Achromobacter xylosoxidans Cellular Pathology Is Correlated with Activation of a Type III Secretion System

Achromobacter xylosoxidans is increasingly recognized as a colonizer of cystic fibrosis (CF) patients, but the role that A. xylosoxidans plays in pathology remains unknown. This knowledge gap is largely due to the lack of model systems available to study the toxic potential of this bacterium. Recently, a phospholipase A₂ (PLA₂) encoded by a majority of A. xylosoxidans genomes, termed AxoU, was identified.

ABSTRACT

Achromobacter xylosoxidans is increasingly recognized as a colonizer of cystic fibrosis (CF) patients, but the role that A. xylosoxidans plays in pathology remains unknown. This knowledge gap is largely due to the lack of model systems available to study the toxic potential of this bacterium. Recently, a phospholipase A₂ (PLA₂) encoded by a majority of A. xylosoxidans genomes, termed AxoU, was identified. Here, we show that AxoU is a type III secretion system (T3SS) substrate that induces cytotoxicity to mammalian cells. A tissue culture model was developed showing that a subset of A. xylosoxidans isolates from CF patients induce cytotoxicity in macrophages, suggestive of a pathogenic or inflammatory role in the CF lung. In a toxic strain, cytotoxicity is correlated with transcriptional activation of axoU and T3SS genes, demonstrating that this model can be used as a tool to identify and track expression of virulence determinants produced by this poorly understood bacterium.

INTRODUCTION

Achromobacter xylosoxidans is a Gram-negative, rod-shaped environmental bacterium that was originally isolated from patients with otitis media (1). Since its identification, A. xylosoxidans has been documented as an opportunistic pathogen in numerous health care-associated infections (2, 3). Bacteremia is a common clinical manifestation in immunocompromised patients (4). Other reported infections include meningitis (5, 6), urinary tract infection (7), endocarditis (8), and pneumonia (9). Treatment of A. xylosoxidans infection is challenging because isolates are intrinsically resistant to numerous antibiotics, including aminoglycosides and cephalosporins (10, 11). More recently, A. xylosoxidans has been recognized as an emerging pathogen in a subset of cystic fibrosis (CF) patients (12, 13).

Patients affected by CF develop polymicrobial infections leading to life-threatening pulmonary exacerbations characterized by airway inflammation. With the improvement of diagnostic techniques, A. xylosoxidans is increasingly identified as a member of the bacterial community that colonizes the CF lung (14–18), yet the role of this species in pathology remains controversial. Epidemic strains of A. xylosoxidans have been isolated in European countries with evidence of patient to patient transmissibility (15, 19, 20). Multiple studies have shown that infection by Achromobacter spp. (21) and A. xylosoxidans (22) is correlated with a decrease in lung function. Patients chronically colonized with Achromobacter spp. are also at an increased risk for lung transplantation or death (23). In contrast, studies performed in CF centers located in North America, Belgium, and Italy found that colonization by Achromobacter spp. did not impact lung function (24–26). It is unclear whether there are circulating virulent clones or whether treatment differences influence infection outcomes.

Several potential pathogenic traits have been identified and tested in A. xylosoxidans, including biofilm formation (27–30), motility (31), and antibiotic resistance (31, 32). Achromobacter species genomes contain virulence-related genes coding for hemolysins, capsule biosynthesis proteins, and iron transport systems, though none of these products have been functionally validated (33). Multiple groups have also identified the presence of a type III secretion system (T3SS) in Achromobacter species genomes (33–35), with a gene organization most similar to that of Bordetella spp. (33). The T3SS is a virulence determinant used by a number of Gram-negative bacteria to inject effectors directly into host cells. While A. xylosoxidans possesses pathogenic phenotypes and encodes potential factors that may impact human infections, there have been few functional studies performed, due to the lack of well-characterized strains that can be genetically manipulated. Recent progress has been reported utilizing allelic replacement and transposon mutagenesis of select clinical isolates (29, 30).

We previously reported that A. xylosoxidans encodes AxoU, an ortholog of the Pseudomonas aeruginosa T3SS toxin and ubiquitin-activated phospholipase ExoU (36, 37). Recombinant AxoU is cytotoxic when transfected into epithelial cells, suggesting that it may be an effector of the A. xylosoxidans T3SS (37). Although functionally similar, ExoU and AxoU share only 22% amino acid identity and are predicted to be structurally dissimilar (37). These data suggest that AxoU may play a different role during infection compared to ExoU. The combination of encoding a potential toxin, possession of a T3SS for delivery, and the poor clinical outcomes associated with A. xylosoxidans infection at some CF centers prompted further analyses of clinical isolates; however, no in vivo or in vitro model systems demonstrating pathology had been developed.

