Structures, mechanisms and applications of RNA-centric CRISPR–Cas13

Abstract

Prokaryotes are equipped with a variety of resistance strategies to survive frequent viral attacks or invading mobile genetic elements. Among these, CRISPR–Cas surveillance systems are abundant and have been studied extensively. This Review focuses on CRISPR–Cas type VI Cas13 systems that use single-subunit RNA-guided Cas endonucleases for targeting and subsequent degradation of foreign RNA, thereby providing adaptive immunity. Notably, distinct from single-subunit DNA-cleaving Cas9 and Cas12 systems, Cas13 exhibits target RNA-activated substrate RNase activity. This Review outlines structural, biochemical and cell biological studies toward elucidation of the unique structural and mechanistic principles underlying surveillance effector complex formation, precursor CRISPR RNA (pre-crRNA) processing, self-discrimination and RNA degradation in Cas13 systems as well as insights into suppression by bacteriophage-encoded anti-CRISPR proteins and regulation by endogenous accessory proteins. Owing to its programmable ability for RNA recognition and cleavage, Cas13 provides powerful RNA targeting, editing, detection and imaging platforms with emerging biotechnological and therapeutic applications.

In many bacteria and most archaea, CRISPR–Cas systems provide adaptive immunity by acquiring immune memory against previous invading mobile genetic elements and harnessing RNA-guided programmable endonucleases for foreign nucleic acid degradation. A typical CRISPR–Cas system consists of a genetic locus of related cas genes and a CRISPR array containing several different spacer units flanked by unique direct repeats (Figs. 1 and 2). The arms race between bacteria and phages triggers activation of highly diverse CRISPR loci and Cas proteins, leading to divergent immune mechanisms. Conversely, phages have evolved divergent bacteriophage- or prophage-encoded anti-CRISPR proteins (Acrs) that inhibit Cas effector complex-mediated targeting of invading nucleic acids (reviewed in refs. 1–11).

Fig. 1 |. Three stages of CRISPR–Cas13 adaptive immunity: adaptation, expression and interference.

CRISPR–Cas13 loci consist of a CRISPR array and neighboring cas genes. In the adaption stage, the adaptation machinery acquires fragments of invading mobile genetic elements and integrates them into the CRISPR array as new spacers, which generates heritable immunological memory. CRISPR repeats are represented as cyan rectangles, and spacers are represented as differently colored diamonds. The cas1 and cas2 genes, which encode adaptation modules, are missing in most Cas13 loci. How these Cas13 systems acquire new spacers remains unclear. During the CRISPR expression stage, the CRISPR array is transcribed as pre-crRNA and is processed into mature crRNA by Cas13, such that each crRNA carries a single spacer flanked with a part of the direct repeat on one side. The topologies of pre-crRNA in types VI-A, VI-B and VI-D systems, highlighting the conserved sequences and secondary structures, as well as the differences among different subtypes are shown in the zoomed-in box. Pre-crRNA processing sites are indicated by red arrows. The mature crRNA assembles with its effector protein Cas13 to form a surveillance complex that recognizes and degrades foreign genetic elements complementary to the crRNA spacer during interference. At the interference stage, guide–target base-pairing requirements enable sequence-specific targeting and lead to activation of Cas13 for the cleavage of target RNA at promiscuous positions outside the pairing region. Additionally, upon target RNA recognition, Cas13 causes collateral RNase activity against environmental substrate RNA, which induces cell dormancy. Acrs encoded by bacteriophages can directly inactivate Cas13 and prevent target cleavage.

Fig. 2 |. Functional module organization of CRISPR type VI Cas13 loci.

