AcrIIIA1 is a protein–RNA anti-CRISPR complex that targets core Cas and accessory nucleases Authors: Lucy Chou-Zheng, Olivia Howell, Tori A Boyle, Motaher Hossain, Forrest C Walker, Emma K Sheriff, Barbaros Aslan, Asma Hatoum-Aslan Published: 2024 ABSTRACT AbstractClustered regularly-interspaced short palindromic repeats (CRISPRs) and CRISPR-associated (Cas) proteins protect bacteria and archaea from their viruses, and anti-CRISPRs (Acrs) are small virus-encoded proteins that inhibit CRISPR-Cas immunity. Over 80 families of Acrs have been described to date; however, only three of these subvert Type III CRISPR-Cas immunity. Type III systems employ a complex network of Cas and accessory nucleases to degrade viral nucleic acids. Here, we discover and characterize AcrIIIA1, the first Type III-A specific anti-CRISPR protein. We demonstrate that AcrIIIA1 binds to Csm2 within the Cas10-Csm effector complex and attenuates Cas10’s DNase activity and second messenger production. Additionally, AcrIIIA1 associates with fragmented t(m)RNAs (acrIIIA1-RNAs), and we show that they co-purify with the Cas10-Csm complex during phage infection. Although the precise role(s) of acrIIIA1-RNAs remain unclear, we found that they bind stably to RNase R, a host-encoded nuclease known to bolster immunity, and RNase R has the capacity to degrade them. Altogether, our results support a model in which AcrIIIA1 and its associated RNAs target both core Cas and accessory nucleases to provide robust protection against Type III CRISPR-Cas immunity. Introduction The co-evolutionary arms race between bacteria and their viruses (phages) has given rise to a vast array of immune systems and counter-defense mechanisms that phages have evolved to overcome them. Over 40% of bacteria and nearly all archaea employ clustered regularly-interspaced short palindromic repeats (CRISPRs) and CRISPR-associated (Cas) proteins to protect against their viruses (1). CRISPR-Cas systems maintain memories of past infections by capturing short stretches (∼30–40 nucleotides) of viral DNA and integrating them into the CRISPR locus as ‘spacers’ in between DNA repeats with similar lengths (2,3). Spacer sequences are transcribed and processed into small CRISPR RNAs (crRNAs), which combine with one or more Cas protein(s) to form an effector complex. Effector complexes surveil the interior of the cell and degrade foreign nucleic acids bearing ‘protospacer’ sequences that are complementary to crRNAs. CRISPR-Cas systems are tremendously diverse, and have been grouped into two classes, six types and dozens of subtypes based upon their effector complex architectures, cas gene content, and mechanistic details (4). In response to the selective pressure imposed by CRISPR-Cas systems, the viruses of bacteria and archaea have evolved a variety of strategies to overcome immunity. Prime among these are anti-CRISPR proteins (Acrs). Acrs are a diverse group of small proteins (∼80–150 amino acids) that typically work by binding Cas effector complexes and blocking their functions (5). Over 80 families of Acrs targeting five of the six CRISPR-Cas Types have been described thus far (6); however, only three of these have the capacity to overcome Type III immunity (7–10). Type III CRISPR-Cas systems are found in bacteria and archaea and are regarded as the most complex (4,11). Of the six subtypes currently identified (A–F), Types III-A and III-B have historically garnered the most attention. These systems employ multi-subunit effector complexes and a variety of accessory/cellular nucleases to detect and destroy invading RNA and DNA. Upon crRNA binding to a complementary transcript, members of the core Cas effector complex perform at least three catalytic activities: Cas7/Csm3/Cmr4 shred target RNA (12–16), Cas10 degrades single-stranded DNA indiscriminately (15,17–20), and Cas10 also produces cyclic-oligoadenylates (cOAs), second-messenger molecules that bind and stimulate accessory nucleases often bearing CRISPR-associated Rossman Fold (CARF) domains (21–25). These include Csm6/Csx1, which are not a part of the core effector complex, but are encoded within or proximal to the CRISPR locus (26). Beyond these Cas accessory enzymes, Type III systems co-opt diverse cellular ‘housekeeping’ nucleases to facilitate crRNA processing and clear phage nucleic acids (27–30). The elaborate network of accessory nucleases in the Type III immune response presents unique challenges for Type III Acrs and implies that targeting the effector complex alone might not be sufficient to completely protect against immunity. Indeed, the first Type III anti-CRISPR described, AcrIIIB1 binds the Cas effector complex and inhibits its function, but it rescues viruses only when late-expressed genes are targeted by the CRISPR system (7). Late gene targeting results in a delayed immune response that gives viruses a window of opportunity to begin replicating undetected. This ‘head start’ in viral replication, coupled with the inhibitory activity of AcrIIIB1, allows viruses to complete their life cycle and escape immunity. In addition, while this manuscript was under preparation, AcrIIIB2 was reported (9,10). AcrIIIB2 binds the Type III-B effector complex and diminishes all three of its catalytic activities by reducing the disassociation rate of cleaved target RNA. Although plasmid-encoded AcrIIIB2 is effective when targeting both early- and late-expressed viral genes (9), a single genomic copy of AcrIIIB2 protects viruses only when late genes are targeted (10). In contrast to these Type III-B specific Acrs, AcrIII-1 does not bind/inhibit effector complexes, but rather it degrades cyclic tetra-adenylates (cOA4), and thus interferes with the downstream activation of CARF-containing accessory nucleases that rely upon cOA4 to propagate the immune response (8). The effectiveness of AcrIII-1 during early- versus late-gene targeting has not been explored; however, since the cOA-responsive accessory nucleases are essential particularly when late-expressed genes are targeted (31), it is likely that AcrIII-1 is also more effective when the target is located in late-expressed genes. To date, Type III-A specific Acrs have not been described. Here, we report the discovery and characterization of AcrIIIA1, the first Type III-A anti-CRISPR protein. We identified AcrIIIA1 in a subset of phages that infect Staphylococcus epidermidis RP62a. Genetic analyses revealed that native expression of acrIIIA1 completely overcomes CRISPR immunity whether early- or late-expressed phage genes are targeted. Biochemical analyses showed that AcrIIIA1 co-purifies with fragmented t(m)RNAs (herein referred to as acrIIIA1-RNAs), and both RNA-bound and RNA-free fractions associate with the effector complex and attenuate its DNase activity and cOA production. Moreover, we found that acrIIIA1-RNAs co-purify with effector complexes in the native host during phage infection, and we demonstrate in a purified system that acrIIIA1-RNAs specifically interact with RNase R, a cellular ‘housekeeping’ nuclease that promotes robust immunity (30). Altogether, our results support a model in which AcrIIIA1 uses a unique multi-layered strategy to mitigate indiscriminate nucleic acid cleavage on multiple fronts in order to provide robust protection against Type III CRISPR immunity. Materials and methods Bacterial strains and growth conditions S. epidermidis RP62a (32), Δcrispr (known as LAM104 in (33)) and Δcrispr-cas (known as LM1680 in (34)) were generous gifts from Luciano Marraffini. S. epidermidis strains and Staphylococcus aureus RN4220 (35) were grown in Brain Heart Infusion (BHI, BD Diagnostics) and Tryptic Soy Broth (TSB, BD Diagnostics), respectively. Escherichia coli DH5α was grown in Luria Bertani (LB) broth (VWR), and E. coli Rosetta 2 (BL21(DE3)) was grown in Terrific Broth (TB, VWR). These media were supplemented with the following antibiotics: 10 μg/ml chloramphenicol (to select for pC194-based plasmids), 10 μg/ml erythromycin (to select for pTet-based plasmids), 15 μg/ml anhydrotetracycline (aTc; to activate the inducible promoter), 15 μg/ml neomycin (to select for S. epidermidis), 30 μg/ml chloramphenicol (to select for E. coli Rosetta2 plasmids) and 50 μg/ml kanamycin (to select for pET28b-10HisSmt3-based plasmids). All bacterial strains were grown at 37°C, except for protein induction at 17°C. Liquid cultures were propagated with constant agitation in an orbital shaker set to 180–200 rpm. Phage propagation and enumeration S. epidermidis phage PhilBB (Genbank: NC_041928.1) was a generous gift from Luís Melo (36). PhilBB, Terranova (Genbank: MH542234.1), Twillingate (Genbank: MH321491.1) and Quidividi (Genbank: MH321490.1) (37) were propagated on S. epidermidis RP62a Δcrispr-cas. Concentrated phage stocks were prepared by picking 1–5 purified phage plaques and resuspending them into 500 μl of TSB by vortexing for 30 s. Suspensions were then subjected to centrifugation at ∼15 000 × g for 2 min to pellet agar and cells. The resulting supernatant was passed through a 0.45 μm syringe filter to create a low-titer phage lysate. Low-titer lysates (300 μl) were combined with overnight host culture (diluted 1:100) in 7 ml of Heart Infusion Agar (HIA, Hardy Diagnostics, prepared at 0.3 × concentration) supplemented with 5 mM CaCl2. Phage-host mixtures were poured atop plates containing a solid layer of Tryptic Soy Agar (TSA) supplemented with 5 mM CaCl2. The top agar layer was allowed to solidify ∼10 min at room temperature and then incubated overnight at 37 °C. The following day, the entire top agar layer was harvested with a sterile cell scraper and resuspended in 20 ml fresh TSB. Phages were released into suspension by vortexing for 5 min. Suspensions were then pelleted via centrifugation at 10 000 ×g for 10 min, and the supernatant was passed through a 0.45 μm bottle filter to obtain high concentrations of each phage. Phage concentrations were determined using the double-agar overlay method as described in (38). Briefly, overnight cultures of host bacteria were diluted 1:100 in molten HIA (prepared at 0.5 × concentration) supplemented with 5 mM CaCl2, and the agar was poured atop a plate containing solidified TSA supplemented with 5 mM CaCl2. The top layer was allowed to solidify at room temperature for 10 min, and 10-fold dilutions (100–10−7) of concentrated phage lysate were prepared. Approximately 5 μl spots of each dilution were dropped atop the agar, spots were allowed to air dry, and plates were incubated overnight at 37°C. The phage concentration (i.e. titer) in plaque-forming units per ml (pfu/ml) was determined by the following formula: [(number of plaques counted on the most diluted spot)/(dilution factor)] × 200. Phage stocks were stored at 4°C. Phage identities were routinely confirmed/authenticated via polymerase chain reaction (PCR) amplification and sequencing of genomic regions unique to each phage. Phage challenge assays For all phage challenge plate assays, concentrated phage stocks were diluted in 10-fold increments (100–10−7), and dilutions were spotted atop lawns of cells. Following overnight incubation, plaques were enumerated using the protocol described in the section above titled ‘phage propagation and enumeration.’ For assays using the anhydrotetracycline- (aTc-) inducible system, plates and top agar were supplemented with 15 μg/ml aTc. For the phage challenge assays in liquid culture, overnight cultures were diluted 1:100 in fresh BHI broth, and phages were added to reach multiplicities of infection (MOIs) (i.e. phage:cell ratios) of 0.1, 0.5 or 1.0. Phage-host mixtures (200 μl) were then aliquoted into triplicate wells of a 96-well plate. Cells were grown at 37°C with agitation for 15 h in a SpectraMax M2e microplate/cuvette reader (Molecular Devices), and OD600 (optical density at 600 nm) measurements were taken every 15 min. For all phage challenge assays, graphs show an average of triplicate measurements (±S.D.) as a representative of at least two independent trials. Constructing pcrispr-spcϕ-, pcrispr-cas-, pcrispr-cas-spcI- and pTet- plasmids All plasmids were constructed using either inverse PCR or Gibson Assembly (39) with the primers and templates listed in Supplementary Table S1. pAH011 (40), pcrispr-cas (41) and pcrispr-cas-spcI (42), all of which are derivatives of pC194 (43), were used as the backbones for plasmids designated as pcrispr-spcϕ-, pcrispr-cas-csm6H6C and pcrispr-cas-spcI-, respectively, in this study. pTarget (15), a derivative of pE194 (44), was used as the backbone for all plasmids designated as pTet- in this study. All assembled plasmids were first introduced into S. aureus RN4220 via electroporation (described in the section below), and inserted sequences were confirmed by PCR amplification and Sanger sequencing (performed by Eurofins MWG Operon). At least two transformants were confirmed by sequencing and at least one of each construct was purified using the EZNA Plasmid Mini Kit (Omega Bio-tek). Where indicated, plasmids were introduced into S. epidermidis RP62a Δcrispr or Δcrispr-cas via electroporation. Electroporation into staphylococci Electrocompetent cells were prepared as described in (45). Briefly, fresh media was used to dilute 10 ml of overnight culture until OD600 of 0.5. Diluted cultures were incubated at 