The goal of this study was to assess the virulence of A. xylosoxidans, an emerging CF pathogen. We demonstrated that the putative A. xylosoxidans T3SS effector AxoU is functionally recognized as a T3SS substrate using P. aeruginosa as a surrogate for delivery. Next, we developed the first in vitro infection model system to test whether a pathological response could be assayed during A. xylosoxidans coculture. These studies reveal that primary macrophages and macrophage-like cell lines are susceptible to infection, while other cell types appear resistant. Importantly, this screen identified a distinct cytotoxic phenotype associated with a subset of CF isolates. Genome sequence analysis of two clinical isolates differing in their cytotoxic phenotype revealed that both isolates contain a complete T3SS locus and axoU. We demonstrate that a cytotoxic strain of A. xylosoxidans expresses axoU and select T3SS genes at the transcriptional level during infection of macrophages, while transcriptional activation of T3SS-related genes is not detectable in a noncytotoxic isolate. Finally, a screen of isolates from CF patients indicates that different A. xylosoxidans strains induce various levels of toxicity in macrophages. These results suggest that regulation of the T3SS or expression of different effectors may modulate virulence in A. xylosoxidans. Defining the virulence potential of this emerging pathogen will aid in understanding the impact of A. xylosoxidans in the polymicrobial environment of the CF lung.

RESULTS

AxoU is a T3SS substrate.

AxoU, encoded by A. xylosoxidans, was identified as an ExoU ortholog in a bioinformatic screen (36). When transfected into HeLa cells, recombinant AxoU (rAxoU) induced cellular cytotoxicity (37). Because of the cytotoxic potential of AxoU and the presence of T3SS loci in sequenced genomes, we postulated that AxoU may be injected into host cells by a T3SS. As a genetic system has not been fully developed for use in A. xylosoxidans, we used P. aeruginosa, whose T3SS secretes at least four enzymes associated with virulence (38–40), as a surrogate host to verify AxoU delivery through a competent T3SS. We amplified the axoU operon, which also contains a predicted chaperone, sacU, from A. xylosoxidans strain GN008 (Table 1) genomic DNA, and cloned the operon into pUCPpS (Table 1), a broad-host-range vector with the exoenzyme S promoter region (pS), to coordinate transcription of target genes with P. aeruginosa T3SS biosynthesis. Constructs were transformed into P. aeruginosa PA103 ΔexoUexoT::Tc (PA103 ΔUT), a T3 effectorless strain producing a functional T3SS (Table 1). PA103 ΔpcrV, a strain that retains expression of the T3SS but is unable to deliver effectors because of a defective tip complex (41), serves as a negative control. A549 lung epithelial cells were infected with strains expressing SacUAxoU, SacUAxoUS201A (catalytically inactive [37]), or ExoU (native substrate), and adenylate kinase (AK) release (42) was measured as an indicator of host cell death (Fig. 1A). At 5 h, PA103 ΔUT pUCPpSsacUaxoU induced significantly more cell death than PA103 ΔUT pUCPpSsacUaxoUS201A or PA103 ΔpcrV pUCPpSsacUaxoU (P < 0.001), suggesting that cytotoxicity is mediated by injection of a catalytically active AxoU through the P. aeruginosa-encoded T3SS.

TABLE 1.

Bacterial strains and plasmids used in this study, with relevant descriptions

Strain, genotype, or plasmid	Description	Reference or source
P. aeruginosa strains
PA103 WT	Wild type	48
PA103 ΔUT	ΔexoUexoT::Tc; produces a functional T3SS but no effectors	49
PA103 ΔpcrV	Unmarked deletion of pcrV; does not produce a functional T3SS	50
A. xylosoxidans strains
GN008	CF isolate	This study
GN050	Ear infection isolate	This study
Ax3	CF isolate	This study
Ax4	CF isolate	This study
Ax5	CF isolate	This study
Ax6	CF isolate	This study
Ax10	CF isolate	This study
Ax11	CF isolate	This study
Ax12	CF isolate	This study
Ax13	CF isolate	This study
Ax14	CF isolate	This study
Ax15	CF isolate	This study
Ax16	CF isolate	This study
Ax17	CF isolate	This study
Ax18	CF isolate	This study
Ax22	CF isolate	This study
Ax23	CF isolate	This study
Ax24	CF isolate	This study
Ax25	CF isolate	This study
Ax26	CF isolate	This study
Ax27	CF isolate	This study
Ax28	CF isolate	This study
Ax29	CF isolate	This study
Plasmids
pUCPpS	Vector plasmid with exoS promoter	51
pUCPexoU	Expresses ExoU	52
pUCPexoUS142A	Expresses noncatalytic ExoU	53
pUCPpSsacUaxoU	Expresses AxoU and chaperone, SacU	This study
pUCPpSsacUaxoUS201A	Expresses noncatalytic AxoU and chaperone	This study
pUCPpShis-sacUaxoU-HA	Expresses HA-tagged AxoU and histidine-tagged SacU	This study
pUCPpShis-sacUaxoUS201A-HA	Expresses HA-tagged noncatalytic AxoU and histidine-tagged SacU	This study

FIG 1.

AxoU is injected by the P. aeruginosa T3SS. (A) AK release assays of A549 lung epithelial cells infected with P. aeruginosa expressing ExoU, AxoU, or AxoUS201A at an MOI of 5:1. Data are from three independent experiments, each performed in triplicate. Data points are the mean, and the shaded region is ± 1 standard deviation (SD). At 5 h, PA103 ΔUT expressing WT AxoU is significantly more toxic than PA103 ΔUT expressing AxoUS201A or PA103 ΔpcrV expressing WT AxoU (P < 0.001; one-way ANOVA with Tukey’s post hoc test). (B) Representative Western blot of A549 cells infected with P. aeruginosa expressing ExoUS142A (lanes 1 and 5), AxoUS201A (lanes 3 and 7), or AxoUS201A-HA (lanes 2, 4, 6, and 8) at 6 h. (C) Representative immunofluorescence images of an infection using strains depicted in panel B at 6 h. Phalloidin, actin; DAPI, nucleus; toxin, ExoU or HA. Scale = 50 μm.

