A naturally DNase-free CRISPR-Cas12c enzyme silences gene expression

Used widely for genome editing in human cells, plants and animals, CRISPR-Cas enzymes including Cas9 and Cas12 provide RNA-guided immunity to microbes by targeting foreign DNA sequences for cleavage. We show here that the native activity of CRISPR-Cas12c protects bacteria from phage infection by binding to DNA targets without cleaving them, revealing that antiviral interference can be accomplished without chemical attack on the invader or general metabolic disruption in the host. Biochemical experiments demonstrate that Cas12c is a site-specific ribonuclease capable of generating mature CRISPR RNAs (crRNAs) from precursor transcripts. Furthermore, we find that crRNA maturation is essential for Cas12c-mediated DNA targeting. Surprisingly, however, these crRNAs direct double-stranded DNA binding by Cas12c using a mechanism that precludes DNA cutting. Cas12c’s RNA-guided DNA binding activity enables robust transcriptional repression of fluorescent reporter proteins in cells. Furthermore, this naturally DNase-free Cas12c enzyme can protect bacteria from lytic bacteriophage infection when targeting an essential phage gene. Together these results show that Cas12c employs targeted DNA binding to provide anti-viral immunity in bacteria, providing a native DNase-free pathway for transient antiviral immunity.

We used wild-type LbCas12a (WT Cas12a) and catalytically deactivated LbCas12a (dCas12a) as controls, which exhibited the expected behavior: expression of WT Cas12a killed the bacteria, while dCas12a silenced gene expression without cell death (Fig. 5B, S3A). In contrast, both WT and dCas12c, when guided by a GFP-or RFP-targeting sgRNA, silenced gene expression without killing cells (Fig. 5C, S3B). These results are consistent with biochemical data showing that Cas12c binds but does not cleave target DNA (Fig. 4).
We next tested whether Cas12c-induced repression of essential gene expression in E.
coli causes cell death. In control experiments using either WT or catalytically deactivated Cas12a, we found that crRNAs targeting an essential gene cause cell death or growth defects (smaller colonies), although the magnitude of the effects depended on the crRNA and were enhanced by the target-activated nuclease activity of WT Cas12a (Fig. S3C). WT Cas12c, paired with guide RNAs targeting essential genes, produced effects similar to those observed using dCas12a (Fig. S3D). This finding suggests that DNA binding by Cas12c blocks transcription of essential genes, inducing cell death or reduced cell proliferation.
Previously it was reported with a slightly different Cas12c ortholog that WT Cas12c but not dCas12c prevented growth of E. coli expressing genome-targeting guide RNAs, leading to the conclusion that Cas12c functions by a genome-cutting mechanism (Yan et al., 2019). Since our Cas12c construct (database identifier: Cas12c_4) varies from the previously tested Cas12c2 (Yan et al., 2019) by seven amino acids, none of which are predicted RuvC active site residues, we also performed the CRISPR interference assay using the Cas12c2 variant. We found that both WT Cas12c2 and dCas12c2, when expressed in cells containing targeting sgRNAs, repress expression of a chromosomally-integrated GFP gene without killing the bacteria (Fig.   5D). This finding supports our conclusion that both Cas12c_4 (focus of this study) and Cas12c2 (Yan et al., 2019) are RNA-guided DNA binding proteins that do not cut DNA.
Notably, we found that expression of tracrRNA and pre-crRNA under separate promoters (Yan et al., 2019), rather than as a single pre-sgRNA transcript, produced a more apparent difference in GFP repression between the WT Cas12c2 and dCas12c2 (Fig. 5D, E).
Since both tracrRNA and pre-crRNA are required for binding and processing by Cas12c (Fig.   S3E), these observations imply that guide RNA maturation limits Cas12c function in cells expressing the dual transcripts, perhaps due to differences in the assembly kinetics of a tripartite versus bipartite complex. This difference in repression between WT Cas12c2 and dCas12c2 was eliminated when the crRNA was produced using a self-cleaving HDV ribozyme at the end of the pre-crRNA transcript (Fig. 5D, E), suggesting that the WT Cas12c2/dCas12c2 discrepancy reported previously (Yan et al., 2019) was due to a difference in pre-crRNA processing capability rather than target dsDNA cleavage activity. Consistent with this conclusion, Cas12c-independent dual RNA processing enabled more robust dCas12c_4mediated gene repression that is comparable to the repression observed in the presence of sgRNA ( Fig. 5F and 5G). We noticed that Cas12c_4-mediated GFP repression is less sensitive to the dual vs. sgRNA switch than Cas12c2, which may be explained if the 7-amino-acid difference yields a difference in the baseline pre-crRNA-processing capacity of each variant.
Regardless of these differences between Cas12c2 and Cas12c_4, our data, together with those presented by Yan et al. 2019, are best explained by a model in which Cas12c does not cleave

