Peptidyl-Prolyl Isomerase ppiB Is Essential for Proteome Homeostasis and Virulence in Burkholderia pseudomallei

Burkholderia pseudomallei is the causative agent of melioidosis, a disease endemic to Southeast Asia and northern Australia. Mortality rates in these areas are high even with antimicrobial treatment, and there are few options for effective therapy. Therefore, there is a need to identify antibacterial targets for the development of novel treatments. Cyclophilins are a family of highly conserved enzymes important in multiple cellular processes.

ABSTRACT

Burkholderia pseudomallei is the causative agent of melioidosis, a disease endemic to Southeast Asia and northern Australia. Mortality rates in these areas are high even with antimicrobial treatment, and there are few options for effective therapy. Therefore, there is a need to identify antibacterial targets for the development of novel treatments. Cyclophilins are a family of highly conserved enzymes important in multiple cellular processes. Cyclophilins catalyze the cis-trans isomerization of xaa-proline bonds, a rate-limiting step in protein folding which has been shown to be important for bacterial virulence. B. pseudomallei carries a putative cyclophilin B gene, ppiB, the role of which was investigated. A B. pseudomallei ΔppiB (BpsΔppiB) mutant strain demonstrates impaired biofilm formation and reduced motility. Macrophage invasion and survival assays showed that although the BpsΔppiB strain retained the ability to infect macrophages, it had reduced survival and lacked the ability to spread cell to cell, indicating ppiB is essential for B. pseudomallei virulence. This is reflected in the BALB/c mouse infection model, demonstrating the requirement of ppiB for in vivo disease dissemination and progression. Proteomic analysis demonstrates that the loss of PpiB leads to pleiotropic effects, supporting the role of PpiB in maintaining proteome homeostasis. The loss of PpiB leads to decreased abundance of multiple virulence determinants, including flagellar machinery and alterations in type VI secretion system proteins. In addition, the loss of ppiB leads to increased sensitivity toward multiple antibiotics, including meropenem and doxycycline, highlighting ppiB inhibition as a promising antivirulence target to both treat B. pseudomallei infections and increase antibiotic efficacy.

INTRODUCTION

Burkholderia pseudomallei is a Gram-negative soil saprophyte found in tropical and subtropical areas around the world, such as in Southeast Asia and northern Australia (1–3). It is the causative agent of melioidosis and has been reported in 45 countries, with a predicted global burden of 165,000 cases and 89,000 deaths annually (4). Melioidosis can present as a variety of clinical syndromes, ranging from nonhealing skin lesions to intraabdominal abscesses to pneumonia and septicemia (5), leading to difficulty in prompt diagnosis, particularly in regions where it is not endemic. Mortality rates vary depending on geographic location, with rates ranging from 14% in Darwin (5) to 49% in northeast Thailand (6). Treatment of melioidosis is prolonged, consisting of two phases: a 2-week intensive intravenous phase followed by a 3- to 6-month oral eradication phase (7, 8). Due to intrinsic resistance to antimicrobials (9), treatment of B. pseudomallei infection can be further complicated by the limited number of viable antimicrobial alternatives. Relapse of infection is common and is associated with increased mortality, particularly in cases where treatment is unsuccessful or an incomplete course of antimicrobial therapy is taken (10, 11).

B. pseudomallei infections can be difficult to overcome due to the bacterium encoding an array of defense mechanisms that enable successful survival in diverse environments, including inside mammalian host cells. The ability of B. pseudomallei to form biofilms allows it to persist in the environment and has been implicated in infection (12–14). B. pseudomallei encodes flagellin, important for disease dissemination and virulence in BALB/c mouse infection models (15, 16). Intracellular survival is reliant on B. pseudomallei rapidly escaping from the phagolysosome and establishing a replicative niche in the cytosol of eukaryotic cells (17). The ability to escape into the cytosol is dependent on multiple secretion systems, which function to deliver specialized secreted proteins known as effectors into the host that enhance bacterial survival and enable the spread of B. pseudomallei into neighboring cells (18). Three different type III secretion systems (T3SS-1, -2, and -3) are found in B. pseudomallei, with only T3SS-3 required for full virulence in a hamster model of infection (17, 19–21). Following phagosome escape and replication in the cytosol, expression of type VI secretion systems (T6SS) is induced and is essential for in vivo virulence (22–24). A well-documented phenomenon of B. pseudomallei is the formation of multinucleated giant cells (MNGC). This formation has been attributed to the T6SS-5 (T6SS cluster 1) effector VgrG-5, which is required to stimulate cell fusion and leads to the spread of infection (22, 25, 26). Six clusters of T6SS are found in B. pseudomallei, with T6SS cluster 1, defined by Shalom et al. (24) as tss5, shown to play a role in the formation of MNGC and cellular cytotoxicity (26, 27). Throughout this paper, the Schell et al. (28) nomenclature for T6SS will be used.

Cyclophilins are part of the immunophilin superfamily, with bacteria generally carrying two cyclophilin genes, ppiA and ppiB, with one located in the cytoplasm and the other in the periplasm or outer membrane, respectively (29). Cyclophilins catalyze the cis-trans isomerization of xaa-proline bonds, a rate-limiting step in protein folding, which is required for proteome homeostasis (30, 31). Cyclophilins are required for optimal protein folding, and in multiple bacterial systems cyclophilins are important for stress response and infectivity, suggesting a role in folding virulence factors. In Brucella abortus, the expression of both CypA and CypB becomes elevated during intracellular infection, with deletion of these genes resulting in virulence attenuation, reduced intracellular survival, and increased susceptibility to acidic and oxidative stress (32). Further evidence of the role of cyclophilins in virulence is demonstrated in Legionella pneumophila, where the cyclophilin gene cyp18 is essential for optimal intracellular survival in Acanthamoeba castellanii (33). Cyclophilins also play an important role in biofilm formation, with Escherichia coli ppiB having been shown to be a negative regulator of both motility and biofilm formation, and mutagenesis of E. coli ppiB results in hypermotility and increased biofilm formation (34). The pleiotropic role of cyclophilins in bacteria is also demonstrated in E. coli, where interaction of PpiB with the protein FtsZ is important for correct cell division, with deletion of ppiB resulting in aberrant cell division and formation of filamentous cells (35).

