Burkholderia pseudomallei as an Enteric Pathogen: Identification of Virulence Factors Mediating Gastrointestinal Infection

Burkholderia pseudomallei is a Gram-negative bacterium and the causative agent of melioidosis. Despite advances in our understanding of the disease, B. pseudomallei poses a significant health risk, especially in regions of endemicity, where treatment requires prolonged antibiotic therapy. Even though the respiratory and percutaneous routes are well documented and considered the main ways to acquire the pathogen, the gastrointestinal tract is believed to be an underreported and underrecognized route of infection.

ABSTRACT

Burkholderia pseudomallei is a Gram-negative bacterium and the causative agent of melioidosis. Despite advances in our understanding of the disease, B. pseudomallei poses a significant health risk, especially in regions of endemicity, where treatment requires prolonged antibiotic therapy. Even though the respiratory and percutaneous routes are well documented and considered the main ways to acquire the pathogen, the gastrointestinal tract is believed to be an underreported and underrecognized route of infection. In the present study, we describe the development of in vitro and in vivo models to study B. pseudomallei gastrointestinal infection. Further, we report that the type 6 secretion system (T6SS) and type 1 fimbriae are important virulence factors required for gastrointestinal infection. Using a human intestinal epithelial cell line and mouse primary intestinal epithelial cells (IECs), we demonstrated that B. pseudomallei adheres, invades, and forms multinucleated giant cells, ultimately leading to cell toxicity. We demonstrated that mannose-sensitive type 1 fimbria is involved in the initial adherence of B. pseudomallei to IECs, although the impact on full virulence was limited. Finally, we also showed that B. pseudomallei requires a functional T6SS for full virulence, bacterial dissemination, and lethality in mice infected by the intragastric route. Overall, we showed that B. pseudomallei is an enteric pathogen and that type 1 fimbria is important for B. pseudomallei intestinal adherence, and we identify a new role for T6SS as a key virulence factor in gastrointestinal infection. These studies highlight the importance of gastrointestinal melioidosis as an understudied route of infection and open a new avenue for the pathogenesis of B. pseudomallei.

INTRODUCTION

Melioidosis is a life-threatening disease with a wide range of complications caused by the Gram-negative saprophytic bacterium Burkholderia pseudomallei (1). The current geographic endemicity of B. pseudomallei includes Southeast Asia and northern Australia, but recent studies have expanded this distribution to be worldwide, including regions of the tropics (1, 2). Human melioidosis is associated with a high mortality rate, which can reach up to 40% in regions of endemicity, and several reports suggest the disease to be underreported throughout the world (1, 2). Current estimates propose that the global disease burden is approximately 165,000 new cases each year, of which ∼89,000 are fatal (2). The intrinsic ability of B. pseudomallei to survive in diverse soil conditions of regions of endemicity has led to the prediction that its geographical distribution includes several countries and tropical regions (3, 4).

Melioidosis can develop after subcutaneous infection, ingestion, or inhalation of contaminated particles or aerosols. Although the percutaneous and inhalational routes of infection are well documented, the intragastric route of inoculation is underrecognized and understudied (1, 5). Reports of intragastric infection, though scarce, have been reported. In a recent study conducted in northeast Thailand, B. pseudomallei was isolated from 9/83 (11%) of stool samples and 9/58 (16%) rectal swabs from 141 patients with melioidosis (6). In a Malaysian study of patients with suspected Helicobacter pylori infection, of 215 gastric biopsy specimens, 15 (7%) contained B. pseudomallei cultures, of which 10 cases were also culture positive for H. pylori, and 5 contained only B. pseudomallei (7). Nonetheless, it is not known whether people with melioidosis develop persistent gut colonization and are capable of long-term shedding of B. pseudomallei in the stool (7). However, it is very likely that individuals that live in areas of endemicity can acquire the organism by the consumption of contaminated food or water.

Despite the predicted global distribution of B. pseudomallei and the potential to acquire melioidosis by the consumption of contaminated products, only two previous reports have tried to elucidate the pathogenesis of B. pseudomallei in the gastrointestinal (GI) tract. The first study suggested that esophageal delivery of B. pseudomallei to C57BL/6 and BALB/c mice did not result in susceptibility to infection, with a 50% lethal dose (LD₅₀) greater than 7 × 10⁸ CFU (8). Further, a later study reported the lethal dose of B. pseudomallei when delivered orally to BALB/c and C57BL/6 mice to be 1.04 × 10⁷ and 7.1 × 10⁶ CFU, respectively (9). These reports suggest that infection with B. pseudomallei by the oral route, though feasible, requires high bacterial doses to cause disease; however, it is evident that further studies are needed to understand the pathogenesis of B. pseudomallei in the GI tract.

Initial adherence to the intestinal mucosa is often one of the critical steps in establishing an infection for most enteric pathogens (10). Nonetheless, studies evaluating virulence factors involved in the adherence of B. pseudomallei to intestinal epithelial cells have not been described. Type IV pili have been shown to be important in B. pseudomallei adherence to lung epithelial cells and contribute to in vivo virulence (10). However, one of the most important mediators of enteric pathogen adherence to the intestinal mucosa is the type 1 fimbria (11). Type 1 fimbriae binds d-mannosylated sugar residues linked to cell receptors on the surface of the epithelium, helping with the establishment of a colonization niche and, in some cases, facilitating internalization (12).