The functional modules involved in adaptation, expression and interference for each subtype are illustrated in the inserted panels. The approximate positions of the two HEPN domains are indicated in dark green within each effector Cas13a (C2c2), Cas13b (C2c6), Cas13c (C2c7), Cas13d (CasRx), Cas13bt (further classified as Cas13X and Cas13Y) and Cas13ct proteins. The average sizes of Cas13 proteins are indicated under the corresponding genes, among which Cas13d (~930 amino acids (aa)) and Cas13bt and Cas13ct (~800 amino acids) proteins are relatively smaller than their Cas13a–Cas13c protein counterparts (larger than 1,100 amino acids). Genes encoding Cas1, Cas2, Csx27, Csx28 and WYL domain proteins are found only in some species in a set of indicated subtypes and are marked by dashed outlines. The dendrogram on the left of the Cas13 systems is adapted from refs. 34,38,45 and shows the potential evolutionary relationships. The branch lengths are not scaled and do not indicate evolutionary distances. HTH, helix–turn–helix. TM, transmembrane.

Despite the diversity and complexity of CRISPR–Cas systems, the immunization process functions in three stages: adaptation, expression and interference (reviewed in refs. 12–18) (Fig. 1). At the initial adaptation stage, short DNA fragments derived from phages or invading plasmids, called protospacers, are integrated into the leader proximal end of a host CRISPR array to create new spacer segments, which provide immunological memory of previous infections (reviewed in refs. 18–22). In many cases, spacer acquisition is mediated by the Cas1–Cas2 integrase complex and sometimes is also associated with reverse transcriptase (RT), Cas4 or Csn2. At the subsequent expression stage, the CRISPR array is transcribed into a single noncoding RNA, known as pre-crRNA, and further processed by related Cas proteins or non-Cas host RNases to generate individual mature CRISPR RNA (crRNA), each of which carries a single spacer together with one adjacent direct repeat^23–25. At the last interference stage, a surveillance ribonucleoprotein complex, called the effector or interference complex, formed by mature crRNA and effector Cas protein(s), recognizes and cleaves foreign nucleic acids that contain sequences complementary to the spacer segments of crRNA^26–28.

Based on the protein constituents of effector modules, locus architectures and cas gene composition, CRISPR–Cas systems are grouped into two classes and further subclassified into six types, labeled I to VI²⁹. Class 1 systems (types I, III and IV) use multiple (typically three to six) Cas subunits to form a multisubunit effector complex for interference³⁰. By contrast, class 2 systems (types II, V and VI) use a single, relatively larger, multiple-domain-containing Cas protein for degradation of invading nucleic acids^31,32.

RNA-centric type VI Cas13

Unlike DNA-targeting systems that could repress virus replication by a single cleavage of viral DNA, RNA-targeting systems need to repel multiple copies of RNA transcripts to achieve immunity. RNA-targeting type VI CRISPR–Cas Cas13 effectors monitor RNA in the host cell and search for RNA targets that are complementary with the crRNA spacer (Fig. 1). Beyond degradation of bound RNA targets, Cas13 is also activated by bound RNA targets, thereby inducing collateral degradation of nonspecific substrate single-stranded RNA (ssRNA) (Fig. 1). Such processes provide immunity against complementary viral or phage RNA targets and cognate DNA phage transcripts, thereby inducing cell dormancy of infected cells, together with neutralization of CRISPR-resistant phages^33–36 (reviewed in ref. 37). Various accessory proteins are found in CRISPR–Cas13 loci (Fig. 2) and are believed to regulate Cas13 activity^33,38.

The capacity of a single Cas protein for recognition and cleavage of target ssRNA, in addition to target-dependent collateral RNase activity of Cas13, in turn empowers applications for programmable RNA targeting, editing, tracking, detection and imaging in both prokaryotic and eukaryotic cells^39–43.

We will describe the current understanding of the three stages of the immunization process mediated by type VI Cas13 systems, with an emphasis on the expression and interference steps, thereby providing up-to-date knowledge of structure-based mechanistic insights into the function and behavior of single-subunit Cas13 proteins during effector complex formation, crRNA maturation, self versus non-self discrimination, substrate recognition and cleavage and activity modulation. Throughout, we will also outline recent advances together with remaining challenges toward the development of powerful and programmable Cas13-based RNA manipulation applications.