B. pseudomallei carries a ppiB gene, the role of which was investigated by construction of a B. pseudomallei ΔppiB (BpsΔppiB) null mutant strain. In vitro characterization of the BpsΔppiB strain demonstrates a loss of multiple virulence determinants, including reduced motility and biofilm production. Intracellular survival of the BpsΔppiB strain was significantly reduced, with bacteria confined within the macrophage cell, lacking the ability to spread cell to cell, indicating ppiB is important for B. pseudomallei virulence. This is reflected in the BALB/c mouse infection model in which the BpsΔppiB strain was avirulent, demonstrating the important role for ppiB in in vivo disease dissemination and progression. Proteomic analysis confirms widespread alterations within the BpsΔppiB strain that are partially restored by complementation of ppiB. Consistent with this, complementation restored multinucleated cell formation and cell disruption. Finally, we demonstrate that loss of ppiB leads to increased susceptibility to first-line treatment antibiotics, such as meropenem and doxycycline. Thus, this study shows the importance of ppiB for virulence of B. pseudomallei and how disruption of proteome homeostasis may be targeted to sensitize B. pseudomallei to antibiotic regimens.

RESULTS

Deletion of ppiB gene in B. pseudomallei.

BPSL2246 (UnitProt entry Q63SS5) encodes a putative cytoplasmic cyclophilin, a ppiB homolog in B. pseudomallei (see Fig. S1 in the supplemental material). There is 66.3% and 57.7% protein identity with cyclophilin B homologs from E. coli and L. pneumophila, respectively, with residues involved in enzymatic activity also being conserved (see Fig. S1 in the supplemental material) (36). To determine the role of ppiB in B. pseudomallei strain K96243, the gene was deleted by construction of an in-frame null mutation strain, the BpsΔppiB strain (37). Deletion of ppiB in the BpsΔppiB strain was confirmed by whole-genome sequencing. Compared to the parent K96243 strain, there was one additional single-nucleotide polymorphism (SNP) in the BpsΔppiB strain that resulted in a missense mutation in rpoZ (RpoZ_Leu10Pro). No differences in growth were observed between the wild-type (BpsWT) and BpsΔppiB mutant strain in either Luria-Bertani (LB) broth or M9 minimal medium (Fig. S2).

BpsΔppiB strain can infect murine macrophage cells in vitro but has reduced intracellular numbers 6 and 9 h postinfection.

Murine macrophages, J774.1 cells, were infected with either the wild-type or BpsΔppiB strain, and bacteria were enumerated at various time points postinfection (Fig. 1). There were no significant differences in the levels of adherence and invasion of macrophages (data not shown). Significant differences in the numbers of intracellular bacteria at 6 (14-fold; P = 0.0159 by Mann-Whitney U test) and 9 (36.8-fold; P = 0.0159 by Mann-Whitney U test) h postinfection were seen with the BpsΔppiB strain, demonstrating reduced survival and/or replication. At 12 h the BpsΔppiB strain was able to overcome the reduced-growth phenotype, showing levels of intracellular bacteria similar to that of the BpsWT parental control. At 24 h, although the levels of intracellular bacteria in both BpsWT and BpsΔppiB strains appeared to be similar, there was a substantial reduction in cell cytotoxicity caused by the BpsΔppiB strain compared to that of the BpsWT using lactate dehydrogenase (LDH) cytotoxicity screening (Fig. S3). Together these results demonstrate the BpsΔppiB strain has reduced growth and/or survival in macrophage cells.

FIG 1.

BpsΔppiB strain shows reduced intracellular survival in J774.A1 murine macrophages. Intracellular growth of BpsWT (●) and BpsΔppiB (▲) strains. J774A.1 murine macrophages were infected at an MOI of 10, and intracellular counts were taken at 0, 3, 6, 9, and 12 h. Graphs are means from five biological replicates, with each having two technical replicates. The error bars display the standard errors of the means. *, P = 0.0159 by Mann-Whitney U test.

ppiB is essential for in vivo infection.

As in vitro results demonstrated a decrease in intracellular counts during early time points, the role of ppiB during infection was further investigated using the BALB/c mouse infection model of B. pseudomallei. Groups of mice were challenged by the intraperitoneal route with either the BpsWT or BpsΔppiB strain (Fig. 2A). At the end of the experiment, 100% of the animals challenged with the BpsΔppiB strain survived, compared to 33% in the group challenged with the BpsWT (P = 0.0183 by log rank [Mantel-Cox] test). Disease progression was also monitored by measuring weight loss during the infection study, and all except two mice infected with the BpsWT showed considerable weight loss (Fig. 2B). In contrast, mice infected with the BpsΔppiB strain demonstrated no weight loss throughout the experiment (Fig. 2C). The lungs, livers, and spleens in survivors were enumerated for bacteremia, and all were clear from infection at the conclusion of the experiment. This demonstrates that ppiB is essential for B. pseudomallei to successfully establish in vivo infection.

FIG 2.

BpsΔppiB strain is attenuated in the BALB/c mouse model of infection. (A) BALB/c mice (n = 6) were injected intraperitoneally with 1.1 × 10⁴ CFU BpsWT (●) and 1.86 × 10⁴ CFU of BpsΔppiB (▲) strains. *, P value of 0.0183 by log-rank test. Weight loss of individual BALB/c mice (labeled 1 to 6) was monitored daily following intraperitoneal infection as a measure of morbidity in BpsWT (B) and BpsΔppiB (C) strains.

BpsΔppiB strain demonstrates reduced ability to form MNGC.

The in vitro cell infection study suggests the BpsΔppiB strain is attenuated; in contrast, the mouse infection studies demonstrate that the BpsΔppiB strain is avirulent. To investigate this in more detail, a complemented strain was constructed, the BpsΔppiB/ppiB strain, and was further characterized in macrophage cells. BpsWT, BpsΔppiB, and BpsΔppiB/ppiB strain-infected cells were analyzed by immunofluorescence microscopy 12 h postinfection. Figure 3A shows that during a later time point of 12 h postinfection, macrophage cells infected with the BpsWT had multiple MNGC formations. In contrast, BpsΔppiB strain-infected cells demonstrate a significant reduction in MNGC formation, although actin protrusions are still observed (Fig. 3B). Complementation of ppiB demonstrates restoration to a BpsWT phenotype (Fig. 3C). Enumeration of nuclei within multinucleated cells relative to the number in mononucleated cells shows that there is a 67.4% reduction in nuclei associated with MNGCs in macrophage monolayers infected with the BpsΔppiB strain, and this is significantly restored in the BpsΔppiB/ppiB complemented strain (Fig. 3D), confirming the role of ppiB in the virulence of B. pseudomallei during intracellular infection.