Successful colonization and adherence to the epithelial surface is a required mechanism of invading bacteria to compete with the host microbiota and/or antagonize host defenses. Some Gram-negative bacteria utilize the type 6 secretion system (T6SS) to inject a myriad of effectors into host cells or competing bacteria, allowing the pathogen to thrive in a wide range of environments (13). The T6SS is a syringe-like structure that resembles a bacteriophage tail injection device. The tail of the needle is structurally composed of polymerized hemolysin coregulated (Hcp) proteins, while the cap end of the needle is composed of VgrG family proteins (14). B. pseudomallei codes for six genomic clusters of T6SS genes, also termed T6SS-1 through T6SS-6 (15, 16). Upon entry into the host and escape from the vacuole, B. pseudomallei utilizes the T6SS-1 for its intercellular dissemination in eukaryotic cells, including the induction of multinucleated giant cells (MNGCs) and, eventually, cell death (15, 16). To mediate various host cellular effects, the T6SS apparatus injects putative effectors with a wide range of functions into the host cell. For example, the VgrG2a of Pseudomonas aeruginosa is a protein that directly interacts with γ-tubulin ring complex (γ-TuRC), while VgrG1 protein from Vibrio cholerae and Aeromonas hydrophila contains an actin cross-linking protein, allowing for the rearrangement of the host cytoskeleton (17, 18). The C-terminal domain of the VgrG-5 of B. pseudomallei is required for MNGC formation and virulence in mice (19). However, no functional effectors have been associated with the pathogenesis of B. pseudomallei in the GI tract.

Therefore, the objective of this study was to first evaluate the ability of B. pseudomallei to adhere, invade, and survive within intestinal epithelial cells in vitro, to establish an in vivo model of gastrointestinal melioidosis, and to study the function of putative virulence factors during B. pseudomallei GI infection. Together, our results highlight the importance of the intragastric route as a portal of entry for B. pseudomallei infections and warrants the recognition of B. pseudomallei as a true enteric pathogen.

RESULTS

B. pseudomallei requires type 1 fimbria to adhere to IECs.

A key event early in the pathogenesis of most gastrointestinal pathogens is the adherence to the intestinal mucosa. It is well-known that many enteric pathogens use fimbrial appendages to start the colonization in the intestine (20). Further, intracellular pathogens utilize complex virulence systems for successful intracellular survival and intercellular movement, such as the T6SS (13). To test whether B. pseudomallei utilizes a functional fimbria to mediate adherence to intestinal epithelial cells (IECs), we evaluated several B. pseudomallei mutants with transposon insertions in two biosynthetic clusters encoding type 1 fimbria-associated proteins (Tn::2047 and Tn::2048) or, in the type 1 fimbriae major subunit, FimA (Tn::2049 and Tn::2050). At least three transposon mutants showed reduced adherence to hIECs, but only a single strain with an interruption of the fimA gene (Tn::2049) was associated with a statistically significant difference in adherence compared to B. pseudomallei WT (strain 1026b) (Fig. S1A and B in the supplemental material). To further test the functionality of FimA, we created an isogenic mutant (B. pseudomallei ΔfimA) in the parent B. pseudomallei WT strain K96243 and analyzed the role of FimA in plaque formation. In addition, to examine the role of the T6SS in mediating intercellular spread, a critical step in bacterial intracellular survival, an isogenic mutant in hcp1 (B. pseudomallei Δhcp1), a gene that encodes a critical component of the T6SS injectosome was constructed and used to evaluate the formation of plaques. Although complementation of each of these mutations would be ideal to restore the phenotypes, a potential gain-of-function experiment in a select agent is not permitted. A characteristic cytotoxic effect at late stages of infection is cell detachment, which can be visualized and quantified by the formation of plaques on cell monolayers. After 24 h of infection, we observed a decrease in plaque formation efficiency, although not a complete elimination was observed, like in the case of the T6SS as visualized by the formation of plaques in Giemsa-stained monolayers (Fig. 1A). However, we noticed a significant decrease in plaque formation efficiency in the absence of FimA (Fig. 1B). Next, we evaluated the role of FimA in adherence to human IECs (hIECs) and primary mouse IECs (mIECs). In the absence of FimA, we noticed a significant reduction in B. pseudomallei adherence in both human and mouse IECs compared to B. pseudomallei wild-type (WT) or B. pseudomallei Δhcp1 infections (Fig. 1C). To corroborate the involvement of type I fimbriae, a well-known mannose-sensitive adhesin, we pretreated hIEC or mIECs with d-mannose and confirmed that addition of this carbohydrate caused a reduction in the adherence of B. pseudomallei WT and B. pseudomallei Δhcp1 to levels comparable to the B. pseudomallei ΔfimA mutant (Fig. 1C). The adherence of B. pseudomallei to IECs was visualized and corroborated by the quantitative results. Using fluorescence microscopy, we observed fewer bacteria adhered to the surface of IECs in the absence of FimA than to B. pseudomallei WT or B. pseudomallei Δhcp1 (Fig. 1D) after 1 h postinfection. To further analyze the effect of FimA in a T6SS mutant background, we evaluated the adherence of the double mutant B. pseudomallei Δhcp1 ΔfimA on IECs and observed an equivalent reduction as B. pseudomallei ΔfimA (Fig. 1D). These results confirmed the functionality of type 1 fimbriae in the adherence of B. pseudomallei to IECs and suggested a role for fimbriae in the adherence of this intracellular pathogen to the intestinal mucosa. Interestingly, the complete ablation of plaques by B. pseudomallei Δhcp1, in the absence of reduced adherence onto IECs, warranted further investigation.

FIG 1.

Type 1 fimbriae, a mannose-sensitive adhesin is involved in the adherence of B. pseudomallei (Bpm) to IECs. (A and B) Caco-2 cells were infected with B. pseudomallei WT, B. pseudomallei Δhcp1, or B. pseudomallei ΔfimA for 24 h using an MOI of 10, washed, fixed, and then stained with Giemsa to visualize the formation of plaques. Plaque-forming efficiency was calculated as described in Materials and Methods. (C) Adhered bacteria to human or mouse IECs were quantified by CFU enumeration after 1 h. Prior to infection, IECs were incubated for 1 h in the presence or absence of 1% of d-mannose. *, P < 0.05 (comparison between hIECs). ^, P < 0.05, and ^̂, P < 0.01 (comparison between mIECs). (D) Fluorescence microscopy of adhered B. pseudomallei WT, B. pseudomallei ΔfimA, B. pseudomallei Δhcp1, or B. pseudomallei Δhcp1 ΔfimA on IECs for 1 h was visualized after staining for B. pseudomallei, actin, or DAPI. Error bars of the mean represent the average ± standard error of the mean of at least three experiments with three biological replicates. Statistical analysis was done using a Student's t test. Statistical differences between adherences were calculated by a two-way analysis of variance (ANOVA) followed by Tukeýs post hoc test. *, P < 0.05.

