Growth Characteristics of Bartonella henselae in a Novel Liquid Medium: Primary Isolation, Growth-Phase-Dependent Phage Induction, and Metabolic Studies

Abstract

Bartonella henselae is a zoonotic pathogen that usually causes a self-limiting infection in immunocompetent individuals but often causes potentially life-threatening infections, such as bacillary angiomatosis, in immunocompromised patients. Both diagnosis of infection and research into the molecular mechanisms of pathogenesis have been hindered by the absence of a suitable liquid growth medium. It has been difficult to isolate B. henselae directly from the blood of infected humans or animals or to grow the bacteria in liquid culture media under laboratory conditions. Therefore, we have developed a liquid growth medium that supports reproducible in vitro growth (3-h doubling time and a growth yield of approximately 5 × 10⁸ CFU/ml) and permits the isolation of B. henselae from the blood of infected cats. During the development of this medium, we observed that B. henselae did not derive carbon and energy from the catabolism of glucose, which is consistent with genome nucleotide sequence data suggesting an incomplete glycolytic pathway. Of interest, B. henselae depleted amino acids from the culture medium and accumulated ammonia in the medium, an indicator of amino acid catabolism. Analysis of the culture medium throughout the growth cycle revealed that oxygen was consumed and carbon dioxide was generated, suggesting that amino acids were catabolized in a tricarboxylic acid (TCA) cycle-dependent mechanism. Additionally, phage particles were detected in the culture supernatants of stationary-phase B. henselae, but not in mid-logarithmic-phase culture supernatants. Enzymatic assays of whole-cell lysates revealed that B. henselae has a complete TCA cycle. Taken together, these data suggest B. henselae may catabolize amino acids but not glucose to derive carbon and energy from its host. Furthermore, the newly developed culture medium should improve isolation of B. henselae and basic research into the pathogenesis of the bacterium.

Bartonella henselae is a medically important gram-negative, zoonotic pathogen. Cats are the reservoir, and infections are generally asymptomatic. B. henselae causes a diverse and emerging disease spectrum in humans, from self-limiting lymphadenopathy to life-threatening conditions such as bacillary angiomatosis. Cat scratch disease (CSD), a generally benign infection characterized by regional lymphadenopathy and persistent fever, is the most-common manifestation of B. henselae infection in humans. CSD occurs in healthy individuals and affects an estimated 24,000 persons a year in the United States (11). In immunocompromised patients, B. henselae causes more-serious infections, including bacillary angiomatosis and bacillary peliosis, which can be fatal when misdiagnosed and improperly treated. A major contributing factor to misdiagnosis is the inherent difficulty in culturing the bacterium. B. henselae takes an average of 21 days to form colonies during primary isolation on blood agar plates, and no reliable liquid growth medium exists (8). In clinical settings, the protracted growth phase often leads to false-negative culture results, making the isolation of new strains very problematic. The lack of a suitable liquid medium has also severely hindered basic studies of B. henselae biology and pathogenesis.

Little is known of the metabolic requirements of B. henselae. When the bacterium was first isolated from the blood of patients with human immunodeficiency virus, preliminary characterization of the pathogen did not detect any carbohydrate utilization (19, 24). Similarly, the closely related pathogen Bartonella quintana does not metabolize glucose (10, 23). This is unusual for a pathogenic organism, since glucose is an abundant, readily catabolizable carbon source in host tissue and blood. Recently, it was reported that the unpublished B. henselae genome contains most of the genes necessary for the Embden-Meyerhoff-Parnas and Entner-Doudoroff pathways but lacks hexokinase and phosphofructokinase (4). Since these pathways are the most widely conserved mechanisms of glucose catabolism in eukaryotic and prokaryotic cells, the authors suggest that B. henselae may utilize a chimeric Embden-Meyerhoff-Parnas-Entner-Doudoroff pathway to generate energy from glucose. This speculation would seem to contradict previous data that B. henselae does not utilize carbohydrates.

To facilitate isolation of B. henselae and answer fundamental questions about its life cycle and pathogenesis, we developed a liquid culture medium that supports consistent B. henselae growth, yielding good cell densities and a 3-h doubling time. This novel medium has been used successfully for isolation of the bacterium from the blood of experimentally infected cats and has allowed us to elucidate the unusual central metabolism of the pathogen.

