Bacterial effector binds host cell adenylyl cyclase to potentiate Gαs-dependent cAMP production

Abstract

Subversion of host organism cAMP signaling is an efficient and widespread mechanism of microbial pathogenesis. Bartonella effector protein A (BepA) of vasculotumorigenic Bartonella henselae protects the infected human endothelial cells against apoptotic stimuli by elevation of cellular cAMP levels by an as yet unknown mechanism. Here, adenylyl cyclase (AC) and the α-subunit of the AC-stimulating G protein (Gαs) were identified as potential cellular target proteins for BepA by gel-free proteomics. Results of the proteomics screen were evaluated for physical and functional interaction by: (i) a heterologous in vivo coexpression system, where human AC activity was reconstituted under the regulation of Gαs and BepA in Escherichia coli; (ii) in vitro AC assays with membrane-anchored full-length human AC and recombinant BepA and Gαs; (iii) surface plasmon resonance experiments; and (iv) an in vivo fluorescence bimolecular complementation-analysis. The data demonstrate that BepA directly binds host cell AC to potentiate the Gαs-dependent cAMP production. As opposed to the known microbial mechanisms, such as ADP ribosylation of G protein α-subunits by cholera and pertussis toxins, the fundamentally different BepA-mediated elevation of host cell cAMP concentration appears subtle and is dependent on the stimulus of a G protein-coupled receptor-released Gαs. We propose that this mechanism contributes to the persistence of Bartonella henselae in the chronically infected vascular endothelium.

Bacterial pathogens use a plethora of molecular mechanisms to interfere with the host cell signaling to promote their replication, propagation, and escape from the immune system. Pathogens often act via posttranslational protein modifications (phosphorylation, ubiquitylation, sumoylation, AMPylation, ADP ribosylation, and deamidation) to alter activities, subcellular localization, and half-lives of the host proteins (1, 2). Many pathogens also elevate cellular cAMP concentrations using at least five distinct mechanisms. (i) Some bacteria, such as Bordetella pertussis and Pseudomonas aeruginosa, increase cAMP concentration in the target cell by secreting toxins that possess the adenylyl cyclase (AC) activity (3). (ii) Bacteria, such as Vibrio cholera and B. pertussis, increase cAMP concentration in the target cell by secreting toxins that possess ADP ribosylation activity toward heterotrimeric G proteins (4, 5). In the case of V. cholerae, cholera toxin-catalyzed ADP ribosylation of a specific arginine residue of the α-subunit of the AC-stimulating G protein (Gαs) converts the α-subunit into a constitutively active form (4). In the case of B. pertussis, pertussis toxin-catalyzed ADP ribosylation of a specific cysteine residue of Gαi prevents the coupling of α-subunit to G protein-coupled receptor (GPCR), thus abolishing the GPCR-mediated AC inhibition (5). (iii) Bacillus subtilis has been reported to produce an acylpeptide, which inhibits the activity of cAMP-degrading phosphodiesterases (PDEs) in vitro, and increases the cytosolic cAMP concentration in vivo (6). (iv) Recently, Mycobacterium tuberculosis was proposed to deliver cAMP from its own cytosol into the cytosol of a macrophage (7). (v) Cry1Ab toxin of Bacillus thuringiensis has been reported to elevate cellular cAMP concentration by binding to BT-R1, a single-pass cadherin-like plasma-membrane receptor present in the host insect cell (8). In the present study, we have identified and characterized a previously undescribed type of molecular mechanism by which bacterial pathogens increase host cell cAMP concentration.

Bartonella spp. are arthropod-borne facultative intracellular bacteria that typically cause a long-lasting hemotrophic bacteremia in their mammalian hosts, including humans (9). Endothelial cells (ECs) are efficiently colonized by these bacteria and it has been reported that Bartonella henselae (Bh) and Bartonella quintana inhibit actinomycin D-induced apoptosis of human dermal microvascular ECs and human umbilical vein ECs (HUVECs) (10). Most likely, pathogen-triggered blockage of host cell death facilitates a slow microbial replication process and enables chronic persistence. Recently, it was shown that the capacity of Bh to inhibit apoptosis of HUVECs, induced either by actinomycin D or by cytotoxic T lymphocytes, is dependent on the VirB/VirD4-type IV secretion system (T4SS) and its Bartonella effector protein A (BepA) (11, 12). Translocation of BepA into ECs during infection coincides with an increase in cellular cAMP concentration (12). Pharmacological elevation of cAMP by combined action of the AC activator forskolin and the PDE inhibitor isobutylmethylxanthine (IBMX) or by addition of the nonhydrolyzable cAMP analog dibuturyl-cAMP similarly protected ECs from apoptosis (12). This direct phenocopy-effect indicates that the BepA-induced cAMP elevation is the molecular basis of BepA-mediated antiapoptosis. However, the molecular mechanism of how BepA induces the cAMP elevation has remained elusive. Here, we provide evidence that BepA of Bh directly binds the host cell AC to potentiate Gαs-dependent cAMP production.

