Na⁺/H⁺ Antiport Is Essential for Yersinia pestis Virulence

Abstract

Na⁺/H⁺ antiporters are ubiquitous membrane proteins that play a central role in the ion homeostasis of cells. In this study, we examined the possible role of Na⁺/H⁺ antiport in Yersinia pestis virulence and found that Y. pestis strains lacking the major Na⁺/H⁺ antiporters, NhaA and NhaB, are completely attenuated in an in vivo model of plague. The Y. pestis derivative strain lacking the nhaA and nhaB genes showed markedly decreased survival in blood and blood serum ex vivo. Complementation of either nhaA or nhaB in trans restored the survival of the Y. pestis nhaA nhaB double deletion mutant in blood. The nhaA nhaB double deletion mutant also showed inhibited growth in an artificial serum medium, Opti-MEM, and a rich LB-based medium with Na⁺ levels and pH values similar to those for blood. Taken together, these data strongly suggest that intact Na⁺/H⁺ antiport is indispensable for the survival of Y. pestis in the bloodstreams of infected animals and thus might be regarded as a promising noncanonical drug target for infections caused by Y. pestis and possibly for those caused by other blood-borne bacterial pathogens.

INTRODUCTION

Yersinia pestis, a Gram-negative bacterium, is the causative agent of plague. Because of its high infectivity and lethality, Y. pestis is considered a potential bioterrorism weapon. Y. pestis is typically introduced into the human body by the bite of an infected flea and is carried to the regional lymph nodes by macrophages. Following replication in the lymph nodes, the bacteria spread throughout the body by the lymph and bloodstream and cause bubonic plague. During the later stages of the disease, the patient usually develops septicemia, disseminated intravascular coagulation, shock, and peripheral gangrene (reviewed in reference 1).

A large number of diverse bacterial species, including pathogens, use sodium as a chemiosmotic coupling ion in addition to, or even instead of, protons (2–5). The sodium motive force (SMF) may be generated either by primary sodium pumps, such as Na⁺-translocating NADH:ubiquinone oxidoreductases (NQR), or by secondary sodium pumps, such as Na⁺/H⁺ antiporters, that are energized by the proton motive force (PMF). Although many pathogenic organisms utilize different elements of the sodium cycle extensively in their physiology, the role of sodium bioenergetics in bacterial virulence in general is not well understood. Since the host presents a potentially unique, and not necessarily energetically favorable, environment for the growth of pathogenic microbes, the capacity to use sodium as a coupling ion can be expected to be of key importance during pathogenesis in a variety of bacterial species (discussed in detail in reference 5).

The reported genome sequence of Y. pestis predicts the presence of the primary sodium pump NQR, in addition to four secondary sodium pumps, the Na⁺/H⁺ antiporters NhaA, NhaB, NhaC, and NhaP, which are expected to contribute to SMF generation across the inner membrane of Y. pestis (Fig. 1) (4). However, recent detailed biochemical characterization of NhaP-type antiporters NhaP-1 and NhaP-2 from Vibrio cholerae revealed that these antiporters mediate K⁺/H⁺ exchange rather than Na⁺/H⁺ exchange under physiological conditions (6, 7). By analogy, we predict that the Y. pestis NhaP protein has a similar cation preference (Fig. 1). Moreover, NhaC-type antiporters are poorly characterized; Y. pestis NhaC (Yp-NhaC) remains a hypothetical protein, and it is not clear at this moment whether it is chemiosmotically active at all. In contrast, one could expect the NhaA-NhaB pair to be the major secondary Na⁺-extruding system of Y. pestis, as has been found in many bacterial species possessing the nhaA and nhaB genes in their genomes (8–10). A comparison of the published genomes of various Y. pestis strains (obtained from the databases of the J. Craig Venter Institute [http://www.jcvi.org/] and the DOE's Joint Genome Institute [http://img.jgi.doe.gov]) revealed the presence of intact genes for NhaA, NhaC, and NhaP antiporters in all of the strains. Interestingly, the nhaB gene of the Y. pestis strain KIM contains a frameshift, whereas this gene is intact in all other Y. pestis strains; thus, the designations of all Y. pestis KIM strains will be followed by “nhaB” in parentheses to indicate the presence of this mutation. Furthermore, genes encoding at least 13 putative Na⁺-dependent symporters with various substrate specificities and two gene orthologs of the Na⁺-dependent multidrug efflux pump NorM are predicted to be the consumers of the SMF in the Y. pestis genome (Fig. 1) (4).

Fig 1.

Sodium cycle in energetics in Y. pestis. Putative generators and consumers of the sodium motive force (SMF) in Y. pestis strain CO92 are shown. The primary Na⁺ pump, NQR (YPO3235 to -3240), generates SMF by direct sodium extrusion. The sodium extrusion activity is coupled to NADH oxidation and ubiquinone (Q) reduction. The Na⁺/H⁺ antiporters NhaA (YPO0470), NhaB (YPO2142), and NhaC (YPO0624), and the predicted K⁺(Na⁺)/H⁺ antiporter NhaP (YPO3630), can convert proton motive force into SMF and vice versa. DAACS, dicarboxylate/amino acid:cation (Na⁺ or H⁺) symporter (YPO05784 and YPO1716); DASS, divalent anion:Na⁺ symporter (YPO0759 and YPO2561); ESS, glutamate:Na⁺ symporter (YPO0035); GPH, glycoside-pentoside-hexuronide:cation symporter (YPO0995 and YPO1226); SSS, solute:sodium symporter (YPO0251, YPO1853, YPO3002, and YPO3657); MATE, multidrug and toxic compound extrusion (YPO1571 and YPO2392).