MATERIALS AND METHODS

Bacterial strains, growth conditions and chemicals.

B. henselae strains Houston-1, passage 4 (P4), LSU16 (P4), and ATCC 49793 (HP), and B. quintana OK 90-268 (P2) were grown in Brucella broth histidine-hematin (BBH-H) (described below and summarized in Table 1), as well as in two previously published liquid media (20, 26). B. henselae Houston-1 was used for all metabolic studies. Liquid cultures were incubated at 37°C with shaking (180 rpm) under different gas mixtures. Aerobic cultures were grown under normal atmospheric oxygen (21% O₂); microaerobic cultures were grown under an atmosphere of 7% O₂, 5% CO₂, and 88% N₂; and anaerobic cultures were grown under 4% H₂, 5% CO₂, and 91% N₂. Microaerobic cultures were prepared by three 10-min cycles alternating between degassing by vacuum and replacing dissolved gas with premixed gas to achieve the desired gas concentrations. Anaerobic medium was prepared by degassing medium overnight in an anaerobic chamber containing the gas mix detailed above. Microaerobic and anaerobic media were prepared in sealed, sidearm flasks, and samples were withdrawn by syringe through a rubber septum to maintain the appropriate atmosphere within the flask. Unless otherwise noted, cultures were grown under aerobic conditions. To determine the cell density and viability of B. henselae in liquid culture, aliquots were serially diluted and plated on BBH-H agar (BBH-HA) (described below) to determine the number of CFU per milliliter and plot growth curves. Plates were incubated at 34°C under atmospheric oxygen with 3% CO₂. Cell viability was also determined using a Molecular Probes (Eugene, Oreg.) BacLight fluorescent staining kit with a Nikon Eclipse E800 fluorescent microscope according to the manufacturer's instructions. All chemicals and reagents were purchased from Sigma Chemical, St. Louis, Mo., unless stated otherwise.

TABLE 1.

Components of liquid media developed for cultivation of B. henselae

Medium	Base medium (g/liter)	Fe source (mg/liter)	Buffer (g/liter)	Supplement(s)	Final pH	Reference
BBH-H	Brucella broth (28)	Histidine-hematin (100)	HEPES (23.8)	Histidine (0.4%, wt/vol)	7.2	This study
Semidefined	Brucella broth (28)	Hemin (250)	None	Fildes solution (5%, vol/vol)	NRa	20
Isolation	RPMI 1640 (10.4)	Hemin (15)	HEPES (10)	l-Glutamine, Na pyruvate, nonessential amino acids (1% [wt/vol] each)	7.0	26

^aNR, not reported in reference.

Growth medium.

A previously described medium (20) consisting of Brucella broth, Fildes solution, and hemin was modified to create an efficient growth medium. BBH-H was formulated as described below. First, 28 g of brucella broth (Fisher Scientific, Pittsburgh, Pa.) and 23.8 g of HEPES (free acid) were dissolved in 850 ml of distilled water, and the pH was adjusted to 7.0 with 10 N NaOH. Next, hematin was conjugated to histidine (14) as follows: 100 mg of hematin was dissolved in 100 ml of a 4% histidine solution, the pH was adjusted to 8.0 with 10 N NaOH, and the reaction mixture was stirred overnight at room temperature. One hundred milliliters of the hematin-histidine conjugate was added to 900 ml of Brucella broth base, and the resulting medium was adjusted to pH 7.2 with 10 N NaOH and filter sterilized, yielding BBH-H. Additional supplements were added to broth medium for some experiments as described in the text. Solid medium (BBH-HA) was made by adding filter-sterilized 2× BBH-H medium to an equal volume of molten (45°C) Bacto Agar (Fisher Scientific) and dispensing into petri plates.

Metabolite and amino acid analysis.

The concentrations of common metabolic substrates and end products in uninoculated and spent media were determined as described below. Glucose, glutamate, ammonia, succinate, acetate, lactate, formate, malate, citrate, and ethanol levels were determined with Boehringer-Mannheim metabolite kits (R-Biopharm, Inc., South Marshall, Mich.). Carbon dioxide levels were measured with a CO₂ assay kit. Except for CO₂ and ethanol assays, all assays were performed using filtered culture supernatant according to the manufacturers' instructions. CO₂ and ethanol assays were performed without filtering to avoid loss of volatile products. Oxygen levels were determined using an oxygen electrode (model 733; Diamond General Development Corp., Ann Arbor, Mich.). To determine which amino acids B. henselae depleted from the growth medium, filtered culture supernatants were analyzed. Free amino acid analyses of uninoculated and spent media were performed on a Beckman Instruments model 6300 amino acid analyzer by the Scientific Research Consortium (St. Paul, Minn.) (www.aminoacids.com).

