Coxiella burnetii-Infected NK Cells Release Infectious Bacteria by Degranulation

NK cells prevent the intracellular establishment and growth of C. burnetii.

NK cells function as essential sentinels and mediators of the immune system and are a primary innate source of IFN-γ produced before the development of an adaptive response (66). Investigation of the interaction between NK cells and C. burnetii could provide new insight into how the intracellular bacteria escape early immune responses. Using KY-2 cells, a murine NK cell line with homogeneous/consistent culture properties, and the C. burnetii strain NMII RSA 439 as a suitable model system for Coxiella infection, we first aimed to characterize whether and how Coxiella invades NK cells. To this end, KY-2 cells were incubated with C. burnetii for 24 h in the presence or absence of monodansylcadaverine (MDC), which blocks clathrin-mediated endocytosis. MDC sharply reduced Coxiella uptake as measured by a previously established flow cytometry-based assay (Fig. 1a) (5, 67), indicating that clathrin-mediated endocytosis is critically involved in Coxiella engulfment. This result is similar to previous observations in epithelial cells (68). Next, we analyzed the time course of Coxiella infection in KY-2 cells. KY-2 cells were infected with Coxiella for 3 h. Afterwards, they were extensively washed to remove all extracellular and cell surface-attached bacteria and then incubated for 0 to 72 h. To differentiate between intracellular and surface-bound bacteria, the ingestion of C. burnetii was analyzed by comparing anti-Coxiella stainings of permeabilized and nonpermeabilized NK cells (Fig. 1b, left). No cell surface-bound bacteria were found at 8 h postinfection (hpi), suggesting that the bacteria were efficiently engulfed by NK cells. The further course of infection was monitored via Western blotting using a Coxiella HSP60 (coxHSP60)-specific antibody (Fig. 1b, middle and right). The immunodetection of the bacterial molecular chaperone (69) (also termed GroEL [70]), which is essentially required for in vivo folding and assembly of proteins, was used as proxy to demonstrate the intracellular presence of bacteria in NK cells as well as their cellular expulsion into the culture supernatant. Interestingly, we observed a marked increase of intracellular coxHSP60 in cell extracts at 24 hpi, suggesting that KY-2 cells were indeed infected (Fig. 1b). However, at 72 hpi, the intracellular coxHSP60 signal had faded and was barely detectable (Fig. 1b). This is unlikely to reflect the degradative elimination of the pathogen in the infected cells, because the intracellular disappearance of coxHSP60 correlated tightly with a time-dependent increase of the bacterial chaperone in the extracellular medium (Fig. 1b). Thus, the total coxHSP60 levels (intracellular plus supernatant) remained nearly constant over the time course (Fig. 1b, right). Next, we analyzed the infection of KY-2 cells using immunofluorescence (Fig. 1c). In agreement with Fig. 1b, infected KY-2 cells did not establish a typical perinuclear Coxiella-containing vacuole (CCV) (67) during infection (Fig. 1c). While multiple peripheral small bacterium-positive vacuoles with diameters of 1 to 3 μm were observed at 24 hpi, at 72 hpi, only very few or no Coxiella structures were found (Fig. 1c). The time-dependent disappearance of bacteria in infected NK cells was further confirmed by flow cytometry (Fig. 1d). In line with the Western blot data (Fig. 1b), we observed a maximum of Coxiella-positive cells at 24 hpi, followed by a continuous reduction of intracellular bacteria between 48 and 72 hpi. During the time course, the number of living KY-2 cells remained largely constant (Fig. 1e), showing that these results cannot be explained by the death of infected host cells. Expulsion of tdTomato-expressing C. burnetii from infected KY-2 cells was observed by live-cell imaging (LCI) at ≥24 hpi (Fig. 1f and Fig. 2; see also Movies S1, S2, and S3 in the supplemental material). Time-lapse image acquisition with short intervals (∼1,700 ms, resulting in about 35 frames per minute) (Fig. 2; Movie S3) revealed that the entire bacterial release process takes about 15 min and is accompanied by a visible local and convex membrane protrusion on the cell surface. Released bacteria seem to remain on the outside of the cell membrane for some time before they are set free.

