Bacterial retrons encode phage-defending tripartite toxin-antitoxin systems

Abstract

Retrons are prokaryotic genetic retroelements encoding a reverse transcriptase that produces multi-copy single-stranded DNA¹ (msDNA). Despite decades of research on the biosynthesis of msDNA², the function and physiological roles of retrons have remained unknown. Here we show that Retron-Sen2 of Salmonella enterica serovar Typhimurium encodes an accessory toxin protein, STM14_4640, which we renamed as RcaT. RcaT is neutralized by the reverse transcriptase-msDNA antitoxin complex, and becomes active upon perturbation of msDNA biosynthesis. The reverse transcriptase is required for binding to RcaT, and the msDNA is required for the antitoxin activity. The highly prevalent RcaT-containing retron family constitutes a new type of tripartite DNA-containing toxin-antitoxin system. To understand the physiological roles of such toxin-antitoxin systems, we developed toxin activation-inhibition conjugation (TAC-TIC), a high-throughput reverse genetics approach that identifies the molecular triggers and blockers of toxin-antitoxin systems. By applying TAC-TIC to Retron-Sen2, we identified multiple trigger and blocker proteins of phage origin. We demonstrate that phage-related triggers directly modify the msDNA, thereby activating RcaT and inhibiting bacterial growth. By contrast, prophage proteins circumvent retrons by directly blocking RcaT. Consistently, retron toxin-antitoxin systems act as abortive infection anti-phage defence systems, in line with recent reports^3,4. Thus, RcaT retrons are tripartite DNA-regulated toxin-antitoxin systems, which use the reverse transcriptase-msDNA complex both as an antitoxin and as a sensor of phage protein activities.

Extended Data Figure 1|. Retron phenotypes are suppressed by inactivating the accessory retron gene, which encodes a cytoplasmic toxin RcaT.

a, ΔrrtT and ΔxseA STm are cold-sensitive. 1536 colony-arrays of the STm gene-deletion library were pinned onto LB plates, and strains were grown at room temperature. Colony sizes were used to calculate a fitness S-score for each strain. S-scores were calculated from n=8 biological. Dashed vertical line denotes the mean S-score of all strains (n=3781); negative or positive S-scores indicate sensitive or resistant mutants. b, Perturbing msDNA biogenesis leads to cold-sensitivity. STm wildtype and retron-deletion strains were serially diluted and spotted on LB plates as in Fig. 1b. Plates were incubated at indicated temperatures (n=4 biological). c, Retron mutants grow less in anaerobic conditions. Growth curves of STm wildtype and retron-deletion strains were obtained by measuring OD578 in microtiter plates, under anaerobic conditions at 37°C; n = 11 biological, symbols denote the mean and error bars the standard deviation (not shown if smaller than symbols). d, Retron mutants are not affected in aerobic conditions. Experiment and plotting as in panel c, but strains were grown aerobically (n=11 biological). e, RNAse H and Exo VII are involved in msDNA biosynthesis. msDNA was extracted from STm wildtype and retron-deletion strains carrying plasmid p-retron-ΔrcaT. Extracted msDNA was electrophoresed in TBE-Polyacrylamide gels (n=3 biological). f, Deleting rcaT reverts the cold-sensitivity of retron mutants. STm strains were grown and spotted as in Fig. 1d (n=2 biological). g, Deleting rcaT reverts the anaerobic sensitivity of retron mutants. Growth curves of STm strains were obtained and plotted as in panel c (n=11 biological). h, RcaT is a soluble protein. Untagged and rcaT-3xFlag STm WT were grown in LB at 37°C, lysed, and samples were separated into soluble and membrane fractions through ultra-centrifugation steps. Different fractions were analysed by SDS-PAGE and immunoblotting. LpoA and RecA were used as controls for the membrane and soluble fraction, respectively (n=2 biological). i-j, Suppressors grow like wildtype in cold temperatures. Suppressors isolated from cold-sensitive STm mutants (ΔrrtT, ΔxseA, Δmsrmsd; panel i) or the catalytic rcaT-E107Q mutants (panel j) were grown, serially diluted, and spotted on LB plates as in Fig. 1b. Identified suppressor mutations are indicated (n=2 biological). k-l, Loss-of-function point mutations do not affect RcaT levels. RcaT protein levels were quantified by MS in mutant strains (indicated by dashed lines in panels i-j). Y-axis is the ratio of RcaT protein levels (log2 fold-change) in retron deletion strains compared to WT (panel k) or the same ratio in RcaT point mutants compared to the background strain, in which suppressor was isolated in (panel l). The grey dotted line represents no change to RcaT protein levels. Black lines denote the mean (n=2–4 biological).