Importantly, sodium pumps are not only essential for maintaining the SMF as an energy source for Na⁺ symporters; they also protect bacterial cells from Na⁺ toxicity. Irrespective of the presence of functional NQR, Na⁺/H⁺ antiporters have been shown to play a leading role in the Na⁺ resistance of bacteria (9, 10). Among the Na⁺/H⁺ antiporters examined, antiporters of the NhaA type are usually extremely kinetically competent, fast, and pH dependent, displaying maximal activity at alkaline pHs, and are indispensable for survival in sodium-rich environments, especially alkaline environments (8–10). Like that of Escherichia coli NhaA, the activity of Y. pestis NhaA (Yp-NhaA) is regulated by pH and selectively exchanges Na⁺ and Li⁺ for H⁺ (11; J. Winogrodzki and P. Dibrov, unpublished data). NhaB-type antiporters are much less active, “housekeeping” cation exchangers that normally play an auxiliary role. Like NhaA, they exchange sodium ions for protons in an electrogenic manner (3 H⁺ per 2 Na⁺), and, unlike NhaA, they are expressed constitutively (9, 10, 12). Although the activity of E. coli NhaB does not show pH sensitivity (13), Vibrio alginolyticus NhaB shows pH-dependent activity, reaching its maximum at alkaline pHs (14). To date, no biochemical analyses of the Yp-NhaB protein have been reported.

Blood-borne bacterial pathogens can survive in blood, and thus, they can cause disease. To survive in blood, bacterial pathogens must be able to resist the multiple innate defense systems present in blood. In Y. pestis, a type III secretion system (T3SS) and secreted Yop effector proteins, encoded on a Yersinia virulence plasmid, are known to prevent phagocytosis and other protective innate immune responses. An outer membrane protein, Ail, is also known to protect Y. pestis cells against complement-mediated lysis (15, 16). In addition to the known essential factors for survival in blood, Y. pestis has to be able to tolerate the salt concentrations in blood in order to establish an infection. Thus, sodium pumps may play a role in its pathogenesis. To date, however, there is no report investigating the roles of the various Y. pestis sodium pumps in the context of its pathogenic potential.

In this study, we investigated the roles of the primary sodium pump, NQR, and the two main secondary sodium pumps, the Na⁺/H⁺ antiporters NhaA and NhaB, in Y. pestis virulence. It was found that the loss of both NhaA and NhaB, but not NQR, abolishes Y. pestis virulence, presumably because these antiporters are essential for protecting Y. pestis from Na⁺ toxicity in blood.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. All bacterial strains were kept at −80°C in 20% glycerol stocks. Y. pestis strains were grown in LB media at 30°C unless otherwise noted. Media were supplemented with antibiotics as appropriate, as follows: streptomycin (Str), 100 μg/ml; ampicillin (Amp), 100 μg/ml; kanamycin (Km), 50 μg/ml.

Table 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source or reference(s)
Strains
Y. pestis
KIM5-3001	NhaB− pCD1+ pPCP1+ pMT1+ Pgm− Strr	11, 24
KIM5-3001 Δnqr	KIM5-3001 ΔnqrABCDEF	This study
KIM5-3001 ΔnhaA	KIM5-3001 ΔnhaA	This study
KIM5	NhaB− pCD1+ pPCP1+ pMT1+ Pgm−	G. Plano lab collection
KIM5 ΔnhaA	KIM5 ΔnhaA	This study
CO92	NhaB+ pCD1+ pPCP1 pMT1+ Pgm+	36
CO92 ΔnhaA	CO92 ΔnhaA; Kmr	This study
CO92 ΔnhaA ΔnhaB	CO92 ΔnhaA ΔnhaB; Kmr Ampr	This study
E. coli
Top10	Cloning host strain	Invitrogen
EP432	melBLid ΔnhaA1::kan ΔnhaB1::cat ΔlacZY thr1	12
Plasmids
pCR2.1-TOPO	Cloning vector	Invitrogen
pKD4	Kmr; template for PCR amplification for lambda Red recombinase-mediated recombination	17
pCP20	Ampr Cmr; temperature-sensitive replicon; FLP recombinase	17
pWKS130	pSC101-based low-copy-number vector; Kmr lacZ′	37
pWKS130-nhaA	nhaA ORF and promoter region cloned into pWKS130	This study
pBADTOPO	Cloning vector with pBAD promoter	Invitrogen
pBAD-nhaA	nhaA ORF cloned into pBADTOPO under the control of the pBAD promoter	This study
pET6×HN-C	Expression plasmid with T7 lac promoter	Clontech Laboratories
pET-nhaB	nhaB ORF cloned into pET6×HN-C under the control of the T7 lac promoter	This study

Construction of Y. pestis strains carrying deletions in nqrABCDEF, nhaA, and nhaB.

To construct the nqrABCDEF and nhaA deletion mutant strains of Y. pestis KIM5-3001 (nhaB), PCR of a kanamycin resistance (Km^r) cassette flanked by the FLP recombination target (FRT) sites and homologous regions of the target gene was performed essentially as described previously (17) by using primer pairs listed in Table 2 (ΔnhaA-KIM-P1 and ΔnhaA-KIM-P2, Δnqr-P1 and Δnqr-P2). The ΔnhaA mutant was also constructed in Y. pestis KIM5, using different sets of primers (ΔnhaA-KIM-P1B and ΔnhaA-KIM-P2B), also listed in Table 2. Y. pestis strains containing plasmid pKD46 were grown first in heart infusion broth (HIB) at 28°C to an optical density at 620 nm (OD₆₂₀) of 0.5 and then for 2 h with 0.2% l-arabinose. Electrocompetent cells were prepared as described previously (18) and were electroporated with 2 μl of the purified PCR product. Gene replacements were confirmed by PCR analysis. The Km^r cassettes were removed using the temperature-sensitive plasmid pCP20, which encodes the FLP recombinase. After FLP/FRT-mediated site-specific recombination, Km-sensitive colonies containing a single FRT site were identified by replicate plating and PCR analysis. Temperature-sensitive plasmids pKD46 and pCP20 were cured from the nqrABCDEF and nhaA deletion mutants by overnight growth at 38°C. The ΔnhaA and ΔnhaA ΔnhaB mutant strains of Y. pestis CO92 were constructed by similar lambda Red recombinase mutagenesis strategies (19), in which 974 bp of nhaA was replaced with a Km^r cassette and 1,453 bp of nhaB was replaced with an Amp^r cassette. The primers used for these constructs (ΔnhaA-CO-P1 and ΔnhaA-CO-P2, ΔnhaB-CO-P1 and ΔnhaB-CO-P2) are also listed in Table 2.

