a gene-by-gene basis through DESeq-2 (41). Differentially expressed genes were determined for the various pairwise combinations of the eight culture conditions. For the purposes of this discussion, a gene is considered differentially expressed if its detection changed >2-fold between conditions with an adjusted P value of <0.001. For each condition-condition comparison, the number of differentially expressed genes (out of 2,368 in the genome) is summarized in Fig. 7B; complete lists of these genes are provided in Data Set S2 in the supplemental material. Of particular interest are genes that were differentially expressed with Caco-2 and l-lactate supplementation (conditions 4 to 7) relative to all other conditions with unsupplemented F12 and MHB (conditions 1, 3, 5, and 8).

FIG 7.

Differential transcription across culture conditions. (A) Relative transcriptional abundance for each gene and culture condition. RNA-seq transcript levels for each ORF in H. cinaedi ATCC BAA-847^T are summarized across eight culture conditions as follows (conditions 1 to 8 indicated by the numbers 1 to 8 within brackets): [1], monophasic F12, H₂-supplemented microaerobic; [2], biphasic F12, H₂-supplemented microaerobic; [3], biphasic F12, H₂-free/CO₂-supplemented microaerobic; [4], F12 plus Caco-2 cells, H₂-supplemented microaerobic; [5], F12 plus Caco-2 cells, aerobic; [6], F12 plus 10 mM l-lactate, H₂-supplemented microaerobic; [7], F12 plus 10 mM l-lactate, aerobic; [8], MHB, H₂-supplemented microaerobic. Rows of the heatmap correspond to individual genes (in sequential order), while columns denote culture conditions. Average RPKM values for each gene/condition are designated according to log₂ values. (B) Differentially expressed genes. Transcription levels of all genes were compared among the above culture conditions in a pairwise manner. The matrix summarizes the number of differentially transcribed genes between the stated conditions (in the column and row headers). A complete list of these genes for each comparison is included in Data Set S2 in the supplemental material.

For these 16 pairwise comparisons (conditions 4 to 7 versus conditions 1, 3, 5, and 8), the overlapping set of differentially expressed genes was limited to two loci, a putative six-gene operon (ORFs 1272 to 1277) encoding the components of a Nap class periplasmic nitrate reductase (napAGHBLD) and a two-gene operon (ORFs 2132 and 2133) encoding a putative trimethylamine-N-oxide/dimethyl sulfoxide (TMAO/DMSO). Mean RPKM values for these genes across all eight conditions are summarized in Table 3 and denoted in Fig. 5. Nap class enzymes are distinct from Nar class nitrate reductases that generate a proton motive force in various proteobacteria as alternate terminal electron acceptors (Nap enzymes are notably absent from the epsilonproteobacterial subdivision) (43). Nevertheless, periplasmic nitrate reductases may still participate indirectly in anaerobic respiration, as well as other processes in dissimilitatory nitrogen metabolism (44–46). In Escherichia coli, the nap operon is preferentially expressed during anaerobic, low-nitrate conditions and is additionally regulated by cyclic AMP (cAMP)-mediated catabolite repression (47). In H. cinaedi, by contrast, observed transcription of these genes decreased significantly (up to ∼10-fold) in the presence of Caco-2 cells or l-lactate, whether in an H₂-enhanced microaerobic atmosphere (conditions 4 and 6) or an aerobic atmosphere (conditions 5 and 7). ORF 2132 is homologous to the torA gene of Campylobacter jejuni, which likewise acts as an alternate terminal electron acceptor through the reduction of TMAO/DMSO (48). Although the altered expression levels of these operons in H. cinaedi do not elucidate the underlying mechanism of aerobic growth per se, they suggest an organism whose redox physiology is adapted to multiple atmospheric/nutritional environments.

TABLE 3.

