The Helicobacter pylori CZB Cytoplasmic Chemoreceptor TlpD Forms an Autonomous Polar Chemotaxis Signaling Complex That Mediates a Tactic Response to Oxidative Stress

ABSTRACT

Cytoplasmic chemoreceptors are widespread among prokaryotes but are far less understood than transmembrane chemoreceptors, despite being implicated in many processes. One such cytoplasmic chemoreceptor is Helicobacter pylori TlpD, which is required for stomach colonization and drives a chemotaxis response to cellular energy levels. Neither the signals sensed by TlpD nor its molecular mechanisms of action are known. We report here that TlpD functions independently of the other chemoreceptors. When TlpD is the sole chemoreceptor, it is able to localize to the pole and recruits CheW, CheA, and at least two CheV proteins to this location. It loses the normal membrane association that appears to be driven by interactions with other chemoreceptors and with CheW, CheV1, and CheA. These results suggest that TlpD can form an autonomous signaling unit. We further determined that TlpD mediates a repellent chemotaxis response to conditions that promote oxidative stress, including being in the presence of iron, hydrogen peroxide, paraquat, and metronidazole. Last, we found that all tested H. pylori strains express TlpD, whereas other chemoreceptors were present to various degrees. Our data suggest a model in which TlpD coordinates a signaling complex that responds to oxidative stress and may allow H. pylori to avoid areas of the stomach with high concentrations of reactive oxygen species.

IMPORTANCE Helicobacter pylori senses its environment with proteins called chemoreceptors. Chemoreceptors integrate this sensory information to affect flagellum-based motility in a process called chemotaxis. Chemotaxis is employed during infection and presumably aids H. pylori in encountering and colonizing preferred niches. A cytoplasmic chemoreceptor named TlpD is particularly important in this process, and we report here that this chemoreceptor is able to operate independently of other chemoreceptors to organize a chemotaxis signaling complex and mediate a repellent response to oxidative stress conditions. H. pylori encounters and must cope with oxidative stress during infection due to oxygen and reactive oxygen species produced by host cells. TlpD's repellent response may allow the bacteria to escape niches experiencing inflammation and elevated reactive oxygen species (ROS) production.

INTRODUCTION

Chemoreceptors operate on the front line of the bacterial chemotaxis response, sensing signals and initiating attractant or repellent responses. Chemoreceptors can be divided into two classes based on whether they reside in the inner membrane or are soluble cytoplasmic proteins. The cytoplasmic class presumably senses signals that occur intracellularly. Cytoplasmic chemoreceptors are relatively understudied but represent 14% of bacterial and 43% of archaeal chemoreceptors (1). As a group, cytoplasmic chemoreceptors are reported to mediate tactic responses to diverse conditions, including to increased cellular energy stores (2, 3), redox conditions (4, 5), and metabolites (6), although significant gaps remain in our understanding of what they sense and how they function (1). One such cytoplasmic chemoreceptor is Helicobacter pylori TlpD. TlpD plays a critical role during H. pylori infection of the mammalian stomach (7, 8). In wild-type mice or gerbils, mutant strains of H. pylori that lack TlpD display colonization defects during the early stages of infection that are more severe than those of mutant strains that lack any other individual chemoreceptor (7, 8).

TlpD, which was previously referred to as HP0599 or HylB, does not have transmembrane domains and resides in both soluble and membrane-associated subcellular fractions (2). TlpD possesses the canonical chemoreceptor domain, called MA or methyl-accepting chemotaxis receptor protein (MCP) signal, which typically interacts with the signal transduction proteins CheW and CheA. TlpD additionally has a C-terminal zinc-binding domain (CZB) (9, 10), which is the second most prevalent domain found in cytoplasmic chemoreceptors (1). Numerous cytoplasmic chemoreceptors contain a CZB domain and thus share a domain structure with TlpD (1, 9). Chemoreceptors, like TlpD, are widespread and found in many bacterial genera, including gastric and nongastric Helicobacter species. CZB domains have been shown to function as zinc-responsive allosteric regulators in diguanylate cyclases (10), but their function in TlpD and other chemoreceptors is not known. Information relating to regulation of the transcription or translation of tlpD is sparse. There is some evidence that tlpD and tlpB are upregulated in gerbil infections. The regulatory mechanisms, however, are not yet known (11).

The H. pylori chemotaxis signal transduction system is reasonably well understood. It contains typical core signaling proteins, the CheA kinase, the CheW coupling protein, and the CheY response regulator, all of which are required for chemotaxis (1, 12, 13). H. pylori encodes three transmembrane chemoreceptors called TlpA (HP0099), TlpB (HP0103), and TlpC (HP0082) (7, 8, 12, 14). H. pylori also possesses a few accessory chemotaxis signaling proteins that supplement the core ones. In addition to CheW, H. pylori expresses three coupling proteins called CheV1, CheV2, and CheV3. CheV proteins combine the CheW coupling domain with an additional phosphorylatable response regulator domain (REC) (15). H. pylori CheV proteins all function in chemotaxis, but their exact roles are unclear (17, 19). In Bacillus subtilis, CheV modulates CheA activity depending on the covalent modification state of the chemoreceptors and the phosphorylation state of CheV (20). H. pylori also encodes two more proteins that are critical for chemotaxis. The first is a CheZ phosphatase, called CheZ_HP, which is capable of dephosphorylating CheY, CheA, and CheV2 (21). The second is a polyglutamate-rich protein called ChePep. ChePep contains a putative response regulator domain and is conserved in Epsilonproteobacteria (22).

Chemotaxis proteins in H. pylori are organized into two distinct polar complexes. One complex contains the chemoreceptors CheA and CheV1 (23), while the other contains CheZ_HP and ChePep (23). CheZ_HP and ChePep physically interact with each other and form a distinct complex at the cell pole that is independent of the one formed by the chemoreceptors and other signaling proteins (23). It is not known why H. pylori organizes its chemotaxis proteins into two distinct complexes.