Table 2.

Primers used in this study

Primer	Sequence (5′ to 3′)a
ΔnhaA-KIM-P1	GATATACCAGTCTCGATAAAAATAGCTTCGCTGGATATCTGTGTAGGCTGGAGCTGCTTC
ΔnhaA-KIM-P2	CAGTAAAGCAATAGCAGCCGCAGCAATACCCAAAGATTGCATATGAATATCCTCCTTAGT
ΔnhaA-KIM-P1B	GGGCTGGAAGTTAAACGTGAGTTGATGGAAGGCTCTTTGTGTGTAGGCTGGAGCTGCTTC
ΔnhaA-KIM-P2B	CCTAGCAATATCCCTAGTTTGGAATAGGTCGTCAGAGCGCATATGAATATCCTCCTTAGT
Δnqr-P1	AAAGGGCTAGATCTTCCCATTGCTGGAGCCCCCGTTCAGTGTGTAGGCTGGAGCTGCTTC
Δnqr-P2	ATAAAATTCACAATCCTCTGGAGCCGGGTGATCTTTCAGCATATGAATATCCTCCTTAGT
ΔnhaA-CO-P1	GCTGCCGCTATCGCCTTGTTGATGGCGAACAGCGCCCTACAAGGGGTTTATCGTTGTCGGGAAGATGCGTG
ΔnhaA-CO-P2	CAGTAAGCTGTAACCCACAACAGCCGCTGTGGTAGAGCCTAGCAATATCCCCGGCTCGTATGTTGTGTGG
ΔnhaB-CO-P1	GGACATTACCAACAGACAAGCTGTACTCAAGAATTTTCTTGGTAACTCTCCGATCTTTTCTACGGGGTCTGACG
ΔnhaB-CO-P2	CCGTTTGTGATGTGTGGCAAACTGATCCATCCAGCCTGAGTCAGCCACTCCTTTTCGGGGAAATGTGCG
nhaA-pWKS-P1	GGGGGGAGCTCATGGCGTGGTTAATTAGCAGCTAGCC
nhaA-pWKS-P2	GGGGGGGATCCTTACCTTACGTTAACAGCCTTCCGCCG
nhaB-pBAD-P1	GAGGAATAATAAATGGACATTACCAACAGACA
nhaB-pBAD-P2	TCAGTGTGGGATGGCAACCCCG
nhaB-pBAD-SDM-P1	GTCTCGTTACAACAATTTGAATTGTTGGCGGTCGCC
nhaB-pBAD-SDM-P2	GGCGACCGCCAACAATTCAAATTGTTGTAACGAGAC
nhaB-pET-P1	GACATTACCAACAGACAAGCTGTA
nhaB-pET-P2	GAAGAGAATTCGTGTGGGATGGCAACCCCGTTTGT

^aRestriction sites are underlined.

Cloning of the nhaA and nhaB genes.

For the complementation analyses, we cloned the nhaA gene as follows. The DNA fragment, which contained the nhaA promoter region and open reading frame (ORF), was amplified by PCR using chromosomal DNA of Y. pestis KIM5-3001 (nhaB) as a template. The primers used for this construct, nhaA-pWKS-P1 and nhaA-pWKS-P2, included SacI and BamHI restriction sites, respectively, and are listed in Table 2. The PCR products obtained were digested with SacI and BamHI, gel purified, and then ligated into the SacI and BamHI sites of vector pWKS130. pWKS130 has a pSC101 origin of replication and is stably maintained along with the native plasmids. The nhaB expression plasmid was constructed as follows. The nhaB ORF was amplified by using primers nhaB-pBAD-P1 and nhaB-pBAD-P2 (Table 2), with chromosomal DNA of Y. pestis KIM5-3001 (nhaB) as a template, and was cloned into pBADTOPO. Because nhaB from KIM5-3001 (nhaB) contained a frameshift mutation, we performed site-directed mutagenesis to restore the function of the gene by using primers nhaB-pBAD-SDM-P1 and nhaB-pBAD-SDM-P2 (Table 2), a high-fidelity DNA polymerase, KOD -Plus-, version 2 (Toyobo), and DpnI (Invitrogen). The site-directed mutagenesis was confirmed by sequencing. The restored nhaB gene was amplified by PCR using primers nhaB-pET-P1 and nhaB-pET-P2 (Table 2). Primer nhaB-pET-P2 included an EcoRI restriction site. The PCR products obtained were digested with EcoRI, gel purified, and ligated into the StuI and EcoRI sites of the pET6×HN-C plasmid, which is located under the control of the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible T7 lac promoter.

Mouse models of septicemic and bubonic plague.

Y. pestis strains were inoculated from frozen glycerol stock to 5 ml brain heart infusion (BHI) and were grown overnight at 28°C without aeration. Overnight cultures were diluted 1:100 into 10 ml Luria-Bertani (LB) medium and were incubated again overnight (16 to 18 h) without aeration, at 37°C. The overnight LB cultures had a density of about 1 × 10⁸ CFU/ml as determined by a Petroff-Hausser chamber count and were diluted in phosphate-buffered saline (PBS). The LB cultures of the mutant and parent strains achieved the same density. Eight- to 12-week-old female Rocky Mountain Laboratory (RML) Swiss Webster mice were used. Groups of 5 to 10 mice were injected intravenously in the tail vein (septicemic-plague model) with 100 μl of PBS containing 1,000 bacteria (KIM5 strains) or intradermally in the right lower lumbar region (bubonic-plague model) with 25 μl containing 10 or 100 bacteria (CO92 strains). The number of bacteria injected was confirmed by a CFU count of the inoculum preparations. Infected animals were observed three times daily and were euthanized upon signs of terminal plague (evidence of lethargy, hunched posture, and reluctance to respond to external stimuli). Heart blood, spleens, or both were collected immediately after euthanasia. Dilutions of blood and triturated spleen were plated onto blood agar containing 1 μg/ml Irgasan to verify plague and to determine the bacterial loads in these tissues. These studies were conducted at biosafety level 3 and were approved by the Rocky Mountain Laboratory Animal Care and Use Committee in accordance with NIH NIAID guidelines.