To biochemically track AxoU delivery, variants in which AxoU was fused to a C-terminal hemagglutinin (HA) tag were constructed. Biological analyses were then performed to ensure that the tagged variant was functionally similar to the native protein (see Fig. S1A and B in the supplemental material). A549 cells were infected with PA103 ΔUT expressing ExoUS142A, SacUAxoUS201A-HA, or SacUAxoUS201A and PA103 ΔpcrV expressing SacUAxoUS201A-HA. Catalytically inactive molecules were used to maintain host cell integrity for downstream analysis. Extracellular bacterial cells (prokaryotic fraction) and the remaining eukaryotic monolayer (eukaryotic fraction) were collected and subjected to SDS-PAGE and Western blot analysis (Fig. 1B). As a control, ExoUS142A was used as a native substrate of the P. aeruginosa T3SS. Using a monoclonal antibody, ExoUS142A was detected in both the eukaryotic and prokaryotic cell fractions (Fig. 1B, lanes 1 and 5, respectively). In the eukaryotic fraction, ExoUS142A appeared diubiquitylated (lane 1), which has been observed previously (43). PA103 ΔpcrV contains a wild-type (WT) copy of exoU in the chromosome, which cannot be delivered due to the defective tip complex but serves as a control for T3SS expression (top panel, lane 8). Similar to ExoUS142A, AxoUS201A-HA was detected in the eukaryotic and prokaryotic fractions (lanes 2 and 6), while the untagged derivative was not (lanes 3 and 7). Further, AxoUS201A-HA was not detected in the eukaryotic fraction when expressed by PA103 ΔpcrV, even though it was present in prokaryotic fractions (lanes 4 and 8). Immunofluorescence microscopy was used to visualize AxoUS201A-HA intracellular delivery (Fig. 1C), and the fluorescence signal was quantified (Fig. S2). Collectively, these data indicate that expression of AxoU is toxic to eukaryotic cells and requires a functional T3SS for intracellular delivery.

A. xylosoxidans specifically targets phagocytic cells.

While our data suggest that A. xylosoxidans encodes a putative toxin that can be delivered by the T3SS of P. aeruginosa, directly testing the toxic potential of A. xylosoxidans requires development of an infection model. As epidemiological evidence regarding human infection remains controversial, developing an animal infection model seemed premature. Instead, two A. xylosoxidans clinical isolates predicted to have unique lifestyles were used to infect a variety of different cell types; GN008 is an isolate from a CF patient (representing a potentially chronic lifestyle), while GN050 is an isolate from a patient with an ear infection (potentially representing an acute lifestyle). To initiate our infection studies, a screen for cellular cytotoxicity was quantified by AK release assays in multiple cell types (Fig. 2). These preliminary data indicate that GN050 induced the greatest cytotoxicity in primary macrophages and macrophage-like cell lines (J774a.1) and suggested that selective toxicity may correlate with the ability to internalize the bacterium. Conversely, the CF isolate GN008 appears to possess relatively poor cytotoxic potential despite harboring an intact copy of axoU.

FIG 2.

Cell screen for A. xylosoxidans-mediated cytotoxicity. Multiple eukaryotic cell types were screened for cytotoxicity after coincubation with GN008 or GN050 at an MOI of 20:1 for 8 h. At 8 h, supernatants from cell infections were harvested and assayed for AK release. For each cell type, the percentage of cell lysis was calculated as a fraction of maximum cell lysis by detergent treatment. Each cell type was screened for cytotoxicity in triplicate on a single day. The bars represent the mean, and the error bars are ± 1 SD.

To better understand the kinetics of GN050 cytotoxicity in J774a.1 macrophages, live cell imaging was performed (Movies S1 and S2). Cell death was examined in real time by monitoring propidium iodide (PI) incorporation into damaged cells (Fig. 3A) and quantifying PI fluorescence intensity as a function of time (Fig. 3B). GN050 induced significantly more cytotoxicity than GN008 or uninfected cells at 50 min (P < 0.005). Approximately 50% of the cytotoxic activity occurred within the first 2 h of infection.

FIG 3.

Characterization of the kinetics of A. xylosoxidans cytotoxicity in macrophages. (A) Bright-field and fluorescent images of PI incorporation into the nuclei of compromised J774a.1 cells coincubated with GN008 or GN050 at an MOI of 5:1 after a light centrifugation. Scale = 50 μm. (B) Quantification of PI fluorescence over time in arbitrary units (AU). The line is the mean of three independent experiments, each quantifying three fields of view, and the shaded area is ± 1 SD. GN050 is significantly more cytotoxic than GN008 or uninfected groups at 50 min (P < 0.005). (C and D) J774a.1 cells treated with either live or heat-inactivated (HI; 56°C, 30 min) GN008 or GN050 at an MOI of up to 50:1 (C) or medium conditioned (CM) by GN008 or GN050 previously (D) for 8 h. Supernatants were assayed for AK release. For each, data bars are the mean of three independent experiments performed in triplicate, and the error bars are ± 1 SD. All analyses were done with one-way ANOVA followed by Tukey’s post hoc test.

