Klebsiella pneumoniae causes bacteremia using factors that mediate tissue-specific fitness and resistance to oxidative stress

Abstract

Gram-negative bacteremia is a major cause of global morbidity involving three phases of pathogenesis: initial site infection, dissemination, and survival in the blood and filtering organs. Klebsiella pneumoniae is a leading cause of bacteremia and pneumonia is often the initial infection. In the lung, K. pneumoniae relies on many factors like capsular polysaccharide and branched chain amino acid biosynthesis for virulence and fitness. However, mechanisms directly enabling bloodstream fitness are unclear. Here, we performed transposon insertion sequencing (TnSeq) in a tail-vein injection model of bacteremia and identified 58 K. pneumoniae bloodstream fitness genes. These factors are diverse and represent a variety of cellular processes. In vivo validation revealed tissue-specific mechanisms by which distinct factors support bacteremia. ArnD, involved in Lipid A modification, was required across blood filtering organs and supported resistance to soluble splenic factors. The purine biosynthesis enzyme PurD supported liver fitness in vivo and was required for replication in serum. PdxA, a member of the endogenous vitamin B6 biosynthesis pathway, optimized replication in serum and lung fitness. The stringent response regulator SspA was required for splenic fitness yet was dispensable in the liver. In a bacteremic pneumonia model that incorporates initial site infection and dissemination, splenic fitness defects were enhanced. ArnD, PurD, DsbA, SspA, and PdxA increased fitness across bacteremia phases and each demonstrated unique fitness dynamics within compartments in this model. SspA and PdxA enhanced K. pnuemoniae resistance to oxidative stress. SspA, but not PdxA, specifically resists oxidative stress produced by NADPH oxidase Nox2 in the lung, spleen, and liver, as it was a fitness factor in wild-type but not Nox2-deficient (Cybb-/-) mice. These results identify site-specific fitness factors that act during the progression of Gram-negative bacteremia. Defining K. pneumoniae fitness strategies across bacteremia phases could illuminate therapeutic targets that prevent infection and sepsis.

Fig 1. Transposon insertion-site sequencing (TnSeq) reveals that K. pneumoniae bacteremia is enhanced by a set of genes representing diverse fitness mechanisms.

(A) Overview of K. pneumoniae bacteremia TnSeq. A KPPR1 transposon library was divided into four pools containing 8,500 unique insertions and administered to mice via the tail vein at a 1x10⁶ CFU dose. After 24 hours, splenic CFU was recovered, and DNA was sequenced. The TnSeqDiff pipeline determined genes influencing fitness. This illustration was created with BioRender.com. (B) The input and splenic CFU burden for each pool and mouse represented in TnSeq, with mean organ CFU colonization represented by the bar. (C) Genes represented in TnSeq (~3,800 genes) that were defined as influencing infection (132 genes), and primary KEGG orthology for genes increasing K. pneumoniae fitness during bacteremia (58 genes). (D) A volcano plot displaying genes represented in the TnSeq screen with log₂ fold change (x-axis) and p-value (y-axis) defined by the TnSeqDiff pipeline. The genes selected for further study, arnD, purD, dsbA, sspA, and pdxA, are highlighted.

Fig 2. K. pneumoniae bacteremia fitness factors directly relay tissue-specific fitness advantages.

Five factors indicated by TnSeq as significantly enhancing bacteremia were selected for in vivo validation using the tail vein injection model. The 1:1 inoculum consisted of KPPR1 and transposon mutants for (A) arnD, (B) purD, (C) dsbA, (D) sspA, or (E) pdxA. Competitions were also performed using strains carrying the empty pACYC vector (_ev) or complementation provided on pACYC184 under native promoter control for (F) arnD (arnD+pACYC_arnD), (G) dsbA (dsbA+pACYC_dsbA), or (H) sspA (sspA+pACYC_sspA). The log₁₀ competitive index at 24 hours post infection is displayed for individual mice with bars representing the median and interquartile range. For all, *p<0.05, **p<0.01, ***p<0.001 by one sample t test with a hypothetical value of zero; for (F-H) ^#p<0.05, ^##p<0.01 by unpaired t test. For each group, n≥8 mice in at least two independent trials.