In conclusion, our findings show that NK cells are able to impair the growth of internalized Coxiella, which is reminiscent of the previously described situation with Chlamydia (5). As with Chlamydia, NK cells expel Coxiella within 24 to 48 hpi by releasing them into the extracellular environment.

C. burnetii-infected NK cells display functional maturation and release infectious bacterial structures by degranulation.

Next, we examined whether the “transient” Coxiella infection, which is characterized by uptake and subsequent expulsion of the bacteria, also activates NK cells, as previously observed for Chlamydia-infected NK cells (5). To this end, KY-2 cells were infected with Coxiella for 48 h, and surface expression of CD107a was measured by flow cytometry (Fig. 3a). In parallel, we analyzed the release of granzyme B (Grzm B) and IFN-γ by enzyme-linked immunosorbent assay (ELISA) (Fig. 3b). All assays demonstrated that C. burnetii-infected KY-2 cells were activated and degranulated. Accordingly, PKCϴ underwent phosphoactivation (Fig. 3c), a step triggering downstream signaling, activation, and degranulation in NK cells (22), and the PKCϴ inhibitor sotrastaurin (71) strongly suppressed Coxiella release (Fig. 3d). To assess whether C. burnetii is able to grow at all inside NK cells when bacterial release is blocked, kinetics of Coxiella-infected cells were analyzed in the presence of sotrastaurin. In contrast to the case with Coxiella-infected L929 reporter cells, which showed a continuous growth of the bacteria over 72 h (Fig. S2), no increase in the bacterial load was observed for infected NK cells with blocked degranulation (Fig. S2). Thus, NK cells do apparently not provide the optimal vacuolar environment for the optimal intracellular growth of Coxiella.

Degranulation of NK cells is a process of regulated secretion (72) of granules with sizes between 0.2 and 1 μm (73). This allows the incorporation of Coxiella small-cell variants (SCVs) (diameter, 0.2 to 0.3 μm) and large-cell variants (LCVs) (diameter, 0.5 to 1 μm) into these vacuoles. Thus, given a nearly simultaneous release of C. burnetii (Fig. 1) and Grzm B (Fig. 3b), we asked whether Coxiella structures are localized within secretory granules of infected NK cells, as is the case with Chlamydia (5). KY-2 cells were infected with Coxiella and analyzed by electron microscopy. In noninfected KY-2 cells (control), secretory granules were detectable as type II vacuolar structures (74) containing lamellar and vesicular material (Fig. 4a). In contrast, in infected cells (24 hpi), most of the secretory granules were filled with electron-dense material (Fig. 4a and b) and large round structures (diameter, 0.2 to 0.3 μm) corresponding in size to Coxiella SCVs or LCVs. A smaller bacterial fraction was associated with outer cell surface membrane structures (24 hpi), suggesting that they were on their way out or had already been released into the environment (Fig. 4a, middle, arrowheads). At 72 hpi, only a small fraction of bacteria remained inside the cells (Fig. 4a, right). This suggests that bacterial material was in physical contact with secretory granules and that the release of Coxiella was linked to degranulation in infected KY-2 cells. Consistent with this, C. burnetii structures colocalized with perforin-positive granules, whereas in noninfected cells, perforin resided in circular structures near the nucleus (Fig. 4c). Analysis of the expulsed Coxiella by immunofluorescence, Western blotting, and electron microscopy revealed that the extracellular bacteria were associated with perforin (Fig. 5a, left, and Fig. S1) and that some bacterial structures were characterized by aberrant large electron-dense material (Fig. 5a, middle). Nevertheless, a direct comparison with C. burnetii from infectious stock cultures (Fig. 5a, right) revealed that a well-detectable amount of intact Coxiella was present in the released pathogen fraction of NK cells. Chlamydia psittaci released by NK cells was also characterized by perforin association (Fig. 5b, left, inset). However, in contrast to the situation with Coxiella, the NK cell-released Chlamydia seemed to be structurally disordered (Fig. 5b, left) and, in contrast to C. psittaci from infectious chlamydial stocks (predominantly elementary bodies [EBs]), had a low electron density and no detectable intact outer membrane structures (Fig. 5b, right).