It was previously demonstrated that A. xylosoxidans cell wall components induce proinflammatory cytokine production in eukaryotic cells (44). To test the hypothesis that a nonspecific bacterial cell wall component or a general secreted factor was responsible for the observed cytotoxicity, J774a.1 cells were incubated with heat-inactivated (HI) GN008 or GN050 (Fig. 3C) or filtered conditioned medium (CM) from GN008 or GN050 cultures (Fig. 3D). Neither HI- nor CM-treated cells induced significant cell death, suggesting that only live A. xylosoxidans induces cytotoxicity. Overall, phagocytic cells appear to be susceptible to infection by certain strains of live A. xylosoxidans. The two isolates tested do not appear to synthesize an extracellular toxin or cell wall component that contributes to the cytotoxic phenotype observed in phagocytic cells.

Internalization of A. xylosoxidans is required for cellular cytotoxicity in J774a.1 macrophages.

As primary macrophages and macrophage-like cell lines were permissive for GN050-mediated toxicity (Fig. 2), we hypothesized that phagocytosis of A. xylosoxidans by host cells may be a prerequisite for this phenotype. To test this possibility, J774a.1 cells were incubated with cytochalasin D to block actin polymerization (45) and phagocytosis (Fig. 4A). At 3 μg/ml of cytochalasin D, there was no significant difference in cytotoxicity between J774a.1 cells infected with GN050 and the uninfected group (P > 0.05). We conclude that internalization is required for cytotoxicity.

FIG 4.

Internalization is required for A. xylosoxidans-mediated cytotoxicity in macrophages. (A) Cytochalasin D titration into J774a.1 cells infected with GN008 or GN050 at an MOI of 5:1. At 8 h, supernatants were processed for AK release. At 3 μg/ml cytochalasin D, there is no significant difference between GN050 and the uninfected group (P > 0.05, one-way ANOVA and Tukey’s post hoc test). (B) Quantification of bacterial cells phagocytosed by J774a.1 macrophages. Cells were infected with GN008 or GN050 at an MOI of 5:1 in the absence or presence of cytochalasin D. There are significantly more GN050 cells than GN008 cells at the time of quantification (P < 0.0001, independent t test). For each, the data are the mean of three independent experiments performed in triplicate, and the error bars are ± 1 SD.

If internalization is required to induce cytotoxicity, the lack of cytotoxicity of GN008 may be due to inefficient host cell uptake. To test this hypothesis, we used polymyxin B protection assays (46) to quantify phagocytosed bacteria (Fig. 4B), as both of the clinical isolates were resistant to gentamicin. Significantly fewer GN008 than GN050 bacteria were observed in cells after incubation and polymyxin B treatment (P < 0.0001). J774a.1 cells pretreated with 2 μg/ml of cytochalasin D were used as a negative control for bacterial internalization. The results from our internalization assays suggest that GN008 was not phagocytosed as efficiently as GN050 or that GN008 was nonviable after phagocytosis. Collectively, these data indicate that phagocytosis of GN050 is required to initiate cell cytotoxicity and that GN008, a nonisogenic CF clinical isolate, clearly displays an alternative phenotype. Currently, it is unclear whether GN008 possesses a defect in uptake or intracellular survival relative to GN050. Alternatively, differences in gene composition or gene regulation may explain these phenotypes.

Analysis and expression of the A. xylosoxidans T3SS.

To determine whether the differences between GN008 and GN050 cytotoxicity are related to their T3SS gene content, both genomes were sequenced, annotated (Fig. S3A), and queried against all A. xylosoxidans genomes available through the NCBI database. Out of 65 genomes analyzed, 72% possessed a complete T3SS locus, and 98% of T3SS-positive strains possessed axoU. Across all strains, AxoU shared a 95% pairwise identity (Fig. S4). These data indicate that possession of an intact T3SS and possession of axoU are positively correlated in A. xylosoxidans.

We next wanted to establish if A. xylosoxidans expresses axoU or T3SS genes during infection with macrophages using quantitative reverse transcription-PCR (RT-qPCR) of total RNA utilizing gene-specific primers. Transcription of axoU (toxin), axlF (needle), axlA1 (V-tip), and axlG (V-tip chaperone) was measured 0 to 3 h postinfection with J774a.1 macrophages (Fig. 5). There was a significant increase in axoU and T3SS gene transcripts by GN050 at 30 min postinfection, followed by a decrease in mRNA levels. Supplementing cytochalasin D into the infection medium abrogated the increase in expression, suggesting that bacterial internalization is required for GN050 T3SS transcriptional activation. Conversely, GN008 did not display an increase in axoU or T3SS gene expression. We hypothesize that this result may be due to the lack of internalized bacteria. These data suggest that GN050 briefly increases expression of axoU and T3SS genes, which is dependent on bacterial internalization.

FIG 5.