Classification and composition

Searches for putative CRISPR–Cas type VI loci led to the discovery of subtypes VI-A, VI-B (including VI-B1 and VI-B2), VI-C and VI-D as well as VI-BT and VI-CT from distinct branches within subtypes VI-B and VI-C^{32–34,38,44–46} (Fig. 2 and Supplementary Table 1). The classification of subtypes is based on subtler differences in locus organization, Cas13 phylogeny and the positions of the pair of conserved HEPN domains in Cas13. Different Cas13 orthologs exhibit extremely low sequence similarity but share a similar overall bilobed architecture common to other members of the single-subunit Cas protein family, consisting of recognition (REC) and nuclease (NUC) lobes. Generally, the REC lobe is composed of N-terminal (NTD) and helical-1 domains that mainly contribute to crRNA recognition, while the NUC lobe is composed of helical-2, split HEPN-1 and HEPN-2 domains that mainly contribute to target RNA accommodation and cleavage. Extra domains include the helical-3 (or linker) domain in type VI-A and RRI (or lid) domain in type VI-B (Supplementary Table 2).

Spacer acquisition is undefined among type VI systems, with little known about how Cas13 systems acquire new spacers and whether the acquisition is from RNA or DNA. Crosstalk on adaptation was observed between co-occurring RNA-targeting type VI-B and DNA-targeting type II-C CRISPR–Cas systems⁴⁷, suggesting that RNA-targeting systems provide protection to overcome DNA-based immunity evasion. Details and underlying mechanisms on co-occurring RNA-targeting and DNA-targeting systems remain to be explored.

crRNA biogenesis in type VI systems

In type VI systems, pre-crRNA processing activity is relegated to effector modules within Cas13 proteins, with a requirement for formation of the CRISPR immune surveillance complex (Fig. 1).

CRISPR array organization and crRNA topology

The CRISPR arrays in different subtypes show unique and conserved features related to length and sequence (Supplementary Table 1), resulting in a defined predicted secondary structure (zoomed-in box in Fig. 1). Generally, the direct repeat contains a stem loop (also called the 5′-handle), flanked by the processing cleavage site at one side and a variant spacer at the other side (reviewed in refs. 48,49). The upstream and downstream regions of the stem loop, called 5′- and 3′-flank segments, exhibit single-stranded conformations with distinct lengths in different subtypes. The pre-crRNA processing sites within the direct repeat as well as the dependence on divalent metal ions are different among subtypes^{33,34,38,50,51} and sometimes distinct even across different species within the same subtype^50,52,53 (Supplementary Table 1), which highlights a notable mechanistic distinction between different subtypes.

crRNA recognition and maturation

The recognition of crRNA in types VI-A, VI-B and VI-D has been studied in great detail^51–60 (reviewed in ref. 49), with the focus below on the type VI-A system. Structural studies have been undertaken on U-cleaving Leptotrichia shahii LshCas13 (Fig. 3a) in the apo state (Fig. 3b) and bound to crRNA (Fig. 3c–e).

Fig. 3 |. Mechanisms of crRNA recognition and maturation in type VI-A systems.

a, Schematic representation of the domain architecture of LshCas13a. The catalytic sites for pre-crRNA cleavage and target RNA cleavage are labeled. b,c, Overall structures of LshCas13a in apo (b; PDB 5WTJ) and crRNA-bound (c; PDB 5WTK) states. The line indicates the boundary between the REC lobe and the NUC lobe. The NTD domain in the REC lobe, which cannot be traced in the apo form, is clearly visible in the crRNA-bound form. The bound crRNA is embedded in the Cas13 scaffold with the entire repeat segment traceable, but only some segments of the spacer are detectable in the structure of the complex. The direct repeat and the spacer are colored in light blue and dark blue, while disordered regions are indicated as dashed lines. The catalytic residues for pre-crRNA cleavage are indicated by orange stars, while the R-X_4–6-H motifs in the two HEPN domains for target RNA cleavage are indicated by green and yellow stars, respectively. d, Topology showing crRNA in the Cas13a-bound state on binary complex formation. Loop, flank and bulge elements are labeled. Nucleotides not observed in the structure are colored in gray. The pre-crRNA processing site is indicated by a red arrow. e, Domain movements induced by crRNA loading. The structure of Cas13a in the apo state (in silver) is superimposed with the structure of Cas13a in the crRNA-bound state (in color). The major domain movements occur in the helical-2 domain that shifts toward the linker domain. A portion of the linker domain is disordered in the crRNA-bound state, which is indicated as a cyan dashed circle. f, Detailed interaction between the pre-crRNA cleavage site nucleotide C(−27) and Cas13a.