RNA-seq analysis of periplasmic nitrate reductase and TMAO/DMSO reductase genes

Conditiona	Mean RPKM value
	Periplasmic nitrate reductaseb						TMAO/DMSO reductasec
	napA	napG	napH	napB	napL	napD	torA	ORF 2133
1	2,377	1,815	1,021	1,046	261	255	679	464
2	3,305	2,483	1,392	1,384	376	224	1,135	892
3	3,060	2,629	1,357	1,284	242	136	1,289	543
4*	341	297	168	153	54	29	102	33
5*	333	327	195	186	79	51	83	36
6*	158	169	78	99	36	33	141	60
7*	243	261	120	133	54	40	104	41
8	2,792	2,724	1,142	1,307	376	169	2,775	920

^aAtmosphere and media of the given culture condition, as defined by conditions 1 to 8 in the body of the text and in the legend to Fig. 7. Caco-2 and l-lactate-supplemented conditions (conditions 4 to 7) are indicated with an asterisk; note downregulation of both operons for these conditions, relative to conditions 1 to 3 and 8.

^bMean RPKM values for the following genes encoding components of the periplasmic nitrate reductase: napA (ORF 1272, NCBI accession number BAM32505), napG (ORF 1273, BAM32506), napH (ORF 1274, BAM32507), napB (ORF 1275, BAM32508), napL (ORF 1276, BAM32509), and napD (ORF 1277, BAM32510).

^cMean RPKM values for the genes encoding components of the putative TMAO/DMSO reductase: torA (ORF 2132, NCBI accession number BAM33348) and ORF 2133 (BAM33349).

Analysis of l-lactate utilization genes.

The genes responsible for l-lactate utilization were recently characterized in two species of Epsilonproteobacteria, Helicobacter pylori (49) and C. jejuni (50). In both organisms, l-lactate permease (LctP) is encoded divergently from a three-gene segment that includes putative iron-sulfur clusters and comprises a nonfermentative l-lactate dehydrogenase (iLDH). These enzymes are distinct from classic NAD-dependent lactate dehydrogenases, which interconvert l-lactate and pyruvate during fermentative metabolism (51). Instead, iLDH generates pyruvate from l-lactate (typically in an NAD-independent fashion) and can potentially serve as a respiratory electron donor. The LctP/iLDH element in H. pylori and C. jejuni is conserved broadly among many Gram-positive and Gram-negative species, and it was first characterized in Bacillus subtilis (as the lut operon) (52) and Shewanella oneidensis (53).

As depicted in Fig. S6 in the supplemental material, H. cinaedi encodes a similar element with putative genes for LctP and iLDH from ORFs 304 to 307 (denoted here as lctP and lldEFG, keeping with the H. pylori nomenclature). The current RNA-seq analysis revealed prominent transcription of both lctP and lldEFG, with RPKM values ranging (per condition) from 1.27 × 10³ to 6.40 × 10³ for lctP and 4.7 × 10² to 2.3 × 10³ for the lld components (Table 4). Across all culture conditions/replicates, observed coverage of these genes fell within the top 15% of all protein-encoding ORFs in the organism's genome. At the same time, only mild differential expression of lctP and lld (less than a twofold change) was observed between most conditions. Upregulation of lctP was noted for bacteria grown with biphasic media (conditions 2 and 3), although the significance of this observation is unclear (for condition 3, an additional contributing factor could be the H₂-free/CO₂-supplemented microaerobic atmosphere). The lack of more-pronounced transcriptional changes during l-lactate supplementation is perhaps surprising; to our knowledge, similar transcriptional studies of this locus have not been conducted in other epsilonproteobacteria. Nevertheless, the current results suggest strong constitutive transcription of l-lactate utilization machinery and a prominent role for this nutrient in the natural physiology of H. cinaedi.

TABLE 4.

RNA-seq analysis of the l-lactate utilization locus

Conditiona	Mean RPKM valueb
	lldG	lldF	lldE	lctP
1	1,059	1,259	1,031	1,706
2	1,114	1,463	1,662	5,534
3	1,036	1,907	2,329	6,395
4*	909	1,295	1,189	1,309
5*	1,232	1,735	1,557	2,157
6*	685	1,791	1,608	1,356
7*	923	1,934	1,566	1,562
8	468	922	966	1,266

^aAtmosphere and media of the given culture condition, as defined by numbers 1 to 8 in the body of the text and the legend to Fig. 7. Caco-2 and l-lactate-supplemented conditions are indicated with an asterisk.

