The chaperone Hsp90 and PPIases of the cyclophilin and FKBP families facilitate membrane translocation of Photorhabdus luminescens ADP-ribosyltransferases

Authors


Summary

TccC3 and TccC5 from Photorhabdus luminescens are ADP-ribosyltransferases, which modify actin and Rho GTPases, respectively, thereby inducing polymerization and clustering of actin. The bacterial proteins are components of the Photorhabdus toxin complexes, consisting of the binding and translocation component TcdA1, a proposed linker component TcdB2 and the enzymatic component TccC3/5. While the action of the toxins on target proteins is clearly defined, uptake and translocation of the toxins into the cytosol of target cells are not well understood. Here we show by using pharmacological inhibitors that heat shock protein 90 (Hsp90) and peptidyl prolyl cis/trans isomerases (PPIases) including cyclophilins and FK506-binding proteins (FKBPs) facilitate the uptake of the ADP-ribosylating toxins into the host cell cytosol. Inhibition of Hsp90 and/or PPIases resulted in decreased intoxication of target cells by Photorhabdus toxin complexes determined by cell rounding and reduction of transepithelial electrical resistance of cell monolayers. ADP-ribosyltransferase activity of toxins and toxin-induced pore formation were notimpaired by the inhibitors of Hsp90 and PPIases. The Photorhabdus toxins interacted with Hsp90, FKBP51, Cyp40 and CypA, suggesting a role of these host cell factors in translocation and/or refolding of the ADP-ribosyltransferases.

Introduction

Photorhabdus luminescens is an entomopathogenic bacterium, which is mutualistically associated with nematodes. The nematodes invade insect larvae and eventually kill the insects by means of bacterial toxin. One of the major toxins produced by P. luminescens is the toxin complex toxin Tc. This toxin consists of the three components TcA, TcB and TcC, each of which occurs in several homologous isoforms. TcA forms pentamers and is involved in host cell binding and toxin translocation. While TcB is probably a functional linker between TcA and TcC, the latter component TcC harbours the biological activity. Two homologues of TcC named TccC3 and TccC5 have been studied in detail recently (Lang et al., 2010). Both Tc components possess ADP-ribosyltransferase activity. While TccC3 ADP-ribosylates actin at threonine-148 to induce actin polymerization, TccC5 modifies Rho proteins at glutamine-63/61 resulting in persistent activation of the switch proteins. The combined action of TccC3 and TccC5 causes actin clustering in host cells and blockade of immune responses of the host (Lang et al., 2010). While the ADP-ribosylation of host target proteins by bacterial toxins and its functional consequences are well understood, the precise mechanism of the translocation of the toxin into the cytosol is still largely enigmatic. Some ADP-ribosylating toxins like the binary actin modifying C. botulinum C2 toxin and C. perfringens iota toxin are transported into the cytosol by means of a heptameric translocation system (Blöcker et al., 2001; Schleberger et al., 2006), which is largely related to the protective antigen of anthrax toxin (Young and Collier, 2007). The characteristic feature of the transport system is the formation of a pore in membranes of acidified endosomal vesicles on the basis of a β-barrel structure, which serves as translocation channel for the unfolded enzyme components (Young and Collier, 2007). Diphtheria toxin, another ADP-ribosylating toxin, translocates across endosomal membranes by a different system, which does not depend on a separate pore-forming moiety (Ren et al., 1999). In this case the membrane crossing of the toxin depends on a built-in translocation domain which forms functional pores by using helical structures. Recently, the structure of TcA, the binding and translocation component of P. luminescens Tc complex toxin has been elucidated by cryo electron microscopy (Gatsogiannis et al., 2013). These data show a novel type of bacterial microinjection machinery. TcA forms a pentamer with a central channel part consisting of 25 helices with a minimal diameter of ∼ 1.5 nm. This structure is surrounded by a shell, which appears to be largely rearranged during membrane insertion, thereby undergoing a movement like the injection of a syringe (Gatsogiannis et al., 2013).

While a role of intracellular factors in toxin uptake has been suggested by Sandvig and Olsnes (1981) for diphtheria toxin more than 30 years ago, more recent studies suggest that cytosolic chaperones are crucially involved in translocation of ADP-ribosylating toxins. First, it was shown that heat shock protein (Hsp) 90 plays an essential role in membrane translocation of diphtheria toxin (Ratts et al., 2003) and binary actin-ADP-ribosylating toxins (Haug et al., 2003; 2004). Later, it turned out that cyclophilin A participates in this action (Kaiser et al., 2009; Dmochewitz et al., 2011). The final progress was the finding that also FK506-binding proteins (e.g. FKBP51) take part in translocation (Kaiser et al., 2012). Cyclophilins and FKBPs are peptidyl prolyl cis/trans isomerases (PPIases), which accelerate the cis/trans isomerization of prolyl peptide bonds thereby assisting chaperones in refolding of proteins (Fischer et al., 1984; 1989; Lang et al., 1987; Schmid, 1993; Schmid et al., 1993; Wang and Heitman, 2005).

So far the precise role of these chaperones and PPIases during membrane transport of the ADP-ribosylating toxins is not known. Of special interest is the question of whether these factors act in concert with the translocation machinery of the toxin or separately. To further clarify this question, we studied the uptake of P. luminescens Tc toxins, which are delivered into host cells apparently but employ a novel type of injection machinery. Here, we report that the cellular uptake and cytotoxicity of the ADP-ribosylating toxins TccC3 and TccC5 is inhibited by radicicol (Rad), cyclosporin A (CsA, Handschumacher et al., 1984) and FK506 (tacrolimus, for review see Galat, 2003), indicating dependence of these toxins on the chaperone Hsp90 and PPIases of the cyclophilin and FKBP families respectively.