B. pseudomallei stimulates the formation of multinucleated giant cells and plaques on intestinal epithelial cells, a phenotype that requires a functional T6SS.

The ability of B. pseudomallei to infect epithelial cells is a well-recognized and principal route for bacterial entry (1). However, despite previous reports showing the ability of B. pseudomallei to cause lethality in mice infected intragastrically, no reports on the establishment of relevant models to study the biology or mechanisms of intestinal infection have been undertaken. Upon internalization, B. pseudomallei is able to move intercellularly by fusing host cells together and subsequently disseminating to distal organs, including the liver and spleen (1). To study the potential of B. pseudomallei to invade intestinal epithelial cells, we infected and subsequently visualized internalized bacteria in human-derived Caco-2 intestinal epithelial cells (hIECs) 24 h after infection (Fig. 2A). B. pseudomallei induced the formation of MNGCs by disruption of cellular membranes, leading to the formation of syncytia and the condensation of nuclei (Fig. 2A, center). In the absence of a functional T6SS, B. pseudomallei was unable to induce the condensation of cell nuclei or the formation of MNGCs in hIECs, and rather, B. pseudomallei remained confined to the intracellular cytoplasmic space (Fig. 2A, right). To visualize plaque formation, we infected hIECs or mIECs and quantified the efficiency of plaque formation 24 h after infection (Fig. 2B and C). The infection with B. pseudomallei WT resulted in the formation of plaques in monolayers of hIECs and mIECs (Fig. 2B) with equivalent efficiency in mediating this effect. In the absence of a functional T6SS, this phenotype is reduced in hIECs and, to a lesser extent, in mIECs (Fig. 2B and C). The mIECs were validated by the phenotypic expression of cytokeratin-18 and ZO-1, two important markers of epithelial cells (Fig. S2A). Together, these results showed the ability of B. pseudomallei to infect human and mouse IECs, leading to the formation of syncytia, plaques, and subsequent cell death, an in vitro phenotype dependent on a functional T6SS.

FIG 2.

The formation of MNGCs and plaques in monolayers of intestinal epithelial cells requires a functional B. pseudomallei T6SS. (A) Human Caco-2 and mouse (C57BL/6) primary small intestine epithelial cells were infected with B. pseudomallei K96243 (WT) or B. pseudomallei Δhcp1 mutant for 24 h at 37°C and 5% CO₂, using an MOI of 10. After infection, cells were washed and stained with Giemsa and examined. (B) Giemsa-stained plaques were visualized by bright field microscopy (×60 magnification). Magnified view from the dotted-outlined area (×6) shows MNGC and cell nuclei contraction in B. pseudomallei WT infection but not in a B. pseudomallei T6SS mutant. (C) Plaque formation efficiency was calculated as described in Materials and Methods. Error bars represent the average ± standard error of the mean of at least three experiments with three biological replicates. Statistical analysis was done using a two-way ANOVA followed by a Tukeýs post hoc test. ****, P < 0.0001.

B. pseudomallei can move from cell to cell and stimulate formation of MNGCs on IECs, a mechanism dependent on the T6SS.

To understand the intracellular lifestyle of B. pseudomallei as a gastrointestinal pathogen, we evaluated the ability of the bacteria to spread from cell to cell in hIECs and primary mIECs. Actin filaments and nuclei were stained using rhodamine-phalloidin and 4′,6-diamidino-2-phenylindole (DAPI), respectively, and bacterial cells were visualized using an antibody recognizing Burkholderia lipopolysaccharide (LPS). Starting at 12 h of infection, B. pseudomallei disrupted the cell cytoskeleton forming unipolar actin tails, allowing for intercellular movement (Fig. 3A and B). By 24 h after infection, this effect led to the formation of MNGCs (cell membranes denoted by dotted line) indicating fusion of cells; these effects were more pronounced in primary mIECs (Fig. 3B, arrows). To test the functionality of the T6SS on bacterial movement and the subsequent formation of MNGCs, we assessed the time course of infection using a T6SS-deficient strain. In the absence of the T6SS, the formation of MNGCs due to membrane fusion was not observed, and, in comparison to the B. pseudomallei WT infection, the integrity of the cell membrane in the monolayer remained intact (Fig. 3A and B). B. pseudomallei Δhcp1 remained trapped in the cytoplasm of individual cells, while the formation of actin tails was not affected (Fig. 3A and B, right panels). However, we observed a reduction in actin tail formation efficiency and length in both hIECs and mIECs in the absence of a functional T6SS (Fig. 3A and B and Fig. S2C). This effect may be due to a potential cross-communication mechanism between the T6SS and the BimA autotransporter, which is the protein associated with the induction of actin tails. These data further confirm that B. pseudomallei causes fusion of IECs, allowing the formation of MNGCs, an effect not seen in a T6SS-deficient strain, while the capacity of B. pseudomallei to form actin tails and move within the cell was minimally impacted.

FIG 3.

B. pseudomallei (Bpm) spreads intercellularly and causes formation of MNGC in IECs using a functional T6SS. Caco-2 (A) and mouse intestinal epithelial cells (mIECs) (B) were infected with B. pseudomallei WT or B. pseudomallei Δhcp1 for 12 and 24 h at 37°C and 5% CO₂, using an MOI of 10. After infection, cells were washed with PBS, fixed, and stained for immunofluorescence prior to microscopic examination (×60). Bacterial cells were stained with mouse anti-Burkholderia LPS polyclonal antibody, followed by an Alexa fluorophore 488 goat anti-mouse IgG secondary antibody. Actin was stained using rhodamine-phalloidin and cell nuclei with DAPI. Magnified view (×10) of dotted-outlined area is shown on the right side of each panel. Arrowheads indicate the presence of MNGC in B. pseudomallei WT-infected IECs. Images were processed and analyzed using ImageJ analysis software.