Chemotaxis appears to have the greatest impact on colonization within the first 3 weeks of infection, consistent with the idea that this signal transduction system plays a particularly important role in early H. pylori colonization (7, 18, 24–28). tlpD mutants are substantially affected in their ability to colonize the stomach at early time points (7, 8), presumably because tlpD-deficient strains are unable to sense signals that are generated by exposure to the host environment and are therefore unable to locate a preferred niche in the stomach. At later time points, colonization defects due to chemotaxis and TlpD are less apparent (16, 18). TlpD is reported to mediate a tactic response to electron transport chain inhibitors in vitro, a behavior ascribed to energy taxis (2), but it is not yet clear to which signal TlpD responds during such treatment. Studies have shown that H. pylori does not display aerotaxis, so a direct oxygen response appears unlikely (16). TlpD has no identifiable domains that would bind reduced flavin adenine dinucleotide (FADH₂) or flavin mononucleotide (FMN) at either its N or C terminus, but it has been reported to interact with iron to a small degree (9). Given the important role TlpD plays during host colonization, we sought to understand the function of this chemoreceptor in greater detail. Our results suggest that TlpD is able to operate independently of transmembrane chemoreceptors and coordinates a chemotaxis complex at the pole that contains at least CheA, CheW, CheV1, and CheV3, confirming that it is a bona fide chemoreceptor. Furthermore, we report that TlpD mediates a repellent response to conditions that promote oxidative stress, including being in the presence of iron, hydrogen peroxide, paraquat, and metronidazole. We propose a model in which TlpD-mediated repellent responses to oxidative stress are vital to H. pylori avoiding mammalian niches with elevated oxygen, such as near the epithelial surface (29, 30), or reactive oxygen species (ROS) that arise from sites of inflammation (31, 32).

MATERIALS AND METHODS

Bacterial strains and culture conditions.

H. pylori strains mG27, G27, and SS1 were used for strain construction and experimental studies (Table 1). Other H. pylori strains were used to analyze chemoreceptor diversity, as noted in Table 1. H. pylori was cultured on either Columbia horse blood agar (CHBA), brucella broth with 10% fetal bovine serum (FBS; Life Technologies) (BB10), or 0.8× Ham's F-12 with 10% FBS (Ham's-10). CHBA consisted of Columbia agar (BD) with 5% defibrinated horse blood (Hemostat Labs, Davis, CA), 50 μg/ml cycloheximide, 10 μg/ml vancomycin, 5 μg/ml cefsulodin, 2.5 U/ml polymyxin B, and 0.2% (wt/vol) β-cyclodextrin. Cultures were incubated at 37°C under 5 to 7% O₂, 10% CO₂, and balance N₂. For the selection of H. pylori mutants, 10% (wt/vol) sucrose, 5 to 10 μg/ml chloramphenicol, and 15 μg/ml kanamycin were used. For the selection of plasmid-bearing Escherichia coli, 20 μg/ml chloramphenicol, 30 μg/ml kanamycin, and 100 μg/ml ampicillin were used. All media, chemicals, and antibiotics were from BD, Fisher, Sigma, Gold Biosciences, or ISC BioExpress.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	KO reference no.	Genotype or description	Reference and/or source(s)
Strains
G27	KO379		69; Nina Salama
mG27	KO625	Mouse-adapted G27	70
mG27 ΔcheA	KO629	KO625 ΔcheA::cat	33
mG27 ΔtlpA	KO1002	KO1001 ΔtlpA	33
mG27 ΔtlpB	KO1004	KO625 ΔtlpB	33
mG27 ΔtlpC	KO1005	KO625 ΔtlpC::aphA3	33
mG27 ΔtlpD	KO1006	KO625 ΔtlpD::cat	33
mG27 ΔtlpA ΔtlpC	KO1008	KO1002 ΔtlpC::aphA3	23
mG27 ΔtlpA ΔtlpB	KO1007	KO1022 ΔtlpB	23
mG27 ΔtlpA ΔtlpD	KO1009	KO1002 ΔtlpD::cat	23
mG27 ΔtlpB ΔtlpD	KO1012	KO1004 ΔtlpD::cat	23
mG27 ΔtlpB ΔtlpC	KO1011	KO1004 ΔtlpC::aphA3	23
mG27 ΔtlpC ΔtlpD	KO1013	KO1006 ΔtlpC::aphA3	23
mG27 ΔtlpA ΔtlpC ΔtlpD	KO1014	KO1009 ΔtlpC::aphA3	23
mG27 ΔtlpA ΔtlpB ΔtlpD	KO1015	KO1009 ΔtlpB	23
mG27 ΔtlpA ΔtlpB ΔtlpC	KO1016	KO1007 ΔtlpC::aphA3	23
mG27 ΔtlpB ΔtlpC ΔtlpD	KO1017	KO1011 ΔtlpD::cat	23
mG27 ΔtlpA ΔtlpB ΔtlpC ΔtlpD	KO1021	KO1015 ΔtlpC::aphA3	23
SS1	KO457		71; Janie O'Rourke
SS1 ΔcheY	KO481	SS1 ΔcheY::cat-102	27
SS1 ΔmotB	KO479	SS1 ΔmotB::cat-1	26
SS1 ΔtlpA	KO661	SS1 ΔtlpA::cat	72
SS1 ΔtlpB	KO1108	SS1 ΔtlpB::cat-2	18
SS1 ΔtlpC	KO565	SS1 ΔtlpC::cat	72
SS1 ΔtlpD	KO914	SS1 ΔtlpD::cat-1	18
SS1 ΔcheV1	KO864	SS1 ΔcheV1::cat	17
SS1 ΔcheV2	KO865	SS1 ΔcheV2::cat	17
SS1 ΔcheV3	KO856	SS1 ΔcheV3::cat	17
SS1 ΔcheA	KO855	SS1 ΔcheA::cat	27
SS1 ΔcheW	KO852	SS1 ΔcheW::aphA3	27
J99	KO479	Wild type	73; Nina Salama
26695	KO668	Wild type	14; ATCC 700392
X47	KO612	Wild type	74; Doug Berg
M6	KO613	Wild type	75; Elizabeth Joyce, Andrew Wright
CPY3401	KO637	Wild type	76; Doug Berg
ATCC 43504	KO669	Wild type	77; ATCC (also known as NCTC 11637)
Alston	KO610	Wild type	78; Elizabeth Joyce
HP 12-1	KO760	Wild type/clinical isolate	Jay Solnick
HP 28-1	KO761	Wild type/clinical isolate	Jay Solnick
HP 116-1	KO764	Wild type/clinical isolate	Jay Solnick
HP 125-1	KO765	Wild type/clinical isolate	Jay Solnick
Plasmid pGEX6P2-CheW		pGEX6p2::CheW	This work

Creation of mG27 mutants.

