key: cord-1050330-5vp21ubk
authors: Grüschow, Sabine; Adamson, Catherine S.; White, Malcolm F.
title: Specificity and sensitivity of an RNA targeting type III CRISPR complex coupled with a NucC endonuclease effector
date: 2021-09-13
journal: bioRxiv
DOI: 10.1101/2021.09.13.460032
sha: 11da058290900db9db0c01694aabf79d75ca7c92
doc_id: 1050330
cord_uid: 5vp21ubk

Type III CRISPR systems detect invading RNA, resulting in the activation of the enzymatic Cas10 subunit. The Cas10 cyclase domain generates cyclic oligoadenylate (cOA) second messenger molecules, activating a variety of effector nucleases that degrade nucleic acids to provide immunity. The prophage-encoded Vibrio metoecus type III-B (VmeCmr) locus is uncharacterised, lacks the HD nuclease domain in Cas10 and encodes a NucC DNA nuclease effector that is also found associated with Cyclic-oligonucleotide-based anti-phage signalling systems (CBASS). Here we demonstrate that VmeCmr is activated by target RNA binding, generating cyclic-triadenylate (cA3) to stimulate a robust NucC-mediated DNase activity. The specificity of VmeCmr is probed, revealing the importance of specific nucleotide positions in segment 1 of the RNA duplex and the protospacer flanking sequence (PFS). We harness this programmable system to demonstrate the potential for a highly specific and sensitive assay for detection of the SARS-CoV-2 virus RNA with a limit of detection (LoD) of 2 fM using a commercial plate reader without any extrinsic amplification step. The sensitivity is highly dependent on the guide RNA used, suggesting that target RNA secondary structure plays an important role that may also be relevant in vivo.

Early bioinformatic analyses of CRISPR associated (cas) genes revealed the existence of a "Repeat associated mysterious protein" (RAMP) operon that included a gene encoding a predicted polymerase (1) . Subsequently, a seminal study of Cas protein families classified the RAMP (Cmr) and Mtube (Csm) subtypes, each including a putative polymerase gene (cmr2 and csm1, respectively) (2) . The burgeoning CRISPR nomenclature, which had become increasingly confusing for everyone, was (first) standardised in 2011, resulting in the classification of the polymerase protein as Cas10 -the signature protein of Type III CRISPR systems (3) .

The first biochemical studies resulted in the observation that type III CRISPR systems comprise a multi-subunit ribonucleoprotein complex that uses CRISPR RNA (crRNA) to target and degrade cognate RNA molecules (4) . In a follow up study, the Terns and Li labs determined the crystal structure of Cas10, confirming the predicted PALM polymerase/cyclase domain and ruling out a role for the N-terminal HD nuclease domain in the targeted degradation of RNA (5) . Target RNA cleavage was subsequently shown to be catalysed by the Cas7 (Cmr4/Csm3) subunit, which forms the backbone of Type III complexes (6) (7) (8) (9) . This left the function of the Cas10 subunit unresolved -an uncertainty that was compounded by the observation that plasmid clearance by a type III system from Sulfolobus islandicus required transcription of DNA target genes and depended on the presence of an uncharacterised Cas protein known as Csx1 (10) . Similarly, plasmid clearance in Staphylococcus epidermidis was shown to be dependent on the active site of the polymerase/cyclase domain and the presence of the Csm6 protein (11) . Csx1 and Csm6 are found adjacent to many Type III CRISPR systems, although not part of the multi-protein complex. Tellingly, both proteins have a C-terminal HEPN nuclease domain (12) , suggesting a possible function as ribonucleases for these proteins. In vivo studies demonstrated that the S. epidermidis type III system tolerated lysogenic phage but destroyed lytic phage via transcription-dependent DNA targeting (13) .

The first convincing evidence that the HD domain of Cas10 was a nuclease came from a combined structure / function study of the protein from Thermococcus onnurineus, which demonstrated that the HD nuclease domain has ssDNA endonuclease activity in vitro (14) .This was confirmed unequivocally by studies of intact type III complexes from

Thermus thermophilus which demonstrated ssDNA-specific HD nuclease activity, activated by target RNA binding (15) (16) (17) (18) (19) .This DNase activity is not ubiquitous however, with some type III systems having little detectable RNA-activated DNase activity, at least in vitro (20, 21) .

In 2017, another major piece of the puzzle was resolved with the demonstration that the PALM domains of Cas10 generate a novel family of cyclic oligoadenylate (cOA) molecules once activated by target RNA binding (22, 23) . cOA, which is generated in a range of ring sizes from cA3 to cA6, functions as a second messenger of infection, activating effector nucleases to provide immunity against mobile genetic elements. These effectors include the widespread Csx1/Csm6 family, which bind cA4 or cA6 in their CARF (CRISPR associated Rossman fold) domains, resulting in activation of the HEPN ribonuclease domain at the Cterminus (22) (23) (24) (25) (26) (27) and extensive non-specific degradation of RNA in the cell (28) .

