High Frequency of Dynamic Rearrangements in CRISPR Loci

Abstract

AO_SCPLOWBSTRACTC_SCPLOWCRISPR-Cas immunization of prokaryotes proceeds by the acquisition of short fragments of invading DNA and integrating them into specific positions within the host genome in a process called adaptation. Adaptation is thought to be polarised, which suggests that CRISPR array spacer order reflects the recentness of the infection. The detailed processes through which CRISPR loci arise, and how they evolve are not completely clear. In this study, we collected 12,461 prokaryotic genomes, and using a combination of four different approaches and a series of conservative filters, we identified CRISPR arrays in 82.7% of Archaea and 40.6% of Bacteria. To understand spacer evolution in these CRISPR loci we firstly tracked point mutations in CRISPR repeats, and secondly, we carried out a comparative analysis of arrays that share multiple similar spacers. Both results indicate that CRISPR arrays are frequently dynamically rearranged. These findings are at odds with a model that suggests that spacer order is likely to reflect the recentness of infection. We conclude that the order of spacers in a CRISPR array, as well as the spacer content of the array, is likely to arise from a combination of events, such as insertion in the middle of the array, recombination within or between arrays, or Horizontal transfer of all or part of an array. We suggest these rearrangements are favoured by natural selection in complex and dynamic environments.