|Home | About | Journals | Submit | Contact Us | Français|
Here, we report the first comprehensive study of Bartonella henselae gene expression during infection of human endothelial cells. Expression of the main cluster of upregulated genes, comprising the VirB type IV secretion system and its secreted protein substrates, is shown to be under the positive control of the transcriptional regulator BatR. We demonstrate binding of BatR to the promoters of the virB operon and a substrate-encoding gene and provide biochemical evidence that BatR and BatS constitute a functional two-component regulatory system. Moreover, in contrast to the acid-inducible (pH 5.5) homologs ChvG/ChvI of Agrobacterium tumefaciens, BatR/BatS are optimally activated at the physiological pH of blood (pH 7.4). By conservation analysis of the BatR regulon, we show that BatR/BatS are uniquely adapted to upregulate a genus-specific virulence regulon during hemotropic infection in mammals. Thus, we propose that BatR/BatS two-component system homologs represent vertically inherited pH sensors that control the expression of horizontally transmitted gene sets critical for the diverse host-associated life styles of the alphaproteobacteria.
The alphaproteobacterium Bartonella henselae is a globally distributed zoonotic pathogen that naturally infects cats by causing an asymptomatic intraerythrocytic infection. Transmission to humans can result in various clinical manifestations, including cat scratch disease in immunocompetent patients and bacillary angiomatosis-peliosis as distinct forms of vascular tumor formation characteristically seen in immunocompromised patients (14). The cat flea (Ctenocephalides felis) is responsible for cat-to-cat transmission (12), whereas transmission to humans is caused by cat scratches and bites. Notably, B. henselae and the closely related human-pathogenic species Bartonella quintana and Bartonella bacilliformis are unique in the bacterial kingdom for their capacity to cause proliferation of the human vasculature (16). Thus, cultured human endothelial cells (HEC) represent a valid in vitro model to study the unique interactions of B. henselae with the human vasculature that culminate in the formation of vascular tumors (17, 19).
Type IV secretion systems (T4SSs) are multicomponent transporters crucial for the pathogenesis of many Gram-negative bacteria (e.g., Helicobacter, Legionella, Bordetella, Brucella, Agrobacterium, and Bartonella). Bacteria use these systems to deliver bacterial effector proteins or DNA-protein complexes into the cytoplasm of their host cells in order to subvert their cellular function (13). For B. henselae, the VirB T4SS and the Bartonella effector proteins (Beps) mediate most of the cellular phenotypes associated with B. henselae infection of HEC (46). The translocation of the Beps into HEC mediates (i) a massive rearrangement of the actin cytoskeleton, resulting in the formation of an invasome that internalizes large bacterial aggregates; (ii) nuclear factor kappa B-dependent proinflammatory activation; and (iii) the inhibition of apoptotic cell death (46, 49).
The importance of the VirB T4SS for Bartonella pathogenicity was also demonstrated by a large-scale signature-tagged mutagenesis screen in Bartonella tribocorum to identify essential genes for the colonization of its rat reservoir host (42). In the same screen, the batR gene encoding the putative response regulator of the predicted BatR/BatS two-component system (TCS) was also identified and thus is likely to be involved in the transcriptional regulation of Bartonella pathogenicity. Bacterial two-component regulatory systems are a key element of the transcriptional regulatory circuits that enable organisms to elicit an adaptive response to changes in their host-associated microenvironments and to mount the appropriate response to successfully establish mutualistic or pathogenic interactions with their respective hosts (6, 36, 37, 51). The closest homologs of BatR/BatS in the alphaproteobacteria were shown to be essential for effective host interaction. For the facultative intracellular pathogen Brucella abortus, the BvrS/BvrR TCS is essential for virulence and is responsible for extensive cell envelope modulation, including the upregulation of the outer membrane proteins Omp3a/Omp3b (25, 32). The ChvG/ChvI TCS of the plant pathogen Agrobacterium tumefaciens is essential for plant tumor induction (8, 33) and controls the expression of acid-inducible genes involved in virulence (29, 31). Similarly, the ExoS/ChvI TCS of the legume-nodulating symbiont Sinorhizobium meliloti plays an essential role in the establishment of symbioses with its host by regulating the production of succinoglycan (11) and was also shown to regulate the expression of genes required for flagellar assembly (57, 60). In a recent publication, Chen et al. identified three direct transcriptional targets of a constitutive active form of ChvI (9).
To date, our knowledge of the coordinated response orchestrated by B. henselae to successfully invade and colonize HEC is sparse, as is our understanding of the environmental signal(s) perceived by the bacteria during host cell interaction. Additionally, the factors implicated in the regulation of the B. henselae adaptive response to host cell interaction remain elusive. To address these important questions, we have used HEC as a model to examine the expression profile of B. henselae during host cell infection, with a specific focus on the BatR/BatS TCS. The data obtained from the characterization of a batR deletion mutant reveal that BatR is essential for B. henselae pathogenicity. By transcriptional-profiling analysis, we demonstrate that batR is required for the upregulation of a critical cluster of genes regulated during the infection of HEC, including genes encoding the VirB T4SS and its cognate secreted effectors (Beps). Evidence is provided that BatR/BatS constitute a functional TCS, and furthermore, that BatR binds directly to the promoters of the virB operon and the bepD gene. BatR is thus the first regulatory protein shown to be directly involved in the regulation of these key pathogenicity factors in Bartonella. Moreover, we show that the BatR/BatS TCS is activated in a neutral pH range (pH 7.0 to 7.8) with an optimum at the physiological pH of blood (pH 7.4). Finally, we show that despite the evolutionary conservation of both the histidine kinase sensor protein and the response regulator across alphaproteobacterial species, a large subset of the BatR/BatS regulon is specific to the genus Bartonella and/or has recently been horizontally transferred into the genus.
B. henselae and Escherichia coli strains were grown as previously described (49). Plasmids were introduced into B. henselae by conjugation from E. coli using three-parental mating. Table Table11 lists all strains and plasmids used in this study. Table S4 in the supplemental material lists all oligonucleotides used in the study. The endothelial cell line Ea.hy926, resulting from a fusion of human umbilical vein endothelial cells (HUVEC) and the lung carcinoma cell line A549 (20), was cultured as reported previously (27).