Taken together, these findings suggest that a large proportion of structurally intact Coxiella survives the uptake into lytic granules and the subsequent export via degranulation. Accordingly, we found that NK cell-released C. burnetii, which is highly resistant to low pH (pH 4 to 5) (75) (Fig. 6a, top) and the bactericidal protease Grzm B (11) (Fig. 6b), maintained its infectivity when measured in standard reporter cells (Fig. 6c). These findings are in clear contrast to the pH and Grzm B sensitivities of Chlamydia (46–48) (Fig. 6a, bottom) and the situation with Chlamydia-infected NK cells, in which the released bacteria completely lose their infectious properties (Fig. 6c) (5).

To see whether NK cell-released C. burnetii is still able to grow and replicate in reinfected reporter cells, we analyzed the reinfection of L929 cells at different time points (24, 48, and 72 h) (Fig. 7a, top). Indeed, flow cytometry analysis to detect intracellular bacteria showed a time-dependent and continuous increase of cellular infection and bacterial load (relative mean fluorescence intensity [MFI] of bacterium-positive cells increased 5-fold between 24 and 72 hpi) (Fig. 7a, bottom). In addition, immunofluorescence studies revealed that CCV formation in reinfected reporter cells was comparable to that in cells infected with original Coxiella stocks (Fig. 7b). Consistent with this, we also observed that C. burnetii released by NK cells was able to initiate successive rounds of L929 reinfection (Fig. 7c).

Our standard reporter cell line used for Coxiella infections and reinfections was L929 (Fig. 1 and 5). To rule out that the observed loss of infectivity of NK-cell released Chlamydia was a BGM cell line-specific phenomenon, additional experiments were carried out in which the L929 cells were also infected or reinfected with C. psittaci (Fig. S3). These studies confirmed further that regardless of the reporter cell line used, C. psittaci was sensitive to low pH (Fig. S3a) and Grzm B (Fig. S3b) treatment and when released by NK cells lost its infectivity (Fig. S3c).

Next, we examined whether primary NK and KY-2 cells show the same properties when infected with C. burnetii. To this end, primary NK cells were isolated from the spleens of C57BL/6 mice (with ≥96% purity) and then infected. coxHSP60-specific immunoblotting of cell extracts and culture supernatants demonstrated efficient Coxiella release into the environment (Fig. 8a). The time-dependent expulsion of bacteria in infected primary NK cells was furthermore confirmed by flow cytometry (Fig. 8b). As for KY-2 cells, infected primary NK cells displayed low numbers of dead cells at all analyzed time points, suggesting that their time-dependent reduction was not related to cell death (Fig. 8c). Moreover, in further agreement with the results obtained with infected KY-2 cells, the release was sotrastaurin sensitive (Fig. 8d), and no perinuclear CCV appeared during the infection (Fig. 8e). Peripheral small bacterium-containing structures colocalizing with perforin were observed inside and outside infected NK cells at 12 to 24 hpi (Fig. 8e). Following our previous observations on Chlamydia (5), this suggests that immortalized and primary NK cells use the same cellular defense strategy (uptake and subsequent degranulation) during infection. However, again as with KY-2 cells, Coxiella released from lytic granules of primary NK cells retained much of its infectivity (Fig. 8f), which was still not observed for NK cell-released chlamydia (regardless of the reporter cell system used) (Fig. 8f and Fig. S4). We also observed in this study that Coxiella released from primary NK cells can grow and multiply in reinfected reporter cells (Fig. 9a), form comparable CCVs there (Fig. 9b), and remain infectious in successive rounds of reporter cell infection (Fig. 9c).