Construction of the ΔcheA, ΔtlpA, ΔtlpB, ΔtlpC, and ΔtlpD mutants has been described (33) (Table 1). Construction of the multiple receptor mutants was described recently (23). In brief, mG27 ΔtlpA or ΔtlpB was used as the starting strain, and then tlpC was eliminated by transformation with plasmid pKO150, or tlpD was eliminated by transformation with the ΔtlpD::cat chromosome from strain KO1006. In all cases, the resulting mutations were verified using PCR.

Motility and chemotaxis analyses.

Soft-agar assays were performed using brucella broth, 2.5% (vol/vol) FBS, and 0.35% Bacto agar, as previously described (26). Soft-agar plates were incubated under microaerobic conditions, at 37°C, for 5 to 7 days. To obtain films of swimming H. pylori, bacteria were inoculated from plates into 5 ml of Ham's-10 in a 50-ml flask and then incubated with shaking for approximately 15 to 17 h prior to filming. To inspect for tactic responses to extracellular iron or hydrogen peroxide, liquid cultures were checked for motility, diluted to an optical density at 600 nm (OD₆₀₀) of 0.12 into prewarmed Ham's-10 containing Fe₂(SO₄)₃, FeCl₃, or H₂O₂ that had been prepared immediately prior to use, at concentrations indicated in the text, and filmed immediately for a maximum duration of 5 min. To inspect for tactic responses to iron chelation or superoxide production, cultures were similarly diluted into 50 μM dipyridyl, 10 μg/ml metronidazole, or 10.5 μM paraquat and then incubated for 15 min under microaerobic conditions. After this period, the cultures were either filmed immediately or treated with Fe₂(SO₄)₃, FeCl₃, or H₂O₂ and subsequently filmed immediately for a maximum duration of 5 min. The 50% effective concentrations (EC₅₀s) were determined by nonlinear regression of direction change frequencies obtained by tracking wild-type (WT) mG27 exposed to a range of FeCl₃ concentrations using GraphPad Prism 6 (GraphPad Software, Inc., San Diego, CA). After mixing, cultures were filmed immediately for a maximum duration of 5 min using phase-contrast microscopy with the SimplePCI software (version 5.2.1.1609; Compix 2003) using a Hamamatsu C4742-95 digital camera mounted on Nikon Eclipse E600 at 400× magnification. Videos were recorded at maximum acquisition speed (approximately 16 frames/s), exported into .AVI format, and then relabeled to blind the analyzer to the condition.

For reversal frequency analysis, individual bacterial swimming paths were analyzed by hand tracing the path of a randomly selected cell that was visible for at least 3 s, with each clear direction change counted, as described previously (34).

Plug bridge spatial chemotaxis assay.

For the spatial agarose-in-plug bridge chemotaxis assay, H. pylori SS1 was grown to an OD₆₀₀ of 0.2 to 0.8 in BB10 with shaking and then checked for motility. Cells were collected by centrifugation at 2,000 rpm for 5 min at room temperature, washed in phosphate-buffered saline (PBS), resuspended in PBS plus 10% fetal bovine serum (FBS) that had been treated by boiling for 30 min in the presence of 50 mM EDTA, frozen overnight, the liquid portion filtered through a 0.2-μm-pore-size filter, and dialyzed 2 times against a 3,500-kDa dialysis membrane in PBS, as described previously (34). Motility was assessed by microscopy, as described above. The agarose-in-plug bridge assay was carried out as described by Yu and Alam 35), except that the inherent precipitation of FeCl₃ was used in place of agarose immobilization. Briefly, two plastic coverslips were placed about 16 mm apart on a glass slide and covered by a glass coverslip. Nine to 100 μl of washed and prepared motile cells was pipetted under the top glass coverslip. Immediately following, 2 μl of 100 mM FeCl₃ was pipetted at the edge of the top glass coverslip. The slides were incubated for 15 min under microaerobic conditions, as this duration was empirically determined to lead to a robust chemotactic response. The area immediately surrounding the crystallized iron was observed for the formation of chemotactic bands at 25× magnification using a Zeiss Axiovert 200 inverted microscope, and images were recorded with under bright-field illumination using a Zeiss AxioCam digital camera.

Protein preparation from whole-cell lysates and subcellular fractionation.

For whole-cell lysates, total cell proteins were prepared from H. pylori cultured on CHBA plates for 2 days. The cells were resuspended in ice-cold PBS. The optical density at 600 nm was determined. The suspension was then mixed with 2× Laemmli sample buffer to give a final OD₆₀₀ of 1.0 and then boiled for 10 min. Five to 10 μl of each sample was separated on a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel.

For subcellular fractionation, H. pylori strains were grown in 100 ml of BB10 for 24 to 48 h, generating OD₆₀₀s ranging from 0.24 to 0.55. All cultures were verified for purity and motility by light microscopy and diluted in BB10 to matching OD₆₀₀s. H. pylori cells were collected by centrifugation at 8,500 × g for 10 min at 4°C and resuspended in 1 to 2 ml of lysis buffer [50 mM Tris-HCl (pH 7.0), 10% glycerol, 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF), 10 mM dithiothreitol (DTT)]. Samples were kept at 4°C on ice for all future steps. Resuspended cells were lysed in bursts by sonication until the lysate appeared clear (3 to 5 min). The lysate was then spun at 349,000 × g for 15 min at 4°C in a Beckman TL-100 ultracentrifuge to separate the cytoplasm and membrane fractions. The supernatant (cytoplasm) was removed, and the pellet (membrane fraction) was rinsed three times with 2 ml of high-salt buffer (lysis buffer plus 2 M KCl). The membrane pellet was then completely resuspended in 2 ml of high-salt buffer using an 18-gauge needle and syringe and then collected by centrifugation at 217,000 × g for 15 min at 4°C. This step was repeated once more and then repeated again using lysis buffer instead of high-salt buffer. The final membrane pellet was resuspended in a volume of lysis buffer equal to that of the original sample and stored at −20°C. Protein concentration was measured using the Bio-Rad protein assay, with bovine serum albumin (BSA) as a standard. The samples were boiled in Laemmli sample buffer for 5 min and loaded onto 10% SDS-PAGE gels, with an equal total protein concentration in each lane.

