In order to further explore the infection capabilities of the Parachlamydiaceae and S. negevensis we analyzed their ability to replicate within cell lines originating from insects. Cells lines tested in this study include the phagocytic S2 cell line derived from the fruit fly Drosophila melanogaster^[42], the Sf9 cell line, which is a clonal isolate of the IPLB-SF-21 cell line originating from the moth Spodoptera frugiperda^[43], and the Aa23T cell line derived from the mosquito Aedes albopictus^[44]. Bacteria, which were routinely cultivated within Acanthamoeba sp. UWC1, were purified from their amoebal host cells and transferred to insect cells at a defined multiplicity of infection (MOI). Addition of bacteria was followed by low speed centrifugation, which facilitates infection with chlamydiae (Fig. S1) ^[45]–^[47], and the medium was then immediately exchanged to synchronize the infection.

Immunostaining of bacteria revealed that S. negevensis was able to efficiently enter and replicate within all three tested insect cell lines (Fig. 1). Although the proportion of infected cells was variable among cell lines, in general a MOI of 40 was sufficient to cause infections in more than 50% of all cells in infected cultures. Bacterial replication was first apparent between 24 h and 48 h post infection (p.i.) (Fig. 2A and S2). At 72 h p.i. most infected cells contained already high numbers of intracellular bacteria that appeared to be localized either within a single large inclusion or numerous smaller vacuoles. Cells that harbored only small inclusions or few single bacteria were, however, also visible at late stages, suggesting that bacterial growth did not proceed equally fast in all infected cells. Cultures infected with an MOI of 20 could be maintained by regular passage for about two to three weeks, after which cell numbers stopped to increase and finally rapidly decreased due to cell lysis. At this time point, culture supernatants contained infectious particles that could establish a second round of infection after transfer to uninfected cultures (data not shown), suggesting that S. negevensis is able to successfully complete its infection cycle within these host cells.

Pa. acanthamoebae was also able to replicate within all three insect cell lines (Fig. 1), but induced severe cytotoxic effects. Bacterial growth was first evident between 24 h and 48 h p.i., when tiny inclusions appeared in a small proportion of infected cells (Fig. 2B and S2). Intracellular vacuoles then grew in size as bacterial numbers progressively increased resulting in heavily infected cells that were first detectable at 72 h p.i. Parachlamydial inclusions seemed to be more densely packed with bacteria than those formed by S. negevensis, and as the infection proceeded, host cell nuclei appeared to be displaced by the growing vacuoles, which could not be observed during infections with S. negevensis. Immunostaining of bacteria did not allow an accurate determination of the percentage of cells containing intracellular Pa. acanthamoebae, because microscopic inspection suggested that a significant proportion of added bacteria remained firmly attached to the extracellular surface of the cells and could not be removed by additional washing steps. Efficient bacterial growth resulting in visible inclusions occurred, however, only in very few cells within infected cultures. Although higher bacterial doses might increase the infection efficiency, relatively low numbers of bacteria (MOI 0.5 to 5) were preferentially used for Parachlamydiaceae due to the strong cytotoxicity observed in cultures infected with these bacteria.

Intracellular bacteria, as well as cytotoxic effects, were also observed in insect cell cultures treated with P. amoebophila (Fig. 1), yet formation of large inclusions could not be observed even at later stages of infection. This finding is consistent with the reported growth behavior of this strain that also in amoebae appears to form rather small inclusions, containing only single or very few bacterial cells, which can be found scattered throughout the host cell ^[48]. In addition, as in the case of Pa. acanthamoebae, complete removal of extracellular P. amoebophila turned out to be impossible, which further complicated monitoring of the infection progress over time.

Altogether these findings indicate that the amoeba-associated chlamydiae P. amoebophila, Pa. acanthamoebae, and S. negevensis have the capacity to infect cells derived from insects and that at least the latter two are also able to replicate within these invertebrate host cells.