Subsequently, CARF-family effector proteins capable of targeting DNA when activated by cOA have also been described (29) (30) (31) . All of these enzymes are presumed to carry out "collateral cleavage" by degrading the nucleic acids of both host and virus, delaying viral replication. This process can lead to cell death if not controlled, but a range of specialised ring nuclease enzymes that degrade cOA allow cells to deactivate the collateral cleavage activity and survive viral assault (reviewed in (32) ).

A subset of Cas10 orthologues, primarily found in type III-B CRISPR systems in the gamma proteobacteria, firmicutes and bacteroidetes, lack the N-terminal HD domain entirely and are thus smaller proteins, around 600 aa in length (33) . The proteins tend to have the cyclase active site residues conserved in the PALM domains, suggesting that they provide interference solely via cyclic nucleotide signalling rather than direct Cas10-mediated DNA degradation. Here we focus on the system from Vibrio metoecus, a type III-B system comprising subunits Cmr1-6 where the Cmr2/Cas10 subunit lacks an HD-nuclease domain, together with a Cas6 enzyme for crRNA generation ( Figure 1A ). The CRISPR system is encoded on a prophage that is also found in several Vibrio cholerae strains, and there are no associated adaptation genes (34) . Adjacent to the cmr genes is a gene encoding NucC, a hexameric DNA endonuclease that was recently shown to be activated by cyclic trinucleotides in cyclic-oligonucleotide-based anti-phage signalling systems (CBASS) (35) .

NucC is thus an interesting example of an effector protein found in both CBASS and CRISPR defence systems. In the latter context it is assumed to be activated by cA3 generated by the activated Cas10 subunit, although this has not been demonstrated directly.

In the past few years, the collateral cleavage activities of Cas12 and Cas13 have been repurposed for the development of rapid and specific assays capable of detecting specific RNA or DNA sequences (36) (37) (38) (39) . Two recent studies have demonstrated that the ability of type III (Cas10) CRISPR systems to detect RNA specifically and synthesise cOA can be harnessed in combination with CARF family effector nuclease Csm6 to provide an alternative assay system (40, 41) . Here, we demonstrate that a prophage-encoded Vibrio type III-B system generates cA3 on specific target RNA binding, resulting in the activation of the NucC effector nuclease for non-specific dsDNA degradation. By coupling these activities to a fluorescent readout, we obtain a limit of detection of 2 fM for SARS-CoV-2 RNA targets.

Enzymes were purchased from Thermo Scientific or New England Biolabs and used according to manufacturer's instructions. Oligonucleotides and synthetic genes were obtained from Integrated DNA Technologies (IDT, Coralville, Iowa, USA). Synthetic genes were codon-optimized for E. coli and restriction sites for cloning incorporated where necessary. All final constructs were verified by sequencing (GATC Biotech, Eurofins Genomics, DE). Vibrio metoecus nucC (NucC) was obtained as a G-Block with flanking restriction sites for cloning. After digestion with NcoI and SalI, nucC was ligated into NcoI and XhoI-digested pEV5HisTEV (42) to allow expression as a N-terminal His8-fusion protein.

The expression construct pACE-Cmr for the V. metoecus Cmr interference complex (VmeCmr) was assembled using the NEBuilder® HiFi DNA Assembly kit (New England Biolabs) following the Gibson assembly strategy (43) . All fragments were PCR-amplified from synthetic genes prior to assembly. The final construct ( Figure 1B ) contained a ColE1 origin of replication and ampicillin resistance gene, copied from pACE (MultiColi™, Geneva Biotech, Genève, CH). The expression of cmr1-3 and cmr4-6 was driven by T7 promoters.

The cmr2 (cas10) gene included a sequence encoding a TEV-cleavable, N-terminal His8-tag.

The V. metoecus CRISPR array was designed as two repeat sequences flanking two oppositely directed BpiI recognition sites (5'-gtgtcttcgtaccttgaagacca) to allow later insertion of the target/spacer sequence of choice. The synthetic cassette also contained flanking NcoI and SalI sites that were used to clone the pre-array into MCS-1 of pCDFDuet™-1 (Novagen, Merck Millipore). Vibrio metoecus cas6f, obtained as a G-Block with flanking NdeI and XhoI sites, was subsequently cloned into MCS-2 of the latter construct to give pCDF-notarget_CRISPR ( Figure 1B) . The VmeCas6 protein is 69 % identical to the Cas6f protein from a Schewanella putrfaciens type I-F system, and the CRISPR sequence is closely related, suggesting a 5'-handle sequence 5'-CUUAGAAA (44) . It has been suggested that a Vibrio phage has co-opted this hybrid type III-B/I-F system for inter-phage conflict (34) .