Extended Data Figure 2|. Retron-Sen2 functions as a toxin/antitoxin system in E. coli and the antitoxin does not shut down RcaT expression.

a, RcaT can be inhibited by msrmsd-rrtT in trans. E. coli carrying binary combinations of plasmids p-rcaT, p-retron-ΔrcaT, and empty vectors, were grown in LB with appropriate antibiotics, serially diluted and spotted as in Fig. 1f (n=3 biological). b, RcaT inhibition requires both msrmsd and rrtT in E. coli. E. coli with plasmids carrying retron-components were grown and spotted as in Fig. 1f. Only the intact retron could restore growth. RcaT expression is sufficient to inhibit growth (n=3 biological). c, RcaT inhibition requires RNAse H and Exo VII in E. coli. E. coli strains (WT, ΔxseA, ΔxseB, ΔrnhA) carrying plasmid p-retron-ΔrcaT or p-retron were grown, serially diluted and spotted as in panel b (n=2 biological). d, RcaT levels are only slightly affected by antitoxin deletions. rcaT-3xFlag STm strains (WT and retron-deletions) and the STm untagged strain (native) were either grown in LB at 37°C, or shifted to 20°C for 5 hours. Protein samples from strains were analysed by SDS-PAGE and immunoblotting. LpoA levels (α-LpoA antibody) were used as loading control. Bars and error bars indicate mean and standard deviation, respectively (n=5 biological). e, Quantification of RcaT-3xFlag signal from 37°C immunoblots in panel d using ImageJ (pixel-density). Bars and error bars depict the mean and standard deviation, respectively (n=5 biological).

Extended Data Figure 3|. Functional Flag-tagged RcaT-RT co-immunoprecipitate independently of temperature.

a-b, Volcano plots of affinity purifications of 3xFlag-tagged RrtT (panel a) and RcaT (panel b) at 37°C and 20°C in wildtype and different mutant backgrounds. Same experiment as in Fig. 2a–b, presenting additionally the AP data from 37°C. The x-axis represents the average log2 ratio of identified proteins (by more than 2 independent peptides) in rrtT-3xFlag and rcaT-3xFlag AP samples compared to an untagged STm WT control strain (n=2 biological), and the y-axis represents p values of these log2 ratios (two-tailed limma). c, Flag-tagged RcaT retains its function. Untagged and rcaT-3xFlag tagged STm strains (WT and retron-deletions) were grown for 5–6 hours at 37°C in LB, serially diluted, spotted on LB plates, and incubated at indicated temperatures. * denotes ΔSTM14_4645::cat, used to co-transduce the scarless Flag-tagged rcaT (co-transduction verified by PCR) (n=2 biological). d, Flag-tagged RrtT retains its function. Untagged and rrtT-3xFlag tagged STm strains (WT and retron-deletions) were grown for 5–6 hours at 37°C in LB, serially diluted, spotted on LB plates, and plates were incubated either at indicated temperatures (n=2 biological). e, Flag-tagging rrtT does not alter retron protein expression, whereas flag-tagging rcaT decreases levels of both retron proteins. Proteins in input (whole proteome) samples used for samples shown in Fig. 2a–b were quantified by MS. Protein levels (log2-fold change) of RcaT and RT were compared between Flag-tagged and untagged STm WT strains (x-axis). Lines denote the mean and come from two biological replicates. RT was downregulated in rcaT-3xFlag tagged strains, whereas RcaT was not even detected (n.d.), presumably due to lower levels (chromosomal RcaT is at the level of detection by MS, so small changes can bring it below detection). Note that despite lower levels of both retron components, rcaT-3xFlag retains its function (panel c) and is expressed (ED Fig. 2d).

