Clostridium difficile toxin B differentially affects GPCR-stimulated Ca²⁺ responses in macrophages: independent roles for Rho and PLA₂

Abstract

Clostridium difficile toxins cause acute colitis by disrupting the enterocyte barrier and promoting inflammation. ToxB from C. difficile inactivates Rho family GTPases and causes release of cytokines and eicosanoids by macrophages. We studied the effects of ToxB on GPCR signaling in murine RAW264.7 macrophages and found that ToxB elevated Ca²⁺ responses to Gαi-linked receptors, including the C5aR, but reduced responses to Gαq-linked receptors, including the UDP receptors. Other Rho inhibitors also reduced UDP Ca²⁺ responses, but they did not affect C5a responses, suggesting that ToxB inhibited UDP responses by inhibiting Rho but enhanced C5a responses by other mechanisms. By using PLCβ isoform-deficient BMDM, we found that ToxB inhibited Ca²⁺ signaling through PLCβ4 but enhanced signaling through PLCβ3. Effects of ToxB on GPCR Ca²⁺ responses correlated with GPCR use of PLCβ3 versus PLCβ4. ToxB inhibited UDP Ca²⁺ signaling without reducing InsP3 production or the sensitivity of cellular Ca²⁺ stores to exogenous InsP3, suggesting that ToxB impairs UDP signaling at the level of InsP3/Ca²⁺coupling. In contrast, ToxB elevated InsP3 production by C5a, and the enhancement of Ca²⁺ signaling by C5a was prevented by inhibition of PLA₂ or 5-LOX but not COX, implicating LTs but not prostanoids in the mechanism. In sum, ToxB has opposing, independently regulated effects on Ca²⁺ signaling by different GPCR-linked PLCβ isoforms in macrophages.

Introduction

Pathologic overgrowth of Clostridium difficile causes pseudomembranous colitis when normal flora are disrupted during antibiotic therapy. C. difficile is the leading cause of hospital-acquired diarrhea in the United States, causing >250,000 cases/year of diarrhea at a cost of more than $1 billion/year in added hospitalization costs [1, 2]. Disruption of intestinal function and integrity results primarily from the effects of two toxins produced by C. difficile, ToxA and ToxB. ToxA, an enterotoxin, has direct effects on enterocyte ion flux and barrier integrity, and ToxB, a cytotoxin, induces cell death via apoptosis [3]. Both toxins contribute to the virulence of C. difficile [4, 5]. A major target of both toxins is the Rho family of small G-proteins [6,7,8], which regulates cellular structure and metabolism via effects on cytoskeletal organization [9, 10]. Both toxins additionally promote inflammation by activating the production of cytokines and eicosanoids. IL-1β, TNF-α, IL-8, IL-6, and MCP-1 are generated, and the activation of PLA₂ also yields AA, PGE2, and LTs [11,12,13,14,15]. The resulting influx of neutrophils and macrophages and the resulting production of inflammatory mediators cause additional tissue injury.

GPCRs are important regulators of enterocyte and leukocyte function and of the inflammation associated with C. difficile infection. Trimeric G-proteins, activated by GPCRs, stimulate PLCβ isoforms to catalyze the conversion of PIP₂ to DAG and InsP3; InsP3 acts on its receptors (InsP3Rs) to release Ca²⁺ from intracellular stores in the ER. Ca²⁺ signaling requires the spatial organization of these components in close proximity, providing a level of cellular regulation by compartmentalization of cellular biochemistry. Ca²⁺ responses from different receptors in the same cell, however, can vary widely, and we are at an early stage of understanding the factors that determine the magnitude and pattern of Ca²⁺ responses [16,17,18].

Gαq- and Gαi-linked GPCRs can activate Rho family members as well as Ca²⁺ responses. Activation of these signaling pathways differs by receptor and cell type or status, and analysis of some systems has demonstrated Rho-dependent regulation of Ca²⁺ responses, suggesting interaction of these signals. Levels at which Rho family members can regulate changes in [Ca²⁺]_i include modulation of receptor activation through control of cytoskeletal organization [19,20,21,22], assembly of Ca²⁺ signaling complexes on scaffolding proteins [18, 23, 24], regulation of phosphatidylinositol kinases [25,26,27,28,29,30], and direct activation of PLC [31,32,33,34,35]. In all of these mechanisms described, Rho family proteins facilitate Ca²⁺ responses.

In our current studies of GPCR-mediated Ca²⁺ signaling in macrophages, we found that ToxB differentially affects Ca²⁺ responses to different GPCRs based on PLCβ isoform dependence. In RAW264.7 cells, ToxB increased Ca²⁺ responses robustly by Gαi-linked receptors, such as C5aR, but it reduced responses by Gαq-linked receptors, such as the P2Y receptors for UDP. Studies in BMDM showed that this difference correlated with differences between these receptors in their use of PLCβ isoforms and that ToxB enhanced Ca²⁺ signaling through PLCβ3 but inhibited Ca²⁺ signaling through PLCβ4. The inhibition of UDP Ca²⁺ signaling by ToxB was mediated by the loss of Rho activity. In contrast, the enhancement of C5a Ca²⁺ signaling was independent of Rho and reflected an increase in InsP3 production, which was dependent on the activation of PLA₂ and the consequent production of LTs. These studies identify for the first time a differential requirement for Rho in GPCR-mediated Ca²⁺ signaling by macrophages, and they identify independent regulation of PLCβ isoforms by ToxB. These features of ToxB may be important in the colitis that follows infection by C. difficile.

MATERIALS AND METHODS

Reagents

Human complement fragment C5a, UDP, UTP, LPA, PAF, and spiradoline were from Sigma-Aldrich (St. Louis, MO, USA). S1P was from Avanti Polar Lipids (Alabaster, AL, USA). C. difficile ToxA and ToxB (VPI 10463) were from Calbiochem (San Diego, CA, USA) or List Biological Laboratories (Campbell, CA, USA). Native Clostridium botulinum C3 exotoxin (C3 transferase) was from Calbiochem, and cell-permeable C3 exotoxin was from Cytoskeleton, Inc., (Denver, CO, USA). PTx was from List Biological Laboratories. Fura-2 and Fluo-3 were from Invitrogen (Carlsbad, CA, USA). Photolabile-caged InsP3 [D-2,3-O-Isopropylidene-6-O(2-nitro-4,5-dimethoxy) benzyl-myo-inositol 1,4,5-trisphosphate-hexakis(propionoxymethyl)ester] was from Axxora, LLC (San Diego, CA, USA). FTA, AFC, and AGGC were from Cayman Chemical Co. (Ann Arbor, MI, USA). Ionomycin, thapsigargin, LY294002, wortmannin, PP2, and PP3 were from Calbiochem. AACOCF3, MAFP, indomethacin, MK-886, and quinacrine were from EMD Biosciences (Gibbstown, NJ, USA). Cytochalasin D and AA861 were from Sigma-Aldrich. Jasplakinolide, latrunculin A, colchicine, and nocodazole were from EMD Biosciences. AA was from NuChek Prep (Elysian, MN, USA). Mouse IgG2a was from BD Biosciences (San Jose, CA, USA), and F(ab′)₂ fragments of goat anti-mouse IgG were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA, USA). Anti-P2Y6R was from Alomone Labs, anti-C5aR (CD88) was from BD Biosciences, anti-FLAG® was from Sigma-Aldrich, and anti-P-AKT, anti-P-ERK, anti-P-p38, anti-P-JNK, anti-P-PLCγ2, and anti-PLCγ2 were from Cell Signaling Technology (Danvers, MA, USA). Annexin V-FITC and propidium iodide staining solutions were from EMD Biosciences. Additional detailed protocols for reagents, procedures, and solutions are available on the AfCS websites (http://www.signaling-gateway.org/data/ProtocolLinks.html and www.afcs.org) and are referenced according to protocol number (e.g., PP00000226).

Cell culture

RAW264.7 cells, originally obtained from the American Type Culture Collection (Manassas, VA, USA), were grown in DMEM supplemented with 10% FBS, 20 mM HEPES, and 2 mM glutamine (Protocol PS00000636) in non-TC dishes. Cells were passed every 2–3 days by harvest using PBS with 2 mM EDTA. Further details are described on the AfCS websites (Protocol PP00000226). J774 cells were grown similarly but in TC flasks. Culture media and ToxB were assayed for endotoxin contamination by limulus lysate assay and found to contain <0.001 EU/ml at final concentrations in our assays.

Mice genetically deficient in PLCβ2, PLCβ3, or PLCβ4 were described previously [36,37,38,39,40,41,42]. PLCβ3^−/− and PLCβ4^−/− mice were on the C57BL/6 background, and PLCβ2^−/− mice were on the 129SV background. PLCβ-deficient mice and corresponding wt strains were bred and housed at California Institute of Technology (Pasadena, CA, USA) under standard conditions. All experiments were conducted according to protocols approved by the Animal Care and Use Committees of The California Institute of Technology and the San Francisco Veterans Affairs Medical Center (San Francisco, CA, USA). To obtain BMDM, femurs and tibias were removed from sex- and age-matched mice (routinely 4–20 weeks of age, matched ±4 weeks). Cells were extracted immediately, or bones were shipped in medium on ice overnight to another laboratory location. Marrow was flushed from bones, erythrocytes were lysed, and the resulting white cells were seeded in non-TC Petri dishes for selection by growth and adhesion. GM (BMDM-GM) contained DMEM plus 10% FBS, 10% CMG14-12 supernatant (containing murine rM-CSF) [43], 2 mM glutamine, 25 mM HEPES, 100 U/ml penicillin, and 50 μM β-ME (Protocol PS00000722). After 6 days, all of the surviving cells were macrophages, as indicated by CD115 (c-FMS) and CD11b (α_M integrin) expression levels, and these cells were maintained for up to 35 days in culture with replacement of medium every 2–3 days and dilution to reduce cell density as needed. Passage was by brief exposure and harvest in PBS with 2 mM EDTA and 0.5% BSA, centrifugation at 250 g, and replating for culture or assay. Cells were cultured overnight in assay plates prior to use in assays (Protocol PP00000172).