Results

Gel-Free Proteomics Identify AC and Gαs as Potential Cellular Target Proteins for BepA.

We hypothesized that after T4SS-mediated translocation, BepA binds a host cell protein to increase cellular cAMP concentrations. To initiate gel-free proteomics screens for the identification of cellular target proteins of BepA, we first cloned stable Bep-expressing cell lines. The homologous T4SS effectors BepA, BepB, and BepC of Bh share ∼30% amino acid identity and contain in their C terminus the ∼140 amino acid large Bep intracellular delivery domain and a short positively charged tail sequence (Fig. 1A), which together compose a bipartite signal for T4SS-dependent protein translocation from the bacterium into the host cell (13). Previous transient transfection studies with truncated forms of BepA demonstrated that the Bep intracellular delivery-domain (E305-S446) is sufficient for the antiapoptotic activity of BepA (12). In the present study, a C-terminal fragment of BepA (E305-S544), and the corresponding fragments of the not antiapoptotic homologs BepB and BepC, were N-terminally tagged with HA-EGFP (Fig. 1A). These constructs localized primarily to the plasma membrane in transiently transfected HEK293T cells, as judged by confocal microscopic imaging (Fig. S1A). To clone cell lines expressing these constructs for the gel-free proteomics (Fig. S2A), expression plasmids were transfected into Ea.hy926 cells. Ea.hy926 cells were chosen for this study as they are readily transfectable and Bh infection induced activation of the canonical cAMP/PKA/CREB (cAMP response element-binding) pathway in a BepA-dependent manner (Fig. S1 B and C). Flow cytometry was used to verify clonality of the cell lines and to analyze the expression levels of the fusion proteins (Fig. 1B). Proteolytic stability of the expressed fusion proteins was verified by anti-EGFP Western blotting (Fig. 1C). Quantitative RT-PCR (qRT-PCR) of crem and pde4B expression was used to demonstrate that the BepA-expressing cell line displays a constitutively activated cAMP signaling (Fig. 1D). Moreover, the concentration of cAMP was significantly elevated in the clone that stably expresses HA-GFP-BepA, but not in any other analyzed cell line (Fig. 1E). Therefore, the cloned cell lines were regarded as useful tools to identify the possible cellular protein targets for BepA by gel-free proteomics.

Fig. 1.

Endothelial hybridoma-based cell line that stably expresses HA-EGFP-tagged BepA displays constitutively activated cAMP signaling. (A) Schematic representation of constructs that were used to select Bep-expressing cell lines. (B) Flow cytometry-profiles of the selected cell lines. (C) Proteolytic stability of HA-EGFP-fusion proteins in the cell lines were analyzed by Western blotting and immunodetection with polyclonal anti-EGFP antibodies (antiactin Western blotting shown as the loading control). Sample numbering 1–5 is based on B. (D) Expression of the cAMP-responsive genes pde4B (phosphodiesterase isoform 4b) and crem (cAMP responsive element modulator) was determined by qRT-PCR, as previously described (12). The mean values ± SDs (SD) from a representative experiment are shown for samples analyzed in triplicate. Sample numbering 1–5 is based on B. (E) Quantification of cellular cAMP concentration in confluent monolayers of the stable cell lines. Sample numbering 1–5 is based on B.

Because BepA, BepB, and BepC appear to associate with the plasma membrane (see Fig. 4B and Fig. S1A) it was reasoned that the putative BepA–cellular protein interaction leading to the elevation of cAMP concentration might be dependent on the lipophilic membrane environment. Therefore, Triton X-100 lysates of the stable cell lines were prepared in the presence of a lipophilic primary amine cross-linker. Anti-HA immunoprecipitated material was eluted from the beads with low pH and the whole trypsin-treated eluate was subjected to LC-electrospray ionization (ESI)-MS/MS analysis. This nonbiased gel-free proteomics approach is schematically represented in Fig. S2A. Anti-EGFP Western blotting was routinely used to control each step of the sample preparation (Fig. S1D). Given that BepA induces an elevation of cellular cAMP concentration in Ea.hy926 cells (Fig. 1E) and activates the canonical cAMP/PKA/CREB pathway (Fig. 1D and Fig. S1 B and C), it was of interest to identify AC isoform 7 and the Gαs as possible cellular target proteins for BepA (Fig. S2B). The data indicate that to induce elevation of cAMP concentration in the host cell, BepA might bind and alter activities of the core components of host cell cAMP generation system, the actual enzyme or its stimulatory G-protein α-subunit.