Extended Data Figure 4|. RT interacts with msDNA.

a, Purification of protein RT-Sen2. An E. coli BL21 (DE3) CodonPlus-RIL strain carrying plasmid p-msrmsd-rrtT-6xHis was used to purify protein RT-Sen2–6xHis (C-terminal fusion), by nickel-column immobilized metal-affinity chromatography (n=2 purifications). b, His-tagging rrtT does not affect its antitoxin activity. E. coli BL21-AI strains carrying binary combinations of plasmids p-msrmsd-rrtT-6xHis, p-rcaT, and empty vectors, were grown for 5–6 hours at 37°C in kanamycin-LB, serially diluted, spotted on LB-kanamycin plates with or without arabinose, and plates were incubated at 20°C (n=6 biological). c, E. coli BL21 does not normally produce DNA of similar size to msDNA-Sen2. msDNA was extracted from STm or E. coli BL21 carrying plasmid p-msrmsd-rrtT or an empty plasmid. Extracted msDNAs were electrophoresed in TBE-polyacrylamide gels (n=4 biological). d, Isolation of msDNA from purified RT-Sen2 from E. coli (same prep as in panel a). Total DNA was extracted from 500 μg of purified RT-Sen2–6xHis protein and electrophoresed on a TBE-polyacrylamide gel (n=2 purifications).

Extended Data Figure 5|. Different retrons inhibit growth in different environments.

a, STm and E. coli DE-COMM-4976 (Eco10), SC402 (Eco3), and REL606 (Eco1) (WT, ΔrnhA, ΔrcaT, or double mutant strains) were grown for 5–6 hours at 37°C in LB, serially diluted, and spotted on LB or LB pH 5.5 plates. Plates were incubated at indicated temperatures (n=2 biological). b, Same strains as in panel a were grown anaerobically at 37°C in LB for 24h in microtiter plates while measuring OD578. Here plotted the yield of mutant strains compared to the yield of the corresponding WT strain. n=4 (2 biological/2 technical). c, Only RcaT-Sen2 and -Eco9 act as toxins when overexpressed alone. E. coli BW25113 strains carrying retron- or rcaT-encoding plasmids (Sen2, Eco9, Eco10, Eco3, and Eco1), were grown for 5–6 hours at 37°C in LB-spectinomycin, serially diluted, spotted on LB-spectinomycin plates with or without arabinose, and plates were incubated at 37°C (n=2 biological).

Extended Data Figure 6|. Retron-Eco9 is similar to Retron-Sen2.