Enzyme assays.

B. henselae cell lysate was assayed for tricarboxylic acid (TCA) cycle and amino acid catabolic enzymes. Cells were grown to a density of 2 × 10⁸ CFU/ml in BBH-H, harvested by centrifugation (5,000 × g, 15 min, 4°C), suspended in 20 mM HEPES-5% sucrose (pH 7.6), and lysed by two passes in a French pressure cell (14,000 lb/in², 4°C). After lysis, cell debris was removed by centrifugation (15,000×g, 15 min, 4°C). Enzymatic assays for citrate synthase (21), aconitase (5), isocitrate dehydrogenase (25), α-ketoglutarate dehydrogenase (7), fumarase (9), glutamate dehydrogenase (6), and histidase (18) were performed as previously described. Succinate dehydrogenase was assayed as previously described (12) with the addition of fresh flavin adenine dinucleotide (0.10 mM, final concentration) to the reaction mixture. Malate dehydrogenase assays (13) were done in 50 mM potassium phosphate buffer, pH 7.3, containing NAD (final concentration, 10 mg/ml). All assays were performed at 37°C using a Beckman Coulter DU 640 spectrophotometer (Beckman Instruments, Palo Alto, Calif.).

Phage isolation.

Growth-dependent phage induction in B. henselae was examined by isolating phage particles and extracting phage DNA from BBH-H culture supernatant at time points throughout the growth curve. Bacteria were removed from the medium by centrifugation (5,000 × g, 15 min, 4°C), and the supernatant was filtered with a 0.22-μm-pore-size filter (Millipore, Bedford, Mass.). Phage particles were harvested from the filtered supernatant (38 ml) by centrifugation (141,000 × g, 4 h, 10°C) in an SW-28 rotor (Beckman Instruments), suspended in phage buffer (5 mM MgCl₂, 1 mM CaCl₂, 0.15 M NaCl in 10 mM Tris-HCl, pH 7.5) overnight at 4°C (15) and examined by electron microscopy as described previously (27). For isolation of phage DNA, 0.5 ml of a phage suspension was treated with 3 μl of Benzonase (Novagen, Madison, Wisc.) and DNase (Roche Applied Science, Indianapolis, Ind.) (1 h, 37°C) to remove any contaminating chromosomal DNA from the suspension. The phage coat proteins were then digested with proteinase K (6.67 mg/ml) in 3.3% sodium dodecyl sulfate-0.167 M EDTA (1 h, 37°C). Phage DNA was extracted with phenol-chloroform-isoamyl alcohol (25:24:1), reextracted with chloroform-isoamyl alcohol, and precipitated with ethanol at −80°C. Purified phage DNA was electrophoresed in 0.5% agarose gels and visualized by ethidium bromide staining.

Isolation from experimentally infected cats.

Cats were experimentally infected by intradermal injection with 6.1 × 10⁸ CFU of B. henselae Houston-1 and monitored for bacteremia as described previously (17). Blood from bacteremic cats was collected in Vacutainer tubes (Becton Dickinson, Franklin Lakes, N.J.) supplemented with either 0.1 ml of erythrocyte lysis solution (4% saponin, 0.96% polyamethodesulfonic acid, and 2.5% polyethylene glycol in 0.9% saline) or 0.1 ml of heparin solution (500 U/ml in 0.9% saline), diluted 1:100 in BBH-H, and incubated as described above. Cultures were examined by microscopy every 24 h for growth, and cells were transferred to plates with BBH-HA when bacterial growth was evident. PCR was used to confirm that colonies from these plates were B. henselae (16). All federal and institutional guidelines were followed in the treatment and handling of animals.

RESULTS

Development of a liquid growth medium for B. henselae culture.