Fig. 4.

In vitro and in vivo studies of the protein–protein interactions of BepA-AC-Gαs triad. (A) Sensorgrams of the binding of MBP (black), MBP-BepC (red), and MBP-BepA (green) on immobilized catalytic domain 2 (C2) of AC2. Interactions were monitored with a Biacore T100 system using five different concentrations of the analytes (1.0, 3.0, 5.0, 7.5, and 10.0 μM). The nonspecific binding of each concentration was determined and subtracted using an untreated control flow chamber. Displayed are representative differential binding curves of the sensor surface carrying immobilized C2 domain. (B) Protein–protein interaction analysis by BiFC in HEK293T cells. Folding of the inherently nonfluorescent YFP1 and YFP2 fragments into a functional YFP upon putative YFP1- and YFP2-fusion partner interactions was analyzed and quantified by flow cytometry. The quantitation is based on triplicate samples (means ± SD) and it refers to the fractions of the cell populations displaying YFP signal over the threshold set by nontransfected HEK293T cells. Western analysis of membrane localization of YFP1- and YFP2-tagged proteins by Western blotting and immunodetection with polyclonal anti-Gαs, and subsequently after membrane stripping with monoclonal anti-HA antibodies. Membrane-enriched cellular fractions were prepared by differential centrifugation after hypotonic cell lysis.

BepA Is a Conditional Gαs-Dependent Activator of Host Cell AC.

To substantiate the proteomics findings (Fig. S2), possible stimulatory effect of BepA on human AC activity was first analyzed in vivo in a heterologous Escherichia coli background. The main reason for this approach was the ability to study possible Gαs-dependent regulation of human AC without the interference of other cellular regulators of G-protein signaling. To this end, the catalytic C1a and C2 subunits of human AC7 (Fig. 2A) were expressed alone or in combination in an E. coli mutant strain, ΔcyaΔcpdA, which lacks the E. coli AC and the cAMP-specific PDE. To activate AC7 in E. coli, Gαs was coexpressed in the ΔcyaΔcpdA strain by another plasmid, either in its native form or as a constitutively active GTPase-deficient Q213L-mutant. As an in vivo read-out for functionality of human AC7 activity under regulation of Gαs in E. coli, growth of bacteria on MacConkey maltose plates was used (14). On these plates, wild-type E. coli colonies are brightly red pigmented because they use maltose as a carbon source through a catabolic activity, which is dependent on cAMP. In contrast, ΔcyaΔcpdA colonies do not get pigmented on the MacConkey maltose plates (Fig. 2B). As shown in Fig. 2B, the ΔcyaΔcpdA strain expressing Gαs-Q213L together with C2-AC7 and C1a-AC7 turned red. This result indicates that the reconstitution of human AC7 activity under the regulation of Gαs in E. coli was successful. Next, the effect of BepA on the Gαs-regulated AC7 activity was studied by introducing bepA into the system in a third plasmid. The smallest antiapoptotic fragment of BepA (E305-S446) (12), and the corresponding fragments of the homologous but not antiapoptotic BepB and BepC, were N-terminally tagged with maltose-binding protein (MBP). Of note, N-terminal tagging of a BepA subfragment with MBP was necessary to acquire a stable form of BepA (see stability of the full-length nontagged BepA in Fig. S3C). As shown in Fig. 2C, expression of BepA_E305-S446 (referred hereafter as MBP-BepA) alone in the ΔcyaΔcpdA strain did not reverse the catabolic defect. Moreover, cAMP production was not significant when BepA was coexpressed with the catalytic subunits of AC7. However, when BepA was coexpressed with both of the catalytic subunits of AC7 and Gαs, a clear reversal of the catabolic defect was detected. Importantly, this phenotype was not detected with the control proteins MBP, MBP-BepB_E303-S444 (referred hereafter as MBP-BepB), or MBP-BepC_D293-G434 (referred hereafter as MBP-BepC). The main findings were recapitulated with an alternative coexpression set-up where BepA was expressed as a full-length form with or without Gαs and the catalytic subunits of AC7. Despite extensive instability of the full-length BepA, the protein still activated the AC in a Gαs-dependent manner (Fig. S3A). In conclusion, data from the E. coli reconstitution system indicate that BepA binds the catalytic subunits of the AC and Gαs to induce cAMP synthesis in a Gαs-dependent manner.

