TracrRNA reprogramming enables direct PAM-independent detection of RNA with diverse DNA-targeting Cas12 nucleases – Nature.com

Cas12 nucleases offer diverse yet complex opportunities for tracrRNA reprogramming

Type V CRISPR-Cas systems comprise numerous systems that involve tracrRNAs that could be amenable to tracrRNA engineering. Of the 14 subtypes of type V systems defined to-date, eight (associated with Cas12b, Cas12c, Cas12d, Cas12e, Cas12f1, Cas12g, Cas12k and Cas12l) exclusively rely on a tracrRNA for gRNA biogenesis (Fig.1b)5,6,11,34,35,36,37,38,39,40,41,42,43. For the remaining systems, the Cas12 nuclease directly recognizes and processes the transcribed repeat, as commonly demonstrated for Type V-A systems and its Cas12a nuclease7,11,44,45,46,47. Apart from collateral cleavage activity9,12,35,38,48, some or all of these DNA-targeting nucleases possess features distinct from more traditional Cas9 nucleases, such as pre-crRNA processing38,49,50, compact nucleases39,40,41,42,48,51, higher optimal temperatures9,36,52, crRNA-guided transposition37,53 and T-rich and C-rich PAM recognition6,11,43,49.

TracrRNA reprogramming involves engineering the anti-repeat region to hybridize with an RNA-of-interest while maintaining the essential sequence and structural features of the natural repeat/anti-repeat (R/AR) duplex recognized by the Cas nuclease. While Cas9-associated RNA duplexes form a simple 2540bp stem typically interrupted by a small bulge54,55,56, Cas12-associated RNA duplexes adopt distinct and more complicated conformations (Fig.1c). In addition to a long repeat/anti-repeat (LR/AR) stem often containing an intervening bulge, the reported duplexes associated with Cas12b, Cas12f1, Cas12g and Cas12l also possess a pseudoknot that includes a 57bp short repeat/anti-repeat (SR/AR) stem6,15,43,57,58,59,60. For Cas12e and Cas12k, the reported RNA duplexes possess a 3-bp triple helix formed by two portions of the anti-repeat sandwiching the repeat in addition to the bulged LR/AR stem35,53,61,62 (Fig.1c and Supplementary Fig.1). Finally, for Cas12c, the reported RNA duplexes form three 4-7bp disjoint R/AR stems49 (Fig.1c). Given the diversity and complexity of these RNA duplexes, we explored the extent to which the RNA duplexes associated with these diverse tracrRNAs can be reprogrammed for RNA detection.

We started with the Bacillus hisashii Cas12b (BhCas12b) due to the relative simplicity of its RNA duplex comprising a 30-bp LR/AR stem with an intervening bulge, and a 5-bp SR/AR duplex between the LR/AR and the guide36,57 (Fig.2a). We sought to investigate the reprogrammability of both stems using a Cas12 cleavage assay conducted with a cell-free transcription-translation (TXTL) system63. As part of the assay, purified BhCas12b protein, a gRNA-expressing plasmid and a plasmid encoding the PAM-flanked dsDNA target upstream of a GFP reporter construct were added to a reaction, and we monitored fluorescence over time. Cleavage of the reporter construct leads to loss of GFP expression through rapid degradation of the linearized DNA (Fig.2b).

