RNA structure modulates Cas13 activity and enables mismatch detection

Abstract

Cas13 is activated by the hybridization of a CRISPR RNA to a complementary single-stranded RNA protospacer in a target RNA. While Cas13 is not activated by double-stranded RNA in vitro, it robustly targets RNA in cellular environments where RNAs are highly structured. The mechanism by which Cas13 targets structured RNAs remains unknown. Here, we systematically probe the effects of secondary structure on Cas13. We find that secondary structure in the protospacer and 3' to it inhibits Cas13 activity and quantitatively explains the former effect through a strand displacement framework. We then harness strand displacement to generate an 'occluded' Cas13, which enhances mismatch discrimination up to 50-fold and enables sequence-agnostic mutation identification at low (<1%) allele frequencies. Using occluded Cas13, we identify human-adaptive mutations in SARS-CoV-2 and human and avian influenza A viruses, as well as oncogenic mutations in KRAS. Our work leverages improved mechanistic understanding of Cas13 to expand the scope of RNA diagnostics and enable structure-informed Cas13 approaches.

Extended Data Fig. 1 |. Multiplexing RNA secondary structure for Cas13-based assays.

a. Criteria used in the design of target RNAs for multiplexed detection assays. b. Histogram showing the minimum and total structure free energies of randomly generated RNA sequences, compared to the free energy of the two target RNAs used in the tiling experiment. c. NUPACK 3 predicted minimum free energy structures for the two experimental targets. d. Overview of the multiplexed tiling experiment.

Extended Data Fig. 2 |. Activity profiles of Cas13 targeting occluded RNAs.

a-d. Cas13 activity as a function of occluder start position in each 96 nt target block; green shading represents protospacer. a: 28-mers. b: 21-mers. C: 14-mers. d: 10-mers. e-h. Same as A-D, with activities normalized to the non-occluded condition for each target block i. Activities of all negative control conditions. j. Activities as a function of occluder start position for a single target with all four occlusion lengths, normalized to the non-occluded condition. k. Effects of occluders on the activities of different Cas13 orthologs and spacer lengths: LwaCas13d complexed with a crRNA with a 28-nt-long spacer sequence (yellow), LwaCas13d with a crRNA with a 21-nt-long spacer sequence (red), RfxCas13d (green), and LbuCas13a (purple). Note on occluder nomenclature: occluders are named according to their start position from the 5’ end of the target block; all occluders here are 28nt in length. Occluder s19 binds to nucleotides 1–13 of the protospacer and extends into the 5’ flanking region, s31 binds to nucleotides 1–25 and extends slightly into the 5’ flanking region, s52 binds to nucleotides 18–28 and extends into the 3’ flanking region, and s61 binds to nucleotide 28 of the protospacer and extends into the 3’ flanking region. Lines in a-h, j, represent the mean, and error bars the standard deviation, of n = 4 replicates (two technical and two biological). Bar heights points in k represent the mean and error bars show the standard deviation of n = 2 technical replicates.

Extended Data Fig. 3 |. Controls and data normalization for multiplexed tiling assay.

a. Fluorescence curves for the set of experimental conditions chosen for the multiplexed tiling assay. b. Scatter plot showing activity correlation between technical replicates in the tiling experiment. c. Raw activity curves from the control (unoccluded) targets across all conditions in tiling assay. d. Cumulative histogram showing activity distribution for control target 1 before and after correction. e. Same as in d, but for control target 10. f. Dot plot showing activity correlations between the two control targets for all conditions before and after correction. g. Scatter plot showing activity correlation between the two target shufflings for all tested conditions, before and after correction. h. Histogram of ratio of activities of points of the two target shufflings, before and after correction, excluding negative controls. In d, e, h, represents standard deviation.