Target/spacer sequences were obtained as synthetic oligonucleotides with a 5'-overhang sequence of 5'-GAAA for the sense strand and 5'-GAAC for the antisense strand. After the two strands were annealed, they were ligated into BpiI-digested pCDF-notarget_CRISPR to give pCDF-target_CRISPR. vmeRepeat and spacer sequences are listed in Table 1 .

RNA oligonucleotides were obtained from IDT. The mRNA for the SARS-CoV-2 N gene was obtained by in vitro transcription as follows. The template for in vitro transcription was constructed by PCR-amplification (primers: 5'-ctagccatgGCGTTTGAGACGGGCGACAG and 5'-ctaggtcgaCAGCTCGCTGGTCCAGAACTG) from the commercially available plasmid 2019-nCoV_N_Positive Control (IDT) and cloned by restriction digest / ligation into MCS-1 of pRSFDuet™-1 (Novagen) to give pSG220. The N gene plus the upstream T7 promoter region was PCR-amplified from pSG220 using Phusion™ High-Fidelity DNA polymerase (Thermo Scientific) and standard primers ACYCDuetUP1 and DuetDOWN1. The agarose gel purified PCR product served as template for in vitro transcription using the MEGAscript® T7 kit (Invitrogen) according to the manufacturer's protocol. After DNase I digest the transcript was purified by phenol:chloroform extraction / iso-propanol precipitation. The integrity of the transcript was verified by agarose gel electrophoresis. RNA species were quantified by spectrophotometry using a dilution series for each sample; for RNA oligomers the molar concentration was obtained from OD260 measurement and the calculated extinctions coefficients (IDT); N gene transcript concentration was estimated using the generic e260 of 0.025 (ng µl -1 ) -1 cm -1 and a molecular weight of 464,564 Da (from the formula: MW = 1,449 nt * 320.5 Da + 159 Da). All RNA species were dispensed into single use aliquots and stored at -70 °C.

The NucC reporter substrate was obtained by heating and slow cooling of the 5'-FAM-and 3'-Iowa Black®-labelled strands (25 µM each) in 10 mM Tris-HCl, pH 8, 50 mM NaCl. The dsDNA substrate was stored in the dark at -20 °C.

All infectious work with human coronavirus SARS-CoV-2 (strain hCOV-19/England/2/2020; The production and purification of NucC was carried out as for the VmeCmr complex with the following exceptions: E. coli C43(DE3) was used as the expression host, size exclusion chromatography was carried out on a HiLoad® 16/60 Superdex® pg 200 column (GE Healthcare) using 20 mM Tris-HCl, 250 mM NaCl, 10 % (v/v) glycerol, pH 7.5 as mobile phase, and no heparin purification was performed. The NucC peaks eluting at approximately 160 ml and 180 ml during gel filtration were collected separately, and likely correspond to hexameric and trimeric species, respectively, as observed previously (35) . The trimeric species was used for this study. Quantitation was carried out as described for the VmeCmr complex using a calculated extinction coefficient of 29,910 M -1 cm -1 , and concentrations are reported for the NucC trimer. SDS-PAGE analysis of all proteins used in this study is shown in Figure S1 .

To determine which cOAs are produced by VmeCmr, 500 nM VmeCmr To test for degradation of cA3 by NucC, 500 nM NucC and 100 µM cA3 were incubated at 37 °C for 1 h under the same conditions as for cOA production. Samples were processed and analysed as described above, however, elution was monitored by UV spectrophotometry alone.

All assays were performed in duplicate unless stated otherwise on a FluoStar Omega plate reader (BMG Labtech) using fluorescence detection (lex / em 485 / 520 nm) in black, non- (Table 1 ) and varying concentrations of target or control RNA unless stated otherwise. Synthetic target RNAs are listed in Table 1 .

The reaction was incubated at 37 °C in the absence of NucC to allow for cyclic oligoadenylates to be produced; NucC was added at t = 10 min to start the nuclease reaction. Fluorescence was measured throughout in 30 s intervals at 37 °C for up to 50 min.

The signal between 2 and 9.5 min was used for baseline subtraction (Mars software, BMG Labtech).

RT-qPCR was performed using the 2019-nCoV RUO Kit (Integrated DNA Technologies, #10006713) in combination with the iTaq™ Universal SYBR ® Green One-Step Kit (BioRad).

