Bacterial defense systems exhibit synergistic anti-phage activity

Abstract

Bacterial defense against phage predation involves diverse defense systems acting individually and concurrently, yet their interactions remain poorly understood. We investigated >100 defense systems in 42,925 bacterial genomes and identified numerous instances of their non-random co-occurrence and negative association. For several pairs of defense systems significantly co-occurring in Escherichia coli strains, we demonstrate synergistic anti-phage activity. Notably, Zorya II synergizes with Druantia III and ietAS defense systems, while tmn exhibits synergy with co-occurring systems Gabija, Septu I, and PrrC. For Gabija, tmn co-opts the sensory switch ATPase domain, enhancing anti-phage activity. Some defense system pairs that are negatively associated in E. coli show synergy and significantly co-occur in other taxa, demonstrating that bacterial immune repertoires are largely shaped by selection for resistance against host-specific phages rather than negative epistasis. Collectively, these findings demonstrate compatibility and synergy between defense systems, allowing bacteria to adopt flexible strategies for phage defense.

Keywords: Druantia; Gabija; Kiwa; Septu; Zorya; co-occurrence; ietAS; phylogroups; prokaryotic immunity; tmn.

Figure 1

Distribution of defence systems across E. coli phylogroups. (A) A phylogenetic tree displaying 26,362 E. coli genomes obtained from the RefSeq database. Phylogroups are colour-coded according to the key. (B) Number of defence systems found per E. coli strain in each phylogroup. The mean number of defence systems is indicated by a red line. (C) Prevalence of defence systems in the E. coli genomes. The defence systems are organised from most prevalent (left) to least prevalent (right) and their total count is shown in the top bar graph. The bars are colour-coded according to the mechanism of the defence system. The remaining bar graphs show the prevalence, in percentage, of the defence systems per phylogroup.

Figure 2

Co-occurrence and negative association among defence systems in E. coli. (A) Graphical representation of the co-occurrence analysis, depicting one pair of defence systems that co-occur (Gabija and tmn), and one pair that is negatively associated (Zorya II and ietAS). The nodes of the E. coli phylogenetic tree are coloured according to the presence or absence of the defence system in each strain. Their location in chromosome, plasmid, or prophage regions is indicated in the middle. For visualization purposes, only leaves that carry at least one system from the pair are shown. (B) Co-occurrence of defence system pairs in E. coli. Co-occurring systems are shown in orange, and negatively associated systems are shown in green. The correlation between pairs was calculated with Pagel test for binary traits. Asterisks show correlations that were significant after the less stringent Benjamini-Hochberg correction (*) or the most stringent Bonferroni correction (**) for multiple testing. In the main text, the results after Bonferroni correction are considered. The defence systems are colour-coded according to their broadly defined mechanism of defence. (C,D) Distance histograms of (C) all and (D) co-occurring defence system pairs in 2,164 complete E. coli genomes. The median distance between genes encoding defence systems is shown by a red line. The analysis considered only those pairs that significantly co-occurred after Bonferroni correction.

Figure 3

Defence system pairs provide synergistic anti-phage activity. (A) Experimental set-up for the assessment of the anti-phage activity of individual defence systems and their combinations. YFP, yellow fluorescent protein. (B) Heatmap of synergy score of protection provided by selected defence system pairs against a panel of 29 phages. The synergy score is the epistatic coefficient for pairs of defence systems (see STAR Methods). Null, EOP equivalent to the defence provide by one system; Additive, EOP corresponds to the combined defence of the two individual systems; Synergy, EOP exceeds the collective defence of the two systems. Data is shown as the average of three biological replicates. ** Statistically significant (p < 0.01). (C) Time post infection assays, measuring T1 or T3 titres over the course of four hours in liquid cultures of E. coli containing individual or combined defence systems. Data is shown as the average and standard deviation of three biological replicates. (D) Bacterial growth under phage predation at different multiplicities of infection (MOIs), represented as area under the curve (AUC) in OD·h. A defence system pair acts synergetic when its dot (red) is above the expected additive effect (blue). Data is shown as the confidence interval of three biological replicates. The raw data and growth curves used to calculate the AUCs are available on the associated Github and Zenodo databases.

