O antigen biogenesis sensitises Escherichia coli K-12 to bile salts, providing a plausible explanation for its evolutionary loss

Abstract

Escherichia coli K-12 is a model organism for bacteriology and has served as a workhorse for molecular biology and biochemistry for over a century since its first isolation in 1922. However, Escherichia coli K-12 strains are phenotypically devoid of an O antigen (OAg) since early reports in the scientific literature. Recent studies have reported the presence of independent mutations that abolish OAg repeating-unit (RU) biogenesis in E. coli K-12 strains from the same original source, suggesting unknown evolutionary forces have selected for inactivation of OAg biogenesis during the early propagation of K-12. Here, we show for the first time that restoration of OAg in E. coli K-12 strain MG1655 synergistically sensitises bacteria to vancomycin with bile salts (VBS). Suppressor mutants surviving lethal doses of VBS primarily contained disruptions in OAg biogenesis. We present data supporting a model where the transient presence and accumulation of lipid-linked OAg intermediates in the periplasmic leaflet of the inner membrane interfere with peptidoglycan sacculus biosynthesis, causing growth defects that are synergistically enhanced by bile salts. Lastly, we demonstrate that continuous bile salt exposure of OAg-producing MG1655 in the laboratory, can recreate a scenario where OAg disruption is selected for as an evolutionary fitness benefit. Our work thus provides a plausible explanation for the long-held mystery of the selective pressure that may have led to the loss of OAg biogenesis in E. coli K-12; this opens new avenues for exploring long-standing questions on the intricate network coordinating the synthesis of different cell envelope components in Gram-negative bacteria.

Fig 1. The biogenesis of O16-LPS in E. coli K-12.

(A) Schematic representation of the early history of E. coli K-12 wild-type strains and derivatives, denoting relevant rfb mutations. Details of the procedures and media used during isolation and early propagation of strains are unknown (marked by question mark) but can be inferred by common practices at the time. (B) Biogenesis of O16-LPS in the original WT E. coli K-12 strain. The enzymatic order of glycosyltransferases WbbK and WbbJ is marked with question marks as it remains to be experimentally determined and the order presented is predicted from the inverted gene order within the operon. The structure of Lipid A-core is shown in green glow with corresponding glycosyltransferases and kinases. Genes that are responsible for the biosynthesis of C₅₅-PP-OAg^O16 and K-12 lipid A-core are shown in box. S-LPS, smooth LPS; R-LPS rough LPS. OM, outer membrane; PP periplasm; PG, peptidoglycan; IM, inner membrane; CP, cytoplasm, n refers to the modal chain length of polymerised OAg RU in MG1655-S. (C) Schematic representation of sequence alignment between regions rfb-50 from EMG2/MG1655 and rfb-51 from WG1. Grey shading marks areas of 100% sequence identity.

Fig 2. Restoration of O-antigen production sensitises E. coli K-12 to bile salts vancomycin.

Bacterial cultures of indicated strains grown in LB media were adjusted to OD₆₀₀ of 1 and spotted (4 μl) in 10-fold serial dilutions (10⁰ to 10⁻⁶) onto MacConkey agar (A) or LB agar (B & E) supplemented without or with 100 μg/ml vancomycin (V100), 0.1% (w/v) sodium deoxycholate (DOC), or both (DV100). prfb^sf, pJRD215 carrying rfb region from Shigella flexneri. The OAg RU structure produced in MG1655[prfb^sf] is shown. (C) Vancomycin minimum inhibitory concentration (MIC, μg/ml) for indicated E. coli strains in the absence (Nil) or presence of DOC. (D) Growth curves of indicated E. coli K-12 strains harbouring plasmids without or with wbbL cultured in LB media supplemented with DV100 and with or without 1 ng/ml anhydrotetracycline (AnTet). Arrow indicates lysis.

Fig 3

Genotypic and phenotypic traits of isolated MG1655-S suppressor mutants. Growth profiles of 51 MG1655-S suppressor mutants (BP1-51) in LB supplemented with 100 μg/ml vancomycin and 0.1% (w/v) sodium deoxycholate (DV) were classified as low (L) and similar to MG1655-S, medium (M) with growth recovery in between MG1655 and MG1655-S, or full (F) and similar to MG1655. The LPS silver staining profile of each mutant was compared to MG1655-S and MG1655 and classified as smooth-like LPS (S), rough-like LPS (R), or intermediately O antigen substituted LPS (I) with additional detailed alterations described in notes. The MIC values (μg/ml) of purified colicin E2 (ColE2) are recorded for each mutant. Suppressor mutants selected for whole genome sequencing are highlighted in yellow shading and their identified mutations are detailed by gene/coding sequence (CDS) name and non-synonymous substitution, early stop or insertion (ins). The mutation in BP49 was identified in wzy by targeted Sanger sequencing.

Fig 4. Accumulation of O antigen in the periplasm sensitises E. coli K-12 to bile salts.

Bacterial cultures of indicated strains grown in LB media were adjusted to OD₆₀₀ of 1 and spotted (4 μl) in 10-fold serial dilutions (10⁰ to 10⁻⁶) onto LB agar (A, D & E) without or supplemented with vancomycin at the concentrations of 100 μg/ml (V100) or 200 μg/ml (V200), 0.1% (v/w) sodium deoxycholate (DOC), a combination of V100 and DOC (DV100) or DOC and 200 μg/ml vancomycin (DV200). OAg RU structures of 2457T (S. flexneri 2a), EC958 (O25b), and CFT073 (O6) are shown. (B) Schematic representation of the synthesis and assembly process of O16 substituted LPS in different MG1655-S mutants. (C) Growth curves of indicated E. coli K-12 strains harbouring plasmids without or with wbbL cultured in LB media supplemented with 0.1%(w/v) DOC and without or with 1 ng/ml anhydrotetracycline (AnTet).

Fig 5. Continuous exposure of MG1655-S to bile salts promotes loss of O antigen.

(A) Bacterial cultures of MG1655-S grown in LB media supplemented without (Nil) or with 0.1% (w/v) deoxycholate (DOC) for 1 day, 7 days or 28 days were adjusted to OD₆₀₀ of 1, and spotted (4 μl) in 10-fold serial dilutions (10⁰ to 10⁻⁶) onto LB agar without or with 100 μg/ml vancomycin and 0.1% (w/v) sodium deoxycholate (DV100), 20 μg/ml crude colicin E2 preparation (ColE2), 200μg/ml vancomycin (V200), 200 μl of phage P1kc or T4GT7 lysate prepared from host MG1655 at the titre of above 10¹⁰ pfu. Data represent at least three independent repeats. (B) Silver staining of SDS-PAGE of LPS samples of MG1655, MG1655-S and its 28-day DOC-adapted isolates sensitive to colE2, P1kc and T4GT7.