The batR in-frame deletion mutant of RSE247 was generated by a two-step gene replacement procedure, as described previously (46, 48), using the suicide plasmid pAB001. The 5′ 939-bp and 3′ 706-bp flanking regions were amplified with primers prAB007, prAB008, prAB009, and prAB010 and combined by megaprime PCR. The resulting fragment was digested by XbaI and inserted into the corresponding site of pTR1000, yielding pAB001. The ΔbatR strain contains an in-frame deletion of 678 bp in batR, resulting in a 45-bp cryptic open reading frame composed of 5′ and 3′ sequences of batR. To generate the batR complementation plasmid pbatR, a 782-bp fragment containing batR and the Shine-Dalgarno sequence of pPG110 was amplified using primers prIT009 and prDT032, ligated in pCR-blunt II TOPO, digested with BamHI, and inserted into the corresponding site of pCD341.
The 366-bp intergenic upstream region of the virB2 gene (47) and the 333-bp intergenic upstream region of the bepD gene were amplified using primers prIT011-prIT012 and prIT015-prIT016, respectively. The terminal EcoRI and BamHI sites were used to insert the fragment in the corresponding sites of pCD366, yielding pPvirB-gfp and pPbepD-gfp. The truncated promoter probe vectors were generated by the same strategy using the primers listed in Table S4 in the supplemental material.
The C-terminally His6-tagged version of BatR (His6-BatR) was generated by amplifying the BatR coding sequence using primers prDT007 and prDT008. Using the flanking NdeI and BamHI sites, the fragment was cloned into the corresponding sites of the expression vector pET15b(+) (Novagen), resulting in pDT009. The glutathione S-transferase (GST)-tagged version of the kinase domain of BatS (GST-BatSΔ1-332) was constructed by PCR amplification using oligonucleotides prDT026 and prDT027. Using the flanking BamHI and EcoRI sites, the fragment was cloned into the corresponding sites of the expression vector pGEX-2T (GE Healthcare), resulting in pDT020. The C-terminally His6-tagged version of VirB5 (His6-VirB5) was generated by amplifying VirB5 coding sequence using primers prSH003 and prSH004. Using the flanking NdeI and BamHI sites, the fragment was cloned into the corresponding sites of the expression vector pET15b(+) (Novagen), resulting in pSH007. The C-terminally His6-tagged version of BepD (His6-BepD) was generated by amplifying BepD coding sequence using primers prPG165 and prPG166. Using the flanking NdeI and BamHI sites, the fragment was cloned into the corresponding sites of the expression vector pET15b(+) (Novagen), resulting in pPG124.
EA.hy926 cells were grown to confluence in Dulbecco's modified Eagle medium (DMEM) with Glutamax (Gibco Invitrogen) supplemented with 10% fetal calf serum (FCS) (Gibco Invitrogen) in 150-cm2 cell culture flasks in a humidified atmosphere at 37°C and 5% CO2. One hour before infection, the cells were washed with DMEM. The bacteria were grown for 48 h on Columbia agar plates containing 5% defibrinated sheep blood (CBA plates) and harvested in phosphate-buffered saline (PBS) (pH 7.4). Four to 6 confluent 150-cm2 flasks were infected at a multiplicity of infection (MOI) of 200. The infected cells were incubated in a humidified atmosphere at 37°C and 5% CO2. To stop the infection, the medium was collected and 9 ml PBS plus 0.5% saponin (35°C) was added to the flasks, followed by 5 min of incubation at 35°C and 5% CO2. The infected cells were harvested using a cell scraper (Corning), pooled with the medium, and centrifuged for 5 min at 4,800 × g and 4°C in a swinging-bucket rotor. The pellet was washed with PBS and resuspended in PBS. After addition of 1/10 volume of ethanol-phenol (95:5) and centrifugation, the supernatant was removed and the pellet was snap-frozen in liquid nitrogen and stored at −70°C until RNA extraction.
RNA was isolated using modified hot-phenol extraction, including DNase I digestion combined with an RNA cleanup using the RNeasy Mini Kit (Qiagen). The frozen bacterial pellet was resuspended in 600 μl Tris-EDTA (TE) buffer (pH 8.0) containing 0.5 mg/ml lysozyme. Lysis was completed by adding 1/10 volume 10% SDS and 1/10 volume 3 M sodium acetate (pH 5.2). RNA was extracted by addition of 1 volume water-saturated phenol and incubation at 64°C for 6 min. The sample was set on ice for 5 min and centrifuged at 15,000 × g for 10 min, and the aqueous phase was subjected to phenol-chloroform-isoamyl alcohol (25:24:1) extraction followed by two chloroform extractions. RNA was precipitated by the addition of 2.5 volumes ethanol and storage at −70°C for 16 h. RNA was pelleted by centrifugation for 1 h at 15,000 × g and 4°C, washed with 1 ml 70% ethanol, and air dried. The pellet was resuspended in 80 μl water, incubated on ice for 10 min, and heated to 64°C for 10 min. DNase treatment was carried out by addition of 10 μl 10× DNase I buffer (400 mM Tris-Cl, pH 7.4, 40 mM MgCl2), 30 U DNase I (GE Healthcare), 2 μl RNA guard (GE Healthcare), 5 μl water, and incubation at 37°C for 30 min. RNA was cleaned up using the RNeasy Mini Kit (Qiagen) and eluted in 50 μl water. The integrity of total RNA was verified on an Agilent 2100 Bioanalyser using the RNA 6000 Nano LabChip kit (Agilent Technologies). The purified RNA was used directly for downstream application or stored at −70°C after precipitation with 2.5 volumes ethanol and 1/10 volume 3 M sodium acetate, pH 5.2.
A PCR-based DNA microarray similar to that previously described (30) was used for genome-wide transcriptional profiling. Further details are provided in the supplemental material.