Transcription of the A. xylosoxidans T3SS during infection of J774a.1 macrophages. (A) GN050 or (B) GN008 coincubated with J774a.1 macrophages at an MOI of 5:1 from 0 to 3 h with DMSO or cytochalasin D supplemented into the medium. Data points represent transcript levels from infections performed on separate days, while the line represents the mean, and the shaded area is ± 1 SD. Each inset P value represents the comparison of DMSO- to cytochalasin D-treated groups at 0.5 h determined by one-way ANOVA followed by Tukey’s post hoc test. Bacterial RNA levels were normalized to the gyrB housekeeping gene and to the amount of transcript present at t = 0 h (dotted line) and analyzed using the ΔΔC_T method.

A subset of A. xylosoxidans CF isolates are cytotoxic toward mouse J774a.1 and human THP-1 macrophages.

To determine if the nontoxic phenotype observed in GN008 coculture with mouse J774a.1 macrophages was a common trait associated with isolation from individuals with CF, an additional 29 A. xylosoxidans isolates were obtained from CF patients (Medical College of Wisconsin [MCW], Department of Pathology). Strains with a mucoid phenotype (n = 6) or strains that grew poorly (n = 2) were excluded from further characterization. The remaining 21 isolates (Table 1) were tested for the ability to induce cytotoxicity in mouse J774a.1 macrophages. Differentiated human THP-1 macrophages were also queried to determine whether differences between mammalian hosts impacted the infection model. Cytotoxicity was assayed by quantifying PI fluorescence at 6 h (Fig. 6A). A spectrum of toxicity was observed that was isolate-dependent and host cell-dependent. In order to group these patterns of cytotoxicity in an unbiased manner, we used k-means clustering to generate two-dimensional nonmetric multidimensional scaling (NMDS) plots (Fig. 6B). Based on the silhouette scores (Fig. S5), the optimal number of clusters was 2 (score of 0.76) with the possibility of 3 (score of 0.654) to 4 (score of 0.644) groups, suggesting that subclusters likely exist. We chose to represent the data as 3 clusters (Fig. 6B and C). Cluster 1 represents the majority of isolates (blue). They were nontoxic or slightly toxic to both human and mouse macrophage cell lines. Isolates that were toxic to both cell lines, similar to GN050, formed cluster 2 (red). Finally, isolates that were more toxic to one macrophage cell line than another formed cluster 3 (green). Generally, THP-1 macrophages appear to be more sensitive to A. xylosoxidans infection than J774a.1 macrophages.

FIG 6.

Cytotoxicity of A. xylosoxidans CF isolates in macrophages. (A) Mouse J774a.1 and human THP-1 macrophages were challenged with A. xylosoxidans isolates at an MOI of 10:1 for 6 h and then assayed for PI staining relative to GN050. Data bars are the mean, and the error bars are ± 1 SD from four separate fields of view. (B) Two-dimensional NMDS plots showing the grouping of A. xylosoxidans isolates by k-means clustering. Grouping was based on the average cytotoxicity for each isolate in J774a.1 and THP-1 macrophages. (C) Corresponding silhouette plot for 3 clusters. C1, cluster 1; C2, cluster 2; C3, cluster 3.

DISCUSSION

Possession of genes encoding a T3SS was recognized as an early indicator of the toxic potential of A. xylosoxidans (33–35), an emerging opportunist associated with infections in CF patients. The lack of infection models to measure a pathological response, however, has restricted the direct testing of pathogenesis of this bacterium. In this study, we developed the first in vitro cell infection model for studying A. xylosoxidans pathogenesis. Incubation of macrophages with a subset of A. xylosoxidans clinical isolates induces a cytotoxic phenotype. Pretreatment of macrophages with cytochalasin D inhibits this affect (Fig. 4A), suggesting that internalization is required for A. xylosoxidans to induce cell death. Furthermore, transcription of T3SS genes in A. xylosoxidans, including the predicted cytotoxin axoU, was briefly induced early during infection of macrophages (Fig. 5A), a phenotype that is dependent on bacterial internalization. Genes that were induced are postulated to encode proteins important for building the T3SS machinery, mediating pore formation in host membranes, and a potential T3SS toxin, AxoU (Fig. S3A and B).

While our data indicate that there is a brief transcriptional pulse of T3SS genes and axoU postinternalization, the functional role of AxoU is not clear. Previous studies have shown that AxoU is toxic to HeLa cells following transfection and is enzymatically active in bacteria when coexpressed with ubiquitin (37). As a genetic system has not been reliably developed for A. xylosoxidans, we could not introduce tagged constructs to monitor AxoU secretion directly. To determine whether AxoU is a T3SS substrate, however, we engineered an effectorless strain of P. aeruginosa to express catalytic and noncatalytic forms of AxoU along with its chaperone, SacU. When coincubated with A549 epithelial cells, P. aeruginosa strains expressing SacU/AxoU, but not the noncatalytic AxoUS201A variant, were toxic (Fig. 1A). Biochemical fractionation and immunofluorescence analyses indicate that a tagged, noncatalytic derivative of AxoU is expressed and detected in A549 lung epithelial cells (Fig. 1B and C). A control strain that is unable to deliver T3SS substrates (PA103 ΔpcrV) was unable to inject AxoU. Based on these findings combined with our cytochalasin D experiments, and the requirement of ubiquitin for activation of its phospholipase activity, we conclude that AxoU likely accesses the intracellular environment through the A. xylosoxidans T3SS. Questions still remain regarding the mechanism of recognition of SacU/AxoU by the P. aeruginosa T3SS and whether AxoU is the only toxin involved in macrophage cytotoxicity. The brief transcriptional pulse at an early point during infection supports the notion that AxoU may be involved in phagosome escape. Alternatively, AxoU may be an especially potent enzyme that only needs to be produced in minimal amounts to induce cell death. The development of a cell infection model now opens the door to mapping virulence pathways and testing their roles in macrophage toxicity.