Results

Pharmacological inhibition of host cell Hsp90 protects HeLa cells from intoxication with PTC3

We used the specific pharmacological inhibitors Rad and the geldanamycin derivative 17-DMAG (Roe et al., 1999) to investigate whether the cellular uptake of the ADP-ribosylating P. luminescens toxins depends on the activity of the host cell chaperone Hsp90. Therefore, HeLa cells were pretreated with the inhibitors followed by the addition of the Photorhabdus toxin complex 3 (PTC3). PTC3 consists of the binding and translocation component TcdA1, the component TcdB2 and the enzymatic active component TccC3, which leads to the ADP-ribosylation of actin at T148. As shown in Fig. 1A, treatment of cells with PTC3 induced cell-rounding due to the redistribution of the actin cytoskeleton. The pretreatment of PTC3-intoxicated cells with Hsp90-inhibitors resulted in decreased cell-rounding (Fig. 1A, upper panel). To analyse in more detail which part of TcdB2-TccC3 is affected by inhibition of Hsp90, the His-tagged ADP-ribosyltransferase domain hvr of TccC3 was applied to cells in combination with the B. anthracis transport component PA (Fig. 1A, lower panel). Because the translocation of His-TccC3hvr into the cytosol is mediated by PA, the interaction of host cell factors with the isolated ADP-ribosyltransferase domain can be investigated independently of the other toxin complex proteins (Blanke et al., 1996; Lang et al., 2010). Again, pretreatment of HeLa cells with Rad or 17-DMAG significantly delayed the intoxication of cells with PA plus His-TccC3hvr, implying that only the ADP-ribosyltransferase domain of TccC3 is crucial for the interaction with Hsp90 during PA-mediated membrane transport (Fig. 1A, lower panel). To verify that this interaction indeed occurs via the ADP-ribosyltransferase domain of TccC3, and not the His-tag, the same translocation experiment was performed with the His-tagged glucosyltransferase domain of Clostridium difficile toxin B (amino acids 1–543) instead of His-TccC3hvr. Full-length toxin B also delivers its enzyme domain from acidified endosomal vesicles into the host cell cytosol, but in contrast to the ADP-ribosylating toxins, this translocation step was not inhibited by Rad (Haug et al., 2003). As expected, pre-incubation of HeLa cells with Rad had no inhibitory effect on cell rounding induced by PA/His-toxin B1–543 (Fig. S1). This indicates that Hsp90 is involved in the PA-dependent transport of ADP-ribosyltransferases but is not required for the PA-mediated transport of His-tagged cargo proteins per se. To investigate the inhibitory effect of Rad and 17-DMAG in more detail, we measured the transepithelial resistance (TER) of CaCo-2 cell monolayer after intoxication by PTC3 (Fig. 1B) or PA plus His-TccC3hvr (Fig. 1C). In both experiments, inhibition of Hsp90 by Rad or 17-DMAG reduced the TccC3-mediated intoxication over time (Fig. 1B and C) and also in a dose-dependent manner (Fig. S2).

Figure 1.

Influence of Hsp90 on the uptake of P. luminescens ADP-ribosyltransferase TccC3/5.

A. HeLa cells were intoxicated with PTC3 (2 μg ml−1 TcdA1 and 1 μg ml−1 TcdB2-TccC3) in the presence of 10 μM Rad or 17-DMAG (upper panel). 0.25 μg ml−1 of the ADP-ribosyltransferase domain of TccC3 (His-TccC3hvr) was delivered by 0.5 μg ml−1 of anthrax PA into HeLa cells in the presence of 10 μM Rad or 17-DMAG (lower panel). Pictures were taken after 3 h for PTC3 and after 1 h for PA-translocated His-TccC3hvr (scale bar, 80 μm).

B–D. Intoxication of human colon carcinoma (CaCo-2) cells with P. luminescens ADP-ribosyltransferase TccC3/5. CaCo-2 cell monolayers were treated with PTC3 (800 ng ml−1 of TcdA1 and 400 ng ml−1 of TcdB2-TccC3) (B) or with 250 ng ml−1 of PA and 125 ng ml−1 of His-TccC3hvr (C) in the presence or absence of 10 μM Rad or 17-DMAG. Control (◊), toxin alone (■), toxin plus Rad (Δ), toxin plus 17-DMAG (●). Reduction of transepithelial resistance (TER) due to cell rounding over time (min) served as a parameter for intoxication. (D) Cells were pretreated with 10 μM Rad followed by the addition of PTC5 (800 ng ml−1 of TcdA1 and 400 ng ml−1 of TcdB2-TccC5). TER was measured after 3 to 7 h. All results are from mean values ± SD of at least three independent experiments and are shown as percentages of starting TER. Significance was tested by using the Student's t-test. Values refer to samples treated with toxin only (*P < 0.05, ns = not significant).

To test that the inhibitory effect of Rad is not only specific for Photorhabdus TccC3, we repeated the TER experiments with PTC5 (TcdA1, TcdB2 and TccC5), which leads to ADP-ribosylation of Rho GTPases, thereby inducing their activation (Lang et al., 2010). As shown in Fig. 1D, PTC5 induced the reduction of TER of Caco-2 cell monolayer due to changes of cell morphology and cell-cell contact induced by persistently activated Rho GTPases. Again, the addition of Rad reduced the TccC5-mediated intoxication, which argues for a role of Hsp90 in the uptake of this ADP-ribosyltransferase into cells.

Role of PPIase activities of cyclophilins and FKBPs for the mode of action of TccC3

Recently, we found that the translocation of actin-ADP-ribosylating toxins from several Clostridium species is facilitated by cyclophilin A and FK506-binding protein 51 (Kaiser et al., 2011; 2012). To test whether the translocation of TccC3 depends on the activities of cyclophilins and FKBPs, HeLa cells were pretreated with CsA and FK506 and challenged with the three components of PTC3, the transport component TcdA1 plus the linker/enzyme component TcdB2-TccC3. To monitor the intoxication of cells, we analysed the characteristic toxin-induced blebbing after 1 h (Fig. 2A) as well as subsequent cell-rounding after 4 h (Fig. 2B). As shown in Fig. 2A and B, treatment of cells with each inhibitor resulted in a decreased number of intoxicated cells. The protective effect of Rad was most prominent as compared with CsA and FK506. The inhibitors alone had no effects on cell morphology under these conditions (not shown) but since we observed in earlier studies that higher concentrations of CsA and FK506 might affect the morphology of HeLa cells, we used not more than 20 μM final concentration of the inhibitors in the present study.

Figure 2.

Rad, CsA and FK506 protect HeLa cells from intoxication by PTC3 and by B. anthracis PA-delivered His-TccC3hvr.