Extended Data Fig. 4 |. Strand displacement-based model accounts for asymmetries in occlusion pattern.

a-b. Correlation between activities resulting from 21-mer occluders and 28-mers occluding the same nucleotides. In a, 28-mers starting at positions 1, 4, 7, etc are matched with 21-mers starting at positions 3, 6, 9, etc; in b they are matched with 21-mers starting at positions 6, 9, 12, etc. c. Activity as a function of DNA occluder length for occluders of different lengths extending inwards from 3’ and 5’ ends of the protospacer. Strand displacement model predictions shown as dashed lines. Error bars show standard deviation of parameter fit. d. Data from tiling experiment plotted as in A and also including oligos occluding the middle of the protospacer (yellow). Small dots show individual data points from each crRNA; large dots show the mean and error bars show standard deviation across the 8 crRNAs. Model predictions shown as dashed lines. e. Ratio of activities resulting from 21-mer occluders targeting 3’ end of protospacer to those resulting from occluders targeting 5’ end, plotted as a function of occluder length. Each curve shows a different crRNA. Model prediction shown as dashed line. Lines show the mean, and error bars the standard deviation, across n = 4 replicates (two technical and two biological). Cumulative histograms of these data are shown in Fig. 2d. f. As in E, for 28-mer occluders. g. Isothermal titration calorimetry (ITC) data showing the differential power (DP) required to equilibrate the temperature of the reaction chamber in response to titration of 3’ occluded target RNA into a Cas13 RNP solution, which corresponds to the change in heat resulting from the binding that occurs. h. Binding curve derived by integrating the data shown in A with respect to time, yielding the enthalpy of interaction in kcal/mol as a function of molar ratio of the target RNA to the Cas13 RNP. The binding stoichiometry was set to 1:1.

Extended Data Fig. 5 |. Occlusion of target RNA and crRNA increase Cas13’s sensitivity to mismatches.

a. Bar charts representing Cas13 activity for perfectly-matched and mismatched targets when occluded by several different target-blocking occluders, as well as crRNA occluders and combinations thereof. Upper dotted line indicates activity saturation point; curves above saturation are not fit well by exponentials and therefore bars in this region are not accurate measurements of activity. b. Effects of crRNA occlusion and target occlusion, as well as both at once, on Cas13 specificity for two single mismatches and a double mismatch. c. NTC (no-target control) raw curves for conditions with and without a crRNA occluder. d. Raw fluorescence curves and fits for the data shown in a. In a-b, bar heights show the mean and error bars the standard deviation of n = 2 technical replicates.

Extended Data Fig. 6 |. Fluorescence curves from experiments probing Cas13 mismatch detection at low allele frequencies and varying target concentration with occlusion.

a. Raw fluorescence curves corresponding to Fig. 3d detecting presence of a mismatch at different target RNA concentrations with and without occlusion. Shaded regions show the range of fluorescence measurements for each condition across replicates. b. Raw fluorescence curves corresponding to Fig. 3g detecting mismatch detection sensitivity at decreasing allele frequency with and without occlusion. c: Raw fluorescence curves of a SHERLOCK assay detecting the Delta strain of SARS-CoV-2 at different concentrations. Data points show the mean and error bars show the standard deviation across n = 2 technical replicates. d: Comparison between normalized final fluorescence measured for KRAS variants (Fig. 4j) at different magnesium concentrations. Mean across n = 2 technical replicates is shown.

Extended Data Fig. 7 |. Mismatch detection using different types of occlusion.

Raw fluorescence curves corresponding to Fig. 3e–f detecting various mismatches within three different target sequences for no occlusion, target occlusion, crRNA occlusion, and target+crRNA occlusion conditions.

Extended Data Fig. 8 |. Fluorescence curves from Cas13 specificity matrix experiment with different types of occlusion.

Raw fluorescence curves corresponding to Fig. 3h using all possible crRNA and target nucleotides at a given position. Semi-transparent dashed lines show curve fits (see Methods).