Infection of insect cells with Parachlamydiaceae, Pa. acanthamoebae or P. amoebophila, was accompanied by morphological changes indicative for apoptotic cell death, including cell shrinkage, cell detachment, and formation of apoptotic bodies (Fig. 3A and S3). This resulted eventually in cell lysis, which can be explained by secondary necrosis, which usually occurs in vitro subsequent to apoptotic cell death ^[49]. Moreover, these phenotypic changes were highly similar to those observed in response to actinomycin D (ActD), a known inducer of apoptotic death in insect cells ^[50] (Fig. 3A and S3). The extent of cell death induction in cultures infected with Parachlamydiaceae was dependent on the bacterial dose, yet formation of apoptotic bodies was even visible at very low MOIs, such as MOI 0.01. The time course of cellular disintegration, however, appeared to be independent of the amount of bacteria added, and morphological changes were detectable as early as 6 to 8 h p.i. Cell death did not proceed synchronously in all cells of infected cultures; its onset seemed to be delayed in some cells and few individual cells even resisted disintegration over several days. In contrast to Parachlamydiaceae, S. negevensis did not induce appreciable extents of early host cell death at comparable low MOIs (≤5) (Fig. 3A and S3), though very high bacterial doses also resulted in morphological alterations indicative for apoptotic cell death.

The pathogenic Chlamydiaceae are able to efficiently block the apoptotic machinery in infected mammalian cells ^[17], ^[18]. Although host cell death that resembles apoptosis has occasionally been reported to occur at the end of their infection cycle, this appears to be a clearly distinct mode of cellular demise, as it proceeds in the absence of caspase activation ^[17], ^[18]. We were therefore interested in a more detailed characterization of the nature of Parachlamydiaceae-induced cell death. In particular, we aimed to clarify whether it was accompanied by typical biochemical hallmarks of apoptotic cell death, such as changes in nuclear morphology and internucleosomal DNA fragmentation, and whether it involved activation of apoptotic effector caspases.

Nuclear morphology was analyzed microscopically after staining with the DNA dye 4′,6-diamidino-2-phenylindole (DAPI). Whereas most nuclei in untreated control cultures and cultures infected with Simkania displayed a normal morphology, a marked increase in condensed and fragmented nuclei, indicative for apoptotic cell death ^[51], was visible not only in cells treated with ActD, but also in cultures infected with Parachlamydiaceae (Fig. 3A and S3). Condensed nuclear fragments were detected in both dying cells and apoptotic bodies.

We next investigated whether Parachlamydiaceae-induced cell death was also characterized by DNA fragmentation. Electrophoretic separation of DNA isolated from Sf9 cultures that were infected with Pa. acanthamoebae or P. amoebophila revealed a DNA ladder consisting of bands that were multiplies of about 180–200 bp in size (Fig. S4). This band pattern, which is typical for apoptotic internucleosomal DNA fragmentation ^[52], ^[53], could also be observed after experimental apoptosis induction by ActD. In contrast, only high molecular weight DNA was detectable in untreated control cells. For unknown reasons, apoptotic DNA ladders could neither be detected in infected S2 cells nor in S2 cultures treated with ActD, despite detectable cell death induction. We therefore additionally applied the TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labelling) assay ^[54], which allows in situ detection of DNA strand breaks in host nuclei, as alternative technique. This method not only confirmed occurrence of DNA fragmentation in infected Sf9 cultures (Fig. S5), but an increase in TUNEL-positive cells, compared to untreated cultures, was also detectable for infected S2 cells (Fig. 3B). TUNEL-staining was also observed after ActD treatment and moreover typically coincided with other features of apoptotic cell death, such as nuclear condensation and formation of apoptotic bodies that frequently also contained TUNEL-positive, condensed nuclear fragments.