Cas13 buries the stem-loop of the direct repeat in the cleft between the NTD and helical-1 domains of the REC lobe (Fig. 3c–e). The stem region of crRNA is recognized mainly by backbone contacts with Cas13 residues, whereas the loop segment adopts a distinct conformation with extrusion out of two to four nucleotides, some of which are recognized in a sequence-specific manner. The 3′-flank segment exhibits a nearly A-form conformation and is stabilized by ribose–phosphate backbone interactions with the HEPN-2 domain. Substitution of nucleotides in the bulge, loop or flanking regions of the stem as well as structural mutations within the stem-loop region abolishes pre-crRNA processing and target RNA cleavage, due to failure of RNP complex formation between Cas13 and crRNA^36,50,53.

The spacer region is bound in a non-sequence-specific manner. Before base pairing of target RNA with the crRNA spacer, several regions of spacer are invisible in the binary structure due to high flexibility (Fig. 3c–e). The central and 3′-segments of the spacer (the repeat region-distal segment) are solvent accessible and adopt a nearly A-form conformation, which allows for target searching and loading^{49,51,54,55,58–60}.

Upon binding crRNA, domain rearrangements occur in the two lobes to generate an RNA-binding cleft^51,54 (Fig. 3e). The Cas13a helical-2 domain moves toward the linker domain in the NUC lobe on complex formation (Fig. 3e), resulting in a more compact conformation and a channel for crRNA accommodation⁵⁴ (Fig. 3e,f).

On the behavior of pre-crRNA cleavage, Cas13a generates a 2′,3′-cyclic phosphate on the 5′-flank product and a 5′-OH on the mature crRNA, with the mature crRNA bound to Cas13 after processing^38,50. The 5′-flank segment usually presents a non-A-form conformation and is positioned in the groove between the helical-1 and HEPN-2 domains, with some nucleotides in this region contributing to the sequence specificity of bound crRNA⁵⁴. In crRNA-bound LshCas13, the conserved Arg438 and the nonconserved Lys441 and Lys471 residues in the helical-1 domain has key roles in cleavage chemistry (Fig. 3f), given that Ala substitutions at these positions severely impacted pre-crRNA processing. By contrast, Ala substitution of residues surrounding the scissile nucleotide (−27 position) in the HEPN-2 domain (Fig. 3f) showed little effect on pre-crRNA cleavage⁵⁴. Interestingly, structural and biochemical data on U-cleaving Leptotrichia buccalis LbuCas13a and A-cleaving Lachnospiraceae bacterium LbaCas13a suggest a conserved processing mechanism involving the helical-1 and HEPN-2 domains, with the surrounding conserved residues forming an acid–base catalytic center^50,52–55, which shows divergence from U-cleaving LshCas13a.

Notably, Cas13 executes pre-crRNA cleavage and target RNA cleavage by using two distinct catalytic pockets (Fig. 3a,f). The two conserved R-X_4–6-H motifs in the two HEPN domains are responsible for target RNA cleavage but have no effect on pre-crRNA cleavage.

Interference mediated by Cas13

When mature crRNA is loaded into Cas13, this ribonucleoprotein complex retains the inactive state and is ready to monitor the presence of foreign invaders to provide RNA immunity. Cas13 is activated only upon binding with target ssRNA containing a region complementary with the crRNA spacer. The target RNA is then cleaved at promiscuous positions outside the guide–target duplex^34,36,60. Before the release of the crRNA-bound segment, Cas13 collaterally cleaves environmental nonspecific ssRNA, resulting in cell dormancy or cell death. Sequence-specific target RNA cleavage (called cis-RNase activity) and collaterally nonspecific RNA cleavage (called collateral trans-RNase activity) are executed by the pair of R-X_4–6-H motifs in the HEPN domains, which is a general property of type VI systems. Cas13 lacks the capacity to open up double-stranded RNA, and thus the cleavage pattern is determined by nucleotide cleavage preference (details in Supplementary Table 1) and structurally accessible ssRNA regions^{34,36,51–53,55}.