Fig. 2.

In vivo reconstitution of BepA-regulated host cell adenylyl cyclase activity in E. coli. (A) Topology model of mammalian AC (17). Arrows indicate amino acid positions, the first and the last amino acid, of the expressed recombinant C1a and C2 subunits of human AC7 (23). (B) Dots of E. coli grown on a MacConkey maltose indicator plate with or without IPTG. AB472(λDE3) turns red because it can use maltose as a carbon source through a catabolic activity, which is strictly dependent on cAMP. AB472(λDE3)ΔcyaΔcpdA double knockout strain cannot use maltose as a carbon source. However, AB472(λDE3)ΔcyaΔcpdA expressing Gαs-Q213L by one plasmid together with HIS-C2-AC7 and C1a-AC7-S-tag by another plasmid turns red on the MacConkey plate as an indication of cAMP production. (C) AB472(λDE3)ΔcyaΔcpdA strain expressing MBP-BepA by one plasmid and Gαs by another plasmid together with HIS-C2-AC7 and C1a-AC7-S-tag by a third plasmid turns red on the MacConkey plate as an indication of cAMP production. See also Fig. S3A.

To substantiate the findings of the E. coli reconstitution system, we used a highly sensitive in vitro AC assay (15) with membrane preparations of Sf9 insect cells that ectopically overexpress full-length human AC2 (16). AC2 is in the same family and shares many regulatory properties with AC7 (17). The stimulatory effect of Gαs on the membrane-bound AC2, with or without Beps, was studied by using a highly pure recombinant and N-terminally HIS-tagged form (Fig. S4). In the E. coli reconstitution system, this form activated AC7 together with MBP-BepA in a similar fashion to the nontagged form of Gαs (Fig. S3D). The effect of Beps on the AC2 activity was studied using purified recombinant and N-terminally MBP-tagged forms (Fig. S4). As shown in Fig. 3A, the basal activity of AC2 could significantly be increased by Gαs-GDP/γS-GTP. With this designation of Gαs we refer to a condition where we add recombinant Gαs, purified from E. coli in its GDP-loaded form, together with a nonhydrolyzable GTP analog (γS-GTP) into the AC assay. During the AC assay, Gαs-GDP gets activated by spontaneous exchange of GDP with γS-GTP. The effect of BepA on the AC2 activity was analyzed by addition of MBP-BepA in an equimolar ratio to Gαs-GDP/γS-GTP. As shown in Fig. 3A, the specific activity of AC2 was significantly higher in the presence of MBP-BepA and Gαs-GDP/γS-GTP compared with the presence of Gαs-GDP/γS-GTP alone. In fact, exclusion of Gαs-GDP/γS-GTP completely abolished the BepA-mediated activation of AC2. Next, the AC stimulatory potential of a preactivated Gαs; that is, Gαs-GDP that is preincubated with a saturating amount of γS-GTP and subsequently purified from unbound nucleotide (referred hereafter as Gαs-γS-GTP) was determined. As shown in Fig. 3B, Gαs-γS-GTP was more potent than Gαs-GDP/γS-GTP to activate AC, as expected. However, this canonically high AC-stimulatory potential of Gαs could significantly be increased by inclusion of MBP-BepA into the reaction mixture in an equimolar ratio to Gαs-GDP/γS-GTP. MBP-BepA could also increase the AC stimulatory potential of Gαs-GDP, which can activate AC, albeit less potently than active Gαs (18). MBP-BepA also increased the AC stimulatory potential of Gαs in the presence of GTP.

Fig. 3.