B. pseudomallei is efficient in IEC intracellular replication and induction of cell death, mechanisms involving a functional T6SS, but not in IEC internalization.

To quantitatively assess the internalization of B. pseudomallei and the involvement of the T6SS in intracellular survival, we measured the internalization and fold change of intracellular replication within IECs. The internalization efficiency of B. pseudomallei WT and B. pseudomallei Δhcp1 at 3 h postinfection is equivalent for both strains, with approximately 10% and 4% for hIECs and mIECs, respectively (Fig. 4A and B). Using hIECs and primary mIECs, we measured the fold change in intracellular replication after 6, 12, and 24 h of infection compared to an early time point of infection (3 h postinfection). Infection of hIECs with B. pseudomallei WT resulted in an increase of 100 times of intracellular replication at 12 h and a larger increment of approximately 600 times by 24 h, compared to 3 h postinfection (Fig. 4C). A similar pattern of intracellular replication was observed using primary mIECs, although in this case, B. pseudomallei was more efficient, with approximately 7,000-fold change in intracellular replication by 24 h compared with 3 h postinfection (Fig. 4D). In the absence of a functional T6SS, the relative fold change of intracellular replication was significantly lower by 24 h in both hIECs and mIECs (Fig. 4C and D). Based on our previous observation showing the formation of plaques and the disruption of cell membrane integrity, we assessed the ability of B. pseudomallei to induce cell cytotoxicity by measuring the release of lactate dehydrogenase (LDH), indicative of the disruption of cell membrane integrity and a release of intracellular LDH (21). Cell culture supernatants were taken after each infection time point, and these results showed that IECs infected with B. pseudomallei WT exhibited a higher percentage of cytotoxicity beginning with 12 h after infection and a more pronounced effect at 24 h postinfection (Fig. 4E and F). Cells infected with B. pseudomallei Δhcp1 showed a significant reduction in cytotoxicity of approximately 10% at 12 h after infection and a reduction of approximately 40% and 20% by 24 h in hIECs and primary mIECs, respectively (Fig. 4E and F). Compared to hIECs, primary mIECs appear to have a higher susceptibility to infection with B. pseudomallei Δhcp1 (Fig. 4E and F). Together, these results indicate that at 24 h postinfection, B. pseudomallei WT is able to replicate more efficiently than B. pseudomallei Δhcp1, an effect likely due to the ability of bacteria to induce cell death and multiplication in the extracellular milieu. Also, these results demonstrate the internalization efficiency of B. pseudomallei in IECs and indicate the dependence of a functional T6SS to cause cell toxicity.

FIG 4.

B. pseudomallei (Bpm) is highly efficient in IEC intracellular replication and in the induction of cytotoxicity, effects partially dependent on a functional T6SS, but not IEC internalization. (A and B) Percentage of internalized bacteria after 3 h postinfection. IECs were infected as described in Materials and Methods, and after 3 h, cells were washed and lysed. (C and D) Intracellular bacterial replication was quantified by CFU enumeration, and the fold change of intracellular replication was calculated from internalized bacteria 3 h postinfection. (E and F) Cell supernatants were collected after each time point and centrifuged, and LDH release was measured. The percentage of cell cytotoxicity was calculated following the manufacturer’s instructions. Error bars of the mean represent the average ± standard error of the mean of at least three experiments with three biological replicates. Statistical analysis was done using a two-way ANOVA followed by a Tukeýs post hoc test. Internalization statistical difference at 3 h postinfection was calculated by Student's t test. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

B. pseudomallei can infect and disseminate in acute and chronic mouse models of infection.

Given the remarkable ability of B. pseudomallei to adhere and invade mIECs and induce cell cytotoxicity, we used a mouse model to assess the virulence of B. pseudomallei when delivered by the intragastric route. The LD₅₀ of B. pseudomallei K96243 was calculated based on the Reed and Muench method and determined to be 1 LD₅₀ = 2.5 × 10⁶ CFU per mouse (22). To assess the virulence of B. pseudomallei after GI infection in an acute murine model, we infected animals with 2.5 LD₅₀ equivalents (∼6.25 × 10⁶ CFU/mouse) of B. pseudomallei WT via gavage and evaluated survival for 21 days (Fig. 5A). Three days after infection, animals infected with B. pseudomallei WT showed 30% survival, with only 3/10 animals surviving to the endpoint of the experiment (Fig. 5B). To assess the role of either the type 1 fimbriae or the T6SS during infection, we inoculated animals with 2.5 LD₅₀ equivalents of B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA, and assessed survival for up to days 21 postinfection. Among animals that were inoculated with B. pseudomallei Δhcp1 or B. pseudomallei Δhcp1 ΔfimA, the lethality was reduced at 21 days postinfection, with a significant difference in percent survival compared to B. pseudomallei WT (Fig. 5B). In contrast, the B. pseudomallei ΔfimA mutant displayed a similar lethality rate as the B. pseudomallei WT. We collected fecal samples for 3 days after infection and were not able to recover any shed bacteria (limit of detection of 10). To further determine whether B. pseudomallei can colonize the GI tract, we collected segments of the GI tract (stomach plus small intestine or cecum plus colon) of surviving animals to quantify bacterial loads (Fig. 5C). Animals infected with B. pseudomallei WT that survived to the endpoint of the experiment had no recoverable bacteria in any of the organs evaluated, with a limit of detection of 1 CFU (Fig. 5C). This suggests that surviving animals infected with B. pseudomallei WT had resolved the infection. However, animals that were inoculated with B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA carried at the 21-day endpoint, approximately 10¹ to 10² CFU/g of tissue in the colon and small intestine, respectively (Fig. 5C).