Sequence-specific target RNA cleavage

Structural studies have been undertaken on U-cleaving LbuCas13a (Fig. 4a) bound to crRNA⁵⁵ (Fig. 4b,d) and on the addition of complementary target RNA⁵⁵ (Fig. 4c,e). The REC lobe binds to the crRNA direct repeat, while the NUC lobe forms an RNA duplex-binding channel to accommodate the guide–target RNA duplex (Fig. 4c). Notably, on target RNA binding, conformational changes occur in the 3′-flank of the crRNA direct repeat (Fig. 4d–g). Before target RNA loading, the three nucleotides (−1 to −3) in the 3′-flank that are connected with the spacer adopt a pre-ordered A-form conformation⁵⁵ (Fig. 4d). Upon target RNA access, these three nucleotides (−1 to −3) in the 3′-flank move toward the crRNA stem loop and form a kink (Fig. 4e,f). Simultaneously, target RNA induces substantial conformational changes in the helical-2 and HEPN-1 protein components of Cas13a (Fig. 4f,g), thereby generating both the positively charged binding channel to accommodate the guide–target RNA duplex (Fig. 4f) and the activated composite HEPN catalytic pocket for both cis- and collateral trans RNA cleavage (Fig. 4g)^51,54,55. The channel for guide–target RNA duplex accommodation is widened, and the residues in the positively charged surface make backbone interactions mainly with the central region of the duplex, with a few backbone interactions also formed in the repeat-proximal and repeat-distal regions (Fig. 4f). Focusing on the composite HEPN catalytic pocket responsible for RNase activity, the distance among the four key catalytic residues (two arginines and two histidines) in the strictly conserved R-X_4–6-H motifs in the two HEPN domains is shortened on complex formation with target RNA due to the movement of the HEPN-1 domain toward the surface and proximal to the HEPN-2 domain (Fig. 4g), making Cas13 catalytically active for ssRNA hydrolysis. Although there is no structure of ssRNA positioned in the composite HEPN pocket yet, it has been proposed, in terms of a catalytic mechanism, that the conserved His acts as a general base to activate the 2′-OH, which then targets the phosphodiester backbone, with the conserved Arg in turn stabilizing the pentavalent transition state intermediate.

Fig. 4 |. Mechanisms of target RNA recognition in type VI-A systems.

a, Schematic representation of the domain architecture of LbuCas13a. The catalytic sites in the pair of HEPN domains for target RNA cleavage are indicated. b,c, Overall structures of the LbuCas13a–crRNA binary complex in the absence (b, pre-target bound; PDB 5XWY) and presence (c, target bound; PDB 5XWP) of target RNA. The REC and NUC lobes are labeled. The R-X_4–6-H motifs in the two HEPN domains for target cleavage are indicated by colored stars. d,e, Topology of crRNA (d) and crRNA–target RNA (e) in the LbuCas13a binary and ternary complexes. The direct repeat, spacer and target RNA are colored in light blue, dark blue and red, respectively. The loop, flank and bulge elements of crRNA are labeled. Nucleotides not observed in the structure are colored in gray. The conformational change in the 3′-flank of the crRNA direct repeat is indicated by a red arrow. The 3′-flank region (−3 to −1) of the direct repeat of LbuCas13a–crRNA moves toward the stem loop as indicated by red arrows. f, Domain movements in the Cas13a–crRNA binary complex induced by bound target RNA on ternary complex formation. The structure of LbuCas13a in the pre-target RNA-bound state (in silver) is superimposed with that in the target RNA-bound state (in color). The major domain movements occur in the helical-2 and HEPN-1 domains and are indicated by colored arrows. g, Conformational changes in catalytic residues of the pair of HEPN domains on proceeding from the Cas13a–crRNA binary complex (in silver) to the ternary complex with added target RNA (in color). Target RNA loading induces the formation of a composite catalytic pocket involving a pair of HEPN domains (in color). Lbu, LbuCas13a.