FIG 3.

BpsΔppiB demonstrates reduced formation of multinucleated giant cells in J774 murine macrophages. Fluorescently stained monolayers infected with the BpsWT (A), BpsΔppiB (B), or BpsΔppiB/ppiB (C) strain were stained with anti-Bps-LPS-FITC, and nuclei were stained with Hoechst. The bar indicates 14 μm. (D) Percentage of nuclei associated with an MNGC for BpsWT (●), BpsΔppiB (▲) and BpsΔppiB/ppiB (■) strains. Graphs are the result from three biological repeats with each biological repeat containing two technical repeats. One thousand nuclei were counted from each coverslip, either being part of a multinucleated cell or mononucleated, and then the percentage of multinucleated cells was calculated. *, P = 0.026; **, P = 0.0022; both by Mann-Whitney U test. ns, not significant.

BpsΔppiB strain reveals marked changes in the proteome.

To understand the changes driving the alterations in virulence, we analyzed the proteome of the BpsΔppiB, its BpsΔppiB/ppiB complement, and BpsWT strains. Using label-free-based quantitative (LFQ) proteomics, we identified 2,091 proteins with high consistency observed across biological replicates, as determined by Pearson correlations (average, 0.95) (Fig. S4). Consistent with the role of PpiB in multiple cellular pathways, 734 proteins underwent statistically significant alterations within the proteome of the BpsΔppiB strain compared to levels in the BpsWT (Table S1), with these proteins predicted to be localized to multiple cellular compartments (Fig. 4B). Consistent with the loss of PpiB in the BpsΔppiB strain, this protein demonstrated the largest fold difference of −10.14 log₂ within the proteome, with the majority of altered proteins also showing a decreased abundance in response to loss of PpiB (Fig. 4A). Upon complementation, PpiB levels were restored to 10% of the level of the wild type, yet consistent with phenotypic assays, this led to restoration of proteins observed to increase (Fig. 4C) and decrease (Fig. 4D) to nearly wild-type levels. Within the categories of proteins that had reduced abundance upon the loss of PpiB, we observed alterations in multiple proteins associated with motility, consistent with the reductions in the BpsΔppiB strain, including BPSL3305 [CheW, −2.78587 log₂; −log₁₀(P value), 4.38] and BPSL3301 [CheBI, −2.48162 log₂; −log₁₀(P value), 6.03] as well as the flagellin [BPSL3319; FliC, −3.38141 log₂; −log₁₀(P value), 5.81]. Again, consistent with phenotypic assay complementation, only partial restoration of these proteins occurred, with FliC only restored by 0.9 log₂. Other changes were observed in response to the loss of PpiB, including reduction in capsule-associated proteins BPSL2799 [WcbI, −1.32 log₂; −log₁₀(P value), 4.30], BPSL2800 [WcbH, −0.47 log₂; −log₁₀(P value), 1.36], BPSL2807 [WcbC, −0.36 log₂; −log₁₀(P value), 1.61] (highlighted in green in Fig. 4D), and BPSL2810 [ManC, 0.80 log₂; −log₁₀(P value), 2.33], as well as increases in multiple components of the T6SS-3s, including BPSS2099 [TssCIII, 4.49 log₂; −log₁₀(P value), 3.38] and BPSS2098 [HcpIII, 3.99 log₂; −log₁₀(P value), 4.28] (highlighted in blue in Fig. 4A). A KEGG pathway analysis was undertaken to determine what functional pathways were being affected in the BpsΔppiB strain (Fig. 5). Metabolism accounted for 268 differentially present proteins, with 170 being increased while 98 were decreased. Genetic information processing, signaling and cellular processes, and environmental information processing-related proteins were the majority of proteins affected. In addition, many hypothetical or unassigned proteins were differentially present.

FIG 4.

Quantitative proteomic analysis of BpsWT versus BpsΔppiB strain. Label-free quantification was undertaken to compare the BpsΔppiB strain to BpsWT. (A) Identified proteins are presented as a volcano plot depicting mean label-free quantitation (LFQ) intensity ratios of the BpsΔppiB strain versus BpsWT plotted against logarithmic t test P values from four biological experiments of each strain. (B) GO term-assigned localization of 42 out of the 213 proteins that undergo statistically significant changes with localization assignment. Only GO localization terms for groups with greater than 3 entries are shown. Complementation of PpiB leads to restoration of proteins observed to increase (C) and decrease (D) to nearly BpsWT levels.

FIG 5.

KEGG pathway analysis of proteins differentially present in BpsWT versus BpsΔppiB strain. Proteins that were differentially present by proteomics (Table S1) were manually curated using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database against the Burkholderia pseudomallei K96243 genome (entry number T00203) and assigned a KEGG Orthology (KO). Proteins in black were increased in the BpsΔppiB relative to the BpsWT strain, while proteins in gray were decreased. Numerous proteins were predicted to be in other functional groups, but only the highest KO was taken down.

BpsΔppiB strain has decreased motility and biofilm formation under nutrient-rich conditions.

Motility of B. pseudomallei has been shown to be important for successful establishment of in vitro and in vivo infections (15, 16). The motility of the BpsΔppiB strain was determined using a swarming assay (Fig. 6A). The mean bacterial spread from the site of inoculation of the BpsWT parental strain was 48.5 mm after 24 h. In comparison, the spread of the BpsΔppiB strain was reduced to 24.5 mm (P = 0.0022 by Mann-Whitney U test) but was not restored in the complemented strain, consistent with the partial restoration of motility-associated protein levels, as shown by the proteomics studies. This suggests a role for PpiB in motility, although further research is required to determine if this is a direct effect on the flagellum or due to regulatory or sensory deficits.

FIG 6.