37°C for 30 min, then placed on ice for 10 min. All subsequent steps were performed on ice. Cells were pelleted at 4000 × g for 10 min and then washed with an equal volume ice-cold water two times. Cell pellets were washed twice more with 10% glycerol (ice-cold) using 1/20 and 1/25 of the culture volume, respectively. Cells were finally resuspended in 10% glycerol (at 1/200 the volume of initial culture) and stored at −80 °C in 50 μl aliquots. For electroporation, constructs/plasmids were dialyzed against sterile water using the drop dialysis method on a 0.022 μm filter for 20 min. At the same time, competent cells were thawed on ice for 5 min, and then left at room temperature for another 5 min. Cells were then pelleted for 1 min at 5 000 × g. Pellets were resuspended in 50 μl of sterilized 10% glycerol supplemented with 500 mM sucrose. Then the entire volume of dialyzed construct/plasmid was added into the cell suspension. Mixtures were transferred into a 2 mm electroporation cuvette (VWR) and pulsed at 29 kV/cm, 100 W and 25 mF with a GenePulser Xcell (Bio-Rad). Cells were then placed in 1 ml of sterile TSB containing 500 mM sucrose and allowed to recover at 37°C with agitation for at least 1 h (S. aureus) or 2 h (S. epidermidis). Next, 200 μl recovered cells were plated on TSA or BHI agar supplemented with appropriate antibiotics. Plates were incubated at 37°C overnight and transformants were recovered on the following day. Phage hybridization and genotyping Fresh overnight culture of S. epidermidis RP62a Δcrispr-cas was diluted 1:100 in BHI (2 ml) supplemented with 5 mM CaCl2. Cells were grown to an OD600 of 0.5 and then infected with two phages: Twillingate at an MOI of 0.05 and PhilBB at a MOI of 8. The bacteria : phage mixtures were incubated in a 37°C water bath for 5 min (without agitation) to allow phage absorption. Then, infected cells were collected via centrifugation at 17 000 × g for 5 min. The pelleted cells were washed with 1 ml of fresh BHI and centrifuged again at 17 000 × g for 5 min to remove unabsorbed phages. The final pellet was resuspended in 150 μl fresh BHI. Ten-fold dilutions (out to 10−5) of the cell suspension were spotted atop a semisolid layer of 0.5 × HIA agar containing 5 mM CaCl2 and a 1:100 dilution of overnight culture of S. epidermidis RP62a Δcrispr bearing pcrispr-spcϕ-polA. Spots were air-dried, and plates were incubated overnight at 37°C. The following day, ten well-isolated plaques, now referred to as putative hybrid phages, were picked from plates and placed into separate tubes containing 0.5 ml fresh BHI. The tubes were vortexed, centrifuged and filtered as described in ‘Phage propagation and enumeration’ to separate the phages from the agar. Ten-fold dilutions (100–10−7) of the supernatant were plated atop a semisolid HIA agar layer containing 5 mM CaCl2 and a 1:100 dilution of overnight S. epidermidis RP62a Δcrispr cells bearing pcrispr-spcϕ-polA. This process was repeated three times to obtain pure phages. These putative hybrids were grown to high titer according to the protocol described in ‘Phage propagation and enumeration.’ Genomic DNA was extracted from the high-titer lysates by combining 300 μl with an equal volume of phenol:CHCl3:isoamyl alcohol, vortexing for 15 s and subjecting to centrifugation at 10 000 × g for 2 min. The upper aqueous phase was removed into a fresh tube and DNA was ethanol precipitated and resuspended into nuclease-free water. For phage genotyping, forward and reverse primer sets were designed to anneal immediately outside the 10 ‘gap’ regions for both phages Twillingate and PhilBB (Supplementary Table S1). PCR products generated from putative recombinants were resolved on a 1% agarose gel and imaged alongside the 1 kb Plus DNA ladder (NEB). Since the ten gap regions in phages PhilBB and Twillingate have different lengths, a determination of which gaps originated from which phage could be made based on PCR product length. Genomic DNA extraction of phage Twill-BB H8 for whole genome sequencing Phage genomic DNA was extracted from a high titer lysate as described in (38). Briefly, phage lysate (20 ml) was digested with DNase I and RNase A (10 μg/ml of each) for 30 min at 37°C. Digested lysate was combined with 10 ml of precipitant solution (30% [wt/vol] polyethylene glycol [PEG] 8000 and 3 M NaCl) and incubated at 4°C overnight to allow phages to precipitate. Precipitated phages were pelleted by centrifugation for 10 min at 10 000 × g and 4 °C. The phage pellet was then resuspended in 250 μl of resuspension buffer (5 mM MgSO4, 10 mM EDTA [pH 8.0]) and incubated with proteinase K (100 μg/ml) at 50°C for 30 min. The phage suspension was then combined with 500 μl of resin contained in the Promega Wizard DNA cleanup kit. The mixture was inverted several times and applied to the mini-column contained within the kit. The resin was then washed with 2 ml of 80% isopropanol and dried by centrifugation for 2 min at 13 000 × g. The DNA was finally eluted from the resin with 100 μl of distilled water preheated to 80°C. Phage DNA sequencing and assembly Library preparation and sequencing were performed at the SeqCenter (Pittsburgh, PA) using the Illumina Library Prep Tagmentation kit. Sequencing was performed on an Illumina NextSeq 2000. Adapters and indexes were removed by bcl2fastq v. 1.8.4 (Illumina), and FastQC v. 0.11.9 was used to confirm data quality. Sequencing reads were assembled using SPAdes v. 3.15.5 (46) in isolate mode with the kmer values of 21, 33, 55, 77 and 99. The resulting assembly graph was inspected using Bandage v. 0.8.1 (47), and the highest coverage contig with the proper length representing the phage genome was extracted as a fasta file. Phage genome sequence alignments with MAUVE Genome sequences of wild-type and/or the newly assembled hybrid phage (TwillBB-H8) were first were re-opened such that position 1 in all phages occurs at homologous loci. Re-opened genomes were then saved as fasta files and multiple genome alignments were generated using MAUVE by downloading the software and following the instructions on the website (https://darlinglab.org/mauve/mauve.html). Generation of phage escapers (E1–E10) and genotyping Ten-fold dilutions (100–10−7) of phage Twillingate were spotted atop a lawn of S. epidermidis RP62a Δcrispr-cas cells bearing pcrispr-spcϕ-gp18. Spots were air-dried, and plates were incubated overnight at 37°C. The following day, ten well-isolated plaques, now referred to as escaper phages (E1–E10), were picked and purified as described for phage hybrids in the section above titled ‘Phage hybridization and genotyping.’ Genomic extracts from escapers were subjected to PCR amplification using primers OH016/OH013 which flank the gap 3 region (Supplementary Table S1). PCR products were resolved on a 1% agarose gel containing ethidium bromide and visualized under UV transillumination using an Azure 400 imager (Azure Biosystems). For samples that gave visible products (E1–E4, E6, E9 and E10), PCR products were purified and subjected to Sanger sequencing to delimit the precise regions that were deleted. Constructing pET28b-10His-Smt3- plasmids pET28b-10His-Smt3-acrIIIA1, pET28b-10His-Smt3-acrIIIA1H63A,H64A and pET28b-10His-Smt3-csm2 were each constructed using a two-piece Gibson assembly with primers and templates listed in Supplementary Table S1. The PCR products derived from the plasmid backbone were treated with 1000 Units of DpnI (NEB) for 1 h at 37°C, followed by heat inactivation at 80°C for 20 min. Then, inserts and backbones were purified using the EZNA Cycle Pure Kit (Omega Bio-tek) and Gibson assembled. Assembled products were introduced into E. coli DH5α by chemical transformation (described in the section below). Four transformants were selected and confirmed via PCR and DNA sequencing with T7P/T7T. Confirmed plasmids were extracted using the EZNA Plasmid DNA Mini Kit (Omega Bio-tek) and transformed into E. coli Rosetta 2 (BL21(DE3)) for protein purification. Transformation of E. coli For preparation of chemically-competent E. coli, overnight cultures were diluted in LB (1:100) and incubated with agitation at 37°C until the OD600 reached ∼0.5. The culture was placed on ice for 10 min and subjected to centrifugation at 4000 ×g for 5 min. The resulting pellet was resuspended in transformation and storage (TSS) buffer (85% LB medium, 10% [w/v] polyethylene glycol [PEG] 8000, 5% [v/v] dimethylsulfoxide [DMSO], 50 mM MgCl2) at 1/10 the starting culture volume. Cells were distributed in 50 μl aliquots and stored at −80°C. For transformation, aliquots were thawed for 10 min on ice and combined with 5 μl Gibson assembled product or 1 μl purified plasmid. The mixture was allowed to cool on ice for 30 min and then heat-shocked at 42°C for 30 s. Cells were immediately placed on ice for 2 min. Approximately 1 ml of fresh LB was added into the tube and cells were allowed to recover at 37 °C for 1 h with agitation. Following recovery, 200 μl cells were plated on LB-agar containing appropriate antibiotics and incubated overnight at 37°C. Purification of AcrIIIA1, Csm2, Csm5, PNPase, RNase R and RNase J2 from pET28b-based plasmids Recombinant AcrIIIA1, Csm2, Csm5, PNPase, RNase R and RNase J2 encoded in pET28b-10His-Smt3- plasmids were overexpressed and purified from E. coli Rosetta 2 (BL21(DE3)) as described previously (27,29,30,48). Briefly, overnight cultures of E. coli Rosetta 2 harboring pET28b-based plasmids were diluted 1:100 in 1 L TB supplemented with appropriate antibiotics. At an OD600 of 0.5–0.6, cells were induced with 0.3 mM isopropyl-1-thio-b-d-galactopyranoside (IPTG) and 2% ethanol, followed by 20 min incubation in an ice-water bath. Cells were then allowed to grow at 17°C with agitation for 16–18 h. Grown cells were harvested and washed with pre-chilled PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4). Final pellets were stored at −80°C until purification. All purification steps were conducted at 4°C. For purification of AcrIIIA1 and AcrIIIA1H63A,H64A, all buffers were set to a pH of 6.5; for purification of Csm5 and Csm2, all buffers were set to a pH of 7.0; and for purification of PNPase, RNase R and RNase J2, all buffers were set to a pH of 7.5. For the first step of purification, each 1 L pellet was suspended in 30 ml of Buffer A (50 mM Tris–HCl, 1.25 M NaCl, 200 mM Li2SO4, 10% sucrose, 25 mM imidazole) containing one complete EDTA-free protease inhibitor tablet (Roche), 0.1 mg/ml lysozyme and 0.1% Triton X-100. After 1 h incubation at 37°C, lysed cells were sonicated on ice (four 30 s pulses with 1 min rest in between). Insoluble materials were removed via centrifugation at 10 000 × g and 13 500 × g for 20 min, respectively, followed by filtration with a 0.22 μm PES (polyethersulfone) membrane. Then, 3 ml of Ni-NTA agarose resin slurry (Thermo Fisher Scientific) was equilibrated with Buffer A and mixed with the filtered lysates for 1 h. The resin was pelleted via centrifugation and washed with 40 ml of Buffer A. Resins were then resuspended with 5 ml of Buffer A, transferred to a 5-ml gravity column (G-Biosciences) and further washed with 25 ml of Buffer A. Proteins were collected in 1 ml fractions using a stepwise elution of 3 ml of immobilized metal affinity chromatography (IMAC) buffer (50 mM Tris–HCl, 250 mM NaCl, 10% glycerol) containing 50, 100, 200 and 500 mM imidazole, respectively. For the second step of purification, the 200 and 500 mM imidazole aliquots were pooled and mixed with small ubiquitin-related modifier (SUMO) Protease (Mclab, 1000 U) and the supplied salt-free buffer to remove the 10His-Smt3 tag. Digested samples were injected into a 10K MWCO Pierce™ Slide-A-Lyzer® Dialysis Cassette (Thermo Fisher Scientific) and dialyzed against 2 L IMAC buffer containing 25 mM Imidazole for 3 h. Then, 2 ml of Ni-NTA agarose resin slurry (Thermo Fisher Scientific) was equilibrated with the dialysis buffer and incubated with the dialysate samples for 1 h. The resins were collected in a 5 ml gravity column, and tag-free proteins were collected and, if necessary, concentrated using a 10K MWCO centrifugal filter (PALL) to obtain desired concentrations. Prior to size-exclusion chromatography (SEC), proteins were stored in IMAC buffer containing 25 mM imidazole. AcrIIIA1 was further purified via SEC using a Superdex 75 Increase 10/300 GL column on an ÄKTA pure™ instrument (Cytiva) at 4°C. One column volume was collected with Buffer B in 1 ml fractions and stored at −80°C. All fractions were resolved on an 10–15% sodiumdodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized with 0.1% Coomassie G-250. Protein sizes were estimated with PageRuler™ Plus Prestained Protein Ladder, 10–250 kDa (Thermo Fisher Scientific). Gels were imaged using an Azure 400 imager (Azure Biosystems). Extraction and visualization of AcrIIIA1-associated nucleic acids (acrIIIA1-RNAs) The acrIIIA1-RNA extraction protocol was adapted from the protocol used for extracting total RNAs from Cas10-Csm in (30) with minor modifications. Briefly, each size-exclusion fraction of AcrIIIA1 was resuspended in 1.5 ml of TRIzol Reagent (Invitrogen). After 5 min incubation, 200 μl of chloroform was added to the mixture and vortexed for 15 s, followed by 10-min incubation at room temperature. Samples were then centrifuged for 15 min at 