Diverse viral cas genes antagonize CRISPR immunity

Abstract

Prokaryotic CRISPR-Cas immunity is subverted by anti-CRISPRs (Acrs), which inhibit Cas protein activities when expressed during the phage lytic cycle or from resident prophages or plasmids¹. Acrs often bind to specific cognate Cas proteins, and hence inhibition is typically limited to a single CRISPR-Cas subtype². Furthermore, although acr genes are frequently organized together in phage-associated gene clusters³, how such inhibitors initially evolve has remained unclear. Here we investigated the Acr content and inhibition specificity of diverse Listeria isolates, which naturally harbour four CRISPR-Cas systems (types I-B, II-A, II-C and VI-A). We observed widespread antagonism of CRISPR, which we traced to 11 previously unknown and 4 known acr gene families encoded by endogenous mobile elements. Among these were two Acrs that possess sequence homology to type I-B Cas proteins, one of which assembles into a defective interference complex. Surprisingly, an additional type I-B Cas homologue did not affect type I immunity, but instead inhibited the RNA-targeting type VI CRISPR system by means of CRISPR RNA (crRNA) degradation. By probing viral sequence databases, we detected abundant orphan cas genes located within putative anti-defence gene clusters. Among them, we verified the activity of a particularly broad-spectrum cas3 homologue that inhibits type I-B, II-A and VI-A CRISPR immunity. Our observations provide direct evidence of Acr evolution by cas gene co-option, and new genes with potential for broad-spectrum control of genome editing technologies.

Extended Data Fig. 1 |. Variation in L. seeligeri strain background affects type I-B CRISPR-Cas immunity.

Plasmid targeting assay in which the indicated L. seeligeri strains were first transformed with a chromosomally integrated type I-B CRISPR-Cas system equipped with a spacer targeting a conjugative plasmid, then challenged with either a non-target plasmid (left columns) or plasmid containing a target protospacer (right columns).

Extended Data Fig. 2 |. Variation in L. seeligeri strain background affects type II-A CRISPR-Cas immunity.

Plasmid targeting assay in which the indicated L. seeligeri strains were first transformed with a chromosomally integrated type II-A CRISPR-Cas system equipped with a spacer targeting a conjugative plasmid, then challenged with either a non-target plasmid (left columns) or plasmid containing a target protospacer (right columns).

Extended Data Fig. 3 |. Variation in L. seeligeri strain background affects type II-C CRISPR-Cas immunity.

Plasmid targeting assay in which the indicated L. seeligeri strains were first transformed with a chromosomally integrated type II-C CRISPR-Cas system equipped with a spacer targeting a conjugative plasmid, then challenged with either a non-target plasmid (left columns) or plasmid containing a target protospacer (right columns).

Extended Data Fig. 4 |. Variation in L. seeligeri strain background affects type VI-A CRISPR-Cas immunity.

Plasmid targeting assay in which the indicated L. seeligeri strains were first transformed with a chromosomally integrated type VI-A CRISPR-Cas system equipped with a spacer targeting a conjugative plasmid, then challenged with either a non-target plasmid (left columns) or plasmid containing a target protospacer (right columns).

Extended Data Fig. 5 |. Anti-defense candidate (adc) gene occurrence across 62 strains of L. seeligeri.

Each row corresponds to either a known anti-CRISPR gene or a particular anti-defense candidate gene identified as frequently encoded nearby acr genes or nearby other well-established anti-defense candidates. Each column corresponds to an individual L. seeligeri strain genome. Filled red boxes indicate occurrence of a putative anti-defense gene in a particular strain.

Extended Data Fig. 6 |. Anti-defense candidate (adc) gene occurrence among L. seeligeri strains that inhibit (or tolerate) type I-B CRISPR immunity.

Each row corresponds to either a known anti-CRISPR gene or a particular anti-defense candidate gene identified as frequently encoded nearby acr genes or nearby other well-established anti-defense candidates. Each column corresponds to an individual L. seeligeri strain genome. The group of columns on the left indicate strains that inhibited type I-B CRISPR immunity in the plasmid targeting assay shown in Fig. 1, while the group on the right tolerated type I-B immunity. Filled red boxes indicate occurrence of a putative anti-defense gene in a particular strain. Gene names in red indicate experimentally validated type I-B Acrs from this study.