Western blotting.

Samples run on SDS-PAGE gels were transferred to immunoblot polyvinylidene difluoride (PVDF) membranes (Bio-Rad). Successful transfer and relative protein loading were visualized by staining the membrane with DB71 dye prior to antibody binding. Membranes were incubated with a 1:2,000 to 1:5,000 dilution of anti-glutathione transferase (GST)-TlpA22 (18) or a 1:60 dilution of anti-CheV1 (23). For visualization, the blots were incubated with the horseradish peroxidase-conjugated chicken anti-rabbit or goat anti-rabbit secondary antibody (Santa Cruz Biotechnology) at a dilution of 1:2,000 and incubated with luminol, p-coumaric acid, and hydrogen peroxide. Luminescent blots were then exposed to BioMax light film (Kodak) or a Bio-Rad ChemiDoc MP.

Immunofluorescence.

Liquid cultures of H. pylori wild-type mG27 and tlpABC (TlpD only) grown in BB10 to a final OD₆₀₀ of 0.6 to 1.0 were inspected for motility and purity by microscopy prior to fixation on poly-l-lysine-coated slides. Bacteria were fixed and permeabilized as described previously (36) (23). Antibodies were preadsorbed as described previously (23). Preadsorbed anti-His-CheAY (1:200) (23), anti-GST-TlpA22 (1:200) (18), anti-CheV1 (1:100) (23), anti-CheZ (1:200) (23), anti-CheV3 (1:100) (J. Castellon, P. Lertsethtakarn, and K. M. Ottemann, unpublished data), anti-CheW (1:100) (see below), or chicken anti H. pylori (AgriSera AB) (1:500) was added, incubated at room temperature for 30 min, and washed with blocking buffer (3% BSA, 0.1% Triton X-100 in 1× PBS) three times. Goat anti-guinea pig antibody–Alexa Fluor 594 or goat anti-chicken antibody–Alexa Fluor 488 (Abcam) was added at a 1:300 or 1:500 dilution, respectively, and incubated in the dark at room temperature for 30 min. The samples were washed with blocking buffer, as described above, and a drop of Vectashield (Vector Laboratories) was added to the samples prior to them being sealed with coverslips. Imaging was performed on a Zeiss LSM 5 Pascal confocal microscope. Images were taken separately in each channel and merged in Adobe Photoshop CS2.

Protein purification and antibody production.

cheW was amplified from SS1 genomic DNA with the primers cheW623bam (5′-CCCCGGATCCGTGAGCAACCAATTA-3′) and cheW624ecoR1 (5′-CACAGAATTCTTAGAAGTCTTTTTTTAAGATTTC-3′), digested with BamH1 and EcoR1, and ligated into pGEX6P2 (GE Healthcare) to create pGEX6P2-CheW. The resulting plasmid was sequenced to confirm its identity. pGEX6P2-CheW was transformed into ArcticExpress E. coli (Agilent), grown to OD₆₀₀ of 0.6, induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and expressed at 10°C for 24 h. After overexpression, GST-CheW was purified essentially as described for other GST chemotaxis proteins (21). Protein concentrations were determined using the Bradford assay (Bio-Rad) with BSA as a standard. Anti-CheW antibodies were generated in guinea pigs with purified CheW (Cocolico).

RESULTS

TlpD is a bona fide chemoreceptor able to drive a chemotaxis response.

Although certain cytoplasmic chemoreceptors have been reported to respond to specific ligands or conditions, it has remained unclear whether those chemoreceptors can function independently of transmembrane chemoreceptors. To begin our studies on TlpD, we first constructed an H. pylori strain that expresses TlpD as its sole chemoreceptor, and for comparison, strains with all other combinations of the chemoreceptors. We confirmed that each chemoreceptor is expressed in the absence of the others by Western blotting using an antibody raised against the conserved MA domain of TlpA (18) (Fig. 1A). As reported before for G27-derived strains (33), mG27 expresses only three of the four chemoreceptors, TlpA, TlpB, and TlpD, and these were expressed at similar levels regardless of which other chemoreceptors were present, although there were some degradation products detected in TlpB-only strains (Fig. 1A). Strains lacking all chemoreceptor genes displayed no antibody cross-reactivity, consistent with the notion that the known chemoreceptors are the only MA domain-containing proteins in mG27 (Fig. 1A). With these strains in hand, we then analyzed whether the sole chemoreceptors were able to confer chemotactic function, using the common brucella broth (BB) plus fetal bovine serum (FBS) soft-agar chemotaxis assay, which has been a valuable method to characterize the chemotaxis system in H. pylori (16, 24, 37). Mutant strains that retained only TlpD, referred to as TlpD-only strains, retained near-wild-type chemotaxis, similar to mutants that retained TlpB only (Fig. 1B). These findings are consistent with previous work reporting that TlpD-only mutants retain chemotaxis function in a different assay consisting of conditions that alter cellular energy (2). Strains expressing TlpA only, in contrast, had poor chemotaxis ability (Fig. 1B). Mutants lacking all of the chemoreceptors did not form expanded colonies in this assay, similar to a mutant lacking the key chemotaxis kinase CheA (Fig. 1B), suggesting that there are no other chemotaxis receptors in H. pylori that function in this medium. These results suggest that TlpD is able to function on its own to confer chemotaxis.

FIG 1.