Finally, we investigated whether Parachlamydiaceae-induced cell death also involved activation of apoptotic effector caspases. We used an in vitro enzyme assay based on a fluorimetric substrate containing the peptide DEVD, which is specifically cleaved by mammalian effector caspases and homologous enzymes in insects ^[10], ^[11]. As initial experiments indicated DEVD cleavage activity in response to infection with Parachlamydiaceae (data not shown), we applied this assay to monitor effector caspase activity in both cell lysates and supernatants of S2 and Sf9 cultures over three consecutive days of infection with Pa. acanthamoebae, P. amoebophila, or S. negevensis in order to obtain a more comprehensive picture of the extent and time course of cell death induction. Compared to untreated control cells, significantly elevated cleavage activity could be observed in S2 and Sf9 cells that were either incubated with ActD or infected with Parachlamydiaceae (Fig. 4 and S6) (ANOVA & Scheffé, p≤0.001). Moreover, the pan caspase inhibitor Z-VAD-FMK ^[55] was not only able to completely inhibit DEVD cleavage, but also all other signs of cell death usually observed at early stages of infections with Parachlamydiaceae. Consistent with the observation that S. negevensis does not induce appreciable host cell death at an equivalent bacterial dose, only a slight, though statistically significant (p≤0.001), increase in DEVD cleavage activity could be observed after infection with these bacteria (Fig. 4 and S6).

An increase of DEVD cleavage activity in S2 and Sf9 cells infected with Parachlamydiaceae was first visible as early as 6 h or 8 h p.i., respectively. A statistically highly significant difference (p≤0.001) compared to 0 h p.i. was first detected after 8 h in S2 cells and after 24 h in Sf9 cells. DEVD cleavage activity in cell lysates initially increased over time, but was then followed by a marked decline, which most likely reflects a transition from early apoptotic to late apoptotic/secondary necrotic cells. Consistently, decreasing cleavage activity in lysates was also accompanied by an increasing activity detectable in culture supernatants, which indicates release of active caspases from secondary necrotic cells.

In conclusion, these data demonstrate that Parachlamydiaceae-induced cell death in insect cells is indeed of apoptotic nature and depends on effector caspase activity. The combination of DAPI and TUNEL staining with immunodetection of bacteria moreover revealed that dying cells were frequently associated with bacteria, yet cell death induction was not restricted to infected cells, but also occurred in uninfected cells within infected cultures. Consistent with our morphological observations of cell death, Parachlamydiaceae-induced apoptosis in insect cells starts at a very early stage of infection and in vitro rapidly proceeds to secondary necrosis.

As a first step to decipher how Parachlamydiaceae may induce apoptosis, we aimed to analyze the specific requirements for cell death induction. In particular, we were interested in the question whether only infectious bacteria had this capacity or whether cell death might in fact be caused by cytotoxic factors or bacterial cell constituents present in chlamydial preparations. In addition, we wanted to exclude that apoptosis was triggered by cell debris originating from the amoebae from which the bacteria were purified. For this purpose S2 and Sf9 cells were subjected to diverse treatments followed by monitoring of morphological changes using phase contrast microscopy and quantitative evaluation of nuclear morphology 48 h after treatment.

Based on morphological observations, cell death induction could only be detected after addition of infectious Parachlamydiaceae (MOI 5) and to a much lower extent also after infection with ten times higher doses of S. negevensis (MOI 50) (data not shown). A statistical significant increase in altered nuclei compared to untreated control cells was, however, only induced by infectious Parachlamydiaceae (ANOVA & Scheffé, p≤0.001), but not by S. negevensis (MOI 5 or 50) (Fig. 5 and S7). Significant changes in nuclear morphology could also not be observed in response to heat- or UV- inactivated Parachlamydiaceae, sterile filtrates of suspensions of purified infectious bacteria or lysates of Acanthamoeba sp. UWC1 (p>0.05), indicating that apoptosis induction depends on viable, infectious Parachlamydiaceae. However, de novo bacterial protein synthesis was dispensable, as induction of cell death was unaffected by doxycycline, an inhibitor of bacterial translation. Doxycycline was applied at a concentration above the minimal inhibitory concentration (MIC) reported for Pa. acanthamoebae strains Bn₉ and Hall's coccus ^[34], which in our study additionally appeared to be nontoxic for insect cells, but sufficient to effectively block growth, but not entry, of Pa. acanthamoebae in S2 cells (data not shown). Moreover, supernatants collected 48 h after infection with Parachlamydiaceae from clearly apoptotic cultures did not induce significant nuclear alterations after transfer to uninfected cultures (p>0.05), suggesting the absence of cell death stimulating factors in culture supernatants at this time (Fig. 5 and S7).