Naturally DNase-free Cas12c protects cells from bacteriophage infection
In nature, CRISPR-Cas systems are thought to protect prokaryotes from viral infections based on RNA-guided cleavage of foreign nucleic acid, and DNase activity is the key enzymatic function of all known DNA-targeting CRISPR systems (Nussenzweig and Marraffini, 2020;Shmakov et al., 2017;Watson et al., 2021). Having no detectable target-activated DNase activity ( Fig. 1B and 5), can Cas12c provide antiphage immunity based on targeted DNA binding? We tested this possibility using a phage plaque assay in which ten-fold dilutions of lambda (λ) phage were plated on lawns of E. coli expressing Cas12c and guide RNAs that either target the essential-for-virulence cro gene in the phage λ genome (Johnson et al., 1978) or contain a non-targeting sequence. Cells that express Cas12c and one out of the two tested λphage targeting sgRNAs resulted in a 5000-fold reduction of plaques, as compared to a non-targeting control (Fig. 6A, B). The extent of plaque reduction induced by Cas12c is comparable to that observed with dCas12a but slightly less robust than wild-type Cas12a (Fig. 6A, B).
Together, these results establish Cas12c as the first demonstrated example of a natural DNAtargeting CRISPR ribonuclease that provides antiviral immunity without target-activated DNase activity.

DISCUSSION
CRISPR-Cas systems have evolved in diverse microbes to provide adaptive immunity against foreign nucleic acids. Until now, immunity provided by the class 2 single-effector type V CRISPR-Cas systems was thought to rely on RNA-guided nuclease activity that targets phage or other foreign molecules, typically at the level of double-stranded DNA cutting. In this study, we show the type V CRISPR effector Cas12c is an RNA-guided DNA-binding enzyme that does not cut target DNA. We found that the RuvC domain of Cas12c, which was previously assumed to cut DNA due to functional comparison to other type V CRISPR enzymes, instead exclusively catalyzes the maturation of pre-crRNA for targeted Cas12c DNA binding. Nonetheless, Cas12c inhibits transcription and can defend bacteria against lytic bacteriophage infection. These results suggest that CRISPR systems can provide anti-phage immunity in the absence of targetdirected nuclease activity and, furthermore, that this is Cas12c's native mechanism (Fig. 6C).
All Cas12-family enzymes contain a RuvC catalytic center, an enzymatic domain resembling Ribonuclease H that catalyzes DNA or RNA phosphodiester cleavage by a "carboxylate-chelated two metal-ion" mechanism (Yang and Steitz, 1995). Enzymes thought to be ancestral to Cas12 including TnpB possess RuvC domain-mediated DNA cutting activity (Altae-Tran et al., 2021;Karvelis et al., 2021), implying that the observed RNA-cutting specificity of Cas12c's RuvC domain represents a lineage-specific departure from the ancestral state.
Notably, the substrate specificities of RuvC domains found within different Cas12 enzymes are divergent. For example, Cas12a's RuvC domain cleaves DNA only, with a separate active site responsible for pre-crRNA processing (Fonfara et al., 2016). In contrast, the RuvC domain of CasPhi and Cas12g can cleave both DNA and RNA (Pausch et al., 2020;Yan et al., 2019).
Based on reported DNA cleavage activity by other Cas12c family members (Wang and Zhong, 2021), the switch to exclusive use of RNA as a substrate by the Cas12c variants tested in this study may have occurred recently.
Repurposing the RuvC active site for exclusive RNA processing suggests that pre-crRNA processing is essential for function. Interestingly, we observed that RuvC-mediated pre-crRNA processing is required for high-affinity DNA binding in vitro, but was not as important for transcriptional silencing in vivo in a heterologous host. A possible explanation for this observation is that guide RNA expression is under the control of a non-native strong promoter in our assays. If some transcripts undergo abortive transcription or cleavage by host nucleases, the resulting mature guide RNAs may be sufficient to direct unimpeded transcriptional silencing by Cas12c. The preservation of RuvC-catalyzed RNase activity in Cas12c suggests that crRNA maturation is an essential function of Cas12c in its natural host, perhaps due to lower pre-crRNA transcriptional levels or a lack of certain host nucleases.
Prior to this study, single-effector CRISPR-Cas systems that are DNA-targeting yet The results presented here suggest that contrary to previous assumptions, immunity against bacteriophage may be achieved solely through RNA-guided DNA binding. Other CRISPR systems for which no immune mechanism has been identified, such as some type IV multieffector systems (Taylor et al., 2021), may work analogously to Cas12c. The targeting parameters of Cas12c, including mismatch tolerance, seed sequence and strand-dependence, remain to be determined and could be distinct from those of DNA-cleaving Cas12 enzymes. In addition, the structural basis for the substrate specificity swap in Cas12c's RuvC active site will be fascinating to uncover. Nonetheless, the specific properties of this CRISPR-Cas system, including its minimal PAM requirement (TN), capacity for multiplexed targeting and uniquely precise pre-crRNA processing activity, offer new tool development potential for transcriptional regulation and engineered base editing.