A. HeLa cells were pre-incubated with Rad, CsA or FK506 (20 μM) for 30 min at 37°C. Subsequently, PTC3 (2 μg ml−1 TcdA1 plus 1 μg ml−1 TcdB2-TccC3) was added and cells were further incubated at 37°C. For control (con) cells were left untreated. As further control, cells were treated with the toxin in the absence of inhibitor. To monitor the toxin-induced morphological changes, pictures were taken after 1 h when blebbing of cells (see arrows) was most obvious (A) and after 4 h when cells rounded up (B) (scale bars, 50 μm). Magnification of blebbing of PTC3-treated cells from inset is indicated as mag.

B. For quantitative analysis the percentage of rounded cells was determined from the pictures. Values are given as mean ± SD (n = 3). Significance was tested by using the Student's t-test. Values refer to the samples treated with toxin only (*P < 0.05, ***P < 0.001, ns = not significant).

C. Time-course of intoxication of HeLa cells with PTC3 in the presence of Rad, CsA and FK506. HeLa cells were treated exactly as described above and the percentage of rounded cells was determined. Values refer to the samples treated with toxin only (n = 3, *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant).

D. CsA and FK506 inhibit intoxication of HeLa cells in a synergistic manner. HeLa cells were pre-incubated for 30 min at 37°C either with CsA or FK506 or with the combination of both inhibitors (20 μM each). Subsequently, the toxin was added for 5 h and the percentage of rounded cells was determined as described above. If not indicated by lines, values refer to the samples treated with toxin only (n = 3, **P < 0.01, ***P < 0.001, ns = not significant).

E. Effect of CsA and FK506 on the intoxication of HeLa cells with His-TccC3hvr plus PA. Cells were pretreated for 30 min at 37°C with CsA or FK506 (10 μM), His-TccC3hvr (50 ng ml−1) plus PA (100 ng ml−1) were added and cells were further incubated at 37°C. The toxin-induced cell-rounding was determined after 3.5 and 4.5 h. Significance was tested by using the Student's t-test. Values refer to samples treated with toxin only (n = 3; *P < 0.05, **P < 0.01, ***P < 0.001).

F. HeLa cells were pretreated with 20 μM of the indicated inhibitors for 1 h at 37°C. Thereafter, PTC3 was added and cells were further incubated for 3 h. Cells were then washed with cold PBS and lysed. Samples (25 μg each) were subjected to an in vitro ADP-ribosylation assay for 15 min at 21°C with 1 μM radioactive [32P]NAD+ and 500 nM His-TccC3hvr (20 ng μl−1). Proteins were detected by SDS-PAGE and radiolabelled actin was visualized by autoradiography (upper panel). Lower panel shows the quantification of post-ADP-ribosylated actin. Values are given as mean ± SD (n = 3) and significance was tested by using the Student's t-test (**P < 0.01, ns = not significant; values refer to the samples treated with toxin only).

A more detailed analysis of these protective effects revealed that each inhibitor caused a time-dependent delay in intoxication with PTC3 (Fig. 2C). In contrast to Rad and CsA, the inhibitory effect of FK506 on intoxication with PTC3 in these experiments was not significant. However, when FK506 was applied to cells in combination with CsA, a much stronger protective effect towards PTC3 intoxication compared with the individual inhibitors was observed (Fig. 2D). While CsA reduced PTC3-induced cell rounding from ∼ 85 to ∼ 65% and FK506 not significantly to about 75%, the combination of CsA and FK506 caused a reduction in toxin-induced cell rounding to ∼ 30%. These findings suggest a synergistic action of CsA and FK506 in inhibiting the intoxication by PTC3, possibly by interfering with a cyclophilin/FKBP-containing Hsp90-multi-chaperon complex (Pirkl and Buchner, 2001). The solvent alone had no inhibitory effect on the intoxication process under these conditions (not shown). Moreover, pretreatment of cells with CsA or FK506 significantly decreased the number of round cells after incubation with His-TccC3hvr plus PA (Fig. 2E), suggesting that cyclophilins and FKBPs facilitate PA-mediated membrane translocation of His-TccC3hvr. Again, we confirmed that this interaction specifically occurred via the ADP-ribosyltransferase domain of TccC3, and not the His-tag by performing the translocation experiment with PA/His-toxin B1–543 in the presence of CsA and FK506 as described for Rad before (see Fig. S1). Noteworthy, CsA and FK506 do not inhibit the uptake of the native glucosylating toxins from C. difficile into the host cell cytosol (Kaiser et al., 2009; 2012).

The results obtained by morphology-based analysis of cell protection (e.g. cell rounding and TER assay) were confirmed by biochemical analysis using the ADP-ribosylation status of actin as an end-point for PTC3-induced intoxication of cells. Therefore, HeLa cells were treated for 3 h with PTC3 in the presence and absence of Rad, CsA or FK506. Then, cells were lysed and lysates subjected to in vitro ADP-ribosylation with fresh ADP-ribosylating toxin component and radio-labelled NAD+ as co-substrate. Radio-labelled actin was detected by autoradiography and the intensity of the bands was quantified (Fig. 2F). In this assay, actin from control cells exhibited a strong signal, because it was not ADP-ribosylated during the initial incubation period. In contrast, treatment of cells with PTC3 resulted in a weak signal, because most of the actin was already modified during incubation of the intact cells by PTC3. Therefore, this portion of actin could not serve as a substrate for subsequent in vitro ADP-ribosylation. Like in Fig. 2C, the inhibitory effect of FK506 on the intoxication with PTC3 was not significant. When cells were incubated with PTC3 in the presence of Rad or CsA, increased labelling of actin was detected in the post-ADP-ribosylation assay, implying less active ADP-ribosyltransferase of PTC3 in the cytosol as in the absence of inhibitors. Taken together, the results indicate that besides the chaperone activity of Hsp90 the PPIase activities of cyclophilins and FKBPs play a role during intoxication of cells with PTC3.