Consistent with these findings, heat- and UV-inactivated Parachlamydiaceae did not trigger internucleosomal DNA fragmentation as indicated by absence of apoptotic DNA ladders in treated Sf9 cells (Fig. S4). Moreover, in response to heat-inactivated bacteria no increase in effector caspase activity was detected in S2 and Sf9 cells (Fig. 4 and S6). Together these results strongly suggest that apoptosis induction by Parachlamydiaceae requires the activity of infectious bacteria, but is independent of de novo synthesis of bacterial proteins.

The time course of apoptosis induction by Parachlamydiaceae suggested that the infection progress might be severely compromised due to an early disruption of chlamydial development. In order to test this hypothesis, we undertook a quantitative comparison of infections of S2 and Sf9 cells with Pa. acanthamoebae or S. negevensis in the absence or presence of the pan caspase inhibitor Z-VAD-FMK. At ten time points during a total period of 14 days the percentage of infected cells and the number of intracellular bacteria per cell were determined. To achieve an accurate quantification, which should be less affected by extracellularly attached bacteria compared to immunostaining, fluorescence in situ hybridization (FISH) targeting ribosomal rRNAs ^[56], ^[57] was used for the detection of bacteria. Although FISH may lead to an underestimation of the infection at early stages, its ability to predominately detect metabolically highly active bacteria, compared to inactive or dead bacteria ^[57], was considered as major advantage of this technique.

Consistent with our earlier results (Fig. 2A and S2) intracellular replication of S. negevensis in absence of Z-VAD-FMK was not detectable in S2 and Sf9 cells within the first 24 h, after which bacterial numbers in infected cells, however, increased (Fig. 6 and S8). Starting from 72 h p.i., when cells that were densely filled with bacteria first appeared, cells reflecting variable stages of infections were visible throughout the two weeks of analysis. In the presence of Z-VAD-FMK, infections with S. negevensis proceeded similarly, although the growth in S2 cells appeared to be slightly slowed down. At the end of the analysis period a significant increase in the proportion of infected S2 cells compared to 0 h p.i. (ANOVA & Scheffé, p≤0.001), probably indicating release of infectious progeny and onset of the second round of infection, was visible in the absence of Z-VAD-FMK. Interestingly, although the same trend was also detectable when the caspase inhibitor was present, the increase in the percentage of infected cells was much smaller and in fact turned out to be not statistically significant (p>0.05). In case of the Sf9 cells, we did not observe a marked increase in the proportion of infected cells during the analysis period, which may indicate inefficient bacterial spread or may be explained by the fact that bacterial growth appeared to be in general delayed in this cell line.

In the absence of Z-VAD-FMK, infections with Pa. acanthamoebae were inefficient in both cell lines, as most infected cells contained only few intracellular bacteria throughout the two weeks of analysis (Fig. 7 and S8). After 24 h increased numbers of bacteria were detectable in a small proportion of infected cells, indicating the onset of bacterial replication. Yet, heavily infected cells (containing more than 100 bacteria) were hardly observed and constituted less than 1% of the infected cells at all time points analyzed. Interestingly, a slight but continuous increase in the percentage of infected cells over time was, nevertheless, observed in both cell lines. This may, however, be explained by delayed entry of bacteria that originally remained attached after the initial inoculation or by engulfment of infected apoptotic cells by neighboring cells.

In S2 cells infections with Pa. acanthamoebae were greatly enhanced by addition of Z-VAD-FMK (Fig. 7). Even though the initial steps including attachment and entry appeared to be unaffected by the treatment, bacterial growth, which was first detectable in few infected cells at 24 h p.i., occurred very rapidly under conditions of apoptosis inhibition. After 48 h, about 45% of the infected cells contained already large inclusions, hence the infection progress was significantly improved compared to infections that occurred in the absence of the caspase inhibitor (χ² test, p≤0.01). Moreover, a strong increase in the percentage of infected cells, which was also accompanied by the onset of detectable cell lysis, was already visible at this early time point. Although infections with Pa. acanthamoebae in Sf9 cells were also significantly improved by caspase inhibition (Fig. S8), reinfection appeared to be less efficient in this cell line, which after an initial increase even resulted in a drop in the percentage of infected cells at later time points, possibly due to dilution effects introduced by passaging.