The difficulty in isolating and studying the biology and pathogenesis of B. henselae is due in part to the lack of a reliable growth medium. Two liquid media have been described for the growth of B. henselae (Table 1) (20, 26), but they are seldom used because they do not support consistent bacterial growth or yields. Our attempts to culture B. henselae in either medium gave very inconsistent results. However, the medium described by Schwartzman et al. (20) was a useful starting point for the development of a reliable medium. Initial growth experiments which finally resulted in the development of BBH-H demonstrated that three factors were critical for optimal growth. The form and quantity of heme were the first modifications. Hematin was substituted for hemin and was conjugated to histidine to improve solubility, and the concentration was reduced from 250 to 100 mg/liter (Table 1). When using this form of BBH-H in static culture, the cells grew very well at the medium-atmosphere interface, suggesting that the amount of dissolved oxygen was affecting growth. Therefore, the gas headspace was increased to 2:1 (vol/vol) in proportion to the culture medium and cultures were shaken at 180 rpm during growth. These modifications significantly improved growth rate and yield.

B. henselae growth was unexpectedly sensitive to the pH of the medium, and optimal growth occurred over a very narrow pH range (pH 6.8 to 7.2). The doubling time in BBH-H at pH 7.2 was approximately 3 h (Fig. 1A), which represents a threefold reduction compared to previously reported growth rates (20). Growth rates under acidic conditions began to decrease at pH 6.6. The growth rate of B. henselae was >6 h at pH 7.4, and no growth was observed at pH 7.6 or more (data not shown). Significantly, cell densities reached 5 × 10⁸ to 1 × 10⁹ CFU/ml in BBH-H (Fig. 1A). Growth of cultured B. henselae on BBH-HA was similar to that on blood agar plates, with colonies forming in 5 to 7 days.

FIG. 1.

Growth of B. henselae in BBH-H. (A) Growth curve of viable cell density in BBH-H. Viable cell counts were determined by plating from a culture grown aerobically as described in the text. (B) Microscopic photographs (magnification, ×400) of viability-stained B. henselae cells in BBH-H of exponential-phase growth (top) and death-phase growth (bottom), illustrating the drastic decline in viable cell density.

BBH-H was tested for its ability to support growth of B. henselae strains LSU16 and ATCC 49793 as well. Both strains exhibited growth levels similar to the Houston-1 strain, with approximately the same doubling time. The closely related pathogen B. quintana was cultured in BBH-H under aerobic and microaerobic conditions. Interestingly, B. quintana did not grow in 21% O₂ but did grow in 7% O₂. These data show that BBH-H supports rapid growth of multiple B. henselae strains, and it may also be used for the cultivation of another pathogenic Bartonella species, B. quintana. The ability of the new medium BBH-H to support rapid and reliable growth of multiple Bartonella strains and species represents a significant advance in Bartonella research.

Substrate utilization and metabolite production by B. henselae.

Little is known about the metabolic properties of B. henselae, and conflicting reports have obscured identification of its primary carbon and energy sources. Therefore, we examined culture supernatants for the depletion of carbon sources and correlated these with the appearance of secondary metabolites and end products. Metabolite concentrations in BBH-H were measured before and after B. henselae growth in the medium. Glucose levels were unchanged during growth, indicating that B. henselae does not utilize this carbohydrate under the conditions tested (Table 2). To determine which metabolic pathways B. henselae was utilizing, the concentrations of a number of fermentation products were assayed in conditioned BBH-H. Acetate, ethanol, lactate, and formate are common end products of fermentation pathways. No changes in the levels of acetate, ethanol, and formate were detected; however, the concentration of lactate decreased, suggesting that B. henselae could metabolize lactate (Table 2). The absence of secondary metabolites suggested that carbon sources utilized by B. henselae were completely catabolized to CO₂. To determine if CO₂ was produced, B. henselae cultures were grown in controlled atmospheres and the CO₂ concentrations were assayed before and after growth. Initial levels of CO₂ in BBH-H were undetectable but increased to 8.75 mM as B. henselae reached a postexponential phase of growth (Table 2). Since the production of CO₂ is accompanied by O₂ consumption, the levels of dissolved O₂ in the culture medium were also determined. As B. henselae grew in BBH-H, dissolved oxygen dropped from 26 mM to undetectable levels (<20 ppm) (Table 2).

TABLE 2.