Each 20 µl reaction contained 10 µl 2X iTaq universal SYBR® Green reaction mix, 0.25 µl iScript reverse transcriptase, 2 µl premixed N2 primer/probe mix, and 2 µl sample. The reverse transcriptase was omitted to test for DNA contamination. Samples were either a 10fold serial dilution of SARS-CoV-2 N gene transcript (final concentrations 1 aM -1 pM) or SARS-CoV-2 extracts. Each condition, including a non-template control, was run in triplicate.

The RT-qPCR reaction was performed in a Stratagene Mx5000P instrument using the 2-step protocol for SYBR Green with Dissociation Curves settings and monitoring ROX as reference dye. Cycle conditions were: 10 min at 50 °C; 1 min at 95 °C; 40x 10 s at 95 °C, 30 s at 60 °C; the dissociation curve was measured at 55 -95 °C. Amplification plots and dissociation curves are shown in Figure S8 , Ct values are provided in Table S1 .

Microsoft Excel and Prism 8 (Graphpad) were used for data analysis. Statistical analyses were performed with Prism 8 (GraphPad). Target specificity was analyzed by extracting the time point (Tt) when the fluorescence intensity crossed a threshold value. The threshold value was set to 1/8 th of the maximal measured fluorescence signal of the reference substrate (pUC target, Table 1 ). Individual Tt values were subtracted from the Tt value obtained in the absence of RNA to give DTt values. Finally, DTt values were normalized to the DTt value of the reference substrate.

To determine the limit of detection, the mean of fluorescence intensities between 29 -31 min reaction time was calculated for each target concentration and independent experiment.

Sample means higher than the mean of the signal for the reference reaction (without RNA or with non-target RNA) plus 10 standard deviations were deemed detected. The lowest measured concentration detected was taken as the LoD. The quoted LoDs represent the average LoD from at least two independent experiments. 

In V. metoecus NucC (NucC hereafter) is the only effector protein encoded by the CRISPR locus ( Figure 1A) . To investigate the activity of the VmeCmr / NucC system, we built two plasmid constructs ( Figure 1B) , one expressing synthetic versions of the codon optimised cmr1-6 genes and a second encoding Cas6 and a mini-CRISPR array (see materials and methods). This system can be programmed to detect any RNA sequence by changing the spacer sequence in the CRISPR array. The VmeCas6 enzyme processes the pre-crRNA transcript to generate mature guide RNAs with the 5'-handle CUUAGAAA (44) . The Cas6 enzyme is not part of type III effectors and as expected VmeCas6 did not co-purify with

VmeCmr. Initially we programmed the system using a single spacer that matched a synthetic target RNA species used previously (21) . The VmeCmr complex expressed at a high level in E. coli and could be purified in milligram quantities. We cloned, expressed and purified NucC separately, using expression plasmid pEV5HisTEV (42) in E. coli, to obtain milligram quantities of pure protein ( Figure 1C ). 

We proceeded to characterise the V. metoecus type III-B (VmeCmr) system by using target RNA complementary to the crRNA spacer sequence of the ribonucleoprotein complex (see Table 1 for sequences) to activate the cyclase activity of Cas10. We confirmed that VmeCmr generates predominantly cA3 when activated by target RNA, with much lower levels of cA4 made, as expected for a system coupled to NucC (Figure 2 ). 

NucC was previously shown to cleave plasmid DNA when activated by synthetic cA3 (35) . To 

NucC has a very high affinity for cA3. As the cOA level generated by Cas10 is directly proportional to the RNA present in the sample (26, 46) , the affinity of effector nucleases for their cOA activators confers intrinsic limitations on the sensitivity of the system. Some Csx1/Csm6 family proteins have an intrinsic "ring nuclease" activity for the degradation of their cOA activator (25, 47, 48) -an activity that is probably important in vivo for the control of the CRISPR mediated immune response (46) . To determine whether NucC degrades cA3, we incubated 100 µM cA3 with 0.5 µM NucC for 60 min and monitored the products by liquid chromatography and UV detection ( Figure S2D ) after removal of the enzyme by spin filtration (MWCO 3 kDa). The assay showed no significant depletion of cA3 and no trace of any degradation products, suggesting that NucC does not degrade its own activator at significant levels. 

We next coupled NucC to specific RNA detection by VmeCmr, monitoring NucC-mediated fluorescent readout as before. We first titrated target RNA concentrations to establish the limits of RNA detection using the NucC-coupled fluorogenic assay (Scheme 1) and a standard reference RNA target sequence (pUC target, Table 1 

Type III CRISPR systems must avoid inappropriate activation by RNA targets such as antisense RNAs transcribed from the CRISPR locus, which could cause toxicity or even cell death. To achieve this, Type III systems sense mispairing of RNA at the Protospacer flanking site (PFS), which is immediately 3' of the RNA duplex formed between the target RNA and the crRNA, corresponding to the repeat-derived 5'-handle of the crRNA. When an anti-sense CRISPR RNA binds, it base-pairs along the length of the PFS, preventing activation of the HD nuclease or cyclase activities of Cas10 (15) (16) (17) 19, 22, 23, 26, 49) . A detailed investigation of the Type III-B system from Thermotoga maritima (TmaCmr) (50) revealed that the three nucleotides at positions -1 to -3 of the PFS are crucial in regulating Cas10 activity, consistent with observations in type III-A systems (51) (52) (53) and furthermore that guanine at position -1 is sensed directly, rather than via base-pairing, to keep the complex in an inactive state (50) .