a, Retron-Eco9 has a similar operon structure to Retron-Sen2. Retron-Eco9 contains msr, msd, rcaT and rrtT regions, 85%, 58%, 43%, and 49% identical to the corresponding Retron-Sen2 regions (first two nucleotide, last two protein level); rrt = retron reverse transcriptase. b, msDNA-Eco9 is similar to msDNA-Sen2. Models of msDNAs-Sen2 and -Eco9. c, msDNA-Eco9 biosynthesis requires ExoVII and RNase H. msDNA was extracted from E. coli BW25113 strains (wildtype, ΔxseA, ΔxseB, ΔrnhA) carrying plasmid PBAD-msrmsd-RT-Eco9 and electrophoresed in a TBE-Polyacrylamide gel (n=2 biological). d, RcaT-Eco9 inhibition requires RNAse H and Exo VII in E. coli. E. coli BW25113 strains (wildtype, ΔxseA, ΔxseB, ΔrnhA) carrying plasmid Ptac-msrmsd-Eco9, along with either PBAD-RT-Eco9 (−), or PBAD-RT-RcaT-Eco9 (+), were grown for 5–6 hours at 37°C in LB with appropriate antibiotics, serially diluted, spotted on antibiotics-LB plates, with/without arabinose, and plates were incubated overnight at 37°C (n=2 biological). e, Non-cognate RT-msrmsd pairs produce msDNA. msDNA were extracted from E. coli BW25113 co-expressing the RT of Retrons-Sen2 (Se) or -Eco9 (Ec) (PBAD-RT; Se or Ec – arabinose induction), and Ec/Se msrmsd (Ptac-msrmsd or an empty vector (−)). Extracted msDNA were electrophoresed in a TBE-polyacrylamide gel (n=2 biological). f, Ec RT cannot act as antitoxin at basal msrmsd (Se) levels, but can when latter is induced (Fig. 2e), in line with lower capacity for Ec RT to reverse transcribe the Se msrmsd (panel e). Experiment conducted as in Fig. 2e (n=2 biological).

Extended Data Figure 7|. TAC low-IPTG, reproducibility, hits heatmap and validation.

a-b, TAC results from low-IPTG induction. Experiments as described in Fig. 3b–c, but with low-IPTG plasmid-library induction for MOB (panel a; p1) and TransBac (panel b; p2). Prophage gene fold enrichment (FE) and p values (one-tailed Fisher’s exact test) are shown on top. c-d, TAC reproducibility in control and experiment plates. Unprocessed colony opacity values of strains from control (ctrl) or experiment (exp) plates, derived from replicate plates 1 and 2, were plotted against each other. The overexpression library and IPTG concentration contained in each plate-pair are denoted above the plots. Results shown for MOB (panel c; p1) and TransBac (panel d; p2). Replication correlation (R²) is shown for each reproducibility plot. Green-colored dots correspond to the hits shown in panels a-b and in Fig. 3b–c. e, Non-parametric comparison of identified triggers across overexpression libraries. Trigger-genes identified from the MOB (p1) and the TransBac (p2) overexpression libraries were rank-ordered in percentiles based on their mean z-score difference value, as calculated from TAC screens conducted in low, or high-IPTG induction. Not available (NA) denotes genes for which measurements were flagged as problematic (see Methods) in the respective library/IPTG-concentration. Absent denotes genes which are missing from one library. Prophage and phage-related trigger-genes are in bold and underlined, respectively. f, Plasmids (p1/p2-trigger gene in green) were conjugated into E. coli BW25113 carrying p-empty, p-retron, p-retron-ΔrcaT, or p-rcaT. 384-colony-arrays of the transconjugants were pinned on LB plates containing appropriate antibiotics, arabinose, and IPTG concentrations (IPTG concentrations with strongest effect shown here). The y-axis represents the triggering-degree of each plasmid, measured as the colony opacity ratio of strains (p-retron-ΔrcaT + p1/p2-trigger) divided by the (p-retron + p1/p2-trigger). p1-control values were derived by measuring the same colony opacity ratio, for p1-blocker genes (n=72 – 36 biological X 2 technical replicates), as a negative control. Ratios were calculated from n=10 – 5 biological X 2 technical replicates for p1-trigger genes (except for p1-yfbO; n=8 – 4 biological X 2 technical replicates), and from n=24 – 12 biological X 2 technical replicates for p2-trigger genes. Representative colonies of strains carrying p1/p2-trigger plasmids are shown below the graphs. Horizontal lines denote the average fitness ratio, and error bars denote the standard deviation. The p1-control fitness data are shown in grey and p1-dam was plotted separately, to avoid compressing the scores of the rest of the trigger-genes. g, Plasmids (p1/p2-trigger gene in green, and p1-empty/p2-empty) from conjugation donor strains were conjugated with E. coli BW25113 carrying plasmids p-retron-ΔrcaT and p-retron. Transconjugants were grown for 5 hours in LB with appropriate antibiotics, serially diluted, spotted on LB plates containing antibiotics, arabinose, and IPTG (no IPTG, low, or high), and incubated at 37°C. Triggers specifically inhibit growth by activating RcaT, and triggering depends on induction levels.

