Reverse transcriptase inhibitors prevent liver abscess formation during Escherichia coli bloodstream infection

Abstract

The presence of bacteria in the bloodstream is associated with severe clinical outcomes. In mice, intravenous inoculation of Escherichia coli can lead to the formation of macroscopic abscesses in the liver. Abscesses are regions of severe necrosis and consist of millions of bacteria surrounded by inflammatory immune cells. Liver abscess susceptibility varies widely across strains of mice, but the host factors governing this variation are unknown. Here, we profiled hepatic transcriptomes in mice with varying susceptibility to liver abscess formation. We found that transcripts from endogenous retroviruses (ERVs) are robustly induced in the liver by E. coli infection and ERV expression positively correlates with the frequency of abscess formation. Hypothesizing that ERV-encoded reverse transcriptase may generate cytoplasmic DNA and heighten inflammatory responses, we tested whether nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) influence abscess formation. Strikingly, a single NRTI dose administered immediately following E. coli inoculation prevented abscess formation, leading to a concomitant 100,000-fold reduction in bacterial burden. We provide evidence that NRTIs inhibit abscess formation by preventing the tissue necrosis that facilitates bacterial replication. Together, our findings suggest that endogenous reverse transcriptases drive inflammatory responses during bacterial bloodstream infection to drive abscess formation. The high efficacy of NRTIs in preventing abscess formation suggests that the consequences of reverse transcription on inflammation should be further examined, particularly in infectious diseases where inflammation drives negative clinical outcomes, such as sepsis.

Keywords: bloodstream infection; endogenous retroviruses; liver abscess; reverse transcriptase inhibitors.

Fig. 1.

ERV expression correlates with abscess susceptibility. (A) Schematic of variation in abscess susceptibility in BALB/cJ, B6J, and B6N female mice. (B) Schematic of RNA-Sequencing experiment. (C) Z-score normalized expression of transcripts that are induced by infection in B6N mice and whose expression correlates with abscess susceptibility. The complete dataset, including non-ERV genes, is in Datasets S1 and S2. Coordinates are from GRCm38. (D) Maximum likelihood tree of abscess-associated and representative ERV loci in B6J. Loci identified in this study are in bold and those identical to previously known loci are in parentheses [e.g., Xmv13 (Chr13: 67.9 Mb)]. Four loci lack env and do not match previously known ERVs (3). For clarity, one xenotropic and several polytropic MLVs are represented as single, collapsed branches (triangles). (E) Sequence alignment of the 11 abscess-associated MLV loci compared to their consensus.

Fig. 2.

NRTI treatment prevents abscess formation. (A) B6N females were inoculated IV with E. coli and NRTIs or vehicle control was injected IP at indicated times (Bottom table) and doses (Top table). CFUs are shown per liver, along with whether the animal possessed (blue) or lacked (black) abscesses, as well as the fraction of animals with abscesses (red). P values are derived from one-tailed Mann–Whitney U tests (mw, to compare CFUs) or Fisher Exact tests (fe, to compare the fraction of animals that developed abscesses). *historical data from ref. for reference. (B) Growth curves of E. coli exposed to NRTIs in LB (n = 4, mean ± SD). (C) Effect of NRTI treatment (all four drugs IP, one dose immediately following inoculation) on the formation of necrotic lesions caused by formalin-killed E. coli (Middle and Bottom rows, blue arrows). Abscesses formed from inoculation with live E. coli are shown for reference (Top row, blue arrows).