To validate the microarray data and to further characterize gene regulation, quantitative reverse transcription (qRT)-PCR was applied. Total bacterial RNA was isolated as described above. Total RNA (1 μg) was reverse transcribed using random primers (Promega) and Superscript II reverse transcriptase (Invitrogen). SYBR green I quantitative RT-PCR was performed as previously described (19) using rpsL expression as a reference. Table S5 in the supplemental material lists all primers used for quantitative PCR in this study.
The His-tagged version of BatR (His6-BatR from pDT009) and the GST-tagged cytoplasmic fragment of BatS (GST-BatSΔ1-332 from pDT020) were induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 27°C for 3 h. All protein purification steps were performed at 4°C. Cells expressing His6-BatR were lysed using a French press and purified with 2 ml His-Select Nickel Affinity Gel (Sigma-Aldrich) according to the manufacturer's instructions. The eluate was dialyzed overnight against 2 liters of buffer A (50 mM Tris [pH 8.5], 300 mM NaCl, 2 mM dithiothreitol [DTT], and 10% glycerol). Cells expressing GST-BatSΔ1-332 were lysed by sonication and purified with 2 ml glutathione-Sepharose (GE Healthcare) according to the manufacturer's instructions. The protein was further purified by ion-exchange chromatography using a Mono Q anionic-exchange column preequilibrated with loading buffer (100 mM phosphate buffer, pH 6.5, 50 mM NaCl, 10% glycerol, and 2 mM DTT). The protein was eluted on a fast protein liquid chromatography (FPLC) system (Äkta purifier; GE Healthcare), using a concentration gradient of elution buffer (100 mM phosphate buffer, pH 6.5, 50 mM NaCl, 10% glycerol, 2 mM DTT, and 1 M NaCl). For polyclonal antibody serum production, His6-VirB5 was expressed from SHE118 with 3 h of induction using 1 mM IPTG at 37°C and purified as His6-BatR. His6-BepD was expressed from PGB31 with 4 h of induction using 1 mM IPTG at 37°C and purified as His6-BatR. Purified His6-VirB5, His6-BepD, and His6-BatR were loaded on preparative SDS-PAGE. After electrophoresis, the gel was stained with 1 M KCl, the band corresponding to the size of the protein was cut out. and the gel pieces were sent for immunization (Laboratoire d'Hormonologie, Marloie, Belgium).
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting for the detection of BatR, VirB5, and BepD proteins were performed as follows. B. henselae wild type or the batR mutant was harvested after 48 h of growth on CBA plates or HEC infection (MOI, 200). To recover bacteria from infected HEC, the cells were pretreated with PBS-0.5% saponin (35°C) for 5 min before being harvested. The cells were washed and resuspended in PBS to an optical density at 600 nm (OD600) of 16 and mixed with an equal volume of 2× Laemmli buffer. Ten microliters of each sample was separated by 12% SDS-PAGE and transferred to a nitrocellulose membrane (Hybond-C Extra, GE Healthcare). The immunoblot was developed with polyclonal rabbit sera raised against recombinant BatR (1:20,000), BepD (1:75,000), or VirB5 (1:50,000), followed by a 1:15,000 dilution of a goat anti-rabbit horseradish peroxidase-conjugated secondary antibody (GE Healthcare).
Autophosphorylation of the purified GST-BatSΔ1-322 protein was assessed in 46 μl reaction buffer (800 pmol GST-BatSΔ1-322, 50 mM Tris [pH 8.5], 300 mM NaCl, 2 mM DTT, 10% glycerol, 5 mM MgCl2, 0.5 mM ATP, and 2 μl [γ-32P]ATP, 110 TBq/mmol) and incubated at 26°C. After various intervals, 5 μl of the reaction mixture was withdrawn and mixed with 3 μl SDS loading buffer. All time points were separated by SDS-PAGE on 12% acrylamide gels. Radioactivity was quantified with a Molecular Dynamics Typhoon 8600 phosphorimager after 3 h of exposure. The gel was subsequently stained with Coomassie blue. Phosphotransfer between GST-BatSΔ1-322 and His6-BatR was assessed as follows: 1 nmol of purified GST-BatSΔ1-322 was allowed to autophosphorylate for 1 h as described above before an equal molar amount of purified His6-BatR was added to the reaction mixture (final volume, 100 μl) and incubated at 26°C. After various intervals, 10 μl of the reaction mixture was withdrawn and treated as described above.
Radiolabeled probes were generated by PCR in the presence of [α-32P]dATP or [α-33P]ATP and purified using a nucleotide removal kit (Qiagen). For each probe, a parallel reaction was performed in the absence of radioactive dATP, and the DNA concentration was determined using a NanoDrop ND-100 spectrophotometer (Thermo Scientific). The PCR primers used to generate the different probes are listed in Table S4 in the supplemental material. Binding reactions were performed in a total volume of 20 μl of buffer A (50 mM Tris [pH 8.5], 300 mM NaCl, 2 mM DTT, and 10% glycerol) in the presence of 1 μg of poly(dI-dC) and 2 to 4 fmol radiolabeled probe. The reaction mixtures were separated on a 5 to 8% polyacrylamide gel in 0.5% Tris-borate-EDTA buffer at 120 V for 1 h, and the gels were dried. Radioactivity was quantified with a Molecular Dynamics Typhoon 8600 phosphorimager. For competition with nonlabeled double-stranded oligonucleotides, the reaction mixture was supplemented with a 1,000 molar excess of annealed competitors.
Induction of expression from the virB and bepD promoters was measured as GFP fluorescence by using a FACSCalibur flow cytometer (Becton Dickinson) with excitation at 488 nm. B. henselae strains carrying a reporter plasmid were streaked from stock (−70°C) on CBA plates with 30 μg/ml kanamycin and grown in a humidified atmosphere at 35°C and 5% CO2 for 3 days, followed by restreaking on fresh CBA plates and growth for 48 h. The bacteria were resuspended in PBS, washed, diluted to a final OD600 of 0.008 in 1 ml of tester medium, and incubated in 24-well plates in a humidified atmosphere at 37°C and 5% CO2. To test the pH dependency of GFP induction of B. henselae strains containing pPvirB-gfp or pPbepD-gfp.IT011, M199 was reconstituted from 10× stock solution (Gibco Invitrogen), supplemented with 10% FCS, and buffered with sodium bicarbonate (0.3 g/liter to 8.3 g/liter) to cover a pH range between 6.3 and 8.1 (35°C).