We first tested the importance of the PFS for VmeCmr activity ( Type III CRISPR systems are tolerant of extensive mis-pairing between crRNA and target RNA, a factor which is postulated to limit viral escape by mutation (26, 50, (54) (55) (56) (57) . The bound target RNA can be divided into 5 bp segments followed by a sixth nucleotide that is flipped out of the duplex by the Cas7 subunit (52, 53, 58, 59) . Segment 1, matching the 5'-end of the spacer sequence, adjacent to the PFS, is particularly important for Cas10 activation (26, 50) , analogous to the "seed" region next to the protospacer adjacent motif in type I, II and V systems. To investigate this, we changed single nucleotides in the target RNA at positions 1 to 9, making the nucleotide at each position identical to, rather than complementary to, the crRNA sequence ( Figure 5D ). At the concentration of target RNA tested (5 pM), single mutations at positions 1 to 5 all resulted in a significant decrease in cA3 production and therefore NucC activity. Mutation at position 6, which is not base-paired in the ribonucleoprotein complex, had no effect on the cyclase activity, as expected, and there were modest reductions in cyclase activity for positions 7 and 8 in segment 2.

VmeCmr is thus exquisitely sensitive to single nucleotide mismatches in segment 1, which contrasts with the findings for TmaCmr, where significant effects on Cas10 activity were only observed when four or five nucleotides were mutated simultaneously (50) , but is in good agreement with findings for the S. solfataricus type III-D and Staphylococcus epidermidis type III-A complexes (26, 60) . A mismatch at position +1 or +3 resulted in almost complete abolition of the fluorescent signal, reflecting effective suppression of the cyclase activity of Cas10. 

We next investigated the sensitivity of the VmeCmr/NucC system for detection of larger RNA species, using the SARS-CoV-2 N gene as an exemplar. To programme VmeCmr to detect the SARS-CoV-2 RNA specifically, we designed, expressed and purified six different VmeCmr complexes carrying guide RNAs designed to match a range of positions in the SARS-CoV-2 N gene ( Figure 6A ). The VmeCmr constructs were named according to the first nucleotide of the N gene matching the crRNA ( Figure 6A , Table 1 ). Each was designed to have a G at position -1 of the PFS, as that provided the highest activity with the reference target set ( Figure 5C ). We generated a ~1250 nt in vitro transcript of the N gene to serve as target RNA for the assays.

In the course of our studies, we noted that the background activity in the absence of any added RNA target became significant for some of the six VmeCmr constructs ( Figure S4A ).

We therefore subjected all complexes to a further purification step by heparin chromatography, which significantly reduced the background activity without affecting the signal generated by the activated complex ( Figure S4B, C) . Accordingly, all further assays were conducted with VmeCmr complexes that had undergone the additional heparin purification step.

As is clear from Figure 6B , we observed a wide range of activities for the six different complexes. The best ones generated a large fluorescent signal within 1-2 min when activated by 2.6 pM transcript while the least sensitive constructs gave only a marginal signal. The difference in activity was reflected in the LoD obtained for each complex ( Figure   6C ). The most sensitive complexes were N209 and N320 with LoDs of 1.9 fM and 8 fM, respectively. The N719 complex was the poorest with an LoD of 1.2 pM; thus, the measured LoDs spanned 3 orders of magnitude for the six investigated VmeCmr complexes. We also tested the Cmr4 D26A variant of the N209 targeting VmeCmr complex. This mutation targets the Cmr4 (Cas7) active site and is known to prevent degradation of the target RNA in all other type III systems studied (6) (7) (8) 19, 26, 52, 60, 61) . It has previously been shown that preventing target RNA degradation in type III CRISPR-Cas complexes leads to increased cOA production (21, 26, 60) which in turn would be expected to lower the target RNA concentration required to trigger NucC activity. For VmeCmr N209 Cmr4 D26A, however, no improvement in sensitivity was observed.