ACKNOWLEDGMENTS
We thank members of the Doudna lab and the Innovative Genomics Institute for helpful discussions. We would also like to acknowledge Josh Cofsky, Dr. Patrick Pausch, and Dr. Brady  (B) Schematics of the termini chemistry after cleavage occurs by a metal-dependent mechanism and by acid-base catalysis, and respective outcomes of CIP-phosphatase treatment on the 3' cleavage product.
(C) CIP phosphatase treatment identifies the 3' crRNA cleavage product to possess a 5' phosphate terminus consistent with metal ion-dependent cleavage. Customized oligonucleotides of the same 11-nt sequence with a 5'-phosphate or a 5'-hydroxyl terminus served as positive and negative controls, respectively.
(D) Cas12c was not able to process a DNA/RNA hybrid pre-crRNA substrate containing four 2'deoxynucleotides spanning the pre-crRNA processing site. The position corresponding to the scissile phosphate is indicated with a scissor icon.
(E) Pre-crRNA processing assays with modified pre-crRNAs containing a 2'-deoxynucleotide at indicated positions revealed a requirement of a 2'-hydroxyl group directly downstream of the scissile phosphate for pre-crRNA cleavage.  filter binding assays with radiolabeled dsDNA, partially base-paired ("bubbled") dsDNA, ssDNA, and ssRNA as a function of Cas12c protein concentration when supplied with a fixed concentration of dual RNAs. KD for fully duplexed dsDNA was 2.7 nM (n=2, 95% CI: 2.6 to 2.9) when NTS was radiolabeled and 3.7 nM (n=2, 95% CI: 3.5 to 3.8) when TS was radiolabeled.
(B) The dependence of dsDNA binding by Cas12c on pre-crRNA processing. Data are from filter binding assays with radiolabeled dsDNA as a function of wild-type Cas12c protein concentration in the presence of tracrRNA and pre-crRNA, mature crRNA, or phosphorothioated pre-crRNA.
Estimated KD values are reported as shown (n=2).  (E) Quantification of GFP intensities in targeting guide expressing cells relative to non-targeting guide expressing cells (from Fig. 5D). Data are represented as mean ± SD (n=3).
(F) Images of the fluorescence interference experiment with the wild-type and RuvC-dead Cas12c_4 (the focus of this study) when GFP-targeting or non-targeting (-) sgRNA, dual RNAs, or mature dual RNAs were used.
(G) Quantification of GFP intensities in targeting guide expressing cells relative to non-targeting guide expressing cells (from Fig. 5F). Data are represented as mean ± SD (n=3).
(H) Cas12c can use sgRNAs targeting GFP and RFP coexpressed in a single transcript for multiplexed targeting. Two non-targeting (-) spacers were in place of the targeting spacers as a negative control. Images shown are representative of the effect seen in replicates (n = 3). (C) Proposed mechanism of the type V-C CRISPR-Cas12c system. After the CRISPR locus is transcribed into a long precursor transcript, Cas12c, with the help of a tracrRNA, recognizes and binds to pre-crRNA. The long pre-crRNA is cleaved into smaller fragments by Cas12c and unknown enzymes (questionmark), allowing Cas12c to bind target DNA molecules by basepairing between the crRNA and the DNA target. In this case, Cas12c does not cause doublestranded DNA (dsDNA) breaks and likely only represses gene expression by transcription block.
The binding alone is sufficient to confer immunity against certain DNA populations such as essential genes in phages, and this may be Cas12c's native immune mechanism.
Plasmid sequences and maps will be made available on addgene.org. To reprogram the guide RNA plasmids for targeting different loci, guide sequences were exchanged via Gibson Assembly or Golden Gate assembly to encode the guide for the selected target site (guide spacer sequences are listed in Table S1B).