Isolation of membrane vesicles.

For the functional expression of Yp-NhaA, EP432, a Na⁺/H⁺ antiporter-deficient strain of E. coli kindly provided by E. Padan (Hebrew University of Jerusalem, Jerusalem, Israel), was used. EP432 is a K-12 derivative (melBLid ΔnhaA1::kan ΔnhaB1::cat ΔlacZY thr1) (12). EP432/pBAD-YpNhaA transformants were grown in LBK medium (modified LB medium in which NaCl was replaced with KCl) supplemented with 100 μg/ml ampicillin, 30 μg/ml kanamycin, 34 μg/ml chloramphenicol (Cm), 25 mM LiCl, and 0.01% arabinose. Cells were harvested at an OD₆₀₀ of 1.5 to 1.8 and were immediately used for the isolation of inside-out membrane vesicles as described previously (20). Briefly, overnight cultures of EP432 transformants were grown in LBK medium containing the antibiotics listed above. These cultures were then used to inoculate the growth medium at a concentration of 1:100. After harvesting, the cells were washed three times in a buffer containing 200 mM sorbitol, 10% (wt/vol) glycerol, and 20 mM Tris-H₂SO₄ (pH 7.5). After the last wash, the bacterial pellet was resuspended in the same buffer containing 1 mM 1,4-dithiothreitol (DTT), 1 μg/ml pepstatin-A, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and approximately 5 mg/liter DNase. The bacterial suspension was then passed twice through a French press (Aminco), and the unbroken cells were pelleted. The membrane fraction was collected by ultracentrifugation (Optima LE-80K ultracentrifuge), resuspended, stored in the same buffer containing all the additions except DNase, and assayed for cation/proton antiport activity as described previously (20–23).

Measurement of transmembrane ΔpH.

For measurements of the transmembrane pH gradient (ΔpH), aliquots of vesicles (200 μg of protein) were added to 2 ml of a buffer containing 200 mM sorbitol, 25 mM KCl, 5 mM MgCl₂, 10% (wt/vol) glycerol, 15 μM acridine orange, and 50 mM bis-tris propane (BTP)-HCl adjusted to pH 8.0. The Na⁺(Li⁺)/H⁺ antiport activity was then registered using the acridine orange fluorescence-quenching/dequenching assay. Respiration-dependent generation of ΔpH was initiated by the addition of 10 mM Tris-d-lactate, and the resulting quenching of acridine orange fluorescence was monitored in a Shimadzu RF-1501 spectrofluorophotometer (excitation at 492 nm; emission at 528 nm). Antiport activity was estimated based on the ability to dissipate the established ΔpH in response to the addition of NaCl or LiCl at the indicated concentrations (see Fig. 7 and 8). A 10 mM concentration of each was used in the determination of the pH profile of activity, and 0.1 mM to 20 mM was used in the determination of apparent K_m. The antiport activities are expressed as percentages of restoration of lactate-induced fluorescence quenching. All assays were performed at room temperature.

Fig 7.

Determination of the concentration of Na⁺ or Li⁺ required for a half-maximal response of Yp-NhaA ([NaCl⁺]_1/2 and [LiCl⁺]_1/2) in inside-out membrane vesicles derived from EP432/pBAD-YpNhaA cells. Assays were performed in a buffer adjusted to pH 8.0 using 50 mM BTP-H₂SO₄ with varying concentrations of NaCl and LiCl as described in Materials and Methods. [NaCl⁺]_1/2 and [LiCl⁺]_1/2 of 3.5 mM and 0.7 mM, respectively, were determined. The average results for two separate isolations are plotted. Error bars indicate standard deviations. DQ, dequenching.

Fig 8.

Effect of 2-AP on the antiport activity of Yp-NhaA in inside-out membrane vesicles isolated from EP432/pBAD-YpNhaA. Inside-out membrane vesicles were isolated from EP432/pBAD-YpNhaA cells as described in Materials and Methods. (A) Vesicles were assayed for ΔpH in a reaction mixture containing 200 μg of protein, 15 μM acridine orange, 200 mM sorbitol, 5 mM MgCl₂, 50 mM BTP-HCl adjusted to pH 8.0, and 2-AP at the indicated concentrations. At the beginning of the experiment, vesicles were added to the reaction mixture (first arrows pointing down), followed by 10 mM Tris-d-lactate (second arrows pointing down), and acridine orange fluorescence quenching was recorded until a steady-state level of ΔpH was reached. LiCl (2 mM) (left) or NaCl (10 mM) (right) was then added (arrows pointing up), and acridine orange fluorescence dequenching was recorded until a new steady state was obtained. All experiments were repeated four times, and the results were virtually the same. (B) For calculation of the IC₅₀ of 2-AP, the activity of Yp-NhaA was plotted against the 2-AP concentration. One hundred percent dequenching corresponds to activity in the absence of 2-AP. The IC₅₀ of 2-AP was 3.2 μM in the presence of 2 mM LiCl and 3.1 μM in the presence of 10 mM NaCl.

Blood and blood serum survival assay.

Defibrinated sheep blood and sheep serum (Colorado Serum Company) were used for the animal blood and serum survival assays. Human blood was purchased from the Biological Specialty Corporation. Human serum was purchased from Lonza. Bacterial strains were grown overnight in LB medium at 30°C, washed with PBS, and suspended in PBS. Approximately 10⁵ or 10⁶ CFU of the diluted bacteria was inoculated into 4 ml of the blood or blood serum and was shaken for 18 h at 30°C. CFU counts were determined before and after shaking in defibrinated sheep blood or sheep blood serum by plating on LB agar and incubating plates for 2 days.