12 000 × g at 4°C. The top aqueous phase containing the RNA was transferred to a new tube and mixed with 0.5 ml of isopropanol to precipitate the RNA. Mixtures were incubated at room temperature for 10 min, followed by 10 min centrifugation at 12 000 × g at 4°C. RNA pellets were washed with 1 ml of 75% ethanol, followed by 5 min centrifugation at 12 000 × g at 4°C. RNA pellets were air-dried and re-dissolved with 25 μl RNase-free water. The entire extracted acrIIIA1-RNAs were end-labeled with T4 Polynucleotide Kinase in a reaction containing γ-[32P]-ATP (PerkinElmer) for 1 h at 37 °C. Then, radiolabeled acrIIIA1-RNAs were subjected to ethanol precipitation to remove excess γ-[32P]-ATP. Purified acrIIIA1-RNAs were mixed 1:1 with 95% formamide loading dye and resolved on 12% urea PAGE gels. The sizes were estimated with a Decade™ Marker RNA ladder (Thermo Fisher Scientific). Gels were exposed to storage phosphor screens and visualized using Amersham Typhoon biomolecular imager (Cytiva). Nuclease digestion of acrIIIA1-RNAs Approximately 200 ng of radiolabeled acrIIIA1-RNAs were treated with 1 unit of DNase I (NEB), 5 μg of RNase A (Thermo Fisher Scientific) or 1 μl of 0.5 M EDTA to serve as control. Reactions were incubated at 37°C for 30 min, quenched by adding an equal volume of 95% formamide loading dye and resolved on a 15% urea PAGE gel. Gels were exposed to storage phosphor screens and visualized using an Amersham Typhoon biomolecular imager (Cytiva). Two independent trials were performed. RNA-seq analysis of acrIIIA1-RNAs Extracted acrIIIA1-RNAs (100–300 ng) were submitted to the Roy J. Carver Biotechnology Center at the University of Illinois at Urbana-Champaign for RNA-seq. The small RNAs were treated with Antarctic Phosphatase and Polynucleotide Kinase, and libraries were constructed with the Next Small RNA Sample Prep kit (NEB) according to the manufacturer’s instructions. The library pool was quantitated by quantitative polymerase chain reaction (qPCR) and sequenced on a MiSeq flowcell for 151 cycles from each end of the fragments using a MiSeq 300-cycle sequencing kit version 2. Fastq files were generated and demultiplexed with the bcl2fastq v2.20 Conversion Software (Illumina), and adaptors were trimmed from the 3′-end of the reads. For analysis, RNA-seq reads were first mapped to the entire E. coli Rosetta 2 (BL21(DE3)) genome (NZ_CP083274.1). Subsequent to the identification of coverage peaks, reads were aligned to individual transfer RNA (tRNA) and transfer-messenger (tmRNA) sequences, which were extracted from the GenBank file and reverse complemented if necessary to match the strand orientation. Coverage profiles were calculated as a function of position using Bowtie2 (v. 2.4.4) and Samtools (v. 1.13) (49). Furthermore, using pysam (https://github.com/pysam-developers/pysam, a python wrapper around HTSlib (50) and the Samtools), the starting and ending positions of every aligned read were binned into respective position histograms and were overlaid atop the coverage profile which was calculated using IGVtools (v. 2.13.2.) (51). For AcrIIIA1 purified from E. coli, three independent RNA-seq analyses of acrIIIA1-RNAs extracted from three different AcrIIIA1 preps were performed. Constructing tRNA read coverage heatmaps To generate heatmaps, all tRNA sequences were extracted from the GenBank file, reoriented, and combined into a single FASTA file and aligned using the cmalign program (part of the Infernal suite) with the ‘-g –notrunc’ flags. Infernal (v. 1.1.4) (52) is a suite of programs for RNA sequence analysis using profiles of RNA sequence and secondary structure consensus. The covariance model utilized in this analysis was created and used by tRNAviz4 (53) (using cmbuild and cmcalibrate programs, part of the Infernal suite). Using a custom Python bioinformatics library developed for this study, each tRNA raw coverage profile previously generated was first normalized to one, absolute value of its derivative (with respect to position) calculated, and mapped onto its corresponding alignment parsed from the cmalign output file. Only unique tRNA sequences (differing at most by two nucleotides), which had a peak raw coverage >2000 were used in the analysis. The tRNA type, isotype and loop positions were also independently verified using tRNAscan-SE (54). Synthesis of tmRNA for nuclease assays The S. epidermidis tmRNA substrate was synthesized using in vitro transcription. Briefly, the transcription template was amplified from S. epidermidis RP62a using primer pairs L328/L329 (Supplementary Table S1). The PCR product was purified using the EZNA Cycle Pure Kit (Omega Bio-tek), and 0.5–1.0 μg of purified PCR fragment was used as template. The tmRNA was synthesized using the HiScribe T7 High Yield RNA Synthesis Kit (NEB). The reaction was carried out at 37°C for 2 h. Following tmRNA synthesis, the DNA template was digested with DNaseI (NEB), and the synthesized RNA was extracted using Trizol, followed by ethanol precipitation. The S. epidermidis tmRNA pellet was rehydrated in RNase-free water and 25 μg of synthesized RNA was subjected to quick CIP reaction (NEB) to dephosphorylate its ends. The dephosphorylated RNA was extracted again using Trizol, followed by ethanol precipitation. The final S. epidermidis tmRNA pellet was rehydrated in RNase-free water. AcrIIIA1 and RNase R nuclease assays For nuclease assays with AcrIIIA1, the in vitro synthesized tmRNA (described above) and a 31 nt linear RNA (Supplementary Table S2) were used as substrates. For nuclease assays with RNase R, acrIIIA1-RNAs extracted from AcrIIIA1 (described above) and a 43 nt linear RNA (ssRNA-01; Supplementary Table S2) were used as substrates. Substrates were radiolabeled on their 5′-ends using T4 polynucleotide kinase and γ-[32P]-ATP, and purified over a G25 column (IBI Scientific). Substrates were combined with various AcrIIIA1 fractions or RNase R (1 pmol) in nuclease buffer (25 mM Tris–HCl pH 7.5, 2 mM dithiothreitol [DTT], 10 mM MgCl2). The nuclease reactions were carried out at 37°C for 1 h (AcrIIIA1) or indicated time points (RNase R), then quenched with an equal volume of 95% formamide loading buffer. Samples were resolved on an 8% urea PAGE gel at 55 W for 1.5 h. The RNA product lengths were estimated with a Decade™ Marker RNA ladder (Thermo Fisher Scientific). Gels were exposed to storage phosphor screens and visualized using an Amersham Typhoon biomolecular imager (Cytiva). Cas10-Csm and Csm6 expression and purification Cas10-CsmCsm2H6N, Cas10-CsmΔCsm2,Csm4H6N, Cas10-CsmCsm2H6N,Csm3D32A and Csm6H6C were overexpressed and purified from S. epidermidis RP62a Δcrispr-cas cells bearing pcrispr-cas-csm2H6N, pcrispr-cas-Δcsm2-csm4H6N, pcrispr-cas-csm2H6N-csm3D32A or pcrispr-cas-csm6H6C, respectively, as previously described (55). Briefly, overnight cultures were diluted 1:100 in 1 L BHI supplemented with appropriate antibiotics. Cells were incubated at 37°C with agitation until an OD600 of 2.0 was achieved. Cells were pelleted and washed with cold water, and final pellets were stored at −80 °C until purification. To purify Cas10-Csm complexes, thawed pellets were resuspended in 10 ml lysis buffer (22 mM MgCl2 and 44 μg/ml lysostaphin) and incubated in a 37 °C water bath for 1 h. Subsequently, lysed cells were combined with 10 ml of 2× resuspension buffer (600 mM NaCl, 100 mM NaH2PO4, pH 8.0) supplemented with 20 mM imidazole, 0.1% Triton X-100, and one tablet of EDTA-free protease inhibitor (Thermo Fisher Scientific). The mixture was homogenized by inverting the tube gently several times and then sonicated on ice (three 30 s pulses with 1 min rest in between). Insoluble materials were removed via centrifugation at 10 000 × g and 13 500 × g for 20 min, respectively. Lysates were then filtered through a 0.2 μm PES membrane. Filtered lysates were passed through pre equilibrated Ni-NTA columns to capture Cas10-Csm complexes. The Ni-NTA columns were packed in 5 ml gravity columns (G-Biosciences) using 1 ml Ni-NTA agarose resin slurry (Thermo Fisher Scientific) and equilibrated with 10 ml of 1× resuspension buffer (300 mM NaCl, 50 mM NaH2PO4, pH 8.0). Ni-NTA columns with bound Cas10-Csm complexes were washed with 10 ml Wash Buffer 1 (resuspension buffer supplemented with 20 mM imidazole) and then 10 ml Wash Buffer 2 (resuspension buffer supplemented with 20 mM imidazole and 10% glycerol). Cas10-Csm complexes were eluted with five 0.5 ml fractions of elution buffer (resuspension buffer supplemented with 250 mM imidazole and 10% glycerol). Samples from each fraction were resolved in 15% SDS-PAGE gels and stained with 0.1% Coomassie G-250. Protein sizes were estimated with PageRuler™ Plus Prestained Protein Ladder, 10–250 kDa (Thermo Fisher Scientific). Gels were imaged using an Azure 400 imager (Azure Biosystems). Protein samples were stored in elution buffer at −80 °C. Native gel electrophoresis For testing interactions between Cas10-Csm and AcrIIIA1, Cas10-CsmCsm2H6N or Cas10-CsmDCsm2,Csm4H6N complexes (30 pmol) were combined with increasing amounts of AcrIIIA1 FT (flow-through containing both RNA-free and RNA-bound fractions) (30, 150, 300 and 1500 pmol) and allowed to chill on ice for ∼5 min. Cas10-Csm complexes (30 pmol) and AcrIIIA1 (1500 pmols) were also loaded individually on the same gel as controls, alongside the Native Mark Protein Standard (Thermo Fisher Scientific). For testing the interactions between cellular nucleases and AcrIIIA1 FT, 60 pmols of exonucleases (PNPase, RNase J2 or RNase R) were combined with 1200 pmols AcrIIIA1 (or an equivalent volume of protein buffer as a negative control) and allowed to chill on ice for 5 min. In an additional experiment, RNase R (60 pmols) was combined with increasing amounts of Acr (600, 1200 and 2400 pmol) and allowed to chill on ice for 5 min. RNase R (60 pmols) and AcrIIIA1 (2400 pmols) were also loaded individually on the same gel as controls, alongside the native protein standard. Proteins were resolved in a 6% native polyacrylamide gel (29:1 acrylamide/bisacrylamide) with 0.75 mm thickness. Tris-Glycine Buffer (25 mM Tris, 250 mM glycine, pH 8.5) was used to cast the native gels and also used as running buffer. Gels were run for 80 min at 100 V in a Mini-PROTEAN Tetra Cell (Bio-Rad) submerged in an ice water bath. For visualization, gels were stained with 0.1% Coomassie G-250 for 10–20 min, and then submerged in destaining solution (50% methanol and 10% acetic acid) for 1–2 h. Destained gels were imaged with an Azure 400 imager (Azure Biosystems). To visualize acrIIIA1-RNAs in native gels, the Coomassie Blue Destaining Solution was completely removed from the gel by soaking gels in dH2O (with agitation) for at least 3 h, while changing out the water every hour. Gels were then soaked in 50 ml TAE buffer (40 mM Tris Base, 20 mM acetic acid and 1 mM EDTA) supplemented with 5 μl ethidium bromide and agitated for 30–45 min. Gels were imaged with an Azure 400 imager (Azure Biosystems) under UV light. Size exclusion chromatography to test protein-protein interactions For detecting interactions between AcrIIIA1 and Csm2/5, the AcrIIIA1 flow-through (containing both RNA-free and RNA-bound forms) was combined with Csm2 (7 nmol) or Csm5 (14 nmol) at 4:1 and/or 8:1 molar ratios in their respective storage buffers. For detecting interactions between AcrIIIA1 and RNase R, the AcrIIIA1 flow-through containing both RNA-free and RNA-bound forms (30 and 60 nmols) was combined with 3 nmols of RNase R in their respective storage buffers. As controls, the highest molar amount of each protein was passed through the column alone. The mixtures containing Csm2/5 were fractionated on a Superdex S75 Increase 10/300 GL column (Cytiva), while the mixtures containing RNase R were fractionated on a Superdex 200 Increase 10/300 GL column (Cytiva). Columns were pre-equilibrated and run with Buffer B (50 mM Tris–HCl, 250 mM NaCl, 5% glycerol, pH 6.5) using an ÄKTA pure™ instrument (Cytiva) at 4°C. As the proteins were passed through the column, one column volume was collected in 1 ml fractions. Fractions were resolved on 15% SDS-PAGE gels and visualized with 0.1% Coomassie G-250. Protein sizes were estimated with PageRuler™ Plus Prestained Protein Ladder, 10 to 250 kDa (Thermo Fisher Scientific). Gels were imaged using an Azure 400 imager (Azure Biosystems). For each pairwise interaction, at least two independent trials were performed. Cas10-Csm RNA cleavage assays The Cas10-Csm RNA cleavage assay was adapted from (24). Briefly, a target RNA that bears complementarity to the crRNA encoded by the first spacer in the S. epidermidis RP62a CRISPR-Cas locus (ssRNA-01 and Supplementary Table S2), was radiolabeled on its 5′ end with T4 polynucleotide kinase and γ-[32P]-ATP and purified over a G25 column (IBI Scientific). Next, Cas10-CsmCsm2H6N complexes (20 pmols) and AcrIIIA1 (200 pmols) were pre-incubated alone or combined for 30 min at room temperature. To start the reaction, tubes were placed at 37 °C and a master mix containing nuclease buffer (25 mM Tris-HCl pH 7.5, 10 mM MgCl2 and 2 mM DTT) and radiolabeled ssRNA-01 were added to each tube. Aliquots (10 μl) were withdrawn from reaction mixes at various time points (0.5, 1, 2 and 4 min) and quenched by adding an equal volume of 95% formamide loading buffer. Reactions