Escherichia coli CRISPR arrays from early life fecal samples preferentially target prophages

Abstract

CRISPR-Cas systems are defense mechanisms against phages and other nucleic acids that invade bacteria and archaea. In Escherichia coli, it is generally accepted that CRISPR-Cas systems are inactive in laboratory conditions due to a transcriptional repressor. In natural isolates, it has been shown that CRISPR arrays remain stable over the years and that most spacer targets (protospacers) remain unknown. Here, we re-examine CRISPR arrays in natural E. coli isolates and investigate viral and bacterial genomes for spacer targets using a bioinformatics approach coupled to a unique biological dataset. We first sequenced the CRISPR1 array of 1769 E. coli isolates from the fecal samples of 639 children obtained during their first year of life. We built a network with edges between isolates that reflect the number of shared spacers. The isolates grouped into 34 modules. A search for matching spacers in bacterial genomes showed that E. coli spacers almost exclusively target prophages. While we found instances of self-targeting spacers, those involving a prophage and a spacer within the same bacterial genome were rare. The extensive search for matching spacers also expanded the library of known E. coli protospacers to 60%. Altogether, these results favor the concept that E. coli's CRISPR-Cas is an antiprophage system and highlight the importance of reconsidering the criteria use to deem CRISPR-Cas systems active.

Keywords: CRISPR; E. coli; bacteriophage; gut; microbiome; phage; phage resistance; virome.

Figure 1

CRISPR diversity in E. coli isolates; (A) each node represents a CRISPR array (isolate) and each edge represents a level of shared spacers between two nodes; the length of the edge is indicative of the Jaccard similarity index; the color of the node represents the module the CRISPR array belongs to; (B) one random CRISPR array per module is illustrated; each colored square corresponds to a spacer; the colored dot next to the array refers to the module it represents; two spacers with the same diamond–square color combination share homologous sequences; the numbers in parentheses correspond to the number of CRISPR arrays in each module.

Figure 2

Specificity of the spacers to the early life gut environment; (A) distribution of the number of spacers per cluster, showing known (blue, identical to previously sampled spacers) and new (orange, unique to our dataset) clusters; (B) mean relative spacer position for each spacer cluster along the 5′–3′ axis, grouped according to known (blue, bottom) and new (top, orange) clusters.

Figure 3

The E. coli spacers preferentially target prophages found in Enterobacteriaceae genomes; (A) list of the 15 most targeted genes in bacterial genomes; genes marked with an asterisk are strictly phage genes; (B) scatter plot of the number of targets and MAD for each bacterial genome targeted by spacers; the dotted vertical line at x = 10 corresponds to the minimum cut-off for bacterial genomes that were investigated for the presence of a prophage; (C) network representation of the viral sequences targeted by spacers; each node is a viral sequence, and each edge between two nodes signifies at least 95% identity over 85% coverage, which is the definition of a viral species; nodes are colored according to their origin (red, prophages in NCBI bacterial genomes; green, NCBI phage genomes; orange, prophage in COPSAC metagenomes; blue, phage in COPSAC virome; and pink, coliphage from the COPSAC E. Coli supernatant), and their shapes (circle, temperate; diamond, virulent) are based on their replication mode.

Figure 4

Genome map of a representative prophage of the large cluster in Figure 3; all arrows represent genes; the gray arrows are genes coding for hypothetical proteins or proteins of unknow functions; the black dots under the genome map are regions targeted by spacers, and the numbers are the coordinates of the prophage in the bacterial genome.