Growth assay.

The artificial serum medium Opti-MEM I Reduced Serum Medium (referred to below as Opti-MEM) (Invitrogen) was used for in vitro growth assays. LB medium (Difco) buffered with BTP (Sigma) at pH 6.5 or pH 7.4 (LBB) was also used for growth assays. The pH of LB medium was adjusted by the addition of HCl or NaOH. Bacterial strains were grown overnight in LB medium at 30°C, washed with PBS, and suspended in PBS to obtain an OD₆₀₀ of 1.0. Portions (1 μl and 10 μl) of the diluted bacteria were inoculated into 5 ml of Opti-MEM or LB medium and were shaken at 30°C for 24 h. Bacterial growth was evaluated by measuring the OD₆₀₀.

RESULTS

Na⁺/H⁺ antiporters are essential for Y. pestis virulence.

To investigate the roles of sodium pumps in the virulence of Y. pestis, we constructed mutant strains with deletions of the primary sodium pump, NQR, or the predicted main secondary sodium pump, the Na⁺/H⁺ antiporter NhaA from Y. pestis KIM5-3001 (nhaB) (11, 24). We tested the virulence of these mutant strains by using an established mouse model of septicemic plague (25). While the KIM5-3001 (nhaB) Δnqr strain showed a level of virulence similar to that of the parent strain (Fig. 2A), the KIM5-3001 ΔnhaA (nhaB) strain was highly attenuated when tested in this model (data not shown). To further confirm these results, a new nhaA deletion mutant strain was constructed in Y. pestis KIM5. Again, high attenuation was observed for this Y. pestis KIM5 ΔnhaA (nhaB) strain in mice. These attenuated phenotypes were partially reversed by providing the nhaA gene carrying its native promoter on a low-copy-number plasmid, pWKS130, to the KIM5-3001 ΔnhaA (nhaB) (data not shown) and KIM5 ΔnhaA (nhaB) (Fig. 2B) strains.

Fig 2.

Effects of Y. pestis sodium pump mutations on Y. pestis virulence. (A and B) The parent Y. pestis strain KIM5-3001 [WT (KIM5-3001)] and its ΔnqrABCDEF derivative (A) or the parent strain Y. pestis KIM5 harboring pWKS130 (WT/pWKS130), a ΔnhaA derivative harboring pWKS130, and the ΔnhaA strain carrying plasmid pWKS130 containing the intact nhaA gene and its native promoter (ΔnhaA-comp) (B) were assessed by intravenous injection of groups of mice with 1,000 CFU. (C and D) The parent Y. pestis strain CO92 [WT (CO92)] and its nhaA (C) or nhaA nhaB (D) deletion derivative were assessed by intradermal injection of groups of 8-week-old female RML mice. A total of 10 CFU (C) or 100 CFU (D) was injected. P values were calculated by the log rank test.

Because the nhaB gene of Y. pestis KIM strains contains a frameshift, our nhaA deletion variants in the KIM5-3001 (nhaB) and KIM5 backgrounds lack both functional NhaA and functional NhaB. Thus, we next asked whether both NhaA and NhaB are essential for Y. pestis virulence. To test this, we used a fully virulent Y. pestis strain, CO92 that has three intact Na⁺/H⁺ antiporters, nhaA, nhaB, and nhaC (Fig. 1). We constructed single (nhaA) and double (nhaA nhaB) mutant derivatives of Y. pestis CO92 and tested their virulence in an intradermal mouse model of bubonic plague. Unlike the KIM5-3001 ΔnhaA (nhaB) or KIM5 ΔnhaA (nhaB) strain, the Y. pestis CO92 ΔnhaA strain showed full virulence, similar to that of the parent strain (Fig. 2C). Since the Y. pestis CO92 ΔnhaA ΔnhaB double mutant was fully attenuated, like the KIM5-3001 ΔnhaA (nhaB) or KIM5 ΔnhaA (nhaB) strain (Fig. 2D), these data suggested that only the loss of both NhaA and NhaB results in an attenuated phenotype.

Na⁺/H⁺ antiporters are essential for Y. pestis survival in sheep blood and sheep serum.

Since survival in blood is key to Y. pestis virulence, we tested the survival of the nhaA (nhaB) mutant, the nqr mutant, and the parent strain in animal blood. For this purpose, we grew the Y. pestis KIM5-3001 (nhaB), KIM5-3001 Δnqr (nhaB), and KIM5-3001 ΔnhaA (nhaB) strains in defibrinated sheep blood and monitored bacterial survival by measuring the CFU after 18 h. Whereas the KIM5-3001 (nhaB) parental strain and the KIM5-3001 Δnqr strain showed significant increases in the CFU after 18 h in sheep blood, the KIM5-3001 ΔnhaA (nhaB) strain showed a significant decrease in the CFU after 18 h in sheep blood (Fig. 3A). Like the Y. pestis KIM5-3001 (nhaB) derivative, the KIM5 ΔnhaA (nhaB) strain also showed decreased survival in sheep blood (Fig. 3B). Provision of the nhaA gene in trans allowed the KIM5 ΔnhaA (nhaB) strain to grow in blood similarly to the parent strain (Fig. 3B), confirming that the nhaA gene is required for the survival of Y. pestis KIM (nhaB) strains in blood. However, our in vivo data suggested that either NhaA or NhaB is essential for Y. pestis virulence (Fig. 2C and D). To confirm that either nhaA or nhaB is required for Y. pestis survival in blood, we cloned the nhaB gene into a pET expression vector under the control of its IPTG-inducible hybrid T7 lac promoter. The resultant plasmid was transformed into the KIM5 ΔnhaA (nhaB) mutant, and the resultant strain was tested for survival in sheep blood. Expression of the nhaB gene by adding IPTG restored the survival of the KIM5 ΔnhaA (nhaB) strain in sheep blood to parental-strain levels (Fig. 3C), suggesting that only the loss of both antiporters produces a survival defect. The CFU assays for KIM5 ΔnhaA (nhaB) survival in sheep blood were also performed by plating on brain heart infusion agar plates, an optimal growth medium for Y. pestis, instead of LB agar plates; however, similar defects in recovered CFU were observed on the two types of plates.