Extended Data Figure 8|. TIC high-IPTG, reproducibility, hits heatmap, and validation.

a-b, TIC results from high-IPTG induction. Experiments as described in Fig. 3d–e, but with low-IPTG plasmid-library induction for MOB (a; p1) and TransBac (b; p2). Prophage gene fold enrichment (FE) p values (one-tailed Fisher’s exact test) are shown on top. yfbO was selected as a blocker, despite not being over the cut-off, due to genetic linkage with yfbN. c-d, TIC reproducibility in control and experiment plates – as in ED Fig. 7c–d. Blue-colored dots correspond to the hits shown in panels a-b and in Fig. 3d–e. e, Non-parametric comparison of identified blockers across overexpression libraries – as in ED Fig. 7e. f, Plasmids (p1-blocker gene in blue and p1-empty in black) were conjugated into E. coli BW25113 carrying p-empty, p-retron, p-retron-ΔrcaT, or p-rcaT. 384-colony-arrays of the transconjugants were pinned on LB plates containing appropriate antibiotics, arabinose, and IPTG concentrations (IPTG concentrations with strongest effect shown here). The y-axis represents the blocking degree of each plasmid, measured as the colony opacity ratio of strains carrying the different blockers (p-rcaT + p1-blocker) divided by the average of a strain carrying trigger plasmids (p-rcaT + p1-trigger), as a negative control. Fitness ratios calculated from n=8 – 4 biological X 2 technical replicates. The mean colony opacity for p1-trigger was derived from n=110 – 55 biological X 2 technical replicates. Representative colonies of strains carrying p1-blocker plasmids shown below the graphs. Horizontal lines denote the average fitness ratio, and error bars denote the standard deviation. The p1-control fitness data are shown in grey (full toxicity of RcaT without any blocker). g, Plasmids (p1/p2-blocker gene in blue, and p1-empty/p2-empty) from conjugation donor strains were conjugated with E. coli BW25113 carrying plasmid p-rcaT or a p1-empty plasmid. Transconjugants were grown as in ED Fig. 7g. Blockers specifically alleviate RcaT-mediated toxicity in a dose-dependent manner.

Extended Data Figure 9|. Dam-mediated retron-TA triggering.

a, Bacterial/phage dam homologues trigger Retron-Sen2. E. coli BW25113 carrying combinations of p-retron-ΔrcaT, p-retron, p1-dam^Ec, p1-dam^STm, p1-dam^P1 and p1-empty, were grown for 5 hours in LB with appropriate antibiotics, serially diluted, spotted on LB plates with antibiotics, with/without arabinose, and IPTG (low), and incubated at 37°C (n=2 biological). b, Phage P1 dam triggers the chromosomal Retron-Eco9 in E. coli NILS-16. E. coli NILS-16 WT and ΔrcaT-Eco9 strains carrying plasmids p2-dam^P1 or p2-empty, were grown for 5 hours in LB-hygromycin, serially diluted, spotted on LB-hygromycin plates, with or without IPTG, and incubated at 37°C (n=2 biological). c, Phage P1 Dam is expressed at comparable levels to the p2-dam^P1 plasmid. An E. coli BW25113 ΔlacY strain carrying an empty plasmid was infected with phage P1vir (MOI 2) for 0, 5, or 35 minutes (uninfected, early, and late infection). The same strain, but carrying a single-copy vector expressing Dam^P1 was grown in LB and the plasmid was induced with 60 μM of IPTG. The levels of the Dam^P1 protein were quantified by proteomics (TMT labeling), and Dam^P1 amounts were compared between phage-infection and plasmid-induction (X-axis). Phage infection or Dam^P1 induction does not alter the expression of endogenous Dam (Table S7), which has diverse enough sequence to be distinguished from Dam^P1 (Dam 39% identical for 266/754 residues of Dam^P1) (n=2 biological). d, The Dam^P1 protein levels produced by P1vir are enough to trigger Retron-Eco9. The same strains as in panel b were grown in LB-hygromycin at 37°C with IPTG (0 – 1000 μM), their optical density was measured at 8 hours post-induction, and the fitness of the WT was compared to the ΔrcaT-Eco9 strain (X-axis). Vertical lines denote the mean fitness ratio, and horizontal lines denote standard deviation (n=3 biological). e, Dam overexpression does not impact msDNA levels. msDNA was isolated from STm carrying combinations of plasmids p-retron-ΔrcaT, p1-dam, or p1-empty. Plasmids were co-induced with arabinose and IPTG (low). Extracted msDNA was electrophoresed in a TBE-polyacrylamide gel (n=4 biological). f, RcaT in p-retron^mut is functional. E. coli BW25113 ΔxseA strains carrying combinations of plasmids p-retron^WT, p-retron^mut, p1-dam, and p1-empty, were grown and spotted as described in panel a (n=3 biological). g, p-retron^mut produces same levels of msDNA as wildtype retron. msDNA was extracted from STm strains carrying p-retron^WT, or p-retron^mut. Extracted msDNA was electrophoresed in a TBE-polyacrylamide gel (n=2 biological).