BpsΔppiB strain has significantly reduced motility and biofilm formation. (A) Swarming motility of BpsWT (●), BpsΔppiB (▲), and BpsΔppiB/ppiB (■) strains through 0.3% agarose plates. Values are the diameter of spread, with readings taken at 24 h postinoculation. Results are from three biological replicates. **, P = 0.0022 by Mann-Whitney U test. (B) Biofilm-forming capacity of BpsWT (●), BpsΔppiB (▲), and BpsΔppiB/ppiB (■) strains in nutrient LB broth. Biofilms were allowed to form over a 48-h period before being fixed with methanol and stained with crystal violet to determine bacterial biomass. Crystal violet was solubilized with 33% glacial acetic acid, and optical density was read with a spectrophotometer (Bio-Rad Xmark) at 590 nm (OD₅₉₀). Bars are representative of the mean optical density, with individual values plotted. Six biological replicates with 6 technical repeats were conducted. **, P = 0.0022 by Mann-Whitney U test.

Another important survival mechanism in the environment and potentially for establishing a chronic infection is the ability to form a biofilm. The BpsΔppiB strain demonstrates significant attenuation in the formation of biofilms under nutrient-rich conditions compared to that of the BpsWT (P = 0.0022 by Mann-Whitney U test); again, this was not restored in the complemented strain (Fig. 6B).

BpsΔppiB strain has greater susceptibility to antimicrobial and intracellular stresses.

The reduced survival of the BpsΔppiB strain under both in vitro and in vivo conditions may be a consequence of incorrect folding of proteins involved in resistance to intracellular stresses such as peroxide and acid tolerance. This was determined by MICs of the BpsΔppiB strain to hydrogen peroxide and hydrochloric acid. The BpsΔppiB strain demonstrates greater sensitivity toward oxidative stress, with a significant 3-fold reduction in the MIC of hydrogen peroxide exposure compared to that of the BpsWT, which is partially restored in the BpsΔppiB/ppiB strain (Fig. 7). There was, however, no increased susceptibility to acid stress, perhaps reflected in the ability of the BpsΔppiB strain to survive in cells (Table S2).

FIG 7.

BpsΔppiB strain demonstrates greater sensitivity toward oxidative stress. Survival of BpsWT (●), BpsΔppiB (▲), and BpsΔppiB/ppiB (■) strains in increasing concentrations of hydrogen peroxide. Values are the mean MICs from three biological replicates. Concentrations are in microliters per milliliter of 30% hydrogen peroxide (BioVar) solution. MIC was determined by measuring the optical density at 590 nm at 24 h after hydrogen peroxide exposure, and the MIC was called the lowest concentration which resulted in less than 20% growth (dotted line) of the unexposed control on that plate.

B. pseudomallei is intrinsically resistant to antimicrobials that are cleared by active efflux pumps (9). It is feasible that PpiB is involved in protein folding of some efflux pumps; as such, the susceptibility to antimicrobials was investigated. In particular, the MICs to antimicrobials that are currently used for B. pseudomallei treatment were determined (Table 1). The BpsΔppiB strain displayed a 4-fold increase in susceptibility (128 μg/ml to 8 μg/ml) to the 3rd-generation cephalosporin ceftriaxone. B. pseudomallei is intrinsically resistant to 3rd-generation cephalosporins, indicating that some mechanism of resistance is being modulated by PpiB. There was a 2-fold decrease in resistance to tetracycline (2 μg/ml to 0.5 μg/ml) and its derivative, doxycycline (1 μg/ml to <0.25 μg/ml). Tetracyclines are involved in protein synthesis inhibition, and the main mechanism of resistance in B. pseudomallei is efflux out of the cell, confirming a potential role for ppiB in the correct folding of efflux pumps. The BpsΔppiB/ppiB complemented strain partially restores resistance to wild-type levels.

TABLE 1.

MIC as determined by broth microdilutions^a

Antibiotic	MIC (μg/μl) for:
	BpsWT	BpsΔppiB	BpsΔppiB/ppiB
Ceftriaxone	128	8	16
Meropenem	2	0.5	1
Tetracycline	2	0.5	1
Doxycycline	1	<0.25	0.5

^aValues are the mean MICs from three biological replicates. MIC was determined by measuring the optical density at 590 nm at 24 h after antibiotic exposure, and the MIC was called the lowest concentration that resulted in less than 20% growth of the unexposed control on that plate.

DISCUSSION

Previous work has demonstrated that cyclophilin B plays a role in modulating virulence in a number of bacterial species, resulting in attenuation in vivo (32, 33, 38, 39). Consistent with these studies, we demonstrate that cyclophilin B in B. pseudomallei influences multiple virulence-associated phenotypes, with loss of the ppiB gene resulting in complete attenuation in the BALB/c mouse infection model (Fig. 2). Although most work has focused on the role of cyclophilins in virulence, recently the direct interaction of cyclophilin B with intracellular proteins important for bacterial growth and survival, such as DnaK, AccC, and FtsZ, has been reported (34, 35, 39–41). It is shown here that B. pseudomallei deletion of ppiB leads to pleiotropic effects, including stress intolerance, reduction in motility, and biofilm formation. Furthermore, the direct effect of ppiB loss on the proteome homeostasis of B. pseudomallei has been defined. Key pathways important for virulence modulation have been identified and disrupted, providing evidence of the importance of PpiB in bacterial protein folding and overall virulence.