Protein sequences of BatR and BatS and concatenated sequences of BatR/BatS were aligned by using ClustalW implemented in MEGA4 (52). Phylogenetic trees were inferred from the single protein or the concatenated alignments of BatR and BatS by the maximum-likelihood method with PhyML 3.0 (24) using the paralogous phosphate-sensing system PhoB/PhoR as the outgroup. An appropriate substitution model was selected by using the Akaike information criterion of ProtTest (1). We used the model LG + I + G + F for inferring the trees based on BatS, as well as on the concatenated sequences, and the LG + G model for the tree based on BatR.
The microarray data have been deposited in the microarray database at EBI under accession numbers A-MEXP-644 and A-MEXP-645 for the array design and E-MEXP-2322, E-MEXP-2323, and E-MEXP-2324 for experimental data.
To identify sets of B. henselae genes that are coregulated during human HEC infection, we performed transcriptional profiling using a PCR-based DNA microarray covering 92.3% of the B. henselae protein-encoding genes (1,373/1,488), designed as previously described (4). The results of a time course experiment of Ea.hy926 cell (HEC) infection with B. henselae in DMEM-10% FCS (DMEM) is presented as color-coded concentric circles in Fig. S1 in the supplemental material (circles 1 to 6). We performed a hierarchical gene tree analysis of the 95 genes that fulfilled our criteria for differential expression (see Table S1 and description in the supplemental material). The gene tree presented in Fig. Fig.11 forms four major clusters corresponding to distinct expression profiles: cluster 1 comprises genes displaying progressive but persistent upregulation during the course of infection; cluster 2 comprises genes displaying rapid upregulation after contact with HEC, followed by slower downregulation; cluster 3 comprises genes that become transiently upregulated, followed by downregulation below the initial expression level; and cluster 4 comprises genes that become persistently downregulated during the course of infection (see Table S1 in the supplemental material). Clusters 1 and 2 best matched the expected expression profile for an adaptive response. Strikingly, cluster 1 includes the majority of the 18 genes encoding the VirB T4SS (virB2 to -7, -10, and -11) and its cognate translocated effector proteins (bepC to -E and -G). Figure Figure22 illustrates the changes in expression at the virB-virD4-bep locus. None of the genes located directly upstream or downstream of the locus displayed differential expression, confirming the specificity of the regulation.
A validation of the microarray data for the genes virB4 and bepD by qRT-PCR is shown in Fig. S2A and B in the supplemental material. This experiment demonstrated congruent data for both methods, except that the fold change values of gene expression derived from the microarray experiment were compressed by a factor of 8 (bepD) to 10 (virB4). This indicates that the amplitude of differential expression might be generally underrepresented in the microarray data, as previously reported (40).
A previous study on B. tribocorum identified the batR gene, encoding the putative response regulator of the BatR/BatS TCS, as essential for the colonization of the rat as the natural reservoir host (42). Therefore, we were interested in assessing the involvement of the BatR/BatS TCS in the adaptive response to B. henselae during HEC infection. The batR gene displayed an expression profile similar to that of cluster 1, although it is not included in the cluster, as it did not pass the quantitative cutoff in fold expression. However, validation of the microarray data for batR by qRT-PCR (see Fig. S2C in the supplemental material) showed that the gene is indeed differentially regulated during HEC infection (fold change after 48 h = 11.3 [range, 6.8 to 21.1]). To test whether BatR might be a master regulator of the genes differentially regulated during HEC infection (Fig. (Fig.1),1), we generated an in-frame deletion in the batR gene and compared the gene expression profile of this batR mutant to the wild-type strain at 48 h of HEC infection in DMEM (see Fig. S1, circle 7, in the supplemental material). In total, we recorded 43 genes that fulfilled our quantitative and statistical criteria for differential regulation (see Table S2 in the supplemental material). Of these genes, 35 were downregulated and 8 were upregulated in the batR mutant. Figure Figure1B1B and Table S1 in the supplemental material display the expression profiles for the batR mutant in comparison to the wild-type for the 95 genes differentially expressed in the wild-type time course experiment. Strikingly, 21 of the 33 genes in cluster 1 (representing the genes gradually upregulated in the microarray time course experiment) are significantly downregulated in the ΔbatR strain, with an additional 6 genes that passed the statistical test but not the quantitative cutoff. These results indicate that BatR is indeed a master regulator for this set of genes, including the virB operon and the adjacent bep genes. The lack of a global effect of batR deletion on the regulation of the genes comprising clusters 2 to 4 suggests the involvement of additional regulators and/or the integration of additional stimuli during the infection process.