Extended Data Figure 10|. racC-recE are a blocker-trigger gene-pair.

a, RecE triggers the Retron-Eco9. E. coli BW25113 carrying combinations of plasmids p-retron-Eco9, p-retron-ΔrcaT-Eco9, p1-recE, and p1-empty, were grown for 5 hours in LB with appropriate antibiotics, serially diluted, spotted on LB plates with antibiotics, arabinose, and IPTG (low), and incubated at 37°C (n=3 biological). b, RecE degrades mature and immature msDNA in vitro. msDNA was extracted from STm strains (WT, or ΔxseA) carrying plasmid p-retron-ΔrcaT. msDNA extracts were incubated with recombinant RecE (see Methods), and electrophoresed on a TBE-Polyacrylamide gel (n=3 biological). c, RacC blocks RcaT-Eco9. E. coli BW25113 carrying combinations of plasmids p-rcaT-Eco9, p1-racC, and p1-empty, were grown and spotted as in panel a with different IPTG concentrations (n=3 biological). d, RacC blocks cold-sensitivity of endogenous STm retron-antitoxin deletions. STm strains (WT, ΔrrtT, ΔxseA, ΔrnhA, Δmsrmsd) carrying plasmid p1-racC or the empty vector were grown and spotted as in panel a (n=2 biological).

Extended Data Figure 11|. Phage defense assay data for Fig. 5a.

Growth curves of strains containing retrons in pSC101 vectors (3–5 copies/cell) under the control of their native promoter or an empty vector (see Methods); n=2 biological for each of the panels: a (37°C) and b (25°C).

Figure 1|. Retron-Sen2 encodes and inhibits a toxin protein, RcaT.

a, Biosynthesis of msDNA-Sen2. RT-Sen2 (rrtT) binds to msr-RNA and reverse transcribes msd-RNA to DNA, while Ribonuclease H (rnhA) degrades the RNA template. msr-RNA and msd-DNA are covalently joined by a 2’−5’ phosphodiester bond. Exodeoxyribonuclease VII (xseA/xseB) cleaves the first four nucleotides from msd-DNA, separating the DNA/RNA hybrid and releasing mature msDNA-Sen2. b, Perturbing any step in msDNA biogenesis leads to cold sensitivity. STm retron-deletion strains were grown for 5–6 hours at 37°C in LB, serially diluted, spotted on LB plates, and incubated at 15°C (n=4 biological). c, Cold-sensitivity suppressor mutations of ΔrrtT, ΔxseA, Δmsrmsd map in rcaT (STM14_4640). d, Deleting rcaT reverts the cold-sensitivity of ΔxseA. STm strains were grown and spotted as in 1b (n=2 biological). e, RcaT has two distinct predicted domains – a nucleotide glycosidase (purple; residues 24–185) and a winged helix-turn-helix (grey; residues 1–23 and 186–320). The structure prediction of RcaT-Sen2 comes from AlphaFold2. Mutations suppressing RcaT’s toxicity are depicted. f, Retron-Sen2 functions as a TA in E. coli. E. coli with plasmids p-rcaT (wildtype or D296V) or p-retron (msrmsd-rcaT-rrtT) were grown as in B (but with antibiotics), spotted on LB-plates with/without arabinose and incubated overnight at 37°C (n=3 biological).

