Adaptation by Type V-A and V-B CRISPR-Cas Systems Demonstrates Conserved Protospacer Selection Mechanisms Between Diverse CRISPR-Cas Types

Abstract

Adaptation of clustered regularly interspaced short palindromic repeats (CRISPR) arrays is a crucial process responsible for the unique, adaptive nature of CRISPR-Cas immune systems. The acquisition of new CRISPR spacers from mobile genetic elements has previously been studied for several types of CRISPR-Cas systems. In this study, we used a high-throughput sequencing approach to characterize CRISPR adaptation of the type V-A system from Francisella novicida and the type V-B system from Alicyclobacillus acidoterrestris. In contrast to other class 2 CRISPR-Cas systems, we found that for the type V-A and V-B systems, the Cas12 nucleases are dispensable for spacer acquisition, with only Cas1 and Cas2 (type V-A) or Cas4/1 and Cas2 (type V-B) being necessary and sufficient. Whereas the catalytic activity of Cas4 is not essential for adaptation, Cas4 activity is required for correct protospacer adjacent motif selection in both systems and for prespacer trimming in type V-A. In addition, we provide evidence for acquisition of RecBCD-produced DNA fragments by both systems, but with spacers derived from foreign DNA being incorporated preferentially over those derived from the host chromosome. Our work shows that several spacer acquisition mechanisms are conserved between diverse CRISPR-Cas systems, but also highlights unexpected nuances between similar systems that generally contribute to a bias of gaining immunity against invading genetic elements.

FIG. 1.

Cas4 and Cas12 are not required for adaptation of type V-A and V-B systems. (A) Schematic of the CRISPR-Cas loci for the type II-A, V-A, and V-B systems. The CRISPR arrays (enlarged for clarity) consist of repeats (red diamond) and spacers (green). (B) Workflow schematic of the adaptation assay using the three-plasmid system (pAdaptation, pEffector, and pTarget) in Escherichia coli. l-Arabinose and IPTG are added to induce expression of cas genes. Cells are grown for 48 h in a medium with selection for pAdaptation and pEffector, but not pTarget, and subsequently used in a population PCR to detect new spacers. The forward PCR primer matches the repeat plus a single 3′ nucleotide that mismatches with the first base of the existing spacer, allowing preferential amplification of expanded (+1) compared with unexpanded (+0) arrays. Amplified CRISPR arrays are visualized on an agarose gel electrophoresis. (C) Population PCR of cells expressing type V-A or V-B cas genes (and variations thereof). CRISPR-arrays were amplified and visualized by agarose gel electrophoresis. Plasmid variants are indicated above the gel. pAdaptation WT = Cas4, Cas1, and Cas2 (V-A) or Cas4/1 and Cas2 (V-B), Δ2 = ΔCas2, Δ4 = ΔCas4, 4^mut = mutated Cas4. Cas4^I-G = wild-type Cas4 domain swapped with Cas4 domain of type I-G from Geobacter sulfurreducens. pEffector WT = Cas12a/b, ΔWT = ΔCas12a/b, RuvC = catalytically inactive Cas12a/b, PI = Cas12a/b mutated in the PI domain. pTarget: Naive = pNaive, Targ = pTargeted (with protospacer and PAM), Prim = pPriming (protospacer containing a mismatch in the seed position 1). Amplicons from expanded arrays containing one new spacer (+1 arrays) are indicated by a black triangle. In addition to the +1 band found in V-B, an approximately +1/2 band was also observed, which sequencing revealed to be a PCR artifact. Heatmaps below each gel indicate the mean number of new (total) spacers detected by high-throughput sequencing of the PCR amplicons. Note, that the PCR approach used to detect spacers dictates that comparisons between samples are predominantly qualitative, whereas quantitative analyses of spacer characteristics can be performed within sample data sets. These heatmaps are intended to illustrate sufficient read depth for further analyses. All experiments were performed using three biological replicates: additional data are displayed in Supplementary Figure S3. CRISPR, clustered regularly interspaced short palindromic repeats; IPTG, isopropyl β-d-1-thiogalactopyranoside; PAM, protospacer adjacent motif; PCR, polymerase chain reaction; PI, PAM-interacting; WT, wild type.