At the ultrastructural level, infection of apparently intact S2 cells with Parachlamydiaceae – including Pa. acanthamoebae, but also P. amoebophila – could be observed in the presence of Z-VAD-FMK, whereas almost all cells appeared to be in a late stage of cellular disintegration after 2 days of infection in the absence of the caspase inhibitor (Fig. S9).

In conclusion, whereas the growth of S. negevensis in insect cells was largely unaffected by the caspase inhibitor Z-VAD-FMK, experimental apoptosis inhibition strongly enhanced infections with Pa. acanthamoebae, most likely as a result of the protection of their replicative niche.

The insect cell lines S2 (Drosophila melanogaster) ^[42] and Aa23T (Aedes albopictus) ^[44] were maintained at 27°C in Schneider's insect medium (Sigma-Aldrich, Austria) supplemented with 10% heat-inactivated (56°C, 30 min) fetal bovine serum (PAA Laboratories, Austria). The insect cell line Sf9 (Spodoptera frugiperda) ^[43] was maintained at 27°C in Grace's insect medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum (PAA Laboratories) and 2 mM L-glutamine (Sigma-Aldrich). Cells were regularly passaged at confluency. Insect cell cultures were shown to be free of contamination with Mycoplasma spp. by using the Venor GeM PCR kit (BioProducts, Austria). Acanthamoeba sp. UWC1 containing the chlamydial symbionts P. amoebophila UWE25 ^[48], Pa. acanthamoebae UV7 ^[36], or S. negevensis Z ^[60], ^[118] and symbiont-free isogenic amoebae were maintained at 20°C in TSY medium (30 g/l trypticase soy broth, 10 g/l yeast extract).

Amoebae containing chlamydial symbionts were harvested at 3200× g and washed with Page's amoebic saline (PAS; 0.12 g/l NaCl, 0.004 g/l MgSO₄×7H₂O, 0.004 g/l CaCl₂×2H₂O, 0.142 g/l Na₂HPO₄, 0.136 g/l KH₂PO₄). Cells were resuspended in a small volume PAS and disrupted by three consecutive cycles consisting of freezing (dry ice/ethanol bath) and thawing (37°C). Then 0.5 volumes sterile glass beads (0.75–1.00 mm) were added, followed by vortexing for 3 min and centrifugation (620× g, 10 min). The supernatant was filtered (1.2 µm) and bacteria were collected by centrifugation (35 700×g, 45 min) in an Optima™ L-100 XP ultracentrifuge equipped with SW41 Ti rotor (Beckman Instruments, USA). Bacteria were resuspended in a small volume of sucrose-phosphate-glutamate (SPG) buffer (75 g/l sucrose, 0.52 g/l KH₂PO₄, 1.53 g/l Na₂HPO₄×2H₂O, 0.75 g/l glutamic acid) and the suspension was homogenized using needles (diameter 0.90 mm and 0.45 mm). Bacteria were collected by centrifugation (as described above), resuspended in SPG, and homogenized. Aliquots of purified bacteria were stored at −80°C. If not stated otherwise, all steps during the purification were carried out at 4°C. Number of bacterial particles were determined by counting of bacteria, which were filtered onto a 0.2 µm filter (Milipore, Austria) and stained with 4′,6-diamidino-2-phenylindole (DAPI), at an epifluorescence microscope (Axioplan 2 imaging; Zeiss, Germany). To verify the identity of the chlamydiae, DNA was extracted from purified bacteria using the DNeasy blood & tissue kit (Qiagen, Austria) and the chlamydial 16S rRNA gene was amplified using a pan-chlamydia primer set described elsewhere ^[119]. PCR products were sequenced on an ABI 3130 XL genetic analyzer (Applied Biosystems, Austria) and compared to sequence data available at NCBI.

Insect cells seeded into 24-well plates (Nunc; 5×10⁵ cells per well), 12-well plates (Nunc; 1×10⁶ cells per well), 8-well Lab Tek™ chamberslides (Nunc; 1×10⁵ cells per well), or culture flasks (Iwaki, 25 cm²; 6×10⁶ cells) were infected using a defined MOI. For microscopic analyses, infections were performed on cells seeded either in chamberslides or on glass coverslips in 24-well plates. Addition of the bacteria was followed by centrifugation (130× g, 15 min, 23°C), exchange of growth medium, and incubation at 27°C until analysis. If indicated apoptosis was induced by addition of 2 µg/ml actinomycin D (ActD; Sigma-Aldrich) to the growth medium followed by incubation for indicated periods of time.