Metabolite concentrations in BBH-H before and after B. henselae growth

Metabolite	Mean concn ± SDa (mM)
	T0b	TFc
Glucose	4.42 ± 0.21	4.45 ± 0.19
Acetate	2.16 ± 0.08	2.08 ± 0.15
Ethanol	0.09 ± 0.01	0.10 ± 0.02
Formate	1.09 ± 0.08	1.12 ± 0.11
Citrate	0.04 ± 0.01	0.03 ± 0.01
Lactate	2.54 ± 0.09	1.96 ± 0.09
Succinate	0.49 ± 0.02	0.26 ± 0.02
Oxygen	26.0 ± 0.1	NDd
Glutamate	2.75 ± 0.19	1.27 ± 0.11
Carbon dioxide	ND	8.74 ± 0.64
Malate	0.02 ± 0.00	0.06 ± 0.00
Ammonia	4.70 ± 0.46	11.6 ± 0.35

^aAll values are from triplicate assays on two independent cultures. Values were determined by enzymatic assay except for oxygen levels, which were measured with an oxygen electrode.

^bT₀ values are from uninoculated medium.

^cT_F values are from conditioned medium from death-phase cultures.

^dND, none detected.

The absence of glucose catabolism indicated that B. henselae was utilizing alternative carbon sources. A very closely related bacterium, B. quintana, has been reported to utilize succinate and glutamate (10, 23), leading us to speculate that B. henselae could utilize these compounds as well. As expected, the concentrations of succinate and glutamate in BBH-H decreased significantly during the growth of B. henselae (Table 2). The catabolism of amino acids usually produces NH₃ when the amine group is removed. To determine indirectly if glutamate was catabolized, we measured the concentration of NH₃ in the culture supernatants. Ammonia was produced at a level greater than the amount of glutamate depleted from the medium (Table 2), suggesting that additional amino acids were being catabolized. To test this hypothesis, we measured the concentrations of free amino acids present in BBH-H before and after growth. The results confirmed that, in addition to consuming glutamate, B. henselae consumed histidine, asparagine, glycine, and serine from the growth medium (Table 3). Tryptophan and cysteine were also consumed but at significantly lower levels than the other amino acids (Table 3). Taken together, these data suggest that B. henselae was catabolizing amino acids rather than glucose as a carbon source.

TABLE 3.

Depletion of free amino acids from BBH-H by B. henselae

Amino acid	Mean concn ± SDa (mM)		Change
	T0b	TFc
Asn	0.67 ± 0.02	0.01 ± 0.00	0.66
Cys	0.18 ± 0.00	0.07 ± 0.00	0.12
Glu	4.19 ± 0.05	1.72 ± 0.01	2.47
Gly	3.75 ± 0.03	3.43 ± 0.07	0.32
His	33.0 ± 0.30	30.7 ± 0.67	2.30
Ser	0.89 ± 0.01	0.45 ± 0.02	0.44
Trp	0.46 ± 0.00	0.40 ± 0.00	0.05

^aValues were determined by mass spectrometry and are from two independent experiments assayed in triplicate.

^bT₀ values are from uninoculated medium.

^cT_F values are from conditioned medium from death-phase cultures.

Enzymatic activities in B. henselae cell lysate.

Based upon amino acid depletion and NH₃ generation, we hypothesized that B. henselae catabolized amino acids. To test this hypothesis, B. henselae lysate was assayed for amino acid catabolism enzymes. Since glutamate was the predominant amino acid depleted from BBH-H, the enzyme responsible for the degradation of this amino acid was assayed. NADH-dependent glutamate dehydrogenase catalyzes the reaction that deaminates glutamate to α-ketoglutarate, a TCA cycle intermediate. Activity was detected for this enzyme in B. henselae lysate, but not for NADPH-dependent glutamate dehydrogenase, a biosynthetic enzyme that produces glutamate from α-ketoglutarate and NH₃ (Table 4). The presence of a glutamate-catabolizing enzyme suggested that histidine, the other amino acid depleted from the medium in large quantities, was also entering a catabolic pathway. To test this, B. henselae lysate was assayed for histidase activity, the first step in the degradation of histidine to glutamate; histidase activity was detected in B. henselae cell lysate (Table 4). Since histidase deaminates histidine, the high levels of this enzyme and NADH-dependent glutamate dehydrogenase in B. henselae could account for the elevated NH₃ measured during growth. The presence of these key enzymes indicated that B. henselae was able to catabolize amino acids to TCA cycle intermediates.