a Predicted tracrRNA-crRNA structure for BhCas12b based on its ortholog BthCas12b (PDB: 5WTI57). R/AR, repeat/anti-repeat. b Setup to assess Rptr functionality using cell-free transcription-translation (TXTL). Expressed Cas-guide RNA complex recognizes and cuts its dsDNA target, causing the degradation of target-encoding GFP reporter plasmid and resulting in lower fluorescence compared to a non-targeting guide control. c 16-hourendpoint fluorescence measurements in TXTL when changing the long and short RNA duplexes. NT, non-targeting guide; T, targeting guide; T-br, targeting crRNA with bulge removed. d Setup to reprogram tracrRNAs to sense a Campylobactor jejuni transcript CJ8421_04975 mRNA. The guide and target components are added in the form of DNA constructs, while the purified BhCas12b protein is used. mRNA(mut), mRNA with point mutations in the predicted seed region of the guide. Rptr(scr-LA), Rptr with the long anti-repeat sequence scrambled; Rptr(scr-SA), Rptr with the short anti-repeat sequence scrambled; Rptr(scr-LA&SA), Rptr with both long and short anti-repeat sequence scrambled. e 16-hourendpoint fluorescence measurements in TXTL when assessing Rptr-guided sequence-specific dsDNA targeting. Nucleotide changes in R/AR stems in c and d are indicated by gray boxes. Bars and error bars in c, e represent the mean and standard deviation from three independently mixed TXTL reactions. Dots represent individual measurements. ***p<0.001 based on a one-sided Students t-test with unequal variance (n=3). Source data are provided as a Source Data file.

To interrogate the reprogrammability of the crRNA-tracrRNA duplex, we began with the intervening bulge in the crRNA-tracrRNA duplex followed by the two stems. Previous studies showed that a bulge in the LR/AR duplex is necessary to maintain the dsDNA targeting activity for SpyCas9 and Sth1Cas955. However, removing this bulge from the LR/AR associated with BhCas12b did not impinge on GFP silencing (Fig.2c), likely due to the bulge falling outside of the nuclease binding region57. Using the bulge-removed variant as a baseline to interrogate programmability of the LR/AR and SR/AR RNA stems (Fig.2c), we found that both stems could be reprogrammed without impinging on GFP silencing, whether changing the lower or upper portion of the LR/AR stem (cr1-4) or the SR/AR stem (cr5-6). The crRNA-tracrRNA duplex could be similarly reprogrammed for the Bacillus thermoamylovorans Cas12b (BthCas12b)57, as changes in the LR/AR and SR/AR of a fusedsingle-guide RNA (sgRNA) were well tolerated (Supplementary Fig.2). Therefore, tracrRNAs associated with Cas12b nucleases are highly amenable to reprogramming.

We next explored the extent to which the BhCas12b Rptr could be applied for RNA detection. We started with the CJ8421_04975 mRNA previously used to evaluate Rptrs associated with different Cas9 nucleases31. Two Rptrs hybridizing at different loci of CJ8421_04975 mRNA were designed based on rules derived from our mutational analysis of the LR/AR and SR/AR (Fig.2c and Supplementary Fig.3). Strong GFP silencing was observed for both BhRptr1 and BhRptr2 when compared with the non-targeting crRNA control, and both Rptrs combined with the mRNA exhibited similar performance as their equivalent crRNA/tracrRNA pairs (Fig.2d, e). Furthermore, dsDNA targeting occurred specifically through the predicted guide sequence, as mutating the predicted seed region or scrambling the tracrRNA anti-repeat (long, short or both) fully inhibited GFP silencing. The one exception was scrambling the anti-repeat of the tracrRNA associated with locus 1 of the mRNA, which still maintained substantial targeting activity likely due to shifted base pairing in the short duplex (Fig.2d, e). Overall, the Cas12b tracrRNA can be reprogrammed to link an RNA-of-interest to sequence-specific dsDNA targeting.

Building on the reprogramming of Cas12b tracrRNAs, we turned to the Acidibacillus sulfuroxidans Cas12f1 (AsCas12f1) from Type V-F CRISPR-Cas systems39,64. While its crRNA-tracrRNA duplex parallels that associated with BhCas12b (Fig.3a), AsCas12f1 is a much smaller protein and forms a homodimer when binding a single crRNA-tracrRNA duplex. Using the TXTL assay with plasmid-expressed AsCas12f1 and an sgRNA, we found that the intervening bulge was also dispensable and the LR/AR and SR/AR could be fully reprogrammed without impinging on GFP silencing (Fig.3b). The base-pairing in the SR/AR was crucial for dsDNA targeting, as deletion of the SR portion of the SR/AR or mismatches in the SR/AR substantially inhibited GFP silencing (Fig.3b). We further demonstrated that three Rptrs designed to hybridize to different loci in the CJ8421_04975 mRNA yielded GFP silencing with comparable performance as their equivalent crRNA/tracrRNA counterparts in TXTL (Fig.3c, d). As before, mutating the seed region in the predicted guide or scrambling the tracrRNA anti-repeat (long, short or both) fully inhibited GFP silencing.