Cells were treated with 5–100 ng/ml ToxB in culture medium for 2–20 h as indicated. Cells were treated with 100 μg/ml native, nonpermeable C3 exotoxin for 20 h or with 1–8 μg/ml cell-permeable C3 exotoxin for 1–20 h. Cells were intoxicated with 50–100 ng/ml PTx by culture overnight (18–22 h). Cytoskeletal, kinase, phospholipase, and LOX inhibitors or the corresponding solvent control solutions were applied for the 30-min equilibration period prior to and during assays as described below. For AA exposure, cells were treated, as described by Rouzer et al. [44], in the presence or absence of 10 μg/ml AA for 4–20 h prior to Ca²⁺ assay.

Expression constructs and transfection

Tobias Meyer and co-workers [45] generously provided dominant-negative (T19N) and constitutively active (Q63L) forms of Rho family proteins as CFP-fusion constructs in the plasmid pECFP (Clontech, Palo Alto, CA, USA). CFP or GFP expressed alone were used as controls. Transient transfection of RAW264.7 cells was performed with Lipofectamine 2000, and cells were assayed 24–36 h post-transfection. Transfection efficiency was 10–40%. Assays were performed on these mixed populations to allow comparison of responses of positive to negative cells in the same field of view during single-cell Ca²⁺ assays. After the Fura-2 Ca²⁺ fluorescence time series, images of CFP or GFP fluorescence were collected to allow the gating of individual cells into positive and negative populations.

Population Ca²⁺ assays

Ca²⁺ responses were measured by monitoring the fluorescence signals of Fura-2-loaded cells in 96-well plates in a fluorescence spectrophotometric plate reader (FLEXstation, Molecular Devices, Sunnyvale, CA, USA). Cells were grown overnight in black-walled TC grade assay plates, loaded with Fura-2-AM (4 μM) for 30 min at room temperature, equilibrated for 30 min at 37°C, and then assayed at 37°C. Baseline readings were collected for 30–40 s prior to ligand addition, and responses were evaluated for 2.5 min after ligand addition. Calibration steps, including additions of a Ca²⁺-minimizing solution (Protocol PS00000607) and Fura-2 Ca²⁺-saturating solution (Protocol PS00000608), were performed at the end of each recording to allow calculation of [Ca²⁺]_i values, according to the method of Grynkiewicz et al. [46], assuming a cytoplasmic K_d of 250 nM for Fura-2. Ca²⁺ signals during the response period were quantified by several features, including the peak-offset response (difference between average baseline Ca²⁺ level and the maximal Ca²⁺ level observed, reported in nM) and the integrated response (integrated Ca²⁺ level above the average baseline over the indicated time period, e.g., 20 s or 2.5 min, reported in nM×s). Further details of the Ca²⁺ assay are described in Protocol PP00000211.

Single-cell Ca²⁺ assays

Cells were plated in GM in chambered coverglasses (Nunc, Rochester, NY, USA; eight wells/coverglass) at 6–60 × 10⁴ cells/well and incubated for 3–24 h at 37°C in 5% CO₂. Ca²⁺-sensitive dyes Fura-2-AM or Fluo-3-AM were loaded by incubation at 4 μM in Ca²⁺ assay buffer for 30 min. Caged InsP3 was loaded as a membrane-permeable PM ester at 3 μM simultaneously with Fluo-3-AM. Cells were then equilibrated in fresh Ca²⁺ assay buffer for 30–45 min prior to assay. Light exposure was minimized during sample preparation. Coverslips were transferred to a microscope stage incubator (Bionomics System, 20/20 Technology, Inc., Wilmington, NC, USA) and kept at 37°C throughout the experiment. Imaging was performed on a Nikon TE-300 fluorescence microscope with a Photometrics HQ2 camera. Illumination was by a 300 W Xe lamp (Sutter Instrument, Novato, CA, USA) with filter/shutter (Lambda 10-3, Sutter Instrument) and dichroic (Conix Research Inc., Springfield, OR, USA) controllers. SimplePCI software was used to control collection parameters and to extract fluorescence intensity data for individual cells. Calibration methods were as described above and were calculated on an individual cell basis. A cytoplasmic K_d of 390 nM was used for Fluo-3 conversion of intensity data to Ca²⁺ concentration. Responses were calculated for each cell and, for transiently transfected samples containing mixed populations of positive and negative transfectants, average peak-offset responses were calculated for each subpopulation, and the ratio of positive/negative responses was determined.

During Fura-2 recordings, illumination was minimized to prevent photobleaching, and image pairs were collected every 2–4 s. Cells were loaded with Fluo-3 and exposed to Fluo-3 excitation wavelength light for minimum duration/image (∼7 ms). To induce photolysis, samples were exposed to DAPI excitation wavelength (360±20 nm) light for bursts of 1 s every 4 s during the photolysis period. Samples were imaged during a 40-s baseline of Fluo-only light, a 60-s photolysis period of alternated DAPI and Fluo light, and a 60-s recovery period of Fluo-only light. Control samples without caged InsP3 loading were imaged similarly to confirm that measured Fluo responses were a result of InsP3 release.

Assessment of the efficacy of cytoskeletal perturbations

The efficacy of dominant-negative forms of Rho was assessed by their effect on formation of membrane blebs in response to ATP, a P2X7R and Rho-dependent phenomenon in RAW264.7 cells [47]. Briefly, RAW264.7 cells were transiently transfected with the T19N version of CFP-labeled Rho or CFP control protein and replated into assay chambers 24 h following transfection. In parallel, cells were treated with C3 exotoxin (1 μg/ml) or control medium for 24 h and then replated into assay chambers. Cells were stimulated with ATP (3.5 mM) for 5 min, fixed with paraformaldehyde, and evaluated for the formation of membrane blebs and/or CFP expression.

The efficacy of cytoskeletal inhibitors was assessed by using F-actin or microtubule-dependent assays. Macropinocytosis is dependent on F-actin, Rac/Cdc42, PLC, and PI3K [48], and was measured by internalization of FITC-dextran as described previously [49]. Briefly, cells were grown overnight in 12-well TC plates prior to assay. The medium was replaced with HBSS with 25 mM HEPES (HBSS), 0.1% BSA, pH 7.4, and cells were treated with cytochalasin D (10 μM) or DMSO (0.1% as control) for 30 min prior to stimulation with MCSF (1.25 nM) in the presence of 1 mg/ml FITC-dextran (150 kDa MW) for 10 min. Samples were then washed and harvested at 4°C for analysis by flow cytometry. Cell morphology, adhesion, and spreading are influenced by F-actin cytoskeletal integrity, and spreading of RAW264.7 cells was assessed by phase-constrast microscopy. Cells were grown overnight in six-well TC plates prior to assay. The medium was replaced with HBSS, 0.1% BSA, pH 7.4, before the cells were treated with cytochalasin D (10 μM), latrunculin A (5 μM), jasplakinolide (5 μM), or DMSO (0.2%) for 30 min and then evaluated morphologically on a phase-contrast microscope. Cell retraction, as evidenced by the presence of rounded cells with residual filopodial extensions, was quantified. Cell-cycle progression and culture expansion are dependent on microtubule function during mitosis, so RAW264.7 culture growth and cell-cycle distribution were assessed by cell-counting and cytometric analysis of DNA content. Briefly, cells were plated into six-well TC plates in the presence or absence of colchicine (25 μM), nocodazole (33 μM), or DMSO (0.1%) and cultured for 48 h. Cells were harvested, counted, fixed with paraformaldehyde, RNase-treated, and stained with propidium iodide for cell-cycle analysis by cytometry. The distribution of cells in G1, S, and G2/M phases of the cell cycle was calculated from the resulting histograms.

SDS-PAGE and Western blot analysis of phosphoproteins

Cells were plated in 35 mm non-TC dishes at 10⁶ cells/dish in 3 ml medium and cultured overnight prior to assay. Medium was replaced with assay medium (DMEM with 2 mM glutamine, 20 mM HEPES, 0.01% BSA) 1 h before stimulation. Ligands were added in 1/10 vol, the well contents mixed gently and cells incubated for the indicated times at 37°C. For FcγRI cross-linking, cells were incubated with 5 μg/ml mouse IgG2a for 10 min prior to addition of 44 μg/ml F(ab′)₂ goat anti-mouse IgG. At the end-point, the buffer was aspirated, the cells were scraped into reducing Laemelli sample buffer, and the samples were heated for 3 min at 95°C. SDS-PAGE gel lanes were loaded with ∼20 μg protein/lane. Western blots of kinases were probed with antibody mixes containing anti-P-AKT or anti-P-ERK and anti-Rho-GDI. Western blots of PLCγ were probed with anti-PLCγ2 or anti-P-PLCγ2 (expression of PLCγ2>>PLCγ1 in macrophages). Signals from P-AKT and P-ERK were normalized to Rho-GDI signals in each lane to account further for loading variations. Levels of P-PLCγ2 were normalized to total PLCγ2 in each sample, detected in lanes blotted in parallel. Further details of SDS-PAGE and Western blotting are in AfCS Protocols PP00000168 and PP00000181.