Extended Data Fig. 9 |. Discrimination of IAV variants.

a: Fluorescence at 180 minutes is shown for different occluder modifications in absence of target (NTC=No target condition). b: Fluorescence curves for unmodified DNA occluders and occluders modified with a locked nucleic acid (LNA) at position 16. crRNA complementary to target shown as yellow circles, NTC as black squares. Data points represent the mean, and error bars represent the standard deviation of n = 2 technical replicates. c: Fluorescence curves corresponding to Fig. 4c. In c-e, each subplot shows results for different targets, with several targets sometimes grouped together in a single subplot and shown in different colors. Different crRNAs detecting these targets are shown as different line styles (solid/dashed/dotted). d: Fluorescence curves corresponding to Fig. 4f. e: Fluorescence curves corresponding to Fig. 4i.

Extended Data Fig. 10 |. Discrimination of SARS-CoV-2, IAV, and KRAS variants.

In a-d, each subplot shows results for different targets, with several targets sometimes grouped together in a single subplot and shown in different colors. Different crRNAs detecting these targets are shown as different line styles (solid/dashed). a: Fluorescence curves corresponding to Fig. 4b. b: Fluorescence curves corresponding to Fig. 4d. c: Fluorescence curves corresponding to Fig. 4e. d: Fluorescence curves corresponding to Fig. 4g. e: Fluorescence curves corresponding to Fig. 4j. Here, different crRNAs are shown as different colors, and each subplot corresponds to a single target.

Fig. 1 |. RNA structure leads to a reduction in Cas13 activity because of strand displacement.

a, Different amounts of structure were introduced into the target through intramolecular structure, RNA oligos (occluders) and DNA occluders. The resulting fluorescence kinetic curves are shown. Target input concentration: 7.5 × 10⁸ cp per μl. Data from n = 3 technical replicates are shown (dots). b, Scatter plot comparing the impact of the different occlusion types depicted in a on Cas13 activity (x axis, Cas13 activity when occluded by DNA oligos; y axis, Cas13 activity when occluded by intramolecular RNA or RNA occluders). c, Cas13 activity versus occluder length for RNA occluders compared to two models: an equilibrium model based on crRNA–target hybridization free energies (light green) and a strand displacement model (dark green). Effects of changing parameters are indicated by arrows. The gray bar represents the NTC. d, Overview of strand displacement reactions. After initial binding to part of the target (blue), the crRNA (pink) and occluder (red) undergo a random walk process until one or the other is fully displaced. Displacement of the occluder leads to Cas13 activation. In b,c, the mean is displayed, with error bars showing the s.d. across n = 3 technical replicates.

Fig. 2 |. A massively multiplexed assay modulating secondary structure.

a, Overview of the multiplexed assay, in which a total of 4,608 simultaneous assays were performed, with oligo occluders of lengths 10, 14, 21 and 28 nt tiling each protospacer region in 3-nt increments. b, Overview of the Cas13 activity data from the multiplexed assay. Each data point represents the mean activity resulting from averaging four timeseries curves (Methods), normalized to the nonoccluded condition; positive and negative controls are not shown (Extended Data Fig. 3). Dashed lines represent the start and end of the protospacer. c, Heat map showing the degree of activity reduction (darker greens) by each 21mer and 28mer occluder. d, Cumulative histogram of inhibition asymmetry, defined as the ratio of activities when the same numbers of nucleotides are occluded at the 3′ versus 5′ ends of the protospacer (Extended Data Fig. 4c,d). e, Normalized Cas13 activity for 21mer (green) and 28mer (purple) occluders with different start positions in the region around the protospacer. Each line represents one crRNA. f, Strand displacement model prediction for activity as a function of occluding oligo, colored by the fraction of seed or switch regions occluded by the oligo. g, EMSA showing Cas13–crRNA complex with various combinations of target and occluders. P and 3′ represent 28mer occluders overlapping the protospacer or the region 3′ to the protospacer, respectively. A representative image of three replicate gels is shown. h, Bar chart showing the inhibitory effect of 28mer occluders overlapping the protospacer or the region 3′ to the protospacer, at different occluder concentrations and when annealing the occluder before or at the same time as the crRNA. In b–e, the mean of n = 4 replicates (two biological and two technical) is displayed. In e, error bars show the s.d. across n = 4 replicates (two biological and two technical). In h, the mean is displayed, with error bars showing the s.d. across n = 3 technical replicates.