TlpD is stably produced when the sole chemoreceptor confers chemotactic function in a soft-agar assay. (A) Whole-cell lysates were analyzed by Western blotting with the anti-TlpA22 antibody that recognizes all four H. pylori chemoreceptors (18). Lane 1, wild-type mG27 (WT); lane 2, mG27 ΔtlpA; lane 3, mG27 ΔtlpB; lane 4, mG27 ΔtlpC; lane 5, mG27 ΔtlpD; lane 6, mG27 ΔtlpA ΔtlpB (ΔtlpAB); lane 7, mG27 ΔtlpA ΔtlpC (ΔtlpAC); lane 8, mG27 ΔtlpA ΔtlpD (ΔtlpAD); lane 9, mG27 WT; lane 10, mG27 ΔtlpB ΔtlpC (ΔtlpBC); lane 11, mG27 ΔtlpB ΔtlpD (ΔtlpBD); lane 12, mG27 ΔtlpC ΔtlpD (ΔtlpCD); lane 13, mG27 ΔtlpA ΔtlpB ΔtlpC (ΔtlpABC); lane 14, mG27 ΔtlpA ΔtlpC ΔtlpD (ΔtlpACD); lane 15, mG27 ΔtlpA ΔtlpB ΔtlpD (ΔtlpABD); lane 16, mG27 ΔtlpB ΔtlpC ΔtlpD (ΔtlpBCD). Not shown is mG27 ΔtlpA ΔtlpB ΔtlpC ΔtlpD, which looks identical to lane 15, as both lack tlpC. Marker sizes are given in kilodaltons on the left side. The migration positions of TlpA, TlpB, and TlpD are indicated at the right. Western blots are representative of 2 to 5 replicates for each strain. (B) Brucella broth-FBS soft-agar chemotaxis assays with the colonial diameter of the indicated H. pylori strains were measured after 5 days. The error bars represent the standard error of the mean. Each strain was analyzed in two to five biological replicates, with ≥3 technical replicates each time. *, P < 0.05; **, P < 0.01, compared to the wild type using Student's t test.

TlpD has all the information necessary to localize to the pole and assemble a multiprotein chemotaxis signaling complex.

The finding that TlpD is able to confer soft-agar migration suggests that it is sufficient to assemble the required chemotaxis signaling proteins. To determine what proteins are assembled by TlpD, we utilized the immunofluorescence of whole cells to visualize the subcellular localization and TlpD dependence of the CheA kinase, the CheW and CheV coupling proteins, and the CheZ_HP phosphatase. Previous published work showed that chemoreceptors and chemotaxis signaling proteins are all polar in wild-type H. pylori (23). In TlpD-only strains, we observed that TlpD localized to the pole, suggesting that it contains the information needed for this subcellular localization (Fig. 2). CheA, CheW, CheV1, and CheV3 likewise localized to the cell pole in a TlpD-dependent manner (Fig. 2). In contrast, the localization of CheZ_HP did not require chemoreceptors, consistent with its localization to an independent complex, as shown previously (23). These results suggest that TlpD has the information necessary to establish and maintain polar localization of itself and to form a chemotaxis signaling complex containing at least CheW, CheA, CheV1, and CheV3.

FIG 2.

TlpD localizes to the pole and coordinates a chemotaxis complex there. Immunofluorescence images of wild-type (WT), TlpD only (D only), and a strain lacking all chemoreceptors (receptorless, ΔtlpABCD) mG27 H. pylori strains. H. pylori whole cells were detected with chicken anti-H. pylori antibodies and stained with goat anti-chicken IgG conjugated to Alexa Fluor 588 (red). Specific chemotaxis proteins were detected as indicated on the left side of the figure, with antibodies as described in Materials and Methods, and subsequently stained with secondary antibodies conjugated to Alexa Fluor 494 (green). The TlpA antibody recognizes all H. pylori chemoreceptors. Scale bar = 4 μm.

TlpD membrane association depends on transmembrane chemoreceptors CheA, CheW, and CheV1.

TlpD has been reported to reside in both the membrane and soluble subcellular fractions (2). It has remained unclear, however, how TlpD associates with the membrane, as it does not have transmembrane domains. Given that TlpD and chemotaxis proteins in general exist in multiprotein complexes (Fig. 2) (38), we examined whether any known chemotaxis proteins were required for TlpD to localize with the membrane fraction. H. pylori whole-cell lysate was separated into cytoplasmic and washed membrane fractions and probed using an antichemoreceptor antibody (18). TlpD was found in both soluble and membrane-associated fractions in wild-type cells (Fig. 3A), consistent with previous reports using a distinct H. pylori strain (2). Both types of subcellular fractions appeared to be substantially pure based on the lack of transmembrane chemoreceptor proteins in the cytoplasmic fraction and the absence of anti-CheV1 cross-reacting cytoplasmic proteins in the membrane fraction (Fig. 3A). In mutants that lack all transmembrane chemoreceptors, the vast majority of TlpD is cytoplasmic (Fig. 3A), suggesting that membrane chemoreceptors promote TlpD membrane association. Consistent with this possibility, the membrane association of TlpD was substantially reduced in strains lacking the core chemotaxis signaling protein CheA or CheW (Fig. 3B), as well as the CheV1 coupling protein (Fig. 3C). Taken together, these data suggest that the TlpD membrane association is indirect, e.g., that TlpD itself does not interact with the membrane and instead relies on the chemotaxis signaling and chemoreceptor proteins.

FIG 3.

TlpD associates with the membrane fractions via interactions with chemoreceptors and chemotaxis signaling proteins. (A) Top, TlpD subcellular localization in wild-type strains and strains lacking H. pylori chemoreceptors, assessed by immunoblotting with anti-TlpA22. W, whole cells; S, soluble; and M, membrane. The positions of the H. pylori chemoreceptors TlpA, TlpB, TlpC, and TlpD are shown at the right. Bottom, control blot stripped and reprobed with an anti-CheV1 antibody that additionally recognizes several cytoplasmic proteins (arrowheads). (B) Top, TlpD subcellular localization in strains lacking chemotaxis signaling proteins CheA and CheW, assessed as described in panel A. Bottom, control blot as described in panel A. (C) TlpD subcellular localization in strains lacking chemotaxis signaling proteins CheV1, CheV2, and CheV3, assessed as described in panel A. In all panels, the amount of protein loaded was normalized based on the Bradford assay and confirmed by staining the blots with DB71 (not shown).

TlpD drives a chemotaxis response to conditions generated by iron.

