The Klebsiella pneumoniae citrate synthase gene, gltA, influences site specific fitness during infection

Abstract

Klebsiella pneumoniae (Kp), one of the most common causes of healthcare-associated infections, increases patient morbidity, mortality, and hospitalization costs. Kp must acquire nutrients from the host for successful infection; however, the host is able to prevent bacterial nutrient acquisition through multiple systems. This includes the innate immune protein lipocalin 2 (Lcn2), which prevents Kp iron acquisition. To identify novel Lcn2-dependent Kp factors that mediate evasion of nutritional immunity during lung infection, we undertook an InSeq study using a pool of >20,000 transposon mutants administered to Lcn2+/+ and Lcn2-/- mice. Comparing transposon mutant frequencies between mouse genotypes, we identified the Kp citrate synthase, GltA, as potentially interacting with Lcn2, and this novel finding was independently validated. Interestingly, in vitro studies suggest that this interaction is not direct. Given that GltA is involved in oxidative metabolism, we screened the ability of this mutant to use a variety of carbon and nitrogen sources. The results indicated that the gltA mutant has a distinct amino acid auxotrophy rendering it reliant upon glutamate family amino acids for growth. Deletion of Lcn2 from the host leads to increased amino acid levels in bronchioloalveolar lavage fluid, corresponding to increased fitness of the gltA mutant in vivo and ex vivo. Accordingly, addition of glutamate family amino acids to Lcn2+/+ bronchioloalveolar lavage fluid rescued growth of the gltA mutant. Using a variety of mouse models of infection, we show that GltA is an organ-specific fitness factor required for complete fitness in the spleen, liver, and gut, but dispensable in the bloodstream. Similar to bronchioloalveolar lavage fluid, addition of glutamate family amino acids to Lcn2+/+ organ lysates was sufficient to rescue the loss of gltA. Together, this study describes a critical role for GltA in Kp infection and provides unique insight into how metabolic flexibility impacts bacterial fitness during infection.

Fig 1. The Kp citrate synthase, gltA, interacts indirectly with Lcn2 during lung infection.

(A) A gltA (VK055_1802) mutant was constructed and used to validate InSeq findings. C57BL/6J mice or isogenic Lcn2^-/- mice were retropharyngeally inoculated with approximately 1×10⁶ CFU of a 1:1 mix of WT KPPR1 and KPPR1ΔgltA. Lung bacterial burden was measured after 24 hours, and log₁₀ competitive index of the mutant strain compared to the WT strain was calculated for each mouse strain (n = 10, mean displayed, *P < 0.05, ***P < 0.0005, ****P < 0.00005, one-sample t test or Student’s t test). (B) WT KPPR1 and various isogenic mutants were grown in RPMI + 10% (v/v) heat-inactivated resting human serum ± purified recombinant human Lcn2 overnight, then total CFU was enumerated by dilution plating on selective media (n = 3–4, mean displayed ± SEM, ****P < 0.00005, Student’s t test).

Fig 2. Deletion of gltA leads to diminished metabolic flexibility and distinct amino acid auxotrophy.

(A) Heatmap summarizing BioLog Phenotype Microarray analysis of WT KPPR1 and KPPR1ΔgltA growth in 282 carbon and nitrogen limited growth conditions indicates multiple conditions that sustained growth of WT KPPR1, but not KPPR1ΔgltA. (B) A subset of growth conditions summarizing glycolysis and non-essential amino acid biosynthesis that indicates a distinct amino acid auxotrophy is induced by deletion of gltA. Arrows from citric acid cycle intermediates indicate amino acids that utilize these intermediates for biosynthesis. (C) A subset of growth conditions summarizing dipeptide utilization that further indicates induction of a distinct amino acid auxotrophy by deletion of gltA (n = 3, mean displayed, *P < 0.05, Student’s t test). Underlined substrates support equivalent or enhanced growth of KPPR1ΔgltA relative to WT KPPR1.

Fig 3. Auxotrophy due to deletion of gltA is functionally complemented by glutamate and glutamate family amino acids.