Nucleic acid preparation
DNA oligonucleotides were synthesized commercially (IDT) and purified by denaturing PAGE, ethanol-precipitated, and resuspended in water before being used in cleavage assays.
RNA substrates were either synthesized commercially (IDT, Integrated DNA Technologies) or generated through in vitro transcription, which was described previously (Cofsky et al., 2020).
For in vitro T7 RNA polymerase transcription, double-stranded DNA templates were assembled from several overlapping DNA oligonucleotides (IDT) by PCR or by annealing a short oligonucleotide containing the T7 promoter sequence to a long ssDNA oligonucleotide template (IDT). Synthesized or transcribed RNA fragments were PAGE-purified and resuspended in RNA storage buffer (0.1 mM EDTA, 2 mM sodium citrate, pH 6.4). All oligonucleotide identities and sequences are listed in Table S2.
Bound proteins were washed with wash buffer until UV baseline and eluted in elution buffer (50 mM HEPES-Na, pH 7.5, 500 mM NaCl, 300 mM imidazole, 1 mM TCEP, 5% glycerol). The buffer of the eluate was exchanged to ion exchange buffer A (50 mM HEPES-K, pH 7.5, 200 mM KCl, 1 mM TCEP, 5% (v/v) glycerol) using a HiPrep 26/10 Desalting Column (Cytiva) before the addition of TEV protease. After overnight cleavage at 4°C, proteins were loaded into a MBPTrap HP column (Cytiva) connected to a HiTrap Heparin HP Column (Cytiva) and washed until UV baseline. The TEV-cleaved proteins were eluted from the Heparin column with a KCl gradient and concentrated to 2 mL before injection into a HiLoad 16/600 Superdex 200 pg column (Cytiva). The gel filtration buffer contained 20 mM HEPES-K, pH 7.5, 200 mM KCl, 1 mM TCEP, 5% (v/v) glycerol. Peak fractions were verified by SDS-PAGE, and the concentrations of purified proteins were determined using a NanoDrop 8000 Spectrophotometer (Thermo Scientific).

Radiolabeled DNA cleavage assays
Radiolabeled DNA cleavage assays were performed as previously described (Harrington et al., 2020). Briefly, the reactions were carried out in the same cleavage buffer described above for pre-crRNA processing. The tested interference substrates of either target strand (TS) or nontarget strand (NTS) were 5′-radiolabeled with T4 PNK (NEB) in the presence of gamma 32 P-ATP. To form dsDNA substrates, the labeled substrate was annealed with excess cold TS or NTS depending on the labeled strand. The final concentrations of Cas12c, tracrRNA, crRNA, and 32 P-labeled interference substrates were 100 nM, 125 nM, 125 nM and 3 nM, respectively.
Before the addition of the labeled interference substrates at 37°C, Cas12c ribonucleoprotein was pre-complexed by incubating with its pre-annealed dual guide RNAs at 37 C for 5 min, then 25 C for 25 min. Reactions were incubated for 60 min and quenched with 2x Quench Buffer (described above for pre-crRNA processing). The quenched reactions were heated to 95ºC for 5 min. The reaction products were resolved by 10% denaturing PAGE and phosphorimaging.
Oligonucleotide identities are shown in Table S3.