InsP3 assay

One hour prior to assay, culture medium was replaced with HBSS, 0.5% BSA, pH 7.4. Cells were stimulated by addition of warmed ligands in 1/10 vol, and, after indicated time periods, dishes were transferred to ice, the buffer aspirated, and the cells washed with ice-cold PBS. Cells were scraped into 125 μl 5.4% perchloric acid solution and transferred to siliconized microfuge tubes on ice. After 20 min, samples were centrifuged at 14,000 g for 15 min at 4°C. Supernatant (120 μl) was neutralized with 5 N KOH containing 60 mM HEPES, and the samples were recentrifuged at 14,000 g for 15 min at 4°C. Samples were stored at −80°C prior to assay of InsP3 content using an Amersham (Piscataway, NJ, USA) InsP3 [³H] Biotrak assay kit, according to the manufacturer’s instructions. Results were reported as pmoles InsP3/100 μl cell lysate and normalized to control samples in each assay.

Phosphatidylinositol measurements

Measurement of cellular PIP content was performed by mass spectrometric analysis of samples prepared from cells by a modified, acidified Bligh/Dyer extraction method [50, 51]. One hour prior to cell assay, culture medium was replaced with HBSS, 0.5% BSA, pH 7.4. Cells were stimulated by addition of warmed ligands in 1/10 vol, and, after different time periods, dishes were transferred to ice, the buffer aspirated, and the cells harvested with ice-cold PBS with 2 mM EDTA. Cells were transferred to tubes and pelleted at 2500 g for 5 min. Pellets were extracted with 400 μl ice-cold CHCl₃:CH₃OH (1:1) with vortexing for 1 min, and the resulting suspension was centrifuged at 7500 g for 5 min at 4°C. The supernatant was discarded, and the pellet was dissolved in 200 μl CHCl₃:CH₃OH (2:1) containing 0.25% concentrated HCl with vortexing for 5 min at 4°C. The suspension was transferred to a clean microfuge tube, 40 μl 1 N HCl was added, and the sample was vortexed for 15 s. The phases were then separated by brief centrifugation, and the lower phase was evaporated and redisolved in 55 μl CHCl₃:CH₃OH:H₂O (1:1:0.3). Samples were stored under Argon at −80°C until analysis. Just prior to analysis, 5 μL 300 mM piperidine was added. Samples were analyzed on an Applied Biosystems (Foster City, CA, USA) 4000 QTrap mass spectrometer, and PIP species were quantified relative to a di-C8 PIP₂ internal standard (16:0 PIP₂, Avanti Polar Lipids), as described [50]. The fatty acids are designated by two numbers—xx:y—where xx is the total number of carbon atoms, and y is the number of double bonds. PIP species are reported as 36:2 and 38:4 in mono-(PIP) and bisphosphate (PIP₂) forms.

Rho activation assay

The activation state of Rho was measured by using a G-LISA™ kit from Cytoskeleton, Inc., according to the manufacturer’s protocol. This kit uses a microplate-bound Rhotekin Rho-binding domain to capture GTP-loaded Rho from cell lysates, and a HRP-coupled polyclonal antibody to RhoA, -B, and -C to detect captured protein by colorimetric assay. RAW264.7 cells were grown overnight in six-well plates, and, 1 h prior to assay, culture medium was replaced with HBSS, 0.5% BSA, pH 7.4. Cells were stimulated by addition of warmed ligands added in 1/10 vol, and, after different time periods, dishes were transferred to ice, the buffer aspirated, and the cells washed with ice-cold PBS. Lysates were prepared, protein concentrations measured, and lysate containing equal amounts of protein assayed immediately for Rho-GTP content. Similar assays were performed using 293T cells as positive controls.

Statistical analyses

The error bars in the graphs depict the sem. The statistical significance of each comparison was evaluated by performing a Student’s t-test or a one-way ANOVA, followed by Dunnett’s test or individual t-tests (with Bonferroni correction). A P value of <0.05 was considered significant.

RESULTS

The action of C. difficile ToxB differentially affects GPCR Ca²⁺ responses in RAW264.7 macrophages: Gαi-linked receptor activity is potentiated, and Gαq-linked receptor activity is inhibited

Our laboratory has been exploring the mechanisms of synergistic Ca²⁺ signaling in response to dual ligand stimulation of GPCRs on macrophages [49]. In these cells, simultaneous stimulation by C5a and UDP elicits a synergistic response. To examine mechanisms of this effect, we determined the effect of C. difficile ToxB, which is known to inhibit Rho family GTPases [6, 8, 52]. ToxB did not alter synergy between these GPCR ligands, but it had sharply distinct effects on the individual Ca²⁺ responses to C5a and UDP; the Ca²⁺ response to C5a was enhanced robustly by ToxB, and the Ca²⁺ response to UDP was reduced substantially (Fig. 1, A–C). The effect of the toxin was apparent after 4 h, but it increased with exposures to 24 h. A similar enhancement of C5a and depression of UDP Ca²⁺ responses were observed in J774 cells (Fig. 1D). There was no acute effect of the toxin alone on Ca²⁺ levels nor change in baseline Ca²⁺ after prolonged exposure (data not shown).

Figure 1.

C. difficile ToxB differentially affects Ca²⁺ responses to C5a and UDP in RAW264.7 cells. (A–C) RAW264.7 cells were treated with ToxB (5–100 ng/ml) or control medium for 4–22 h prior to assay. Ca²⁺ assays were performed on Fura-2-loaded adherent cells with (A) 30 nM C5a or (B) 500 nM UDP, applied after a 40-s baseline determination. HBSS served as the buffer control. Lines show the average of n = 4–5 replicate samples from one representative assay of n = 6 with similar results. (C) Ligand specificity of ToxB (TxB) Ca²⁺ response phenotypes in RAW264.7 cells. RAW264.7 cells or transfectants expressing the Ro2 receptor were treated with ToxB (100 ng/ml) or control medium for 4 h prior to assay. Cells were stimulated with 500 nM UDP, 500 nM UTP, 5 nM PAF, 500 nM LPA (∼EC₅₀), 30 nM C5a, or 500 nM spiradoline (Spir; ∼EC₁₀₀). Responses were measured by the integrated area under the peak (difference of [Ca²⁺]_i from basal) over 0–2.5 min after ligand addition and reported as the ratio of ToxB responses to control responses. The dotted line corresponds to a ratio of 1.0, i.e., no change with ToxB treatment. Values are mean ± sem from n = 3–6 assays; *, P < 0.05. (D) Ligand specificity of ToxB Ca²⁺ response phenotypes in J774 cells. Values are mean ± sem from n = 3 assays.

C5a signals through Gαi family G-proteins [49], and ToxB also enhanced signaling through the Gαi-linked κ opioid receptor Ro2 [53], stimulated with spiradoline (Fig. 1C). UDP signals through Gαq-coupled receptors in macrophages [49], and ToxB also depressed Ca²⁺ signaling by UTP, LPA, or PAF GPCRs, each of which signals predominantly through Gαq in macrophages [49] (Fig. 1C). The similar effect of ToxB on multiple receptors coupled to Gαi- versus Gαq-linked pathways suggests that it perturbs Ca²⁺ signaling at steps common to each respective pathway rather than in a receptor-specific manner.

Signaling through C5a invokes an early monophasic Ca²⁺ response in RAW264.7 cells, and signaling through UDP causes an early Ca²⁺ peak and a prolonged, sustained phase. For C5a, ToxB enhanced the early response and introduced a prolonged, sustained phase. In the case of UDP, ToxB inhibited both phases of the Ca²⁺ response, and UDP Ca²⁺ responses were often inhibited by ToxB to levels below those of C5a (Fig. 1, A and B), indicating that the defect is not merely a result of a reduction in general cellular Ca²⁺ signaling capacity. These opposing effects of ToxB on signaling by Gαi- versus Gαq-linked receptors in RAW264.7 cells led us to examine mechanistic differences in the Ca²⁺ signaling pathways by these different groups of GPCRs and to define how these are modulated by the pathogenic toxin.

Loss of Rho activity accounts for the inhibition of UDP Ca²⁺ responses by ToxB in RAW264.7 cells but not for the enhancement of C5a responses

A known principal target for ToxB is the Rho family of small GTPases, which are inhibited by glucosylation of residues Thr35 or Thr37. There are more than 20 members of the Rho family [54, 55], and ToxB inactivates several members selectively: RhoA, -B, -C, and -G, Rac1, -2, and -3, Cdc42, and TC10 but not RhoD or RhoE [8, 52]. Rho, Rac, and Cdc42 isoforms have been implicated in mechanisms of signal transduction for Ca²⁺ responses. In different systems, enhancement by Rho of Ca²⁺ responses has been attributed to the activation of PLCε [34, 56], the production and availability of PIP₂ [25,26,27,28,29,30], and/or the assembly of Ca²⁺ signaling complexes by cytoskeletal scaffolding functions [18, 23, 24, 57,58,59]. Thus, loss of Rho activity could explain the reduced Ca²⁺ responses to UDP in RAW264.7 cells. To test this, we first intoxicated RAW274.7 cells with C. botulinum C3 exotoxin, which inactivates Rho specifically but not Rac or Cdc42. As with ToxB, C3 exotoxin caused a reduction in the peak and sustained levels of intracellular Ca²⁺ in response to UDP (Fig. 2A), but it did not alter responses to C5a (Fig. 2B). The peak and sustained Ca responses to UDP were reduced by ∼40% for all doses tested. These results are consistent with the hypothesis that ToxB reduces the Ca²⁺ response to UDP by inhibiting Rho, but they imply that inhibition of Rho does not enhance Ca²⁺ signaling in response to C5a.

Figure 2.

