We first evaluated whether chlamydial infection can prevent Bax activation during apoptosis since activation of Bax is known to induce mitochondrial cytochrome c release (10, 11). Upon induction of apoptosis, Bax undergoes a conformational change and translocates to mitochondrial outer membranes, where it inserts itself and mediates the release of cytochrome c from the intermembrane space into the cytosol (18). Also importantly, the Bax conformational change and translocation to mitochondrial membranes can be detected with an NH₂ terminus-specific antibody in an immunofluorescence assay (5, 14), which was used in the present study. HeLa cells (American Type Culture Collection, Manassas, Va.) were infected with C. trachomatis serovar L2 at an MOI (multiplicity of infection) of 0.5 (an ~50% infection rate) for 40 h. Both control and infected cultures were treated with staurosporine at 2 μg/ml (Sigma, St. Louis, Mo.) for 5 h before being processed for triple staining (Fig. ​(Fig.1).1). The chlamydial inclusions were labeled with either a rabbit or mouse antichlamydial antibody (produced in our own laboratory) in combination with Cy5-conjugated goat anti-rabbit or -mouse immunoglobulin G (IgG; blue; Jackson ImmunoResearch Laboratories, Inc., West Grove, Pa.). The fragmented DNA was labeled with fluorescein isothiocyanate-tagged dUTP via a terminal transferase reaction with a terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling kit (green; Promega, Madison, Wis.). Active caspase 3 (top row), cytochrome c (middle row), or active Bax (bottom row) was labeled with a rabbit anti-active caspase 3 (Promega), a mouse anti-cytochrome c (Pharmingen, San Diego, Calif.), or a rabbit anti-active Bax (Santa Cruz Biotechnology, Inc., Santa Cruz, Calif.) antibody in combination with Cy3-conjugated goat anti-rabbit or -mouse IgG (red; Jackson ImmunoResearch Laboratories, Inc.). The images for the various colors were acquired individually with an Olympus confocal microscope and overlaid to make tricolor images. Although cells without chlamydial infection were induced to undergo DNA fragmentation (panels c, g, and k), caspase 3 activation (panel c), mitochondrial cytochrome c release (panel g), and Bax activation (panel k), the chlamydia-infected cells were prevented from any of these responses (panels d, h, and l). As we have previously demonstrated (8), the convincing evidence of chlamydial antiapoptotic activity comes from the cultures with both infection and apoptosis induction when the infection rate is kept at ~50%. As shown in panels d, h, and l of Fig. ​Fig.1,1, the apoptosis responses including Bax activation were detected only in uninfected cells and not in infected cells although both cell populations were maintained in the same culture. When these two cell populations from the same culture were counted under an Olympus AX-70 fluorescence microscope for Bax activation, we found that 93% of uninfected cells were induced to express active Bax while only 8% of the chlamydia-infected cells were induced to do so. Five random views with a total of ~200 cells were counted for each coverslip, and the results were consistent in three independent experiments. These observations have not only confirmed our previous observations that chlamydial infection profoundly inhibits nuclear apoptosis, caspase 3 activation, and mitochondrial cytochrome c release (6, 8) but, more importantly, allowed us to map the chlamydial antiapoptotic activity to Bax, an upstream step of mitochondrial cytochrome c release.

To further understand the mechanism of chlamydial antiapoptotic activity, we evaluated the effect of chlamydial infection on the mitochondrial localization of Bak, another proapoptotic member of the Bcl-2 family. As with Bax, upon apoptosis stimulation, Bak can undergo conformational changes by exposing its cryptic epitope at the NH₂ terminus and localize to mitochondria to induce mitochondrial cytochrome c release (1, 17). Bak activation and localization to mitochondria from cell cytosol are also traceable via immunofluorescence staining with antibodies specifically recognizing the NH₂-terminal cryptic epitope of Bak molecules (3, 12). We used experimental conditions similar to those described for Fig. ​Fig.11 to evaluate the effect of chlamydial infection on Bak activation (Fig. ​(Fig.2).2). The final cultures were stained with the Hoechst DNA dye for visualizing host cell nuclei (blue), a mouse antibody to chlamydia in combination with a Cy2-conjugated goat anti-mouse IgG for visualizing chlamydial inclusions (green), and a rabbit anti-Bak antibody (catalog no. 196150; Calbiochem, La Jolla, Calif.) in combination with Cy3-conjugated goat anti-rabbit IgG for visualizing the activated Bak molecules that are localized to mitochondria (red). The triple staining was observed under an Olympus AX-70 fluorescence microscope, and single-color images were acquired individually and then overlaid to form tricolor images. Again, the convincing evidence comes from the culture with both infection (~50% infection rate) and apoptosis induction. As shown in the final panel of Fig. ​Fig.2,2, although the uninfected HeLa cells were induced to undergo both nuclear condensation and Bak activation, the chlamydia-infected cells in the same culture were prevented from doing so. When both cell populations on the same coverslips were counted for Bak activation in three separate experiments (cell counting was carried out as described above for the Bax experiment), we found that Bak activation was induced in 93% of the uninfected cells but only 7% of the chlamydia-infected cells, demonstrating that chlamydial infection profoundly inhibited staurosporine-induced Bak activation. This observation has allowed us to map the chlamydial antiapoptotic activity to Bak, another upstream step of mitochondrial cytochrome c release.