Extended Data Fig. 7 |. Anti-defense candidate (adc) gene occurrence among L. seeligeri strains that inhibit (or tolerate) type II-A CRISPR immunity.

Each row corresponds to either a known anti-CRISPR gene or a particular anti-defense candidate gene identified as frequently encoded nearby acr genes or nearby other well-established anti-defense candidates. Each column corresponds to an individual L. seeligeri strain genome. The group of columns on the left indicate strains that inhibited type II-A CRISPR immunity in the plasmid targeting assay shown in Fig. 1, while the group on the right tolerated type II-A immunity. Filled red boxes indicate occurrence of a putative anti-defense gene in a particular strain. Gene names in red indicate experimentally validated type II-A Acrs.

Extended Data Fig. 8 |. Anti-defense candidate (adc) gene occurrence among L. seeligeri strains that inhibit (or tolerate) type II-C CRISPR immunity.

(a) Each row corresponds to either a known anti-CRISPR gene or a particular anti-defense candidate gene identified as frequently encoded nearby acr genes or nearby other well-established anti-defense candidates. Each column corresponds to an individual L. seeligeri strain genome. The group of columns on the left indicate strains that inhibited type II-C CRISPR immunity in the plasmid targeting assay shown in Fig. 1, while the group on the right tolerated type II-C immunity. Filled red boxes indicate occurrence of a putative anti-defense gene in a particular strain. Gene names in red indicate experimentally validated type II-C Acrs from this study. (b) Alignment of AcrIIA3 from L. seeligeri, which inhibited type II-C but not type II-A CRISPR in our study, with AcrIIA3 from L. monocytogenes, which has been shown to inhibit type II-A CRISPR (Rauch et al.).

Extended Data Fig. 9 |. Anti-defense candidate (adc) gene occurrence among L. seeligeri strains that inhibit (or tolerate) type VI-A CRISPR immunity.

Each row corresponds to either a known anti-CRISPR gene or a particular anti-defense candidate gene identified as frequently encoded nearby acr genes or nearby other well-established anti-defense candidates. Each column corresponds to an individual L. seeligeri strain genome. The group of columns on the left indicate strains that inhibited type VI-A CRISPR immunity in the plasmid targeting assay shown in Fig. 1, while the group on the right tolerated type VI-A immunity. Filled red boxes indicate occurrence of a putative anti-defense gene in a particular strain. Gene names in red indicate experimentally validated type VI-A Acrs from this study.

Extended Data Fig. 10 |. Anti-Defense Candidates with no detectable anti-CRISPR activity.

Each anti-defense candidate was tested for inhibition of plasmid targeting by the indicated CRISPR type in L. seeligeri strain LS1. Adcs were first introduced into LS1 carrying the indicated CRISPR type, then a targeted conjugative plasmid was introduced, and the cells were plated on media selecting for the target plasmid. *We previously tested adc31 against the L. seeligeri type VI-A CRISPR system in Meeske & Jia et al. Science 2020 (Fig. 1c, pgp1), no inhibition was detected. nd, not determined. Each image is a representative of three biological replicates.

Extended Data Fig. 11 |. Predicted structural differences between AcrIB3 and Cas5 are essential for Acr function.