a AsCas12f1 sgRNA structure (PDB: 8J1264). See the detailed information in Supplementary Fig.1. b 16-hourendpoint fluorescence measurements in TXTL when reprogramming the long and short RNA duplexes in theAsCas12f1 sgRNA. NT, non-targeting crRNA; T, targeting crRNA. c Setup to detect the Campylobacter jejuni transcript CJ8421_04975 mRNA using AsCas12f1 Rptrs in TXTL. d 16-hourendpoint fluorescence measurements in TXTL for Rptr-guided sequence-specific dsDNA targeting by AsCas12f1 in TXTL. e Structure of DpbCas12e sgRNA (PDB: 6NY3)35. In the triple-helix region, a cis Hoogsteen/Watson-Crick base pair is formed between the U.A and a cis Watson-Crick/Watson-Crick base pair between the A-U. f 16-hourendpoint fluorescence measurements in TXTL when assessing the changeability of the LR/AR region. Dpb_T-br, targeting sgRNA with thebulge and G.U wobble base pair removed. g 16-hourendpoint fluorescence measurements in TXTL when changing the RNA triple-helix region. h 16-hourendpoint fluorescence measurements in TXTL when changing the RNA triple-helix surrounding region. i, Setup to detect the Campylobacter jejuni CJ8421_04975 mRNA using DpbCas12e Rptrs in TXTL. j, 16-hour endpoint fluorescence measurements for Rptr-guided sequence-specific dsDNA targeting by DpbCas12e in TXTL. Rptr(scr-dplx), Rptr with a scrambled anti-repeat sequence; Rptr(scr-tplx), Rptr with the RNA triple-helix sequence scrambled; Rptr(scr-d&tplx), Rptr with the RNA duplex and triple-helix sequence scrambled. Nucleotide changes in AsCas12f1 sgRNA and DpbCas12e sgRNA in b, f, g and h are indicated by gray boxes. Bars and error bars in b, d, f, g, h, and j represent the mean and standard deviation from three independently mixed TXTL reactions. Dots represent individual measurements. No error bars are shown when only two replicates were successfully collected. *: p<0.05. **: p<0.01. ***:p<0.001 based on a one-sided Students t-test with unequal variance (n=3). Source data are provided as a Source Data file.

Deviating from Cas12b and Cas12f1, Cas12e nucleases rely on crRNA-tracrRNA duplexes containing an RNA triple helix instead of a pseudoknot (Fig.3e and Supplementary Fig.1)35,53,61,62, posing an even greater challenge for RNA detection with PUMA. We selected the previously characterized Deltaproteobacteria Cas12e (DpbCas12e)35 and evaluated the reprogrammability of the bulged stem as well as the triple helix. Paralleling BhCas12b and AsCas12f1, removing the bulge and a G-U wobble pair in the context of an sgRNA did not compromise GFP silencing, and the stem could be readily reprogrammed (Fig.3f). Turning to the triple helix, this helix is formed by two separate tracts of three uracils at the 5 end of the tracrRNA sandwiching three adenosines in the repeat (Fig.3e and Supplementary Fig.1)35. A cis Hoogsteen/Watson-Crick base pair forms between the U.A and a cis Watson-Crick/Watson-Crick base pair forms between the A-U, assigning the triple helix to the cWW/cHW triple family65. RNA triple-helix motifs are found in various functional RNAs, such as telomerase RNAs66,67, riboswitches68 and long noncoding RNAs69,70. Despite its diversified distribution, the changeability of RNA triple helix in these biologically important RNAs has not been systematically investigated.