TABLE 4.

Metabolic enzyme activities present in B. henselae lysate

TCA enzyme or amino acid degradation	Activitya
TCA enzyme
Citrate synthase	+
Aconitase	+
Isocitrate dehydrogenase	+
α-Ketoglutarate dehydrogenase	+
Succinate dehydrogenase	+
Fumarase	+
Malate dehydrogenase	+
Amino acid degradation
Glutamate dehydrogenase
NADH dependent	+
NADPH dependent	−
Histidase	+

^aSymbols: +, presence of enzymatic activity; −, no activity detected.

Utilization of the TCA cycle produces CO₂ and consumes O₂, and during B. henselae growth, we observed O₂ consumption and CO₂ production. This led us to hypothesize that B. henselae harbors the enzymes for a complete TCA cycle. Enzymatic activities for all of the TCA cycle enzymes or enzyme complexes were detected in cell lysate, indicating that B. henselae does possess the enzymatic components of a full TCA cycle (Table 4). To generate energy from the reducing potential produced by the TCA cycle, organisms typically utilize the electron transport chain (ETC). Under aerobic conditions, oxygen is consumed as the terminal electron acceptor of the ETC. During growth, B. henselae consumed all of the oxygen present in the medium (Table 2). However, oxidative decarboxylation reactions of the TCA cycle cannot account for all of the O₂ depleted, based on the amount of CO₂ produced (Table 2). These data, coupled with the presence of a complete TCA cycle, suggest that B. henselae uses the ETC to generate ATP from the NADH and flavin adenine dinucleotide-H₂ produced by the TCA cycle. Other bacteria are able to utilize alternate electron acceptors, such as NO₃ and NO₂, when grown anaerobically. A number of terminal electron acceptors (NO₃, fumarate, dimethyl sulfoxide, NO₂, NO₂-formate, or Na₂S₂O₃) were used at concentrations of 20 mM to supplement anaerobic BBH-H; however, none supported B. henselae growth (data not shown). These data suggest that during in vitro growth, B. henselae utilizes the TCA cycle to generate reducing potential, which then enters the ETC to generate energy with oxygen as the terminal electron acceptor and that alternate electron acceptors are not utilized by B. henselae.

Postexponential growth phase of B. henselae in BBH-H.

When B. henselae cultures grown in BBH-H reached the postexponential phase, there was a rapid decline in the number of viable cells in the medium as determined by plating on BBH-HA (Fig. 1A). This precipitous drop in CFU over a 24-h period was defined as the “death phase.” These results were confirmed using the BacLight live/dead stain, a cell viability assay based upon membrane integrity. All of the cells visible at peak density (5 × 10⁸ CFU/ml) fluoresced green, indicating intact cell membranes, whereas 24 h later, >99% of the bacterial cells fluoresced red, indicating that they were not viable (Fig. 1B). The death phase of B. henselae was examined in more detail to determine the potential cause of the rapid decrease in cell viability.

Substrate depletion or toxic metabolite (i.e., NH₃) production were possible explanations for postexponential death. To ascertain if this was the case, spent BBH-H from a death-phase culture was filtered and reinoculated with B. henselae. This “spent” medium supported growth, with doubling times and cell densities very similar to those observed in fresh medium, suggesting that loss of viability was not due to the depletion of an essential nutrient or the production of a toxic metabolite (data not shown). B. henselae does not grow in BBH-H under anaerobic conditions, but does deplete O₂ from the medium as it grows. To determine if the bacteria were dying because of O₂ depletion, cells were grown under atmospheric (21%) and microaerobic (7%) oxygen levels. We hypothesized that if O₂ deprivation was the cause of the death phase, then the decline in viable cells would occur at a lower cell density in microaerobic cultures due to the O₂ being depleted more rapidly. However, both conditions produced the death phase phenotype at similar cell densities, suggesting that O₂ deprivation is not the cause (data not shown). Taken together, these data suggest that substrate depletion, accumulation of toxic metabolites, and production of anaerobic conditions were not responsible for the death phase.

Phage induction in B. henselae.