(a) Quantitative beta-galactosidase assay corresponding to strains tested in Fig. 3d. (b) Cas6-3xFlag and Cas5-his12 function in immunity against a plasmid containing a protospacer recognized by the type I-B CRISPR system. (c) His6-AcrIB3 functions in inhibition of type I-B CRISPR immunity in the plasmid targeting assay. (d) AlphaFold2 structural model of L. seeligeri Cas5 (pink) superimposed onto structure of Cas5 from Pyrococcus furiosus type I-A Cascade (cyan). Structure from Hu et al. 2022 Mol. Cell. RMSD, root mean squared deviation. (e) Overlay of L. seeligeri Cas5 (pink) and AcrIB3 (orange) structural predictions. Two key structural distinctions are highlighted: a “hook” in Cas5 that is missing from AcrIB3, and an extended loop specific to AcrIB3. (f) L. seeligeri Cas5 (pink) modeled into the P. furiosus Cascade complex, with Cas8 shown (cyan). Cas5 hook is predicted to contact Cas8. (g) Same as (f), but with AcrIB3 (orange) instead of Cas5. (h) L. seeligeri Cas5 (pink) modeled into the P. furiosus Cascade complex, with nearest Cas7 protomer shown (green). (i) Same as (h), but with AcrIB3 (orange) instead of Cas5. Extended AcrIB3 loop is predicted to contact Cas7. (j) Plasmid targeting assay demonstrating that restoration of the Cas5 hook in AcrIB3 and deletion of the AcrIB3 loop extension abolishes Acr activity. Representative example of three biological replicates is shown.

Extended Data Fig. 12 |. No detectable interaction between AcrVIA2 and Cas13, and lack of AcrVIA2 activity on pre-formed RNP complexes.

(a) AcrVIA2-3xFlag is partially functional in inhibition of immunity against a plasmid expressing an RNA protospacer recognized by the type VI-A CRISPR system. (b) No detectable co-immunoprecipitation of Cas13-his6 and AcrVIA2-3xFlag. The housekeeping sigma factor σA is shown as a non-interacting control. L, load, UB, unbound, IP, immunoprecipitate. Molecular weight in kDa. (c) Anti-Flag immunoblots of 3xFlag tagged WT AcrVIA2 and DEFD > AAFD mutant allele. Molecular weight in kDa. (d) Northern blot analysis of RNA co-immunoprecipitated with Cas13-his6 using crRNA-specific probe. Molecular weight in nucleotides. (e) No effect of AcrVIA2 on crRNAs from pre-formed Cas13:crRNA RNP complexes. Lysates expressing dCas13-his6 were incubated with lysates expressing either no Acr, AcrVIA2, or AcrVIA2.1 for 1 hr at 30 °C in the presence of 1 mM ATP. Cas13 was then immunoprecipitated, and Cas13-associated RNAs were extracted, separated by denaturing PAGE, and analyzed by SYBR Gold staining. Molecular weight in nucleotides.

Fig. 1 |. Variation in L. seeligeri genomes affects CRISPR–Cas function.

a, Schematic of mobilizable chromosomally integrating CRISPR–Cas loci, each equipped with a single plasmid-targeting spacer. b, Plasmid-targeting assay demonstrating sequence-specific interference by all four CRISPR–Cas types in strain LS1. c, Schematic of strategy to detect activity of endogenous Acrs by introducing CRISPR–Cas loci into diverse strain backgrounds and challenging them with target plasmid. d, Functionality of four transplanted CRISPR types across 62 L. seeligeri strains. e, Natural occurrence of CRISPR types across the strain collection.

Fig. 2 |. Identification of 11 type I-B, II-C and VI-A anti-CRISPR families.

a, Schematic of strategy to test Acr candidates. Acrs were expressed from a plasmid and introduced into strain LS1 harbouring an active CRISPR–Cas system, then challenged with a target plasmid. b, Inhibition spectrum of tested Acrs. Each Acr candidate was tested against all four CRISPR–Cas systems in a plasmid-targeting assay. c, Quantification of Acr activity from three biological replicates of plasmid-targeting assay. T/NT, ratio of transconjugants with target plasmid to those with non-target plasmid. d, Genetic loci encoding known and previously unknown Acrs. Known Acrs are shown in purple. Anti-defence candidate genes used in prediction of acr loci are shown in yellow. Previously unknown Acrs with activity demonstrated are shown in orange. n represents the number of instances of the indicated gene in the L. seeligeri collection. e, Acrs discovered in this study. Predicted protein domains noted, with HTH domains depicted in black. MGE, mobile genetic element; φ, phage; aa, amino acids.