FIG 5.

Functional virulence factors are required for pathogenesis of B. pseudomallei (Bpm) in acute and chronic gastrointestinal mouse models of melioidosis infection. (A) Timeline of gastrointestinal infection. Animals (n = 10) were infected via gavage with a low (acute)- or high (chronic)-dose equivalents of 2.5 LD₅₀ (∼6.25 × 10⁶ CFU/mouse) or 1 LD₅₀ (2.5 × 10⁶ CFU/mouse), respectively, of B. pseudomallei WT, B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA. The oral gavage LD₅₀ of B. pseudomallei K96243 was calculated using the Reed and Muench method (22) and estimated to be ∼2.5 × 10⁶ CFU per mouse. (B and D) Survival curves of animals infected with each strain in either the acute or chronic infection models. Twenty-one days (acute) or 35 days (chronic) after infection, the stomach, small intestine, colon, cecum, liver, and spleen were collected and processed to quantify bacterial load by CFU. (C and E) Bacterial colonization in the acute and chronic studies was determined by CFU per gram of tissue. Error bars of the mean represent the average ± standard error of the mean of at least three experiments with three biological replicates Statistical analysis was done using a log-rank test. *, P < 0.05; **, P < 0.01.

Given the lethality of B. pseudomallei in our acute model of infection and the role of T6SS in the in vivo pathogenic process, we assessed whether B. pseudomallei disseminates to distal organs using a chronic melioidosis infection model (Fig. 5A). Animals were inoculated with a sublethal dose (∼1 LD₅₀) of either B. pseudomallei WT, B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA, and survival was assessed for 35 days (Fig. 5A). Animals that were inoculated with B. pseudomallei WT exhibited 40% (4/10) survival at 35 days postinfection (dpi), while 90% of animals that were inoculated with B. pseudomallei Δhcp1 or B. pseudomallei ΔfimA, and 100% (10/10) receiving B. pseudomallei Δhcp1 ΔfimA, survived to the end of the study (Fig. 5D). Bacterial load was evaluated by collecting individual sections of the GI tract (stomach, small intestine, cecum, and colon), as well as the liver and spleen of surviving animals, which were processed for CFU enumeration (Fig. 5E). Animals inoculated with B. pseudomallei WT exhibited persistent colonization of the stomach, small intestine, and colon, with the largest CFU numbers in the stomach (∼10² CFU/gram of tissue) (Fig. 5E). In the case of the single and double B. pseudomallei mutants, only the B. pseudomallei ΔfimA strain persisted and colonized the stomach, small intestine, and colon. No bacteria were recovered in the liver of any infected group, and only the B. pseudomallei WT-infected animals developed dissemination to the spleen (Fig. 5E). Overall, our data demonstrate the ability of B. pseudomallei to colonize the stomach and the intestine with subsequent dissemination to the spleen, a virulence phenotype mainly dependent on GI-expressed virulence factors.

DISCUSSION

Melioidosis infections are associated with a high mortality rate (1), and acquisition of the pathogen occurs through different routes of infection, including percutaneous inoculation, and inhalation, and more recently, oral ingestion has been reported in clinical cases (6). A previous report of B. pseudomallei infection by the intragastric route suggested the stomach as the primary site of B. pseudomallei gastrointestinal colonization, although no mechanistic studies were performed to understand the pathogenesis of B. pseudomallei following oral inoculation (9). These observations are consistent with our results showing the stomach with the highest level of B. pseudomallei colonization after oral inoculation.

The B. pseudomallei T6SS forms an injectosome that has been associated with different mechanisms of interbacterial competition as well as the delivery of mediators involved in host-pathogen interactions. The T6SS-1 of Burkholderia thailandensis and B. pseudomallei has been demonstrated to be involved in the formation of MNGCs and plaque formation in different in vitro cell models (15, 16). To study GI infection and the contribution of different virulence factors to infection, we optimized a model to study B. pseudomallei pathogenesis, allowing us to evaluate bacterial adhesion, invasion, and dissemination. Our results demonstrate that B. pseudomallei efficiently induces these hallmarks of pathogenesis in a type 1 fimbria- and T6SS-dependent manner. MNGCs are a result of cellular membrane fusion by mechanisms not entirely understood by either mechanical forces involving bacterial motility or bacterial effectors involved in membrane fusion (23). The formation of MNGCs leads to cellular death and detachment (24). It is interesting to note the differences observed in bacterial infectivity between mIECs and hIECs. As such, mIECs appear to be more susceptible to B. pseudomallei Δhcp1 infection, indicating that other virulence factors not associated with the T6SS and additional factors expressed in murine primary cells may be partially involved in MNGC and plaque formation. We also observed the condensation of cellular nuclei, also known as pyknosis, indicative of the induction of cell death (23). In contrast, these effects were not observed in a T6SS-deficient strain. The results suggest the importance of other T6SS-associated factors in mediating these processes, as it occurs in the absence of VgrG, the only other B. pseudomallei T6SS factor which is unable to mediate membrane fusion during intercellular spread and MNGC formation (19, 25). Given that Hcp functions as a T6SS-associated effector protein (26, 27), exogenous addition of the Hcp1 protein was unable to restore in vitro MNGC formation or plaques (Fig. S3 in the supplemental material). These results confirm that an intact T6SS is required to translocate the effector protein(s) that are needed to mediate cell-to-cell spread and MNGC formation. By utilizing a human-derived Caco-2 cell line, we were able to recapitulate in vitro certain aspects of the pathogenesis of B. pseudomallei. However, due to the transformed nature of these cellular models, they are often unable to express and localize receptors or have disrupted intracellular signaling (28–30). Considering these limitations, we decided to test our findings in primary intestinal epithelial cells obtained from C57BL/6 mice (the same genetic background as the mice used during challenges). Because our findings appeared to recapitulate the natural pathogenesis process of B. pseudomallei in vitro, we recommend that future studies of B. pseudomallei in IECs should use primary cells as a more biologically relevant infection model.