Mismatches in the guide–target duplex segment affect both in vitro and in vivo RNase activities. Usually, Cas13, which could somewhat tolerate a single-nucleotide mismatch, is dramatically affected by two tandem mismatches^{36,44,52,54,55,58,61–63}. The most sensitive mismatch region is usually located within the central region of the spacer^36,61–63 (Supplementary Table 1), which is defined as the seed region. The seed region is consistent with the observed solvent-accessible region of the spacer in Cas13–crRNA binary complexes and extends further in some cases^{49,51,54,55,58–60}.

Collateral nonspecific RNA cleavage

The crRNA-bound portion of target RNA is not cleaved when it base pairs with crRNA^34,36,60, which indicates that Cas13 remain active before the guide–target duplex is unwound and the next new target RNA gains access (Fig. 1). This enables broad immunity against viral transcripts that are not recognized by crRNA or escape from DNA immunity.

In all available Cas13 structures, the catalytic pocket is solvent accessible and positioned away from the guide–target RNA duplex (Fig. 4c), thereby allowing access of the region in target RNA beyond the paired complementary portion and surrounding RNA in a non-sequence-specific manner, which explains the promiscuous cleavage site and target-dependent cis- and nonspecific trans-cleavage activity. So far, no structure is available of Cas13 bound simultaneously to crRNA, target RNA and nonspecific RNA, thereby limiting our knowledge of RNA substrate positioning. Therefore, whether sequence-specific and nonspecific target RNA species access the catalytic pocket using the same path remains unclear. Surface mutations in Ruminococcus flavefaciens XPD3002 Cas13d (RfxCas13d, also called CasRx) generated a variant with high sequence-specific cleavage and reduced collateral cleavage in HEK293 cells⁶⁴. Notably, this variant exhibited reduced in vitro target-dependent cis- and nonspecific trans-cleavage activities, suggesting a global defect in the HEPN domains. It remains unclear whether there is a way to modulate sequence-specific and nonspecific RNase activities separately.

Very few studies have been done to investigate the preference of non-sequence-specific RNA targeting and the effects of global RNA degradation. It has been proposed that LshCas13-mediated collateral cleavage of tRNA occurs at anticodon loops, thereby resulting in suppression of translation and cell dormancy⁶⁵. In addition, massive cleavage of tRNA leads to ribosome stalling, which in turn activates type II toxin RNases that cleave mRNA and 16S ribosomal RNA, further abrogating protein synthesis⁶⁵.

Self–non-self-discrimination

The ability to discriminate invading foreign elements (non-self) from intrinsic components (self), thereby preventing self-targeting, is usually imperative for successful immune systems. Whether self–non-self-discrimination naturally occurs in Cas13 systems in their native hosts remains unclear. Some Cas13a³⁶, Cas13b³³ and Cas13bt⁴⁵ systems show preference for a short sequence adjacent to the guide–target duplex, termed the protospacer flanking sequence (PFS), which drives PFS-dependent RNA targeting in vitro and in Escherichia coli cells but not in mammalian cells (Supplementary Table 1). The accurate mechanism underlying PFS preference and the related biological function in native host cells remain unclear.

