The following bacterial strains were used in this study: (1) B. thailandensis DW503, a derivative from the environmental B. thailandensis E264 (gift from Dr. Mary Burtnick, University of South Alabama) (Burtnick et al., 2001); (2) B. thailandensis CDC2721121, a clinical isolate obtained from the CDC (Glass et al., 2006); and (3) Yersinia enterocolitica WA (pYV⁺, ATCC 27729). Both Burkholderia strains were grown on LB agar plates at 37°C and stored at 4°C for up to a week or were cultured on LB broth with aeration at 37°C prior to the infection of host cells. The B. thailandensis DW503-GFP strain was generated by introduction of the BHR4-GFP plasmid (gift from Dr. M. Burtnick) into bacterial cells via electroporation and selection of clones using LB agar plates containing 50 μg/ml gentamicin. For cell infection experiments, Y. enterocolitica WA was grown at 26°C in brain heart infusion (BHI, Difco, Detroit, Mich) broth for 18 h in an orbital shaker at 180 rpm, followed by dilution of the bacterial culture to obtain 0.1 OD₆₆₀ and additional growth for 2 h at 37°C (100 rpm) to stimulate production of the T3SS.

THP-1 human monocytes (ATCC TIB-202) were maintained in RPMI-1640 GlutaMax media (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (HyClone, Logan, UT). A549 cells (ATCC CCL-185) were cultured in DMEM high glucose medium containing 10% FBS (Invitrogen) at 37°C and 5% CO₂. The RFP-LC3B stable cell line was generated by transfection of A549 cells with pmRFP-LC3 (Addgene plasmid #21075) (Kimura et al., 2007) and flow cytometry sorting using the BD FACSAria. Phorbol 12-myristate 13-acetate (PMA) and mannan from S. cerevisiae were obtained from Sigma-Aldrich (St. Louis, MO).

We validated host gene function in intracellular bacterial replication using a differential two-step labeling method to distinguish between intracellular and extracellular B. thailandensis. Bt CDC2721121 cells (10⁸) were incubated in 1 ml PBS (pH6) containing 1 μg/ml FITC (green) for 15 min at room temperature, washed four times in 1 ml PBS, and resuspended in LB. THP-1 cells were treated with siRNAs as described for the screen and infected with FITC-labeled Bt CDC2721121 at MOI 10, 72 h post-siRNA treatment. Growth of extracellular bacteria was suppressed by adding kanamycin (250 μg/ml) 2 h post-infection. After another 3 h, the infected THP-1 cells were pelleted at 800 g for 2 min and resuspended in fresh antibiotic-free RPMI/10% FBS. THP-1 cells were collected 24 h post-infection, fixed for 20 min in 1% paraformaldehyde, and incubated for 1 h in blocking buffer containing 10% sheep serum and 1% bovine serum albumin (BSA). Rabbit anti-B. thailandensis serum (gift of Dr. Paul Brett, University of South Alabama) was then added (1:500 dilution) to the blocking buffer, and cells were incubated for an additional 1 h. Following three rounds of washing in PBS, THP-1 cells were incubated for 30 min in PBS containing 1% BSA and PE (red)-conjugated anti-rabbit goat IgG (1:1,000 dilution). Thus, extracellular adherent Burkholderia were double-labeled with FITC and PE fluorescence, whereas intracellular bacteria were single-labeled with FITC. The labeled THP-1 cells were analyzed using imaging flow cytometry (Amnis ImageStream^X) with the following data collection parameters: slow flow rate, 60 × magnification, and extended depth of field. Images were collected from the darkfield, brightfield, FL1 (488ex/560em nm), and FL2 (488ex/595em nm) channels. Uninfected THP-1 cells treated with rabbit anti-B. thailandensis sera and PE-conjugated anti-rabbit IgG treatment, and Bt CDC2721121-FITC infected THP-1 cells not treated with PE-conjugated anti-rabbit IgG, served as single color controls to establish compensation that excluded channel cross-talk. The IDEAS 5.0 software was used to define the localization of internalized bacteria and to determine the percentage of single vs. double-labeled THP-1 cells.