In RAW264.7 cells, inactivation of Rho inhibits Ca²⁺ responses to UDP but not C5a. Rho family members were targeted with C3 exotoxin (C3 exo) and dominant-negative isoform constructs, and the Ras superfamily was broadly addressed with methyltransferase inhibitors. (A and B) RAW264.7 cells were treated with C3 exotoxin (10 μg/ml) or control medium for 20 h prior to assay. Ca²⁺ assays were performed on Fura-2-loaded adherent cells with 500 nM UDP (A) or 30 nM C5a (B) applied after a 40-s baseline determination. (C and D) RAW264.7 cells were treated with a cell-permeable version of C3 exotoxin (1 μg/ml) or control media for 20 h prior to assay of the Ca²⁺ response to UDP (C) or C5a (D). (E) RAW264.7 cells were treated with varied concentrations of cell-permeable C3 exotoxin for 4 h or 20 h prior to assay of the Ca²⁺ response to UDP. Values are mean ± sem of n = 5 samples from a representative assay; *, P < 0.05. (F) Ca²⁺ responses to 500 nM UDP, 30 nM C5a, 500 nM LPA, 5 nM PAF, or 100 nM spiradoline were assessed by peak-offset feature (normalized to control responses for each ligand). Values shown are mean ± sem from n = 4–10 assays; *, P < 0.05. (G) RAW264.7 cells were transiently transfected with dominant-negative (T19N) versions of CFP-labeled RhoA, -B, or -C versus GFP or CFP as controls. Cells were replated into assay chambers 24 h following transfection, and Ca²⁺ assays were performed on Fura-2-loaded adherent cells 48 h post-transfection. Ligands were 25 μM UDP or 30 nM C5a, applied after a 30-s baseline determination. Responses of transfectants (fluorescent protein-positive) were quantified by peak offset and normalized to the response of nontransfectants (fluorescent protein-negative) within the same sample. Transfection efficiencies were 10–40%. Results are shown as mean ± sem from n = 3 assays/condition; *, P < 0.05. (H) Fluorescence distribution and gating of cells in CFP-RhoA-T19N versus no-CFP (untransfected) samples. (I) RAW264.7 cells were transiently transfected with the T19N version of CFP-labeled RhoA dominant-negative (CFP-RhoA-DN) or CFP control protein or pretreated with C3 exotoxin (1 μg/ml) or control medium for 24 h and then plated into assay chambers. Cells were stimulated with ATP (3.5 mM) for 5 min, fixed with paraformaldehyde, and evaluated for presence of membrane blebs and/or CFP. Values are shown as percent cells positive for membrane blebs, indicated as mean ± sem of n = 3–10 samples in a representative assay of two with similar results; *, P < 0.001. (J) RAW264.7 cells were treated with cell-permeable C3 exotoxin (1 μg/ml) or control medium for 24 h, with or without ToxB at 50 ng/ml for the final 4 h. Ca²⁺ assays were then performed, and peak-offset responses to 5 nM C5a were calculated. Results are shown as mean ± sem from n = 6 assays per condition; *, P < 0.01. (K) RAW264.7 cells were treated with the methyltransferase inhibitors FTA (40 μM) and AFC (30 μM) for 4 h prior to stimulation in Ca²⁺ assays, which were performed on Fura-2-loaded adherent cells with 500 nM UDP or 30 nM C5a, applied after a 40-s baseline determination. The peak-offset feature of the responses was quantified and presented as the mean ± sem from four assays with four samples/assay; *, P < 0.01.

C3 exotoxin enzymatically inactivates Rho by ADP ribosylation of Asn41 residues in the switch I region [60], which reduces Rho GTP-loading, as a result of increased association with Rho-GDI in the cytosol [61], and interferes with effector activation by the GTP-bound G-protein [62]. The substrates of C3 transferase exotoxin are limited to Rho isoforms A, B, and C, and other Rho or Ras family members (such as Rac, Cdc42, Ras, or Rap) are not targeted [63,64,65]. As the permeability of the cell membrane to native C3 exotoxin is low, effective enzymatic inhibition required incubation of cells with relatively high concentrations of exotoxin for prolonged periods (e.g., 10 μg/ml overnight). Therefore, we also tested a membrane-permeable version of C3 exotoxin. As shown in Figure 2, C and D, this resulted in a similar reduction in the Ca²⁺ response to UDP but not to C5a, acting in as little as 4 h and with as little as 1 μg/ml exotoxin (Fig. 2E). Inhibition of the UDP response was dependent on toxin dose and on time, and the maximum inhibition observed was ∼75% (Fig. 2E). The effect of Rho inhibition on UDP signaling was not a result of a reduced maximal Ca²⁺ signaling capacity in toxin-treated cells, as responses to low doses of UDP, which approximate the response to C5a, were inhibited to levels well below those of C5a, which was unaffected by C3 exotoxin (Fig. 2, C and D). Thus, C3 exotoxin yielded the same pattern of inhibition as seen with ToxB; Gαq-linked ligand responses were reduced, and Gαi-linked ligands were unaffected (Fig. 2F). To confirm and to characterize the Rho involvement in Ca²⁺ responses to UDP but not C5a, we performed assays on RAW264.7 cells transiently transfected with dominant-negative forms of Rho. As shown in Figure 2G, CFP-tagged T19N mutants of RhoA, -B, and -C isoforms yielded reduced Ca²⁺ responses to UDP but not to C5a. Responses of CFP expressors were normalized to nonexpressors in the same sample for optimal sensitivity (Fig. 2H). Expressed RhoA-T19N was confirmed to inhibit membrane bleb formation in response to ATP (Fig. 2I), a phenotype demonstrated to be mediated by P2X7R in RAW264.7 cells in a Rho-dependent manner [47]. RhoA, -B, and -C have overlapping and distinct activities [66]. The dominant-negative forms of Rho act by inhibiting Rho-associated GEFs, and RhoA, -B, and -C share several GEFs [55, 67], as do additional members of the Rho family. Thus, the Rho isoform specificity, if any, is not defined by these experiments, but a factor or activity common to these isoforms is implicated. C3 exotoxin and dominant-negative Rho isoform experiments support a role for RhoA-like proteins. In contrast, transient transfection of RAW264.7 cells with dominant-negative constructs of Rac1 or Cdc42 failed to affect the UDP or C5a responses (data not shown), confirming that these members of the Rho family, unlike RhoA-like proteins, do not regulate these Ca²⁺ responses.

In some systems, ToxB intoxication can lead to increased expression of RhoB, and when RhoB exceeds the inhibitory capacity of the toxin, this can yield a contradictory increase in net RhoB activity [68, 69]. RhoB has been implicated in the regulation of vesicle trafficking [70, 71] and receptor surface expression in macrophages [72], and it could impact GPCR Ca²⁺ responses. To evaluate RhoB escape from inhibition in the enhancement of C5a Ca²⁺ by ToxB, we combined ToxB treatment with C3 exotoxin treatment, which will maintain inactivation of RhoB by ADP-ribosylation [69]. C3 exotoxin treatment did not prevent the up-regulation of C5a Ca²⁺ responses (Fig. 2J), indicating that Rho does not contribute to the elevation of C5a Ca²⁺ responses. Additionally, expression of constitutively active RhoB did not elevate C5a Ca²⁺ responses (data not shown). Thus, escape of RhoB from ToxB inhibition does not appear to contribute to the elevated C5a Ca²⁺ responses.

Members of the Ras superfamily of small GTPases, of which the Rho family is a subset, are isoprenylated proteins that undergo carboxy-terminal methylation for activation. Inhibitors of isoprenylation or methylation disrupt the function of small GTPases by affecting subcellular localization and effector interactions [73, 74]. We used this alternative biochemical means of inhibition of Ras superfamily proteins as an independent test of their involvement in UDP versus C5a Ca²⁺ responses. Our results with more directed inhibitors of Rho proteins predicted that broader inhibitors of Ras proteins would similarly inhibit Ca²⁺ responses to UDP, and these inhibitors would help clarify if non-Rho Ras superfamily targets of ToxB contribute to the elevated C5a Ca²⁺ phenotype. As predicted, methyltransferase inhibitors inhibited Ca²⁺ responses to UDP (Fig. 2K). Effects on Ca²⁺ signaling were seen after as little as 30 min inhibition (data not shown), but a 4-h treatment was needed for consistently significant results [75,76,77]. Similar results were seen with FTA plus AFC (Fig. 2K) or FTA plus AGGC (data not shown). Methyltransferase inhibitors had no effect on Ca²⁺ responses to C5a, however, indicating that members of the Ras superfamily are not required for C5a Ca²⁺ signaling, nor does their inhibition induce an elevation of C5a responses.

Studies of BMDM reveal that ToxB differentially affects PLCβ3- versus PLCβ4-dependent pathways

The differential effects of ToxB on C5a and UDP Ca²⁺ signaling in RAW264.7 cells could reflect differential effects of ToxB on Gαi versus Gαq trimeric G-proteins themselves or, instead, events more distal in these Ca²⁺ signaling pathways. Previous studies from our laboratory indicated that Ca²⁺ signaling by C5a in both RAW264.7 and BMDM is largely dependent on PLCβ3 but not PLCβ4 [49]. Ca²⁺ signaling by UDP in RAW264.7, however, is largely dependent on PLCβ4, but in BMDM, UDP signaling uses both PLCβ3 and PLCβ4 [49]. In these prior studies, Ca²⁺ responses to UDP by BMDM were diminished by ∼80% in the absence of PLCβ3 and by ∼20% in the absence of PLCβ4, whereas in RAW264.7 cells, UDP responses were diminished by RNAi of PLCβ4 but not RNAi of PLCβ3. PLCβ2 is expressed in RAW264.7 cells and BMDM, but it appears dispensable for normal Ca²⁺ responses to C5a or UDP, and PLCβ1 is not expressed in either cell type [49].