Insect cells were fixed with 100% methanol (ice-cold, 5 min), washed once with phosphate-buffered saline (PBS), and blocked for 20 min in 2% bovine serum albumin (BSA) in PBS. Cells were then incubated for 1 h with primary antibodies diluted in blocking solution. Primary antibodies used in this study include polyclonal antibodies (Eurogentec, Belgium) raised against P. amoebophila UWE25 or Pa. acanthamoebae UV7, which were purified on gastrografin gradients ^[120], or against purified recombinant protochlamydial heat shock protein DnaK (pc1499). Antisera were pre-adsorbed with host proteins as described recently ^[120]. After three washing steps in PBS, cells were incubated for 1 h with Cy2- or Cy3-labelled secondary antibodies (Dianova, Germany) diluted in blocking solution, followed by three additional washing steps. DNA was stained for 10 min with DAPI (100 ng/ml in PBS). Cells were then washed twice with PBS and embedded in Mowiol ^[120]. Images were taken with a CCD camera (AxioCam HRc; Zeiss) connected to an epifluorescence microscope (Axioplan 2 imaging; Zeiss).

Insect cells seeded in 24-well plates were infected (as described above) and the pan-caspase inhibitor Z-VAD-FMK (10 µM; Promega, USA) was added to half of the wells, but was omitted in the others. Infections were analyzed at 0, 5, 10, 24, 48, 72, and 96 h p.i., however at 48 h p.i. the caspase inhibitor was replenished in the respective remaining wells to assure efficient caspase inhibition. In order to enable analysis of later time points (7, 10, and 14 days p.i.), cells from additional wells were passaged (1∶3) into new wells at 72 h p.i. At the time points indicated, cells were fixed by addition of one volume of a 4% formaldehyde solution to the growth medium, followed by incubation for 1 h at room temperature. Cells were then washed once with PBS and dehydrated by incubation in increasing concentrations of ethanol (50%, 80% and 96%, 3 min incubation with each). Hybridizations were performed for 3 h at 46°C at a formamide concentration of 25%, using hybridization and wash buffers described elsewhere ^[121]. The following fluorescently labelled (Cy3, Cy5, or Fluos) probes (Thermo Fisher Scientific, Germany) were applied in this study: EUK-516 ([gene: 5′- ACCAGACTTGCCCTCC -3′]) ^[122] for the detection of eukaryotic host cells, Chls0523 ([gene: 5′-CCTCCGTATTACCGCAGC-3′]) in combination with the competitor probe ([gene: 5′-CCTCCGTATTACCGCGGC-3′]) ^[56] as general probe for the detection of all chlamydiae used in this study, and UV7-763 ([gene: 5′-TGCTCCCCCTTGCTTTCG-3′]) ^[36] and Simneg183 ([gene: 5′-CAGGCTACCCCAGCTC-3′]) for the specific detection of Pa. acanthamoebae UV7 and S. negevensis Z, respectively. Cells were embedded in Mowiol and images were taken at the epifluorescence microscope and at a confocal laser scanning microscope (LSM 510 Meta; Zeiss). The percentage of infected cells and the number of intracellular bacteria within infected cells were determined. Both normal and dying cells were considered. For each time point and condition, data from three replicate infections were collected. In general, at least 500 cells were counted per sample. However, in case of infections of S2 cells with Pa. acanthamoebae UV7 in presence of Z-VAD-FMK, only a lower number of cells (at least 200 cells per well) could be analyzed for some replicates of later time points (7, 10, and 14 days p.i.) due to highly reduced cell numbers caused by cell lysis.