We reasoned that other RNA triple helices in the same cWW/cHW family might preserve dsDNA targeting by DpbCas12e. Using the RNA Base Triple Database as a reference71, we tested all nine RNA triple helix combinations reported in existing functional RNAs (s11-s18, and the native U.A-U), three expected to form a triple helix but not observed to-date (s19-21), and two not expected to form a triple helix and not observed to-date (s22-23, Fig.3g). Among the 14 tested triple-helix combinations, two (C.G-U_s11, C.G-C_s18) yielded GFP silencing comparable to that of the native U.A-U. In addition, installing the combination of U.A-U and C.G-C base triples in the RNA triple-helix region (s24-26) also yielded comparable GFP silencing. As expected, disrupting the RNA triple-helix conformation in one of thethree triples in the triple-helix region abolished dsDNA targeting (s27-32, Fig.3g), indicating a stringent triple-helix conformation required by DpbCas12e.

The RNA triple-helix region is surrounded by one C-G base pair at the 3 end and three unpaired nucleotides (AUC) at the 5 end of the repeat that may also represent necessary sequence or structural features (Fig.3h). For the C-G base pair, we found that introducing a C.A mismatch (s33) fully abolished silencing, while changing the base pair to U-A (s34) only modestly reduced GFP silencing (Fig.3h). For the AUC at the 5 end, mutating the C to A, G and U (s35-s38) resulted in similar or even improved GFP silencing (Fig.3h). The U could also be replaced with other nucleotides (s39-s41) without compromising activity (Fig.3h). Changing the A to C or G (s42, s44) was also well tolerated, while changing the A to U (s43) substantially inhibited GFP silencing (Fig.3h). Together, the RNA duplex and triple-helix regions are reprogrammable, albeit with less flexibility for the triple-helix region (Supplementary Fig.2).

Based on the insights from the systematic mutational analyses to DpbCas12e sgRNA, we designed three Rptrs targeting different loci in CJ8421_04975 mRNA (Fig.3i). We observed substantial GFP silencing for all three designed Rptrs, with comparable performance to that of their equivalent crRNA:tracrRNA pairs. As before, mutating the seed region in the predicted guide or scrambling the tracrRNA anti-repeat inhibited GFP silencing (Fig.3j). Overall, Rptrs could be extended to different Cas12 nucleases with varying tracrRNA-crRNA structures.

In contrast to Cas9, Cas12 non-specifically cleaves ssDNA upon target recognition, enabling signal amplification as part of CRISPR-based diagnostics2. We therefore reasoned that combining Rptrs, dsDNA targets, and ssDNA reporters would couple RNA detection by Cas12 to an amplified readable outputthe basis of PUMA. To assess the collateral effects of BhCas12b, we devised an in vitro collateral cleavage assay using purified BhCas12b protein, in vitro-transcribed sgRNAs or sensed RNAs and Rptrs, linear dsDNA targets and a ssDNA fluorophore-quencher reporter (Fig.4a). Upon recognition and cleavage of its dsDNA target, the nuclease non-specifically cleaves the fluorophore-quencher reporter, resulting in an increase in fluorescence.

a Schematic ofthe in vitro trans-cleavage assay. The assay includes purified aCas12 nuclease, anin vitro transcribed Rptr, and alinear dsDNA target. The Cas12-guide RNA ribonucleoprotein (RNP) recognizes and cleaves its dsDNA target, which triggers non-specific cleavage activity on ssDNA. Specifically, cleavage of the non-target strand (NTS) occurs before cleavage of the target strand (TS). F, fluorophore; Q, quencher. Yellow circle, PAM; b Impact of unprocessed or processed targets on in vitro trans-cleavage activity by BhCas12b. TS cleavage is the rate-limiting step. Red arrow, cleavage site. The cleavage site of TS is set as position 0. -, truncating the target sequence on NTS or TS. +, adding an overhang on NTS or TS. The PAM is in brown and the target is in blue. c, Direct detection of the full-length CJ8421_04975 mRNA by BhCas12b based on in vitro collateral cleavage activity. Yeast RNA is added in the same mass amount as the 1000nM sensed mRNA, and the best-performing dsDNA target NTS-6: TS-2 is used. d, Impact of unprocessed or processed targets on in vitro collateral cleavage activity by DpbCas12e. e Direct detection of the full-length CJ8421_04975 mRNA by DpbCas12e based on in vitro collateral cleavage activity. Yeast RNA is added in the same mass amount as the 1000nM sensed mRNA, and the best-performed dsDNA target NTS-8: TS-4 is used. 16h end-point values were used to make theplots in c and e. See Supplementary Figs.6b and 10a for the complete time courses. Curves in b and d represent the mean from two independent collateral assays. Bars anddots in c and e represent the mean andindividual measurements, respectively, from two independent collateral cleavage assays. Light blue bars indicate the limit-of-detection (LOD) conservatively estimated as the lowest concentration yielding an average fluorescence exceeding 50% of that of the no-RNA control. Source data are provided as a Source Data file.