THP-1 cells treated with 50 nM siRNA were washed twice with chilled PBS and lysed on ice for 45 min with RIPA Lysis and Extraction Buffer (Thermo Scientific, Grand Island, NY) supplemented with a cocktail of protease and phosphatase inhibitor mixture (Thermo Scientific). Lysates were centrifuged (12000 rpm, 15 min), supernatants were collected, and total protein was quantified with the BCA kit (Pierce). Equal amounts of protein (10 μg) were run on SDS–PAGE and transferred to PVDF membranes (Millipore, Billerica, MA) blocked with 5% nonfat dried milk in TBST (25 mM Tris [pH 7.6], 137 mM NaCl, and 0.2% Tween 20). Membranes were incubated with primary antibodies to PKC-α (sc-8393, H-7), PKC-η (sc-215, C-15), and actin (sc-58673, 2Q1055) (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C overnight, washed with TBST, and incubated with HRP-conjugated secondary antibody. Immunoreactive proteins were visualized with luminol reagent (Santa Cruz Biotechnology).

The CytoTox-ONE™ Assay (Promega, Madison, WI) was used to measure the release of lactate dehydrogenase (LDH) from cells with a damaged membrane. Normalized LDH activity was calculated by subtracting the fluorescent readings in the supernatants of untreated control cells from the experimental sample readings. The percentage of necrotic cells was determined as the fluorescence signal measured in the experimental sample relative to the signal obtained from a control sample treated with 0.1% Triton and processed in parallel. The results were analyzed using the Synergy2 Multi-Mode Microplate Reader (BioTec, Winooski, VT) at 560 nm excitation and 590 nm emission.

A549 cells (2 × 10⁵) were washed twice in ice-cold PBS and resuspended in 250 μl ice-cold Cell Lysis Buffer (Cell Signaling Technology, Danvers, MA) supplemented with 1 mM PMSF and 1X Halt Protease Inhibitor cocktail (ThermoScientific). Whole cell lysates were subjected to centrifugation at 12,000 g for 10 min at 4°C, and the resulting supernatants were analyzed using the PathScan Phospho-MARCKS (Ser152/156) Sandwich ELISA Kit (Cell Signaling Technology) following the manufacturer's instructions. Whole cell lysates (in triplicate) were incubated with pMARCKS antibody-coated microwells overnight at 4°C.

Overnight cultures of A549 cells were infected with B. thailandensis CDC2721121 at the indicated MOI for 2 h at 37°C. Cells were washed with PBS and resuspended in DMEM containing 250 μg/ml kanamycin to suppress the growth of extracellular bacteria. After 3 h of antibiotic exposure, cells were washed 3 × with PBS and further cultured in antibiotic-free DMEM. Prior to analysis, cells were washed with warm PBS and lysed in 0.1% Triton X-100 in PBS for 5 min. The lysis mixture was diluted, plated on LB agar plates, incubated overnight at 37°C for growth, and then quantified by colony counts. The replication index was calculated as the fold increase of intracellular bacteria between two time points.

A549 cells infected with Bt CDC2721121 in 6-well plates were washed with PBS and treated with acid ethanol [5% acetic acid (v/v), 5% dH2O and 90% ethanol (v/v)] for 30 min at room temperature. Cells were washed twice with PBS and stained with Giemsa solution (0.1% w/v) for 30 s at room temperature. After washing with dH2O, cells were air-dried and analyzed using a light microscope. At least 10 fields per view at 10 × magnification were analyzed to obtain the percentage of MNGCs, where a cell was considered a MNGC if cell fusions with 3 or more nuclei were present.