Cell culture.

The murine NK cell clone KY-2 (a kind gift from W. Yokoyama, Washington University School of Medicine) (105) was cultivated at 37°C and 7.5% CO₂ in RPMI 1640 medium supplemented with 2 mM l-glutamine, 10% fetal calf serum (FCS), β-mercaptoethanol (10 μM), and 200 U/ml of interleukin 2 (IL-2). Immortalized murine fibroblasts (L929) were used as reporter cells and were obtained from the ATCC (CCL-1). The epithelial African green monkey kidney cell line (BGM) was obtained from the National Reference Laboratory for Chlamydiosis of the Friedrich-Loeffler-Institut, Jena, Germany (CCLV-RIE 136). Cells were grown at 37°C and 7.5% CO₂ in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% FCS. Cellular uptake was analyzed by the use of a specific inhibitor (monodansylcadaverine [MDC] for inhibition of clathrin-dependent endocytosis) (Sigma-Aldrich). Sotrastaurin [3-(1H-indol-3-yl)-4-(2-(4-methylpiperazin-1-yl)quinazolin-4-yl)-1H-pyrrole-2,5-dione; Cayman Biochemical] as a pan-protein kinase C (PKC) inhibitor was used to block PKCϴ-dependent cellular degranulation.

C. burnetii NMII RSA 439.

Infection studies were performed with C. burnetii NMII RSA 439 (phase II, low virulence) and tdTomato-expressing C. burnetii NMII RSA 439 (phase II, low virulence). The tdTomato gene was amplified from pKM244mod-tdTomato (106) by standard PCR using the primers a533 (5′-GATTTAAGAAGGAGATCTGCAGATGGTGTCAAAAGGAG-3′) and a534 (5′-AAGCTTGCATGCCTCAGTCGACTTATTTATAAAGTTCATCCATGC-3′). The vector pJB-CAT-ProA-mCherry (kindly provided by Robert Heinzen and Paul Beare) was digested with PstI and SalI. Digested vector and the purified PCR product were ligated using the Gibson assembly method (107). The tdTomato-encoding plasmid pJB-CAT-ProA-TdTomatoCC was used to electroporate and transform the bacteria cultured in host-cell free medium ACCM-D (108). Transformation of C. burnetii was performed as described before (106, 109) and was approved by the government of Lower Franconia, Germany (8791.27.36.25). Selection was carried out on ACCM-D agar plates containing 3 μg/ml of chloramphenicol (CA). Screening for positive clones was performed after 10 to 14 days of growth by PCR using the primers a535 (5′-CACAGCTAACACCACGTCGTCC-3′) and a537 (5′-CTGCATCACTGGCCCATCGG-3′). Positive clones were expanded in liquid ACCM-D medium containing 3 μg/ml of CA and analyzed by fluorescence microscopy. For the preparation of bacterial stocks, murine fibroblasts (L929) were infected with C. burnetii at a multiplicity of infection (MOI) of 30 to 50, corresponding to 30 to 50 infection-forming units (IFU)/cell, and cultured at 37°C and 7.5% CO₂ for 2 weeks. Subsequently, cell culture supernatants were collected and centrifuged at 5,000 × g for 30 min at 4°C and stored on ice. The pellet was washed with H₂O and centrifuged again (5,000 × g for 30 min at 4°C). The remaining cell fraction was incubated with H₂O for 45 min for hypoosmotic cell lysis, collected, and centrifuged at 250 × g for 10 min at 4°C. The supernatant was collected and stored on ice. The corresponding pellet was mixed with H₂O, resuspended with different cannulas (G18 and G22), and centrifuged at 250 × g for 10 min at 4°C. All supernatants were collected and centrifuged at 5,000 × g for 30 min at 4°C. The resulting pellet containing purified C. burnetii was mixed with sterile phosphate-buffered saline (PBS) and stored at –80°C. The vital titer (IFU per milliliter) and MOI (IFU per cell) were determined as previously described (67).