Figure 4.

Patterns of defence system co-occurrences across bacterial taxa. Heatmap of defence system co-occurrence patterns in E. coli (n = 26,362) and in four bacterial orders: Enterobacterales including E. coli (n = 9,124), Bacillales (n = 3,952), Burkholderiales (n = 2,199), and Pseudomonales (n = 1,288). Grey squares indicate that at least one system in the pair is not present in the taxonomic group. * Co-occurrence significant after Benjamini-Hochberg correction; ** Co-occurrence significant after Bonferroni correction.

Figure 5

Synergistic defence system pairs provide an evolutionary advantage to bacteria. (A) Set-up of experimental evolution assay of defence systems using chromogenic plasmids (left). Cells containing a single defence system carry a second plasmid that expresses a chromogenic reporter. Cells with defence system combinations were mixed in equal proportions with cells with single systems, and infected with phage. After 1, 2, and 3 days, cells were platted and the colonies of different colours were enumerated. A surface receptor assay (right) assessed the influence of receptor mutants on the outcome of the evolution assays, by subjecting colonies of different colours to phage infection in plaque assays. (B) Prevalence of receptor mutants in the bacterial population during the evolution assay. The proportion of colonies with the wild-type receptor is represented as positive in the vertical axis, while the proportion of colonies with mutated receptor is shown as negative. Colonies with mutated receptor were identified by their complete resistance to phage infection in spot assays. (C) Percentage of colonies from the evolution assay that carry each individual defence system or their combinations at 1, 2 and 3 days post infection with phage at low or high multiplicity of infection (MOI), compared to a non-infected control. Data is shown as the average of three biological and three technical replicates with individual counts shown. * p value < 0.01, ** p value < 0.001 determined by multiple comparison for nested one-way ANOVA, comparing to the corresponding uninfected control.

Figure 6

Mechanistic insight into the synergistic interaction between tmn and Gabija or Septu I. (A) Whole-plasmid alignment of 104 plasmids containing tmn and Gabija from complete E. coli genomes with plasmid CP083423.1 as a reference (see STAR Methods). The histogram shows the percentage of plasmids where the corresponding block from CP083423.1 was found. Annotated genes are coloured by function. (B) Efficiency of plating (EOP) of phages T1 and 670 on cells expressing Gabija (G), tmn (T), Gabija and tmn (GT), and alanine mutants of specific functional domains. The mutations are organised by functional domains of Gabija (ATPase, TOPRIM, and UvrD-like) and tmn (P-loop NTPase). Unfilled circles indicate instances where it was not possible to determine the number of phage plaques, hence a value of 1 was assumed at the corresponding dilution. Asterisk (*) indicates cases of synergy. (C) EOP of phages T1 and 670 on cells expressing tmn (T), Gabija (G), and tmn with either GajA or GajB. Unfilled circles indicate instances where it was not possible to determine the number of phage plaques, hence a value of 1 was assumed at the corresponding dilution. (D) EOP of phages T1 and phi113 on cells expressing tmn (T), Septu I (S), PrrC (P), Septu I and tmn (ST), or PrrC and tmn (PT). Unfilled circles indicate instances where it was not possible to determine the number of phage plaques, hence a value of 1 was assumed at the corresponding dilution. Asterisk (*) indicates cases of synergy. (E) EOP of phages T1 and 670 on cells expressing tmn (T), Septu (S), and Septu and tmn (ST), and variants with point mutations in specific functional domains. The mutations are organised by functional domains of Septu I (ATPase and HNHc) and tmn (P-loop NTPase). Unfilled circles indicate instances where it was not possible to determine the number of phage plaques, hence a value of 1 was assumed at the corresponding dilution. Asterisk (*) indicates cases of synergy.