To explore possible relations between ToxB and PLCβ isoforms, we examined effects of ToxB on BMDM from mice with selective genetic deletion of different PLCβ isoforms. In BMDM from wt mice, the maximal Ca²⁺ responses to C5a and UDP were greater than in RAW264.7 cells, and C5a responses were more robust than those of UDP; in RAW264.7 cells, UDP gave greater responses. In BMDM, the Ca²⁺ responses to C5a were elevated further by exposure to ToxB (Fig. 3A), as they were in RAW264.7 cells, but in contrast to RAW264.7 cells, signaling in response to UDP and other Gαq-linked receptors was enhanced rather than inhibited by ToxB (Fig. 3B), although the enhancing effect of ToxB on signaling by Gαq-linked receptors was significantly less than its enhancement of signaling by the Gαi-linked C5aR (Fig. 3C). Thus, a distinction in ToxB effect on Gαi versus Gαq Ca²⁺ responses persisted in BMDMs, but instead of the opposing effects seen in RAW264.7 cells, the difference was in the magnitude of enhancement.

Figure 3.

In BMDM, C. difficile ToxB enhances Ca²⁺ responses to C5a and UDP to different extents, and C3 exotoxin inhibits Ca²⁺ responses to UDP but not C5a. (A–C) BMDM were treated with ToxB (10 ng/ml) or control medium for 18 h prior to assay. Ca²⁺ assays were performed on Fura-2-loaded adherent cells stimulated with (A) 10 nM C5a or (B) 10 μM UDP (∼EC₁₀₀ s) after a 40-s baseline determination. (C) Ligand specificity of ToxB Ca²⁺ response phenotypes in BMDM. Cells were stimulated with 1 μM UDP, 1 μM UTP, 2 nM PAF, 1 μM LPA, 20 nM LTD4, or 1 nM C5a (∼EC₅₀). Values shown are mean ± sem from n = 3–8 assays. (D–F) Inhibition of Rho by C3 exotoxin inhibits UDP but not C5a Ca²⁺ responses in BMDM, which were treated with 1 μg/ml cell-permeable C3 exotoxin for 20 h and then stimulated with 5 nM C5a (D) or 10 μM UDP (E) in Ca²⁺ assays. (F) Quantitation of BMDM responses by peak-offset feature. Values are mean ± sem of n = 4 samples/condition from a representative assay of five with similar results; *, P < 0.01. (G) PLCβ3 expression is required for ToxB elevation of Ca²⁺ responses. BMDM cultures established from mice genetically deficient (−/−) in PLBβ2, -3, -4, or -2 and -3 (2+3) were treated with or without ToxB (10 ng/ml) for 18 h and assayed for Ca²⁺ responses to 10 nM C5a or 10 μM UDP (∼EC₁₀₀). Responses were measured by the integrated area from 0 to 2.5 min after ligand addition, and results of ToxB-treated cells were normalized to those of the corresponding untreated cell type. C5a responses in untreated PLCβ2 + 3-deficient cells were minimal and therefore not assessed (NA) for changes with ToxB treatment. Values shown are mean ± sem from 16 replicate samples/condition from five assays; *, P < 0.05.

As in RAW264.7 cells, however, inhibition of Rho using C3 exotoxin selectively inhibited UDP responses of BMDM with no effect on C5a (Fig. 3, D–F). This result suggested that in BMDM, the Rho-dependent inhibitory effects of ToxB on UDP signaling may be offset by an enhancing effect. That is, in BMDM, enhancement of UDP Ca²⁺ signaling by ToxB outweighs the Rho-dependent inhibition, whereas in RAW264.7 cells, inhibition by ToxB dominates. As UDP signaling in RAW264.7 cells is largely dependent on PLCβ4, and in BMDMs, it uses PLCβ4 and PLCβ3 [49], we tested the hypothesis that ToxB differentially affects PLCβ isoform pathways to Ca²⁺, inhibiting Ca²⁺ signaling through PLCβ4 but enhancing signaling through PLCβ3.

In accord with this hypothesis, ToxB failed to enhance Ca²⁺ responses to C5a or UDP in BMDM from mice genetically deficient in PLCβ3 (Fig. 3G). In BMDM from mice deficient in PLCβ2 or PLCβ4, ToxB still enhanced C5a and UDP responses comparably with wt BMDM. Thus, ToxB appears to promote PLCβ3-mediated Ca²⁺ signaling. In contrast, Ca²⁺ responses to UDP were inhibited by ToxB in BMDM deficient in both PLCβ2 and PLCβ3. In these mice, Ca²⁺ responses are dependent solely on remaining PLCβ4 (mediated by Gαq with no PLCβ1, -2, or -3 expressed). Thus, although ToxB enhances signaling by PLCβ3, it inhibits signaling by PLCβ4. This accounts for the quantitative differences in the effects of ToxB on BMDM and RAW264.7 cells. In RAW264.7 cells, UDP signals primarily through PLCβ4, and it is inhibited by ToxB. In BMDM, UDP instead signals ∼20% through PLCβ4 but ∼80% through PLCβ3. Here, enhancement of PLCβ3 signaling by ToxB dominates inhibition of PLCβ4 signaling. When PLCβ3 and PLCβ2 are absent, however, the inhibition of PLCβ4 signaling can be seen. Because RAW264.7 cells readily reveal the differential effects of ToxB, we turned to them for further studies about the mechanisms by which ToxB affects signaling.

ToxB affects the release of Ca²⁺ from intracellular stores in response to UDP or C5a

Changes in Ca²⁺ responses could reflect changes in the initial release of Ca²⁺ from intracellular stores, the influx of extracellular Ca²⁺ through SOCs, or both. Release of intracellular Ca²⁺ stores is a prerequisite for the influx of extracellular Ca²⁺, but the rapid coupling of this response to SOCs may contribute to the magnitude of the early peak of [Ca²⁺]_i. Thus, the observed reduction in the early [Ca²⁺]_i peak and the sustained levels of [Ca²⁺]_i could result from deficiency in the release from Ca²⁺ stores alone or in SOCs coupling. To address this, we examined the effects of ToxB on Ca²⁺ signaling in the absence of extracellular Ca²⁺ so that changes in [Ca²⁺]_i reflected only release of Ca²⁺ from intracellular stores. Under these conditions, ToxB again reduced the rise in [Ca²⁺]_i elicited by UDP and increased the response to C5a (Fig. 4, A and B). ToxB did not affect the content of intracellular stores, as reflected by similar Ca²⁺ peaks with release of intracellular stores using thapsigargin/ionomycin (data not shown). A change in Ca²⁺ stores, moreover, would not alone explain the dichotomous effect of ToxB on UDP versus C5a. We conclude that ToxB differentially affects signaling pathways that alter the release of intracellular Ca²⁺ stores by UDP versus C5a. This indicates that ToxB affects proximal signaling steps, although it does not preclude additional effects on Ca²⁺ flux through SOCs.

Figure 4.

ToxB affects the release of Ca²⁺ from intracellular stores and does not reflect a change in C5aR expression, G-protein coupling, or Src-kinase dependence. RAW264.7 cells were treated with ToxB (10 ng/ml) or control medium for 20 h prior to assay. Ca²⁺ assays were performed on Fura-2-loaded, adherent cells. Buffer with or without EGTA (2 mM final) was added to samples 30 s prior to addition of 25 μM UDP (A) or 30 nM C5a (B). (C) The C5a Ca²⁺ response remains sensitive to PTx after enhancement by ToxB. RAW264.7 cells were treated with or without PTx (50 ng/ml) and/or ToxB (10 ng/ml) for 18 h prior to Ca²⁺ assay. Each line represents the mean of five replicate samples. Shown are data from a representative assay of four with similar results. The mean peak-offset values were calculated for each response from these four experiments and are presented (D). Values are mean ± sem. (E) Src activity is not required for the enhancement of C5a Ca²⁺ responses by ToxB. RAW264.7 cells were treated with ToxB (10 ng/ml) for 20 h, and the Src inhibitor PP2 or control PP3 (10 μM) was added for 30 min prior to and during the assay. Ca²⁺ assays were performed with 5 nM C5a, and peak-offset responses were calculated. Shown are mean ± sem from n = 3 assays. (F) Src activity is not required for normal Ca²⁺ responses to C5a or UDP. RAW264.7 cells were treated with PP2 or control PP3 for 30 min prior to assay of Ca²⁺ responses to 5 nM C5a, 500 nM UDP, or FcγRI cross-linking. Responses to each stimulus in the presence of PP2 were normalized to those in the presence of PP3. Shown are mean ± sem from n = 3–5 assays.

Enhanced Ca²⁺ responses to C5a do not reflect changes in receptor expression or Gαi dependence

The effects of ToxB on Ca²⁺ signaling by multiple GPCRs suggested perturbation of signaling at steps common to these receptors, rather than receptor-specific changes. We nonetheless examined C5aR surface expression to determine if it was up-regulated by ToxB in conjunction with the elevation in C5a Ca²⁺ responses. No change in C5aR surface expression was seen with ToxB treatment (data not shown), confirming that the ToxB perturbation was distal to receptor availability.

In macrophages, signal transduction by the C5aR is mediated principally by trimeric G-proteins that are sensitive to PTx. We have shown previously, however, that a modest PTx-insensitive component exists in the Ca²⁺ response to high concentrations of C5a, which may represent signaling through Gα15 [49]. We therefore assessed whether ToxB might enhance C5aR Ca²⁺ responses by selectively enhancing PTx-insensitive pathways. It did not. In the presence of ToxB, most C5a Ca²⁺ signaling remained sensitive to PTx, although at high concentrations of C5a, a minor PTx-insensitive path could again be detected (Fig. 4, C and D). PTx-sensitive and insensitive components of the C5aR responses were increased by ToxB, indicating a lack of specificity of ToxB between these two pathways. These results indicate that ToxB does not alter G-protein coupling by C5aR.