Coxiella organisms released by infected NK cells were centrifuged to remove soluble proteins and nonassociated factors (10,000 × g for 30 min at 4°C). Coxiella-containing pellet fractions were then washed three times with large volumes of cold PBS. After resuspension in culture medium, the pellet fraction was used for GE determination and reporter cell infection. Immunoblots (Fig. S1) were used to analyze the pelleted bacteria from NK cell supernatants.

C. psittaci DC15.

The nonavian C. psittaci strain DC15 (110) was grown in BGM cells. Briefly, the cells were cultivated in antibiotic-free medium, and confluent cultures were infected with 5 × 10⁷ IFU. After 48 h of cultivation at 37°C and 7.5% CO₂, Chlamydia-containing cells were harvested, and the bacterial suspension was sonicated three times for 10 s at 100 W in an ultrasonic bath. The supernatant was carefully transferred to ultracentrifuge tubes after centrifugation (4,000 × g for 3 min at 4°C). Then, the suspension was underlaid with Visipaque (Nycomed) solutions of different concentrations (2 ml of 8% solution, 3 ml of 15% solution, and 5 ml of 30% solution) (111). Afterward, the tubes were centrifuged at 40,000 × g for 50 min and 4°C. The pellet fraction was resuspended in PBS and used for a second ultracentrifugation whereby the obtained fraction was again carefully underlaid with different Visipaque solutions (1 ml of 8%, 1 ml of 15%, 1 ml of 30%, 12 ml of 36%, 8 ml of 40%, and 5 ml of 47%). After the second ultracentrifugation (50,000 × g for 50 min at 4°C), infectious elementary bodies (EBs) were found between the 40% and 47% layers. This gradient fraction was diluted in PBS and centrifuged again (30,000 × g for 50 min at 4°C). Finally, the pellet of enriched EBs was resuspended in sucrose-phosphate-glutamic acid buffer (SPGA) and stored at –70°C.

Determination of genome equivalents (GEs).

For quantitative real-time PCR (qPCR) analysis, C. burnetii DNA was extracted from stocks, infected cells, and culture supernatants using the DNeasy blood and tissue kit (Qiagen) according to the manufacturer’s instructions for Gram-negative bacteria. The dotA PCR assay amplifies a 65-bp fragment of the C. burnetii singular chromosomal dotA gene (GenBank accession no. ABS78182.2). dotA primers (forward, dotA_for [5′-GCGCAATACGCTCAATCACA-3′; positions 1336 to 1355], and reverse, dotA_rev [5′-CCATGGCCCCAATTCTCTT-3′; positions 1382 to 1401]; synthesized by Metabion) were adapted from the literature (33, 53). As a positive control, a pcDNA3.1_dotA vector was constructed by ligating a 109-bp fragment derived from C. burnetii dotA, which contains the primer binding sites, into pcDNA3.1 (Thermo Fisher Scientific). We used 10-fold serial dilutions of the dotA vector to generate standard curves for the determination of gene copy numbers. qPCR was performed using the iScript one-step real-time PCR kit in combination with SYBR green (Bio-Rad). All PCRs were carried out in duplicate in a Bio-Rad CFX96 real-time system (Bio-Rad) (10 s at 95°C, 30 s at 50°C, 44 cycles, followed by a melting-curve analysis: 52°C to 95°C with a temperature increase of 0.5°C). For C. psittaci, we amplified a 151-bp PCR fragment of the singular chromosomal gyrA gene (GenBank accession numbers CP002806.1 and AEG87507.1) with the following gyrA primers: forward, gyrA_for (5′‐GCGAAGCATCGTAAATGCGC‐3′ [positions 181 to 200]), and reverse, gyrA_rev (5′‐AGCCAAAGTTTCCTTGACCAT‐3′ [positions 311 to 331]).

Antibodies.