The immunofluorescence staining with the NH₂ terminus-specific anti-Bax or -Bak antibodies described here has been extensively used by others for detecting Bax and Bak mitochondrial localization in apoptotic cells (9, 12, 14). In fact, the exact same anti-Bax antibody was previously used to detect Bax activation induced by chlamydial infection (14a). We further carried out a preabsorption experiment to verify the specificity of the anti-Bax and anti-Bak antibodies used in this study (Fig. ​(Fig.3A).3A). HeLa cells infected with chlamydia at an ~50% infection rate were induced to undergo apoptosis with staurosporine 40 h after infection as described for Fig. ​Fig.11 and ​and2.2. The cell samples were triply stained with the rabbit anti-Bax or -Bak antibodies (visualized with a goat anti-rabbit IgG Cy3 conjugate; red) plus mouse anti-cytochrome c oxidase subunit II (Molecular Probes, Eugene, Oreg.; visualized with a goat anti-mouse IgG Cy2 conjugate; green) plus Hoechst for staining of DNA (blue). To test whether the anti-Bax and anti-Bak antibodies are specific to the activated Bax and Bak molecules in apoptotic cells, we preabsorbed these antibodies with either medium or cell lysates made from either apoptotic or normal HeLa cells. As shown in Fig. ​Fig.3A,3A, the anti-cytochrome c oxidase II antibody positively stained all of the cells regardless of infection or apoptosis (second column for single-color images, fourth column for tricolor images). The anti-Bax (panels a and d) or anti-Bak (panels m and p) antibodies after preabsorption with medium alone stained the apoptotic cells that were not infected, confirming the results shown in Fig. ​Fig.11 and ​and2.2. More importantly, preabsorption with the apoptotic but not normal HeLa cell lysates completely blocked both the anti-Bax (panel i versus panel e for single-color images, panel l versus panel h for tricolor images) and anti-Bak (panel u versus panel q, panel x versus panel t) antibody staining, which demonstrated that these antibodies specifically recognized the corresponding active Bax and Bak molecules that are present only in apoptotic cells and not in normal cells. To further evaluate whether the activated Bax or Bak molecules are localized to mitochondria in apoptotic cells, we used confocal microscopy to analyze the same sets of samples (Fig. ​(Fig.3B)3B) and found that most of the anti-Bax and anti-Bak staining colocalized with the anti-cytochrome c oxidase subunit II staining. Since cytochrome c oxidase subunit II is known to localize to mitochondrial inner membrane, colocalization of activated Bax and Bak with cytochrome c oxidase subunit II demonstrated that activated Bax and Bak are indeed localized to mitochondria, which is consistent with the observations previously made by others (5, 9, 13, 14, 16, 18).

To check whether the chlamydial inhibition of Bax and Bak activation is due to the decreased Bax and Bak protein levels in the infected cells, we compared the total protein levels of caspase 3, cytochrome c, Bax, and Bak between HeLa cells with or without infection and with or without apoptosis induction on a Western blot (Fig. ​(Fig.4).4). A culture condition similar to that described for Fig. ​Fig.11 was used for the present experiment except that the infection rate was increased to an MOI of 5 to ensure a greater than 95% infection rate. The large infection dose was designed to limit apoptosis responses in the infected cultures by the uninfected cells. After apoptosis induction, the cultures were lysed with sodium dodecyl sulfate sample buffer for a Western blot assay as previously described (6). Mouse antibodies to caspase 3, cytochrome c (both from Pharmingen), and HSP70 and rabbit antibodies to Bax and Bak (all three from Santa Cruz Biotechnology, Inc.) in combination with horseradish peroxidase-conjugated goat anti-mouse or -rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.) were used for the Western blot assay. Besides obvious cleavage of caspase 3 induced by staurosporine, no significant differences were observed in the cytochrome c, Bax, Bak, and HSP70 protein levels among the various HeLa cultures. Host cell HSP70 served as a loading control since we found that neither chlamydial infection nor apoptosis significantly altered its expression. This observation suggests that chlamydial infection only blocked the mitochondrial localization of Bax and Bak without significantly affecting the total levels of these two proapoptotic Bcl-2 family member proteins.