In type VI-A LbuCas13a, Listeria seeligeri LseCas13a and LshCas13a systems, 5–8-bp pairing between the 5′ of the direct repeat flanking spacer (called the 5′ tag) and the 3′ extended region of the target RNA flanking region complementary to the spacer segment (called the anti-tag) abolished both cis and trans RNA cleavage activities and impaired Cas13a-mediated knockdown in bacterial cells, indicating that tag–anti-tag complementarity inhibits self-targeting in native bacteria^66,67. The structure of the U-cleaving LshCas13a (Fig. 5a) bound to anti-tag target RNA (Fig. 5b) highlights the bound crRNA–anti-tag RNA alignment⁶⁷ (Fig. 5c). Structural data reveal that tag–anti-tag base pairing induces conformational changes of the crRNA tag region and domain movements in the NUC lobe (Fig. 5c), thereby blocking the formation of a composite HEPN catalytic pocket required for RNA cleavage activity⁶⁷ (Fig. 5d). Thus, the composite HEPN pocket adopts a noncatalytic conformation following tag–anti-tag pairing in the ternary complex (Fig. 5d), thereby preventing cleavage of self-RNA, but adopts a catalytically competent conformation when this segment is unpaired, resulting in cleavage of non-self-RNA (Fig. 4g), providing a structural explanation for self–non-self-discrimination. We anticipate that the pre-ordered 3′-flank may adopt a favorable alignment for detection and nucleation with anti-tag during self–non-self-discrimination and stronger inhibition would result from formation of tag–anti-tag base pairing.

Fig. 5 |. Mechanism of self–non-self-discrimination in type VI-A systems.

a, Schematic representation of LshCas13a domain architecture. b, Overall structure of the LshCas13a–crRNA binary complex bound to target RNA (PDB 7DMQ) in which the target RNA anti-tag segment pairs with the crRNA tag segment. The NTD domain is disordered as indicated as the dashed box in a. c, Topology showing pairing alignment between crRNA and anti-tag containing target RNA. The 8-nucleotide tag segment is highlighted with a black box. The 8-nucleotide anti-tag and 28-nucleotide target regions of target RNA are colored in black and red, respectively. The conformational changes of the tag segment are indicated by a red arrow. d, Comparison of catalytic residues of the pair of HEPN domains on proceeding from the Cas13a–crRNA binary complex (in silver) to the ternary complex (in color) with added anti-tag target RNA. The catalytic residues remain apart from each other, indicating that Cas13a is inactive after binding with anti-tag target RNA. Lsh, LshCas13a.

Regulation of Cas13 RNase activity

It has been shown that Cas13-mediated immunity can be modulated by other factors including endogenous accessory proteins and phage-encoded Acrs. These proteins repress or enhance Cas13 RNase activities either through physical interactions or crosstalk with other potential signaling pathways.

Regulation by endogenous accessory proteins

Given the profound consequences of global RNA degradation, it becomes necessary to tightly control Cas13 activity. One approach relies on sufficient base pairing between target RNA and the crRNA spacer region, low mismatches in the central seed region and preferential tag–anti-tag base pairing in Cas13a systems. The alternative approach relies on intrinsic accessory proteins in CRISPR–Cas loci. Accessory genes encoding Csx27, Csx28 and WYL (named for three conserved amino acids found in a subset of domains of this superfamily) domain-containing proteins (WYL1 and WYL-b (WYL-b1–WYL-b5)) were found in types VI-B1, VI-B2 and VI-D loci, respectively^33,38 (Figs. 2 and 6a). These accessory proteins share no similarity in primary sequence and domain organization, indicating divergent modulation mechanisms of type VI effectors.

Fig. 6 |. Modulation mechanisms of type VI systems involving anti-CRISPRs and accessory proteins.

a, Proposed modulation mechanisms mediated by anti-CRISPRs and accessory proteins. b, AlphaFold2-predicted overall structure (left) and electrostatic surface (right) of BzCsx27. The hydrophobic surface might be associated with membrane. c, Overall structure of the PbCsx28 octamer (PDB 8GI1). Transmembrane regions are indicated. The predicted key residues R152 and H157 in the R-X_4–6-H motif are shown as a stick model in the zoomed-in box. d, Overall domain alignment and structure of the RspWYL1 dimer (top; PDB 6OAW). Monomer A (Mol A) is shown in a ribbon mode, and monomer B (Mol B) is shown as a surface mode. e, Schematic representation of the domains of LseCas13a and AcrVIA1. f, Two views of a ribbon diagram showing Acr protein AcrVIA1 (in surface) bound to LseCas13–crRNA (in ribbon) (PDB 6VRB). AcrVIA1 directly binds to and interacts with distinct domains of Cas13, thereby blocking access to target RNA.