The effects of ToxB are independent of src kinases and PLCγ

In some cells, activation of GPCRs can lead to transactivation of tyrosine kinase pathways and receptors and associated activation of MAPKs as well as PLCγ-dependent Ca²⁺ responses [78,79,80]. In particular, activation of src kinases by GPCRs [81, 82] has been identified as a critical link for GPCR responses to epidermal growth factor or vascular endothelial growth factor receptors [78, 79, 83], as well as activation of PLCγ via GPCR kinase-interacting proteins [80, 84]. In mast cells, synergy has been seen between PLCβ and PLCγ with simultaneous stimulation of PGE₂ and FcεRI receptors [85]. Thus, GPCR Ca²⁺ responses in macrophages could reflect the net activities of PLCβ and PLCγ, and augmented GPCR Ca²⁺ responses could result from alteration of transactivation signaling pathways by ToxB. To evaluate the potential contribution of src kinases and PLCγ to ToxB effects, we evaluated the effect of src inhibition on Ca²⁺ responses and the phosphorylation state of PLCγ with ToxB treatment. There was no effect of the src inhibitor PP2 on the elevation of C5a Ca²⁺ signaling by ToxB nor was there a significant effect on Ca²⁺ responses by C5a or UDP alone, although PP2 did block FcγRI responses as expected (Fig. 4, E and F), and as expected, cross-linking of FcγRI led to phosphorylation of PLCγ, but no phosphorylation was detected in response to ToxB, with or without C5a stimulation (data not shown). Thus, we found no evidence of a contribution of these transactivation mechanisms to the ToxB effect.

Availability of PIP₂ and production of InsP3 are maintained or increased in the presence of ToxB

Release of Ca²⁺ from intracellular stores is mediated by the opening of Ca²⁺ channels in response to secondary messengers, including InsP3, cyclic ADP ribose, nicotinic adenine dinucleotide, and S1P [16, 17]. PLCβ cleaves PIP₂ to yield InsP3 and DAG. The activity of PLCβ is limited by the availability of PIP₂, and studies in other cell systems have demonstrated previously a role for Rho in the activation of phosphatidylinositol 4 and 5 kinases, which are responsible for production of PIP₂ [25, 28, 86]. To determine if ToxB causes a loss of PIP₂, which could lead to reduced Ca²⁺ responses, we measured the availability of PIP₂ in RAW264.7 cells following treatment with ToxB. ToxB treatment did not diminish the levels of PIP₂ (Fig. 5A), indicating that an adequate supply of substrate is available for PLCβ prior to receptor activation. Our studies were of total PIP₂, and functionally distinct pools of PIP₂ have been identified that differentially contribute to GPCR Ca²⁺ responses [86,87,88,89]. Selective effects of ToxB on these pools could escape detection in our assays, but this explanation was excluded by the next set of experiments, in which we assessed the production of InsP3 directly.

Figure 5.

Availability of PIP₂ and production of InsP3 are maintained or increased following exposure to ToxB. RAW264.7 cells were treated with ToxB (10 ng/ml; A, C, E), the cell-permeable version of C3 exotoxin (1 μg/ml; D), and/or control medium (B) for 20 h prior to assay. (A) Cells were harvested and lipids extracted and quantified by mass spectrometry. Samples were spiked with a fixed mass of lipid standard, and the amounts of each PIP lipid are reported as a fraction of the standard. Shown are mean ± sem from n = 3 samples/condition from a representative assay of three with similar results. (B) Untreated cells were stimulated with 10 μM UDP, 50 nM C5a, or buffer alone (Control Baseline), and samples were harvested at the indicated times to measure cellular InsP3 (IP3) content. Shown are results from n = 11 assays. (C–E) Cells were stimulated with ligand or buffer alone (no stim) for 3 min, and samples were processed to measure cellular InsP3 content. Amounts are reported as InsP3/cell equivalent and normalized to control cells in the same assay. Values shown are mean ± sem from n = 3–6 assays with three samples/condition/assay; *, P < 0.05. (F and G) Inhibition of Rho does not alter the Ca²⁺ response of RAW264.7 cells to exogenous InsP3. RAW264.7 cells, with or without C3 exotoxin exposure (1 μg/ml for 20 h), were loaded with Fluo-3 and photolysable-caged InsP3-PM. Single-cell Ca²⁺ recordings were performed, during which cells were flashed repeatedly with 360 nM light to release InsP3 (the period indicated as “photolysis” between 40 and 100 s). (F) Representative traces of control or C3 exotoxin-treated cells showing population mean ± sem, where n = 54–128 individual cells/population. (G) The peak-offset feature of the responses was measured and normalized to the average control value for each assay. Results shown are mean ± sem with n = 8 samples/treatment from three experiments with similar results.

To determine if ToxB treatment perturbed Ca²⁺ signaling mechanisms proximal or distal to InsP3 production, we measured cellular InsP3 content at baseline and after stimulation with UDP or C5a. As shown in Figure 5B, a robust increase in cellular InsP3 content, peaking at ∼3 min, was observed following stimulation with UDP, and the 3-min time-point was used for subsequent analyses. When RAW264.7 cells were stimulated with UDP in the absence of ToxB, InsP3 was increased approximately threefold, and the levels of ligand-stimulated InsP3 were not reduced by ToxB or by C3 exotoxin (Fig. 5, C and D). Although similar levels of InsP3 were achieved upon stimulation with UDP, with or without ToxB treatment, baseline levels were elevated by ToxB treatment alone, and thus, a reduced change from baseline was observed. This did not, however, explain the effects of Rho inhibition on Ca²⁺ signaling, as C3 exotoxin had no significant effect on baseline levels of InsP3, and, like ToxB, it did not reduce levels of InsP3 in response to UDP. Thus, the inhibition of the UDP response following inhibition of Rho lies distal to the generation of InsP3.

In the absence of ToxB, a rise in InsP3 could not be detected in RAW264.7 cells following C5a (Fig. 5E), despite the dependence of the Ca²⁺ response on PLCβ. We presume that the activation of Ca²⁺ signaling by C5a reflects efficiency of coupling InsP3 to calcium stores or rapid kinetics of InsP3 that cannot be detected by our assays. Regardless, in cells treated with ToxB, C5a stimulated a robust increase in InsP3 content, which reached or exceeded that of UDP-stimulated responses (Fig. 5E). Taken together, these results indicate that ToxB inhibits Ca²⁺ signaling by UDP at a point distal to the production of InP3, but it enhances C5a Ca²⁺ signaling, at least in part, by augmenting the production of InsP3.

Inhibition of Rho in macrophages does not reduce the release of Ca²⁺ stores in response to InsP3

As InsP3 generation in response to UDP is intact in the presence of ToxB or C3 exotoxin, but release of Ca²⁺ from intracellular stores is reduced, either the sensitivity of the Ca²⁺-release mechanism to InsP3 is reduced, or the InsP3 that is generated in response to UDP does not reach the InsP3Rs in a concentration or a manner that allows an adequate engagement with the receptors. We tested the former possibility by measuring Ca²⁺ responses to exogenous InsP3, introduced by loading RAW264.7 cells with photolyzable caged InsP3. Single-cell imaging was used to monitor changes in [Ca²⁺]_i following release of the InsP3 by a photolytic light impulse. As shown in Figure 5F, the responses included a rapid primary peak in [Ca²⁺]_i, followed by a sustained phase during the maintained photolytic exposure. Upon discontinuation of photolytic exposure, [Ca²⁺]_i levels began a return to baseline. Levels of [Ca²⁺]_i were similar in cells treated with C3 exotoxin compared with controls (Fig. 5G), suggesting no loss in sensitivity of intracellular stores to InsP3 when Rho is inhibited. As levels of InsP3 are not reduced by ToxB, and sensitivity to InsP3 is not reduced by inhibition of Rho, we conclude that the loss in Ca²⁺ signaling in response to UDP reflects a modification of factors that promote the local engagement of InsP3 with its receptors or a change in the intracellular distribution of released InsP3, such that it is not available to the InsP3Rs.

Cytoskeletal organization is not necessary for coupling of InsP3 to Ca²⁺ release in response to UDP

In several cell systems, the relative proximity of InsP3 generation sites to sites of intracellular Ca²⁺ store release affects the magnitude of receptor-stimulated Ca²⁺ responses. In particular, the organization of signaling complexes by protein–protein interactions is thought to regulate differences in receptor localization in neurons and fibroblasts [18, 23, 24]. Clustering of GPCRs, PLCβ, transient receptor potential channels, shank, and InsP3Rs, among other factors, provides efficient linkage of InsP3 to InsP3Rs on the ER. Differential formation of these Ca²⁺ signaling complexes (“Ca²⁺ signalosomes”) may thus augment Ca²⁺ responses in a manner that amplifies proximal signaling and InsP3 generation. Scaffolding proteins that affect these clusters include F-actin and Homer. Rho is an important regulator of the F-actin cytoskeleton, and it interacts with Homer. Thus, inhibition of Rho by ToxB could disrupt Ca²⁺ signalosome formation.

To evaluate possible cytoskeletal contributions to UDP signaling, we tested inhibitors of microfilament organization. Treatment of RAW264.7 cells with cytochalasin D did not alter the UDP Ca²⁺ response (Fig. 6A), although it inhibited macropinocytosis (Fig. 6B) and phagocytosis (data not shown) and had evident changes on cell morphology (acute reduction in cell spreading within 5 min of addition and maintained throughout the Ca²⁺ assay period; Fig. 6C). Similarly, other inhibitors of microfilament organization (latrunculin A, jasplakinolide) or of microtubule organization (colchicine, nocodazole) did not alter UDP Ca²⁺ responses (Fig. 6A), despite confirmation of their efficacy: Latrunculin A and jasplakinolide caused acute reductions in cell spreading (Fig. 6C), and colchicine and nocodazole treatment arrested cell growth at the G2/M phase of the cell cycle (Fig. 6D). These results indicate that such cytoskeletal organization is not required for maintenance of the Rho-dependent coupling of InsP3 to Ca²⁺ release following UDP signaling. These data indicate that some of the actin-dependent scaffolding functions identified in previous studies [58, 90,91,92,93,94,95,96] do not contribute to Ca²⁺ signaling mechanisms of the macrophage receptors tested in these cells.