In vitro reconstitution of BepA-regulated host cell AC activity. Activity assays of membrane bound human AC in the presence and absence of purified recombinant His-tagged Gαs and MBP-tagged Beps. Specific activity of AC was determined radiometrically by quantification of produced cAMP as described in Materials and Methods. (A) AC was preincubated before the substrate (α-[³²P]ATP) addition with either buffer, GDP-bound Gαs (100 nM), or with a combination of GDP-bound Gαs and equal molar amounts of MBP-BepA. γS-GTP was added to the reaction (100 μM) to activate GDP-bound Gαs during the assay (referred as Gαs-GDP/γS-GTP). (B) AC was preincubated before the substrate (α-[³²P]ATP) addition with either buffer, GDP-bound Gαs, or preactivated γS-GTP-bound Gαs (Gαs-γS-GTP) in the presence or absence of equal molar amounts of MBP-BepA (100 nM). γS-GTP and GTP were added to the reactions (100 μM) as indicated to activate GDP-bound Gαs during the assay (referred as Gαs-GDP/γS-GTP and Gαs-GTP, respectively). Gαs-GDP refers to an experimental condition where no external guanine nucleotide was added to the reaction mixture. (C) AC was preincubated either with buffer, MBP-BepA, MBP-BepB, MBP-BepC, or MBP in the presence of Gαs-GDP/γS-GTP (100 nM). (D) AC was preincubated either with buffer, MBP-BepA, MBP-BepB, MBP-BepC, or MBP in the absence of Gαs-GDP/γS-GTP. As a positive control, MBP-BepA was analyzed in the presence of equal molar amount of Gαs-GDP/γS-GTP (100 nM). (E) Concentration-dependent activation of AC by Gαs in the presence and absence of a constant amount of MBP-BepA. (F) Concentration-dependent activation of AC by forskolin in the presence and absence of a constant amount of MBP-BepA. Statistics based on Student t test with two sample equal variance (homoscedastic): n.s., nonsignificant; *P < 0.05, **P < 0.01, and ***P < 0.005.

To verify BepA-specificity, MBP-BepB, MBP-BepC, and MBP were analyzed in parallel in the presence and absence of Gαs-GDP/γS-GTP. Only BepA increases the specific activity of AC2 (Fig. 3D). To verify the Gαs-dependency, Gαs dose–response analysis of AC2 activation was performed in the presence and absence of a fixed concentration of MBP-BepA. If BepA were stimulating AC activity in a Gαs-independent manner, we would expect the Gαs dose–response curve to shift to consistently higher values across all Gαs-concentrations. Alternatively, if BepA were stimulating AC2 activity in a Gαs-dependent manner, we would expect a right-end shift in the Gαs dose–response curve. The data of the experiment (Fig. 3E) strongly indicate that the latter model applies for BepA. We also performed forskolin dose–response analysis of AC2 activation in the presence and absence of a fixed concentration of MBP-BepA. Forskolin activates the AC by interchalating the C1 and C2 subunits into a catalytically active form (19). As shown in Fig. 3F, MBP-BepA significantly potentiates the stimulation of AC by forskolin. Taken together, the data indicate that BepA is a conditional Gαs-dependent activator of AC, and that BepA is able to use the AC stimulatory potential of Gαs or forskolin and may directly bind AC.

BepA Directly Binds the C2 Catalytic Subunit of the Host Cell AC.

To examine protein–protein interactions of the BepA-Gαs-AC triad, single-cycle surface plasmon resonance (SPR) analyses with purified recombinant components were conducted. To this end, a highly pure recombinant C2 catalytic subunit of AC2 (Fig. S4) was immobilized via amine coupling on a CM5 sensorchip flow chamber, followed by binding measurements of MBP-BepA, MBP-BepC, and MBP that were injected on the C2-coated surface in flow pulses of increasing concentrations. As shown in Fig. 4A, the increase in SPR signal was significantly higher during the flow pulses with MBP-BepA than with MBP or MBP-BepC. It is concluded that BepA directly binds the C2 catalytic subunit of the host cell AC.

BepA Is in Close Proximity to Gαs in Living Cells.