To test for the requirement of bacterial activity for cell death induction, insect cells seeded in 24-well plates were either treated with infectious bacteria (in the presence or absence of 10 µg/ml doxycycline) or with an equivalent amount of SPG buffer or heat-inactivated (56°C, 30 min) or UV-inactivated (Transilluminator UST-30M-8E (Biostep, Germany), 15 min) bacteria, three times the amount of a sterile-filtrate (0.2 µm) of the suspension of purified bacteria, or different amounts of an amoebal lysate (20 µl or 100 µl). Centrifugation and medium exchange were carried out as described for the standard infection procedure, yet these steps were omitted in wells where 100 µl amoebal lysate were added. In an additional experiment, the growth medium of hitherto untreated cells was replaced by sterile-filtered supernatant of an infected culture, collected 48 h p.i. For the preparation of amoebal lysates, four culture flasks (25 cm²) containing Acanthamoeba sp. UWC1 (in total approximately 2×10⁷ cells) were harvested (3200× g, 10 min). Amoebae were resuspended in 1 ml PBS and lysed by 3× freeze-and thaw (see above). The lysate was filtered (1.2 µm) and stored at −80°C. The efficiency of heat- and UV-inactivation of bacteria was tested by infection of amoebae, which revealed strongly decreased, but not completely abolished, infectivity of inactivated bacteria (data not shown). After incubation for 48 h at 27°C, treated cells were fixed with methanol and subjected to immunostaining and DAPI staining, as described above. The percentage of condensed and/or fragmented nuclei was determined for three independent experiments (of each condition mentioned above), each consisting of two replicate wells. At least 500 nuclei per replicate were counted.

Infected and control cells grown in 24-well plates or chamberslides were fixed with formaldehyde as described for FISH. Cells were then permeabilized in 0.1% Triton-X 100, 0.1% sodium citrate for 10 min and blocked with 2% BSA in PBS for 20 min. Immunostaining of bacteria occurred as described above. The TUNEL reaction was carried out according to the instructions of the in situ cell death detection kit (Roche, Austria). In case of the TUNEL positive control, cells were pre-incubated with DNase I (Sigma-Aldrich; 100 U/ml in 50 mM Tris/HCl, 1 mg/ml BSA, pH 7.5) for 10 min. For the negative control, cells were incubated in labelling solution devoid of the enzyme terminal deoxynucleotidyl transferase (TdT). Cells were finally embedded in Mowiol and analyzed at the epifluorescence microscope.

Infected and control cells grown in 12-well plates were incubated for indicated periods of time. Culture supernatants were collected and centrifuged (3800× g, 10 min, 4°C) to remove residual cells. Cell lysates were prepared by addition of 150 µl lysis buffer (50 mM HEPES (pH 7.4), 5 mM CHAPS, 5 mM DTT) to each well followed by incubation on ice for 20 min. Lysates were centrifuged (20 800× g, 10 min, 4°C) to remove insoluble cell debris. Samples were stored at −80°C until analysis. Effector caspase activity was measured using a fluorimetric caspase 3 assay kit (Sigma-Aldrich) and a Tecan Infinite M200 plate reader (Tecan, Austria). For each measurement, 5 µl lysate or 20 µl supernatant were applied. Each sample was analyzed in duplicate (technical replicates) and data were collected for at least four replicate infections (biological replicates). Values are presented as arbitrary relative fluorescence units without normalization.

Data obtained for the analysis of nuclear morphology, for monitoring of caspase activity, and for the analysis of the proportion of infected cells during the course of infection, were analyzed using one-way analysis of variance (ANOVA) to test for statistical significant (p≤0.05) differences between groups. Scheffé's post hoc test was then applied to specify groups that significantly (***, p≤0.001; **, p≤0.01; *, p≤0.05) diverged from the untreated control (nuclear morphology) or 0 h p.i. (caspase activity, percentage of infected cells). The unpaired student's t-test was additionally applied to compare the percentage of infected cells between infections that occurred in the absence or presence of Z-VAD-FMK for each time point. All three tests were carried out using the software PASW statistics version 17. In the case of the infection analysis by FISH, infected cells were additionally classified into different groups according to their number of intracellular bacteria. A χ² test was used to compare the distribution of cells among these groups for each time point between infections carried out in absence or presence of Z-VAD-FMK (***, p≤0.001; **, p≤0.01; *, p≤0.05).