Glyburide reduces bacterial dissemination in a mouse model of melioidosis

Abstract

Background: Burkholderia pseudomallei infection (melioidosis) is an important cause of community-acquired Gram-negative sepsis in Northeast Thailand, where it is associated with a ~40% mortality rate despite antimicrobial chemotherapy. We showed in a previous cohort study that patients taking glyburide ( = glibenclamide) prior to admission have lower mortality and attenuated inflammatory responses compared to patients not taking glyburide. We sought to define the mechanism underlying this observation in a murine model of melioidosis.

Methods: Mice (C57BL/6) with streptozocin-induced diabetes were inoculated with ~6 × 10(2) cfu B. pseudomallei intranasally, then treated with therapeutic ceftazidime (600 mg/kg intraperitoneally twice daily starting 24 h after inoculation) in order to mimic the clinical scenario. Glyburide (50 mg/kg) or vehicle was started 7 d before inoculation and continued until sacrifice. The minimum inhibitory concentration of glyburide for B. pseudomallei was determined by broth microdilution. We also examined the effect of glyburide on interleukin (IL) 1β by bone-marrow-derived macrophages (BMDM).

Results: Diabetic mice had increased susceptibility to melioidosis, with increased bacterial dissemination but no effect was seen of diabetes on inflammation compared to non-diabetic controls. Glyburide treatment did not affect glucose levels but was associated with reduced pulmonary cellular influx, reduced bacterial dissemination to both liver and spleen and reduced IL1β production when compared to untreated controls. Other cytokines were not different in glyburide-treated animals. There was no direct effect of glyburide on B. pseudomallei growth in vitro or in vivo. Glyburide directly reduced the secretion of IL1β by BMDMs in a dose-dependent fashion.

Conclusions: Diabetes increases the susceptibility to melioidosis. We further show, for the first time in any model of sepsis, that glyburide acts as an anti-inflammatory agent by reducing IL1β secretion accompanied by diminished cellular influx and reduced bacterial dissemination to distant organs. We found no evidence for a direct effect of glyburide on the bacterium.

Figure 1. Bacterial loads are higher in diabetes.

Note. BALF = bronchoalveolar lavage fluid; cfu = colony-forming units; IL = interleukin; STZ = streptozocin; TNFα = tumor necrosis factor-α. Sixty mice were injected with STZ 60 mg/kg daily for five days (marked by five vertical arrows on horizontal axis) and compared to 60 controls (treated with citrate buffer pH 4). Only 48 animals were eventually used for infection experiments. A. Glucose levels in streptozocin-treated animals plateaued 14 days after the first dose of streptozocin was given. The error bars in the graph are ranges. The horizontal interrupted line is the 16.7 mM cut off used to define diabetes in mice: streptozocin-treated animals treated were not used if their blood glucose concentrations fell below this level. The graph on the right shows glucose concentrations at time of sacrifice, and the error bars shown as standard deviations. B. Animals were inoculated with ∼6×10² cfu B. pseudomallei 4–5 weeks after streptozocin treatment. All mice were treated with ceftazidime starting 24 h after inoculation and continued until sacrifice at 48, 72 or 96 h (8 animals per group per time point). Bacterial loads were higher in lung, BALF and liver in diabetic animals. Error bars are 95% confidence intervals of the mean. C. When comparing diabetic animals with controls, no differences were seen in any of the cytokines measured.

Figure 2. Bacterial loads in blood, liver and spleen are lower in glyburide-treated mice.

Note. BALF = bronchoalveolar lavage fluid; cfu = colony-forming units. Mice were treated with glyburide or vehicle for seven days prior to inoculation with ∼6×10² cfu B. pseudomallei. All mice were treated with ceftazidime starting 24 h after inoculation and continued until sacrifice at 48, 72 or 96 h (eight animals per group per time point). Error bars are standard deviations. The horizontal interrupted line marks the limit of detection. A. There were no differences found in bacterial loads at the primary site of infection (lungs and BALF). B. Bacterial loads in glyburide-treated animals were lower in liver and spleen compared to untreated controls. In blood, there was a trend to bacterial loads being lower in glyburide-treated animals compared to untreated controls, but no p-value is reported because a large number of points were below the limit of detection (3 cfu/50 µl).

Figure 3. Diminished pulmonary leukocyte influx and IL1β production in glyburide-treated mice.

Note. BALF = bronchoalveolar lavage fluid; IL1β = interleukin-1beta. Mice were treated with glyburide or vehicle for seven days prior to inoculation with ∼6×10² cfu B. pseudomallei. All mice were treated with ceftazidime starting 24 h after inoculation until sacrifice (eight animals per group per time point). Error bars indicate standard deviations. A single p-value is reported unless there is evidence from a test of interaction that effects at each time point are different. A. No differences were found in wet weight of lung or in protein infiltration. B. Cellular infiltrate was reduced in glyburide-treated mice compared to untreated controls at the 48-hour time point and this difference was seen in both neutrophil and monocyte fractions. C. IL1β concentrations were lower in glyburide-treated mice compared to untreated controls at the 48-hour time point. This difference was greater in BALF than in lung tissue.

Figure 4. Glyburide does not influence production of IFNγ.

Note. IFNγ = gamma interferon. Mice were treated with glyburide or vehicle for seven days prior to intranasal inoculation with ∼6×10² B. pseudomallei. All mice were treated with ceftazidime starting 24 h after inoculation and continued until sacrifice (eight animals per group per time point). Error bars indicate standard deviations. A horizontal interrupted line marks the limit of detection for the assay. No influence of glyburide on IFNγ responses in both the pulmonary and systemic compartment was found.

Figure 5. Glyburide reduces macrophage IL1β secretion.

Note. Gb = glyburide; IL = interleukin. Primary bone marrow-derived macrophages were infected with B. pseudomallei for 30 min (multiplicity of infection 25 cfu/cell) and incubated with 0 µM (control), 100 µM or 1000 µM glyburide. IL1β was assayed in supernatant 24 hours later. The graph shows the results of three separate experiments, each linked by solid lines. The limit of detection for this assay was 55.2 pg/ml.