were resolved on a 15% urea PAGE gel at 55 W for 2 h. Gels were exposed overnight to a storage phosphor screen and visualized using an Amersham Typhoon biomolecular imager. ImageQuant software was used for densitometric analysis. Fractions of substrates cleaved were determined using the following equation: density of uncut substrates for each timepoint were divided by the density of uncut substrate in the lane where no protein was added. Three independent trials were performed, and averages (±S.D.) were calculated. Cas10-Csm cOA production assays The cOA production assay was carried out as described in (24) with minor modifications. Briefly, Cas10-CsmCsm2H6N (20 pmols) and AcrIIIA1 (20, 100 or 200 pmols) were combined and pre-incubated at room temperature for 1 h. As controls, Cas10-Csm (20 pmols) and AcrIIIA1 (200 pmols) were incubated alone for the same amount of time. cOA production was initiated in these reactions by adding TNG Buffer (50 mM Tris–HCl, pH 8.0, 150 mM NH4Cl, 5% [v/v] glycerol, 500 μM ATP and 10 mM Mg2+), 5 μCi α-[32P]-ATP and unlabeled ssRNA-01 (Supplementary Table S2). Reactions were incubated at 37°C for 10 min and quenched by adding an equal volume of 95% formamide loading buffer. Produced cOAs were resolved on a 15% urea PAGE gel at 55 W for 1.25 h alongside A10, an oligonucleotide marker which consists of ten consecutive adenine residues. Gels were exposed overnight to a storage phosphor screen and visualized using an Amersham Typhoon biomolecular imager. ImageQuant software was used for densitometric analysis. Fractions of cOAs produced were determined using the following equation: density of cOAs from each reaction containing AcrIIIA1 was divided by the density of cOAs produced in the absence of AcrIIIA1. Three independent trials were performed, and averages (±S.D.) were calculated. AcrIIIA1 cOA degradation assay Radiolabeled cOAs were first produced by Cas10-Csm complexes according to the protocol described above. cOAs were then extracted from reaction mixtures by adding an equal volume of phenol:CHCl3:isoamyl alcohol (25:24:1), vortexing for 15 s and subjecting to centrifugation at 10 000 × g for 2 min. The upper aqueous phase was removed into a fresh tube and extracted as above with an equal volume of CHCl3. The aqueous phase was removed into a fresh tube and purified over a G25 column (IBI Scientific). To test for AcrIIIA1-mediated cleavage of cOAs, purified cOAs were added into a reaction mixture containing nuclease buffer (25 mM Tris–HCl, pH 7.5, 10 mM MgCl2 and 2 mM DTT) and AcrIIIA1 (200 pmols). Reactions were allowed to proceed for 1 h at 37°C and then quenched with an equal volume of 95% formamide loading buffer. Reactions were resolved on a 15% urea PAGE gel at 55 W for 1.25 h alongside A10, an oligonucleotide marker which consists of ten consecutive adenine residues. ImageQuant software was used for densitometric analysis. Fractions of cOAs cleaved were determined using the following equation: density of cOAs from each reaction containing AcrIIIA1 was divided by the starting density of cOAs added into the reaction. Three independent trials were performed, and averages (±S.D.) were calculated. Cas10-Csm DNA cleavage assay For DNA cleavage assays, Cas10-CsmCsm2H6N,Csm3D32A (30 pmols) was pre-incubated with AcrIIIA1 (30, 150 or 300 pmols) in nuclease buffer (20 mM Tris, pH 8.5, 40 mM KCl, 5 mM MgCl2 and 1 mM DTT) at 37 °C for 5 min. As controls, Cas10-CsmCsm2H6N,Csm3D32A (30 pmols) and AcrIIIA1 (300 pmols) were pre-incubated alone under the same conditions. Next, a target RNA that bears complementarity to the crRNA encoded by the first spacer in the S. epidermidis RP62a CRISPR-Cas locus (ssRNA-01 and Supplementary Table S2) was added to a final concentration of 200 nM. A reaction mixture containing the Cas10-Csm complex without ssRNA-01 was included as a control. To start reactions, ϕX174 DNA (0.05 μg) was added to each tube. Reaction mixtures were incubated at 37°C for 4 h and then quenched by the addition of quench buffer (2:2:1 vol/vol/vol Proteinase K:water:0.5 M EDTA, pH 8.0). The samples were then treated with RNase A (20 μg) and incubated at room temperature for 5 min. Samples were resolved on a 1.5% agarose gel at 90 V for 70 min. Gels were stained with Invitrogen™ SYBR™ Gold Nucleic Acid Gel Stain (Fisher Scientific) according to the manufacturer’s instructions and imaged using an Azure 400 imager (Azure Biosystems). ImageJ software was used for densitometric analysis. Fractions of substrates cleaved were determined using the following equation: density of uncut substrates for each reaction containing AcrIIIA1 divided by the density of uncut substrate in the lane where no target RNA was added. Three independent trials were performed, and averages (±S.D.) were calculated. AcrIIIA1 crRNA cleavage assay To test for AcrIIIA1-mediated cleavage of complex-associated crRNAs, reactions were set up exactly as above using the maximum amount of AcrIIIA1 (300 pmols) in the absence of target RNA and ϕX174 DNA. Reactions were allowed to proceed for 4 h at 37°C. Next, RNAs contained within each reaction mixture were extracted and visualized using 750 μl TRIzol Reagent (Invitrogen) and following the rest of the protocol described in ‘Extraction and visualization of AcrIIIA1-associated nucleic acids (acrIIIA1-RNAs).’ Two independent trials were performed. One-step growth curve for Twillingate WT and E6 The protocol for the one-step growth curves was adapted from (56) with minor modifications. Briefly, 50 ml of mid-log S. epidermidis RP62a Δcrispr was combined with phage lysate to achieve an MOI of 0.1. Phages were allowed to adsorb to the host for 5 min at room temperature without agitation. Cells were pelleted by centrifugation at 8000 ×g for 10 min and washed with 50 ml of TSB twice. Following washes, cells were incubated with agitation at 37°C, and 200 μl samples were taken at 10 min intervals and passed through 0.45 μm syringe filters. Supernatant titers were determined as described in the section above titled ‘Phage propagation and enumeration’ for each time point. The experiment was carried out in triplicate, and averages (±S.D.) were calculated. The latent period (30 min) was defined as the time period post-adsorption to the beginning of the first burst. Infection assays with Twillingate WT and E6 for genomic DNA sequencing by Nanopore Overnight cultures of S. epidermidis RP62a Δcrispr bearing pcrispr-spcϕ-EV or pcrispr-spcϕ-polA were diluted (1:100) in BHI broth supplemented with 5 mM CaCl2. Cells were incubated in a 37 °C orbital shaker until an OD600 of 1.0 was reached. Cells were distributed into separate tubes and phages were added to reach an MOI of 1.0. Phages were allowed to adsorb to cells for 5 min at room temperature and placed back into the incubated shaker for 20 min to allow for a single cycle of infection. Cells were pelleted at 5000 × g for 5 min at 4°C, and the pellets were washed with an equal volume of BHI broth, followed by another wash with PBS. Supernatants were discarded, and the pellets were stored at −80°C. For DNA extraction, cell pellets were resuspended in 200 μl of sterile water and transferred into microtubes. Resuspended pellets were then incubated with lysostaphin (100 μg/ml) and MgCl2 (5 mM) at 37°C for 2 h. The Wizard® Genomic DNA Purification Kit (Promega) was used to extract the genomic DNA according to the manufacturer’s instructions. The final DNA pellets were dissolved in 50–60 μl of prewarmed DNase-free water. DNA concentrations were measured using the NanoDrop™ 2000 Spectrophotometer (Thermo Fisher Scientific), and the samples were stored at −20 °C until being sent for sequencing. Phage DNA cleavage analysis via Nanopore sequencing Library preparation and Nanopore sequencing were performed at the SeqCenter (Pittsburgh, PA). For library preparation, the PCR-free Ligation Sequencing Kit (Oxford Nanopore Technologies, SQK-NBD114.24) was used. Sequence reads within the resultant fastq files were aligned to the corresponding phage genomes using Minimap2 aligner (v. 2.26-r1175) (57). Subsequent alignment files were sorted and indexed into bam format using Samtools program with the respective ‘sort’ and ‘index’ commands. Normalized read-length histograms (i.e. probability distribution function) were then generated for each alignment file utilizing the Pysam Python library with only those reads that were successfully mapped to the reference. Then we proceeded to calculate and plot the complementary cumulative distribution functions (CCDF) on a semi-logarithmic scale. Comparing CCDF curves highlights the distribution of ‘rare’ events (i.e. very long reads), allowing for easier identification of discrepancies in tail behavior between the data sets. Large-scale infection with Twillingate WT and E6 for Cas10-Csm purification and RNA-seq analysis Overnight cultures of S. epidermidis RP62a Δcrispr-cas bearing pcrispr-cas-csm2H6N were diluted 1:100 into 1 L of BHI broth. Flasks were incubated in a 37°C orbital shaker until an OD600 of 1.80–2.00 was obtained. Flasks were supplemented with 5 mM CaCl2 and subjected to infection by Twillingate WT or E6 at an MOI of 1.0. One flask was left uninfected with an equal volume of BHI added as negative control. Phage adsorption was allowed to proceed at room temperature for 10 min. Subsequently, flasks were transferred to the orbital shaker for an additional 20 min to allow for a single round of infection. Cultures were then pelleted, and Cas10-Csm complexes were purified according the protocol above titled ‘Cas10-Csm expression and purification.’ Next, RNAs were extracted from ∼500 pmols of Cas10-Csm complex using 750 μl TRIzol Reagent (Invitrogen) as described in the section titled ‘Extraction and visualization of AcrIIIA1-associated nucleic acids (acrIIIA1-RNAs).’ Seven biological replicates were performed for untreated samples and samples treated with Twillingate WT. Three biological replicates were performed for Twillingate E6. Cas10-Csm associated RNAs were analyzed using DESeq2 according to the section below. Analysis of Cas10-Csm associated RNAs following phage infection RNA extracted from Cas10-Csm complexes (∼300 ng) were submitted to the Roy J. Carver Biotechnology Center at the University of Illinois at Urbana-Champaign for RNA-seq. RNA-seq was performed exactly as described in the section above titled ‘RNA-seq analysis of acrIIIA1-RNAs.’ Raw Illumina reads in fastq format were aligned to S. epidermidis RP62a Δcrispr-cas genome using Bowtie2 (v. 2.4.4), producing a Sequence Alignment Map (SAM) file. Following alignment, SAM files were sorted and indexed to generate Binary Alignment Map (BAM) files using Samtools (v. 1.13). Utilizing the -pC and –countReadPairs flags, the alignment outputs, along with an annotation file were given to featureCounts (v. 2.0.3) program which quantified the number of fragments mapped to each genomic feature of the reference genome for every replicate and condition. The -pC flag ensured only read pairs that mapped to the same chromosome and strand were counted, while the –countReadPairs flag treated each fragment pair as a single unit of evidence. The raw counts derived from featureCounts were compiled into a singular CSV file. This, in conjunction with another CSV file outlining the experimental design and conditions, served as the input for the R script that runs the DESeq2 (v. 1.34.0) analysis pipeline in order to perform a differential gene expression analysis. The resulting DESeq2 outputs were then employed to create volcano plots, which showcase the significantly differentially expressed genes between conditions. Codon usage analysis ofS. epidermidisRP62a and Twillingate Raw frequency data for codons and corresponding amino acids in S. epidermidis RP62a and phage Twillingate were generated using the online application at https://www.bioinformatics.org/sms2/codon_usage.html. The frequency data for synonymous codons—those that code for the same amino acid—were first aggregated, and then absolute counts and the relative percentages of each codon within its respective amino acid group were calculated and exported into a table. Quantification and statistical analysis Student’s t-tests were performed to determine if observed drops in plaque counts (Figure 2) or enzymatic activities (Figure 5) were statistically significant. Tests were performed using Microsoft Excel, and a difference was deemed significant if the P-value was below 0.05. Details for numbers of replicates for specific experiments can be found in the corresponding figure legends and methods. Phage Twillingate gp17 (acrIIIA1) is necessary and sufficient for anti-CRISPR activity. (A) Illustration of candidate acr genes within ‘gap 3′ of Twillingate. Arrows below genes show locations of CRISPR targeting spacers tested in phage challenge assays in panels (B) and (E). (B) Phage challenge plate assays with WT Twillingate. (C) PCR products generated from the gap 3 region of ten ‘escaper’ phages (E1–E10). (D) Illustration showing deleted segments (dashed lines) in selected escapers. (E) Phage challenge plate assays with Twillingate E6. (F) Illustration of the inducible system used to screen candidate genes for anti-CRISPR activity. (G) Phage challenge plate assays with ISP using the inducible system in panel (F). aTc, anhydrotetracycline. For all phage challenge plate assays, averages of triplicate measurements (±S.D.) of plaque-forming units per milliliter (pfu/ml) are shown on top, and representative plate images from at least two independent trials are shown on bottom. Dotted line, limit of detection. Asterisks indicate significant reductions in pfu/ml when compared to the EV control in two-tailedt-tests (*P< 0.05, **P< 0.005). AcrIIIA1 attenuates cOA production and DNA cleavage by Cas10-Csm. (A) Urea-PAGE analysis of cOAs produced by Cas10-CsmCsm2H6N following pre-incubation with AcrIIIA1 at indicated ratios. A10, ladder composed of ten adenine nucleotides. (B) Relative amounts of cOAs produced in three trials of the experiment in panel (A). (C) Agarose gel showing cleavage of ϕX174 DNA by Cas10-CsmCsm2H6N,Csm3D32A in the absence or presence of AcrIIIA1 at indicated ratios. KL, 1 kb DNA ladder; kb, kilobase. (D) Relative fractions of DNA cleaved from three trials of the experiment shown in panel (C). For panels (B) and (D), average measurements (±S.D) are shown. Asterisks indicate significant reductions in two-tailed t-tests (*P< 0.05, **P< 0.005, ***P< 0.0005). (E) Illustration of experiment used to detect DNA cleavage in vivo. (F) Nanopore sequencing reads mapped to genomes of Twillingate (Twill) WT and E6 following infection of S. epidermidis Δcrispr bearing pcrispr-spcϕ plasmids. (G) Fractions of Nanopore reads with various lengths that map to phage genomes following infection of indicated strains. (H) Length distributions of Nanopore reads that map to phage (blue) and bacterial (black) genomes following the experiment in panel (E). Results A subset of staphylococcal phages exhibit type III-A anti-CRISPR activity With a goal of identifying new Type III Acrs, we screened a diverse collection of phages for the ability to evade the Type III-A CRISPR-Cas system in the clinical isolate S. epidermidis RP62a (32). This system employs the Cas10-Csm effector complex (composed of Cas10, Csm2/Cas11, Csm3, Csm4, Csm5 and a crRNA) (41), an accessory nuclease that responds to cOA signaling (Csm6) (31,58) and three non-Cas cellular nucleases (PNPase, RNase J and RNase R) (29,30) to mount a successful defense (Figure 1A). To screen for anti-CRISPR activity, we used S. epidermidis RP62a Δcrispr, in which the native repeat-spacer array has been deleted, but harbors pcrispr-spcϕ, a plasmid with a single repeat downstream of which spacer(s) targeting phage(s) of interest can be inserted (33,40). This system typically exhibits robust immunity against diverse phages with a single targeting spacer (42). However, while screening through a group of related phages with myovirus morphology from the family Herelleviridae (PhilBB, Terranova, Quidividi and Twillingate), we noticed that some have the ability to actively escape immunity. To demonstrate this, spacers complementary to three conserved genes within these phages (hyp [hypothetical], holin and polA) were designed in both forward and reverse directions (Figure 1B). Two of the targeted genes (polA and hyp) have consensus σ70 promoters that are known to drive gene expression early during phage infection, while the third gene (holin) has a promoter with an extended -10 motif and lacks a -35 element, which are hallmarks of genes expressed late in the life cycle of these phages (59). As expected, strains bearing pcrispr-spcϕ encoding a non-targeting spacer (empty vector, EV) or spacers producing crRNAs that are identical to the target sequence (polA-RC, hyp-RC and holin-RC) are completely susceptible to phage infection, as evidenced by the appearance of billions of plaques (zones of bacterial death) when dilutions of phage are plated on lawns of each strain (Figure 1C). In contrast, strains expressing crRNAs complementary to the transcript of each targeted gene were completely protected against PhilBB and Terranova, as evidenced by the absence of plaques. However, these same targeting strains remained sensitive to Quidividi and Twillingate, suggesting that these phages have the capacity to resist immunity. To rule out the possibility that escape from CRISPR targeting occurred due to the passive acquisition of mutation(s) in targeted loci in the latter phages, we purified surviving phages and sequenced the targeted regions to confirm that the three loci indeed maintained the wild-type (WT) genotype. Further, to test the limits of this anti-CRISPR activity, we performed phage challenge experiments in liquid culture with representative sensitive and resistant phages PhilBB and Twillingate, respectively, at different phage:host ratios (MOIs) (Figure 1D). The bacterial growth curves revealed that even when Twillingate is outnumbered by bacteria 10:1 (MOI = 0.1), the phage remains completely resistant to CRISPR activity, irrespective of whether the target is located in early- or late-expressed genes. Altogether, these data demonstrate that a subset of S. epidermidis phages employ a robust proactive strategy to overcome Type III-A CRISPR-Cas immunity. A subset of S. epidermidis myophages overcome Type III-A CRISPR-Cas immunity. (A) Illustration of the phage-host system used for Type III-A CRISPR-Cas functional assays in panels (C) and (D). CRISPR-associated cas and csm genes encoded in the chromosome (gray and colored polygons, respectively) and plasmid-encoded repeat and spacer (white and colored boxes, respectively) are shown. Cas/Csm proteins involved in interference are colored according to corresponding genes in the CRISPR locus. Filled and open triangles indicate cleavage events by Cas and cellular nucleases, respectively. Gray diamonds, cOAs; R, RNase R; P, PNPase; J, RNase J1/2; RNAP, RNA polymerase. (B) Alignment of phage genomes generated using MAUVE (http://darlinglab.org/mauve/mauve.html). Similarly colored histograms indicate homologous regions in the nucleotide sequences of the different phages. Arrows indicate genetic loci targeted by CRISPR spacers in functional assays. (C) Phage challenge plate assays. Averages of triplicate measurements (±S.D.) of plaque-forming units per milliliter (pfu/ml) are shown on top, and representative plate images from at least two independent trials are shown on bottom. Dotted line, limit of detection; RC, reverse-complement. (D) Phage challenge assays in liquid culture. Averages of triplicate measurements (±S.D.) are shown; OD600, optical density at 600 nm. Twillingate gp17 (acrIIIA1) is necessary and sufficient for anti-CRISPR activity We next sought to identify the gene(s) responsible for anti-CRISPR activity. We started by taking advantage of the fact that the CRISPR-sensitive and -resistant phages share a high degree of sequence similarity. For instance, PhilBB and Twillingate share ∼88% identity at the nucleotide level in a pairwise alignment; however, there are ten ‘gaps’ in which the resistant phage Twillingate possesses genes that PhilBB lacks (Supplementary Figure S1A). Thus, we reasoned that gene(s) conferring anti-CRISPR activity must reside within these gaps. To begin to narrow down which gap region confers the CRISPR resistance phenotype, a two-step hybridization strategy was employed (Supplementary Figure S1B). First, PhilBB and Twillingate were combined and allowed to co-infect and recombine within S. epidermidis RP62a Δcrispr-cas, which bears a large genomic deletion encompassing the CRISPR-Cas locus (34). The resulting phage lysate enriched with recombinants was then passaged through S. epidermidis RP62a Δcrispr bearing pcrispr-spcϕ-polA in order to recover phage hybrids that acquired anti-CRISPR activity. Eight hybrid phages were purified, and their DNA was extracted and subjected to PCR amplification across the 10 gap regions to determine from which phage (PhilBB or Twillingate) each gap region was derived. This analysis revealed that all eight CRISPR-resistant hybrids shared a short stretch of Twillingate-derived DNA (<10 000 nucleotides) encompassing gaps 1–3 (Supplementary Figure S1C). We further sequenced the whole genome of ‘Twill-BB H8,’ the hybrid phage that acquired the shortest genomic segment from Twillingate, and confirmed that its sequence was identical to PhilBB across the majority of its genome, with the exception of the short region encompassing gaps 1–3 (Supplementary Figure S1D). These observations narrowed our search for the acr to the six candidate genes encoding hypothetical/uncharacterized proteins in gaps 1–3—Twillingate gp11, gp14, gp16, gp17, gp18 and gp19 (Supplementary Figure S1E). We next took several approaches to further narrow the gene(s) required for Acr activity. In silico analyses were first performed using the webtools HHPred (60) and I-TASSER (61) to identify potential structural homologs; however, this analysis failed to produce high-confidence hits across the entire proteins. We next attempted to introduce each of the six candidate genes into separate plasmids; however, multiple attempts failed to produce viable transformants, suggesting the gene products are toxic to the host. We also performed additional hybridization experiments using Twill-BB H8 and PhilBB to further narrow the list, but we were unable to isolate hybrids with shorter Twillingate-derived segments. Finally, we reasoned that since the gene encoding the Acr must necessarily be transcribed and translated before it can perform its function, then CRISPR targeting of the acr itself (or genes nearby within the same transcript) might allow CRISPR to exhibit some activity and thus reveal the acr’s location. To test this, we began by designing spacers targeting the candidates within gap 3—gp16-gp19 (Figure 2A). Spacers were introduced into pcrispr-spcϕ, and S. epidermidis RP62a Δcrispr strains harboring the plasmids were challenged with Twillingate (Figure 2B). The results showed that Twillingate produces 10- to 1000-fold fewer plaques when plated atop strains targeting acr candidates versus strains that target the conserved genetic loci (polA, hyp and holin). In addition, many plaques on the former strains appeared slightly larger than those produced when targeting the conserved genetic loci and were comparable in size to plaques on the EV control plate, suggesting that these surviving ‘escaper’ phages might have subverted immunity by losing the targeted locus altogether. To test this, we purified ten such escapers, extracted their DNA and PCR-amplified across the gap 3 region. PCR products were obtained for only seven escapers, and the products were all shorter than expected (Figure 2C). Accordingly, sequencing of the PCR products revealed that these phages harbor large deletions within/beyond gap 3, resulting in complete loss of the targeted locus, and possibly the acr itself (Figure 2D). To test the latter possibility, we plated the escaper with the smallest deletion (Escaper 6 [E6], Δgp17-gp19) atop the same set of targeting strains and found that the phage had indeed completely lost anti-CRISPR activity (Figure 2E). This analysis allowed us to further narrow our search for the acr to three of the candidates—gp17, gp18 and gp19. Finally, to pinpoint which of these gene(s) confer Acr activity, we developed an inducible system in S. aureus (Figure 2F). In this system, S. aureus RN4220 lacks a chromosomal CRISPR-Cas system but harbors a multi-copy plasmid (pcrispr-cas-spcI) encoding the Type III-A CRISPR-Cas system of S. epidermidis RP62a with a single spacer targeting polA in phage ISP (Bari et al., 2017). In addition, a second plasmid was introduced into this strain (pTet-gpx), which has an aTc-inducible promoter downstream of which candidate gene(s) were inserted. To test for Acr activity, ISP dilutions were plated atop these strains in the presence and absence of aTc. As expected, strains bearing the EV versions of each plasmid permitted the phage to form millions of plaques, and the CRISPR system on its own completely eliminated plaque production regardless of aTc addition (Figure 2G). Additionally, in the presence of gp16, gp18 or gp19, the CRISPR system maintained its ability to block ISP plaque formation. However, when the gp17-gp18 operon, or gp17 alone were introduced into pTet-gpx, ISP formed millions of plaques under inducing conditions despite the presence of the targeting CRISPR system. Altogether, these data demonstrate that Twillingate gp17 is necessary and sufficient in this genetic context to subvert Type III-A CRISPR-Cas immunity. Here onward, we refer to its protein product as AcrIIIA1 (the first Type III-A specific anti-CRISPR), according to the recommended naming convention for newly discovered Acrs (6). AcrIIIA1 co-purifies with fragmented t(m)RNAs We next sought to elucidate the biochemical features of AcrIIIA1. To do so, acrIIIA1 was introduced into the pET28b vector, overexpressed in E. coli Rosetta 2 (BL21(DE3)) and purified using a three-step process consisting of tandem Ni2+-affinity purifications followed by SEC (Figure 3A). AcrIIIA1 has a theoretical molecular weight of ∼10.2 kDa (Supplementary Figure S2A); however, it was found to elute off of the sizing column earlier than expected in fractions 10–12 (Figure 3B). Additionally, we observed that the A254 peak was more prominent than the A280 peak in these fractions, suggesting that the protein likely co-purifies with nucleic acids. To investigate further, we