The specific parameters sensed by TlpD remain unclear. Prior work had shown that this chemoreceptor responds to conditions generated by blockage of the electron transport chain, but the exact TlpD signal is unknown (2). We thus sought to determine additional conditions that are sensed by TlpD. For this approach, we employed the agarose-in-plug bridge spatial gradient-based chemotaxis assay described by Yu and Alam (35) to test many compounds, including metals, amino acids, and organic acids. We detected a positive response with FeCl₃. Specifically, we observed that wild-type H. pylori formed a sharp band near the edge of a FeCl₃ precipitate, allowing for the bacteria to respond to gradients emerging from the source (Fig. 4). In this assay, there were two features. The first was the sharp bacterial band (indicated by arrows in Fig. 4), which was dependent on chemotactic motility, while the other was a clearing zone, which was independent of chemotactic motility (Fig. 4). We therefore focused on the sharp band as an indicator of chemotaxis. Mutants lacking tlpD did not form the sharp bacterial band, whereas other single-receptor mutants lacking tlpA, tlpB, or tlpC formed sharp bands similar to the one seen with wild-type cells. No band formed in response to MnCl₂, suggesting that the cells respond to the iron and not the chloride portion of FeCl₃. We considered the possibility that tlpD mutants are generally defective for chemotaxis but concluded this is not the case because they retain nearly normal soft-agar migration (Fig. 1B) and retain the ability to respond chemotactically to acid (34). We determined that iron is released from the immobilized FeCl₃ in this assay, using inductively coupled plasma mass spectrometry analysis of the medium around the FeCl₃. We found levels that were 2.5-fold higher than that of background, equivalent to 92 nM. We do not know the form of this iron, as the medium contains FBS proteins, so the iron may be chelated by some of these. Additionally, the medium surrounding the iron has a pH of ∼5.5 (data not shown), because iron interacts readily with water to form iron hydroxide to liberate protons. The TlpB pH chemoreceptor, however, does not appear to play a role under these conditions. Taken together, these data show that iron is released from FeCl₃ in our chemotaxis assay, causes H. pylori to form a chemotaxis-dependent band near but not directly adjacent to the immobilized iron, and that TlpD is the only receptor required for this response. These data, however, do not indicate whether the TlpD-dependent chemotactic response to FeCl₃ is an attractant or repellent response. The position of the bacterial band a bit away from the iron source, however, suggests that iron may have some repellent properties, as observed previously for other bacteria in their spatial response to oxygen (39).

FIG 4.

TlpD mediates tactic response to iron in the agarose-in-plug bridge assay. Modified agarose-in-plug bridge assays were carried out using H. pylori strain SS1 and its isogenic mutants. The arrows mark chemotactic bacterial rings, and asterisks mark crystallized FeCl₃ or MnCl₂ in an agarose matrix. Clearing around the plugs is nonspecific, as it appears independently of chemotaxis or motility. Images are representative of the following number of replicates: wild-type, n = 17; cheW, n = 1; cheA, n = 4; cheY, n = 4; tlpA, n = 2; tlpB, n = 4; tlpC, n = 2; tlpD, n = 14; motB, n = 1.

Iron acts as a TlpD-detected repellent.

To determine whether iron was acting as a chemotaxis attractant or repellent, we employed a well-established microscopic assay for H. pylori in which the number of bacterial reversals is monitored in response to specific conditions (2, 33, 34). Repellents trigger high numbers of bacterial reversals, while attractants suppress reversals; changes in flagellar rotation have been used in many instances as an accurate indicator of a chemotaxis response (22, 33, 40, 41). This assay has been used to monitor an H. pylori repellent response to acid and autoinducer 2 (AI-2); these two conditions cause a 1.5- to 2-fold increase in bacterial reversals (33, 34). Wild-type H. pylori showed a statistically significant increase in the frequency of direction changes upon FeCl₃ exposure, with an apparent half-maximal effective concentration (EC₅₀) for this response of 100 μM (Fig. 5A). The magnitude of this response, a 1.5-fold increase, is similar to that reported for other H. pylori repellents, including acid and AI-2 (33). A concentration of 100 μM of Fe(II) iron sulfate evoked a response similar to that of 100 μM Fe(III)Cl₃ (data not shown). TlpD and chemotaxis signaling were required for this iron-triggered increase in reversals (Fig. 5B). TlpD-only H. pylori retained the ability to respond to 100 μM FeCl₃ as a repellent, although the change effected by iron was muted compared to that of the wild type and complicated by the fact that this strain already displays a high frequency of reversals (Fig. 5B). Taken together, these results suggest that TlpD mediates a repellent response to iron exposure.

FIG 5.

H. pylori displays a repellent chemotaxis response to oxidative stress. (A) WT mG27 grown overnight in Ham’s-10 was exposed to various concentrations of FeCl₃ immediately before being filmed and tracked. Direction changes were tracked over a 3-s window. (B) mG27 strains grown overnight in Ham’s-10 were exposed to 100 mM FeCl₃, 50 μM dipyridyl, or 1 mM hydrogen peroxide prior to being filmed and tracked. For dipyridyl exposure, cultures were incubated with 50 μM dipyridyl for 15 min prior to addition of FeCl₃ or H₂O₂. Direction changes were tracked over a 3 s window. (C) mG27 strains were incubated for 15 min in the presence of either 10 μg/ml metronidazole or 10.5 μM paraquat, swimming behavior was recorded, and direction changes were tracked over a 3-s window. The error bars represent the standard error of the mean. *, P < 0.05; **, P < 0.01, compared to the wild type using Student's t test.

TlpD mediates a repellent response to conditions that promote oxidative stress.

While iron is an essential nutrient, it is also toxic because it can react with cellular hydrogen peroxide via Fenton-mediated chemistry to generate reactive oxygen species (ROS), including hydroxyl radical and hydroxide ion (42). Given that iron acted as a repellent, we examined whether ROS might also be a chemotaxis signal for H. pylori. We employed the same microscopic swimming assay and assessed the response of H. pylori to chemicals shown to generate ROS in H. pylori: hydrogen peroxide (43), paraquat (44), and metronidazole (45).