Fig. 3 |. AcrIB3 and AcrIB4 are Cas protein homologues that inhibit type I-B CRISPR immunity.

a, Schematic of genetic loci encoding AcrIB3 and AcrIB4 (opaque orange) and type I-B CRISPR–Cas locus. Percentage sequence identity between AcrIB3 and Cas5, and AcrIB4 and the CTD of Cas8b, is noted. b, Plasmid-targeting assay demonstrating that expression of AcrIB3 and AcrIB4, but not their cognate Cas proteins Cas5 and Cas8b, inhibits type I-B CRISPR immunity, and quantitation from three biological replicates. c, Predicted phylogeny of AcrIB3 homologues. Black circles indicate nodes with greater than 80% bootstrap support. The orange circle indicates Acr characterized experimentally. The scale bar indicates branch length (arbitrary units (a.u.)). d, CRISPRi lacZ silencing assay using a nuclease-deficient CRISPR system, demonstrating that both AcrIB3 and AcrIB4 inhibit target DNA recognition by Cascade. e, Silver stain analysis of Cas6–3×Flag (or untagged) immunoprecipitate fractions in the presence or absence of Acrs. Molecular weight marker, kDa. f, Co-immunoprecipitation of His6–AcrIB3 and Cas6–3×Flag. The housekeeping sigma factor σ^A is shown as a non-interacting control. g, Co-immunoprecipitation of Cas6–3×Flag with Cas5–His12 is abolished in the presence of AcrIB3. *Nonspecific Flag-reactive band. L, load; UB, unbound; IP, immunoprecipitate; nt, not-targeting. Scale bar, 0.10 a.u.

Fig. 4 |. AcrVIA2 is a Cas3 homologue that inhibits type VI-A CRISPR immunity.

a, Schematic of genetic loci encoding AcrVIA2 (opaque orange) and type I-B CRISPR–Cas locus. Percentage sequence identity between AcrVIA2 and Cas3 is noted. b, Predicted phylogeny of AcrVIA2 homologues. The black circle indicates a node with 100% bootstrap support. The orange circle indicates experimentally characterized Acr. The scale bar indicates branch length (a.u.). c, Plasmid-targeting assay demonstrating that expression of AcrVIA2, but not Cas3 or an AcrVIA2 DEAD-box mutant allele, inhibits type VI-A CRISPR immunity. n = 3 biological replicates. d, Plaque assay demonstrating that AcrVIA2 inhibits type VI-A immunity against a phage target. nt represents non-targeting and spc59 represents a spacer targeting an early lytic transcript of ɸLS59. The image is representative of three biological replicates. e, AcrVIA2 inhibits trans-RNase activity of Cas13 in vivo. Strain LS1 harbouring an aTc-inducible type VI-A protospacer, plus AcrVIA2, was plated on medium with or without aTc, as indicated. The image is representative of three biological replicates. f, The effect of AcrVIA2 on crRNA associated with affinity-purified Cas13–His6. Molecular weight in nucleotides. g, Northern blot using crRNA probe on total RNA extracted from cells with or without AcrVIA2. Scale bar, 0.20 a.u.

Fig. 5 |. Diverse viral cas genes reside in putative anti-defence loci.

a, Frequency of orphan viral cas genes found in the IMGVR database, organized by Cas protein query and predicted viral host phylum. Cas queries are coloured by CRISPR type (green, type I; blue, type II; pink, type III; yellow, type IV; purple, type VI). b, Example locus schematics for viral cas genes found in the vicinity of known anti-CRISPRs or other predicted anti-defence genes. c, Schematic showing percentage amino acid identity between L. seeligeri Cas3, the indicated Acrs and each other, along with protein lengths. d, Plasmid-targeting assay demonstrating the CRISPR inhibition spectrum of viral Cas3 protein and dependence on DEAD box. e, Quantitation of plasmid-targeting assay (three biological replicates). f, Effect of AcrVIA2.1 on levels of crRNA affinity-purified with Cas13–His6 or Cas6–3×Flag. Molecular weight in nucleotides.