The qRT-PCR data for time points corresponding to the microarray time course experiment displayed in Fig. Fig.33 demonstrate that both virB4 (representative of the virB2-virB11 operon) and bepD (representative of genes encoding VirB/VirD4-translocated effector proteins) are strongly upregulated during HEC infection in DMEM, while these genes are only marginally upregulated in cell-free DMEM (Fig. 3A and B). These data could suggest that BatR-dependent upregulation of the genes is meditated by the presence of HEC. However, in a similar experiment replacing DMEM with medium 199-10% FCS (M199), we noticed prominent induction of the expression of these two genes in cell-free M199 (38-fold compared to 7-fold induction for virB4 and 21-fold compared to 3-fold for bepD in cell-free DMEM) (Fig. 3C and D). This induction was strictly dependent on BatR, since neither virB4 nor bepD was upregulated in the ΔbatR strain, either in the presence of HEC or in cell-free M199-10% FCS (Fig. 3E and F). These results indicate that contact with HEC is not a prerequisite for the induction of these genes. To determine whether the same set of genes would be differentially regulated between the wild-type and the ΔbatR strains during HEC infection in cell-free M199 as in DMEM, we determined the gene expression profile of the batR mutant compared to that of the wild-type strain after 48 h of M199 induction (see Table S2 in the supplemental material). Indeed, the data presented in Fig. Fig.1B1B and Fig. S1 in the supplemental material (circle 8) revealed a significant overlap between the 37 genes differentially regulated by BatR in the cell-free medium and the 35 differentially regulated during the infection of endothelial cells, with 22 genes fulfilling both our statistical and quantitative criteria for each condition. Moreover, the combined list of differentially regulated genes for these two conditions demonstrated a high degree of congruency, with 31 of the 43 genes differentially regulated during HEC infection showing a statistical difference in the M199 induction experiment and 29 of the 35 genes differentially regulated in the M199 induction significantly regulated in the HEC infection experiment. In summary, the regulatory cascade triggered by BatR was determined in two different sets of experiments using the wild-type and ΔbatR strains grown under both infectious and noninfectious conditions, with a high degree of congruency for both experiments. We thus defined the BatR regulon as the set of genes differentially regulated in either data set. These 58 genes (see Table S2 in the supplemental material) represent a minimal set of target genes in the BatR regulon that we feel highly confident about and consist of 46 genes that are downregulated and 12 genes that are upregulated in the ΔbatR strain.
To confirm that the global gene expression regulatory phenotype of the batR mutant is due to the deletion of the batR gene itself and is not the result of a coincidental secondary mutation, we complemented the mutant in trans by expressing batR from a low-copy-number plasmid under the control of the tac-lac promoter (ΔbatR-pbatR). qRT-PCR analysis of the BatR-regulated genes virB4 (Fig. (Fig.44 A) and bepD (Fig. (Fig.4B)4B) demonstrated complementation on the mRNA expression level for both genes during HEC infection and cell-independent induction in M199. The marginal downregulation measured for the expression of the virB4 and bepD genes in the ΔbatR strain grown on CBA plates compared to the wild-type (Fig. 4A and B, CBA) strongly supports the hypothesis that these genes are expressed only at a basal level under these conditions. Moreover, the marked induction of the virB4 and bepD genes in the complemented mutant grown on CBA plates in the presence of IPTG demonstrates that overexpression of BatR under these noninducing conditions is enough to bypass the sensory signal necessary for the induction of these genes in the wild type. Phenotypic restoration of the complemented strain was also demonstrated at the level of host cellular phenotypes provoked by the BatR-regulated genes. Based on the lack of upregulation of the entire virB-virD4-bep gene cluster in the ΔbatR mutant (Fig. (Fig.2C),2C), the mutant should be deficient for known T4SS-mediated phenotypes of infected HEC. Indeed, the batR mutant did not provoke the formation of the ring-like F-actin rearrangements that are a characteristic of invasome-mediated uptake of B. henselae (17, 41, 46), while the complemented ΔbatR-pbatR mutant strain triggered these cytoskeletal rearrangements as efficiently as wild-type bacteria (Fig. (Fig.4C).4C). Western blot analysis using polyclonal sera raised against VirB5 (a representative for the operon encoding the VirB2-VirB11 proteins) and BepD confirmed the requirement for BatR for the expression of these proteins, since neither VirB5 nor BepD was detectable in the batR mutant, either from bacteria grown on CBA plates or after HEC infection (Fig. (Fig.4D4D).
The genetic organization of the batR and batS genes in one locus and the genetic and biochemical data obtained for the closely related TCSs of B. abortus, A. tumefaciens, and S. meliloti suggest that BatS and BatR also constitute a functional TCS. To prove this hypothesis, we designed an in vitro experiment to demonstrate the histidine kinase activity of BatS and the phosphorelay between BatS and BatR. We could demonstrate that the purified cytoplasmic kinase domain of BatS (BatSΔ1-332) autophosphorylates in the presence of [γ-32P]ATP (Fig. (Fig.55 A, lane 1) and that a phosphotransfer event could take place between the phosphorylated BatSΔ1-332 and BatR (Fig. (Fig.5A,5A, lane 2). Conversely, no BatR phosphorylation occurred when BatSΔ1-332 was omitted from the reaction (Fig. (Fig.5A,5A, lane 3). A time-course experiment of BatS autophosphorylation showed a gradual increase of radiolabeled BatSΔ1-332 when incubated in the presence of [γ-32P]ATP, reaching 50% phosphorylation after about 15 min (Fig. (Fig.5B).5B). A similar time course for the phosphotransfer between prelabeled BatSΔ1-332 and BatR is shown in Fig. Fig.5C.5C. Maximal BatR phosphorylation was reached after 2 min, followed by a gradual loss of signal. BatR dephosphorylation was not caused by the presence of GST-BatSΔ1-332, since the same loss of signal was observed if the protein was removed by glutathione-Sepharose (data not shown). These results confirm that BatR/BatS constitute a bona fide TCS in B. henselae.
Based on its sequence homology with the OmpR/PhoB subfamily of response regulators, BatR is likely to bind DNA and act as a transcription factor. As BatR regulates the virB operon and the bepD gene, we decided to test whether BatR directly binds to the promoters of these genes. First, promoter-gfp fusions were created for the entire upstream intergenic region of the two genes. A series of truncated derivatives was created to delineate the promoter region, and finally, the direct binding of BatR to these refined promoter regions was assessed by EMSA.