Rad, CsA and FK506 have no effects on the ADP-ribosyltransferase activity of TccC3 and the early steps of PTC3 uptake into cells

To get more insights into the role of host cell factors like Hsp90 and PPIases during cellular uptake of PTC3, we analysed the molecular mechanisms underlying the inhibitory effects of Rad, CsA and FK506 in more detail. The protective effect of the inhibitors used could be due to inhibition of the ADP-ribosyltransferase activity of TccC3 or to blockade of the uptake of the toxin into the cytosol. First we excluded that Rad, CsA or FK506 inhibited the TccC3-catalysed ADP-ribosylation of actin in vitro (Fig. 3A). Addition of Rad, CsA or FK5006 did not alter the ADP-ribosylation of actin in vitro. These data rather suggested that the inhibitors interfere with the TcdA1-mediated uptake of TcdB2-TccC3 into the host cell cytosol. Using two different approaches, we also excluded an inhibitory effect of Rad, CsA and FK506 on binding of PTC3 to receptors on the cell surface. First, HeLa cells were incubated with PTC3 in the absence and presence of each inhibitor at 4°C, a procedure, which does not allow endocytosis of toxins. Then, cells were washed, lysed and employed in an ADP-ribosylation assay. With and without inhibitors, a similar amount of actin was ADP-ribosylated, suggesting that the same amount of active toxin was initially bound and the inhibitors did not interfere with receptor binding and/or cell membrane interaction of PTC3 (Fig. 3B). Second, we performed FACS experiments. To this end, HeLa cell suspensions were incubated with the inhibitors (Rad, CsA and FK506) for 1 h at 37°C, prior to the addition of DyLight488-labelled TcdA1 (TcdA1DL488), followed by incubation of the cells for 20 min at 4°C to prevent endocytic uptake. Importantly, mock- and inhibitor-pretreated cells exhibited an identical amount of cell surface-bound fluorescence, indicating that binding of TcdA1 to cell surface receptors was not affected by Rad, CsA or FK506 (Fig. 3C). In addition, we applied the same FACS approach with HeLa cells suspensions to test whether the inhibitors influence the binding of DyLight488-labelled TcdB2-TccC3 to cell-surface bound, non-labelled TcdA1. As expected, efficient binding of fluorescence-labelled TcdB2-TccC3 to the cell surface occurred only upon pre-addition of TcdA1 to the cells, and this interaction was not influenced by pretreatment of cells with the inhibitors (Fig. S4).

Furthermore, we studied whether the inhibitors of Hsp90 and PPIases had any effects on the pore formation induced by TcdA1. Therefore, 86Rb ion release from preloaded HT-29 cells induced by TcdA1 was determined in the presence of the inhibitors. Pore formation by the toxin and subsequent release of 86Rb ions was induced by a pH shift to pH 5.0, which mimics endosomal pore formation at the plasma membrane (Fig. 3D). All inhibitors did not affect release of 86Rb ions, indicating that the pH-dependent pore formation in cell membranes was not altered by the compounds and occurred independently from Hsp90 and the PPIases.

Rad, CsA and FK506 inhibit the pH-dependent membrane translocation of TcdB2-TccC3

Prompted by our recent findings with clostridial toxins, we expected that inhibitors of Hsp90 and PPIases interfere with intracellular membrane translocation of TcdB2-TccC3 from acidified endosomes to the cytosol (Kaiser et al., 2011; 2012). To test this hypothesis, we first established the pH-dependent translocation assay with intact cells for PTC3. To this end, HeLa cells were pretreated with Baf A1 to inhibit the v-ATPase and thereby endosomal acidification. In Baf A1-treated cells, the ‘normal’ uptake of TcdB2-TccC3 into the cytosol from acidified endosomal vesicles is inhibited and therefore, cells do not round up (Fig. 4A). Baf A1-treated cells were exposed to PTC3 (TcdA1 + TcdB2-TccC3) at 4°C to enable binding of the toxin to the cell surface. Then, cells were either exposed to warm, acidic medium (37°C, pH 4.5, Baf A1) or to warm, neutral medium (37°C, pH 7.5, Baf A1). Subsequently, all cells were further incubated in neutral medium at 37°C in the presence of Baf A1. Only cells, which were exposed to low pH, rounded up (Fig. 4A), indicating that enzymatic active TcdB2-TccC3 (or an active part of it) reached the cytosol of the cells. Noteworthy, the acidic medium alone had no effect on cell morphology. These results clearly demonstrate that acidic conditions are essential to trigger translocation of the active ADP-ribosyltransferase TccC3 across the cytoplasmic membrane to the host cell cytosol, as described by us and others for several bacterial toxins earlier (Milne and Collier, 1993; Barth et al., 2001; Giesemann et al., 2006). When this assay was performed with cells pretreated with Rad, CsA or FK506, significantly fewer cells rounded up within 3.5 h compared with cells treated with PTC3 in the absence of inhibitors (Fig. 4B). This finding implied that Hsp90, cyclophilins and FKBPs were crucial for pH-dependent uptake of TccC3 in living cells and suggested that these host cell factors facilitate translocation of functionally active TccC3 ADP-ribosyltransferase across membranes using the specific translocation component TcdA1.

In cells, TccC3hvr interacts with Hsp90, CypA, Cyp40 and FKBP51

To study whether the isolated ADP-ribosyltransferase domain TccC3hvr interacts directly with Hsp90 and immunophilins CypA, Cyp40 and FKBP51, we performed pull-down experiments. Noteworthy, we recently identified these host cell factors as binding partners of the enzyme moieties of clostridial ADP-ribosylating toxins (Kaiser et al., 2011; 2012). We made use of the fact that the His-tagged toxin is readily transported by PA and, on the other hand, can be easily precipitated and concentrated by binding to cobalt-linked beads. To this end, HeLa cells were incubated with PA plus His-TccC3hvr or left untreated for control. Subsequently, the cells were lysed and cytoplasmic fractions obtained by centrifugation. For pull-down of His-TccC3hvr, the cytoplasmic fractions were incubated with Talon™ CellThru beads. After washing, the beads were boiled in SDS-sample buffer to remove the precipitated proteins which were then separated by SDS-PAGE. The co-precipitated host cell proteins were identified by Western blot analysis with specific antibodies. As shown in Fig. 5, Hsp90, CypA, Cyp40 and FKBP51 co-precipitated with His-TccC3hvr in lysates from cells treated with PA plus His-TccC3hvr. In contrast, only negligible amounts of these proteins were detected in co-precipitates from control cells. To demonstrate the specific interaction between His-TccC3hvr and these host cell factors, we confirmed by Western blotting that other host cell proteins such as RhoA did not co-precipitate with His-TccC3hvr from cells treated with PA plus His-TccC3hvr (not shown). Western blot analysis of the amount of actin from the cell lysates used as input for the co-precipitation confirmed that comparable protein amounts were analysed for toxin-treated and untreated cells (Fig. 5, right panel). Moreover, we confirmed by Western blotting that the toxin-treatment had no effect on the overall cellular amounts of Hsp90, CypA, Cyp40 and FKBP51 (not shown).