We began with an sgRNA and a 334-bp linear dsDNA containing a 27-bp PAM-flanked target, with the resulting in vitro reaction conducted at 37C (Supplementary Fig.4a). We observed slight background fluorescence without the dsDNA target and monotonically increasing fluorescence with the dsDNA target that plateaued after 12hours (kobs=0.03h1, Supplementary Fig.4b, Supplementary Data1), in line with cis-cleavage of the dsDNA target triggering multi-turnover collateral cleavage of the fluorescent ssDNA reporter by BhCas12b. The activity exhibited by BhCas12b was weaker compared to that by FnCas12a (kobs=0.11h1), DpbCas12e (kobs=0.65h1) and LbCas12a (kobs=2.1h1) under equivalent conditions (Supplementary Fig.4b, Supplementary Data1). Elevating the temperature from 29 to 42C increased the reaction rate by 1.7-fold (kobs=0.19h1 at 42C) (Supplementary Fig.4c, Supplementary Data1), in line with higher temperatures yielding optimal cleavage activity for Cas12 nucleases9,36,52.

With an in vitro collateral cleavage assay in place, we next turned to the dsDNA target. Standard Cas12-based diagnostics have little control over the composition of the dsDNA target without extensive manipulations. In contrast, the dsDNA targetis provided as part of PUMA, granting complete control over its sequence, length, and chemistry. This control in turn could be leveraged to enhance the reaction. As a start, we evaluated the impact of using targets encoded on shorter linear DNA, perceivably by reducing the search time for the target sequence. In line with this rationale, we observed a 5.6-fold increase in collateral cleavage activity at 37C when shortening the dsDNA target length from 334bp (kobs=0.03h1) to 94bp (kobs=0.17h1). However, collateral cleavage activity decreased when shortening the DNA length to 60bp (kobs=0.12h1) or to 48bp (kobs=0.08h1) (Supplementary Fig.4d, Supplementary Data1). We also tested ssDNA targets, which exhibited at least a 2-fold increase in collateral activity than dsDNA targets of equivalent size (Supplementary Fig.5), in line with circumventing PAM recognition and DNA unwinding. We continued to use dsDNA targets though due to their more stringent and specific target recognition2.

The observed impact of DNA length on signal production led us to explore a distinct aspect of the dsDNA target: the extent of cleavage by Cas12 nucleases. Upon target recognition, Cas12 nicks the non-target strand followed by the target strand of the dsDNA target through the nucleases RuvC domain72,73, leading to a cleaved dsDNA target with a 5 overhang (Fig.4a). Complete cleavage of the dsDNA target normally precedes collateral cleavage72, with target strand cleavage posing the rate-limiting step36,73,74,75. We therefore hypothesized that using a dsDNA target with a processed target strand would increase the observed rate of collateral cleavage. In line with this hypothesis, a dsDNA target with a processed non-target strand yielded similar collateral cleavage rates to that of an unprocessed dsDNA target (kobs=0.06 - 0.16h1) at 37C for dsDNA lengths ranging between 45 and 94bp (Fig.4b, Supplementary Fig.6a, b, and Supplementary Data1). In contrast, a dsDNA target with a processed target strand yielded increased collateral cleavage rates (kobs up to 0.46h1 for NTS+55: TS+0), in line with target strand cleavage posing the rate-limiting step (Fig.4b and Supplementary Fig.6a, b). A similar collateral cleavage rate (kobs=0.44h1) was observed for a dsDNA target with both strands processed (NTS-6: TS+0) (Supplementary Fig.6a, b). Finally, trimming the target strand by two additional nts towards the PAM can further enhance the observed collateral cleavage activity (NTS-6: TS-2, kobs=0.56h1) (Supplementary Fig.6a, b). With conditions established for enhanced RNA detection using BhCas12b, we turned to detecting the full-length CJ8421_04975 mRNA using BhRptr4 in vitro. Under the optimal conditions with the shortest and processed dsDNA target (NTS-6:TS-2) at 42C, the sensed mRNA was detected at 1M in 45minutes and at 10nM in 16hours based on endpoint measurements compared to a no-RNA control (Fig.4c and Supplementary Fig.6c).