We analyzed two nonisogenic clinical isolates, only one of which was cytotoxic to macrophages. The nontoxic isolate, GN008, is from a CF patient and may represent a type of strain that chronically colonizes the CF lung and has reduced toxic potential, a trait common among bacterial CF isolates (47). The toxic potential of GN008 appears to be unrelated to the possession of axoU. In fact, the open reading frames for axoU are 99.5% identical between GN008 and GN050. Moreover, axoU cloned from GN008 was toxic when expressed and delivered by the P. aeruginosa T3SS. GN008 did not induce transcription of the T3SS upon coculture with macrophages, but it also was not internalized in J774a.1 cells compared to GN050. Bacterial internalization as a virulence signal could explain why GN008 does not induce the expression of its T3SS.

To determine the significance of A. xylosoxidans toxicity to infections in adult CF patients, we tested nonmucoid strains of A. xylosoxidans from different CF patients in coculture with mouse J774a.1 and human THP-1 macrophages (Fig. 6A). Grouping the average cytotoxicity profiles by k-means clustering revealed 3 groups (Fig. 6B and C). Isolates belonging to cluster 1 were slightly more cytotoxic to human THP-1 macrophages than J774a.1 macrophages. Isolates belonging to cluster 2 were cytotoxic to both macrophage cell types at levels comparable to GN050, a cytotoxic strain. Variation in toxicity between the isolates may be related to differences in bacterial gene composition or regulation or the ability to bind to macrophages, be internalized, or induce the T3SS. Unexpectedly, a third cluster emerged in which isolates were more cytotoxic in one cell line than another. Ax3, Ax12, and Ax24 were more cytotoxic to THP-1 cells than to J774a.1 cells, and Ax25 was more cytotoxic to J774a.1 cells than to THP-1 cells. We speculate that this discrepancy could be a result of differences in host cell receptor composition, bacterial trafficking, or survival following uptake. Importantly, the differences between the two cell types will help identify host cell factors or events leading to a more cytotoxic phenotype.

These data suggest that not all CF strains have lost their toxic potential, and cytotoxicity of phagocytic cells may contribute to inflammation observed in patients colonized by A. xylosoxidans (12, 27). To directly test this hypothesis, it will be critical to develop genetic systems that permit the construction of isogenic strains that vary in specific loci. Alternatively, identifying the genes that may be expressed during macrophage infections may lead to new therapeutic targets. Importantly, these studies combined with information on patient health status relative to when the isolate was obtained, the presence or absence of other microbes, and the genomic composition of each isolate will be helpful in understanding whether A. xylosoxidans contributes to the inflammatory environment of the CF lung.

A. xylosoxidans is an emerging pathogen in CF patients and in patients who are immunocompromised. We developed the first reproducible tissue culture models that can be used to identify A. xylosoxidans virulence factors, screen for cytotoxic clinical isolates, or generate biomarkers for virulent phenotypes. This represents an important first step in determining the toxic potential of A. xylosoxidans strains isolated from various infection sites. Future studies will be focused on determining the role of the A. xylosoxidans T3SS and AxoU during infection of CF patients.

MATERIALS AND METHODS

Bacterial culture.

A. xylosoxidans GN008, GN050, and other clinical isolates were obtained from MCW clinical laboratories and cultured on Luria-Bertani (LB) agar at 37°C for 1 day and then at room temperature for at least 1 day. P. aeruginosa strains transformed with pUCP plasmids containing exoU, exoUS142A, sacUaxoU, sacUaxoUS201A, and sacUaxoUS201A-HA were cultured on Vogel-Bonner medium (VBM) supplemented with 400 μg/ml carbenicillin for 1 day at 37°C and then at room temperature for 1 day.

Tissue culture.

A549 lung epithelial cells (ATCC) were cultured in Ham’s F12-K (Kaighn’s) medium supplemented with 10% heat-inactivated, filter-sterilized newborn calf serum (Thermo Scientific). J774a.1 macrophage cells (ATCC) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Thermo Scientific) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Atlanta Biologicals). Cultures were maintained at 37°C in humidified air containing 5% CO₂. Additionally, HeLa (ATCC) MRC-5 cells (provided by Scott Terhune at Medical College of Wisconsin) were cultured in DMEM supplemented with 10% FBS. THP-1 monocytes (ATCC) and B6 primary macrophages (provided by Daisy Sahoo at Medical College of Wisconsin) were cultured in RPMI 1640 medium (Thermo Scientific) supplemented with 10% FBS.