Figure 3.

Effect of Rad, CsA and FK506 on ADP-ribosylation of actin by Photorhabdus TccC3, on binding of PTC3 to cells and pore formation of TcdA1.

A. Rad, CsA and FK506 do not interfere with PTC3-catalysed ADP-ribosylation of actin in vitro. HeLa lysate (40 μg) was incubated for 30 min at 37°C with Rad, CsA or FK506 (20 μM) or without inhibitor for control. Subsequently, TcdB2/TccC3 (500 ng) and biotin-labelled NAD+ (10 μM) were added and the samples were incubated for further 15 min at 37°C. Samples were subjected to SDS-PAGE, blotted and the biotin-labelled, i.e. ADP-ribosylated actin was detected with streptavidin-POD and the ECL system (upper panel). The amount of total actin is shown for loading control (lower panel).

B. Rad, CsA and FK506 do not inhibit binding of PTC3 to cells. HeLa cells were pre-incubated for 30 min at 37°C with Rad, CsA or FK506 (20 μM) or left untreated for control. Subsequently, the cells were cooled to 4°C, PTC3 (2 μg ml−1 TcdA1 plus 1 μg ml−1 TcdB2-TccC3) was added to enable binding to the cell surface and cells were incubated at 4°C for further 30 min. After washing twice with ice-cold PBS, cells were harvested in ADP-ribosylation buffer and lysed by using a syringe. In vitro ADP-ribosylation of actin was performed exactly as described in A to detect the cell-bound PTC3 via its enzyme activity. Comparable amounts of total protein were confirmed by Ponceau S staining (see Fig. S3.).

C. Rad, CsA or FK506 do not inhibit binding of TcdA1 to cells. HeLa cell suspensions (1 × 105 cells in 1 ml PBS) were pre-incubated for 1 h at 37°C with 20 μM Rad (+ Rad, left panel), 20 μM CsA (+ CsA, right panel), 20 μM FK506 (+ FK506, right panel) or left untreated (− Rad; − Inh). Following the addition of 0.8 μg ml−1 of DyLight488-labelled TcdA1 (TcdA1DL488) and incubation of the cells for 20 min at 4°C, or without addition of labelled TcdA1 (mock), cells were washed twice in PBS and applied to FACS analysis. Results are presented as histogram plots, where single cell events are plotted against cell surface-bound fluorescence (Log FL intensity).

D. 86Rb+-loaded HT-29 cells were pre-incubated without or with 20 μM of the indicated inhibitors for 1 h at 37°C. Afterwards, cells were washed twice with PBS and TcdA1 (0.5 μg ml−1) was allowed to bind at 4°C for 1 h in the presence of the inhibitors. Cells were then exposed to an acidic shift (pH 5.0) at 4°C to initiate membrane insertion of TcdA1. After 40 min, 86Rb+ release was determined in aliquots of the medium by liquid scintillation counting.

Figure 4.

Effect of Rad, CsA and FK506 on the pH-dependent translocation of PTC3 across the cytoplasmic membrane.

A. Membrane translocation of PTC3 is triggered by low pH. HeLa cells were pre-incubated for 30 min at 37°C with 100 nM BafA1 to inhibit toxin uptake. Subsequently, the cells were incubated at 4°C with PTC3 (2 μg ml−1 TcdA1 plus 1 μg ml−1 TcdB2-TccC3) to enable binding to the cytoplasmic membrane. For control (con) cells were incubated without PTC3. After 20 min at 4°C the medium was adjusted to pH 4.5 and the cells were incubated for further 15 min at 37°C in the acidic medium. For control, cells were incubated for 15 min at 37°C in neutral medium. Afterwards, the acidic medium was replaced by neutral FCS-containing medium and the cells were incubated at 37°C. To monitor the toxin-induced changes in cell morphology, pictures were taken after 3.75 h and the percentage of rounded cells was determined from the pictures (scale bar, 50 μm). Values are given as mean ± SD (n = 3) and significance was tested by using the Student's t-test (***P < 0.001).

B. Rad, CsA and FK506 inhibit the pH-dependent translocation of PTC3 across the cytoplasmic membrane. HeLa cells were pre-incubated for 30 min at 37°C with 100 nM BafA1 in the presence or absence of Rad (10 μM), CsA or FK506 (20 μM) and then treated exactly as described in A. For control (con) cells were incubated without toxin. Pictures were taken after 3.5 h and percentage of rounded cells was determined as described above. Values are given as mean ± SD (n = 3) and significance was tested by using the Student's t-test (**P < 0.01). Values refer to samples treated with toxin only.

Figure 5.

Co-precipitation of His-TccC3hvr with Hsp90, FKBP51, Cyp40 and CypA from lysates of HeLa cells after incubation of intact cells with His-TccC3hvr plus PA. HeLa cells were incubated for 45 min at 37°C with PA (2 μg ml−1) plus His-TccC3hvr (1 μg ml−1) or without toxin for control (con). Subsequently, cells were lysed and the cytoplasmic fraction was obtained by centrifugation. For pull-down of His-TccC3hvr, cytoplasmic fractions were incubated with Talon™ CellThru beads over night at 4°C. The beads were washed and boiled with SDS-sample buffer. The co-precipitated proteins were separated by SDS-PAGE and blotted onto nitrocellulose. The membrane was cut and probed with specific antibodies against Hsp90, FKBP51, Cyp40 and CypA (left panel). To confirm comparable amounts of protein input for the co-precipitation samples, lysate controls probed with an antibody against actin are shown in the right panel for each corresponding co-precipitated protein.