Following initial growth on CBA plates, the wild-type and ΔbatR strains carrying pPvirB-gfp (bp −366 to +21) were cultivated for various periods of time in M199, and the GFP-mediated fluorescence intensities of individual bacteria were determined by flow cytometry (Fig. (Fig.66 A, −366 to +21). The results corroborated those obtained by microarray and qRT-PCR, showing upregulation during M199 induction and a strict BatR-dependent activity for the promoter. Deletion analysis of the virB promoter allowed us to map the BatR regulatory sequence between bp −366 and −279 (Fig. (Fig.6A).6A). Based on this information, we assessed direct binding of BatR to the virB promoter regions by EMSA and showed a BatR concentration-dependent mobility shift of the radiolabeled PvirB (Fig. (Fig.6B,6B, lanes 1 to 4). The EMSA results correlated with the results obtained with the promoter deletion analysis (Fig. (Fig.6A),6A), since we observed a similar shift using a shorter fragment covering the putative BatR binding region on PvirB (bp −366 to −134; Fig. Fig.6B,6B, lanes 9 to 11), but not when using a truncated fragment in which transcription activity was abolished (bp −153 to +13; Fig. 6A and B), confirming the specificity of the binding assay. To further confirm the specificity of BatR binding to PvirB and to better delineate the BatR binding motif, we designed a set of overlapping oligonucleotides spanning the shortest positive probe in our assay (bp −366 to −134) and used them as double-stranded unlabeled competitors in our EMSA experiments in the presence of the radiolabeled probe. Two sets of overlapping competitors prevented the binding of BatR to the radiolabeled probe (Fig. (Fig.5C,5C, competitors 3 and 4), whereas the other competitors did not show any interference in the assay (Fig. (Fig.6C,6C, competitors 1 and 2 and 5 to 8). These results further confirmed the specificity of the observed shift in the case of the virB promoter and allowed us to locate the BatR binding site between bp −306 and −277 relative to the virB ATG.
The same approach was applied to the bepD promoter (Fig. (Fig.7).7). Analysis of PbepD-gfp (bp −333 to +13) in wild-type and ΔbatR strains grown in M199 showed clear BatR-dependent GFP expression (Fig. (Fig.7A,7A, −333 to +13). Notably, only some of the wild-type bacteria exhibited an increase of fluorescence, with the remaining bacteria staying in the uninduced state, indicating a possible bistability phenotype, as described for the Salmonella enterica serovar Typhimurium type III secretion system (2). This biphasic induction pattern was also observed for the virB promoter (Fig. (Fig.6A,6A, −366 to +21 and −279 to +21), although most of the bacteria reached the active state under the tested conditions. EMSA demonstrated direct binding of BatR to PbepD (Fig. (Fig.7B),7B), in agreement with the promoter deletion analysis (Fig. (Fig.7A).7A). Moreover, the competition experiments with unlabeled double-stranded oligonucleotides allowed us to locate the BatR binding site between bp −176 and −137 relative to the bepD ATG.
It should be mentioned that the shifts observed in the autoradiograms presented in Fig. Fig.6B6B and and7B7B are not conventional shifts but are more likely to represent DNA-dependent protein oligomerization, since we were not able to resolve this DNA-protein complex into the polyacrylamide gel (the material stayed at the bottom of the well). This phenomenon of oligomerization has already been described for other members of the OmpR subfamily of response regulators (7, 10, 28, 54).
Nevertheless, the clear correlation between the deletion analyses of these two promoters, combined with transcriptional fusion to gfp and the EMSA experiments (including competition experiments with unlabeled oligonucleotides), demonstrates that both the virB operon encoding the VirB T4SS and the bepD gene encoding one of the cognate translocated effectors are direct targets of the BatR/BatS TCS.
We next addressed the question of the signal perceived and transduced by the BatR/BatS TCS. In Fig. 3A and B, there is a striking difference in the upregulations of the virB operon and the bepD gene between two different cell culture media (DMEM versus M199) in the absence of HEC. A plausible explanation for this phenomenon could be that it is the result of pH-dependent regulation, since the steady-state pH at 35°C/5% CO2 for DMEM is 7.8 versus 7.35 for M199 in the absence of cells, as a function of the bicarbonate buffer system (3.7 g/liter versus 2.2 g/liter bicarbonate, respectively). This inherent property of the two media was balanced in the presence of HEC by the progressive acidification observed during the culture of the cells, reaching steady-state pHs of 7.2 in M199 and 7.4 in DMEM.
We hypothesized that BatS may represent a pH sensor. To test this hypothesis, we used the wild-type and ΔbatR strains carrying pPvirB-gfp or pPbepD-gfp as biosensor strains and cultivated them for various periods in M199-10% FCS adjusted to different pHs with appropriate concentrations of bicarbonate. The fluorescence histograms for a representative time course experiment performed at acidic (6.3), neutral (7.35), or basic (8.0) pH are illustrated in Fig. Fig.88 A and B. The pH-dependent induction of the two promoters investigated is summarized in Fig. Fig.9.9. In the wild-type strain, the two promoters displayed similar pH-dependent upregulation over a narrow range of neutral to slightly basic pHs (pH 7.0 to 7.8 for PvirB and pH 7.3 to 7.9 for PbepD), whereas no induction was detectable for the batR mutant under any of the tested conditions. In the wild type, the virB promoter responded earlier (as early as 24 h) than the bepD promoter (at 48 h), in agreement with the qRT-PCR results presented in Fig. Fig.3.3. The minor differences in the measured pH optima of the tested promoters are in the range of the experimental variation (error bars in Fig. 9A and B) and therefore are not statistically significant. While none of the promoters was induced at acidic pH (Fig. 8A and B), the induction of both PvirB and PbepD at basic pH was not fully abolished but was greatly reduced. In a control experiment, both wild-type B. henselae and ΔbatR strains carrying an IPTG-inducible tac-lac promoter fusion to gfp showed upregulation to the same extent at all three tested pH values, indicating that none of these conditions interfered with transcription or translation in B. henselae (Fig. 8C and D). Plating of the bacteria incubated at different pHs revealed only minor effects on growth for both wild-type and batR strains when acidic (pH 6.4) and neutral (pH 7.6) conditions were compared (see Fig. S3 in the supplemental material). For the basic (pH 8.1) condition, however, a drop in CFU for the wild-type but not the batR strain was observed after 48 h of incubation. These results confirmed that the differences in induction of both PvirB and PbepD between acidic and neutral pH were not a consequence of different survival of the wild-type bacteria. However, this cannot be excluded for induction at basic pH. Figure 9C and D demonstrates the same pH-dependent regulation of virB4 and bepD using qRT-PCR, validating the results obtained with the biosensor strains. In summary, we demonstrated that the BatR-mediated upregulation of PvirB and PbepD is pH dependent, with optimal activation around physiological pH. Therefore, we postulate that BatS constitutes a pH sensor that, together with its cognate response regulator, BatR, mediates the adaptive response of B. henselae during hemotropic infection of HEC.