Figure 2|. RcaT proteins interact and coevolve with their cognate retron-RTs.

a-b, RT and RcaT reciprocally co-immunoprecipitate, independent of msDNA. Chromosomally tagged rrtT-3xFlag or rcaT-3xFlag STm strains (WT, ΔxseA, Δmsrmsd/Δmsd) grown at 37°C and shifted for 5 hours at 20°C were used to affinity purify (AP) RT-3xFlag or RcaT-3xFlag. Protein abundances in APs of the tagged strains versus the untagged STm WT, measured by quantitative MS (y-axis; averages of n=2 biological, see ED Fig. 3a–b for volcano plots, also from 37°C; Table S2 for all data). c, RcaT is phylogenetically widespread. Maximum-likelihood phylogenetic tree of 778 RcaT proteins from genomes (4929) and metagenomes (1009) (clustered at 99% identity). The inner and first rings depict phyla and proteobacterial classes; the second ring the rcaT-rrt genetic distance (Δ; bp); and the third ring the instances of each unique RcaT (bar-graphs). Positions of RcaT-Sen2, -Eco9, -Eco10, -Eco3, and -Eco1 are annotated, and the dashed grey line marks the Eco1-like (Δ>120) and Sen2-like (Δ<15) subfamilies. For explanations of outliers within the two subfamilies see Table S3. d, RcaT and RT coevolve. Two-dimensional densities of pairwise leaf-to-leaf distances between RcaT and RT phylogenetic trees. e, RT needs msDNA to inactivate its cognate RcaT. E. coli was co-transformed by plasmids carrying RcaT-RT combinations of Retron-Sen2 (Se) and -Eco9 (Ec) (PBAD-RcaT-RT; Se-Se, Se-Ec, Ec-Se, or Ec-Ec – arabinose), and plasmids carrying msrmsd (Ptac-msrmsd; Se, or Ec – IPTG), or an empty vector (−). Strains were grown for 5–6 hours at 37°C in LB with antibiotics, serially diluted, spotted on LB plates with IPTG, antibiotics, and arabinose (+/−), and incubated overnight at 37°C (n=2 biological).

Figure 3|. Toxin Activation/Inhibition Conjugation (TAC/TIC): a genetic screen to discover TA triggers and blockers.

a, Two systematic, arrayed, mobilizable E. coli gene-overexpression libraries (Ptac-gene 1-X; library-plasmids^,) are conjugated with recipient strains carrying PBAD-controlled plasmids, expressing a TA (PBAD-TA; msrmsd-rcaT-rrtT) or just the toxin (PBAD-T; rcaT). Inducing PBAD-T, but not PBAD-TA, inhibits growth. Co-inducing library-plasmids carrying TA-triggers leads to TA-mediated growth inhibition (TAC), while co-inducing library-plasmids carrying TA-blockers alleviates toxin-mediated growth inhibition (TIC). b-c, TAC screen using the IPTG-inducible plasmids from the MOB (B; p1) and TransBac (C; p2) libraries, respectively. Arrayed libraries (384-density format) were conjugated with an E. coli recipient carrying an arabinose-inducible p-retron-Sen2 plasmid. Transconjugants carrying both plasmids were selected, colony integral opacities were measured, and fitness z-scores of strains were calculated on experiment plates containing arabinose + high (1 mM) IPTG, or on control plates containing only high IPTG (see Methods). Z-score differences per strain (y-axes) were derived by subtracting z-scores in control plates from experiment plates (n=2; see Table S4–5 for all scores and S6 for functions of hits). Scores shown from high-IPTG induction (low-IPTG in ED Fig. 7a–b). X-axes depict gene chromosomal position in E. coli MG1655. The hit cut-off (dashed line), prophage gene fold-enrichment (FE) and p values are shown, prophage or phage-related trigger-genes are bolded or underlined, respectively. d-e, TIC screen using the same library-plasmids as in 3b-c. Conjugation was performed as in 3b-c, but with an E. coli recipient carrying a p-rcaT plasmid. Axes and data analysis as in 3b-c, but z-score differences derived by subtracting experiment plates from control plates. Scores shown from low-IPTG induction (high-IPTG in ED Fig. 8a–b).