We reasoned that target RNA secondary structure might in part be responsible for the observed range of LoDs. To test this hypothesis, we selected three VmeCmr complexes (N719, N57 and N209, representing the highest, middle and lowest LoDs, respectively) and compared their activity when presented with either the N gene transcript or an RNA oligonucleotide containing the target sequence plus flanking region (Figure 7) . The best performing N209 complex generated equivalent fluorescent signals when activated with the transcript and oligonucleotide targets at either 10 or 100 fM concentration, suggesting that this region in the N gene is equally available for VmeCmr binding in the oligonucleotide and transcript. However, the mid-performing N57 complex did exhibit a significant improvement in activation when presented with the oligonucleotide target compared to transcript. With the oligonucleotide target, the N57 complex showed a similar activity to the best performing N209 complex (Figure 7 ). This is consistent with the possibility that the N57 target site in the RNA transcript is partially obscured by the secondary structure of the RNA. Finally, the N719 complex was much more active against the oligonucleotide target than the transcript, but still did not approach the sensitivity of the other two complexes studied, suggesting that RNA secondary structure is not the only variable that influences VmeCmr activity. Finally, we tested our best N gene-targeting VmeCmr complex, N209, against SARS-CoV-2 viral RNA (Figure 8 , Figure S7 ). SARS-CoV-2 was propagated by infecting Vero E6 cells and culture supernatant was harvested to give viral stock at 6·10 6 PFU ml -1 . The viral stock was 10-fold serially diluted and polyA RNA was added as carrier RNA to each sample before RNA extraction using a commercial kit to give extracts 1 (highest concentration) to 6 (lowest concentration). The extracts were diluted 20-fold in triplicate into our coupled VmeCmr N209 / NucC assay. Both the wild type and the Cmr4 D26A (cas7 mutant) complexes were activated by SARS-CoV-2 RNA extracted from 6·10 3 PFU ml -1 -6·10 6 PFU ml -1 stocks (Extracts 4 -1) in the presence of polyA carrier RNA. As it is likely that many more RNA molecules than viable viral particles existed in the viral stock, we sought to estimate the N gene concentration in the extracts by comparison to a standard curve of known N gene transcript concentrations using RT-qPCR ( Figure S8 , Table S2 ). In this manner, we obtained a concentration of 8 fM for extract 4, the lowest dilution to yield a clear signal in the assay.

Extract 5, at < 1 fM concentration, was not reliably detected. This is consistent with the LoD for the N gene transcript of 2 fM suggesting that VmeCmr can detect target RNA with high sensitivity in complex mixtures of nucleic acids. There was no significant difference between the wild type and Cmr4 D26A variant complexes. 

Here, we investigated a type III-B CRISPR system from V. metoecus. The locus is highly conserved in strains of V. cholerae and V. metoecus sampled over many years and is always prophage-encoded (34) . It has been described as a hybrid III-B/I-F system as it a type III-B effector with a type I-F Cas6 and CRISPR sequence (34) . The associated CRISPR locus varies in length and content, and some examples have spacers with 100% identical matches to Vibrio plasmids, prophage and phage genes, suggesting that this CRISPR system may play a role in inter-phage conflict (62, 63) . These spacers share a CRISPR type I-F 3'-GG PAM, which is not characteristic of "pure" type III systems (34) . It remains to be determined whether this system functions with a CRISPR-type immunity mechanism, clearing invading phage, or with a CBASS/NucC-like abortive infection mechanism (35) , killing the infected cell. (26, 50, (54) (55) (56) (57) , reducing the ability of viruses to mutate target sites to avoid detection by CRISPR defence (54) . In the protospacer flanking sequence (PFS), extensive base-pairing at positions -2 to -5 limits activation of Cas10, a common property for type III systems that ensures anti-sense RNA generated from the CRISPR locus does not activate type III CRISPR defence. We observed a modest sequence preference at the -1 position where a G is slightly favoured over other nucleotides whilst other systems do not discriminate at the -1 position (15, 54) and one, TmaCmr, is strongly inhibited by a G at position -1 (50) . Notably, this preference for a G at position -1 fits with the presence of a 3'-GG PAM in the protospacers present in the CRISPR locus (34) .

For VmeCmr, a single base pair at position -2 in the PFS is sufficient to severely reduce Cas10 activation whilst other systems tend to tolerate single base pairs in positions -2 through -5 (50, 53, 54) .