The optimized experimental setup with BhCas12b allowed us to assess how the ability of the sensed RNA and Rptr to hybridize impacts collateral cleavage activity. One potential factor is the formation of internal secondary structures that hinder hybridization. To test this factor directly, we introduced extensions to the 5 extensions to the ncrRNA associated with BhRptr4 (Supplementary Fig.7). The hairpins reduced collateral cleavage activity, with an internal hairpin inhibiting more strongly than a flanking hairpin. In the absence of these structures, introducing an annealing step did not enhance collateral cleavage activity (Supplementary Fig.8). Of note, collateral cleavage activity resulting from pairing of thepartial CJ8421_04975 mRNA fragment and BhRptr4 was higher than that obtained with the equivalent sgRNA, indicating that hybridization between a sensed RNA and Rptr is not necessarily a bottleneckto RNA detection.

With factors influencing RNA detection with BhCas12b established, we asked whether increasing the reaction temperature and truncating the dsDNA target also apply to DpbCas12e, which exhibited much higher collateral cleavage activities (Supplementary Fig.4). We tested DpbCas12e with DpbRptr1 against the full-length CJ8421_04975 mRNA along with different-sized dsDNA targets at different temperatures (Supplementary Fig.9ac). Similar to BhCas12b, DpbCas12e exhibited increased activity when elevating the temperature (kobs=0.07h1 at 29C, 0.43h1 at 37C and 0.67h1 at 42C) (Supplementary Fig.9b) and when shortening the length of the dsDNA target (kobs=0.43h1 for 331bp and 0.54h1 for 44bp) (Supplementary Fig.9c). Moreover, introducing a processed dsDNA target increased the collateral cleavage rate (for 91-bp target, kobs=0.46h1 for unprocessed strands, 0.78h1 for processed target strand) (Fig.4d and Supplementary Fig.10a). Finally, under the optimized conditions using the double-strand processed 38-bp dsDNA target at 42C, the sensed mRNA could be detected at a concentration of 1M in 9minutes and at a concentration of 0.1nM in 16hours (Fig.4e and Supplementary Fig.10b). Therefore, different tracrRNA-dependent Cas12 nucleases can be co-opted for direct, PAM-independent RNA detection in vitro.

When comparing collateral cleavage activities across Cas12 orthologs (Supplementary Fig.4b), we noticed that the LbCas12a-gRNA complex produced substantial fluorescence even in the absence of its corresponding dsDNA target, reaching approximately 70% of the levels seen when its dsDNA target is present after 16hours of incubation (Supplementary Fig.4b). A high background activity was also reported for AsCas12a in previous studies76,77. To assess the prevalence of this background activity, we tested four BhCas12b sgRNAs (#1-4, with the #4 guide used with other Cas12 orthologs in FigureS4b) using processed dsDNA targets (NTS-6: TS-2). Substantial DNA target-independent activity was observed for sgRNA#1 and #2, with comparable fluorescence levels to those with the dsDNA targets after 16hours (Fig.5a). Intriguingly, sgRNA#1 exhibited high cleavage activity (kobs=1.02-1.16h1) regardless of the presence or absence of the dsDNA target. This phenomenon was not isolated, as 5 out of 10 additional sgRNAs we tested exhibited DNA target-independent collateral activity higher than that of sgRNA#4 (Supplementary Fig.11).