Since B. henselae and other Bartonella species harbor defective phages (1, 3, 22), phage induction was examined as a possible cause of the death phase. When phage are induced, bacterial membrane integrity becomes compromised, resulting in host cell death. To investigate whether this was the cause of the B. henselae death phase, culture supernatants taken from bacterial cultures were examined for the presence of intact phage by electron microscopy (Fig. 2A). Phage particles with multiple head sizes were detected, although it was not clear if the variation in particle size was an artifact of the isolation protocol, as ultracentrifugation can alter the appearance of phage particles. Consistent with the observations of Anderson and Barbian (1, 3), no phage tails were evident in the phage preparations.

FIG. 2.

Growth-dependent phage induction in B. henselae grown in BBH-H. (A) Electron micrograph (magnification, ×200,000) of phage from death-phase culture supernatant revealing different particle sizes present in the medium. The numbers present on the image represent the diameter in nanometers. (B) Extracted phage DNA separated in 0.5% agarose. Lane 1, exponential-phase culture; lane 2, early-death-phase culture. The arrow indicates a 14-kb DNA band isolated from death-phase culture supernatant. Standards (in kilobase pairs) are indicated to the left.

To verify that particles in the culture supernatants were phage, samples were treated with nucleases to remove any free DNA, as phage DNA packaged in capsid proteins is DNase resistant. The putative phage DNA was extracted and analyzed by agarose gel electrophoresis. A 14-kb DNA segment was recovered from death-phase B. henselae culture supernatants that was not evident in exponential-phase culture supernatants (Fig. 2B). The size of the DNA fragment is consistent with the size of the Bartonella phage DNA described by other researchers (1, 3). The appearance of this DNase-protected, extrachromosomal DNA correlates with the onset of death phase, suggesting that the rapid decline in B. henselae cell density is the result of phage induction and host lysis.

Isolation from experimentally infected cats.

Current methods for isolating B. henselae utilize blood agar plates; however, colonies typically take 2 to 6 weeks to form when the bacteria are first isolated from the mammalian host. To determine if BBH-H would be effective for the isolation of B. henselae from experimentally infected cats, blood samples from bacteremic cats (3 × 10² to 5 × 10⁴ CFU/ml blood) and from noninfected cats were diluted 1:100 in 10 ml BBH-H and incubated at 37°C with aeration. No bacterial growth was detected in BBH-H inoculated with uninfected cat blood. Within 4 days, cultures inoculated from samples with 5 × 10² CFU reached maximum cell density (5 × 10⁸ CFU/ml), while those inoculated with <50 CFU required an additional 24 h to reach the same cell density. Interestingly, treatment of the blood sample during collection had a dramatic effect on isolation time. BHH-H inoculated with samples collected in Vacutainer-saponin resulted in the shortest isolation times (above) while cultures inoculated from Vacutainer-heparin blood took 11 days to reach similar cell densities. These data suggested that lysing the red blood cells during sample collection improved isolation efficiency. When these same samples were used to inoculate BBH-HA, colonies formed within 5 days. PCR screening confirmed that the bacteria isolated from these blood samples were B. henselae (data not shown). The reduction in incubation time from 2 to 6 weeks using blood agar plates to 5 days in BBH-H medium or BBH-HA represented a significant improvement and suggested that BBH-H could be used for more rapid and reliable isolation of B. henselae.

DISCUSSION

Bartonella research has been hampered by the lack of a reliable growth medium. By modifying a previously published formula, we have developed a reliable B. henselae liquid growth medium. Several factors ultimately were key for more rapid, reproducible growth. First, was the form of heme. When hematin was substituted for hemin and conjugated to histidine to improve solubility, growth yields and rates improved. Second, unreacted histidine in the conjugation reaction mixture was utilized by B. henselae as a carbon source. Third, aerating the cultures provided sufficient dissolved oxygen to support maximum growth. B. henselae would not grow under anaerobic conditions and consumed all available dissolved O₂ in the growth medium. And finally, although seemingly trivial, the addition of 100 mM HEPES to maintain the pH within tolerable limits was critical. B. henselae growth was sensitive to the pH of the medium, with optimal growth occurring over a very narrow pH range (pH 6.6 to 7.2). The growth rate of B. henselae in BBH-H was 3 h, and cell densities reached 5 × 10⁸ to 1 × 10⁹ CFU/ml.