Finally, we were interested in examining the extent to which the BatR/BatS system has coevolved with its regulon. To this end, we constructed a phylogenic tree of the concatenated sequences of BatR/BatS orthologous proteins in the alphaproteobacteria using the maximum-likelihood method and estimated the congruence of the resulting tree with previously inferred species tree topology (43, 58). The results showed that the BatR/BatS tree was consistent with the topology of the species tree for all nodes with significant bootstrap support (Fig. (Fig.10),10), indicating strict vertical inheritance of the BatR/BatS system. Similar tree topologies were obtained when only the BatR or BatS protein sequence was used (see Fig. S4 in the supplemental material), confirming coinheritance of the two components. To investigate the conservation among the alphaproteobacteria of the 58 genes representing the BatR regulon (as determined by our transcriptional-profiling analysis), we analyzed their presence in the species included in the BatR/BatS gene tree. To this end, we used BlastP and determined the degree of sequence identity for each pair of homologs by calculating their conservation scores (CS) (see Table S3 in the supplemental material) (38). This analysis revealed that only 15 of these 58 genes have homologs within most alphaproteobacteria (Fig. (Fig.10,10, Nod1; see Table S2 in the supplemental material), only 6 of which are relatively well conserved in the analyzed species (see Table S3 in the supplemental material). Overall, these well-conserved genes essentially encode proteins with housekeeping functions (e.g., the chaperonin genes groEL-groES, the serine protease gene htrA1, and the GMP synthase gene guaA) or with a role in transcription regulation (the sensor histidine kinase genes BH04790 and envZ; the response regulator genes BH04780, BH13850, and batR; and the transcription regulator gene rosR). Conservation analysis of the remaining genes showed that, except for six genes conserved in several, but not all, of the Rhizobiales (Fig. (Fig.10,10, node 2), including hbpA and hbpB, encoding hemin-binding proteins, the remaining 37 genes appeared to be unique to the bartonellae or their presence in the bartonellae resulted from horizontal gene transfer, as in the case of the virB operon (Fig. (Fig.10,10, nodes 3 and 4) (42). Moreover, conservation analysis of the 37 Bartonella-specific genes revealed that 27 were absent from the genome of B. bacilliformis and were therefore restricted to the radiating lineage of Bartonella (Fig. (Fig.10,10, node 4) (42). In summary, conservation analysis of the BatR regulon within the alphaproteobacteria suggested that a highly conserved monitoring system modulates the expression of a rapidly evolving set of target genes to adjust the structure and physiology of the bacterium to its environment.
In this study, we have addressed the coordinated response orchestrated by B. henselae to successfully invade and colonize HEC using transcriptional profiling. Specific attention has been paid to the BatR/BatS TCS and its role in the regulation of the VirB T4SS and its translocated effector proteins. The interaction of B. henselae with HEC has been extensively studied (14, 15, 17, 19, 46), and substantial progress has been achieved toward the understanding of the role of the VirB T4SS and its secreted Beps in the mediation of this interaction (41, 44, 45, 49). However, the regulation mechanism(s) controlling the expression of these virulence factors and the global response of the bacteria during HEC interaction remain elusive.
In this paper, we showed that the virB operon and the bep genes are induced during endothelial cell infection, together with a set of coregulated genes (Fig. (Fig.1,1, cluster 1). The progressive but persistent upregulation of these genes best fit the expected gene expression profile enabling an adaptive response. The results obtained from the analysis of a batR mutant revealed that most of the genes found in this cluster are under the positive control of the response regulator BatR. Among these genes, we also found two of the four B. henselae heme receptor gene family members, hbpB and hbpC, whereas hbpA and hbpD were found to be downregulated during infection (Fig. (Fig.1,1, cluster 4). Different expression patterns for the members of this gene family have been reported in B. quintana (5), where hbpA and hbpD were expressed at low heme concentrations while hbpB and hbpC were expressed at high heme concentrations. We described a second cluster of genes displaying rapid upregulation after contact with HEC, followed by slower downregulation during the time course of infection (Fig. (Fig.1,1, cluster 2), reminiscent of the expression pattern expected for a shock response. Indeed, at least six of the genes present in this cluster have been described as being induced and/or implicated in responses to various stress conditions. They include trmD, encoding a tRNA (guanine-n1) methyltransferase (55); rpsU, encoding the 30S ribosomal protein s21 (35); cspA, encoding a conserved transcription and translation enhancer (39); and acnA, encoding an aconitate hydratase (26, 56). Perhaps most interesting in this cluster is ompR, encoding the response regulator from the EnvZ/OmpR two-component system, identified as the major regulatory factor for transcriptional osmotic regulation in E. coli (34, 53). Furthermore, ompR was shown to be involved in the virulence and survival of Brucella melitensis during macrophage infection (59), and a recent study suggested the involvement of ompR in the ability of B. henselae to invade HEC (23). Gene cluster 2 thus seems to reflect the B. henselae response to the drastic change in environment encountered at the onset of the in vitro infection and may comprise factors important for the establishment of contact with the HEC. Since the deletion of batR did not result in any significant alteration in gene expression for this gene cluster, other transcriptional regulators must be involved in their regulation, and the OmpR/EnvZ TCS represents a likely candidate to fulfill this function. The third cluster of genes upregulated during HEC infection and displaying transient upregulation mostly contains genes within a Bartonella-specific island including phage genes (21). The relevance of this transient upregulation is currently unknown; however, the precise timing of this regulation suggests a role in the early stage of interaction with the endothelial cells or a response to the process. Overall, we have presented here the first comprehensive analysis of B. henselae transcriptional response to HEC infection, which can be grouped into four main clusters of coregulated genes. The three clusters showing a complementary upregulation pattern may represent the sequential physiological changes and/or critical factors required to successfully invade and colonize the HEC.