Intracellular pathogens exploit the host cell actin cytoskeleton for intercellular movement within the cytoplasm (31). This mechanism is crucial for intracellular enteric pathogens such as Salmonella, Listeria, and Shigella, allowing for their dissemination upon entry (31). In order for an intracellular bacterium to hijack the host cytoskeleton, the T3SS plays an important role in the subversion of host processes and polarly localized proteins as initiators of actin-based motility (32). These hallmarks of pathogenesis were recapitulated in our study with both IECs, obtaining comparable results. In the absence of a functional T6SS, we were still able to see the formation of actin tails, though smaller in length in the primary mIECs, suggesting a potential difference in the efficiency of actin tail formation in these cells, but the underlying molecular mechanism is yet to be defined. Interestingly, the size differences in actin tail length in primary mIECs between B. pseudomallei WT and B. pseudomallei Δhcp1 is also suggestive of an association between the T6SS and the virulence factors which work in concert for efficient actin tail formation. This observation has also been reported for Burkholderia mallei, a closely related pathogen with a high degree of genetic identity to B. pseudomallei (33). Consistent with this prediction, we were also able to see the formation of actin tails, MNGCs, and plaques by B. mallei in mIECs (Fig. S1B). Despite the inability to disrupt the formation of actin tails in the absence of a T6SS, we did not observe membrane fusion or MNGC formation, leaving the bacteria trapped in the cytoplasm. These findings further support that a functional T6SS is essential to mediate membrane fusion and subsequent intercellular bacterial spreading in the host (34).

Given the effective way B. pseudomallei multiplies intracellularly and evade host defense mechanisms in both phagocytic and nonphagocytic cells (15), it is surprising that the bacterial mechanisms required to colonize and invade intestinal epithelial cells, as well as the disruption of the cell membrane barrier to promote dissemination, are not understood. We have shown that B. pseudomallei enters hIECs and primary mIECs, with a greater efficiency observed in mIECs. The intracellular replication of B. pseudomallei and the induction of a cytopathic effect in both IECs is dependent on a functional T6SS, suggesting that this mechanism is important for bacterial survival and dissemination to other cells and tissues. The induction of cell cytopathology by other enteric bacteria has been shown to be a strategy for subsequent dissemination and establishment of a new niche (35). The detection of LDH released from infected IECs is a key characteristic of necrotic cells and is linked to the permeabilization of the cell plasma membrane (36). As observed here, in the absence of the T6SS, B. pseudomallei remains confined within the cell cytoplasm, and it is associated with a reduced cytopathic effect, suggesting that other factors involved earlier in the pathogenesis, like BopE (37), may be involved in mediating this effect.

Many gastrointestinal pathogens exploit mannose residues on mucosal epithelial cells for internalization using a variety of adhesin molecules, including type 1 fimbriae (12, 38, 39). Our results showed that although we did not completely block plaque formation in the absence of type 1 fimbriae, a delay in adherence to IECs was observed compared to the B. pseudomallei WT. The reduction in adherence in the presence of exogenous mannose supports the notion that B. pseudomallei carries functional mannose-dependent type 1 fimbriae. Our cumulative in vitro studies demonstrate that we have identified two virulence factors involved in the pathogenic process of B. pseudomallei upon contact with intestinal epithelial cells. Although we have not performed complementation experiments because B. pseudomallei is a select agent and, therefore, a potential gain of function experiment is not permitted, our studies are opening the door for future studies to define the potential mechanism required for gastrointestinal melioidosis.

Different surrogate small animal models have been developed to study various aspects of melioidosis while using different routes of infection, including inhalational and subcutaneous inoculation (40, 41). However, few studies have focused on the establishment of in vivo models that recapitulate either the acute or chronic (sublethal) stages of infection by the GI route. A previous report showed that an intragastric mouse model of melioidosis infection, using BALB/c and C57BL/6 mice, required a high infection dose of 7 × 10⁸ CFU/mouse for successful B. pseudomallei infection (8). In contrast, our studies demonstrate that the lethality of B. pseudomallei in the C57BL/6 mouse acute infection model was achieved with a lower dose and that bacterial dissemination to distal organs was observed using our sublethal model of infection. Another study in BALB/c mice using a sublethal dose (5 × 10⁵ CFU) of B. pseudomallei led to the establishment of persistent intragastric infection and concluded that B. pseudomallei was infectious in both the acute and chronic murine models, an effect dependent on the challenge dose (9). However, it is important to highlight that differences in bacterial strains, mouse genotypes, or mouse microbiomes may account for some of the differences observed. Our studies were consistent with previous reports showing no to low levels of bacterial shedding in the stools of acutely infected animals, suggesting that B. pseudomallei may be efficiently and quickly internalized into IECs and bacteria do not remain in the intestinal lumen (8, 9). In addition, we have reported that B. pseudomallei colonizes the GI tract during chronic infection, with the largest bacterial load recovered from the stomach, an interesting finding that warrants further studies to understand the potential mechanism of bacterial entry or survival in a nonpermissive environment for bacterial colonization. Further, our results are consistent with clinical studies that have found B. pseudomallei in human gastric biopsy specimens (42, 43). However, it is important to emphasize the remarkable ability of B. pseudomallei to survive in the acidic environment of the stomach, which led us to speculate that B. pseudomallei might share some features found in H. pylori, which is capable of producing urease, an enzyme which catalyzes the hydrolysis of urea to yield ammonia, thus elevating the pH of its environment (44). Another interesting observation of our in vivo murine model was the inability of B. pseudomallei to infect the spleen in the absence of a functional T6SS, strengthening its functionality in mediating intercellular bacterial movement and dissemination from the intestine to distal organs.