B. pseudomallei is able to infect a wide range of cells in order to survive and cause disease (42). Infection of murine macrophages demonstrated that the BpsΔppiB strain retained its ability to adhere and invade macrophage cells with reduced survival 6 and 9 h postinfection (Fig. 1), with intracellular counts similar to that of the parental control reached by 12 h. This delayed growth phenotype has been shown with the disruption of type VI/III secretion systems and is important in cell-to-cell spread of B. pseudomallei (27, 43, 44). The BpsΔppiB strain is unable to effectively spread intracellularly, as determined by immunofluorescence, and cause cell fusion into MNGC, with complementation studies showing restoration of the BpsWT phenotype (Fig. 3B). As seen 12 h postinfection, the BpsWT and the BpsΔppiB/ppiB complemented strain display marked bacterial movement throughout the monolayer and cellular fusion into MNGC (Fig. 3A and C). Enumeration of nuclei also indicates a significant reduction in the formation of MNGC in the BpsΔppiB strain, which is restored in the BpsΔppiB/ppiB complement (Fig. 3D). Complementation is observed despite only a 10% restoration of PpiB protein levels, demonstrating that even low levels of PpiB are sufficient to overcome some virulence deficiencies, something which has been previously noted in Saccharomyces cerevisiae (45). This lack of cell-to-cell spread is characteristic of various mutants of T6SS cluster 1 (26, 27, 46), in particular T6SS cluster 1 Δhcp1 and ΔvgrG1 mutants (26, 27). It is hypothesized that PpiB plays a role in either folding or regulating the expression of type VI secretion systems in B. pseudomallei. It was reported that T6SSs in B. pseudomallei are kept under strict transcriptional control and are only induced upon invasion of macrophages (24), yet despite this, our proteomic data showed an increase in BPSS2098 [Hcp-1, 1.10 log₂; −log₁₀(P value), 3.05] and BPSS2099 [Tss-1, 4.06 log₂; −log₁₀(P value), 1.8], two proteins belonging to T6SS-3, a cluster which is usually not expressed in nutrient media (27, 47). Interestingly, levels of BPSL3097 [−1.28 log₂; −log₁₀(P value), 6.12], BPSL3099 [−0.87 log₂; −log₁₀(P value), 2.89], BPSL3105 [−0.75 log₂; −log₁₀(P value), 3.31], BPSL3106 [−0.51 log₂; −log₁₀(P value), 2.45], and BPSL3108 [−0.82 log₂; −log₁₀(P value), 4.37] were all decreased in the BpsΔppiB strain. These belong to T6SS cluster 6, which has been shown to be the only T6SS expressed in nutrient media (27). This again points to a dysregulation of transcriptional or translational control. Additionally, a MarR-family regulator (BPSL3431) shown to be involved in regulation of T6SS transcription was downregulated [−0.67 log₂; −log₁₀(P value), 3.40] (48). This indicates that T6SS proteins are escaping their tight transcriptional control in nutrient media, and it is hypothesized that the same is occurring upon infection in cells, resulting in a malfunctional T6SS and, hence, the loss of MNGC formation. Further assessment of the transcriptome and proteome of the BpsΔppiB strain upon invasion of macrophages would be useful to verify if PpiB is playing a role at the transcriptional or translational level of T6SS regulators or machinery. This also explains the clearance of infection in the BALB/c mouse studies where no viable bacteria were recovered at the end of the experiment, demonstrating the essential role of PpiB and its potential as a novel antivirulence target.

Motility and biofilm formation are important for establishing B. pseudomallei infection (13, 15). Deletion of the flagellum (ΔfliC) has been shown to be important for virulence in the BALB/c mouse infection model (15, 49). Here, a decrease in the protein levels of FliC [BPSL3319, −3.38 log₂; −log₁₀(P value), 5.81] is observed, consistent with the reduction in motility in the BpsΔppiB strain, with recent studies having demonstrated that PpiB from Clostridioides difficile interacts with FliC using bacterial two-hybrid systems (50). Furthermore, reduced levels of CheBI [BPSL3301, −2.48 log₂; −log₁₀(P value), 6.02] and CheW [BPSL3305, −2.78 log₂; −log₁₀(P value), 4.38] (Fig. 6B), important for chemotactic directed motility (51–54), may also contribute to the observed reduction in motility. As the BpsΔppiB strain displayed only a decrease in motility, not a complete loss, there may be a dysregulation of the signal transduction pathways leading to a delay or absence of appropriate signaling to begin movement rather than elimination of the flagellum in the BpsΔppiB strain, although additional studies are required to validate this. Motility has also been implicated as an important factor for biofilm production, with ΔfliC mutants showing a decrease in biofilm production (55). Transcriptomics have identified B. pseudomallei genes important in biofilm formation (56), and of these genes flagged as differentially regulated, 12 were present at opposing protein abundances in the BpsΔppiB strain, possibly explaining the decrease in biofilm formation by the BpsΔppiB strain (Fig. 6B), with genes such as universal stress proteins [BPSS0837, −1.16 log₂; −log₁₀(P value), 4.66; BPSS1140, −0.57 log₂; −log₁₀(P value), 2.03], receptors [BPSS1742, −1.22 log₂; −log₁₀(P value), 4.10], and efflux pumps [BPSL0816, 0.35 log₂; −log₁₀(P value), 2.13] being differentially expressed. This decrease in biofilm formation is in stark contrast to what has been reported in E. coli, where PpiB is a negative regulator of both biofilm and motility, with deletion of ppiB resulting in hypermotility and increased biofilm production (34). This disparity indicates different roles for PpiB in E. coli and B. pseudomallei, but a lack of in vivo data with the ΔppiB strain makes it difficult to determine the overall effect on virulence.

Cyclophilin B in Gram negatives is known to play a role in response to a variety of stresses encountered during infection (32). B. pseudomallei is exposed to reactive oxidizing species within phagocytes, a natural defense mechanism for eukaryotic cells (57), with loss of ppiB increasing the susceptibility to oxidative stress (Fig. 7). Others have shown that there are a variety of mechanisms by which B. pseudomallei responds to and tolerates oxidative stress, with quorum sensing regulating expression of genes important in protecting the cell against DNA damage, and polyphosphate kinases also play a role (55, 58, 59). Although none of these genes appear in the proteomics screen, levels of proteins involved in stress [BPSS0837, −1.3 log₂; −log₁₀(P value), 3.63; BPSS1140, −0.57 log₂; −log₁₀(P value), 2.03] and OmpR, an oxidative stress two-component system transcriptional regulator [BPSL2094, −0.53 log₂; −log₁₀(P value), 2.69], are decreased and may play an unknown role in oxidative stress response. These results indicate that other currently unknown mechanisms exist to combat oxidative stress, and this warrants further investigation.