Hydrogen peroxide exposure at 1 mM evoked a repellent response that was TlpD dependent and was retained in a TlpD-only strain (Fig. 5B). A similar response was noted for lower concentrations of hydrogen peroxide at 50 μM, 100 μM, and 500 μM (data not shown). To assess the contribution of Fenton-mediated chemistry in generating conditions sensed by TlpD, H. pylori was incubated with the membrane-permeable iron chelator dipyridyl prior to exposure to hydrogen peroxide, using an exposure shown previously to result in intracellular affects within this time frame (46). Dipyridyl pretreatment abolished the TlpD-mediated response to hydrogen peroxide (Fig. 5B). These data suggest that TlpD mediates a chemotaxis response to ROS, and that cytoplasmic iron pools contribute to this response.

To further confirm whether TlpD mediates a response to ROS, we next tested the ability of two superoxide generators, metronidazole and paraquat, to generate TlpD-dependent repellent responses. H. pylori strains were incubated in the presence of either metronidazole or paraquat, and swimming behavior was recorded and tracked. Both metronidazole and paraquat exposure induced a TlpD-dependent increase in direction change frequency (Fig. 5C), which was insensitive to the addition or chelation of iron (data not shown). The diluted metronidazole and paraquat did not affect the pH of the medium with which H. pylori was tracked, removing the possibility that the induced repellent responses involved pH sensing. Taken together, these data suggest that TlpD responds to ROS generated by Fenton chemistry as well as superoxide.

TlpD is expressed in all H. pylori strains.

TlpD is critical for animal infection (7, 8), but previous reports suggested that the expression of TlpD, TlpA, and TlpC varied between H. pylori isolates. Specifically, G27 strains express only three chemoreceptors (Fig. 1), and previous studies had suggested that H. pylori strains can lack either tlpC or tlpD (47) or have an interrupted tlpA gene (48). It thus was not clear whether TlpD was present, and therefore potentially utilized, by all H. pylori strains. We therefore examined the expression of TlpD across different H. pylori strains using Western blotting. Of 14 strains examined, all strains expressed TlpD (Fig. 6). All strains also expressed TlpB, but the amount varied substantially (Fig. 6), consistent with a report showing that TlpB expression varies due to a G repeat in the 5′ untranslated region (49). Several strains lacked either TlpC (28%) or TlpA (14%), but no strains lacked both (Fig. 6). The ubiquity of TlpD suggests that it plays critical roles in human infection, in line with its importance in animal models.

FIG 6.

TlpD is expressed by all examined H. pylori strains. Chemoreceptor content was analyzed from total cell lysate of the indicated strains using Western blotting with the TlpA22 antibody that recognizes the conserved domain present in all H. pylori chemoreceptors (18). Marker sizes are given in kilodaltons on the left. The migration positions of TlpA, TlpB, TlpC, and TlpD are indicated at the right. Western blots are representative of at least 3 replicates for each sample.

DISCUSSION

We report here that the cytoplasmic chemoreceptor TlpD assembles a chemotaxis signaling complex at the pole independently of the transmembrane chemoreceptors, suggesting that TlpD has all the information required for polar localization and signaling complex formation. We furthermore determined that TlpD mediates a repellent response to compounds that generate ROS. Oxygen and ROS are ubiquitous in the stomach, where H. pylori resides, arising from both epithelial and immune cells (31, 50). Our findings thus suggest a model in which TlpD is used to help H. pylori avoid regions with high ROS, as discussed below.

Previous studies had not addressed whether cytoplasmic chemoreceptors are able to function on their own (1). Here, we show clearly that TlpD is capable of independent function. Specifically, we report that TlpD-only strains can mediate chemotaxis in the soft-agar assay in the absence of transmembrane chemoreceptors, and TlpD-only strains furthermore retain TlpD-mediated chemotaxis responses to ROS. Previous work had shown that TlpD-only strains in a different H. pylori background retained the ability to chemotactically respond to electron transport chain inhibition (2), in line with what we report here. We also report that TlpD is sufficient for the formation of polar complexes that contain multiple chemotaxis signaling proteins. Taken together, this work provides strong evidence that TlpD is able to carry out all the necessary functions of a chemoreceptor.

Recent work has visualized cytoplasmic chemoreceptors inside intact cells. Briegel and colleagues (51) identified clusters of cytoplasmic chemoreceptors in Rhodobacter sphaeroides and Vibrio cholerae that were positioned adjacent to and separate from the transmembrane receptors. These structures showed high similarity in many respects to those produced by transmembrane chemoreceptors, in that the clusters contained trimers of chemoreceptor dimers organized in 12-nm hexagonal arrays (51). These studies, however, were done in wild-type cells that possess multiple chemoreceptors. Our work shows that a soluble chemoreceptor can be sufficient to organize a signaling complex, consistent with the observation that some prokaryotes have only soluble chemoreceptors (1). Polar localization of chemoreceptors is achieved using a variety of mechanisms, including stochastic self-assembly, the use of polar landmark proteins, and ParA orthologues (52). Presumably, TlpD can utilize one or more of these mechanisms, although it is not known which operates in H. pylori.

TlpD exists in both soluble and membrane-associated forms, as reported previously (2), and similarly to other cytoplasmic chemoreceptors (1). We found that the association of TlpD with the membrane depends on the transmembrane chemoreceptors CheA, CheW, and CheV1. This finding suggests that TlpD is normally part of the H. pylori chemoreceptor signaling complex, interacting with the chemoreceptors via the signaling proteins. Studies in other microbes have also found a role for CheA and CheW in chemoreceptor organization, but the role of the CheV proteins was not well understood. CheV of B. subtilis promotes interchemoreceptor clustering of the 10 B. subtilis chemoreceptors (53), suggesting that it acts similarly to H. pylori CheV1. Of the cytoplasmic chemoreceptors reported to localize to the cell pole, some remain primarily colocalized with transmembrane chemoreceptors (54–56), and others cycle between a diffuse cytoplasmic distribution and a defined polar location (3). It appears that a large fraction of TlpD is soluble in wild-type H. pylori strains. It is not clear whether TlpD exists in equilibrium between soluble and membrane-associated states or if membrane association is affected by any signals.

