A ubiquitous mobile genetic element changes the antagonistic weaponry of a human gut symbiont

Abstract

DNA transfer is ubiquitous in the human gut microbiota, especially among species of the order Bacteroidales. In silico analyses have revealed hundreds of mobile genetic elements shared between these species, yet little is known about the phenotypes they encode, their effects on fitness, or pleiotropic consequences for the recipient's genome. In this work, we show that acquisition of a ubiquitous integrative conjugative element (ICE) encoding a type VI secretion system (T6SS) shuts down the native T6SS of Bacteroides fragilis. Despite inactivating this T6SS, ICE acquisition increases the fitness of the B. fragilis transconjugant over its progenitor by arming it with the new T6SS. DNA transfer causes the strain to change allegiances so that it no longer targets ecosystem members with the same element yet is armed for communal defense.

Fig. 1.

Acquisition of a GA1 ICE shuts down firing of the B. fragilis GA3 T6SS and its ability to antagonize. (A) Gene maps of the GA3 T6SS loci of 13 B. fragilis strains. Genes are color coded as shown below. Strain names are colored the same if they have the same variable region 1 and 2 (V1, V2), which contain genes encoding effector and immunity proteins. These 13 strains contained 3 different V1 (1A-1C) and five different V2 regions (2D-2H). (B) Western immunoblot analysis of the presence of the GA3 T6SS needle protein in the cell fraction or the supernatant. The same color coding used in panel A is used for these strains. Strains designated with an * have a GA1 T6SS ICE in their genomes. (C) Transfer frequencies of the GA1 ICE from donor strain B. finegoldii CL09T03C10 to either B. fragilis 638R or 638RΔT6, deleted for genes necessary for firing the GA3 T6SS. The B. fragilis strains have the ermG gene inserted at a null site and the B. finegoldii GA1 ICE contains the tetQ gene inserted at a null site to allow for selection of transconjugants. Five different ratios of donor:recipient were assayed for transfer frequency (fig. S3). (D) Western immunoblot analysis of the cellular fraction and supernatants of 638R and 638R-GA1 transconjugants 5 and 6. BSAP-1 is a secreted protein of 638R used as a loading control. (E) Antagonism assays showing that the two 638R-GA1 transconjugants are defective in killing B. thetaiotaomicron VPI-5482 compared to the WT strain using cefoxitin selection plates. (F) Antagonism assays with the transconjugant strains as the target strains to analyze their ability to protect themselves from GA3 T6SS antagonism. 638RΔT6 is unable to antagonize and 638RΔV1ΔV2 (10) and 638R-GA1 ΔV1ΔV2 have both of the GA3 variable regions deleted and are unable to protect themselves. Following co-culture to allow antagonism, 638RΔV1ΔV2 was selected based on integration of the ermG gene into the chromosome, and 638R-GA1 transconjugants are selected based on tetracycline resistance encoded in the GA1 ICE. (G) Volcano plot showing the differentially expressed genes of 638R-GA1 transconjugant 6 compared to WT 638R. The red or green dots identify genes that are downregulated or upregulated, respectively, in both 638R-GA1 transconjugants 5 and 6 (transconjugant 5 volcano plot shown in fig. S4). Full RNASeq data from broth grown bacteria provided in Table S1. Fold change and adjusted p-values or FDRs were calculated by DESeq2 (v. 1.42.0) and/or EdgeR (v. 4.0.1).

Fig. 2.