In conclusion, the results indicate that the chaperone activity of Hsp90 as well as the PPIase activity of cyclophilins and FKBPs is crucial for cellular uptake of Photorhabdus luminescence PTC3 toxin, in particular for pH-dependent membrane translocation of the functionally active ADP-ribosyltransferase domain of TccC3, which interacts with Hsp90, FKBP51, Cyp40 and CypA in intact cells.

Discussion

While the molecular mode of action of ADP-ribosylating toxins and the resulting cellular consequences are well defined, our knowledge about the uptake of these toxins is still severely limited. This is especially true in respect to cytosolic factors which interfere with toxin uptake. Here, we investigated the participation of host cell chaperones during the uptake of the ADP-ribosylating toxins from P. luminescens, which are apparently taken up by a novel type of translocation machinery. The results indicate that the actin-ADP-ribosylating toxin PTC3 is taken up by a process, depending on the chaperone Hsp90 and PPIases of the Cyp and FKBP families and imply that CypA, Cyp40 and FKBP51 might be crucial for this process. The involvement of these factors in toxin uptake is deduced from the finding that established pharmacological inhibitors against Hsp90, cyclophilins or FKBPs protected cells from intoxication by PTC3, because they delayed the toxin-induced cell rounding in a dose-dependent manner. The inhibitors had no effect on the ADP-ribosyltransferase activity of the enzyme component TccC3, binding of PTC3 to cell membranes or pore formation by TcdA1 but prevented the pH-dependent membrane translocation of TccC3 in intact cells. Moreover, precipitation experiments suggest a direct or bridged interaction of Hps90, CypA, Cyp40 and FKBP51 with the enzyme domain of PTC3. So far, however, our data do not indicate which phase of the translocation process is facilitated by the cytosolic factors. Considering the minimal diameter of 1.6 nm of the TcdA1 pore, measured by cryo-electron microscopic studies (Gatsogiannis et al., 2013) and estimated by biophysical methods in artificial lipid membranes (Lang et al., 2013), efficient translocation of toxins through the TcdA1 pore depends on unfolding of the proteins, peptide translocation through the pore-channel and refolding of the toxin at the cytosolic side. Interaction of chaperons identified (e.g. Hps90, CypA, Cyp40 and FKBP51) might occur with partially unfolded toxin components reaching the cytosol. Such an interaction probably couples translocation with refolding, because proper folded toxin is blocked in sliding back into the pore-channel. Interestingly, same cytosolic factors were recently identified as interaction partners of the enzyme components of binary actin ADP-ribosylating toxins from clostridia (Kaiser et al., 2009; 2011; 2012) and it was demonstrated that the chaperone activity of Hsp90 as well as the PPIase activities of cyclophilins and FKBPs are required for translocation of the ADP-ribosyltransferase subunits of these toxins across endosomal vesicles into the cytosol of target cells (Haug et al., 2003; Kaiser et al., 2009; 2011; 2012). Hsp90 and CypA also interact with the ADP-ribosyltransferase domain of diphtheria toxin (Dmochewitz et al., 2011) and a role of Hsp90 in membrane translocation of a diphtheria fusion toxin was demonstrated (Ratts et al., 2003). As described for many bacterial protein toxins, the enzyme moieties of these toxins translocate across membranes of endosomal vesicles to reach the cytosol of target cells and most of the toxins (if not all) are at least partially unfolded during their translocation and, therefore, become targets of chaperones. However, a participation of Hsp90, cyclophilins and FKBPs in membrane translocation was exclusively found for ADP-ribosylating toxins but not for other bacterial toxins which deliver their enzyme moieties also across membranes of acidified endosomal vesicles such as C. difficile toxin A and B (Haug et al., 2003; Kaiser et al., 2012) or the binary anthrax toxins (Zornetta et al., 2010; Dmochewitz et al., 2011). The hypothesis that Hsp90/PPIase-dependent translocation across intracellular membranes is specific and selective for ADP-ribosylating toxins was strengthened by experiments with recombinant fusion toxins. When the protease domain of anthrax lethal factor was replaced by the ADP-ribosyltransferase domain of diphtheria toxin, the PA-dependent translocation became Hsp90/PPIase dependent (Dmochewitz et al., 2011). When the actin-ADP-ribosylating C–terminal domain of C2I, the enzyme component of the binary C2 toxin, was replaced by a non-ADP-ribosyltransferase, the C2IIa-mediated membrane translocation became independent from Hsp90 and PPIases but when it was replaced by another ADP-ribosyltransferase domain such as the Rho-modifying C3 transferase, the translocation was Hsp90/PPIase-dependent (Kaiser et al., 2012). The results obtained with PTC3 in the present study do not only demonstrate that the enzyme moiety TccC3 requires host cell chaperones/PPIases for its pH-dependent membrane translocation but confirm the current model that Hsp90, cyclophilins and FKBPs facilitate the intracellular membrane transport of ADP-ribosylating toxins, independent from their translocation machinery since the uptake of the isolated ADP-ribosyltransferase domain TccC3 via the anthrax PA system, which is largely related to the cellular uptake of binary actin ADP-ribosylating toxins, also depends on the host cell factors. This is a further strong hint that only the ADP-ribosyltransferase domain might determine the interaction with Hsp90 and the PPIases.

Although the precise mode of action of Hsp90 and the PPIases in membrane translocation of the ADP-ribosyltransferases is not known, there is evidence that Hsp90 and the PPIases prefer binding to the (partially) unfolded ADP-ribosyltransferases (Kaiser et al., 2012). Interestingly, the uptake of clostridial Rho-modifying C3 ADP-ribosyltransferases (Aktories et al., 1987; Vogelsgesang et al., 2007) into macrophages was not affected by Hsp90 inhibitors (Fahrer et al., 2010). Although C3 toxins translocate in a pH-dependent manner from acidified endosomal vesicles into the cytosol (Fahrer et al., 2010), they might not unfold during membrane translocation. It is also unclear whether the identified chaperones/PPIases facilitate membrane translocation of ADP-ribosylating toxins as Hsp90-containing multi-protein translocation complex. Such a cytosolic translocation factor complex was found for translocation of diphtheria fusion toxins in vitro (Ratts et al., 2003). However, PPIases were not identified in this complex so far. Hsp90, FKBP51/52 and Cyp40 are part of multi-chaperone complexes which have been studied in detail in the context of activation and translocation of steroid receptors in mammalian cells (Pirkl and Buchner, 2001; Richter and Buchner, 2001; Ratajczak et al., 2003; Wegele et al., 2004; Stechschulte and Sanchez, 2011; Sanchez, 2012). Several additional co-chaperones are involved in this process, whether this is also true for the uptake of ADP-ribosylating toxins remains to be clarified.