extracted nucleic acids from each fraction, labelled their 5′-ends with 32P, and resolved them on a denaturing gel. This analysis revealed that fractions 10–12 are indeed enriched with nucleic acids ranging from ∼30–80 nucleotides (nt), while fractions 13 and 14, which contain some AcrIIIA1, are devoid of nucleic acids (Figure 3C). Based on the elution profile of the nucleic acid-free fractions, we speculate that AcrIIIA1 may form a dimer on its own. AcrIIIA1 co-purifies with fragmented t(m)RNAs. (A) Schematic of the purification process for AcrIIIA1. (B) Top: Size-exclusion chromatogram for AcrIIIA1 (blue) and protein markers (gray); BSA, bovine serum albumin; mAU, milli-absorbance units at 280 nm (solid line) and 254 nm (dashed line). Dotted vertical line indicates column void volume. Bottom: SDS-PAGE analysis of proteins from AcrIIIA1 fractions. FT is the input into the column; M, prestained protein ladder. (C) Radiolabeled nucleic acids extracted from AcrIIIA1 fractions. Dashed line separates adjacent lanes in the same gel with different exposures. (D) Nuclease digestion of AcrIIIA1-bound nucleic acids. For panels (C) and (D), representatives of at least two independent trials are shown; DL, decade RNA ladder. (E) RNA-seq analysis of AcrIIIA1-associated RNAs showing Illumina reads mapped to the genome of the overexpression strain. (F) Read coverage profile of a representative tRNA (gray bars). Vertical gray shading delimits positions of loops in the tRNA structure. D, D-loop; A, anticodon loop, V, variable loop, T, TψC loop. (G) Heatmap showing changes in normalized raw coverage across tRNAs that co-purify with AcrIIIA1. (H) Read coverage profile of the tmRNA (gray bars); TLD, tRNA-like domain. For panels (F) and (H), numbers of reads that start (green upward bars) and stop (red downward bars) at each position are overlaid across coverage profiles. Red asterisks indicate regions that sustain significant cleavage. For panels (E–H), representatives from three independent trials are shown. We then further explored the nature of the co-purifying nucleic acids. The nucleic acids were first subjected to nuclease digestion and found to be completely sensitive to RNase A (Figure 3D). These data indicate that AcrIIIA1 associates with small RNAs. To determine their identity, purified RNAs were sequenced using an Illumina MiSeq system, and the reads were mapped to the E. coli genome. The results showed prominent coverage peaks corresponding to diverse tRNAs as well as the tmRNA (Figure 3E). Importantly, three independent trials from different AcrIIIA1 preps showed similar results (Supplementary Table S3). Since intact tRNAs are typically ∼70–100 nt, while most AcrIIIA1-associated RNAs are about half those lengths (Figure 3D), we speculated that the latter are likely fragmented. To investigate further, we examined read coverage profiles across individual tRNAs. Since the Illumina reads (150 bp) are longer than intact tRNAs, cleavage sites can be inferred from the pile-up of shorter reads terminating at similar internal positions. This analysis revealed abrupt changes in coverage in discrete regions corresponding to single-stranded loops, especially the anti-codon loops (Figure 3F and G). A similar analysis of the tmRNA showed a steep drop in coverage in an analogous region (the single-stranded mRNA-like region [MLR]) (Figure 3H). Taken together, these data demonstrate that AcrIIIA1 associates with t(m)RNA fragments, here onward referred to as acrIIIA1-RNAs. The close association of AcrIIIA1 with fragmented t(m)RNAs led us to speculate that AcrIIIA1 itself may facilitate/catalyze their cleavage. To test this, AcrIIIA1 was incubated together with a radiolabeled t(m)RNA substrate, and degradation was observed in the presence of divalent metals (Supplementary Figure S2B). We noted that the tmRNA cleavage products generated in this assay range between ∼70 and 100 nts, while the major tmRNA fragments that co-purify with AcrIIIA1 are much more defined at ∼90 nts (inferred by the read coverage profile, Figure 3H). This length discrepancy may be due to the sequence differences between S. epidermidis tmRNA (which was used in the nuclease assay) and E. coli tmRNA (which co-purifies with AcrIIIA1), and/or additional tmRNA trimming by cellular nucleases in E. coli that might occur during the protein purification process. Nonetheless, the generation of fragments of the approximate expected lengths in this in vitro assay inspired us to investigate further. Thus, we searched for candidate active site residues in AcrIIIA1 that might be responsible for metal coordination, specifically aspartate, glutamate and histidine residues, which are commonly found in nuclease active sites (62–64). Due to the abundance of aspartate and glutamate residues in this small protein (nine of each, Supplementary Figure S2A), we settled upon mutating its only two histidine residues to create AcrIIIA1H63A, H64A. We overexpressed and purified this mutant, and observed that it remains stable during purification, with an elution profile comparable to that of wild-type AcrIIIA1 (Supplementary Figure S2C). Subsequent nuclease assays revealed that the RNA-free fractions of AcrIIIA1 (13 and 14) harbor nuclease activity, while the RNA-bound fractions (10–12) are devoid of this activity. This would be the expected result if cleaved RNAs are indeed the products of AcrIIIA1’s activity. Moreover, the activity of AcrIIIA1H63A, H64A is diminished when compared to that of wild-type (Supplementary Figure S2D), lending further support to the notion that AcrIIIA1 is the cause of tmRNA cleavage. Notably, a similar assay using a linear RNA substrate showed more modest degradation (Supplementary Figure S2E), suggesting a preference for structured RNAs. We also tested for AcrIIIA1H63A, H64A function in our inducible system and observed diminished anti-CRISPR activity (Supplementary Figure S2F). Altogether, these data support the hypothesis that AcrIIIA1 catalyzes t(m)RNA cleavage, and suggest that this activity is important for robust anti-CRISPR function. AcrIIIA1 is unlikely to function by impacting translation dynamics Our observation that AcrIIIA1 binds/cleaves t(m)RNAs led us to speculate that AcrIIIA1 might function by modulating translation dynamics. We first considered the possibility that t(m)RNA cleavage may be a way for the phage to leverage codon usage bias to cripple translation of cellular machinery while enhancing translation of phage-encoded proteins. Although Twillingate does not carry its own t(m)RNAs (37), subtle differences in codon usage in the phage might facilitate such a mechanism. To test this, we examined the codon usage statistics of S. epidermidis RP62a and Twillingate, and found that all codons are used by both entities with similar frequencies (Supplementary Table S4). This observation argues against the notion that codon usage bias comes into play in AcrIIIA1’s function. In addition, since the Type III CRISPR-Cas system must compete with ribosomes for access to targeted transcripts, we also considered the possibility that tRNA fragmentation may be a way to slow translation and induce ribosomal queuing, which in turn might prevent Cas10-Csm from accessing targeted RNAs. To test this compelling hypothesis, we took advantage of the fact that Twillingate harbors multiple introns, particularly within the gene that encodes the large terminase subunit (terL) (37). We reasoned that if ribosomal queueing is necessary for AcrIIIA1 activity, then AcrIIIA1 should only function when the CRISPR target is located in exons but not within untranslated regions of introns. Accordingly, we tested CRISPR activity in strains that target terL in both intron and exon regions, and found that Twillingate maintains CRISPR resistance irrespective of target location (Supplementary Table S5). This observation argues against the notion that AcrIIIA1 activity relies upon protection of targeted transcripts by slowed translation machinery. AcrIIIA1 binds to Cas10-Csm via interactions with Csm2 We next investigated the direct impacts of AcrIIIA1 on CRISPR activity. Since the majority of known Acrs bind directly to Cas effector complexes (5), we tested for binding to Cas10-Csm using different protein–protein interaction assays. First, the complex was overexpressed and purified from S. epidermidis RP62a Δcrispr-cas bearing pcrispr-cas-csm2H6N, a multi-copy plasmid that encodes the native CRISPR locus of S. epidermidis RP62a plus a 6-His tag on the N-terminus of Csm2 (41). Complexes were resolved by denaturing SDS-PAGE to assess purity and then by native PAGE to assess integrity. As expected, all five subunits are represented in the complex (Figure 4A), and it exhibits an approximate molecular weight of 270–312 kDa (Figure 4B), which agrees with predicted stoichiometries—Cas101:Csm22/3:Csm33/4:Csm41:Csm51 (65). However, when increasing amounts of AcrIIIA1 were pre-incubated with the complex and then resolved on a native gel, a prominent shifted band with slower migration appeared, suggesting that AcrIIIA1 binds to the complex. Intriguingly, the band corresponding to unbound complex failed to completely shift upward, even when AcrIIIA1 was added in excess (up to 50:1 molar ratio). In light of this observation, and the fact that Cas10-Csm exists in vitro as a heterogenous mixture of subcomplexes, some of which may be devoid of Csm2 and/or Csm5 (65,66), we speculated that AcrIIIA1 may specifically bind to Csm2 or Csm5. AcrIIIA1 binds to Cas10-Csm via interactions with Csm2. (A) SDS-PAGE analysis of Cas10-CsmCsm2H6N. (B) Native PAGE analysis of Cas10-CsmCsm2H6N alone and combined with increasing amounts of AcrIIIA1 (1:1, 1:5, 1:10 and 1:50 molar ratios). (C) Size-exclusion chromatograms for AcrIIIA1 and Csm2 fractionated individually and combined in indicated ratios. mAU, milli-absorbance units at 280 nm (solid line) and 254 nm (dashed line). Dotted vertical line indicates void volume of the column. (D) SDS-PAGE analyses of proteins within fractions corresponding to chromatograms on the left. (E) SDS-PAGE analysis of Cas10-CsmΔCsm2,Csm4H6N. (F) Native PAGE analysis of Cas10-CsmΔCsm2,Csm4H6N combined with increasing amounts of AcrIIIA1 (1:1, 1:5, 1:10 and 1:50 molar ratios). kDa, kilodaltons; M, Protein Ladder. For panels (B) and (F), representatives of three independent trials are shown. M, NativeMark unstained protein ladder. To test this, we overexpressed and purified Csm2 and Csm5, and passed them through a size exclusion column individually or combined with a mixture of RNA-bound and RNA-free AcrIIIA1 (i.e. the FT following the second purification step, Figure 3A). Although AcrIIIA1 and Csm5 failed to show binding (Supplementary Figure S3A and B), we observed a stable interaction between AcrIIIA1 and Csm2 (Figure 4C and D). This is evidenced by the pronounced leftward shift in the elution profile when protein mixtures are passed through the column as compared to when they are fractionated on their own. Notably, both RNA-free and RNA-bound versions of AcrIIIA1 completely shift into the Csm2-bound fractions when combined, suggesting that binding is mediated by protein-protein interactions, rather than by RNA-bridging interactions. Interestingly, while AcrIIIA1 is predicted to form a dimer on its own (according to the size exclusion chromatogram, Figure 3B), the predicted binding stoichiometry to Csm2 is 8:1 (AcrIIIA1:Csm2). The reason for this discrepancy is unclear, although it is conceivable that Csm2 might alter the stoichiometry of AcrIIIA1. The data thus far exclude Csm5 as a binding partner for AcrIIIA1, but does not rule out other members of the complex. Thus, to explore the possibility that additional subunits within the complex may interact with AcrIIIA1, we purified Cas10-CsmΔCsm2,Csm4H6N from S. epidermidis RP62a Δcrispr-cas bearing pcrispr-cas-Δcsm2-csm4H6N, which harbors an in-frame deletion of csm2 and encodes a 6-His tag on the N-terminus of Csm4. Complexes were resolved by denaturing SDS-PAGE (Figure 4E), and then resolved by native PAGE in the absence/presence of AcrIIIA1 (Figure 4F). The results showed that AcrIIIA1 is unable to produce a shifted band when combined with Cas10-CsmΔCsm2,Csm4H6N. Altogether, these data indicate that AcrIIIA1 binds to Csm2 alone within the Cas10-Csm complex. To help conceptualize AcrIIIA1 and its binding to the Cas10-Csm complex, we used AlphaFold 3 (67) to generate structural models. We first attempted to model AcrIIIA1 alone, in various oligomerization states, in combination with various t(m)RNA fragments, and/or with Csm2; however the models had poor confidence scores. For example, the AcrIIIA1 monomer has a predicted template modeling score (pTM) of 0.23, and the dimer has a pTM of 0.18, where structures with pTM < 0.5 are unlikely to represent the true structure (68,69). Similarly, a model with eight copies of AcrIIIA1 folded together with one copy of Csm2 has a pTM of 0.25. Nonetheless, we also modeled the entire Cas10-Csm complex (Cas101:Csm22:Csm33:Csm41:Csm51:crRNA1) with four copies of AcrIIIA1 and observed a pTM score of 0.7 (Supplementary Figure S4). Importantly, the model displays the individual subunits of the Cas10-Csm complex in the appropriate architecture, similar to the experimentally-determined ‘short apo’ complex (PDB ID 