Identification of the GA1 ICE genetic region responsible for quelling GA3 T6SS firing. (A) ORF map of the GA1 ICE of B. finegoldii CL09T03C10 showing the region of the B. fragilis chromosome where the ICE inserted in 638R-GA1 transconjugant 6. Genes of the GA1 T6SS are color coded as shown in fig. S2. Genes on the left side of the ICE involved in conjugation are colored green. tetQ added to the ICE to allow selection of transconjugants is highlighted gold. Large deletions made within this region in 638R-GA1 are indicated. The region comprising the GA1 T6SS locus is underlined. (B) Deletions made in the GA1 T6SS locus of strain 638R-GA1 analyzed in Fig 2E. (C) Western immunoblots of cell lysates and supernatants of WT and 638R-GA1 and large deletion mutants indicated in panel A, probed with antisera to the GA3 and GA1 Hcp proteins, or an antiserum to BSAP-1, a secreted protein of 638R used as a loading control. (D) Antagonism assay using the strains listed on the top as the antagonist and B. thetaiotaomicron VPI-5482 used as the target strain. The plates contain cefoxitin and are selective for B. thetaiotaomicron. (E) Western immunoblot performed in a manner similar to panel C except with the mutants shown in panel B. Corresponding lysis control western shown in fig S7C. (F) Schematic of the Bacteroidota T6SS apparatus adapted from Bongiovanni et al. (22) showing outer and inner membranes and localizations of T6SS proteins. Figure made with Biorender.com. (G) Western immunoblot of individual genes encoding components of the baseplate and transmembrane complex. Corresponding lysis control western shown in fig S7D. (H) Western immunoblot showing deletions of genes encoding the GA1 and GA3 T6SS transmembrane complex and their effects on firing of the GA1 T6SS. Corresponding lysis control western shown in fig S7E.

Fig. 3.

GA1 effects on the 638R GA3 T6SS in the mammalian gut. (A) Competitive colonization assay in gnotobiotic mice using three female (F) and three male (M) mice showing the percentage of the two strains in the initial inoculum and then in the feces at day 14 and day 28 post-gavage. (B) Western immunoblot analysis of day 14 fecal samples from mice from the competition experiment in panel A. (C) Western immunoblot analysis of fecal samples from three mice mono-colonized with either 638R or 638R-GA1 after 7 days of colonization. (D) Western immunoblot analysis of cells from broth-grown bacteria or from fecal samples from mice mono-colonized with three additional B. fragilis strains containing a GA3 T6SS and a GA1 ICE. (E) Volcano plot showing differentially expressed genes of 638R-GA1 compared to 638R from feces of mice monocolonized with each strain. Blue dots indicate genes of the GA3 T6SS. Red dots indicate other significantly downregulated genes and green dots indicate significantly upregulated genes (significance defined as at least 2.0-fold and adjusted p-value (DESeq2) and FDR (EdgeR) both were less than or equal to 0.05). (F) Transcript per kilobase million (TPM) values for the GA3 T6SS genes from feces of mice monocolonized with 638R or 638R-GA1.

Fig. 4.

Analyses of regulation and competition in vivo. (A) Western immunoblot analysis of Hcp synthesis in WT B. fragilis strains, GA3-GA1 B. fragilis strains, and tetR_GA1 deletion mutants from feces of mono-colonized mice. Two mice were colonized with each ΔtetR_GA1 mutant. Complementation of the tetR_GA1 for native GA3-GA1 containing B. fragilis strain 1284 is shown in fig S9A. (B) Volcano plot showing differential expression of genes of 638R-GA1ΔtetR_GA1 compared to 638R-GA1 from feces of mice monocolonized with each strain. Blue dots indicate genes of the GA3 T6SS locus. Brown dots indicate genes of the GA1 T6SS locus (significance defined as at least ≥2 -fold and adjusted p-value (DESeq2 v. 1.42.0) and FDR (EdgeR(v. 4.0.1) both were less than or equal to 0.05). (C) Volcano plot showing differential expression of genes of 638R compared to 638R-GA1ΔtetR_GA1 from feces of mice monocolonized with each strain. Fold change and adjusted p-values or FDRs were calculated by DESeq2 (v. 1.42.0) and/or EdgeR (v. 4.0.1). (D) Competitive colonization assay in gnotobiotic mice competing 638R versus 638R-GA1ΔtssBC_GA1 in three female (F) and three male (M) mice. The percentage of the two strains in the initial inoculum and then in the feces at day 14 and day 28 post-gavage is shown. (E, F) Competitive colonization assay in gnotobiotic mice competing 638RΔV1ΔV2 against either 638R-GA1Δ4ΔtetR_GA1 or 638R-GA1ΔRHSΔtetR_GA1 in three female (F) and three male (M) mice. The percentage of the two strains in the initial inoculum and then in the feces at day 14 post-gavage is shown. (G) Number of 9343-GA1 transconjugants arising over time in gnotobiotic mice colonized with equal numbers of 9343ermG and B. finegoldii (GA1-tetQ) in four mice. M1 and M2 were co-housed as were M3 and M4. The percentages of each inoculated strain at the end of the experiment are shown in fig S9B.