BepA (E305-S544) and the corresponding fragments of the homologous but not antiapoptotic BepB and BepC (Fig. 1A) were N-terminally tagged with HA-YFP2, whereas Gαs was C-terminally tagged with YFP1. Folding of the nonfluorescent YFP1 and YFP2 fragments into a functional YFP upon YFP1- and YFP2-fusion partner interactions (20) were quantified by flow cytometry. Importantly, none of the single or double transfections were deleterious for the cell viability (Table S1). As shown in Fig. 4B and Fig. S5, cotransfection of HA-YFP2-BepA and Gαs-YFP1 resulted in YFP signal from a significant portion of the cell population (43 ± 2%). In contrast, the number of YFP⁺ cells from expression of HA-YFP2-BepB and HA-YFP2-BepC together with Gαs-YFP1 was not significantly different from control HA-YFP2 (16 ± 1%, 17 ± 1%, and 15 ± 1%, respectively). For HA-YFP2-BepC, this result is particularly significant, because the subcellular fractionation data (Fig. 4B) indicate that HA-YFP2-BepC is even more abundant membrane-associated protein than HA-YFP2-BepA. Furthermore, when the HA-YFP2 was forced to interact with the plasma membrane by the means of a C-terminal fusion to the K-ras plasma-membrane trafficking domain (HA-YFP2-CAAX), the size of the YFP⁺ population was significantly lower (23 ± 2%) than with HA-YFP2-BepA and Gαs-YFP1 cotransfections (43 ± 2%). In conclusion, the bimolecular fluorescence complementation (BiFC) data indicate that BepA might directly interact with Gαs. However, the BiFC data could also indicate an indirect BepA-Gαs interaction; that is, BepA binds a cellular protein that is in very close proximity to Gαs at some stage of its activity cycle, such as the AC (Fig. 4A). The latter possibility is supported by the fact that the C2 catalytic subunit of the AC is also the primary binding domain for Gαs (19).

Discussion

Bacillary angiomatosis (BA) and bacillary peliosis (BP) are clinical manifestations of chronic Bh infection in immunocompromised individuals, such as AIDS patients (9). The tumor-like lesions of BA and BP are composed of proliferated and misshapen ECs, a mixed leukocytic infiltrate and bacteria that are associated with the proliferated ECs (9). One of the bacteria-derived factors, which may influence the disease progression, is the antiapoptotic VirB/VirD4 T4SS effector BepA. Translocation of BepA into primary human ECs coincides with an increase in cellular cAMP concentration and pharmacological elevation of cAMP similarly protects the primary human ECs from apoptosis (12). In the present study, we identified and characterized the most proximal signaling event of BepA-mediated inhibition of EC apoptosis, namely how this protein elevates the host cell cAMP concentration.

cAMP is a ubiquitous second messenger. In mammals, nine multipass plasma membrane-bound isoforms of cAMP-producing AC, as well as one soluble AC isoform, have been identified (17). Regulation of the membrane-bound ACs is primarily mediated by the activation of GPCR, release of the GPCR-associated α-subunit of a heterotrimeric G protein (αβγ), and subsequent binding of the Gα with the C1 and C2 catalytic domains of the ACs. Depending on the nature of the α-subunit, ACs can either be inhibited (αi) or activated (αs) (17). Regulators of G-protein signaling proteins act as GTPase activating proteins for α-subunits, thereby reducing the amplitude and duration of signaling (21). The AC activity is also regulated by protein phosphorylation (17). The intracellular concentration of cAMP is further controlled by PDEs, which degrade cAMP to the inactive 5′-AMP (22).

In the present study, we hypothesized that after the T4SS-mediated translocation into the host cell, BepA binds and alters the activity of a protein that influences cellular cAMP concentration. Gel-free proteomics identified AC but also Gαs as potential cellular target proteins of BepA. To study the putative activation of host cell AC by BepA in vivo, human AC activity was first reconstituted under the regulation of Gαs in an E. coli mutant strain that is devoid of the AC and the cAMP-specific PDE. The main reason for this approach was the ability to study Gαs-dependent regulation of human AC without the regulatory interference of receptors, G proteins, or other regulatory proteins that are present in the cell, such as the regulators of G-protein signaling proteins. The catabolic defect of ΔcyaAΔcpdA E. coli strain was complemented by coexpression of BepA with the C1a and C2 catalytic subunits of AC7, but only when Gαs was also expressed in the system. Therefore, BepA appeared to act as a Gαs-dependent activator of AC. In principle, BepA could have acted in analogy to ADP ribosylating bacterial toxins and covalently modify Gαs into a constitutively active form. To study this possibility, we took advantage of a highly sensitive in vitro AC assay (15) with membrane preparations of Sf9 cells overexpressing full-length human AC2 (16) and recombinant BepA and Gαs. BepA activated AC in a Gαs-dependent manner, even in the absence of NAD⁺, a necessary substrate for ADP ribosylation. Most importantly, the in vitro AC activity assays recapitulated the main finding of the heterologous expression system, (i.e., BepA-mediated activation of AC is dependent on Gαs).