7V02) (65) (compare Supplementary Figure S4A–C). Moreover, the models show the four copies of AcrIIIA1 bound specifically to the two copies of Csm2 within the complex, consistent with our experimental findings. These data help to bolster our conclusion that AcrIIIA1 binds specifically to Csm2 within the Cas10-Csm complex. AcrIIIA1 attenuates cOA production and DNA cleavage by Cas10-Csm After having established a direct interaction between AcrIIIA1 and Cas10-Csm, we wondered if and how AcrIIIA1 impacts the complex’s functions. Although Csm2 has no known catalytic activity, we reasoned that AcrIIIA1 might allosterically inhibit one or more activities of other members in the complex—target RNA binding/cleavage by the Csm3 active site, nonspecific DNA cleavage by the Cas10 HD domain and cOA production by Cas10’s palm domain. Since the Cas10 HD domain is dispensable for anti-plasmid immunity under certain conditions (70), we first sought to determine which of these activities are essential for anti-phage immunity in our inducible system (Figure 2F). To test this, pcrispr-cas-spcI variants were constructed that encode mutations in the Csm3 active site, Cas10 HD and Cas10 palm domains (csm3D32A, cas10H14A,D15A and cas10G584A,G585A,D586A,D587A, respectively). These plasmids were introduced into S. aureus RN4220 cells that harbor pTet-EV or pTet-gp17 (-acrIIIA1), and transformants were challenged with ISP in the presence/absence of aTc to assess CRISPR activity (Supplementary Figure S5). Consistent with previous observations (24), we found that the Csm3D32A mutant maintains robust immunity, thus making Csm3’s activity an unlikely target for AcrIIIA1. In contrast, CRISPR immunity was lost in the presence of Cas10H14A,D15A or Cas10G584A,G585A,D586A,D587A, supporting the hypothesis that Cas10 may be the primary target of AcrIIIA1 mediated inhibition. To test this hypothesis, we sought to determine the impact of AcrIIIA1 on the catalytic activities of Cas10-Csm using a series of in vitro biochemical assays. In these assays, both RNA-bound and RNA-free fractions of AcrIIIA1 were tested to explore possible function(s) for acrIIIA1-RNAs. We first assayed for target RNA cleavage by Cas10-Csm. In this assay, purified complexes were combined with a radiolabeled target RNA, reactions were incubated for increasing amounts of time, and RNAs were resolved on a denaturing gel. Consistent with previous observations (15), complexes alone cleaved the RNA target at 6-nt increments, one cut for each copy of Csm3 within the complex (Supplementary Figure S6A and C). In contrast, AcrIIIA1 caused little/no cleavage of the RNA target, regardless of the presence/absence of acrIIIA1-RNAs. Importantly, when AcrIIIA1 was pre-incubated with complexes at a 10:1 ratio, Cas10-Csm maintained full ability to cleave targeted RNA (Supplementary Figure S6B and D). These data demonstrate that AcrIIIA1 has no effect on target RNA binding/cleavage by Cas10-Csm. We next tested for Cas10-mediated cOA production. For this assay, Cas10-CsmCsm2H6N was combined with unlabeled target RNAs in the presence of α-32P-ATP, which is used by Cas10 as a substrate for cOA production, resulting in radiolabeled cOAs (Figure 5A). Interestingly, when complexes were pre-incubated with increasing amounts of AcrIIIA1, the cOA signal was diminished by up to 40%, and both RNA-free and RNA-bound versions of AcrIIIA1 had a similar effect (Figure 5A and B). To determine whether AcrIIIA1 causes cOA degradation in a manner similar to AcrIII-1 (8), we purified radiolabeled cOAs and incubated them together in the same reaction buffer with both versions of AcrIIIA1 for 1 h. The results showed little/no decrease in the cOA signal (Supplementary Figure S7A and B). Taken together, these data demonstrate that rather than cleaving cOAs, AcrIIIA1 dampens cOA production by Cas10-Csm. We finally tested for Cas10-mediated DNase activity. For this assay, complexes were incubated for 4 h with unlabeled target RNA in the presence of single-stranded circular DNA from ϕX174 as substrate. Additionally, we used Cas10-CsmCsm2H6N,Csm3D32A, which is unable to cleave the targeted RNA and thus exhibits constitutive DNase activity that can be visualized on an agarose gel (Figure 5C). As expected, in the absence of target RNA, little/no cleavage of the DNA substrate was observed, while in the presence of the target, complexes caused nearly complete DNA degradation. Interestingly, AcrIIIA1 significantly attenuated DNase activity by up to 40% (Figure 5D). Again, both RNA-free and RNA-bound fractions of AcrIIIA1 showed a similar effect. Since AcrIIIA1 works against a constitutively active complex (Figure 5C and D; Supplementary Figure S5), it is unlikely that AcrIIIA1 functions by reducing the target RNA turnover rate in a manner similar to AcrIIIB2 (9,10). However, since AcrIIIA1 likely harbors nuclease activity (Supplementary Figure S2), we envisioned that one mechanism by which it might render Cas10-Csm nonfunctional is by cleaving its crRNAs. Thus, to test this possibility, we assayed crRNA integrity after incubating complexes with 10-fold excess of AcrIIIA1 under the same reaction conditions used for the DNase activity assays (Supplementary Figure S7C). The results showed that neither RNA-bound nor RNA-free fractions of AcrIIIA1 have the capacity to cleave complex-associated crRNAs. Taken together, these data demonstrate that AcrIIIA1 attenuates DNA cleavage by Cas10-Csm through allosteric inhibition. The results thus far show that AcrIIIA1 dampens both of Cas10’s catalytic activities in a purified system. However, in light of the incomplete inhibition observed in our in vitro assays, we wondered whether AcrIIIA1 provides more robust inhibition in vivo, where presumably, complexes are more likely to be intact, and additional cellular factors that might be necessary for full Acr function (e.g. ions, ligands and/or proteins) are present. To test this, we designed an experiment to track DNase activity as a proxy for Cas10 function in cells during phage infection (Figure 5E). In this experiment, DNA abundance and integrity were assayed following infection with Twillingate WT versus E6, which harbors a deletion encompassing acrIIIA1 (Figure 2D and E). Both phages were allowed to infect S. epidermidis RP62a Δcrispr bearing pcrispr-spcϕ-polA or the EV control for 20 min, which is ∼10 min shorter than the latent period of these phages (determined by one-step growth curve, Supplementary Table S6). This ensures a single round of infection. Following infection, cells were washed to remove free phage, DNA was extracted from cells, and DNA extracts were subjected to long-read sequencing by Nanopore. Since the Nanopore platform directly sequences DNA samples without prior fragmentation or amplification (71), the abundance and lengths of sequencing reads should reflect the abundance and lengths of DNA fragments in the original sample. We first analyzed this dataset for CRISPR function by mapping reads to the corresponding phage genomes and examining the depths of coverage. As expected, similar coverage was observed across both phages following infection of non-targeting (EV) strains, and significantly reduced depth of coverage was observed across Twillingate E6 following infection of the targeting strain as the phage succumbs to CRISPR immunity (Figure 5F). In contrast, the level of coverage across Twillingate WT following infection of the CRISPR targeting strain was identical to that observed with the non-targeting strains, consistent with the robust anti-CRISPR activity in this phage. We next examined the length distributions of reads mapping to both phage and bacterial genomes in this dataset (Figure 5G and H). Consistent with direct DNA cleavage by the CRISPR system, Twillingate E6 showed a progressively diminishing representation of longer reads in its distribution, as evidenced by the complementary CDF (1 − CDF), when grown on the polA targeting strain versus the EV non-targeting strain (Figure 5G). In contrast, the read length distributions of Twillingate WT on both strains appeared comparable. These results corroborate the reduction of Cas10 DNase activity observed in biochemical assays and suggest that Cas10 repression by AcrIIIA1 may be more pronounced in vivo. Cas10-Csm co-purifies with acrIIIA1-RNAs following Twillingate infection Since acrIIIA1-RNAs have little/no impact on Cas10-Csm function in a purified system, we wondered whether they may play other roles in vivo. To explore this idea, we first sought to confirm that they co-purify with Cas10-Csm in the native S. epidermidis background. As mentioned earlier, multiple attempts to overexpress AcrIIIA1 in S. epidermidis failed, thus preventing us from purifying it directly from the native host. However, now knowing that AcrIIIA1 binds to Csm2 within the Cas10-Csm complex, we designed an experiment to capture acrIIIA1-RNAs through purification of the complex (Figure 6A). In this experiment, S. epidermidis RP62a Δcrispr-cas cells harboring pcrispr-cas-csm2H6N were infected with Twillingate WT or E6 for 20 min, Cas10-CsmCsm2H6N complexes were extracted from infected cells, and RNAs were further extracted from complexes. Complex-associated RNAs were then subjected to Illumina sequencing and the reads were mapped to S. epidermidis RP62a Δcrispr-cas. The results showed that t(m)RNAs are indeed significantly enriched in complexes purified from cells infected with Twillingate WT but not E6 (Figure 6B and C). We also investigated the integrity of the tRNAs by examining their depth of coverage profiles (Figure 6D). This analysis revealed that complex-associated acrIIIA1-RNAs purified from S. epidermidis exhibit more extensive fragmentation than acrIIIA1-RNAs purified from E. coli in the absence of the complex (compare heatmaps in Figures 6D and 3G). This difference may be explained by the activities of nucleases specific to each host. Notably, since this experiment was performed under non-targeting conditions (i.e. pcrispr-cas-csm2H6N does not harbor a Twillingate-targeting spacer), the observed degradation is unlikely to have been caused by Csm6, which is activated by cOAs released by Cas10 upon phage transcript recognition (26). Cas10-Csm co-purifies with acrIIIA1-RNAs following Twillingate infection. (A) Schematic of the experiment used to capture Cas10-Csm associated acrIIIA1-RNAs in cells following phage infection. (B) RNA-seq analysis of Cas10-Csm associated RNAs that map to S. epidermidis RP62a Δcrispr-cas following infection with Twillingate WT (top) and E6 (bottom). Fold enrichment of each RNA species was calculated as a ratio of normalized reads from infected/uninfected cells. Peaks corresponding to tRNAs (red/oval) and the tmRNA (black/asterisk) are highlighted. (C) Volcano plots showing RNA species that are significantly enriched in Cas10-Csm following seven and three independent trials of infecting with Twillingate WT (left) and E6 (right), respectively. Red dots, tRNAs; black dot and *, tmRNA. (D) Heatmap showing changes in normalized raw coverage across tRNAs that co-purify with Cas10-Csm following infection with Twillingate. D, D-loop; A, anticodon loop, V, variable loop, T, TψC loop. AcrIIIA1-RNAs bind stably to RNase R in a purified system Having established that acrIIIA1-RNAs indeed co-localize with Cas10-Csm during phage infection, we wondered whether their intended targets might be one or more of the non-Cas cellular nucleases involved in immunity (27,29,30). To explore this idea, we tested for direct interactions between AcrIIIA1 (a mixture of RNA-bound and RNA-free versions) and all of the accessory nucleases known to promote immunity in this Type III-A system—Csm6, PNPase, RNase R and RNase J (Figure 7). Each of these nucleases were purified and combined with AcrIIIA1, and the mixtures were resolved by native gel electrophoresis. The results showed that while the migration of Csm6, PNPase and RNase J remain unchanged in the presence of AcrIIIA1, a pronounced shift in migration of RNase R occurs when combined with AcrIIIA1 (Figure 7A–C). However, we noticed that the protein band corresponding to unbound AcrIIIA1 in these gels persists in the presence of RNase R, leading us to speculate that rather than interacting with AcrIIIA1 directly, RNase R may be interacting with acrIIIA1-RNAs, which are refractory to Coomassie staining. To test this, the native gels were soaked in ethidium bromide and imaged under UV light to visualize acrIIIA1-RNAs (Figure 7D–F). The results revealed that RNase R indeed forms a stable complex with acrIIIA1-RNAs. To confirm this interaction, we passed RNase R and AcrIIIA1 through a size exclusion column alone and combined in different ratios, and found that while the bulk of AcrIIIA1 retains similar elution profiles whether in the presence or absence of RNase R, the A254 peak corresponding to acrIIIA1-RNAs co-localizes with RNase R (Figure 7G and H). These data suggest that RNase R may be the intended target for acrIIIA1-RNAs. RNase R captures and cleaves acrIIIA1-RNAs. (A–C) Native PAGE analyses of indicated nucleases alone and combined with RNA-bound AcrIIIA1 in a 1:20 molar ratio (A and B) or 1:10, 1:20 and 1:40 molar ratios (C). R, RNase R; PNP, PNPase; J2, RNase J2. (D–F) Native gels from panels (A–C) soaked in ethidium bromide