Total RNA from Burkholderia was isolated using the RNeasy Miniprep Kit (Qiagen, Germantown, MD) and subjected to the TURBO DNA-free Kit (Life Technologies, Grand Island, NY) for DNase treatment and removal. 100 ng purified RNA was used for quantitative real-time PCR with the TaqMan RNA-to-C_T 1-Step kit following the manufacturer's instructions (Life Technologies). Transcripts were normalized to 18S ribosomal RNA levels, using the VIC-MGB primer limited set, and calculated relative to the control sample using the 2T-ΔΔC method.

A549 cells expressing RFP-LC3B were grown on Corning-Costar 96-well plates with coverslip bottoms and treated with 10 μM rapamycin or infected with Bt CDC2721121 at MOI 10. Cells were fixed with 4% paraformaldehyde 5 h post-treatment, washed with PBS, and stored in ProLong Gold mount with DAPI nuclear stain (LifeTechnologies). Samples were acquired at optimized laser strength (100 mW for 488) with an area classifier (number of pixels in μm2) set at 50 using a Zeiss inverted fluorescence microscope. Images were acquired for each cell at 60 × magnification and enhanced depth of perception. ~2,000 cells were analyzed for each experimental or isotype control sample, and 500 cells were acquired for each single positive compensation control sample. The integrated software INSPIRE (version 6.0.154, Amnis) was used for data collection. Gating was determined for each experimental sample and cut-offs were defined using an untreated A549-RFP-LC3 control sample. All single and focused cells used in our analysis had zero saturated pixels. LC3 positive spots were identified by creating a spot mask for peak intensities that were greater than background. The peak to background ratio was determined for each experiment based on untreated controls. This spot mask was then converted into a spot count feature. The bacteria and LC3 spot counts were plotted against a normalized frequency of cells to generate histograms. To determine the bacterial counts in the cytoplasm that were not associated with LC3-positive spots, we used the following algorithm: Spot {Intensity (MO2, Ch02, 114-4095), Ch02 Bright, 16.25, 1}, and Not Spot {Intensity (M04,ch04, 240-4095), Ch04, Bright, 4, 1}. Analysis was performed on the compensated image files using algorithms in IDEAS (version 5.0, Amnis) image analysis software.

One-way ANOVA and Z-test were used to evaluate the significance of responses in the HTS experiments. For samples n ≥ 3, the statistical significance of the difference between experimental and control data sets was determined via Student's t-test and the normal distributions for multiple-comparisons of experimental samples were confirmed using the Tukey-Kramer test where we specified that the entire set of comparisons should have a family error rate of 0.05.

We performed a functional genomic screen using a siRNA library against 718 annotated human kinases or kinase-interacting proteins to identify host proteins required for B. thailandensis survival in human monocytic THP-1 cells (Supplementary Materials and Methods and Supplementary Table 1). We identified 35 initial hits (Table 1), in which siRNA-mediated gene silencing resulted in ≥50% reduction in host cell-associated bacteria. Genes encoding kinases that play a role in host cell survival pathways were highly represented in our screen. Screen hits also included kinases that regulate cell adhesion and actin dynamics, cell processes that are essential for phagocytosis of intracellular bacteria (Kuijl et al., 2007).

For subsequent validation studies, we used the clinical isolate B. thailandensis CDC2721121. We applied differential fluorescent labeling of extracellular and intracellular Bt CDC2721121 to assess the efficiency of pathogen internalization and survival within THP-1 cells using imaging flow cytometry (Figure 1A). We observed that siRNA-mediated silencing of six genes regulating actin dynamics (PTK7, EPHB2, and PRKCH), calcium signaling (CALM1), cell adhesion (STK35), and cell proliferation (STK38L) resulted in a significant (z>-2) reduction of B. thailandensis intracellular load in THP-1 cells, compared to cells treated with control siRNA (Figure 1B, Supplementary Table 2). The RNAi silencing of target gene activity in monocytic THP-1 cells was efficient (>75%), transcript-specific (Supplementary Figure 1), and did not affect host cell viability when compared to the control cells treated with non-targeting siRNA (Supplementary Figure 2).