Fig 3.

Survival of the Y. pestis sodium pump mutants and parent strains in defibrinated sheep blood. Approximately 10⁵ CFU of bacteria was inoculated into 4 ml of defibrinated sheep blood, and the mixture was shaken at 30°C for 18 h. The numbers of surviving bacteria (CFU) in each sample were determined by plating on LB agar plates and incubating the plates at 30°C for 2 days. The growth increase is the ratio of inoculated bacteria to bacteria recovered after the 18-h shaking. (A) The parent strain Y. pestis KIM5-3001 (WT) and its nqrABCDEF and nhaA deletion derivatives were tested. (B) The parent strain Y. pestis KIM5 carrying an empty plasmid (WT/pWSK130), its nhaA deletion derivative carrying an empty plasmid (nhaA/pWKS130), and the nhaA deletion strain carrying plasmid pWKS130 containing the intact nhaA gene (nhaA-comp) were tested. (C) The parent strain Y. pestis KIM5 (WT/pET), its nhaA deletion derivative (nhaA/pET), and the nhaA deletion strain carrying plasmid pET containing the intact nhaB gene under the control of the T7 lac promoter (nhaA/pET-nhaB) were tested. IPTG was added at a final concentration of 1 mM. The error bars indicate standard deviations.

We next tested the survival of the ΔnhaA (nhaB) mutant, the Δnqr mutant, and the parent strain in animal serum. As in the blood survival assay, the Y. pestis KIM5-3001 ΔnhaA (nhaB) strain showed significantly decreased CFU, whereas the KIM5-3001 (nhaB) parent strain and the KIM5-3001 Δnqr (nhaB) strain showed significantly increased CFU, after 18 h in sheep serum (Fig. 4). The Y. pestis KIM5 ΔnhaA (nhaB) strain also showed decreased survival in sheep serum, and providing the nhaA gene in trans completely restored the ability of the KIM5 ΔnhaA (nhaB) strain to survive in sheep serum (data not shown). These data indicate that the combined loss of two Na⁺/H⁺ antiporters, NhaA and NhaB, strongly inhibited Y. pestis growth in blood, suggesting that reduced survival in blood might be responsible for the attenuated in vivo phenotype of the nhaA nhaB mutants. Since a similar survival defect phenotype of the nhaA nhaB double mutant was also observed in blood serum, the immune cells in blood are clearly not responsible for the survival defect of the nhaA mutant in blood. It should be noted that the Y. pestis double mutant did not show a survival defect phenotype in blood and serum when we used a higher inoculation dose (>10⁶ CFU) (data not shown).

Fig 4.

Survival of Y. pestis sodium pump mutants and parent strains in sheep serum. Approximately 10⁵ CFU of the parent strain Y. pestis KIM5-3001 (WT) or its nqr or nhaA deletion derivative was inoculated into 4 ml of sheep serum, and the mixture was shaken at 30°C for 18 h. The number of surviving bacteria (CFU) in each sample was determined by plating onto LB agar plates and incubating the plates at 30°C for 2 days. The growth increase is the ratio of inoculated bacteria to bacteria recovered after the 18-h shaking. The error bars indicate standard deviations.

Y. pestis Na⁺/H⁺ antiporter mutants show a growth defect in an artificial serum medium.

Since serum has complement-mediated lysis and antimicrobial peptide-mediated killing activities, we next tested whether such antimicrobial activities in serum are responsible for the decreased survival rates of the Y. pestis antiporter mutants in blood and serum. For this purpose, we tested the survival of the Y. pestis KIM5-3001 ΔnhaA (nhaB) mutant in heat-inactivated sheep serum. The heat-inactivated serum should not contain active complement but might retain heat-stable antimicrobial peptides. Y. pestis KIM5-3001 (nhaB) showed increased CFU counts after 18 h in heat-inactivated sheep serum, whereas the KIM5-3001 ΔnhaA (nhaB) strain showed decreased CFU counts in heat-inactivated sheep serum (data not shown). This suggested that the complement system present in serum is not responsible for the decreased CFU count of the antiporter mutant in serum. To eliminate any role of heat-resistant antimicrobial peptides, we next investigated the growth of our Y. pestis strains in the artificial serum medium Opti-MEM I Reduced Serum Medium (Opti-MEM). Strain Y. pestis KIM5 (nhaB) grew relatively well in Opti-MEM, whereas the KIM5 ΔnhaA (nhaB) strain showed a severe growth defect in Opti-MEM at 24 h after inoculation (Fig. 5A). However, at 48 h after inoculation, the KIM5 ΔnhaA (nhaB) strain showed a growth yield similar to that of the parent strain (data not shown). This growth phenotype of the KIM5 ΔnhaA (nhaB) strain in Opti-MEM could be complemented by expressing either nhaA or nhaB in trans (Fig. 5A and B). Since Opti-MEM contains no biological factors (e.g., complement or antimicrobial peptides) to kill bacterial cells, these data indicated that the various antimicrobial activities found in serum are not responsible for the growth defect of the Y. pestis antiporter mutant in blood and serum.

Fig 5.

Growth of Y. pestis in Opti-MEM. Approximately 10⁵ CFU of bacteria was inoculated into 5 ml of Opti-MEM, and the mixture was shaken at 30°C for 24 h. Bacterial growth was measured by the OD₆₀₀. (A) The parent strain Y. pestis KIM5 carrying an empty plasmid (WT/pWKS130), its nhaA deletion derivative strain carrying an empty plasmid (nhaA/pWKS130), and the nhaA strain carrying plasmid pWKS130 containing the intact nhaA gene (nhaA-comp) were tested. (B) The parent strain Y. pestis KIM5 carrying an empty plasmid (WT/pET), its nhaA deletion derivative strain carrying an empty plasmid (nhaA/pET), and the ΔnhaA strain carrying plasmid pET containing the intact nhaB gene under the control of the T7 lac promoter (nhaA/pET-nhaB) were tested. IPTG was added at a final concentration of 1 mM. The error bars indicate standard deviations.