Experimental procedures

Materials

Cell culture medium (DMEM, MEM and McCoy's 5A) and fetal calf serum were purchased form Invitrogen (Karlsruhe, Germany). Cell culture materials were obtained from TPP (Trasadingen, Switzerland). Streptavidin-peroxidase and Complete® protease inhibitor were from Roche (Mannheim, Germany). Protein molecular weight marker Page Ruler prestained protein ladder was obtained from Thermo scientific (Rockford, USA). Biotinylated NAD+ was purchased from R&D Systems GmbH (Wiesbaden-Nordenstadt, Germany). Bafilomycin (Baf) A1 was from Calbiochem (Bad Soden, Germany). Radicicol (Rad), 17-dimethylaminoethylamino-17-demethoxygeldanamycin (17-DMAG) and FK506 were from Sigma (Steinheim, Germany) and CsA from Fluka (Munich, Germany). The enhanced chemiluminescence (ECL) system was from Millipore (Schwalbach, Germany). Talon™ Cell Thru Resins were purchased from Clontech (Mountain View, USA). [32P]NAD+ was from PerkinElmer (Rodgau, Germany).

Protein expression and purification

Proteins were expressed and purified as described previously (Gatsogiannis et al., 2013). Escherichia coli BL21-CodonPlus cells (Stratagene) were transformed with TcdA1, TcdB2-TccC3 or TcdB2-TccC5 and protein expression was induced by the addition of IPTG to a final concentration of 25 μM. For B. anthracis PA and TccC3hvr, E. coli BL21 (DE3) were transformed and protein expression was induced with 75 μM IPTG. TcdA1-expressing cells were grown for 24 h at 22°C, whereas all other cells were grown overnight at 28°C. For P. luminescens proteins, all cells were resuspended in lysis buffer (300 mM NaCl, 20 mM Tris-HCl, pH 8.0, 1 mM DTT, 500 μM EDTA and 10% glycerol) supplemented with DNase (5 μg ml−1), lysozyme (1 mg ml−1) and Protease Inhibitor Cocktail (Roche). After sonication, cell lysate was incubated with Ni-IDA resin (Macherey-Nagel) and loaded onto empty PD-10 columns. His6-tagged proteins were eluted with 500 mM NaCl, 20 mM Tris-HCl, pH 8.0, 0.05% Tween-20, 500 mM imidazole, and 5% glycerol. The protein-containing fractions of several columns were pooled and dialysed against 100 mM NaCl, 50 mM Tris, pH 8.0, 0.05% Tween-20, and 5% glycerol. For B. anthracis PA, same lysis and elution buffer as for the other proteins was used, but without glycerol and the pH was adjusted to 8.5. Furthermore, the dialysis of PA was performed with buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 8.5) and 1 mM EDTA. The glucosyltransferase domain of C. difficile toxin B (amino acids 1–543) was produced in the expression host Bacillus megaterium as a C-terminally His-tagged protein (Toxin B1–543-6His) and purified by nickel affinity chromatography as described previously (Genisyuerek et al., 2011).

Cell culture and cytotoxicity assays

HeLa cells were cultivated at 37°C and 5% CO2 in DMEM containing 10% heat-inactivated fetal calf serum, 1.5 g l−1 sodium bicarbonate, 1 mM sodium-pyruvate, 2 mM l-glutamine, 0.1 mM non-essential amino acids, 4 mM penicillin and streptomycin. Human colon carcinoma cells HT-29 were cultivated in McCoy's 5A medium, supplemented with 10% fetal calf serum, penicillin (4 mM) and streptomycin (4 mM). All cells were trypsinized and reseeded for at most 20 times. For cytotoxicity experiments, cells were seeded in culture dishes and incubated in serum-free medium with the respective toxin. For inhibition of the PPIase activity of Cyps or FKBPs or the activity of Hsp90, the cells were incubated for 30 min with the indicated concentrations of CsA, FK506 or Rad respectively. Subsequently, the toxin was added and cells were further incubated at 37°C with toxin plus inhibitor. After the indicated incubation periods, cells were visualized by using a Zeiss Axiovert 40CFI microscope (Oberkochen, Germany) with a Jenoptik Progress C10 CCD camera (Jena, Germany). The cytopathic effects caused by the toxins were analysed in terms of morphological changes. The percentage of round cells was determined from the pictures.

Preparation of cell extracts, SDS-PAGE and immunoblot analysis

After incubation with the toxin at 4°C to enable binding of the toxin, cells were washed twice with ice-cold PBS and lysed in a buffer containing 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 mM DTT, 5 mM MgCl2 and Complete® protease inhibitor using a syringe with an 18G needle. Subsequently, lysates were incubated with 10 μM NAD+ for 30 min at 37°C. Samples were boiled in Laemmli buffer at 95°C and subjected to SDS-PAGE. Proteins were transferred to nitrocellulose membrane (Whatman, Dassel, Germany) and the membrane was blocked for 1 h with 5% non-fat dry milk in PBS containing 0.1% Tween-20 (PBS-T) before detection of ADP-ribosylated actin by which the amount of bound toxin can be estimated. Therefore, the membrane was probed with streptavidin-peroxidase and then washed three times with PBS-T. Subsequently, the amount of ADP-ribosylated actin was visualized by using the ECL system according to the manufacturer's manual. For detection of Hsp90 and FKBP51, antibodies from Santa Cruz (Heidelberg, Germany) were used. Anti-CypA antibody was from Calbiochem (Darmstadt, Germany) and Anti-Cyp40 was purchased from Pierce (Rockford, USA). Anti-mouse and anti-rabbit secondary antibodies were from Santa Cruz (Heidelberg, Germany).

