The spacer size of I-B CRISPR is modulated by the terminal sequence of the protospacer

Prokaryotes memorize invader information by incorporating alien DNA as spacers into CRISPR arrays. Although the spacer size has been suggested to be predefined by the architecture of the acquisition complex, there is usually an unexpected heterogeneity. Here, we explored the causes of this heterogeneity in Haloarcula hispanica I-B CRISPR. High-throughput sequencing following adaptation assays demonstrated significant size variation among 37 957 new spacers, which appeared to be sequence-dependent. Consistently, the third nucleotide at the spacer 3΄-end (PAM-distal end) showed an evident bias for cytosine and mutating this cytosine in the protospacer sequence could change the final spacer size. In addition, slippage of the 5΄-end (PAM-end), which contributed to most of the observed PAM (protospacer adjacent motif) inaccuracy, also tended to change the spacer size. We propose that both ends of the PAM-protospacer sequence should exhibit nucleotide selectivity (with different stringencies), which fine-tunes the structural ruler, to a certain extent, to specify the spacer size.