Na⁺ concentrations in blood, serum, and Opti-MEM are maintained at relatively high levels (135 to 145 mM). Since our data suggested that the growth defect of the Y. pestis antiporter mutant is not due to antimicrobial factors found in blood and serum, we hypothesized that Na⁺/H⁺ antiporters may be essential for bacterial survival in such relatively high salt concentrations. Consistent with this idea, the Y. pestis KIM5 ΔnhaA (nhaB) strain showed slower growth than the parent strain in LB medium containing 140 mM NaCl, buffered at pH 7.4 [LBB (pH 7.4)] (Fig. 6). Such growth retardation was specific to NaCl and was not observed in LBB (pH 7.4) containing 140 mM KCl (Fig. 6). Moreover, the KIM5 ΔnhaA (nhaB) strain showed no growth defect phenotype in LBB (pH 6.5) containing 140 mM NaCl (data not shown), indicating that such a growth defect phenotype is pH dependent. As in the blood and blood serum survival assays, we again noticed that the KIM5 ΔnhaA (nhaB) strain showed a growth defect phenotype in Opti-MEM and LBB (pH 7.4) containing 140 mM NaCl at 24 h, but not at 48 h, after inoculation (data not shown). Also as in the blood and blood serum survival assays, the KIM5 ΔnhaA (nhaB) strain showed no growth defect in Opti-MEM or LBB (pH 6.5) containing 140 mM NaCl when the inoculation size was greater than 10⁶ CFU (data not shown).

Fig 6.

Growth of a Y. pestis mutant strain in LBB (pH 7.4) with either 140 mM NaCl or 140 mM KCl. Approximately 10⁵ CFU of the parent strain Y. pestis KIM5 (WT) or its nhaA deletion derivative was inoculated into 5 ml of LBB (pH 7.4) containing 140 mM NaCl (open bars) or KCl (filled bars), and the mixture was shaken at 30°C for 24 h. Growth was measured by the OD₆₀₀. The error bars indicate standard deviations.

Pharmacology of Na⁺(Li⁺)/H⁺ antiport catalyzed by Y. pestis NhaA.

To probe the pharmacological profile of Yp-NhaA, Yp-NhaA activity was measured directly in everted membrane vesicles. For this purpose, we cloned the Y. pestis nhaA gene into the pBAD-TOPO vector and transformed the resultant pBAD-YpNhaA plasmid into the antiporter-deficient E. coli strain EP432 (12). In the presence of arabinose, Yp-NhaA restored the ability of strain EP432 to grow at high concentrations of LiCl and NaCl, verifying the functional expression of this antiporter in the heterologous host E. coli (data not shown). Further, we isolated sub-bacterial membrane vesicles from EP432/pBAD-YpNhaA cells and analyzed the pH profile and activity of Yp-NhaA, as well as its affinities for Na⁺ and Li⁺, as in a previous study (11). The results obtained for LiCl⁺ and NaCl⁺ were essentially the same. In both cases, the antiporter was inactive at pHs below 6.5 and displayed manyfold alkaline activation, reaching maxima for Na⁺/H⁺ and Li⁺/H⁺ exchange at pH 8.0. We also confirmed that Yp-NhaA has an almost 5-fold higher affinity for Li⁺ than for Na⁺, as was found in reference 11, although our measurements yielded Na⁺ and Li⁺ concentrations required for half-maximal response of 3.5 mM (Fig. 7, top) and 0.7 mM (Fig. 7, bottom), respectively. These numbers are somewhat higher than those reported for Yp-NhaA by Arkin and collaborators (0.86 mM and 0.18 mM) (11). A probable reason for these differences is the use of different host strains for the isolation of vesicles and/or the different compositions of the experimental buffers used in the fluorescence assays.

Considering Yp-NhaA as a prospective target for antibacterial treatments, we checked its susceptibility to the known NhaA-type antiporter inhibitor 2-aminoperimidine (2-AP). It is well known that amiloride and its analogues are competitive inhibitors of mammalian Na⁺/H⁺ antiporters that belong to the Na⁺/H⁺ exchanger (NHE) family (26). Amiloride also inhibits the bacterial Na⁺/H⁺ antiporter NhaB, but not NhaA (27). In a previous study, it was found that 2-AP, a guanidine-containing derivative fused to a naphthalene moiety and a distant analogue of amiloride, acted as a specific inhibitor of NhaA-type antiporters from E. coli and V. cholerae (28). We therefore tested the effect of this compound on Yp-NhaA (Fig. 8). As expected, we found that Yp-NhaA was sensitive to 2-AP, with 50% inhibitory concentrations (IC₅₀) at pH 8.0 of 3.1 μM for 10 mM Na⁺ and 3.2 μM for 2 mM Li⁺ (Fig. 8). Given the differences in the apparent K_ms of Yp-NhaA for Li⁺ and Na⁺, as well as the fact that 2-AP acted as a competitive inhibitor, it is not surprising that the Li⁺/H⁺ antiport measured (with 2 mM LiCl added as a substrate) was inhibited by the same 2-AP concentrations as those inhibiting Na⁺/H⁺ antiport measured with 10 mM NaCl (Fig. 8A, compare left and right sides).