FIG. 2.

Cas4 is required for PAM selection in type V-A and V-B systems and spacer trimming in type V-A. (A) The relative proportions of spacers acquired from each of the plasmids and Escherichia coli genome. Data represent the mean of three replicates. The pAdaptation, pEffector, and the E. coli genome contain identical copies of lacI, which resulted in some ambiguous protospacers whose origins we could not differentiate (gray). Cas4-1-2 = Cas4, Cas1, and Cas2 (V-A) or Cas4/1 and Cas2 (V-B). Cas4^mut is Cas4-1-2 containing a catalytically inactive Cas4. ΔCas4 = Cas1 and Cas2 (V-A). Cas4^I-G = Cas4 domain of type I-G fused to Cas1 and Cas2 (V-B). +Cas12 = Cas4-1-2 and Cas12a/b. In conditions lacking Cas12, an empty plasmid or backbone was used instead. (B) The bias toward adaptation from foreign elements (pTarget) versus the host genome (E. coli) differed between the WT type V-A and V-B samples. Data represent the mean of three replicates, and statistical significance was tested using an unpaired two-sided t-test. ***p < 0.001 or *p < 0.05. (C) The locations of spacers acquired from each of the three plasmids present in the WT (black lines) and Cas4 deletion or mutant samples (magenta lines). Data were smoothed with a sliding window with a width of 250 bp and represent the mean (solid lines) ± SEM (shaded) for three replicates. Protospacers mapping to the forward or reverse strands are plotted above or below the x axis, respectively. The lacI regions (dashed boxes) that are shared between pAdaptation, pEffector, and the E. coli genome were excluded from the mapping analysis. (D) Sequence motif preferences for the 5′-PAM positions −6 to −1 for type V-A and V-B, based on unique spacers; the consensus interference-proficient PAMs for these systems are TTTV and NTTN, respectively.^, The enrichment and depletion scores were generated using EDLogo. (E) Histogram of unique spacer lengths for type V-A and V-B. Error bars represent mean ± SEM and were calculated using biological replicates (n = 3). SEM, standard error of the mean.

FIG. 3.

No primed adaptation in type V-A or V-B systems. (A) The relative proportions of spacers acquired from each of the plasmids and Escherichia coli genome. The data represent the mean of three replicates. pAdaptation, pEffector, and the E. coli chromosome contain identical copies of lacI, which resulted in some ambiguous protospacers whose origins we could not differentiate. (B) The protospacer mapping distributions to the plasmids in each sample. Data were smoothed with a 250 bp sliding window and represent the mean (solid lines) ± SEM (shaded) for either two or three replicates; for some samples with low numbers of reads, the normalized mapping for one of the three replicates contained outlying data points, and so, in these cases, where the other two replicates were in close agreement, the outlying replicate was excluded from the analyses. Samples contained the WT pAdaptation and pEffector (+Cas12) plasmids with different versions of pTarget: pNaive (black), pTargeted (magenta), or pPriming (cyan). Protospacers mapping to the forward or reverse strands are plotted above or below the x axis, respectively. The lacI regions (dashed boxes) that are shared between pAdaptation, pEffector, and the E. coli genome were excluded from the mapping analysis.

FIG. 4.

Naive adaptation is biased toward locations of RecBCD activity. (A) Spacers acquired from the host chromosome for the WT Cas4-1-2 samples. The Escherichia coli genome is displayed linearized starting at the OriC, and spacers mapping to lacI were excluded. Data represent the median for six replicates and are smoothed with a rolling-sum window of 10 kb, and then normalized to the total number of spacers in each sample. (B) The distribution of protospacers around Chi sites in the E. coli genome, with the sum of all spacers acquired across six replicates binned into 1 kb sections, as indicated. Protospacers and Chi sites within the genome terminus region (2.0–2.6 Mb) were excluded from this analysis. Note, although the total number of spacers detected was higher for type V-B than type V-A (due to differences in sequencing depth), the type V-A system acquired a higher proportion of all spacers from E. coli, relative to other sources, than type V-B (Fig. 2A).