Our results also showed that the absence of FimA does not have a strong effect on the lethality of the mutant strain compared to WT infection at high infection dose, which suggests that type 1 fimbriae are not a critical virulence factor during this in vivo condition. This observation could be due to a dose-response effect, whereby a higher number of bacteria might overcome the absence of type 1 fimbria. Further, our in vitro observations reflect the attenuation of B. pseudomallei ΔfimA in the low-dose infection model. In addition, these results could also suggest that B. pseudomallei has a wide repertoire of factors that promote efficient colonization and internalization during acute infection. However, the difference in survival in the chronic infection model between the B. pseudomallei ΔfimA and the B. pseudomallei WT suggests that FimA, in a low infectious dose, is required for full lethal infection, and its absence delays the establishment of effective colonization along the GI tract. This effect may be more translatable to human infections from the ingestion of low-dose B. pseudomallei-contaminated sources.

The use of our acute and chronic murine melioidosis models, together with the previous report (9), should be used to define additional virulence factors required for gastrointestinal colonization, information that could be useful to expand our understanding of human gastrointestinal melioidosis infection. Taken together, we were able to highlight the fact that B. pseudomallei uses specific virulence factors to cause gastrointestinal infection and warrant the classification of B. pseudomallei as an enteric pathogen. In addition, the establishment of both the in vivo and in vitro gastrointestinal models of B. pseudomallei infection opens the door to further studies in identifying additional virulence factors, such as those T6SS-injected effectors associated with intracellular survival, MNGC formation, cell death, and/or bacterial dissemination (Fig. 6). Finally, we have highlighted the importance of gastrointestinal melioidosis as an understudied route of infection.

FIG 6.

Graphical model of B. pseudomallei gastrointestinal infection. A model depicting the key findings of B. pseudomallei gastrointestinal pathogenesis, both in vivo and in vitro. MNGC, multinucleated giant cell; SI, small intestine; T6SS, type VI secretion system.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

B. pseudomallei and B. mallei strains were routinely grown at 37°C in Luria-Bertani (LB) medium containing 1% NaCl, LB with 4% glycerol (LBG), or Ashdown’s selection medium with gentamicin (Sigma-Aldrich) specific for B. pseudomallei.

Transposon mutants.

The B. pseudomallei transposon insertion mutants in fimbriae biosynthetic genes used in this study were identified from a sequence-defined library of T24 transposon mutants in B. pseudomallei 1026b as previously described (45). All work with B. pseudomallei 1026b T24 mutants at Colorado State University was performed in a biosafety level 3 (BSL3) facility approved to work with tier 1 select agents. Transposon insertion mutants Tn::2047 (Bp1026b_I1572), Tn::2048 (Bp1026b_I1572), Tn::2049 (Bp1026b_I1759), and Tn::2050 (Bp1026b_I1759) were confirmed by PCR followed by sequencing. B. pseudomallei 1026b T24 mutants were maintained on media containing 300 μg/ml kanamycin (Gold Biotechnology, Saint Louis, MO).

Construction of mutants.

All isogenic mutants, including the B. pseudomallei Δhcp1 (gene BPSS1498), B. pseudomallei ΔfimA (gene BPSL1629), and B. pseudomallei ΔfimA Δhcp1 strains were constructed using a select agent-compliant suicide vector allelic exchange system as previously described (46). For the single mutants, the 500 bp upstream and downstream of hcp1 or fimA genes were cloned into plasmid pMo130 and introduced into the donor Escherichia coli S17-1 λpir strain. Five hundred microliters of E. coli donor strain were mixed with 500 μl of B. pseudomallei K96243, centrifuged, and resuspended in 100 μl of 10 mM MgSO₄ as a mix or individual strains for controls. Each conjugation was spotted (25 μl) on LBG agar and incubated for 8 to 16 h at 37°C. Selection of merodiploids was performed by resuspending the mixed reactions in Dulbecco’s phosphate-buffered saline (DPBS), and the diluted solutions were plated on LBG agar containing 500 μg/ml of kanamycin to select for integration and 30 μg/ml of polymyxin B to select against the donor E. coli strain. The plates were incubated for 48 to 96 h at 37°C. Merodiploids were screened by exposing the plates to 0.45 M pyrocatechol solution for selection of yellow colonies that integrated the plasmid. Counterselection was performed by subculturing merodiploids in 2× yeast extract-tryptone (2×YT) broth without salt for 4 h and then plating on 2×YT agar without salt and supplemented with 15% sucrose for 48 to 72 h. The resulting colonies were then screened with pyrocatechol, and nonyellow colonies were screened by PCR to confirm the mutation of hcp1 or fimA. Colonies positive by PCR were then sequenced to confirm the mutation. The double mutant strain, B. pseudomallei ΔfimA Δhcp1, was constructed by using B. pseudomallei strain CLH010 (Δhcp1) and introducing the fimA mutation using pMo130 plasmid as described previously (47, 48). The single Δhcp1 mutant was confirmed using primers described previously (48). The single and double ΔfimA mutations were then confirmed by PCR, followed by sequencing using the following primers: forward, 5′-CCG CCC TGC AGC GGA TCC CTC TAG ACA ACC GCT CGG CGA GAC C, and reverse, 5′-GAC AAG CCC GTC GCA TGC ATC TAG ACT GGC CGA GCA GCG TCA CGT TGA CCT GC.

Cell infection assays and immunofluorescence microscopy.

