DISARM is a widespread bacterial defence system with broad anti-phage activities

Abstract

The evolutionary pressure imposed by phage predation on bacteria and archaea has resulted in the development of effective anti-phage defence mechanisms, including restriction-modification and CRISPR-Cas systems. Here, we report on a new defence system, DISARM (defence island system associated with restriction-modification), which is widespread in bacteria and archaea. DISARM is composed of five genes, including a DNA methylase and four other genes annotated as a helicase domain, a phospholipase D (PLD) domain, a DUF1998 domain and a gene of unknown function. Engineering the Bacillus paralicheniformis 9945a DISARM system into Bacillus subtilis has rendered the engineered bacteria protected against phages from all three major families of tailed double-stranded DNA phages. Using a series of gene deletions, we show that four of the five genes are essential for DISARM-mediated defence, with the fifth (PLD) being redundant for defence against some of the phages. We further show that DISARM restricts incoming phage DNA and that the B. paralicheniformis DISARM methylase modifies host CCWGG motifs as a marker of self DNA akin to restriction-modification systems. Our results suggest that DISARM is a new type of multi-gene restriction-modification module, expanding the arsenal of defence systems known to be at the disposal of prokaryotes against their viruses.

Figure 1. Two common classes of DISARM systems occur in bacteria and archaea.

(a) Class 1 DISARM systems are composed of the core gene triplet of drmA, a gene with a helicase domain (orange); drmB, DUF1998 domain-containing gene (yellow); and drmC, containing a PLD domain (green). This core gene triplet is preceded by drmD, an SNF2-like helicase (pink) and drmMI, an adenine methylase (peach). RefSeq genome accessions are indicated; system positions appear in Table S2. (b) Class 2 DISARM systems contain, in addition to the core triplet of drmABC, also drmMII, a cytosine methylase (blue). In Bacilli and some haloarchaea the systems also include a ˜800 aa gene of unknown function named here drmE (red).

Figure 2. DISARM provides protection against phages.

(a) The DISARM locus of Bacillus paralicheniformis ATCC 9945a. Numbers below axis represent position on the B. paralicheniformis genome. Locus tags are provided for each gene. (b) Insertion of the DISARM locus into the Bacillus subtilis BEST7003 genome does not impair growth. Curves show the mean of 2 independent experiments with 3 technical repeats each. Error bars are 95% confidence interval of the mean. (c-e) DISARM provides protection against phi3T, Nf and SPO1 phages. Bacteria were infected at time=0 at multiplicities of infection (MOI) of 0.05, 0.5 and 5. Curves show the mean of 2 independent experiments with 3 technical repeats each. Error bars are 95% confidence interval of the mean. (f) Plaque formation of 7 phages on DISARM-containing strains. Y axis represents concentration of plaque forming units (PFU). Shown is mean of 3 replicates, error bars are SD of the mean. Grey bars represent efficiency of plaque (EOP) on DISARM- cells, red bars are EOP in DISARM+.

Figure 3. Phage phi3T adsorption and DNA replication in DISARM-containing cells.

(a) Adsorption of phages to DISARM-containing cells (red) is not impaired compared to control cells (grey). After infection of logarithmic stage cultures (OD₆₀₀=0.3) with phi3T at MOI=1, samples were taken at 5 minutes intervals, and the extracellular (unadsorbed) phage concentration was measured and compared to the initial phage concentrations (Methods). Bars represent mean of 3 experiments, error bars are SEM. (b) Ratio of phage DNA to bacterial DNA during infection. Total DNA was extracted from infected bacteria (MOI=1) at the indicated time points and sequenced using an Illumina sequencer. Y-axis represents relative phage DNA concentrations, compared to bacterial genome equivalents, normalized to the value at t=5 minutes post infection. Each curve represents an independent repeat of the experiment. (c) DISARM prevents lysogeny of phi3T. Agarose gel of multiplex PCR with 3 primer sets, aimed to detect bacterial DNA, phage DNA and lysogen. Lanes are marked with minutes post infection; U lane is the uninfected control. (d) DISARM prevents phage circularization. Outward-facing primers at the edges of the phi3T genome were used to detect phage genome circularization. (e) Schematic representation of fragments amplified in panel d.

Figure 4. Fluorescence microscopy of phage DNA in DISARM- and DISARM+ cells.

(a) DISARM-lacking, RFP-expressing cells (red cells) were co-incubated with DISARM-containing (light blue) cells in a microfluidic device that allows visualization of a single bacterial layer (Methods). Both strains express LacI-CFP constitutively. SPP1 phages containing a LacO array (10⁵ pfu/µl) were flowed into the device from t=15 minutes to t=45 minutes, and image was taken every 5 minutes. Upon injection of the phage DNA, a fluorescent focus of LacI-CFP is formed on the LacO array in the phage DNA. White arrows show foci in DISARM-lacking cells, which do not disappear and grow in size through the time course. Foci on DISARM-containing cells appear (full yellow arrowheads) but later disappear (empty yellow arrowheads). Scale bar is 5 µm. Representative results of 2 independent experiments. (b) Quantification of phage foci over time in the microscopy field of which a subsection is shown in panel a. Phage foci appear starting t=25 minutes, and become established in DISARM-lacking cells. Similar numbers of foci appear in DISARM-containing cells, but these foci disappear over time. Shaded area represents the time where phages were continuously flowed in.

Figure 5. Deletion of DISARM components.

(a-c) Deletion of drmE, drmA or drmB abolished DISARM defence against phi3T. (d) Deletion of drmC has no effect on the defence against phi3T. (e-f) Deletion of drmC reduces DISARM protection against Nf and SPO1, but the deletion strains are still protected compared to the control bacteria. Curves in a-f are mean of 2 independent experiments with 2 technical repeats each, error bars represent 95% confidence interval of the mean. Infections were performed at MOI=0.5.