Figure 4|. Triggers activate retron toxins by directly affecting the msDNA.

a, Overexpressing Dam triggers the endogenous Retron-Sen2. STm strains carrying p1-dam or the empty vector were grown for 5 hours in LB-ampicillin, serially diluted, spotted on LB-ampicillin plates containing IPTG (low), and incubated at indicated temperatures (n=4 biological). b, Mutating the Dam motif in msDNA, without affecting duplex formation, abolishes retron triggering. E. coli carrying combinations of p-retron^WT, p-retron^mut (5’-GTTC-3’ msDNA mutation), p1-dam, and empty vectors were grown, diluted and spotted as in 4a; LB plates with antibiotics, IPTG (low), arabinose were incubated overnight at 37°C (n=3 biological). c, Dam methylates the 5’-GATC-3’ site on msDNA. msDNA was isolated from STm strains carrying plasmids p-retron^WT or p-retron^mut, and purified by gel extraction. Purified msDNA was digested with DpnI and electrophoresed on denaturing TBE-polyacrylamide gel (n=2 biological). d, Dam triggers the retron-TA (purple RcaT) by methylating the msDNA. e, RecE degrades only mature msDNA in vivo. msDNA was extracted from WT or ΔxseA STm strains carrying p-retron-ΔrcaT with p1-recE, or p1-empty. Plasmids were co-induced with arabinose and IPTG (low). Extracted msDNA was electrophoresed in a TBE-polyacrylamide gel. Mature msDNA runs close to the 50bp ladder band, whereas unprocessed DNA runs higher and accumulates upon RecE overexpression (WT/ΔxseA n=4/n=2 biological). f, Rac prophage genes racC-recE in E. coli K-12 are a blocker-trigger gene-pair. g, RecE triggers the retron-TA (purple RcaT) by degrading the msDNA.

Figure 5|. Retron-TAs protect against phages.

a, Retron-TAs defend against diverse phages. E. coli BW25113 carrying Retrons-Sen2, -Eco9, or -Eco1 under their own promoter, or an empty plasmid, were grown at indicated temperatures until OD595 = 0.05, infected with phages at different MOIs, and growth was monitored over time – indicated when retrons allowed host cells to resist phage lysis. Primary data shown in ED Fig. 11 (n=2 biological). b-c, Retron-Sen2 protects against phage T5 at lower temperatures, whereas Retron-Eco9 protects against a broad phage range, independent of temperature. Experiment as in 5a (lines depict the average of n=2 biological). d, A bacterial-phage arms race. Phages encode proteins (green-blue circle) to counteract bacterial Restriction-Modification systems (early-defense systems; RM). These anti-RM proteins (Dam, RecE) can trigger retron-TAs by directly disrupting the msDNA part of the antitoxin. This frees RcaT, which inhibits the growth of the infected cell, and thereby stops phage propagation – abortive infection (Abi). Thus, phage blockers of early bacterial defenses can trigger secondary Abi defense systems. To counteract, phages have evolved blockers (blue) to inhibit the toxin of Abi systems (RacC).