B. pseudomallei has a number of chromosomal genes associated with antimicrobial resistance and, hence, is intrinsically resistant to most antibiotics used to treat serious infections (9). The BpsΔppiB strain has increased susceptibility to ceftriaxone, tetracycline, and doxycycline, antibiotics currently used for treatment of melioidosis (Table 1) (7, 8). Resistance to these antibiotics is shown to be moderated by resistance-nodulation-division (RND) efflux pumps (60), of which B. pseudomallei strain K96243 has 10 annotated within its genome, as well as by beta-lactamases (9, 61). Differences in the protein levels of the efflux pump components AmrA [BPSL1804, −1.66 log₂; −log₁₀(P value), 2.98], BpeB [BPSL0815, 0.2 log₂; −log₁₀(P value), 1.97], and OprB [BPSL2094, −0.53; −log₁₀(P value), 2.69] indicate that loss of PpiB results in malfunctioning pumps. It is hypothesized that deletion of ppiB results in a reduction of the pump components, making ineffective pumps and restoring susceptibility to certain antibiotics.

There are many reports on the pleiotropic effects that immunophilin proteins have in cells, and this study demonstrates that virtually every compartment within the cell displays gross proteomic changes, especially those involved in metabolism and genetic information processing (32, 34, 38, 50, 62). It has recently been shown by Rasch et al. (38) that proteins from the immunophilin family also have the ability to compensate for one another; in this case the macrophage infectivity potentiator (Mip) protein, belonging to the FK506-binding protein family, is able to compensate for the loss of PpiB in L. pneumophila (38). This compensatory effect has always been theorized, and these studies show that there is an increase in the protein levels of three immunophilin proteins, Mip [BPSS1823, 0.98 log₂; −log₁₀(P value), 2.42], PpiA [BPSL2245, 1.1 log₂; −log₁₀(P value), 3.05], and SurA [BPSL0659, 0.53 log₂; −log₁₀(P value), 2.53]. Whether these proteins can compensate for PpiB loss in B. pseudomallei, and to what degree, requires further investigation.

To conclude, PpiB in B. pseudomallei is essential for virulence, with the BpsΔppiB deletion mutant displaying pleiotropic effects on virulence determinants such as the flagella, biofilm production, and antimicrobial susceptibility. Infection of macrophages with the BpsΔppiB strain resulted in a delayed growth phenotype and an inability to cause fulminant disease in BALB/c mice. On closer investigation it was shown that this was due to the BpsΔppiB strain being unable to spread cell to cell and form MNGCs, indicating that clearance of infection occurs in vivo. Whole-cell proteomic analysis reveals marked changes in the proteome, including proteins previously shown to be important for cell-to-cell spread and virulence of B. pseudomallei, and has also identified a plethora of new proteins potentially playing an important role in infection. Although further work still needs to be conducted to demonstrate the direct interactions of PpiB with its folding partners, it is clear that PpiB is essential for the correct protein folding of virulence determinants in B. pseudomallei, making it indispensable for virulence.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains used in this study are shown in Table 2. All bacterial strains were grown in Luria-Bertani (LB) broth overnight at 37°C with agitation unless stated otherwise. Antibiotics were used at the following final concentrations: ampicillin, 50 μg/ml; chloramphenicol, 30 μg/ml; kanamycin, 50 μg/ml.

TABLE 2.

Bacterial strains used in this study

Strain	Genotype or description	Source or reference
Escherichia coli
TOP10	Chemically competent cloning strain	Invitrogen
S17-1 λpir	S17-1 with a λ prophage carrying the pir gene, conjugal strain for the movement of pDM4	75
ST18	S17-1 λpirΔhemA, conjugal strain for the movement of pBBR1-MCS1	76
Burkholderia pseudomallei
K96243 (BpsWT)	Clinical isolate	Dstl; 61
ΔppiB	K96243 derivative; unmarked deletion ΔppiB	This study
ΔppiB/ppiB	K96243 derivative; unmarked deletion ΔppiB; ppiB_3XFLAG_pBBR1-MCS1	This study

Construction of in-frame deletion mutant of ppiB.

Construction of B. pseudomallei in-frame deletion mutants was performed using the technique previously described by Logue et al. (37). For ppiB, a 449-bp upstream flanking region and a 403-bp downstream flanking region were amplified by PCR from B. pseudomallei K96243 genomic DNA (obtained using the Qiagen Gentra Purgene Yeast/Bact kit) using the primer pairs ppiB_UP_F/ppiB_UP_R (TCTAGATTCCATCGCGTGATCAAGGG/AGATCTTGGTTCCTTCGATGGATGGG) and ppiB_DN_F/ppiB_DN_R (AGATCTGGGATGTTGCAGGAGACACC/TCTAGATTGCCGAACGCGACGATG). Restriction sites (underlined above) were incorporated into the primers to allow for the ligation of the flanks to one another (using BglII) and XbaI to allow for the insertion of the joint flanks into the suicide plasmid pDM4. Upon construction of an upstream-downstream fragment and its subsequent ligation into pDM4, the construct was transformed by heat shock into E. coli S17-1 λpir, which was made calcium competent, and selected for with the antibiotic chloramphenicol. Following conjugation with B. pseudomallei K96243, merodiploid integrants that had successfully integrated the upstream-downstream pDM4 construct were identified using double antibiotic selection of ampicillin and chloramphenicol. A merodiploid integrant was plated onto LB agar lacking sodium chloride but containing 10% sucrose. sacB counterselection was used to select for the excision of the pDM4 backbone, resulting in an in-frame unmarked deletion. Colonies were subsequently screened for chloramphenicol sensitivity and analyzed by PCR to determine their phenotype, either wild-type revertant or in-frame deletion mutant. Colonies determined to be in-frame deletion mutants had the site of recombination sequenced (Sanger sequencing) to confirm a 492-bp deletion of ppiB. The B. pseudomallei BpsΔppiB mutant strain and the parent B. pseudomallei K96243 strain were sequenced using Illumina MiSeq (Murdoch University, Perth, Western Australia, Australia) or HiSeqMMD (Australian Genome Research Facility, Melbourne, Australia), respectively. Whole-genome sequencing data were aligned to the K96243 reference genome (versions NC_006350.1 [chromosome 1] and NC_006350.1 [chromosome 2]), and variants were identified using the SPANDx pipeline (62).

J774A.1 murine macrophage infection assay.