DISCUSSION

We have proposed previously that the essential role of the sodium cycle in the energy metabolism of many pathogenic organisms should provide a distinct set of targets for the development of novel intervention strategies (4). Some bacteria, including Y. pestis, can use sodium as a coupling ion, utilizing a number of primary and secondary Na⁺ pumps that are important for overall bacterial physiology. In this study, we start to investigate the roles of sodium pumps in Y. pestis virulence. We found that while the primary sodium pump, NQR, did not play a major role in virulence, the combined loss of two major secondary sodium pumps, NhaA and NhaB, resulted in a strongly attenuated phenotype when strains were tested in vivo. To our knowledge, this is the first report that establishes a role for Na⁺/H⁺ antiporters in bacterial virulence.

The data presented here show that a Y. pestis nhaA nhaB double mutant has a survival defect in sheep blood and serum. However, the nhaA nhaB double mutant survived and replicated in RAW 264.7 macrophages similarly to the parent strain (data not shown). Furthermore, we found that the double mutant still showed decreased survival in heat-inactivated serum. Taken together, these data suggest that the antibacterial factors present in blood and serum are not responsible for the decreased survival of the Y. pestis nhaA nhaB mutant in blood or serum. Indeed, the double mutant also showed growth retardation when grown in an artificial serum medium, Opti-MEM, further confirming this idea. Blood plasma normally contains 135 to 155 mM Na⁺; similar levels of sodium in bacterial cytoplasm are considered lethal (29). Thus, these observations led us to consider the possibility that the Na⁺/H⁺ antiporters NhaA and NhaB are essential for the survival of Y. pestis in blood, since they protect the bacterial cells from Na⁺ toxicity. Indeed, the double mutant strain showed a growth defect in high sodium but not high potassium concentrations.

In addition, we found, unexpectedly, that the ΔnhaA ΔnhaB double mutant showed survival rates similar to those of the parent strain and the nqr mutant in human blood but was defective for growth in human serum (data not shown). Because human blood samples were treated by adding heparin, whereas the sheep blood samples were processed only to remove fibrin, with no additives or preservatives added, we hypothesized that heparin may affect bacterial survival in blood. Indeed, we observed that the ΔnhaA ΔnhaB double mutant showed survival rates similar to those of the parent strain in sheep blood when sodium heparin was added to the blood (data not shown). However, our data indicated that the Y. pestis antiporters are not required for survival mechanisms related to immune factors, so likely heparin exerts its effect via an alternative mechanism, such as altering the sodium concentrations. It is noteworthy that heparin has the highest negative charge density of any known biological molecule (30), and clearly, future studies are required to explain this intriguing finding.

The lack of relevance of NQR to the survival of Y. pestis under these conditions might not be surprising, because this primary sodium pump creates a transmembrane electrical gradient (Δψ) (“plus” outside) on the membrane as it expels Na⁺ ions; therefore, its ability to build up a sizable chemical transmembrane gradient of Na⁺ (which would lead to lower internal sodium levels) is very limited. On the other hand, the secondary sodium-proton exchangers, NhaA and NhaB, being electrogenic (they exchange 2 H⁺ ions per Na⁺ ion and 3 H⁺ ions per 2 Na⁺ ions, respectively), are able to consume Δψ for the very efficient removal of cytoplasmic Na⁺ (see references 9 and 10 and references therein).

We observed that the mice that survived infection with either the KIM5 ΔnhaA (nhaB) or the CO92 ΔnhaA ΔnhaB strain never showed any signs of illness, and no CFU were recovered from any of their spleen and blood tissues (data not shown). Mice with bubonic plague with lymph node infection but prior to disseminated sepsis typically show no obvious signs of disease, and illness is detected only after disseminated spread. The concentrations of sodium ions in the majority of mammalian extracellular fluids, including those in liver and spleen, are indeed rather high and comparable to that in blood plasma, i.e., ∼100 to 140 mM (reference 31 and references therein). It is reasonable to expect, therefore, that Na⁺/H⁺ antiport-deficient Y. pestis would be unable to grow there, as in blood. Thus, we conclude that the ΔnhaA (nhaB) mutant did not spread to the internal organs; it died out at the intradermal injection site or in the draining lymph node. We also did not notice any swelling or pathology in these lymph nodes.

We found that the Y. pestis nhaA nhaB double mutant strain recovered the ability to grow to the wild-type (WT) level in Opti-MEM and LBB (pH 7.4) containing 140 mM NaCl after a longer incubation (48 h) (data not shown). Similar growth recovery has been observed in an E. coli nhaA nhaB double mutant in LB medium containing 300 mM NaCl, where it was attributed to the expression of the “silent” primary Na⁺ pump in the respiratory chain (32). However, no such recovery was observed for the Y. pestis mutant in sheep blood and in vivo experiments. Thus, such a difference in growth abilities might be caused simply by the fact that LB medium (and Opti-MEM) is a richer medium than serum/blood.

Of note, when the Y. pestis KIM5 and KIM5 ΔnhaA strains were evaluated in growth and secretion experiments using standard TMH medium (38) with or without calcium, both strains showed typical calcium-dependent growth and Yop secretion, which were indistinguishable between them, eliminating the possibility that the low-calcium-response-mediated Na⁺-sensitive phenotype (33–35) is responsible for the Na⁺ sensitivity of the KIM5 ΔnhaA (nhaB) strains.

Competitive inhibitors of mammalian Na⁺/H⁺ antiporters belonging to the NHE family, including the diuretic drug amiloride and its numerous derivatives, have been studied extensively (26). However, until recently, there were no inhibitors specific for the antiporters of the NhaA type (28). The data presented in this paper show that 2-AP acts efficiently on Yp-NhaA, similarly to its inhibition of the NhaA-type enzymes in V. cholerae or E. coli (29). This offers the potential for 2-AP to become a platform for the design of medically relevant inhibitors of the bacterial Na⁺/H⁺ antiporters of the NhaA type. This could be particularly important for the development of novel antimicrobials targeting blood-borne bacterial infections.

In summary, the data presented in this communication demonstrate the essential role of Na⁺/H⁺ antiport in Y. pestis virulence, possibly by protecting against sodium toxicity in blood. Because Na⁺/H⁺ antiporters are ubiquitous proteins, these findings might be applicable to other blood-borne pathogens.