Pull-down experiments with His-TccC3hvr

HeLa cells were grown in 10 cm dishes and incubated in serum-free medium for 45 min at 37°C with PA (2 μg ml−1) plus His-TccC3hvr (1 μg ml−1) or for control without toxin. The cells were lysed with an 18G needle. The cytoplasmic fraction was obtained by centrifugation and incubated with 25 μl bed volume (1:1 in PBS) of Talon™ Cell Thru beads overnight at 4°C. The precipitated proteins were washed, heated for 5 min at 95°C in Laemmli sample buffer and subjected to SDS-PAGE. Proteins were transferred onto nitrocellulose membrane and the membrane was cut and incubated with specific antibodies against Hsp90 (mouse, 1:500 in PBS-T), FKBP51 (rabbit, 1:2500 in PBS-T), Cyp40 (rabbit, 1:5000 in PBS-T) and CypA (rabbit, 1:5000 in PBS-T). After incubation with secondary anti-mouse or anti-rabbit antibodies (1:2500 in PBS-T) the membranes were washed and co-precipitated proteins were visualized using the ECL system. For loading controls supernatant of precipitates was subjected to SDS-PAGE, blotted onto a nitrocellulose membrane and probed with the respective antibodies.

ADP-ribosylation of actin by TcdB2-TccC3 in a cell free system

HeLa lysate (40 μg of protein) was incubated for 30 min at 37°C with Rad, CsA or FK506 (20 μM). Subsequently, TcdB2-TccC3 (500 ng) and biotin-labelled NAD+ (10 μM) were added and samples were incubated for 15 min at 37°C. Samples were subjected to SDS-PAGE, blotted onto nitrocellulose membrane and ADP-ribosylated actin was detected with streptavidin-peroxidase. For post-ADP-ribosylation of inhibitor-pretreated cells, PTC3-intoxicated HeLa cells were washed with cold PBS and lysed in buffer containing 100 ml NaCl, 2 mM MgCl2, 10% glycerol, 1% Igepal, 50 mM Tris (pH 7.4) and protease inhibitors. 25 μg lysate of each sample was then subjected to an in vitro ADP-ribosylation assay for 15 min at 21°C with 1 μM radioactive [32P]NAD+ and 500 nM His-TccC3hvr (20 ng μl−1). Samples were then subjected to SDS-Page followed by autoradiography. For detecting of membrane-bound PTC3 via its enzyme activity, inhibitor-pretreated HeLa cells were cooled to 4°C and PTC3 (2 μg ml−1 TcdA1 plus 1 μg ml−1 TcdB2-TccC3) was added to enable binding to the cell surface. After 30 min at 4°C, cells were washed with cold PBS and lysed. Forty micrograms of each sample was incubated with 10 μM biotin-labelled NAD+ for 15 min at 37°C. Afterwards, all samples were subjected to SDS-PAGE, blotted onto nitrocellulose membrane and ADP-ribosylated actin was detected with streptavidin-POD and the ECL system.

Toxin-translocation assay with intact cells

The pH-dependent translocation of PTC3 across endosomal membranes was experimentally mimicked on the surface of intact cells as described for another toxin earlier (Barth et al., 2000). In short, after binding of the toxin (2 μg ml−1 TcdA1 plus 1 μg ml−1 TcdB2-TccC3) to HeLa cells at 4°C the cells were exposed to an acidic pulse (pH 4.5) and further incubated at 37°C in neutral FCS-containing medium. The morphological changes of the cells were documented by photography and the percentage of round cells was determined from the pictures. To analyse the effect of Rad, CsA and FK506 on membrane translocation, cells were pre-incubated with the inhibitors for 30 min.

In vivo pore-forming assay

86Rb+ release experiments were performed as described recently (Lang et al., 2010). Human HT-29 cells were seeded in medium containing 86Rb+ (1 μCi ml−1) at a density of ∼ 2 × 105 cells per well in 24-well cell culture plates. After 48 h, cells were incubated with 20 μM of Rad, CsA and FK506 for 1 h at 37°C. Cells were then washed twice with PBS and 0.5 μg ml−1 of TcdA1 was added in cold medium (4°C) in the presence of the inhibitors. Proteins were allowed to bind for 1 h at 4°C, followed by washing with cold medium to remove unbound protein. To initiate membrane insertion of TcdA1, cells were exposed to an acidic shift (pH 5.0) at 4°C for 40 min. Afterwards, aliquots of the medium were removed and 86Rb+ release was determined by liquid scintillation counting.

Analysis of the binding of DyLight488-labelled proteins to the cell surface by FACS

HeLa cells were detached from culture dishes with 10 mM EDTA in PBS, washed twice with PBS and kept on ice (either in PBS or PBS/BSA) prior to addition of inhibitors and/or DyLight488-labelled proteins to 1 × 105 cells in 1 ml PBS (or PBS/BSA). After the indicated time of incubation (see respective figure legends), cells were washed twice in PBS (or PBS/BSA) and subjected to FACS analysis using the BD FACSCalibur platform. Cell surface-bound fluorescence was detected with an argon ion laser (488 nm) and 530-nm-bandpass filter. Proteins were labelled with DyLight488 according to the manufacturer's recommendations (Thermo Scientific). Excess dye was removed with Micro Bio-Spin 6 columns (Bio-Rad Laboratories, Munich, Germany).

Reproducibility of the experiments and statistics

All experiments were performed independently at least two times. Results from representative experiments are shown in the figures. In each individual immunoblot panel shown in the figures, the protein bands were originally detected on the same membrane and cut out and recombined for presentation in the figures. Values (n ≥ 3) are calculated as mean ± standard deviation (SD) using the Prism4 Software (GraphPad Software).

Acknowledgements

This work was supported by the Deutsche Forschungsgemeinschaft (AK 6/22-1 to K.A. and BA 2087/2-2 to H.B.) and the Center for Biological Signaling Studies (BIOSS) in Freiburg (Germany) . K.E. is a fellow of the International Graduate School in Molecular Medicine Ulm (IGradU). We thank Dr Cordelia Schiene-Fischer and Dr Gunter Fischer (Halle, Germany) for fruitful discussions.

Ancillary