Abstract
Recent discovery of ectopic repeats (outside CRISPR arrays) provided unprecedented insights into the nondefense roles of CRISPR-Cas. A striking example is the addiction module CreTA (CRISPR-regulated toxin-antitoxins), where one or two (in most cases) ectopic repeats produce CRISPR-resembling antitoxic (CreA) RNAs that direct the CRISPR effector Cascade to transcriptionally repress a toxic RNA (CreT). Here, we demonstrated that CreTA repeats are extensively degenerated in sequence, with the first repeat (ψR1) being more diverged than the second one (ψR2). As a result, such addiction modules become highly specific to their physically-linked CRISPR-Cas loci, and in most cases, CreA could not harness a heterologous CRISPR-Cas to suppress its cognate toxin. We further disclosed that this specificity primarily derives from the degeneration of ψR1, and could generally be altered by modifying this repeat element. We also showed that the degenerated repeats of CreTA were insusceptible to recombination and thus more stable compared to a typical CRISPR array, which could be exploited to develop highly stable CRISPR-based tools. These data illustrated that repeat degeneration (a common feature of ectopic repeats) improves the stability and specificity of CreTA in protecting CRISPR-Cas, which could have contributed to the widespread occurrence and deep diversification of CRISPR systems.