The human colorectal adenocarcinoma cell line (Caco-2; ATCC HTB-37) or C57BL/6 mouse primary small intestinal epithelial cells (catalog no. C57-6051, Cell Biologics Inc., Chicago, IL) were routinely grown in Dulbecco’s modified Eagle medium (DMEM) or complete primary cell culture medium following the manufacturer’s instructions (catalog no. M6621, Cell Biologics Inc.). DMEM medium was supplemented with 0.1 mM nonessential amino acid, 100 U penicillin/ml, 100 μg of streptomycin/ml (Gibco), and 10% fetal bovine serum (FBS) (Gibco). Cells were incubated at 37°C and 5% CO₂. For microscopic analysis, 5 × 10⁵ CFU were grown in 12-well cell-culture grade plates (Corning) or coverslips previously treated with type IV collagen from human placenta (BioReagent) and incubated overnight prior to treatment. Cells were infected with B. pseudomallei wild type (WT), B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA using a multiplicity of infection (MOI) of 10 for 1 h. After infection, cells were washed twice with PBS and fresh medium containing 1 mg/ml of kanamycin (Km) for an additional hour. For plaque formation analysis, upon extracellular bacterial killing for 1 h, a semisolid 0.25% agarose overlay containing 500 μg/ml of Km in the medium was added for 24 h. After infection, cells were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 30 min, stained with Giemsa stain (Gibco) for 30 min, and washed with PBS. For immunofluorescence microscopic analysis, infected human and mouse epithelial cells were fixed with 4% PFA and permeabilized with 0.1% Triton X-100 in PBS for 5 min prior to staining. Bacteria were detected with serum from lipopolysaccharide (LPS)-immunized mice (1:500), followed by a goat anti-mouse IgG, IgM (H+L) secondary antibody conjugated to Alexa 488. Polymerized actin and DNA were visualized using rhodamine isothiocyanate-phalloidin (Molecular Probes, Invitrogen, USA) or DAPI (Molecular Probes, Invitrogen), respectively. Cells were mounted using ProLong Gold antifade (Molecular Probes, Invitrogen) prior to visualization. Quantification of actin tail length was done by measuring the length of at least 200 tails from representative images at 24 h postinfection using the Olympus cellSens imaging software v2.3. Images were examined using an Olympus BX51 upright fluorescence microscope and analyzed using ImageJ software, National Institutes of Health (49).

Bacterial adherence, internalization, and intracellular replication assays.

Human and mouse intestinal epithelial cells were grown to confluence (5 × 10⁵) in a 12-well plate prior to infection as described above with some modifications. For adherence, prior to infection, bacteria were incubated in the presence or absence of 1% of d-mannose (Sigma) for 1 h at 37°C. Prior to infection, the inoculum was calculated based on optical density (OD) of the B. pseudomallei culture and plated to quantify the input inoculum. After 1 h of incubation, cells were infected with each strain for 1 h. After infection, cells were washed three times with PBS and lysed in the presence of 0.1% Triton X-100, and adhered bacteria were quantified by CFU enumeration. To measure the percentage of bacterial internalization, cells were infected for 1 h with each strain and then washed prior to replacing with fresh medium containing 1 mg/ml of Km for 1 h. After 3 h, internalized bacteria were quantified by CFU enumeration compared to the input inoculum. Fold change in intracellular replication was determined by CFU enumeration at 6, 12, and 24 h postinfection, in relation to 3 h postinfection after three washes with PBS and cell lysis with 0.1% Triton X-100. To assess cell cytotoxicity, supernatants were collected at each time point, spun at 5,000 × g for 5 min, and immediately used to assay for lactate dehydrogenase (LDH) release. The percentage of internalized bacteria was quantified according to the equation internalized bacteria at 3 h/input inoculum × 100. The fold change in intracellular replication was quantified according to the equation surviving bacteria at each time point/internalized bacteria at 3 h postinfection.

Lactate dehydrogenase release assay.

The supernatants of infected cells after each time point of infection were collected for quantification of LDH release following the manufacturer’s protocol (Sigma-Aldrich, St. Louis, MO). The percent cytotoxicity was calculated using the following formula: % cytotoxicity = 100 × (corrected reading from test well − corrected reading from untreated well)/(corrected maximum LDH control − corrected reading from untreated well). The corrected readings indicate the read from a test well minus the reading from the media-only control.

Ethics statement.

All manipulations of B. pseudomallei were conducted in CDC/USDA-approved and registered BSL3 facilities at the University of Texas Medical Branch (UTMB) and Colorado State University (CSU) in accordance with approved BSL3 standard operating practices. The animal studies at UTMB were carried out humanely in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals by the National Institutes of Health. The protocol (IACUC no. 0503014D) was approved by the Animal Care and Use Committee of UTMB.

Animal studies.

Female 6-to-8-week-old C57BL/6 mice (n = 10) were purchased from Jackson Laboratory (Bar Harbor, ME, USA) and maintained in an animal biosafety level 3 (ABSL3) facility. Animals were housed in microisolator cages under pathogen-free conditions with food and water available ad libitum and maintained on a 12-h light cycle.

Acute (high-dose) and chronic (low-dose) infection models.

Food was restricted 12 h before infection but was administered throughout the remainder of the study. For the acute model of infection, animals were infected with 2.5 LD₅₀ equivalents of B. pseudomallei WT (K96243), B. pseudomallei Δhcp1, B. pseudomallei ΔfimA, or B. pseudomallei Δhcp1 ΔfimA using a plastic oral gavage needle. Five fecal pellets were taken from each mouse every day, homogenized, centrifuged (5,000 × g for 5 min), plated on Ashdown B. pseudomallei-selective medium containing gentamicin, and incubated at 37°C. The LD₅₀ of B. pseudomallei K96243 was determined using an in-house calculation (1 LD₅₀ = 2.5 × 10⁶ CFU per mouse) calculated based on the Reed and Muench method (22). For the chronic infection model, mice (n = 10) were infected with ∼1 LD₅₀ equivalent as described above. At 21 or 35 days postinfection, mice were humanely euthanized, and the organs from surviving animals were collected and divided for determining bacterial loads. To enumerate bacterial loads, the gastrointestinal tract was divided into stomach and small intestine or cecum and colon in the acute model or all four separate compartments for the chronic model. Also, for the chronic study, the liver and spleen of surviving animals were collected. All organs were homogenized in 1 ml of PBS, serially diluted, and plated in either LB agar (spleen and liver) or Ashdown selective medium (GI tract) to quantify bacterial loads.