Figure 6.

Disruption of cytoskeletal organization does not affect UDP Ca²⁺ responses, and Rho is not activated by UDP or C5aR stimulation in RAW264.7 cells. (A) RAW264.7 cells were treated with cytochalasin D (CytoD; 10 μM), latrunculin A (LatA; 5 μM), jasplakinolide (Jaspl; 5 μM), colchicine (Colc; 25 μM), nocodazole (Noco; 33 μM), or respective control buffers (low percent DMSO or ethanol) for 30 min prior to measurement of Ca²⁺ responses to 500 nM UDP. Similar results were observed for cytochalasin D after treatment for 5 min or 4 h (data not shown). Results shown are the mean ± sem from n = 10–15 replicates from two to five assays/condition. Controls for these results are in panels B–D. (B) Macropinocytosis was assessed in cells treated with cytochalasin D or DMSO (0.1%) as control for 30 min prior to assay. Values shown are normalized to the uptake of FITC-dextran by control samples. (C) Cell morphology was evaluated after 30 min treatment with the inhibitors shown. Values shown are percent of cells with a retracted morphology (mean±sem). (D) Cell-cycle distribution of cells treated with inhibitors overnight. Values shown are percent of cells in phases G1/S versus G2/M (mean±sem). Cell number was reduced by ∼80% (not shown). (E) UDP does not activate Rho. Normal RAW264.7 cells were serum-starved for 1 h and then stimulated with 25 μM UDP, 30 nM C5a, 2.5 μM LPA, or 1 μM S1P in HBSS-BSA. Ligand activation of Rho was assessed at 1 min by G-LISA. Activation of Rho by UDP stimulation was evaluated further by (F) time course and (G) dose response measured at 1 min.

The requirement for Rho in UDP-induced calcium responses does not reflect activation of Rho by UDP

In our studies, the inhibition of Rho by ToxB or other agents required pretreatment of cells. Thus, the requirement for Rho in the Ca²⁺ response to UDP could involve activation of Rho by UDP or instead, depend only on prior or concurrent Rho activity derived from other stimuli. We examined the activation of Rho by UDP or C5a in RAW264.7 macrophages and found no detectable change in Rho activity in response to either ligand (Fig. 6E). In contrast, Rho was activated robustly by LPA or S1P, which activate Gα12/13 as well as Gαq in RAW264.7 cells [97]. To ensure that we examined a broad range of conditions for activation of Rho by UDP, we analyzed the time course and the dose response of Rho to UDP. UDP did not increase Rho activation in RAW264.7 macrophages under any circumstances tested (Fig. 6, F and G), although UDP did activate Rho in 293T cells (data not shown), as reported by others [98]. We conclude that although Rho is important for the Ca²⁺ response to UDP, this requirement is likely met by baseline levels of Rho activity derived from alternate stimuli rather than by UDP-mediated augmentation of Rho activity upon GPCR stimulation.

In sum, with regard to PLCβ4-dependent UDP responses, ToxB inhibition of Rho reduces calcium signaling by interrupting steps between the generation of InsP3 and the recognition of InsP3 by the InsPRs. In contrast, ToxB enhancement of calcium signaling by PLCβ3-dependent C5a involves the generation of increased InsP3, and it thus acts at a different step or steps in the signaling cascade, perhaps at PLCβ3 itself. We therefore examined further pathways in C5a signaling that are altered by ToxB.

The increase in Ca²⁺ responses to C5a in the presence of ToxB is dependent on PLA₂ activity

ToxB is known to have effects on cellular physiology beyond inhibition of Rho family small GTPases. The Rho-inhibitory effects have been distinguished from other cytopathic effects through the use of selectively resistant cell lines and by contrasting the kinetics of different cellular responses [99, 100]. ToxB and ToxA activate PLA₂, which cleaves the sn-2 bonds of phospholipids to release AA and other lysophospholipids, resulting in the production of PGE₂ and LTB4 [99,100,101]. These have direct cytopathic effects on enterocytes, and they contribute to the pathogenesis of pseudomembranous colitis by signaling the release of chemotactic factors for neutrophils and macrophages [2, 14, 102]. We therefore tested the role of cPLA₂ in the elevation of C5a-stimulated Ca²⁺ signaling by ToxB. We first tested three different inhibitors of cPLA₂: AACOCF3, MAFP, and quinacrine. Inclusion of any of these compounds before and during ToxB treatment blocked the elevation of C5a-stimulated Ca²⁺ above that of control cells (Fig. 7, A–C, and data not shown), implicating a product of cPLA₂ in the up-regulation of the Ca²⁺ response to C5a. Although the specificity of any inhibitor may limit interpretation of such results, subsequent experiments also supported a role for AA metabolism in the effect of ToxB on C5a Ca²⁺ responses.

Figure 7.

Elevation of Ca²⁺ responses by ToxB is dependent on AA metabolism. (A and B) RAW264.7 cells were treated with 20 μM AACOCF3 (inhibitor of PLA₂) or vehicle control 1 h before and during a 4-h treatment with 100 ng/ml ToxB. Ca²⁺ responses to 30 nM C5a were then assessed in the continued presence of the inhibitors. (A) Each line represents the mean of five replicate samples/condition. (B) Quantitation of the responses by peak-offset feature. Values are mean ± sem from n = 3 assays; *, P < 0.05. (C) Similar experiments were performed using the PLA₂ inhibitor MAFP at 5 μM. Values shown are mean ± sem of n = 5 replicate samples/condition from a representative assay of six with similar results; *, P < 0.05. (D) Elevation of Ca²⁺ responses by ToxB is dependent on the activity of 5-LOX and not COX. RAW264.7 cells were treated with the COX inhibitor indomethacin (Indo; 50 μM) or the LOX inhibitors NDGA (33 μM), AA861 (15 μM), or MK-886 (MK; 10 μM) or vehicle control 1 h before and during a 4-h treatment with 100 ng/ml ToxB. Ca²⁺ responses to 30 nM C5a were then assessed in the continued presence of the inhibitors. Peak-offset responses were quantified. Shown are the means ± sem from n = 4–8 assays/condition; *, P < 0.01. (E and F) Exposure of RAW264.7 to AA elevates C5a Ca²⁺ responses. RAW264.7 cells were cultured in the presence or absence of 10 μg/ml AA for 20 h prior to Ca²⁺ assay. Cells were then loaded with Fura-2 and assayed for Ca²⁺ responses to 30 nM C5a. (E) Results from an individual assay, in which each line represents the mean ± sem of four replicate samples/condition. (F) Values shown are mean ± sem from n = 8 assays; *, P < 0.01. (G) Elevation of Ca²⁺ responses by AA or ToxB is dependent on LOX activity. RAW264.7 cells were cultured in the presence or absence of NDGA (33 μM) for 30 min prior to addition of 10 μg/ml AA for an additional 4 h prior to Ca²⁺ assay. Cells were then loaded with Fura-2 and assayed for Ca²⁺ responses to 30 nM C5a in the continued presence or absence of NDGA.

5-LOX activity is required for ToxB elevation of Ca²⁺ responses

AA can undergo lipoxygenation, the first step in the formation of LTs, or it may be converted by COX to PGs. LT and PG synthesis in RAW264.7 cells are stimulated by TLR ligands and certain Ca²⁺ agonists [103,104,105,106]. Prior studies have shown that untreated RAW264.7 cells express 5-LOX and FLAP [103]. We similarly found transcripts for 5-LOX and FLAP in our RAW264.7 cells but not transcripts for 12- or 15-LOX (data not shown). To assess the role of these pathways in the action of ToxB on Ca²⁺ signaling, RAW264.7 cells were treated with indomethacin to inhibit COX or with NDGA to inhibit LOX, and Ca²⁺ responses were then assessed. As seen in Figure 7D, indomethacin failed to inhibit the elevation of C5a Ca²⁺ responses, but NDGA blocked the ToxB-induced elevation of responses effectively. Two other 5-LOX-specific inhibitors, AA861 and MK-886, were similarly effective in blocking the ToxB-induced elevation of responses. Thus, the 5-LOX pathway appears necessary to augment C5a Ca²⁺ responses.

As AA is released by the action of PLA₂, we next tested whether the addition of AA would mimic the effect of ToxB in promoting a Ca²⁺ response. RAW264.7 cells were loaded with AA in culture overnight [44] and then stimulated with C5a. This exposure yielded elevated C5a Ca²⁺ responses (Fig. 7, E and F), similar to those observed with ToxB treatment. To determine if this AA effect shared a similar requirement for LOX activity, we treated cells with NDGA prior to and during AA exposure. NDGA inhibited the elevation of C5a Ca²⁺ responses induced by AA exposure, comparable with its inhibition of ToxB effects. Thus, AA released by cPLA₂ activity associated with ToxB treatment may be sufficient to induce the changes in GPCR Ca²⁺ signaling observed. In all, our results support the hypothesis that ToxB enhances Ca²⁺ responses to C5a in a manner that is selective for PLCβ3 and that is dependent on the generation of the LTs but not PGs.

DISCUSSION

We have demonstrated for the first time selective and opposing effects of C. difficile ToxB on different GPCR signaling pathways. The results reveal notable differences in the regulation by ToxB of PLCβ3 versus PLCβ4 pathways, and they open new areas for study in the regulation of the clinical effects of ToxB. ToxB effects have often been attributed to its known inhibition of Rho family GTPases. In our studies, ToxB effects on PLCβ4-dependent Ca²⁺ signaling pathways reflected a requirement for Rho, and the contribution of Rho was shown to be at the level of InsP3/Ca²⁺ coupling. In contrast, ToxB enhanced Ca²⁺ signaling by PLCβ3-dependent pathways in a manner that was independent of Rho, and it depended instead on cPLA₂ and lipoxygenation.