and imaged under UV light. (G) Size exclusion chromatograms for AcrIIIA1 and RNase R fractionated on an S200 column individually and combined in indicated ratios. mAU, milli-absorbance units at 280 nm (solid line) and 254 nm (dashed line). Dotted vertical lines indicate void volume of the column. (H) SDS-PAGE analyses of proteins within fractions corresponding to chromatograms on the left. (I and J) Urea PAGE analysis of radiolabeled acrIIIA1-RNAs (I) or a linear 43 nucleotide target RNA (J) that were subjected to degradation by RNase R for various times (0.5, 1, 5 and 10 min). Dotted line separates non-adjacent wells in the same gel. All data are representatives of two independent trials. kDa, kilodaltons; M, Protein Ladder. One possibility is that the interaction between RNase R and acrIIIA1-RNAs might antagonize RNase R’s function, while another is that the interaction might facilitate AcrIIIA1’s activity. As an example of the latter, our previous work showed that RNase R is recruited to Cas10-Csm through interactions with Csm5 (30), thus we wondered whether the interaction between RNase R and acrIIIA1-RNAs might facilitate AcrIIIA1 recruitment to the complex and promote its function. To test this, we assayed for Twillingate’s ability to overcome CRISPR targeting in S. epidermidis cells harboring an in-frame deletion of rnr, the gene that encodes RNase R (30). The controls showed the expected results—while WT Twillingate maintains the ability to form plaques on WT S. epidermidis RP62a cells bearing plasmids with spacers that target early or late phage genes (pcrispr-spcϕ-polA and pcrispr-spcϕ-holin, respectively), Twillingate E6 was unable to plaque on these targeting strains (Supplementary Figure S8). In addition, consistent with previous observations, RNase R is dispensable for CRISPR function against both phages in this system, presumably because targeted transcripts are expressed to sufficient levels to allow the CRISPR system to mount a strong enough immune response (30). Importantly, the results showed that anti-CRISPR activity is also maintained in S. epidermidis RP62a Δrnr, indicating that RNase R is unlikely to facilitate AcrIIIA1’s function. In light of these results, we sought to explore the nature of the interaction between RNase R and acrIIIA1-RNAs. First, we modeled S. epidermidis RNase R in complex with Mg2+ using AlphaFold 3 (67), and observed that the model closely resembles the experimentally-determined structure of E. coli RNase R (PDB ID 5XGU) (72) and has a confidence score that surpasses the 0.5 threshold (pTM = 0.76) (Supplementary Figure S9A and B). We next modeled RNase R-Mg2+ together with representative S. epidermidis t(m)RNAs, both full-length and 40-nt fragments starting from the 5′-end. Interestingly, the interface predicted template modeling (ipTM) scores for RNase R-Mg2+ in complex with full-length t(m)RNAs were consistently lower than those for RNase R-Mg2+ in complex with t(m)RNA fragments (averages of 0.39 ± 0.11 versus 0.75 ± 0.07, respectively) (Supplementary Figure S9C and Supplementary Table S7). Here, it bears mentioning that ipTM scores below 0.6 suggest a failed prediction (67–69). Moreover, the models with RNase R in complex with t(m)RNA fragments showed single-stranded RNA threaded through RNase R’s RNA channel, with the 3′-end poised at the active site Mg2+ (Supplementary Figure S9D and E), similar to a model with RNase R bound to linear RNA (Supplementary Figure S9F). In contrast, models with full-length t(m)RNAs bound to RNase R do not show such strand separation and RNA threading through the active site (Supplementary Figure S9G and H). These data suggest that RNase R might have the capacity to cleave acrIIIA1-RNAs. To test this, we performed nuclease assays and found that RNase R can indeed degrade acrIIIA1-RNAs, albeit with less efficiency than it degrades a linear RNA substrate (Figure 7I and J). Altogether, these data allow us to speculate that RNase R is the intended target of acrIIIA1-RNAs, and the nature of their interaction is likely antagonistic. The precise impacts of acrIIIA1-RNAs on RNase R’s function will require further investigation, and some possibilities are discussed in the section below. Discussion Here, we describe the first Type III-A specific anti-CRISPR protein, AcrIIIA1. AcrIIIA1 provides robust protection against immunity whether early- or late-expressed genes are targeted by the CRISPR system (Figure 1). This inhibition is due, at least in part, to direct interactions with the Cas10-Csm complex—AcrIIIA1 binds to Csm2 (Figure 4 and Supplementary Figure S4) and causes diminished DNase activity and cOA production by Cas10 (Figure 5). Reduced cOA production, in turn, would result in diminished activation of the cOA-responsive accessory nuclease Csm6. Thus, AcrIIIA1 on its own has the capacity to inhibit collateral damage to phage nucleic acids by both core Cas and accessory nucleases. This may not be the whole story, as AcrIIIA1 also co-purifies with small RNAs—t(m)RNA fragments herein referred to as acrIIIA1-RNAs (Figure 3 and Supplementary Table S3). We found that acrIIIA1-RNAs associate with the Cas10-Csm complex in the native host during Twillingate infection (Figure 6), which establishes their physiological relevance. These t(m)RNAs are necessarily derived from the host since Twillingate is devoid of its own t(m)RNAs (37), and AcrIIIA1 is very likely to be responsible for their fragmentation (Supplementary Figure S2), although more experimentation is required to identify the precise active site. We show that t(m)RNA fragmentation is unlikely to impact translation dynamics (Supplementary Tables S4 and S5), and these RNAs have little/no effect on AcrIIIA1’s ability to bind Cas10-Csm and inhibit Cas10’s functions (Figures 4 and 5). However, our observation that acrIIIA1-RNAs bind stably to RNase R (Figure 7 and Supplementary Figure S9) suggests the latter is their intended target. Based on our collective findings, we propose a model in which AcrIIIA1 and its associated RNAs mitigate collateral damage on multiple fronts to provide robust protection against Type III CRSIPR immunity (Figure 8). Proposed model for AcrIIIA1 function. The interference stage of Type III-A CRISPR-Cas immunity is shown in the absence (A) and presence (B) of AcrIIIA1. The impact of acrIIIA1-RNAs on RNase R remains to be determined. RNase R is a nonspecific 3′-5′ exoribonuclease that has the capacity to cleave through structured RNAs (73,74). It plays critical roles in cellular RNA metabolism, including rRNA quality control (75) and the degradation of nonstop mRNAs following rescue of stalled ribosomes (76). In addition to these well-established ‘housekeeping’ functions, RNase R bolsters Type III CRISPR-Cas immunity—it interacts with Csm5 within the Cas10-Csm complex and collaborates with PNPase (another 3′-5′ exonuclease) to catalyze crRNA maturation and targeted transcript degradation, particularly when targeted transcripts are expressed at low levels (30). In light of our finding that RNase R has the ability to degrade acrIIIA1-RNAs (Figure 7I), we speculate one possible mechanism by which acrIIIA1-RNAs might interfere with RNase R’s functions is through competitive inhibition. In other words, acrIIIA1-RNAs might constitute excess nonspecific substrates that absorb collateral damage dealt by RNase R at the site of phage infection. Another possibility is that acrIIIA1-RNA binding might impact RNase R’s localization and/or stability. Indeed, RNase R is subjected to tight regulation in the cell through its interactions with tmRNA and associated protein SmpB, which confine RNase R to the ribosomes and reduce RNase R’s stability (reviewed in Bechhofer and Deutscher, 2019) (77). We also cannot rule out the possibility that acrIIIA1-RNAs might have an impact on other accessory nucleases that are co-opted for immunity, including PNPase and RNase J2 (29,30). Notably, PNPase and RNase J2 are members of the RNA degradosome, a large protein complex that catalyzes bulk RNA decay (78) and plays a critical role in quality control of tRNAs (79). Thus, it is possible that acrIIIA1-RNAs might impact degradosome-associated nucleases through competitive inhibition or some other mechanism yet to be identified. Our future work aims to tease apart these different possibilities. AcrIIIA1 is not widely distributed in the sequenced phage collection, as a BLASTp search for homologs returned a single hit (an uncultured phage). Such an observation is typical of Acrs, which are highly variable and non-homologous (5). However, acr genes are often co-transcribed with more highly conserved anti-CRISPR associated (aca) genes, which encode proteins that regulate transcription of the operon. Accordingly, we searched for homologs for the gene encoded directly downstream of acrIIIA1 (gp18) and found them in a handful of other staphylococcal myophages. We also performed in silico analyses to determine whether gp18 resembles known transcriptional regulators. Specifically, we obtained a predicted structure with AlphaFold2 (80) (pLDDT = 92.9, not shown) and submitted it to the DALI server (81) to identify structural homologs. Interestingly, there were no transcription regulators among the highest confidence hits. Although gp18 is not required for anti-CRISPR activity (Figure 2G), its presence in the same operon as acrIIIA1 implies that it may have a related function. Future work will also investigate the role of gp18 and other candidate Acrs that lie upstream of its homologs in staphylococcal phages. In sum, our results uncover a Type III anti-CRISPR with unique composition and mechanism of action. Here, it bears mentioning that the robust protection afforded by AcrIIIA1 makes it amenable for use in therapeutic and biotechnological applications. For instance, according to CRISPRCasdb (82), Type III-A CRISPR-Cas systems are found in many staphylococci, mycobacteria, and other pathogens that cause drug-resistant infections. As phage therapy gains traction as an alternative treatment modality (83), CRISPR-Cas and other anti-phage defenses threaten to undermine its effectiveness. Thus, therapeutic phages engineered with AcrIIIA1 are more likely to be effective when targeting bacteria with Type III-A CRISPR-Cas systems. In addition, AcrIIIA1 can be used as a selectable marker (in conjunction with Type III-A CRISPR targeting as a counter-selection tool) to isolate phage variants that have taken up desired genetic payloads. These and other applications of AcrIIIA1 have the potential to benefit both basic and applied research. Supplementary Material Acknowledgements We are grateful to Tina Liu for advice on developing the assay for DNA cleavage by the Cas10-Csm complex. Contributor Information Lucy Chou-Zheng, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Olivia Howell, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Tori A Boyle, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Motaher Hossain, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Forrest C Walker, Department of Biological Sciences, University of Alabama, 1325 Hackberry Lane, Tuscaloosa, AL 35401, USA. Emma K Sheriff, Department of Biological Sciences, University of Alabama, 1325 Hackberry Lane, Tuscaloosa, AL 35401, USA. Barbaros Aslan, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Asma Hatoum-Aslan, Department of Microbiology, University of Illinois, 601 S. Goodwin Avenue, Urbana, IL 61801, USA. Data availability Raw Illumina and Nanopore sequencing reads have been deposited at NCBI Sequence Read Archive (SRA) under BioProject ID PRJNA1039552 and is publicly available. This paper analyzes existing, publicly available data for draft genome sequences of PhilBB, Terranova, Quidividi and Twillingate. Accession numbers are listed in the methods section. All original code has been deposited at GitHub at https://github.com/ahatoum/AcrIIIA1/blob/main/anticrispr.py and has been linked to Zenodo to maintain a permanent publicly available record of the code submitted with this manuscript (DOI: 10.5281/zenodo.13845045). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Supplementary data Supplementary Data are available at NAR Online. Funding U.S. National Science Foundation (NSF) [2054755 to A.H.-A]; Burroughs Wellcome Fund [1020298to A.H.-A.]; National Institutes of Health (NIH) Division of Intramural Research [1R01AI173022 to A.H.-A.]. Funding for open access charge: NIH [1R01AI173022]. Conflict of interest statement. A provisional patent application has been submitted for technology related to this work. References Associated Data Supplementary Materials Data Availability Statement Raw Illumina and Nanopore sequencing reads have been deposited at NCBI Sequence Read Archive (SRA) under BioProject ID PRJNA1039552 and is publicly available. This paper analyzes existing, publicly available data for draft genome sequences of PhilBB, Terranova, Quidividi and Twillingate. Accession numbers are listed in the methods section. All original code has been deposited at GitHub at https://github.com/ahatoum/AcrIIIA1/blob/main/anticrispr.py and has been linked to Zenodo to maintain a permanent publicly available record of the code submitted with this manuscript (DOI: 10.5281/zenodo.13845045). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.