To verify bacterial infection in Western blots, immunofluorescence, and flow cytometry, anti-Coxiella HSP60 (anti-coxHSP60; Enzo Life Science) and an anti-C. burnetii antiserum (67) were used. Further, antibodies against P-PKCϴ/PKCϴ, NK1.1, perforin, CD107a/LAMP1, IFN-γ, Grzm B, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were obtained from Abcam, Cell Signaling, and Millipore. Secondary antibodies were purchased from Dianova, Thermo Fisher, and BioLegend.

NK cell isolation.

Spleens were isolated from uninfected female C57BL/6 mice (8 weeks old). These were transferred to 15-ml tubes and stored on ice. To release cells, spleens were cut into sections and passed through a cell sieve (70 μm) with slight pressure. The resultant cell suspension was centrifuged (10 min at 300 × g and 4°C) and washed with sterile PBS. Afterward, NK cells were isolated by using the mouse NK cell isolation kit from Miltenyi Biotec. This isolation was followed by a negative-selection process in which all “non-NK cells” were bound by a mix of biotin-coupled antibodies and depleted with anti-biotin antibodies from the cell suspension. The separation was carried out by applying a magnetic field through a magnetic activated cell sorter (MACS) from Miltenyi Biotec. Eluted NK cells were collected in culture medium containing IL-2 (200 U/ml). To determine the purity of the isolated NK cells, the cells were stained with a phycoerythrin (PE)-coupled anti-NK1.1 antibody and measured in a MACSQuant flow cytometer (Miltenyi Biotec). The final control revealed that 96% of the purified cells were positive for NK1.1.

Western blotting.

Cells were lysed on ice in RIPA buffer (150 mM NaCl, 50 mM Tris-HCl, 1% NP-40, 0.25% sodium deoxycholate, cOmplete protease inhibitor [Roche], and 50 mM NaF) with 4 M urea. After centrifugation (14,000 rpm for 30 min at 4°C), postnuclear supernatants were analyzed in Western blots as described before (112). SDS-PAGE protein markers were obtained from Serva and Thermo Fisher Scientific. Fluorographs were scanned and quantified with GelEval 1.32 (FrogDance Software).

Flow cytometry.

For intracellular bacterial staining and titer determination, cells were collected, once washed, and fixed with 2% paraformaldehyde (PFA; 2% in PBS) for 10 min at room temperature (RT). Fixed cells were permeabilized in flow cytometric washing buffer (0.5% bovine serum albumin [BSA; Carl Roth], 0.5% saponin [Sigma-Aldrich] in PBS) at RT for 30 min. Subsequently, cells were incubated with the primary (1 h) and the respective secondary (30 min) antibodies (diluted in flow cytometric washing buffer) at RT. Staining of surface molecules occurred without PFA fixation and permeabilization. Antibody staining of cells was performed at 4°C and analyzed via a MACSQuant analyzer and MACSQuantify software (version 2.11; Miltenyi Biotec).

Immunofluorescence microscopy.

For fluorescence microscopy, cells were grown on coverslips and fixed for 20 min in 2% paraformaldehyde, blocked with 3% BSA, permeabilized with 0.1% saponin (Sigma-Aldrich), and incubated serially with the desired primary and corresponding secondary antibodies. Images were taken with an Axiovert 200M/ApoTome microscope. Colocalization was measured using AxioVision colocalization and Zen 2009 software (Zeiss).

IFN-γ and Grzm B ELISA.

To detect the amounts of IFN-γ and granzyme B (Grzm B) in the cell culture supernatant of C. burnetii-infected or uninfected NK cells, the mouse IFN-γ Platinum ELISA and the mouse Grzm B Platinum ELISA from eBioscience were used. For the analysis, 1 × 10⁵ immortalized NK cells were cultured in 6-well plates and infected with C. burnetii (MOI, 10) for 0 to 72 h. Cell culture supernatants were collected at the respective time points, centrifuged for 30 min at 14,000 rpm (4°C), and analyzed via ELISA according to the manufacturer’s instructions for culture supernatants (50 μl of undiluted supernatants as well as supernatants diluted 1:50 and 1:100 in PBS).