Filter binding assays
The filter binding assays were performed as previously described (Cofsky et al., 2020). Briefly, complexes were formed in 1X binding buffer (20 mM Tris-Cl, pH 7.9 at 25°C, 150 mM KCl, 5 mM MgCl2, 1 mM TCEP, 50 µg/mL heparin, 50 µg/mL bovine serum albumin, 5% glycerol). In a typical binary complex binding assay, Cas protein was first diluted down from 600 nM in series in binding buffer, added to a fixed concentration of tracrRNA (750 nM for all protein dilutions), and was incubated with radiolabeled RNA (100 pM) at 37 C for 5 min, then 25 C for at least 1 hr. For ternary complex binding assays, Cas protein was first diluted down from 600 nM in series in binding buffer, added to a fixed concentration of guide RNAs (750 nM), and incubated at 37 C for 5 min, then 25 C for 25 min. This complex was then added to the radiolabeled DNA probe (100 pM) and incubated at 37 C for 5 min, then 25 C for at least 1 hr. HT Tuffryn (Pall), Amersham Protran, and Amersham Hybond-N+ (GE Healthcare) membranes were equilibrated in 1X membrane wash buffer (20 mM Tris-Cl, pH 7.9 at 25°C, 150 mM KCl, 5 mM MgCl2, 1 mM TCEP, 5% glycerol) and assembled on a vacuum dot-blot apparatus. The membranes were washed once with 30 µL 1X wash buffer before radioactive samples were applied to the membranes by low vacuum. Membranes were then washed once with 40 µL 1X wash buffer, air-dried, and analyzed by phosphorimaging. Data were quantified with ImageQuant TL Software (GE Healthcare) and fit to a binding isotherm using Prism (GraphPad Software).
"Fraction bound" is defined as (background-subtracted volume of Protran spot)/(total background-subtracted volume of Protran spot + Hybond N+ spot). For DNA binding, data were best fit by a model that included an exponent on the concentration terms. The physical basis for this dependency is unknown. Dissociation constants (KD) with 95% confidence intervals (CI) and number of independent replicates (n) are reported in the figure legends, when appropriate. For assays testing complex assembly in EDTA-containing buffer, 25 mM EDTA was substituted for 5 mM MgCl2. Oligonucleotide identities are shown in Table S3.

Dual-color fluorescence interference assay
The strain used for all in vivo assays in this study was E. coli MG1655 with sfGFP and mRFP chromosomally integrated at the nsfA locus, originally described by Qi et al., 2013 and modified with the removal of KanR. Guide RNA plasmids were transformed into electrocompetent cells containing the Cas protein using a MicroPulser Electroporator (Bio-Rad). Cells were recovered for 100 min at 37ºC in LB broth, and tenfold serial dilutions were plated on Kan+CAM media in Nunc Rectangular Dishes (Thermo Scientific). Plates were incubated at 37ºC for 13-17 hours.
GFP images were taken on an Amersham Typhoon Biomolecular Imager (GE Healthcare) with a Cy2 525BP20 filter or on a ChemiDoc MP Imaging System (Bio-Rad) with Blue Epi illumination as excitation source and a 530/28 emission filter. RFP images were taken on the Typhoon Imager with a Cy3 570BP20 filter or on the ChemiDoc imager with Green Epi illumination as excitation source and a 605/50 emission filter. Colonies were visualized using the ChemiDoc imager with White Trans illumination as excitation source and a standard filter, and colony forming units (CFUs) were counted. For experiments comparing WT vs. Dead Cas12c transcriptional repression levels, the GFP intensity of undiluted spots was quantified using Image Lab software (Bio-Rad) and divided by the CFU intensity of the same spots. This ratio is normalized by the ratio of the corresponding non-targeting guide sample so that % GFP intensity is 100% for non-targeting guide samples. All assays were performed in triplicate, and independent transformation was performed for each replicate. Plasmid identities are shown in Table S3.