a, Measured in vitro collateral cleavage activity with BhCas12b and an sgRNA with or without a dsDNA target. b, Measured in vitro collateral cleavage activity with BhCas12b and a Rptr and a dsDNA target with or without the sensed RNA. c, Sensitivity comparison between sgRNA and Rptr. In a-b, 37-bp NTS-2:TS-2 processed dsDNA targets were used for both sgRNA and Rptr. In c, 334-bp DNA fragments containing the core PAM-flanking target were used with the sgRNAs and 37-bp NTS-2:TS-2 processed dsDNA targets were used with the Rptrs. In a-c, sgRNA#1 and sgRNA#4 share the same guide sequences as those generated by Rptr#1 and Rptr#4, respectively. Dots represent individual measurements from two independent collateral cleavage assays. Bars represent the mean of the dots. In a-b, values represent fluorescence measurements after reaction times of 2hours and 16hours. In c, values represent fluorescence measurements after reaction times of 16hours. Light blue bars indicate the limit-of-detection (LOD) conservatively estimated as the lowest concentration yielding an average fluorescence exceeding 50% of that of the no-RNA control. Source data are provided as a Source Data file.

This DNA target-independent collateral activity would reduce the sensitivity of nucleic-acid detection, making it more challenging to identify low-concentration biomarkers. In contrast, we hypothesized that any background activities would be greatly reduced using Rptrs, as the guide RNA is principally formed only in the presence of the sensed RNA. Supporting this hypothesis, combining a sensed RNA and Rptr for BhCas12b drove collateral activity even in the absence of the DNA target (Supplementary Fig.12). In the absence of the sensed RNA, each Rptr alone resulted in endpoint fluorescence levels 3.5-fold to 29.9-fold lower than those observed in the presence of the corresponding sgRNA (Fig.5b). Based on this difference, we directly compared the sensitivity of BhCas12b detecting dsDNA with an sgRNA or detecting the equivalent RNA with a Rptr. The limit-of-detection was around 10-fold lower using a Rptr than an sgRNA for one site (#4), while RNA detection (with a Rptr) but not DNA detection (with an sgRNA) was possible at another site (#1) (Fig.5c). Thus, the sensitivity of nucleic-acid detection with Cas12 nucleases can be enhanced by detecting RNA with Rptrs rather than detecting DNA with sgRNAs, at least depending on the nuclease and detected sequence.

Given the enhanced sensitivity when detecting RNA with PUMA versus dsDNA traditionally detected with Cas12 nucleases, we asked how PUMA compares to the two standard CRISPR-based diagnostic approaches DETECTR for DNA detection with Cas122 and SHERLOCK for RNA detection with Cas134 (Supplementary Fig.13). We chose to detect three loci within the CJ8421_04975 DNA/mRNA and used sensitivity as the basis of comparison. BhCas12b was used for both PUMA and DETECTR to ensure a direct comparison, while PbuCas13b was used for SHERLOCK78. No pre-amplification was included to directly gauge the sensitivity associated with each Cas nuclease. Two of the sites lacked the PAM recognized by BhCas12b, in line with the requirement for a PAM inherent to DETECTR. Of the detected loci, the three approaches performed similarly, with the measured limit of detection either at 1nM or 10nM. Thus, PUMA can perform similarly to DETECTR and SHERLOCK, at least with the tested Cas nucleases, with PUMA targeting a broader range of sites than DETECTR.