Taking into account the lessons learned from the reference RNA target, we designed six guide RNAs for the detection of the SARS-CoV-2 N gene transcript. This revealed stark differences in the sensitivity, ranging over 3 orders of magnitude, between the best and worst performing guide RNAs. By comparing the sensitivity of the system when presented with short RNA oligos versus larger transcripts, we confirmed that RNA secondary structure is an important factor for VmeCmr, with the worst performing complex N719 improved by two orders of magnitude when provided with an RNA oligonucleotide target rather than a transcript ( Figure 7 ). This is a familiar concept in molecular biology, where the availability of RNA target sequences can be modulated by changes in RNA secondary structure -for example in riboswitches (reviewed in (64)). Indeed, wide variations in the sensitivity of the Cas13 effector have been observed with different guide RNA sequences during SARS-CoV-2 assay development (38) . In contrast, a recent study from the Wiedenheft lab (40) detected only minor differences in the sensitivity of 10 guides designed to detect different regions of the N gene using a type III effector. A notable difference is that our assays were carried out at 37 °C whereas the Wiedenheft group used a thermostable TtCsm complex active at 60 °C. It is perhaps not entirely coincidental that type III CRISPR systems predominate in thermophiles (65) , where the secondary structure of RNA targets might not present so much of a problem.

One unexpected finding was that the background cyclase activity of VmeCmr in the absence of any target RNA varies between constructs and is proportional to the maximal activity when target RNA is present ( Figure S4A ). Thus, the complex with the N209 guide had a much higher background activity than the one with N719. This background rate was strongly reduced by heparin chromatography and a concomitant decrease in the A260/A280 ratio of the purified complexes. One possibility is that E. coli mRNA with partial matches to the crRNA sequence can co-purify with the VmeCmr complex and are sufficient to provide partial activation of the cyclase.

The combination of VmeCmr and NucC with an optimal guide RNA and fluorescence readout provided a LoD of 2 fM for detection of the SARS-CoV-2 N gene transcript, consistent with the observation that a reaction with fewer than 1 x 10 5 copies of the viral RNA (8 fM) can be readily detected (Figure 8 ). This compares with a LoD using the T.

thermophilus Csm (TtCsm) system of 1 x 10 7 copies (100 pM) (40) and 1.2 x 10 10 copies (1 nM) detected by the TtCmr system (41) . The >100-fold higher sensitivity of the assay we describe here is likely due to a combination of factors. These include the high affinity of NucC for cA3, low background activity of the nuclease, low levels of ring nuclease activity, high levels of cA3 production by VmeCmr and the use of a dsDNA species as a molecular beacon rather than a less stable RNA reporter and screening for optimal guide RNAs. We observed little difference between the wild-type VmeCmr complex and the Cmr4 D26A variant in our assays, despite the fact that this mutation abolishes the ability to degrade bound target RNA (7) . One potential explanation is that we are approaching the limits of target RNA binding affinity instrinsic to the Cmr complex, but this remains an open question for future study.

Our work provides a proof of concept for the use of cA3 generating type III CRISPR systems coupled to the NucC effector nuclease as the basis for a rapid, sensitive and specific assay for any desired RNA sequence. The sensitivity of our best performing complex is on a par with first generation Cas13 assays; by way of comparison, Lbu-Cas13a programmed with a single guide RNA has a LoD of 10 fM of target RNA (38, 66) . There is clear potential for further improvement, for example by optimising and multiplexing guide RNAs, which has yielded clear benefits with Cas13a (38) . Approaches that reduce the secondary structure of RNA samples are also likely to provide benefits in sensitivity. The signal generated by NucC could also be readily adapted to alternative assay formats such as lateral flow. Complexities arising from sample complexity (for example detection of RNA in saliva samples) have yet to be explored but would be expected to reduce assay sensitivity -there remains the option to couple the VmeCmr/NucC assay to an isothermal amplification step to boost the signal to noise ratio.

A DNA repair system specific for thermophilic Archaea and bacteria predicted by genomic context analysis

A guild of 45 CRISPR-associated (Cas) protein families and multiple CRISPR/Cas subtypes exist in prokaryotic genomes

Evolution and classification of the CRISPR-Cas systems

RNA-guided RNA cleavage by a CRISPR RNA-Cas protein complex

Structure of the Cmr2 Subunit of the CRISPR-Cas RNA Silencing Complex

RNA targeting by the type III-A CRISPR-Cas Csm complex of Thermus thermophilus

Programmable RNA shredding by the type III-A CRISPR-Cas system of Streptococcus thermophilus

Target RNA capture and cleavage by the Cmr type III-B CRISPR-Cas effector complex

Structure of the CRISPR interference complex CSM reveals key similarities with cascade

A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus

Genetic Characterization of Antiplasmid Immunity through a Type III-A CRISPR-Cas System

Comprehensive analysis of the HEPN superfamily: identification of novel roles in intra-genomic conflicts, defense, pathogenesis and RNA processing

Conditional tolerance of temperate phages via transcription-dependent CRISPR-Cas targeting

Crystal structure of the Csm1 subunit of the Csm complex and its single-stranded DNA-specific nuclease activity

Bipartite recognition of target RNAs activates DNA cleavage by the Type III-B CRISPR-Cas system