We focused our study on PRKCH, which encodes for PKC-η, a novel protein kinase C isoform that requires diacylglycerol (DAG), but not Ca²⁺, for activation. Interestingly, we did not recover any classical Ca²⁺-dependent PKC isoforms (α, β, and γ), which are known to be key regulators of opsonized antigen phagocytosis (Underhill and Ozinsky, 2002). To further examine the differential functions of conventional and novel PKC isoforms in B. thailandensis CDC2721121 survival within host cells, we compared the intracellular bacterial load in THP-1 cells siRNA-depleted of PKC-η or PKC-α. Upon downregulation of PKC-η, the average number of intracellular B. thailandensis per THP-1 cell dropped by >4-fold compared to control (CTL) cells (Figure 2A). In comparison, downregulation of PKC-α resulted in decreased intracellular bacterial count per host cell by < 2-fold. By Western blot, we observed that siRNA directed toward the PRKCH transcripts was specific, causing the depletion of PKC-η, but not PKC-α protein (Figure 2B). To further test PKC isoform specificity, we examined the induction of host cell death by B. thailandensis CDC2721121 infection in lung epithelial A549 cells. This cell model has been previously used to examine Burkholderia replication in non-phagocytic cells via formation of polymorphonuclear cells (Wongprompitak et al., 2009; Kunyanee et al., 2016). We transiently expressed dominant-negative mutants of PKC-η (PKC-η-DN) and PKC-α (PKC-α-DN), and a constitutively-active PKC-η (PKC-η-CAT), which lacks the N-terminal regulatory domain, in A549 cells (Soh and Weinstein, 2003). Despite the great data variability originating from transient transfection experiments we observed that cells expressing PKC-η-DN demonstrated reduced host cell death by ~3-fold in response to Bt CDC2721121 infection, whereas expression of PKC-α-DN had no significant effect, compared to A549 cells carrying empty vector (CTL) or PKC-η-CAT (Figure 2C). These results indicate that the novel PKC-η isoform plays a more significant role in B. thailandensis intracellular survival and induction of host cell death, compared to the conventional PKC-α isoform.

The respiratory epithelium and nasal-associated lymphoid tissue serve as primary entry points for inhalational melioidosis (Owen et al., 2009). Burkholderia invades host cells via phagocytosis, escapes the membrane-bound phagosome, and disseminates in the host cytoplasm to maintain an intracellular niche (Jones et al., 1996). We investigated a potential role for PKC-η function in the activation of a downstream effector protein, MARCKS, upon Burkholderia thailandensis phagocytosis. MARCKS, an ~87 kD protein, functions as an actin cross-linking protein and a signal transducer in PKC-dependent pathways. MARCKS is activated by PKC-catalyzed phosphorylation of a conserved 25 aa basic effector domain (ED). Although MARCKS phosphorylation by the conventional PKC-α has been previously found to facilitate zymosan adherence to macrophage cells and promote phagocytosis (Allen and Aderem, 1995b), the role of novel PKC isoforms on MARCKS function in response to bacterial infection has not yet been investigated.

We detected a ~2–3-fold increase in phosphorylated MARCKS (pMARCKS) protein levels when normalized to total MARCKS protein in A549 cells infected with Bt CDC2721121 for 30, 60, or 180 min, compared to pMARCKS in uninfected cells, using an ELISA specific for MARCKS phosphorylation at Ser175/160 of the ED (Figure 3A). Interestingly, we also observed a time-dependent response to infection with a different Gram-negative pathogen, Yersinia entercolitica, in which pMARCKS levels increased from ~1.5-fold at 30 min post-infection to ~5-fold at 60 min, but decreased to basal levels at 180 min. This finding is consistent with the known virulence mechanisms of pathogenic Yersinia spp. employing a T3SS to inject Yop effector proteins into the host cells within180 min of infection to inhibit actin polymerization and the host phagocytic response (Zauberman et al., 2006). In comparison, the potent PKC stimulator PMA induced a ~15-fold increase in pMARCKS levels after 30 min and an ~8-fold increase after 60 and 180 min. We also observed that siRNA-mediated knockdown of either MARCKS or PKC-η expression in A549 cells reduced pMARCKS levels by >3-fold in response to 30 or 60 min of infection with Bt CDC2721121, compared to host cells treated with non-targeting siRNA (siR-CTL) (Figure 3B). These data indicate that PKC-η participates in MARCKS ED phosphorylation upon infection of lung epithelial cells with Gram-negative bacteria.