We report here that TlpD mediates a repellent response to conditions that promote oxidative stress, including being in the presence of iron, hydrogen peroxide, paraquat, and metronidazole. Iron was the initial TlpD chemotaxis signal we identified. Chemotactic responses to iron, mediated by TlpD in the agarose-in-plug bridge assay, were confounded by the fact that these tactic responses could be interpreted as repellent, attractant, or mixed attractant and repellent. To clarify the nature of the response, we studied the swimming behavior of H. pylori in the presence of iron and noted that the behaviors elicited were consistent with a repellent response. The correlation of swimming behavior and flagellar motor direction changes to chemotactic responses has been used in numerous studies of bacterial chemotaxis across many species (22, 33, 40, 41), and it leant confidence to our conclusion that H. pylori was responding to extracellular iron as a repellent. It is possible that iron is a repellent at high concentration and an attractant at lower concentrations, but we could not detect any such an attractant response even at low concentrations of iron (Fig. 5A).

Because iron acted as a repellent, we focused our attention on its toxic effects, namely, the generation of ROS. Extracellular iron has been documented to rapidly enter the H. pylori cytoplasm (57, 58). Specifically, within 5 min of treatment with either iron(III) or iron(II), H. pylori had significantly higher levels of intracellular iron than bacteria that did not receive excess iron (57, 58), suggesting that our exposure here would result in intracellular iron increases. Excess iron has been shown carry out Fenton chemistry with endogenous hydrogen peroxide to generate hydroxyl radical and hydroxide anion. Thus, one hypothesis is that iron exposure causes cytoplasmic ROS that are, in turn, the signals sensed by TlpD. Consistent with the idea that elevated cytoplasmic iron propagates ROS, others have found that iron causes a transcriptional response in H. pylori genes coding for ROS-detoxifying enzymes (59). In line with this hypothesis, we found that hydrogen peroxide similarly triggered a repellent response, and this response was dependent on free iron (Fig. 5). Iron-ROS connections have additionally been noted in that oxidative stress increases the free cytoplasmic iron pool in H. pylori (60). Further support for the idea that TlpD senses ROS came from analysis of other compounds that generate ROS, paraquat and metronidazole, which similarly resulted in a TlpD-mediated repellent response. These data on paraquat and metronidazole suggest that TlpD can respond to superoxide and the hydroxyl radical generated by Fenton chemistry. Although these stresses may be linked (61), TlpD appears to respond independently to the two oxidative stresses in our analyses, because iron chelation did not affect the response to superoxide. Therefore, our results support a model in which TlpD detects several forms of cytoplasmic oxidative stress and leads to a repellent chemotaxis response.

Previous work showed that TlpD additionally mediates a repellent response to electron transport chain inhibitors (2). Interestingly, electron transport chain inhibition is known to generate both hydrogen peroxide and superoxide (61, 62), so it is possible that electron transport inhibitors and ROS affect chemotaxis via the same mechanism. Furthermore, Behrens and colleagues (8) report that tlpD mutants are more sensitive than other mutants to oxidative damage, providing an additional connection between TlpD and oxidative stress.

While we know that TlpD is necessary for the ROS response, we do not yet know whether it is sufficient. Chemoreceptors are known to sense signals either directly or via interacting proteins. TlpD has a reported protein interaction partner, HP0697, a predicted subunit of acetone carboxylase (63). This or other unidentified proteins may be required for TlpD's ROS response. Alternatively, TlpD may sense ROS directly, or possibly respond to a small molecule whose concentration changes during oxidative stress. Other proteins use redox-active cofactors, such as FADH, FMN, or iron, to sense alterations in the redox state. These cofactors are typically bound by domains, such as the PAS domain, which are located in the N-terminal direction to the MA domain (1). TlpD has no identifiable domains that would bind FADH or FMN at either its N or C terminus, but it has been reported to interact with iron (9). There are no identifiable domains in the TlpD N-terminal region at all (9), but it does possess a CZB domain at its C terminus, which binds zinc via histidine and cysteine residues (9). As some of us have noted previously (9), other ROS-sensing proteins use cysteine as a way to sense redox status, so this mechanism may operate in TlpD (9).

TlpD is critical during colonization in animal models (7, 8), particularly in the first weeks of infection. H. pylori is exposed to ROS produced in the stomach by both epithelial and phagocytic cells (31, 50, 64). The ability to survive ROS appears to be critical during H. pylori colonization, as antioxidant enzymes are highly expressed by this microbe, and mutants lacking them are attenuated in colonizing animal hosts (43, 65). Others have documented that epithelial cells of the stomach release hydrogen peroxide that exposes the related Helicobacter felis to ROS, consistent with the idea that gastric Helicobacter spp. are regularly exposed to ROS (50). Additionally, there is a recently appreciated radial oxygen gradient emanating from the tissue to the lumen of the gastrointestinal tract. While not yet measured in the stomach, the intestinal oxygen gradient is extremely steep, ranging from a tissue level of 160 mm Hg (5.25%) to <1 mm Hg (30). These and other findings suggest that near-tissue bacteria are exposed to higher oxygen levels than those in the lumen (29, 66). Indeed, our swimming analysis suggests that room oxygen (21%) acts as a modest H. pylori repellent, because dipyridyl decreased the basal bacterial reversal frequency (Fig. 5B), which could be explained by prior reports suggesting that H. pylori experiences oxidative stress in atmospheric oxygen (60). ROS affect H. pylori cytoplasmic iron concentrations (67), suggesting that the oxidative stress experienced by H. pylori is sensitive to oxygen levels in which the bacterium resides. H. pylori has been measured to localize within 15 μm of the epithelial surface (68), and it may maintain this distance to limit the oxidative stress it experiences from oxygen and/or reactive oxygen species emanating from the epithelial surface.

In summary, we report that the cytoplasmic chemoreceptor TlpD possesses the key features of a bona fide chemoreceptor: the ability to drive a chemotaxis response and to organize a chemotaxis signaling complex at the pole. TlpD mediates a repellent response to cytosolic ROS. Our work thus adds substantially to our understanding of cytoplasmic chemoreceptors by clearly showing that they independently retain properties, such as polar localization and signaling complex organization, and can promote responses to specific signals.