J774A.1 murine macrophages were seeded into a 24-well tissue culture treated plate at a concentration of 4 × 10⁵ cells/ml in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) supplemented with a final concentration of 1% GlutaMAX (Gibco, Life Technologies) and 10% heat-inactivated fetal calf serum (lot number 1939338; Gibco, Life Technologies) and incubated for 20 h at 37°C with 5% CO₂. B. pseudomallei strains were grown overnight at 37°C for 18 h and adjusted in Leibovitz L-15 medium (Gibco) supplemented with 1% GlutaMAX and 10% heat-inactivated fetal calf serum to an absorbance at 580 nm between 0.35 and 0.4 using a PLP Colorimeter. Strains were serially diluted in L-15 medium, and 1 ml of bacteria was added to each well at a multiplicity of infection (MOI) of 10 and incubated for 30 min at 37°C. To determine the exact starting inoculum at the time of infection, bacteria were further serially diluted and plated on LB agar. Bacteria were aspirated off the cell monolayer, and infected cells were gently washed three times with phosphate-buffered saline (PBS; autoclaved and filter sterilized; Life Technologies) and then incubated with L-15 medium containing 1 mg/ml kanamycin for a further 30 min at 37°C to kill extracellular bacteria. The supernatant was removed, and infected cells were then incubated with L-15 medium containing 250 μg/ml kanamycin for 12 h. At 0, 3, 6, 9, and 12 h postinfection, cell monolayers were lysed with 1 ml MilliQ water, serially diluted in 1 ml PBS, and plated onto LB agar for bacterial enumeration.

Determination of MIC and susceptibility to stress.

Broth microdilutions were tested against a variety of antibiotic classes as described previously (63), with the following modifications. Strains were incubated overnight in Mueller-Hinton broth (MHB) at 37°C. Overnight cultures were diluted 1:50 in fresh MHB and were incubated at 37°C for 1 h with agitation. Antibiotics were 2-fold serially diluted across a 96-well plate in MHB with a final volume of 100 μl, with an antibiotic range of 256 to 0.25 μg/ml. Following bacterial incubation, 100 μl of each strain was added to antibiotic-containing medium in a 96-well plate and incubated statically at 37°C for 24 h. The optical density of plates was read at 590 nm using a spectrophotometer (Bio-Rad Xmark). The MIC was defined as the minimum antibiotic concentration needed to keep overnight growth to under 20% of the unexposed bacterial growth control.

Motility assay.

Assessment of motility was performed as described previously (64). Briefly, B. pseudomallei strains were incubated at 37°C overnight with agitation. One microliter of overnight culture was stabbed into the middle of a 0.3% motility agar using a sterile inoculation loop, and plates were incubated upright for 24 h, upon which the distance of bacterial spread was measured.

Biofilm-forming capacity assay.

Biofilm assays were performed according to the methodology in reference 65 but with the following modifications. B. pseudomallei strains were incubated overnight at 37°C with agitation. The following day, 2% (vol/vol) of the overnight culture was inoculated into fresh medium and incubated for a further 24 h at 37°C with agitation. Overnight cultures (200 μl) were added to a 96-well plate and incubated for 3 h at 37°C to allow for adhesion. Supernatant was gently aspirated to avoid disturbing the adhered cells, and fresh Luria-Bertani broth was added and incubated at 37°C for a further 24 h. Supernatant was aspirated and biofilms were washed once with PBS, fresh Luria-Bertani broth medium was added, and the solution was incubated for a further 24 h. On the final day, supernatant was removed and biofilms were washed three times with PBS before being fixed with methanol and allowed to air dry. Cells were stained with 2% crystal violet, excess stain was removed with running double-distilled water, and plates were allowed to air dry. Dye bound to cells was solubilized with 33% glacial acetic acid, and the optical density at 590 nm was read on a spectrophotometer (Bio-Rad X-Mark).

BALB/c murine infection model.

Investigations involving animals were carried out according to the requirements of the United Kingdom Animal (Scientific Procedures) Act 1986 under project license PPL 30/3026. This project license was approved following an ethical review by Dstl’s Animal Welfare and Ethical Review Body. Studies were performed using female BALB/cAnNCrl mice (BALB/c; Charles River, UK) implanted with a subcutaneous Pico transponder (Uno BV, Netherlands) to allow individual mice to be tracked throughout the study. On arrival into containment level 3 animal facilities, mice were randomly allocated into cages of five animals and acclimatized to their new surroundings for 5 days before any procedures were performed. Animal husbandry practices and environmental conditions during the study were as described previously by Scott et al. (66). Challenges were performed with B. pseudomallei K96243 prepared as described previously by Scott et al. (66) and delivered via the intraperitoneal route. Mice received 1.1 × 10⁴ CFU B. pseudomallei K96243 and 1.86 10⁴ CFU of the BpsΔppiB strain, with 6 mice in each group. Mice were checked at least twice daily following challenge, and clinical signs for each mouse were recorded for 5 weeks postinfection. Humane end points were used throughout these studies to minimize suffering, with culls performed via cervical dislocation at the end point. At the end of the study, animals were culled and organs removed for enumeration of bacterial burden (lungs, liver, and spleen). These were homogenized through 40-μm sieves into PBS, serially diluted, and plated onto LB agar.

Complementation studies.

The open reading frame of BPSL2246 (ppiB) was amplified from genomic DNA of B. pseudomallei K96243 using the primers ppiB_For/ppiB_Rev (CTGCAGATGGTCGAACTGCATACG/CTGCAGGGACCACGACGGCCTTCT), and the resulting product was ligated into pJR3XFLAG, which incorporated a 3× FLAG tag at the C-terminal end of the gene. This was then amplified with the primers ppiB_pET_For/BamHI_stop_FLAG_Rev (CATATGGTCGAACTGCATACGAAC/GGATCCTTACTTGTCATCGTCATCCTTAT). The PCR product was inserted into the SmaI/BamHI restriction sites of pBBR1-MCS1. The complementation construct was transformed into E. coli ST18 and conjugated into the BpsΔppiB strain. Conjugates were selected on LB agar containing 30 μg/ml chloramphenicol. In experiments the BpsΔppiB/ppiB complemented mutant strain was grown in LB broth containing 30 μg/ml chloramphenicol.

Immunofluorescence.