A. xylosoxidans cell infections.

Unless stated otherwise, J774a.1 cells were seeded at 7 × 10⁴ to 8 × 10⁴ cells per cm² in 6-well or 24-well tissue culture plates the day before infection. THP-1 human monocytes were differentiated into macrophages by supplementing phorbol 12-myristate-13-acetate (PMA; Sigma) at 100 ng/ml in RPMI medium supplemented with 10% FBS for at least 48 h. THP-1 macrophages were seeded at 1.1 × 10⁵ cells per cm². At 24 h prior to infection, the medium was replaced with PMA-free RPMI medium supplemented with 10% FBS, and cells were washed once with Hanks’ balanced salt solution (HBSS; Thermo Scientific). All bacterial infections were carried out at an MOI of 20:1 (initial cell screen), 10:1 (CF isolate screen with macrophage cell lines from mouse and human origin), or 5:1 (all subsequent assays). GN008 and GN050 were cultured (see “Bacterial culture,” above) and emulsified from an agar plate into serum-free cell culture medium. The optical density at 600 nm (OD₆₀₀) was measured, and the culture was diluted to the proper concentration for the desired MOI in serum-free cell culture medium. Dilutions were plated on LB agar plates to confirm the MOI. In all assays, excluding the initial cell screen or CF isolate screen, bacteria were subjected to centrifugation onto eukaryotic cells at 600 × g for 5 min.

P. aeruginosa cell infections.

For all infections involving P. aeruginosa, A549 lung epithelial cells (ATCC) were seeded at 5.5 × 10⁴ cells per cm² in 12-well, 24-well, or 60-mm tissue culture plates the day before infection. For immunofluorescence (IF) imaging, cells were seeded in 12-well plates containing 22-mm glass cover slides precoated with 100 μg/ml fibronectin bovine plasma (Sigma) in phosphate-buffered saline (PBS) for 1 h of shaking at room temperature the prior day. The next morning, A549 cells were washed with HBSS, and bacterial cultures were emulsified from agar plates and diluted in the same way as A. xylosoxidans, except the bacteria were not centrifuged onto the cell monolayer. All infections involving PA103 stains were carried out at an MOI of 5:1.

For Western blot analyses, each infection group was carried out in two 60-mm dishes for 6 h. As P. aeruginosa is an extracellular pathogen for A549 cells, bacterial fractions were obtained by collecting the cell infection supernatant into a new tube followed by centrifugation at 2,300 × g for 5 min at room temperature. Each pellet was suspended in 100 μl of lysis buffer (1× cOmplete, mini, EDTA-free protease inhibitor cocktail [Sigma], 0.5% Triton X-100 to selectively lyse any contaminating eukaryotic cells, 100 μg/ml RNase, and 100 μg/ml DNase in PBS). The suspension was subjected to centrifugation at 16,100 × g for 15 min at 4°C. The resulting pellet was suspended in 90 μl and mixed with 22.5 μl of 5× SDS loading buffer (0.1 M dithiothreitol). To obtain the eukaryotic fraction, the cellular monolayer left from the first step was gently collected using 750 μl of PBS per dish and washed with an additional 750 μl PBS. The suspension was subjected to centrifugation at 2,300 × g for 5 min at room temperature. The pellet was suspended in 100 μl of lysis buffer (0.5% Triton X-100 added) and incubated on ice for 15 min. The lysate was subjected to centrifugation at 16,100 × g for 15 min at 4°C. Approximately 90 μl of the supernatant was mixed with 22.5 μl of 5× SDS loading buffer (0.1 M dithiothreitol). Samples were boiled for 5 min prior to analysis with SDS-PAGE and Western blotting. Fractions from infections with PA103 pUCPexoUS142A were probed with a mouse anti-ExoU monoclonal antibody (1:5,000, 4°C, overnight), and fractions with PA103 pUCPpSsacUaxoUS201A or pUCPpSsacUaxoUS201A-HA were probed with a mouse anti-HA antibody (1:3,000, 4°C, overnight [Sigma]). Anti-sigma factor 70 (1:10,000, 4°C, overnight [Abcam]) and anti-GAPDH (1:2,000, 4°C, overnight [Santa Cruz Biotechnology]) were used as loading controls for bacterial and eukaryotic fractions, respectively. Blots were washed and incubated with a horseradish peroxidase-conjugated anti-mouse IgG (1:10,000, room temperature, 2 h [Invitrogen]).

Adenylate kinase (AK) release assays for J774a.1 cells.

Supernatants (300 μl) from infections carried out in 24-well tissue culture plates were harvested, and cell debris was removed by centrifugation at 16,100 × g for 15 min at room temperature. Approximately 100 μl of supernatant containing host-derived adenylate kinase was transferred to a black, clear-bottom, 96-well microplate (Greiner Bio-One). A ToxiLight bioassay (Lonza) was performed per the manufacturer’s instructions. AK reagent was suspended in assay buffer 15 min prior to analysis, and 100 μl of the assay mixture was added to each well from infected supernatants. After 5 min of incubation, the microplate was transferred to a SpectraMax plate reader for luminescence measurements. A normalization well included fresh tissue culture medium. Maximum cell lysis was achieved by treatment of uninfected cells with 0.5% Triton X-100 for 5 min.

Cytochalasin D treatments.

J774a.1 cells seeded for bacterial infections were treated with 0.1 to 3 μg/ml of cytochalasin D reconstituted in DMSO or treated with DMSO only for 1 h prior to infection. After treatment, macrophages were washed once with HBSS, and the MOI was set to 5:1 in culture medium containing either 0.1 to 3 μg/ml of cytochalasin D or an equal volume of DMSO.