In its inactive GPCR-coupled heterotrimeric βγ-bound conformation, the α-subunit of the G protein binds GDP. Agonist stimulation of GPCR promotes the release of GDP from the α-subunit and subsequent binding of GTP, which is accompanied by extensive conformational changes. The GTP-bound active α-subunit dissociates from the βγ-dimer and activates the AC by scaffolding the C1 and C2 catalytic domains. The cycle ends by GTP hydrolysis and the α-subunit returns into a conformation that favors sequestration to the βγ-dimer and GPCR (17). It was of interest to detect under in vitro conditions that BepA-mediated Gαs-dependent stimulation of AC was not dependent on the nature of the Gαs-bound guanine nucleotide. BepA could increase the AC stimulatory potential of inactive Gαs-GDP, but also of Gαs-γS-GTP, which is a preactivated form because of an exchange of GDP with a nonhydrolyzable GTP analog γS-GTP. With respect to Gαs-GDP, the result is not completely unexpected, because Gαs-GDP has been reported to be only 10-fold less potent than Gαs-γS-GTP to activate AC in vitro (18). The authors proposed that the full inactivation of Gαs requires sequestration of the Gαs-GDP with the βγ-complex and GPCR. However, the fact that BepA could increase the stimulatory potential of a canonically 100% active Gαs, Gαs-γS-GTP, was a surprise. Three possibilities were envisioned: (i) BepA binds Gαs and acts as a guanine nucleotide independent activator of Gαs; (ii) BepA binds Gαs and AC and thereby scaffolds the AC/Gαs-complex into an active cAMP producing conformation; or (iii) BepA binds AC and potentiates the Gαs-mediated activation of the AC. We experimentally addressed these possibilities by performing protein–protein interaction studies based on SPR. A direct interaction was detected between recombinant BepA and the C2 catalytic subunit of the AC. Although we could not detect direct interaction between BepA and Gαs by SPR, BepA still appeared to be in very close proximity to Gαs in the cell, as indicated by the data of our BiFC-analyses.

Forskolin binds at the interface of the C1 and C2 domains of AC, in the pocket structurally related to the AC active site, and converts the AC into a catalytically active form (20). It was therefore of interest to detect that BepA potentiates the stimulation of AC activity not only by Gαs but also by forskolin. We thus propose that BepA acts allosterically on AC to favor C1 and C2 subunit association, and via this mechanism is able to increase the efficacy of AC activation by Gαs and forskolin. At the cellular level, we propose that BepA directly binds the AC to increase the efficacy of AC activation by Gαs-GTP that has been released from the GPCR/G protein βγ-complex upon endogenous agonist stimulation of GPCRs, but might also additionally decrease the dissociation rate of Gαs-GDP from the AC (Fig. 5). BepA is a bacterial protein that directly binds host cell adenylyl cyclase to potentiate Gαs-dependent cAMP production.

Fig. 5.

The molecular basis of BepA-mediated elevation of host cell cAMP concentration. Canonically, in its inactive GPCR-coupled heterotrimeric βγ-bound conformation, the α-subunit of the stimulatory G protein binds GDP. Agonist stimulation of GPCR promotes the release of GDP from the α-subunit and subsequent binding of GTP, which is accompanied by extensive conformational changes. The GTP-bound active α-subunit conformation dissociates from the βγ-dimer and activates the AC. The cycle ends by GTP hydrolysis and the α-subunit returns into a conformation that favors sequestration to the βγ-dimer and GPCR. BepA is translocated by VirB/VirD4-T4SS into the host cell. BepA directly binds the C2 catalytic subunit of AC to increase the efficacy of AC activation by Gαs. BepA also potentiates the efficacy of AC activation by forskolin, a plant (Coleus forskohlii)-derived compound that activates the AC by interchalating the C1 and C2 subunits into a catalytically active form (19). We propose that BepA acts allosterically to favor C1 and C2 subunit association and via this mechanism is able to increase the efficacy of AC activation by both Gαs and forskolin. Physiologically, the BepA-mediated elevation of host cell cAMP concentration that leads into host cell survival relies on endogenous GPCR signaling (i.e., on agonist stimulation of GPCRs and subsequent release of the stimulatory α-subunit). This apparently subtle mechanism corroborates well with the lifestyle of Bartonella spp. because these bacteria typically cause chronic and relapsing infections (9).