The development of BBH-H has allowed us to investigate the metabolism of B. henselae, which appeared to be unusual when compared to other blood-borne bacterial pathogens. Almost all bacterial pathogens that thrive and grow in blood preferentially utilize glucose as a carbon and energy source. Atypically, B. henselae did not use glucose under the growth conditions tested. These results were contradictory to a recently published genomic study which concluded that B. henselae had the metabolic potential to catabolize glucose (4). However, the same study noted that B. henselae lacked the gene encoding phosphofructokinase, a key regulatory enzyme of glycolysis. Since glucose was not depleted from the culture medium during growth, B. henselae must be deriving carbon and energy from alternative sources. The accumulation of NH₃ and the coordinate depletion of five amino acids (Asn, Glu, Gly, His, and Ser) from the culture medium during growth, suggested a possible mechanism for B. henselae metabolism. Amino acids can be used for biosynthesis or they can be deaminated and used as carbon and/or energy sources resulting in the production of NH₃. Therefore, biochemical analysis of culture supernatants before and after growth suggested that B. henselae catabolized amino acids to provide carbon and energy.

One way to utilize amino acids efficiently is to convert them to TCA cycle intermediates. In the obligate intracellular pathogen Rickettsia prowazekii, glutamate is catabolized to TCA intermediates while glucose is not utilized (2). If amino acids were catabolized by B. henselae via the TCA cycle, the pathogen must (i) possess the enzymes for a complete TCA cycle, (ii) demonstrate TCA cycle activity (i.e., CO₂ production from the oxidative decarboxylation reactions and O₂ consumption), (iii) contain the enzymes necessary to convert amino acids into TCA cycle intermediates, and (iv) have a source of acetyl coenzyme A and either a C₄ or C₅ intermediate to maximize the reducing potential and ATP production from the TCA cycle. Our results clearly demonstrated that B. henselae had enzymatic activities for all the TCA cycle enzymes or enzyme complexes, produced CO₂, and consumed O₂ during growth (Table 2). Additionally, we detected enzymes which could convert histidine or glutamine to glutamate, and glutamate to α-ketoglutarate, a C₅ intermediate in the TCA cycle. B. henselae also removed lactate, serine, and glycine from the medium during growth. These compounds are readily converted to pyruvate, a potential source of acetyl coenzyme A. Taken together, these data strongly suggest that B. henselae derives carbon and energy from the catabolism of amino acids rather than glucose. Based upon the pattern of substrate utilization in BBH-H, we have developed a model for central metabolism in B. henselae (Fig. 3). While other pathogenic bacteria can derive energy using the same metabolic pathways, B. henselae is unusual in its preference for this catabolic strategy.

FIG. 3.

Proposed model of B. henselae metabolism. Substrates utilized are shown in green, and secondary metabolites produced are shown in blue. Enzyme complexes present in B. henselae lysate are shown in red. Arrows without labels are reactions inferred from the substrates utilized and pathways present.

A curious aspect of B. henselae growth was the rapid decline in cell density after the culture reached stationary phase. Several Bartonella species, including B. henselae, harbor defective phages. Conditions for induction have been investigated by several groups; however, mitomycin C and UV radiation treatments did not lead to phage induction (1). We have found that phage induction was dependent on B. henselae's growth phase and correlated with the death phase of the bacterium. At this point, we cannot determine if phage induction was the cause or consequence of the death phase.

We have developed a liquid medium for rapid growth of B. henselae and used it to identify the unusual metabolic strategy of this pathogen. Importantly, the medium was also used to isolate B. henselae from the blood of infected cats. The growth rates of B. henselae in BBH-H or BBH-HA inoculated with infected cat blood samples represented a significant decrease in isolation time over previously described media, including blood agar. However, it should be noted that B. henselae strains LSU-16 and Houston-1 can be successfully isolated in 5 to 7 days from experimentally infected cats using chocolate agar. Unfortunately, chocolate agar does not permit the reproducible isolation of Bartonella species from naturally infected cats or humans (8). Studies are currently under way to determine if BBH-H and BBH-HA will support primary isolation of B. henselae from the blood of naturally infected cats. Overall, the development of BBH-H medium represents a major step forward for the study of B. henselae pathogenesis and may ultimately prove efficacious for the isolation of B. henselae and B. quintana.