One core feature of Rptrs is that base pairing with a sensed RNA is somewhat flexible, whereas the flanking guide sequence should direct dsDNA targeting that is highly sensitive to mismatches31. To exploit this feature, we applied Cas12 Rptrs to differentiate bacterial pathogens based on their 16S rRNA79,80. Differentiating pathogens can be important to select appropriate courses of treatment for different indications such as acute sepsis, urinary tract infections, or sexually transmitted diseases. Traditional CRISPR diagnostics based on collateral cleavage by Cas12 or Cas13-based diagnostics have taken strides in this direction81, with one example using multiple guide RNAs to detect different bacterial pathogens82. In contrast, with this core feature of Rptrs, a single Rtpr could be designed to pair next to a variable region of 16S rRNA indicating the genus. The variable region would then be matched to a dsDNA target, with its cleavage and subsequent collateral activity indicating which pathogen is present.

To determine how to best design the Rptr, we began by evaluating the specificity of the three different Cas12b homologs (BthCas12b, BhCas12b and AacCas12b), with the goal of identifying at least one homolog exhibiting high guide-target mismatch sensitivity. We assessed collateral cleavage activity of each homolog using a Rptr-sensed RNA encoding the same guide sequence along with dsDNA targets containing two consecutive mismatches sliding through the guide-target region (Fig.6a). Among the three orthologs, BthCas12b was the most sensitive to guide-target mismatches, especially in positions 5-12 proximal to the PAM in which the mismatches reduced collateral cleavage activity between 102-fold and 105-fold (Fig.6a). We also evaluated the extent to which BthCas12b accepts mismatches between the sensed RNA and the Rptr (Supplementary Fig.14). Mismatches in the long or short repeat consistently reduced but rarely eliminated activation of collateral cleavage activity even with four consecutive mismatches. This flexibility lends to pairing with conserved 16S rRNA regions with some variability, even if unintended RNA duplexes bound by Cas12 could be generated in the process. We therefore proceeded with BthCas12b and aimed for sequence differences to fall within the most sensitive positions of the target.

a Tolerance of guide-target mismatches for three different Cas12b orthologs based on in vitro collateral cleavage activity. The DNA target is the same as the one used in Fig.4B (BhsgRNA4 DNA target). Heat maps represent the mean kobs valuesfrom two independent collateral assays. See the kobs values in Supplementary Data1. b Setup to differentiate 16S rRNA from five different pathogens using only one universal BthCas12b Rptr binding to a conserved region of 16S rRNA. A truncated long anti-repeat of 18 nts instead of the usual 31 nts is used in the universal Rptr. In the alignment, sequences that match the E. coli 16S rRNA are in black, while those that do not match are shown in red. c Detection of pathogen 16S rRNAs with a universal Rptr and corresponding dsDNA targets based on in vitro collateral cleavage activity with BthCas12b. Partial 16S rRNA fragments of different pathogens at a final concentration of 100nM were used. Values represent 36-minute reaction times. Values in c represent the mean and standard deviation from two independent collateral assays. Source data are provided as a Source Data file.

Following this approach, we designed a single BthCas12b Rptr that hybridizes to a conserved region of bacterial 16S rRNA, with the downstream variable region serving as the guide sequence. We specifically focused on five common bacterial pathogens, E. coli, Klebsiella pneumoniae, Staphylococcus aureus, Enterococcus faecalis, and Listeria monocytogenes (Fig.6b), where the sequence differences fall within the region of mismatchsensitivity for BthCas12b. As before, a PAM did not need to appear within the sensed RNA, as this was encoded within the dsDNA targets. We then assessed collateral cleavage activity for each 16S rRNA fragment and each dsDNA target. The fragment was introduced at a final concentration of 100nM, reflecting the output of isothermal pre-amplification and in vitro transcription2. The presence of the 16S rRNA fragment from one specific pathogen triggers fluorescence release only when paired with its corresponding dsDNA target (Fig.6c and Supplementary Fig.15). We noticed 16S rRNA from L. monocytogenes also gave rise to substantial fluorescence when pairing with the dsDNA target from S. aureus, likely due to the high similarity between their 16S rRNA fragments with only three mismatches present outside of the seed region (Fig.6b, c and Supplementary Fig.15). Thus, specific detection of different pathogens based on 16S rRNA can be achieved via a single Rptr.

Read this article:
TracrRNA reprogramming enables direct PAM-independent detection of RNA with diverse DNA-targeting Cas12 nucleases - Nature.com