Bacterial TCSs have been associated with different strategies to achieve niche adaptation. A widespread strategy relies on the acquisition of new TCSs by horizontal gene transfer and/or the lineage-specific expansion of existing gene families, as revealed by the analysis of nearly 5,000 histidine kinase genes in 200 bacterial genomes (3). In comparison to these broader studies, we have focused here on one particular TCS, for which we show that both the histidine kinase and the response regulator have evolved by vertical descent over a long evolutionary distance. The conserved mode of evolution for the sensing components of the system is contrasted by rapid diversification of the regulon involved in formulating the response, as revealed by a striking number of genus-specific genes. The rapid diversification of genes regulated by conserved regulatory proteins has been reported as an alternative strategy to achieve niche adaptation (37), with the regulon of orthologous transcription factors being typically divided into shared target genes and species-specific genes. The shared genes of a regulon allow the bacteria to cope with the environmental change that activates the regulon and to control the amount of the active form of the transcription factor, whereas the species-specific targets help each species to proliferate in the particular niches in which they live (37), as exemplified by a recent study on the evolution of the PhoP/PhoQ regulon in the enterobacteria (38). Based on our conservation analysis, the BatR regulon seems to respect this categorization, as it consists of a small set of genes found in most of the alphaproteobacteria that encode either housekeeping (e.g., groEL, groES, guaA, and rlmN) or transcriptional regulators, and a large subset of genus-specific genes, most of which are absent in B. bacilliformis and thus are restricted to the radiating lineage of the bartonellae. Prime among these genes we found the virB locus encoding the VirB T4SS and genes of the flanking bep cluster, encoding the T4SS-translocated effector proteins. We also found two components of the Trw T4SS (trwN and trwJ1), the second T4SS of B. henselae, and one of its regulatory proteins (korA). The Trw T4SS was shown to be essential for intraerythrocytic parasitism of B. tribocorum in the rat infection model (42, 50) but appears to be dispensable for the infection of HEC (15). We also found hecB, a horizontally acquired gene present in multiple copies in the B. henselae genome, that together with fhaC forms a two-partner secretion system. The fhaC-hecB gene products mediate the transport of filamentous hemagglutinin encoded by fhaB (4). All of these genes appear to be under the positive control of BatR, and strikingly, most of them were more than 2-fold upregulated after 48 h of HEC infection (see Table S2 in the supplemental material), supporting the idea that these genes may represent important factors for B. henselae to colonize HEC. Recently, the ChvI regulon of S. meliloti was described based on the differential gene expression between a partial-loss-of-function chvI mutant and a gain-of-function chvI mutant compared to wild-type expression (9). Strikingly, only 4 of these 59 genes appear to be conserved in B. henselae (data not shown), and none of them is under the control of BatR. Furthermore, none of the three direct transcriptional target genes described for S. meliloti ChvI (9) share homologs with B. henselae, emphasizing the rapid diversification of the genes under the control of this conserved TCS, as demonstrated by our conservation analysis of the BatR regulon. Taken together, the BatR regulon consists of a small core of genes that are conserved among the alphaproteobacteria (although these genes do not appear to be under the control of the S. meliloti homolog) and a large proportion of genes specific to the radiating lineage of the bartonellae. This second group encodes a panel of factors critical for the host cell interaction, as exemplified by the VirB T4SS and its secreted substrates.
Our knowledge of the importance of TCSs in enabling mutualistic or pathogenic host interactions by diverse alphaproteobacteria (29, 32, 60) is contrasted by a sparse knowledge of the environmental signals these systems sense, reflecting the diverse host-associated microenvironments the different alphaproteobacteria colonize. Reporter fusions of the virB and bepD promoters to gfp revealed a pH-dependent activation of these BatR-regulated genes, with an optimum at the physiological pH of mammalian blood (pH 7.4). However, contact with HEC did not appear to be necessary for this response, suggesting that pH sensing is one of the mechanisms used by B. henselae to discriminate the host environment from the arthropod vector. This pH dependency resembles the pH-sensitive system ChvG/ChvI in A. tumefaciens. This plant-associated species infects wounded plant cells and genetically transforms them by the transfer of the so-called T-DNA via the T4SS VirB/VirD4. Plants release acids at sites of wounding, and the external acidification is sensed by the BatS orthologue ChvG. This condition can be mimicked under laboratory conditions by acidifying the growth medium to pH 5.5 (29). Under these conditions, chvI was shown to be upregulated, suggesting positive-feedback regulation (61), which is also observed for batR during HEC infection. Although the transcriptional analysis of the A. tumefaciens response to acid conditions has been analyzed in detail (61), the ChvI regulon has not been resolved to date. It is notable that T4SSs were also among our set of genes regulated by BatR/BatS. However, previous studies have shown the T4SS to be frequently horizontally transmitted (22) and that the two operons in Bartonella and Agrobacterium have been acquired independently (42). Nonetheless, the similarity in design is intriguing; both species contain a pH-sensing TCS that monitors the status of the environment of the bacteria and, at the appropriate pH, activates a T4SS, which in turn modulates the host cell to the benefit of the bacterium. Assessing whether the closely related systems BvrS/BvrR of Brucella and ExoS/ChvI of S. meliloti and their orthologues in free-living alphaproteobacteria also display pH-dependent activity would help us to understand whether this group of TCSs represents a vertically inherited versatile pH sensory system that controls the expression of genes that are critical for the adaptive response to the different host-associated life styles of the alphaproteobacteria.
We thank Arto Pulliainen and Jacob Malone for critical reading of the manuscript. We thank Marc Folcher for helpful suggestions regarding the EMSA experiments.
This work was supported by grant 3100AO-109925 from the Swiss National Science Foundation, grant 55005501 from the International Research Scholar program of the Howard Hughes Medical Institute, and grant 51RT-O_126008 (InfectX) in the framework of the SystemsX.ch Swiss Iniative for Systems Biology (to C.D.), as well as by grants from the Swedish Research Council, the Göran Gustafsson Foundation, and the Swedish Foundation for Strategic Research (to S.G.E.A.).
Published ahead of print on 23 April 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.
‡In memory of Hillevi L. Lindroos.