The effects of ToxB on PLCβ4-dependent Ca²⁺ signaling reveal new requirements for RhoA-like proteins in this pathway. Some of the mechanisms described previously by which Rho family members may regulate Ca²⁺ responses are associated with a defect in proximal signaling, yielding a reduction in net InsP3 production. In some studies, this reflected a reduction in cellular PIP₂ supply [25, 107, 108]. In our studies, however, PIP₂ was sustained in ToxB-treated cells, as was InsP3 production, and thus, the activity of PLCβ4 itself appears unaffected. We also observed normal p38 activation by UDP (data not shown), further supporting maintenance of proximal signaling by the UDP GPCR in the presence of ToxB. The distal signaling step of InsP3-stimulated release of Ca²⁺ from intracellular stores was also unaffected. Thus, the contribution from Rho appears distal to InsP3 production and proximal to release of Ca²⁺ from intracellular stores. A reduction in exposure of the ER to generated InsP3 could explain this uncoupling of InsP3 from Ca²⁺. Our findings indicate that a RhoA-like protein is required for coupling of responses through PLCβ4.

Full inhibition of UDP Ca²⁺ signaling in RAW264.7 cells by ToxB required pretreatment for ≥4 h; shorter treatments were less effective. Coupled with the observations that the effect of ToxB on UDP Ca²⁺ is dependent on Rho, but UDP does not activate Rho, these results imply that Rho intrinsically maintains cellular mechanisms for coupling PLCβ4-mediated InsP3 signaling to InsP3Rs in the ER. The baseline activity of Rho derived from serum factors, adhesion signals, and other stimuli [109, 110] under normal culture conditions may thus provide for normal Ca²⁺ responses.

Reduced coupling of InsP3 production to Ca²⁺ release from intracellular stores for a distinct subset of GPCRs suggests a change in the location of a relevant component as the most likely explanation for the effects of ToxB via Rho [18, 23, 24, 92, 111]. We have attempted to identify such changes in the distribution of the sites of production of InsP3 (GPCR, Gα, PLCβ) in relation to the sites of action of InsP3 (InsP3R, ER), but our studies have not revealed distinct distributions or consistent changes so far. Some gross morphological changes occur with ToxB exposure, but these are not reproduced by inhibition of Rho through other means, which similarly perturb Ca²⁺ responses, so these morphologic changes appear unrelated to effects of ToxB on Ca²⁺ signaling. Nonetheless, as UDP-stimulated InsP3 production is normal in the presence of ToxB, and exogenous InsP3 stimulates normal Ca²⁺ release in the presence of ToxB, differences in the intracellular distribution of InsP3 likely account for the effects of ToxB on UDP signaling. Although the cytoskeleton plays an important role in organization of Ca²⁺ signaling complexes for some Gαq-linked GPCRs in neurons [24, 92], our data suggest a lesser contribution from F-actin to signaling by UDP receptors in macrophages. In sum, we have narrowed the mechanism for the effects of ToxB on UDP signaling to disruption of the delivery of InsP3 signals leading to the release of Ca²⁺ from the ER, but the mechanism of this disruption remains to be explored.

A contribution of RhoA-like proteins, but not Rac or Cdc42, to the ToxB Ca²⁺ phenotypes is supported by the selectivity of the inhibitors used. Although inhibition of Rho reproduced the UDP phenotype, inhibition of Rac, Cdc42 or the broad Ras superfamily did not reproduce the effect of ToxB on the C5a phenotype. An inverse relationship can exist between the activities of some family members, e.g., Rho and Rac [112, 113], but our methods, which did not prevent reciprocal up-regulation, failed to elevate C5a Ca²⁺ responses. We did not exhaustively test all members of the Rho family targeted by ToxB but can dismiss any factors reciprocally activated by inhibition of Rho, Rac, or Cdc42. Thus, although changes in Rho activity are sufficient to explain the depression of UDP Ca²⁺ responses, the elevation of C5a responses does not appear linked to ToxB effects on this family of G-proteins.

In contrast to the inhibitory effects of ToxB on PLCβ4-mediated Ca²⁺ signaling, ToxB enhanced PLCβ3-mediated Ca²⁺ signaling through enhanced production of InsP3. This selective effect of ToxB on PLCβ3 activation required cPLA₂ activity and lipoxygenation of AA. In macrophages, C5a signals primarily through PLCβ3 [49], so ToxB enhances the Ca²⁺ response to C5a. Gαi-linked receptors, such as the C5aR, activate PLCβ2 and PLCβ3 through the release of Gβγ, and the selective use of PLCβ3 by the C5aR in macrophages is not yet understood nor is the selective enhancement of PLCβ3 signaling by ToxB. ToxB does not promote effects of Gβγ globally, as ToxB did not enhance PLCβ2 signaling, and it inhibited rather than enhanced the activation of PI3K, another effector of Gβγ signaling (data not shown). Our studies do not, however, exclude the possibility that ToxB selectively regulates the contribution of Gβγ to these signaling pathways. It remains unclear if the PLA₂-dependent enhancement of PLCβ3 is dependent on the glucosyltransferase activity of ToxB toward unidentified members of the Rho family or if its effect on PLA₂ is independent of Rho family involvement.

Several possible mechanisms for this enhancement of PLCβ3 activity were evaluated, but none was supported by our experimental results. Possible changes in subcellular distribution of PLCβ3 versus PLCβ4 were examined, but we found no discernible changes to explain the enhanced PLCβ3-dependent responses (data not shown). The sufficiency of AA exposure and requirement for 5-LOX activity focused our attention on a short list of likely LT mediators, several of which can couple to Gαq-linked GPCRs. Previous studies of C5aR synergy with Gαq-coupled GPCRs [49] showed a similar elevation of PLCβ3-mediated InsP3 production and reduced phosphorylation of AKT, and LT-stimulated GPCRs could provide synergistic elevation of PLCβ3-mediated responses to C5a. Thus, we evaluated the effect of pre-exposure and simultaneous exposure to the LTs B4, D4 (same receptor as C4), 5-HETE, and 5-oxo-HETE at varied doses and times on C5a responses but observed no significant enhancement of C5a responses. 5-LOX activity may be necessary but not sufficient for the ToxB effect, and other AA metabolites may contribute to provide for enhanced PLCβ3 responses.

The dichotomous effects of ToxB on GPRC-mediated Ca²⁺ signaling in macrophages provide new insights into the mechanism of Ca²⁺ signaling in these cells and may have consequences for the pathogenicity of C. difficile. The Rho-mediated effects of ToxB on UDP Ca²⁺ signaling may be seen as part of a broad attack by C. difficile on macrophage function. In macrophages, the Rho family influences cell adhesion, cell migration, phagocytosis, macropinocytosis, immunoreceptor activation, and the production of reactive oxygen species [9, 10, 54]. Inhibition of these processes in macrophages (and neutrophils) may provide clear advantages to C. difficile toward immune evasion. Similarly, reduction in Rho-dependent Ca²⁺ signals from some GPCRs may impede the host response. The enhancing effects of ToxB on C5a Ca²⁺ signaling, however, present a less-obvious advantage to the pathogen and could instead reflect host responses to C. difficile or a combination of these competing goals. Thus, the generation of AA metabolites may potentiate the inflammatory response to facilitate pathogen clearance even as it exacerbates tissue injury. cPLA₂ activity and AA can confer resistance to apoptotic stimuli in macrophages [114,115,116], and PLCβ3-deficient macrophages are more susceptible to apoptosis [117], suggesting that PLCβ3 imparts antiapoptotic effects in these cells. Thus, generation of AA may improve macrophage survival and facilitate host defense. Through other mechanisms, however, AA can also contribute to apoptotic responses [118, 119], so the net effect is uncertain. It would be interesting to assess the resistance of PLCβ3-deficient mice to C. difficile infection to identify if these elevated responses benefit host or pathogen.

In sum, our studies of macrophage signaling reveal selective effects of ToxB and of Rho on different GPCR signaling pathways, demonstrating that ToxB, by inhibiting Rho, selectively inhibits Ca²⁺ responses mediated by PLCβ4 but not PLCβ3. The requirement for Rho in PLCβ4 activity is shown to lie at the level of Ca²⁺ release by InsP3, implying that PLCβ4 differs mechanistically from PLCβ3 in promoting this step. In contrast to its inhibition of PLCβ4, ToxB promotes the activation of PLCβ3 by the Gαi-coupled C5aR, through a pathway that is dependent on the metabolism of AA to LTs. These differential effects of ToxB on PLCβ3 versus PLCβ4 pathways elucidate new regulatory pathways for signal transduction in macrophages, and they may lead to new targets for blocking the clinical effects of ToxB.

AUTHORSHIP

Robert A. Rebres: project design, statistical analyses, authorship, supervision, cell culture, expression constructs and transfections, population and single-cell Ca²⁺ assays, Western blots, Rho assays, and cytoskeletal assessments; Christina Moon: cell culture, transfections, population Ca²⁺ assays, Western blots, InsP3 analyses, and phosphatidylinositol analyses; Dianne DeCamp: InsP3 analyses; Keng-Mean Lin: expression constructs and InsP3 analyses; Iain D. Fraser: project design, cell culture, and expression constructs; Stephen B. Milne: phosphatidylinositol analyses; Tamara I. A. Roach: authorship and supervision; H. Alex Brown: project design and supervision; and William E. Seaman: project design, authorship, and supervision.

ACKNOWLEDGMENTS

This work was supported by National Institutes of Health grant GM 62114. We thank Amanda Norton (San Francisco Veterans Affairs Medical Center) and Jamie Liu, Leah Santat, and Lucas Cheadle (Caltech) for excellent technical assistance.