To further link PKC-η and MARCKS function to bacterial survival within host cells, we demonstrated that PKC-η or MARCKS depletion via siRNA resulted in a ~10-fold and ~2.5-fold reduction, respectively, in the number of viable intracellular B. thailandensis in A549 cells following infection for 24 h, compared to cells treated with non-targeting siRNA (siR-CTL) (Figure 4A, light gray bars). This decrease in intracellular Burkholderia counts correlated with increased A549 viability of PKC-η– or MARCKS-depleted cells, as measured by a ~2-fold reduction in LDH release (Figure 4A, dark gray bars). We also examined multi-nucleated giant cell (MNGC) formation as a result of Burkholderia-induced host cell fusion (Suparak et al., 2005; Galyov et al., 2010). Reduction of gene expression via MARCKS transcript-specific siRNA resulted in ~2-fold less MNGC counts as compared to A549 cells treated with non-targeting siRNA (siR-CTL) (Figure 4B). A statistically marginal difference between controls and experimental samples was observed in most assays where the MARCKS transcript levels were reduced via RNAi. There could be two logical explanations for this phenomenon, (i) native MARCKS transcript levels increase upon bacterial infection (see Figure 6B) counteracting the effect of the siRNA on MARCKS transcripts and (ii) MARCKS most probably competes with other PKC-η substrates, including MacMARCKS and proteins referred to as STICKs (substrates that interact with C-kinase) (Chapline et al., 1998). Altogether, our data suggest that MARCKS activation is dependent on PKC-η function, and both proteins are involved in Burkholderia-induced host cell pathology following infection.

Pathogen uptake via phagocytosis is initiated by a recognition event between a specific host receptor and microbial cell surface molecules, such as lipopolysaccharide or opsonins (e.g., antibodies) that coat the bacteria. Mannose receptors (MR) play a significant role in the uptake of unopsonized bacteria (Geier and Celli, 2011), but have not yet been found to play a role in microbicidal phagocytic pathways (Astarie-Dequecker et al., 1999). Interestingly, MARCKS was previously reported to function in the phagocytosis of zymosan particles that engage MRs (Carballo et al., 1999). To investigate whether B. thailandensis CDC2721121 can activate MARCKS phosphorylation via MR engagement, we quantified pMARCKS levels in A549 cells treated with purified mannan to saturate the MRs. A549 cells pre-incubated with mannan exhibited a ~1.5–2-fold decrease in pMARCKS levels following infection with Bt CDC2721121, compared to infected cells not treated with mannan (CTL), suggesting that MARCKS phosphorylation is at least partly dependent on MR engagement by the pathogen (Figure 5A, light gray). Alternatively, coating Bt CDC2721121 with an opsonin, rabbit anti-B.th serum, to block potential interactions between MRs and the pathogen, resulted in ~2-fold reduction of pMARCKS levels in A549 cells. Notably, both conditions decreased the efficiency of Bt CDC2721121 infection, reducing the replication index of intracellular Bt CDC2721121 in A549 cells by ~2-fold (Figure 5A, dark gray). These data suggest that MARCKS phosphorylation is triggered upon interaction of Bt CDC2721121-associated cell surface molecular patterns (eg. mannan polysacharides) with lung epithelial A549 cell receptors, including MR.

