Bioinformatics Identification of Anti-CRISPR Loci by Using Homology, Guilt-by-Association, and CRISPR Self-Targeting Spacer Approaches

Authors

Yanbin Yin, Northern Illinois University & University of Nebraska – LincolnFollow
Bowen Yang, University of Nebraska—Lincoln
Sarah Entwistle, Northern Illinois University

ORCID IDs

https://orcid.org/0000-0001-7667-881X

Document Type

Article

Date of this Version

9-10-2019

Citation

2019 Yin et al.

Comments

Yin Y, Yang B, Entwistle S. 2019. Bioinformatics identification of anti-CRISPR loci by using homology, guilt-by-association, and CRISPR self-targeting spacer approaches. mSystems 4:e00455-19. https://doi.org/10 .1128/mSystems.00455-19.

Abstract

Anti-CRISPR (Acr) loci/operons encode Acr proteins and Acr-associated (Aca) proteins. Forty-five Acr families have been experimentally characterized inhibiting seven subtypes of CRISPR-Cas systems. We have developed a bioinformatics pipeline to identify genomic loci containing Acr homologs and/or Aca homologs by combining three computational approaches: homology, guilt-by-association, and self-targeting spacers. Homology search found thousands of Acr homologs in bacterial and viral genomes, but most are homologous to AcrIIA7 and AcrIIA9. Investigating the gene neighborhood of these Acr homologs revealed that only a small percentage (23.0% in bacteria and 8.2% in viruses) of them have neighboring Aca homologs and thus form Acr-Aca operons. Surprisingly, although a self-targeting spacer is a strong indicator of the presence of Acr genes in a genome, a large percentage of Acr-Aca loci are found in bacterial genomes without self-targeting spacers or even without complete CRISPR-Cas systems. Additionally, for Acr homologs from genomes with self-targeting spacers, homology-based Acr family assignments do not always agree with the self-targeting CRISPR-Cas subtypes. Last, by investigating Acr genomic loci coexisting with self-targeting spacers in the same genomes, five known subtypes (I-C, I-E, I-F, II-A, and II-C) and five new subtypes (I-B, III-A, III-B, IV-A, and V-U4) of Acrs were inferred. Based on these findings, we conclude that the discovery of new anti-CRISPRs should not be restricted to genomes with selftargeting spacers and loci with Acr homologs. The evolutionary arms race of CRISPRCas systems and anti-CRISPR systems may have driven the adaptive and rapid gain and loss of these elements in closely related genomes.