RNA-activated DNA cleavage by the Type III-B CRISPR-Cas effector complex

Spatiotemporal Control of Type III-A CRISPR-Cas Immunity: Coupling DNA Degradation with the Target RNA Recognition

A type III-B CRISPR-Cas effector complex mediating massive target DNA destruction

RNA and DNA Targeting by a Reconstituted Thermus thermophilus Type III-A CRISPR-Cas System

Multiple nucleic acid cleavage modes in divergent type III CRISPR systems

Cyclic oligoadenylate signalling mediates Mycobacterium tuberculosis CRISPR defence

A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems

Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers

Regulation of the RNA and DNA nuclease activities required for Pyrococcus furiosus Type III-B CRISPR-Cas immunity

Structure of Csx1-cOA4 complex reveals the basis of RNA decay in Type III-B CRISPR-Cas

Control of cyclic oligoadenylate synthesis in a type III CRISPR system

CRISPR-Cas III-A Csm6 CARF Domain Is a Ring Nuclease Triggering Stepwise cA4 Cleavage with ApA>p Formation Terminating RNase Activity

Non-specific degradation of transcripts promotes plasmid clearance during type III-A CRISPR-Cas immunity

Structure and mechanism of a Type III CRISPR defence DNA nuclease activated by cyclic oligoadenylate

The Card1 nuclease provides defence during type III CRISPR immunity

The CRISPR ancillary effector Can2 is a dual-specificity nuclease potentiating type III CRISPR defence

Cyclic oligoadenylate signalling and regulation by ring nucleases during type III CRISPR defence

Structure and activity of the RNA-targeting Type III-B CRISPR-Cas complex of Thermus thermophilus

CRISPR-Cas systems are present predominantly on mobile genetic elements in Vibrio species

Structure and Mechanism of a Cyclic Trinucleotide-Activated Bacterial Endonuclease Mediating Bacteriophage Immunity

Multiplexed and portable nucleic acid detection platform with Cas13, Cas12a, and Csm6

CRISPR-Cas12-based detection of SARS-CoV-2

Amplificationfree detection of SARS-CoV-2 with CRISPR-Cas13a and mobile phone microscopy

Detection of SARS-CoV-2 with SHERLOCK One-Pot Testing

Intrinsic signal amplification by type III CRISPR-Cas systems provides a sequence-specific SARS-CoV-2 diagnostic

SCOPE enables type III CRISPR-Cas diagnostics using flexible targeting and stringent CARF ribonuclease activation

Investigation of the cyclic oligoadenylate signalling pathway of type III CRISPR systems

Enzymatic assembly of DNA molecules up to several hundred kilobases

Structural Variation of Type I-F CRISPR RNA Guided DNA Surveillance

Analysis of Inactivation of SARS-CoV-2 by Specimen Transport Media, Nucleic Acid Extraction Reagents, Detergents, and Fixatives

The dynamic interplay of host and viral enzymes in type III CRISPRmediated cyclic nucleotide signalling. eLife

A Type III CRISPR Ancillary Ribonuclease Degrades Its Cyclic Oligoadenylate Activator

Activation and self-inactivation mechanisms of the cyclic oligoadenylate-dependent CRISPR ribonuclease Csm6

RNA activation-independent DNA targeting of the Type III CRISPR-Cas system by a Csm complex

Target sequence requirements of a type III-B CRISPR-Cas immune system

Self versus non-self discrimination during CRISPR RNA-directed immunity

Type III-A CRISPR-Cas Csm Complexes: Assembly, Periodic RNA Cleavage

Structure Studies of the CRISPR-Csm Complex Reveal Mechanism of Co-transcriptional Interference

Broad Targeting Specificity during Bacterial Type III CRISPR-Cas Immunity Constrains Viral Escape

Unexpectedly broad target recognition of the CRISPR-mediated virus defence system in the archaeon Sulfolobus solfataricus

In vivo activity of CRISPR-mediated virus defence in a hyperthermophilic archaeon

Incomplete prophage tolerance by type III-A CRISPR-Cas systems reduces the fitness of lysogenic hosts

Crystal structure of a CRISPR RNAguided surveillance complex bound to a ssDNA target

Structural biology. Structures of the CRISPR-Cmr complex reveal mode of RNA target positioning

Regulation of cyclic oligoadenylate synthesis by the S. epidermidis Cas10-Csm complex

Co-transcriptional DNA and RNA Cleavage during Type III CRISPR-Cas Immunity

Virus-borne mini-CRISPR arrays are involved in interviral conflicts

CRISPR-Cas in mobile genetic elements: counter-defence and beyond

Riboswitches and Translation Control

An updated evolutionary classification of CRISPR-Cas systems

RNA Targeting by Functionally Orthogonal Type VI-A CRISPR-Cas Enzymes