Consistent with these observations, both mannan pre-treatment and infection with serum-optimized Bt CDC2721121 had a positive effect on A549 host cell viability, as measured by a decrease in extracellular LDH activity (Figure 5B). In contrast, survival of THP-1 cells treated with mannan did not significantly differ from untreated cells (CTL) following infection. This difference in host cell survival between A549 (epithelial) and THP-1 (monocytic) cells upon mannan treatment is likely due to the lower expression levels of MR on monocytes prior to their differentiation into macrophages (Rouleux-Bonnin et al., 1995; Wollenberg et al., 2002). We also observed that serum-opsonized Bt CDC2721121 induced a rapid decline in THP-1 viability upon infection, suggesting activation of cell death mechanisms, possibly through Fc receptor (FcR) signaling triggered by antibody opsonins (Underhill and Ozinsky, 2002). Finally, we siRNA-depleted MRC1 (mannose receptor C type 1, CD206) transcripts from A549 cells and observed ~60% reduction in bacterial uptake after 1 h post-infection, relative to host cells treated with non-targeting siRNA (Figure 5C). However, we detected no difference in the intracellular Bt CDC2721121 counts between siR-MRC1-treated and control A549 cells after 24 h of infection. These results suggest that mannose receptors facilitate initial B. thailandensis uptake but do not significantly contribute to the activation of cellular defense mechanisms that restrict intracellular bacterial growth over time.

To determine whether the opsonization state of Burkholderia can affect stimulation of macroautophagy, we quantified expression of ATG7, a regulator of autophagosome formation (Geng and Klionsky, 2008), in response to infection with unopsonized and serum-opsonized Bt CDC2721121. We observed ~3-fold increase in ATG7 expression at 5 h post-infection of THP-1 cells with unopsonized Bt CDC2721121, but a ~70% decrease in ATG7 levels upon infection with serum-opsonized pathogen, relative to uninfected monocytes (Figure 6A, light gray), suggesting that unopsonized pathogen may preferentially activate autophagosome formation. We also observed a ~5-fold increase in MARCKS transcript levels in THP-1 cells infected with unopsonized Bt CDC2721121, whereas there was very little change in MARCKS levels in cells infected with opsonized bacteria (Figure 6A, dark gray). Based on our previous findings that MARCKS function was at least partially regulated by PKC-η (Figure 3B), we also examined a putative role for PKC-η in the regulation of MARCKS and ATG7 gene transcription. We found that both MARCKS and ATG7 transcript levels were reduced ~2- and ~5-fold, respectively, in cells depleted of PKC-η, compared to control siRNA-treated cells (CTL) upon infection with unopsonized Bt CDC2721121 (Figure 6B).

To investigate whether PKC-η may be linked to activation of host macroautophagy in response to B. thailandensis infection, we monitored autophagosome formation in A549 cells that stably expressed a RFP-tagged fusion to the key autophagy protein LC3/ATG8. We observed formation of RFP-LC3-labeled autophagosome vesicles 6 h after infection with unopsonized Bt CDC2721121, similar to treatment with rapamycin, a potent inducer of autophagy flux (Figure 6C). The average number of RFP-LC3-associated autophagosome vesicles in A549 cells increased >3-fold in infected compared to uninfected cells (Figure 6D). We observed that siRNA depletion of PKC-η did not significantly change the number of RFP-LC3-labeled vesicles in infected cells, compared to the untreated (CTL), rapamycin-treated, or siR-CTL-treated controls (Figure 6D). However, we did observe a >75% decrease in intracellular cytoplasmic pathogen that was not associated with RFP-LC3-labeled autophagosome vesicles in PKC-η-depleted A549 cells, compared to untreated (CTL) and siR-CTL treated cells (Figure 6E). Taken together, these results suggest that upon infection with unopsonized Burkholderia, PKC-η functions in phagocytic pathways that facilitate bacterial escape into the cytoplasm, which in turn activates host macroautophagy. Here, we demonstrate that inhibition of PKC-η function reduces the events of intracellular bacterial escape into the cytoplasm using the autophagosome flux activation as a reporter marker.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.