Antibiotic protection assays.

J774a.1 cells seeded for bacterial infections were infected with GN008 or GN050 at an MOI of 5:1 in cell culture medium with DMSO or cell culture medium with 2 μg/ml of cytochalasin D for 1 h to allow for phagocytosis. Wells were washed 3 times with HBSS and replaced with cell culture medium containing 50 μg/ml of polymyxin B and 2 μg/ml of cytochalasin D or DMSO only. After 1 h of incubation, cells were washed 3 times with HBSS. J774a.1 cells were then lysed with 0.1% Triton X-100 in PBS. Bacterial cells were diluted serially and plated on LB agar to measure CFU. The percent bacterial uptake was calculated as a fraction of the total initial inoculum.

Microscopy and image processing.

Immunofluorescent images and live imaging were both acquired with a Nikon Eclipse Ti inverted microscope system equipped with a CoolSNAP ES2 charge-coupled-device (CCD) camera (Photometrics) with NIS-Elements AR imaging software (v.4.60). Images were captured using a Nikon 20× lens (Plan Apo 0.75 NA), while live images were captured using a Nikon 20× lens (Plan Fluor 0.45 NA). The live imaging microscope was equipped with a motorized stage and a Tokei Hit environmental chamber temperature controlled at 37°C and 5% clinical blood gas (Airgas).

P. aeruginosa infection of A549 cells was carried out for 6 h at an MOI of 5:1 on glass coverslips. The infection well was then washed 3 times with HBSS and fixed with 4% paraformaldehyde (PFA) in HBSS for 10 min with shaking at room temperature. Cells were washed 3 times with HBSS, and the coverslip was transferred to a new plate. Cells were incubated in 3% bovine serum albumin (BSA) in HBSS for 20 min with shaking at room temperature, followed by a primary antibody incubation (mouse anti-HA, 1:1,500 or mouse anti-ExoU monoclonal antibody, 1:1,500) for 1 h with shaking at room temperature in HBSS 3% BSA. Cells were washed 3 times with HBSS and incubated in secondary antibody (anti-mouse IgG Alexa Fluor 488; 1:1,000) for 1 h with shaking at room temperature in HBSS 3% BSA followed by 3 more washes using HBSS 3% BSA. Cells were then incubated in Texas Red-X phalloidin (Thermo Scientific) at the manufacturers’ recommended concentrations for 20 min with shaking at room temperature. Coverslips were washed 3 times with HBSS and mounted on a glass microscope slide with ProLong Gold antifade with DAPI (4′,6-diamidino-2-phenylindole; Thermo Scientific). The mounted coverslips were cured overnight protected from light and sealed prior to imaging analysis. For all groups, the images were acquired and analyzed in a similar manner, and fields were quantified by 2 individuals who were blinded as to the experimental groups. Each individual counted 10 randomly selected fields per group.

For live imaging analysis of A. xylosoxidans infection of J774a.1 or THP-1 macrophages, images were taken every 30 sec from 30 min to 6 h postinfection (Movies S1 and S2) or every 5 min from 30 min to 8 h postinfection (Fig. 3A and B, Fig. 6). Each individual image stack from the DSRed filter was quantified using Cell Profiler software (v.2.2.0) identifying fluorescent nuclei.

Gene expression analysis using RT-qPCR.

J774a.1 cells seeded for bacterial infections in 6-well tissue culture plates were infected with GN008 or GN050 at an MOI of 5:1 in tissue culture medium containing DMSO or tissue culture medium with 2 μg/ml of cytochalasin D. At the indicated time points, the infection supernatant was harvested and subjected to centrifugation at 16,100 × g for 5 min at room temperature, while the rest of the well was suspended in 500 μl of TRIzol reagent (Ambion). The suspension was then combined with the pellet, and samples were either processed or frozen at –80°C. Total RNA was extracted using the RNeasy plus universal minikit (Qiagen) and quantified using a Qubit fluorometer. For each group, DNase and reverse transcriptase treatments were performed using 2 μg of total RNA and SuperScript IV VILO (Thermo Scientific) per the manufacturers’ instructions. To assay for contaminating DNA, reactions excluding reverse transcriptase were performed. Approximately 75 ng of total cDNA was used in each qPCR experiment performed in triplicate using bacterial, gene-specific primers (0.5 μM) generated by the IDT PrimerQuest tool. Template and primers were mixed with nuclease-free water and 10 μl of 2× SsoAdvanced Universal SYBR green Supermix (Bio-Rad). The 20-μl qPCRs were performed on a CFX Connect real-time system (Bio-Rad) and analyzed using CFX Maestro software (v.4.1).

Clustering analysis, statistical analysis, and plotting.

K-means clustering analysis of A. xylosoxidans clinical isolates and corresponding plots were generated by Orange data mining software (v.3.24.1) using the average cytotoxicity of each isolate in J774a.1 and THP-1 macrophage cell lines. All other plotting and statistical analyses were performed using Python (v.3.4) in a Jupyter Lab notebook environment.

Data availability.

The complete genome sequences for A. xylosoxidans GN008 and GN050 are available from NCBI under accession numbers CP053618 and CP053617, respectively.