Bacteriophage plaque assays
Bacteriophage assays were conducted using a modified double agar overlay protocol using phage λ cI857 (Knott et al., 2019). Briefly, E.coli (NEB® 10-beta) containing both a Cas effector and gRNA plasmid (Table S3) were grown overnight 37˚C, 200 rpm. To perform plaque assays, 100 µL of saturated overnight culture was mixed with molten LB Lennox top agar supplemented with appropriate antibiotics and decanted onto a corresponding LB Lennox Agar plate (to final overlay concentrations of 1% (w/v) tryptone, 0.5% (w/v) yeast extract, 0.5% (w/v) NaCl, 0.7% (w/v) agar, 50 µg/mL kanamycin sulfate, and 34 µg/mL chloramphenicol hydrochloride). This overlay was left to dry for 15 minutes under microbiological flame. 10X Serial dilutions of λ cI857 were performed in SM buffer (Teknova), and 2 µL of each dilution were spotted onto the top agar and allowed to dry for 10 minutes. Plaque assays were incubated at 37˚C for 12-16 hours.
After overnight incubation, plaques were scanned using a standard photo scanner and plaque forming units (PFUs) enumerated. In cases where individual PFUs were not enumerable, but clearings were observed at high phage concentrations, the most concentrated dilution at which no plaques/clearings were observed was counted as 1 PFU. Efficiency of plaquing (EOP) calculations for a given condition were performed by normalizing the mean of PFU for a condition to the mean PFU of a non-targeting control: mean(PFUcondition)/mean(PFUnegativecontrol).
All plaque assays were performed in biological triplicate (three experiments carried out on different days using independent bacterial cultures, independently prepared bottom and top agar, and freshly prepared bacteriophage dilutions). targeting crRNA killed cells, while dead Cas12a with targeting crRNA had no effect in survival but silenced gene expression, as compared to non-targeting (-) controls. It is normal if dead Cas12a with a targeting crRNA produced no effect in gene repression since interference by repression is known to be highly guide-dependent. Images shown are representative of the effect seen in replicates (n = 3).
(B) Images of fluorescence interference experiment with Cas12c. Both wild-type Cas12c and dead Cas12c with selected targeting sgRNAs silenced GFP or RFP expression without killing the cells, as compared to non-targeting (-) controls. Images shown are representative of the effect seen in replicates (n = 3).
(C) Interference assay to test whether dead Cas12a with crRNAs targeting essential host gene(s) can cause cell death. Data suggest that dead Cas12a with selected crRNAs (targeting dnaA and rplJ_1) could result in depletion of cells containing those guides, as compared to non-targeting (-) controls. WT Cas12a was included for comparison. Images shown are representative of the effect seen in replicates (n = 3).
(D) Interference assay to test whether wild-type Cas12c with sgRNA targeting essential host gene(s) can cause cell death. Data showed that wild-type Cas12c with selected crRNAs could cause cell death (murD_1 and rplJ_1) or slower growth (rpoB_2), resulting in depletion of cells containing those guides, as compared to non-targeting (-) controls. Images shown are representative of the effect seen in replicates (n = 3).
(E) Cas12c binds to the complex from tracrRNA and crRNA. Data are from filter binding assays with radiolabeled tracrRNA or crRNA as a function of Cas12d protein concentration when they were alone or when two RNAs were present in the binding reaction.   Cas12c can process engineered pre-sgRNA Pre-sgRNA 2: Cas12c

Genome cleavage Transcription block
Cell death Survival with loss of fluorescence

WT LbCas12a
Dilution 10 0 10 1 10 2 10 3 10 4 10 5 10 6 P r o m o te r R B S R F P 1 R F P 2 G F P 1 G F P 2

Dead LbCas12a
CFU GFP RFP P r o m o te r R B S R F P 1 R F P 2 G F P 1 G F P 2 P r o m o te r R B S R F P 1 R F P 2 G F P 1 G F P 2 Supplemental Figure S3. Related to Figure 5.