(A) WT KPPR1, KPPR1ΔgltA, and KPPR1ΔgltApGltA were grown in M9 minimal media + 0.4% glucose with or without 10 mM glutamate (n = 3, mean displayed ± SEM). (B) AUC analysis of WT KPPR1, KPPR1ΔgltA, and KPPR1ΔgltApGltA growth in M9 minimal media +0.4% glucose with 10 mM glutamate (n = 3, ****P < 0.00005 compared to all other groups, Tukey’s multiple comparison test following ANOVA, mean displayed ± SEM). WT KPPR1 and KPPR1ΔgltA were grown in M9 minimal media + 0.4% glucose with 10 mM (C) glutamine, (D) proline, and (E) 2-α-ketoglutarate (n = 3, mean displayed ± SEM). “+AA” label indicates addition of amino acids to growth media at concentrations indicated in graph title. (F) AUC analysis of WT KPPR1 and KPPR1ΔgltA growth in M9 minimal media + 0.4% glucose + specific amino acid (n = 3, ***P < 0.0005, ****P < 0.00005, Tukey’s multiple comparison test following ANOVA, mean displayed ± SEM). Data presented in panels C-E were generated simultaneously, but graphed separately for ease of visualization.

Fig 4. Bronchoalveolar lavage fluid from Lcn2^-/- mice can sustain growth of KPPR1ΔgltA due to increased amino acid levels.

(A) Murine bronchoalveolar lavage fluid (BALF) was obtained from uninfected C57BL/6J mice or isogenic Lcn2^-/- mice, and WT KPPR1 and KPPR1ΔgltA were grown in BALF (representative curve displayed). (B) Area under curve (AUC) analysis was used to compare growth of WT KPPR1 and KPPR1ΔgltA in Lcn2^+/+ and Lcn2^-/- BALF. (n = 7–11 per group, paired t test, mean displayed ± SEM, **P < 0.005). (C) Heatmap of amino acid concentrations in BALF obtained from uninfected C57BL/6J mice or isogenic Lcn2^-/- mice subjected to metabolomic analysis (n = 6–7 mice per group). Blue histogram in inset indicates composite amino acid concentration values in heatmap matrix. (D) Murine bronchoalveolar lavage fluid obtained from uninfected C57BL/6J mice with or without amino acids was inoculated with a 1:1 mix of WT KPPR1 and KPPR1ΔgltA or WT KPPR1 and KPPR1ΔgltApGltA. Bacterial burden was measured after 24 hours, and log₁₀ competitive index of the mutant strain compared to the WT strain was calculated for each sample (n = 4 per group, **P < 0.005, ***P < 0.0005, ****P < 0.00005, one-sample t test or Tukey’s multiple comparison test following ANOVA).

Fig 5. gltA is dispensable for growth in mouse serum.

(A) WT KPPR1 and KPPR1ΔgltA were grown in M9 minimal media + 20% heat-inactivated murine serum without glucose. (B) AUC analysis of WT KPPR1 and KPPR1ΔgltA growth in M9 minimal media + 20% heat-inactivated murine serum without glucose (n = 3, Student’s t-test, mean displayed ± SEM). (C) WT KPPR1, KPPR1ΔgltA, and KPPR1ΔgltApGltA were grown M9 minimal media + 0.4% glucose with physiological levels of amino acids present in human serum (n = 3, mean displayed ± SEM). (D) AUC analysis of WT KPPR1, KPPR1ΔgltA, and KPPR1ΔgltApGltA growth M9 minimal media + 0.4% glucose with physiological levels of amino acids present in human serum (n = 3, ****P < 0.00005 compared to all other groups, Tukey’s multiple comparison test following ANOVA, mean displayed ± SEM). “+AA” label indicates addition of amino acids to growth media at concentrations indicated in graph title.

Fig 6. gltA influences site-specific fitness during bacteremia and oral infection.

(A) C57BL/6J mice or isogenic Lcn2^-/- mice were intraperitoneally inoculated with approximately 5×10⁵ CFU of a 1:1 mix of WT KPPR1 and KPPR1ΔgltA. Bacterial burden in the blood, spleen, liver, and lung was measured after 24 hours, and log₁₀ competitive index of the mutant strain compared to the WT strain was calculated for each mouse strain (n = 9–11 per group, mean displayed, *P < 0.05, ****P < 0.00005, one-sample t test or Student’s t test). (B) C57BL/6J mice or isogenic Lcn2^-/- mice were orally inoculated with approximately 5×10⁶ CFU of a 1:1 mix of WT KPPR1 and KPPR1ΔgltA. After 48 hours, mice were euthanized, cecum bacterial load was measured, and log₁₀ competitive index of the mutant strain compared to the WT strain was calculated for each mouse strain (n = 8–9 per group, mean displayed, *P < 0.05, Student’s t test).