^bMean RPKM values of l-lactate dehydrogenase component lldG (ORF 304), l-lactate dehydrogenase component lldF (ORF 305), l-lactate dehydrogenase component lldE (ORF 306), and l-lactate permease, lctP (ORF 307) are shown.

Coexpression analysis.

To evaluate global transcriptional relationships, coexpression analysis was conducted across all biological replicates and culture conditions. This includes the eight scenarios described above, as well as H. cinaedi grown under H₂-supplemented microaerobic conditions in F12 medium supplemented with 10% human serum (n = 5). The latter condition was excluded from the previous analyses due to citrate supplementation of the serum (as a chelating agent), confounding the influence of the serum itself on bacterial physiology. RPKM-based Pearson coefficients (r_i,j) were calculated for each gene-gene pair, summarized in Fig. 8 as a hierarchically clustered heatmap. Data Set S3 in the supplemental material includes a complete list of gene-gene pairs with prominent coexpression (|r_i,j| > x, with analyses provided for x = 0.7 and 0.8); a file with all pairwise Pearson coefficients and corresponding P values is available for download via figshare [see “Accession number(s)” above].

FIG 8.

Transcriptional coexpression analysis. RPKM-based Pearson coefficients were calculated for each gene-gene pair in the H. cinaedi genome, with absolute values (abs) summarized here as a heatmap. Green clusters represent genes with high coexpression trends (either positive or negative).

For several genes discussed above, napA and lctP, coexpression partners are summarized explicitly in Tables S2 and S3 in the supplemental material. For napA, this includes inverse correlation with an operon encoding a molybdenum-binding periplasmic protein and ABC transporter (ORFs 299 to 302), appropriate given that periplasmic nitrate reductases are known molybdoenzymes. Strong positive correlation was observed between transcription of lctP and several enzymes of central carbon metabolism (e.g., pyruvate:ferredoxin oxidoreductase [ORF 403/405], acetate kinase [ORF 2183], and acetyl coenzyme A [acetyl-CoA] synthetase [ORF 1787]), as well as additional carbon uptake proteins. The latter include a putative acetate permease (ORF 523), serine transporter (ORF 58), alanine transporter (ORF 1209), and carbon starvation protein A (ORF 861). Of note, both acetate and serine demonstrated a mild aerobic growth-promoting effect in the initial supplement screen of Fig. 2. Together, these correlations further suggest that lctP is a central component in the organism's network of nutrient acquisition and metabolism, in response to changing environmental milieus.

DISCUSSION

The work presented here demonstrates that H. cinaedi is not exclusively dependent on H₂-enhanced microaerobic conditions for proliferation. Epithelial monolayers, l-lactate (given the appropriate preculture conditions), and FBS (variably) all have the potential to support H₂-free aerobic growth of this species. Practically speaking, we recommend the following approach for propagating H. cinaedi in liquid culture without the use of supplemental H₂ gas. Under CO₂-supplemented microaerobic conditions (i.e., commercial sachets), either serum-supplemented brucella broth or l-lactate-supplemented F12 medium can be employed. The latter medium can then be transitioned to CO₂-supplemented aerobic conditions for further subculture.

The basic phenomenon of microaerophilicity, generally attributed to oxidative stress tolerance (54), has still not been characterized in full. Our findings challenge the traditional paradigm that many epsilonproteobacteria are invariably microaerophilic, instead demonstrating that the atmospheric and small-molecule requirements of H. cinaedi are interconnected. This phenomenon has implications for improving in vitro culture methodologies and increasing our understanding of H. cinaedi metabolism during infection. The ability to work with this organism under aerobic conditions could represent a useful tool, particularly in clinical laboratories—where H₂-supplemented microaerobic environments are not universally employed—or when high quantities/concentrations of H₂ are not desired for safety reasons. In general, l-lactate might serve as a nutritional supplement for H. cinaedi propagation. One could likewise hypothesize a role for l-lactate metabolism during H. cinaedi infections, especially given the relative abundance of l-lactate in blood and tissue as an endpoint of eukaryotic nonrespiratory metabolism. Although the gut is characterized by low O₂ levels and flora-generated H₂ (29), the same is not true of the bloodstream or peripheral soft tissue. At these extraintestinal sites of infection where H₂ levels are low/absent, l-lactate could serve as an important host-derived metabolite. Additional studies will be needed to test these hypotheses.