Transmission electron microscopy.

For transmission electron microscopy (TEM) analysis, 5 × 10⁵ NK cells were infected with C. burnetii (MOI, 10; washed 3 hpi). At different time points (0, 24, and 72 hpi), either cells or cell culture supernatants were processed further. Cells were carefully removed from the culture flasks and centrifuged at 300 × g for 5 min at 4°C. The resulting pellet was treated with fixing solution (2.5% glutaraldehyde buffered in 0.1 M sodium cacodylate [pH 7.2], 300 mosmol; Merck) and embedded in 1.8% low-melting-point agarose (Biozym). Small pieces were postfixed in 1.0% aqueous OsO₄ and stained for contrast with uranyl acetate. After stepwise dehydration in ethanol, cells were cleared in propylene oxide, embedded in Glycid Ether 100 (Serva), and polymerized at 60°C for 3 days. Ultrathin sections counterstained with uranyl acetate and lead salts were analyzed with a Tecnai-Spirit TEM (FEI). For the collection and analysis of released bacteria, cell culture supernatants were centrifuged at 5,000 × g for 60 min at 4°C. Resulting bacterial pellets were directly embedded in 1.8% low-melting-point agarose (Biozym) and afterwards treated with a respective PFA fixing solution. The remainder of the procedure was carried out as described above.

Grzm B assay.

For activation, recombinant mouse Grzm B (100 μg/ml; R&D Systems) was first incubated with 10 μg/ml of recombinant mouse cathepsin C (R&D Systems) in activation buffer (50 mM morpholineethanesulfonic acid [MES], 50 mM NaCl [pH 5.5]) at 37°C for 4 h. The activated Grzm B was then diluted to 1 μg/ml in assay buffer (50 mM Tris [pH 7.5]) containing 10⁶ IFU of Coxiella or Chlamydia. A corresponding experimental approach without Grzm B was chosen for the controls. After incubation for 4 h at 37°C, preincubated bacteria were directly used for reporter cell infection and then analyzed by flow cytometry.

Live-cell imaging.

A total of 1 × 10⁵ NK cells were seeded in μ-dishes (ibidi; 35 mm, low) and infected with C. burnetii (tdTomato; MOI, 30; washed 3 hpi) to perform live-cell imaging (LCI). During the LCI experiments, the culture medium was supplemented with 20 mM HEPES. Time-lapse image series were acquired with a Leica DMi8 microscope (objective, HC PL APO 63×/1.30 GLYC UV; camera, DFC7000GT) and a Leica DMI6000 TCS SP5 confocal laser-scanning microscope (objective, HCX PL APO lambda blue 63×/1.40 oil UV; detection via hybrid detector [HyD]; scan speed, 1,400 Hz; pinhole, 280 μm; z-step size, 2 μm) at 37°C using a heating chamber (HT200; ibidi) and a heated incubation chamber, respectively. Time-lapse imaging was started at 24 hpi and continued up to 48 hpi. The frame rates are indicated in the respective figure legends. In the case of confocal microscopy, intra- and extracellular localization of the bacteria was checked by z-stack analysis at each recorded time point. A full z-stack (8 planes; z-step size, 2 μm) was acquired every 1,690 ms, resulting in a frame rate of about 35 per minute (frame = maximum z-projection of the entire stack per time point). To reduce the mobility of the examined cells during high-speed image acquisition at the confocal laser-scanning microscope, they were overlaid at 20 hpi with 0.6% low-melting-point agarose in regular KY-2 cell culture medium, supplemented with 20 mM HEPES as indicated above. Postprocessing, analysis, and annotations were performed with Fiji (113).

Statistical analysis.

Analysis of the obtained data is shown as the means ± standard deviations (SD) of three individual experiments and was estimated using GraphPad Prism 6 (GraphPad Software). Data were analyzed by t test.

Data availability.

All data generated or analyzed during this study are included in this article.