Fig. 3 |. Designed secondary structure enhances Cas13 mismatch detection.

Unless otherwise noted, the assay duration was 180 min. a, Schematic showing strand displacement by Cas13 with a perfectly matched target sequence versus one containing a mismatch. b, ODE-based model predictions of crRNA–target hybridization kinetics with and without occlusion and mismatches. c, Kinetic curves showing detection of a target sequence with and without a single A>U mismatch at spacer position 5, in the presence and absence of occlusion. The shaded region shows the range of fluorescence measurements for each condition across replicates. d, Maximum fluorescence ratios with and without occlusion at a variety of target input concentrations (Methods). Error bars represent the error as measured by the s.d. of n = 3 technical replicates propagated through the ratio calculation (Methods). e, Violin plots showing the ability of Cas13 to distinguish between WT targets (for crRNAs numbered as in Fig. 2c) and targets containing mutations at four different positions in the protospacer both with and without occlusion; the position is relative to the 5′ end of the protospacer. Each data point is the discrimination ratio of a perfectly matched to mismatched sequence (Methods). f, Data from e, but organized by mutation type. g, Heat map showing the ability of Cas13 to detect spiked-in target in a background of mismatched sequence at decreasing allele frequencies, both with and without occlusion. Asterisks indicate statistically significant detection over the no-spike-in control. Significance was determined using a one-tailed t-test at P < 0.05. Activity discrimination is defined analogously to mismatch discrimination (Methods). h, Specificity matrix showing Cas13 activity normalized for each target to its corresponding crRNA, with and without occlusion, for all possible crRNA and target nucleotides at position 5. In g,h, the means across n = 3 and n = 2 technical replicates, respectively, are shown.

Fig. 4 |. crRNA occluders enable consistent and sensitive mismatch discrimination in diagnostic settings.

a, Schematic of detection workflow. b, Detection of Delta and Omicron SARS-CoV-2 spike gene RNA from amplified viral seedstocks. The fluorescence at the time point corresponding to the maximum discrimination ratio is shown, normalized independently for each target to its maximum. c, Cas13-based discrimination of ancestral (627E) IAV variant from multiple mammalian-adapted (627K) strains, as well as the rare 627V variant, all distinguished by a single-nucleotide substitution. In c,d, each target’s final fluorescence at 180 min was normalized independently to its maximum. d, Discrimination of a single-nucleotide substitution in IAV strains conferring oseltamivir resistance in six different isolates, using a single guide pair per NA subtype. e, Variant detection in US samples infected with Delta and Omicron strains of SARS-CoV-2. In e–g, the final fluorescence (x axis) is used to distinguish positive from negative calls, whereas the maximum fluorescence ratio (y axis) is used to distinguish variants. -, not detected. f, Discrimination of 627E from 627K in UK samples infected with seasonal IAV. g, Variant detection in Dutch samples infected with oseltamivir-sensitive (H) and oseltamivir-resistant (Y) single-nucleotide IAV variants. h, Discrimination of Delta versus Omicron SARS-CoV-2 variants in samples using a simple fluorescence readout. i, Discrimination of IAV 627E from 627K variants in samples and clinical isolates from persons tested positive for H5N1 since 2023 in Cambodia. The fluorescence at the time point corresponding to the maximum discrimination ratio was normalized independently for each target to its maximum. j, Using occluded Cas13 to distinguish seven variants of codon 12 of the KRAS gene with mCARMEN. The final fluorescence at 180 min was normalized independently for each target to its maximum. In b–d,i,j, the means across n = 2 technical replicates are shown.