As described in Results (see “Analysis of l-lactate utilization genes”), other members of the Epsilonproteobacteria likewise possess the ability to utilize l-lactate as a carbon source. Although l-lactate metabolism has not previously been investigated in the context of atmospheric requirements, limited previous work has demonstrated that C₃- and C₄-carboxylates (in particular, pyruvate and fumarate) can improve the aerobic growth of C. jejuni (55, 56). Otherwise, the relationship between nutrients and ambient gas in the epsilonproteobacteria remain poorly defined. In future work, it will be of interest to test the ability of l-lactate to promote aerobic growth of other species of the class, especially emerging pathogens whose cultivation is difficult without supplemental H₂ (e.g., Campylobacter concisus, Helicobacter fennelliae) (27). The current work has less potential relevance to cultivation of the gastric pathogen Helicobacter pylori, which does not demonstrate a strong H₂ dependence and exhibits only scant growth in F12 medium, modestly enhanced by supplemental cholesterol (57). In preliminary experiments (unpublished), l-lactate failed to enhance H. pylori proliferation under aerobic or microaerobic conditions. At the same time, it is noteworthy that several reports have highlighted that H. pylori is not invariably microaerobic (58, 59). Although unrelated to l-lactate, the ability to propagate certain strains of H. pylori under aerobic conditions (in serum-supplemented, complex media with sufficient CO₂ supplementation) is routinely exploited within many research laboratories.

The underlying mechanism by which l-lactate induces aerotolerance in H. cinaedi likewise requires additional study. The organism's alternate growth conditions might reflect its energetic/carbon source requirements and the various biochemical combinations that satisfy them (for instance, H₂ gas versus lactate as a respiratory electron donor). In this regard, H₂-free aerobic growth was promoted to a lesser degree by several other small molecules (e.g., pyruvate, serine), all metabolites interconvertible with l-lactate by one or two enzymatic reactions. Notably, whereas l-lactate-induced aerobiosis variably required microaerobic presupplementation, growth was universally observed when H. cinaedi was transitioned from microaerobic monoculture to aerobic epithelial coculture. It is certainly possible that the growth-promoting effect of epithelial cells is attributable to multiple factors, as the relationship between H. cinaedi and epithelial cells is potentially complex. This is underscored by recently published data on the physical interactions between the organism and polarized Caco-2 monolayers (60). Although that study did not address the phenomenon of induced aerobic growth, it reported the ability of H. cinaedi to adhere to, invade, and translocate across the cells.

Finally, the current RNA-seq data support the importance of l-lactate as an H. cinaedi nutrient, as well as the organism's potential respiratory flexibility. The l-lactate utilization locus demonstrated prominent transcription across culture conditions, along with strong coexpression trends with other nutrient acquisition proteins and enzymes of central carbon metabolism. In addition, supplemental l-lactate and Caco-2 cells significantly downregulated transcription of respiratory electron acceptors, namely, a periplasmic nitrate reductase and TMAO/DMSO reductase. Future work might be able to evaluate the organism's metabolism, as well as its transcriptional regulation, in the context of a colonized/infected mammalian host. In general, RNA-seq represents a valuable method both to generate and test hypotheses for organisms like H. cinaedi, where extensive methods for genetic manipulation are currently lacking.

To conclude, the present work demonstrates context-dependent atmospheric requirements for H. cinaedi, an emerging opportunistic pathogen of diverse anatomic sites whose fastidious growth can hinder both laboratory diagnosis and investigative efforts. Going forward, a greater understanding of this organism's metabolism could shed light on its enigmatic pathogenesis and optimize methods for identifying cases of infection.

Funding Statement

This work, including the efforts of Jonathan E. Schmitz, was funded by the Department of Pathology, Microbiology, and Immunology, Vanderbilt University School of Medicine (Clinical and Translational Research Enhancement Award). This work, including the efforts of Timothy L. Cover, was funded the Vanderbilt Digestive Diseases Research Center (VDDRC). The VDDRC is supported by HHS | National Institutes of Health (NIH) (P30DK058404).