CRISPR/Cas9-mediated phage resistance is not impeded by the DNA modifications of phage T4

Abstract

Bacteria rely on two known DNA-level defenses against their bacteriophage predators: restriction-modification and Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-CRISPR-associated (Cas) systems. Certain phages have evolved countermeasures that are known to block endonucleases. For example, phage T4 not only adds hydroxymethyl groups to all of its cytosines, but also glucosylates them, a strategy that defeats almost all restriction enzymes. We sought to determine whether these DNA modifications can similarly impede CRISPR-based defenses. In a bioinformatics search, we found naturally occurring CRISPR spacers that potentially target phages known to modify their DNA. Experimentally, we show that the Cas9 nuclease from the Type II CRISPR system of Streptococcus pyogenes can overcome a variety of DNA modifications in Escherichia coli. The levels of Cas9-mediated phage resistance to bacteriophage T4 and the mutant phage T4 gt, which contains hydroxymethylated but not glucosylated cytosines, were comparable to phages with unmodified cytosines, T7 and the T4-like phage RB49. Our results demonstrate that Cas9 is not impeded by N6-methyladenine, 5-methylcytosine, 5-hydroxymethylated cytosine, or glucosylated 5-hydroxymethylated cytosine.

Figure 1. Native E. coli spacers target phage with modified DNA.

In a BLASTn search, 1749 unique spacers from sequenced E. coli CRISPR arrays were queried against T4-like phage genomes. (A) Spacer S641 matches 25 of 32 nucleotides in phage T2. The putative protospacer has a permissible E. coli CRISPR PAM AAG and the matching nucleotides are concentrated at the 5′ end as a seed sequence. The spacer originated from the CRISPR1 locus of E. coli strain 579, a human-associated isolate from France. (B) Spacer S134 matches 29 of 32 nucleotides in phage CC31. While the protospacer in phage CC31 has five nucleotides inserted in the center of the sequence, there are 15 exactly matched nucleotides at the 5′ end in addition to 14 matched nucleotides after the insertion. The PAM GAG and strongly matched seed region suggest it is a plausible E. coli CRISPR target. This spacer was found in several strains, including E. coli C str. ATCC 8739, ECOR strains 17 through 21, one farm pig and two human fecal samples in France, duck and cattle fecal samples in Australia , and enterotoxigenic E. coli (ETEC) strain UMNK88. The spacer and matching protospacer are in blue, the transcribed CRISPR RNA (crRNA) in bold black, and PAM sequence in red.

Figure 2. Cas9 cuts methylated cytosines and adenosines in E. coli.

(A) Synthetic targets were designed to contain one to two dam (orange) or dcm (blue) sites. A control unmethylated sequence (+) was included. The PAM sequence NGG for SpCas9 recognition is underlined. (B) In serial transformations, we selected for the coexistence of DS-SPcas, the protospacer plasmid, and each spacer plasmid. The number of transformants was divided by the number of colonies resulting from a control transformation using a spacer plasmid (-) that did not target the protospacer plasmid. This relative number of transformants is plotted for E. coli K-12 and E. coli K-12 dam⁻/dcm⁻ from three independent experiments. Lines represent the median.

Figure 3. Cas9 reduces E. coli susceptibility to phages T7 and RB49.

(A) Spacers against T7 were targeted against the primase/helicase gene (gene 4A and 4B). The PAM is underlined in the sequence and shown as a black box in the diagram showing the orientation and location of the protospacer (white box) on the gene. In a representative T7 plaque assay of protected and unprotected strains, there is substantial lysis on wild-type (wt) E. coli K-12, visible plaquing on cells with spacer 2 (sp 2), and no plaques on cells with spacer 1 (sp 1). (B) The efficiency of plating of T7 was calculated for each protected strain relative to the unprotected wild-type strain. Independent replicates of E. coli B (n = 4, 3, 3) and E. coli K-12 (n = 5, 5, 7) are plotted. Lines represent the median. (C) Spacers against RB49 were constructed against the major capsid protein (gp23). In a typical RB49 plaque assay, there is notable lysis on wild-type E. coli B, some plaques on cells with spacer 1, and a few plaques on cells protected with spacer 2. (D) The efficiency of plating of RB49 was quantified for each protected strain relative to the unprotected wild-type strain. Shown are independent replicates of E. coli B (n = 5, 3, 3) and E. coli K-12 (n = 3, 3, 3). Lines represent the median.

Figure 4. Cas9 reduces E. coli susceptibility to phages T4 and T4 gt.

(A) The structures of cytosine and modified cytosines are shown. T4 gt has 100% hydroxymethylated cytosines (hmCs). T4 has 100% glucosyl-hydroxymethylated cytosines (ghmCs), specifically 70% α- and 30% β-ghmCs. The ghmC structure shown is in the β-configuration. (B) Spacers against T4 were also designed against the major capsid protein (gp23), which is homologous to that of RB49. For comparison, the RB49 protospacers are aligned below in italics, where dots indicate identical nucleotides. In the T4 sequences, the PAM is underlined. The PAM (black box) and protospacer (white box) are represented on the gene. (C) In a typical plaque assay with T4 gt (left plate), there was complete lysis on wild-type (wt) restriction-less (r-l) E. coli K-12 and few plaques on cells with spacers 1, 2, or 3 (sp 1, sp 2, or sp 3). In an assay with T4 (right plate), there was complete lysis on wild-type E. coli K-12 MG1655, numerous plaques on cells with spacer 1 or 3, and about a dozen on spacer 2. (D) The efficiency of plating of T4 and T4 gt was quantified for each protected strain relative to the unprotected wild-type strain. Independent replicates of restriction-less E. coli K-12 (n = 5, 3, 3, 5), E. coli K-12 (n = 4, 4, 5, 6), and E. coli B (n = 5, 3, 3, 3) are plotted. Lines represent the median.