J774A.1 macrophages (seeded at approximately 4 × 10⁵ cells/well) were grown overnight on 13-mm round coverslips in a 24-well plate at 37°C with 5% CO₂. Macrophages were infected with B. pseudomallei strains at an MOI of 10 as described above for the J774.A1 murine macrophage infection assay. At 12 h postinfection, monolayers were washed 3 times with PBS for 5 min each time and fixed with 100% methanol for 30 min and then washed 3 times with PBS for 5 min each time. Monolayers were stained at room temperature using the following protocol. Monolayers were blocked with 5% (vol/vol) fetal calf serum-PBS for 2 h to block nonspecific binding and then washed 3 times for 2 min each time. Cells were incubated with anti-B. pseudomallei lipopolysaccharide (LPS) at 1 μg/ml (1:100; Mab4VIHXII) for 1 h, after which they underwent three 2-min washes. Monolayers were incubated with secondary anti-mouse-whole IgG-fluorescein isothiocyanate (FITC) (1:64; Sigma-Aldrich) for 1 h, followed by three 2-min washes. Nuclei were stained using Hoechst (1:10,000; ThermoFisher Scientific) for 15 min, followed by two 2-min washes. Coverslips were mounted onto glass slides using Prolong Gold antifade reagent (Invitrogen). Fluorescence microscopy was performed using a Nikon Eclipse Ts2R microscope, and images were acquired using NIS-Elements software (Nikon).

MNGC.

Evaluation of MNCG formation was conducted using fluorescently stained cell monolayers as described above. Using previously published metrics (67), 1,000 nuclei per coverslip were counted, and the percentage of MNGC formation was calculated using the following equation: MNGC (%) = (no. of nuclei within multinucleated cells/total no. of nuclei counted) × 100.

Protein clean-up and in-solution digestion.

Cell preparations were solubilized in lysis buffer (4% SDS, 10 mM dithiothreitol, 100 mM Tris, pH 8.5) by boiling for 10 min, and the protein content was assessed by bicinchoninic acid protein assay according to the manufacturer’s instructions. One hundred micrograms of protein from each sample was acetone precipitated by mixing 4 volumes of ice-cold acetone with one volume of sample. Samples were precipitated overnight at –20°C and then spun down at 4,000 × g for 10 min at 4°C. The precipitated protein pellets were resuspended with 80% ice-cold acetone and precipitated for an additional 4 h at –20°C. Samples were spun down at 17,000 × g for 10 min at 4°C to collect precipitated protein, the supernatant was discarded, and excess acetone was driven off at 65°C for 5 min. Dried protein pellets were resuspended in 6 M urea, 2 M thiourea, 40 mM NH₄HCO₃ and reduced/alkylated prior to digestion with Lys-C (1/200, wt/wt) and then trypsin (1/50, wt/wt) overnight as previously described (68). Digested samples were acidified to a final concentration of 0.5% formic acid, desalted with homemade C₁₈ stage tips (69, 70), and eluted with buffer B (80% acetonitrile, 0.1% formic acid), and bound peptides were eluted with buffer B and then dried.

LFQ-based quantitative proteome liquid chromatography-mass spectrometry.

Purified peptides were resuspended in buffer A* (0.1% trifluoroacetic acid, 2% acetonitrile in MilliQ water) and separated using a two-column chromatography setup comprised of a PepMapC C₁₈ 20-mm by 75-μm trap and a PepMap C₁₈ 500-mm by 75-μm analytical column (ThermoFisher Scientific). Samples were concentrated onto the trap column at 5 μl/min for 5 min and infused into an Orbitrap Elite mass spectrometer (ThermoFisher Scientific) at 300 nl/min via the analytical column using a Dionex ultimate 3000 UPLC (ThermoFisher Scientific). Ninety-minute gradients were run, altering the buffer composition from 1% buffer B to 28% B over 60 min, from 28% B to 40% B over 10 min, and then from 40% B to 100% B over 2 min, after which the composition was held at 100% B for 3 min, dropped to 3% B over 5 min, and held at 3% B for another 10 min. The Elite mass spectrometer was operated in a data-dependent mode, automatically switching between the acquisition of a single Orbitrap MS scan (120,000 resolution) followed by 20 data-dependent collision-induced tandem mass spectrometry events (35 nominal collision energy), performed with 30-s dynamic exclusion enabled.

Mass spectrometry data analysis.

Identification of proteins was accomplished using MaxQuant (v1.5.3.1) (71). Searches were performed against the Burkholderia pseudomallei strain K96243 (UniProt proteome identifier [ID] UP000000605; downloaded 7 October 2018, 5,717 entries) proteomes with carbamidomethylation of cysteine set as a fixed modification. Searches were performed with trypsin cleavage specificity, allowing 2 miscleavage events and the variable modifications of oxidation of methionine and acetylation of protein N termini. The precursor mass tolerance was set to 20 parts per million (ppm) for the first search and 10 ppm for the main search, with a maximum false discovery rate (FDR) of 1.0% set for protein and peptide identifications. To enhance the identification of peptides between samples, the match between runs option was enabled with a precursor match window set to 2 min and an alignment window of 10 min. For label-free quantitation, the MaxLFQ option within Maxquant (72) was enabled in addition to the requantification module. The resulting protein group output was processed within the Perseus (v1.4.0.6) (73) analysis environment to remove reverse matches and common protein contaminates prior. Gene Ontology (GO) terms and associated annotation were downloaded from UniProt (UniProt proteome ID UP000000605). For LFQ comparisons, missing values were imputed using Perseus and Pearson correlations visualized using Perseus and R. Determination of significant changes was undertaken using a two-sample t test within Perseus, where proteins were considered significant if the mean difference between groups was greater than or less than a 1-fold change and satisfied a Benjamini-Hochberg multiple-hypothesis-corrected FDR of below 0.05, which corresponds to a −log₁₀(P value) of 1.73 or P value of 0.018.

Statistical analysis.

All numerical results were analyzed using Microsoft Excel 2010. Statistical analyses were performed using GraphPad Prism, version 8.0. For growth curves, motility, biofilm, MIC, intracellular infection, and cell cytotoxicity assays, a Mann-Whitney U test was used to determine the difference between strains. The log rank (Mantel-Cox) test was used for the animal studies. A Student's t test was used for MNGC formation.

Data availability.

The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE (74) partner repository with the data set identifier PXD012956.