Extended Data Fig. 2: New spacers are preferentially acquired from MDAΦ and NmeCas9’s PID dictates PAM enrichment for acquisition.

a, SDS-PAGE of purified NmeCas1 and Cas2 proteins used for custom antibody production. b, Cas1 protein induced from a cas1-2 OE cassette integrated into the host genome. 0.1 mM IPTG is used in subsequent experiments. Top, anti-Cas1 western blot using custom antibody; bottom, GroEL probed as loading control. c, Overview of the informatic analysis for new spacers. d, MDAΦ vs. host genome breakdown for DNA source of new spacers from “+1” band in lanes 1 and 4 of Fig. 1d and lane 4 of Fig. 1f. Red dashed line, 3.5 % should be MDAΦ-derived if there is no viral versus self- discrimination by acquisition. Data are mean ± s.d., n = 3. e, New spacers from “+1” band in lane 4 of Fig. 1d are analyzed in e, g, h. Distribution of spacer abundance (reads per million of phage-matching across the MDAΦ genome. f, Transcriptome profiling of MDA transuctants shows that protospacer sampling is not driven by viral transcript abundance. g, Nucleotide frequency plot for viral protospacers. Black line, 25% G/A/T/C equal frequency. h, All 256 possible 4-mers for positions 5-8 nts 3′ of protospacers ranked in descending order of abundance. N₄GATT PAM, red dot; all other 4-mers, blue dots. i, Domain architecture of WT, dcas9, cas9^H1024A and cas9^PID-swap alleles, with D16A and H588A mutations denoted by black stars and H1024A PAM-binding-disrupting mutation by red star. j, Genomically-complemented cas9 mutants are expressed to similar level as WT control in strains in Fig. 1f, g. Top, anti-NmeCas9 western blot; bottom, GroEL as loading control. k, Interference assay showing that dcas9 and cas9^H1024A are interference defective. cas9^PID-swap altered PAM to 3′-N₄CC in interference. Data are log-scale of CFU/mL (mean ± s.d., n = 3) for transformants (red bars) and total cells (blue bars). NS, not significant (P ≥ 0.05), * 0.005 ≤ P < 0.05, ** P < 0.005; P values calculated by two-tailed Welch’s t-tests.