Fig 3. K. pneumoniae serum replication requires purine synthesis and is maximized by endogenous vitamin B6 biosynthesis.

K. pneumoniae strains were grown in M9 salts supplemented with (A) 10% mouse serum or 20% human serum that was either (B) active or (C) heat inactivated. K. pneumoniae strains carrying the empty vector pACYC (_ev) or pACYC expressing (D,E) purD (purD+pACYC_purD) or (F) pdxA (ΔpdxA+pACYC_pdxA) were grown in 10% mouse serum (D, F). Chemical complementation for purD was measured by supplementation of 1mM purines prior to growth (E). For all, the OD₆₀₀ was measured every 15 minutes for 12 hours. Differences in growth compared to KPPR1 or KPPR1_ev were detected by area under the curve using a one-way ANOVA with Dunnett’s multiple comparison for each strain compared to wild-type; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. For each group, n = 3 in independent trials.

Fig 4. K. pneumoniae bacteremia factors convey fitness advantages through tissue-specific interactions within blood-filtering organs.

Ex vivo competitions were performed using uninfected murine spleen or liver homogenate with a 1:1 mixture of KPPR1 and transposon mutants for (A) arnD, (B) purD, (C) dsbA, (D) sspA, (E) or pdxA. The log₁₀ competitive index compared to wild-type KPPR1 at 3 hours post incubation is displayed for individual mice with bars representing the median and interquartile range. *p<0.05, **p<0.01, ****p<0.0001 by one-sample t test with a hypothetical value of zero and n = 3–7 with points representing mice in independent experiments.

Fig 5. Primary site initiation of bacteremia enhances resolution of splenic fitness defects and illuminates requirements of factors across phases of pathogenesis.

To model bacteremic pneumonia, mice were infected with 1x10⁶ CFU K. pneumoniae. Competitive infections were performed with a 1:1 mixture of KPPR1 and a transposon mutant for (A) arnD, (B) purD, (C) dsbA, (D) sspA, or (E) pdxA or (F) a pdxA knockout (ΔpdxA). (G) Competitions were also performed with strains carrying the empty pACYC vector (_ev) or complementation provided on pACYC for pdxA only (pACYC_pdxA) or pdxA and downstream members of the operon (pACYC_operon). The log₁₀ competitive index at 24 hours post infection is displayed for individual mice with bars representing the median and interquartile range. For all, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by one-sample t test with a hypothetical value of zero; ^#p<0.05, ^##p<0.01, ^###p<0.001, ^####p<0.0001 by unpaired t test. For each group, n≥10 mice in at least two independent infections.

Fig 6. K. pneumoniae splenic fitness is maximized by factors relaying resistance to oxidative stress.

(A) Resistance to oxidative stress was measured by incubating K. pneumoniae strains with H₂O₂. Complementation was performed by comparing strains carrying empty pACYC (_ev) to those with pACYC expression of (B) sspA (sspA+pACYC_sspA) or (C) pdxA (pdxA+pACYC_pdxA). (D-H) In a model of bacteremic pneumonia, mice were infected with 1x10⁶ CFU K. pneumoniae containing a 1:1 mix of KPPR1 and a transposon mutant for (D) dsbA, (E) sspA, or (F) pdxA in Ccr2^-/- mice and (G) sspA or (H) ΔpdxA in Cybb^-/- mice. For (D-H), the log₁₀ competitive index at 24 hours post infection is displayed for individual mice with bars representing the median and interquartile range. In (A-C), *p<0.05, **p<0.01 by one-way ANOVA with Dunnett’s correction for each strain compared to wild-type, n = 4 in independent trials. Bars represent the mean percent survival for each group. In B-C, lines connect samples from the same replicate. In (D-H), *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by one-sample t test with a hypothetical value of zero and ^##p<0.01, ^###p<0.001 by unpaired t test. For each group, n≥7 mice in two independent infections.