Physiologically, the direct BepA-AC interaction is interesting because BepA-mediated elevation of host cell cAMP concentration appears to rely on endogenous GPCR signaling (i.e., on agonist stimulation of GPCRs and subsequent release of the stimulatory α-subunit). This mechanism appears to corroborate well with the lifestyle of Bartonella spp. These bacteria typically cause chronic and relapsing infections, which are pronounced in their reservoir hosts, but also evident in the case of vascular tumor formation by B. henseale in immune-compromised individuals, such as AIDS patients (9). The majority of bacterial mechanisms increasing the cellular cAMP concentration, such as the ADP ribosylation of Gαs by cholera toxin (4), cause acute and severe symptoms, including death of the host organism. The BepA-mediated activation of host cell AC is subtle and therefore more suitable for the typical persistence of B. henselae in the infected vascular endothelium.

Materials and Methods

Detailed descriptions of the bacterial strains and growth conditions, generation of the CyaA- and CpdA-deficient E. coli strain, DNA manipulations, cloning and characterization of Bep-expressing stable cell lines, gel-free proteomics, expression and purification of recombinant proteins, activation of Gαs, the BiFC analysis, cAMP quantitation from cellular lysates, and preparation of polyclonal antibodies specific for BepA are provided in SI Materials and Methods and in Tables S2–S4.

Reconstitution of BepA-Regulated AC Activity in E. coli.

To reconstitute the BepA-regulated human AC activity in E. coli, plasmids encoding for catalytic subunits of AC7, bovine Gαs, and Beps were introduced into the AC and cAMP-specific PDE-deficient E. coli strain APE304 by the polyethylene glycol-method. Colonies from freshly transformed plates were subcultured overnight in Luria Bertani (LB) medium with appropriate antibiotics at 37 °C in a 96-well format. The following day, the strains were dotted (3 μL/dot) on MacConkey agar (MacConkey Agar Base; BD Biosciences, 281810) plates supplemented with appropriate antibiotics, 1% (wt/vol) maltose and 5 μM isfopropyl-β-d-thiogalactopyranoside (IPTG) when indicated. The plates were incubated at 30 °C.

In Vitro AC Assays.

Activity of a membrane-bound full-length AC2 was monitored in the presence and absence of highly pure recombinant Gαs and Beps using an assay described by Salomon et al. (15). AC2-containing membranes (15 μg) (16) and various amounts of Gαs or forskolin were incubated for 2 min on ice with indicated concentrations of recombinant MBP, MBP-BepA, MBP-BepB, or MBP-BepC in a final volume of 25 μL. In parallel, reaction mixture containing 0.2 mM ATP, 100 mM MgCl₂, 1 mM ETDA and 10⁶ counts per tube [α-³²P]adenosine 5′-triphosphate (Hartmann Analytic; SRP-207, 3000 Ci/mmol) was heated at 30 °C. To initiate the reaction, 25 μL of the preheated reaction mix was added into the AC2-containing samples and the samples were incubated at 30 °C. After 10 min, reactions were stopped by the addition of 800 μL of stopping solution [0.25% (wt/vol) SDS, 5 mM ATP, 0.175 mM cAMP] and spiked with 10⁴ counts per tube [2,8-3H]-cAMP (Hartmann Analytic; MT616, 15–20 mCi/mmol). cAMP was then purified via a two-column system and quantified by liquid scintillation analysis as described by Salomon et al. (15).

Protein Interaction Analysis by SPR.

To immobilize C2 domain of AC2 on a CM5 sensorchip flow chamber (GE Healthcare), the surface was first activated with a mixture of 77 μL of N-hydroxysuccinimide and 77 μL of N-ethyl-N′-(dimethylaminopropyl)-carbodiimide at 25 °C. After activation, the flow chamber was washed for 4,000 s with 10 μM C2 domain solution at a flow rate of 5 μL/min. Saturation of unreacted carboxymethyl sites was achieved using 1 M ethanolamine solution (pH 8.0). An untreated flow chamber was used as reference during the experiment. MBP-BepA, MBP-BepC, and MBP were diluted in running buffer [100 mM Tris-HCl; pH 8.0, 200 mM NaCl, 5 mM β-mercaptoethanol, 0.05% (wt/vol) N-dodecyl-β-d-maltoside] to a final concentration of 1.0 μM, 3.0 μM, 5.0 μM, 7.5 μM, and 10.0 μM, and injected over the different surfaces with a flow rate of 10 μL/min for 10 min. Samples were kept at 8 °C, experiments were performed at 30 °C. Binding was monitored with Biacore T100 system (GE Healthcare) and surfaces were regenerated in-between samples by washing for 1 min with running buffer. The blank bulk refraction curves from the control flow chamber of every sample [Biacore T100 Evaluation Software (2.0.3.)] as well as buffer response of every channel (OriginPro8) were subtracted.