Csx27, which contains two to four predicted transmembrane helices (an AlphaFold2 model of BzCsx27 is shown in Fig. 6b), was shown to repress Prevotella buccae PbCas13b- and Bergeyella zoohelcum BzCas13b-mediated interference activity in vivo, implying that Csx27 may perform general regulatory effects and function across multiple VI-B1 loci^33,68. However, BzCsx27 was not membrane associated when expressed in E. coli, indicating that additional unknown factor(s) in the native hosts of VI-B1 systems might be essential for membrane localization. The accurate role and repression mechanism of Csx27 remain to be determined. Bioinformatic analysis of the neighboring genes located both upstream and downstream of the Csx27-encoding gene suggest that Csx27 might form a membrane channel and associate with natural competence or ubiquitin signaling, where it may bind with Cas13b and keep Cas13b in an inactive state⁶⁸.

Csx28 was only found in type VI-B2 systems and predicted to contain a transmembrane region and a divergent HEPN domain³³. Csx28 might enhance the collateral activity of Cas13b to increase its interference activity³³. A recent cryo-EM structure of PbCsx28 revealed that it exists as an octamer, with the transmembrane region forming a positively charged pore (Fig. 6c)⁶⁹. The predicted HEPN domains adopt a noncanonical arrangement and present a new topology of unknown function with no similarity to reported structures. Notably, the predicted key histidine and arginine residues in the R-X_4–6-H motif are located apart from each other and point in opposite directions and cannot form a catalytically competent composite nuclease pocket (zoomed-in box in Fig. 6c). Based on a combination of structural data and functional assays, it was proposed that Csx28 is activated by Cas13b, resulting in Csx28 pore formation and concomitant membrane depolarization⁶⁹. Further assays are required to explore more details of the proposed hypothesis, specifically the nature of the trigger for activation as well as an understanding at the molecular level of the principle underlying membrane depolarization.

WYL domain-containing proteins contain a WYL domain and a helix–turn–helix or a ribbon–helix–helix DNA-binding domain³⁸. The WYL domain-containing protein has been reported as a negative regulator of type I-D systems through transcriptional repression and is predicted to sense nucleic acids or nucleotide derivatives^70,71. In subtype VI-D systems, Ruminococcus sp. RspWYL1 increases in vitro cis- and trans-RNase activities of RspCas13d and its ortholog Eubacterium siraeum DSM 15702 (EsCas13d) as well as RspCas13d-mediated in vivo RNA knockdown, suggesting a distinct transcription-independent mechanism³⁸. Biochemical and structural data revealed that RspWYL1 consists of three domains (NTD, WYL and C-terminal domain (CTD)) and forms a homodimer through interactions between the NTD–NTD and CTD–CTD domains (Fig. 6d)⁷². RspWYL1 binds to ssRNA through WYL and CTD domains but has weak binding affinity for Cas13d⁷². RspWYL1 may act as an RNA sponge by positioning more RNA into the Cas13d catalytic pocket and/or allosterically enhancing Cas13d activity by physical interactions. The exact modulation mechanism remains unclear.

Inhibition by Acr proteins

Recent studies have identified AcrVIA1 as an anti-CRISPR inhibitor for Cas13a and AcrVIB1 for Cas13b^56,73,74. Among these Acrs, the prophage-encoded AcrVIA1 physically interacted with the LseCas13a–crRNA binary complex, thereby reducing the RNase activity and abolishing Cas13a-induced cell dormancy⁵⁶. Structural and biochemical analysis of the binary LseCas13a–crRNA (structure not shown) and the ternary LseCas13a–crRNA–AcrVIA1 complex (Fig. 6e,f) elucidated that AcrVIA1 is positioned along the crRNA-exposed face of Cas13a in the ternary complex, thereby forming intermolecular contacts with the NTD, helical-1, linker and HEPN-1 domains of Cas13a as well as the 3′ segment of the spacer. These data indicate that AcrVIA1 not only prevents target RNA access and base pairing with the spacer but in addition blocks the conformational changes required for nuclease activation, thereby retaining Cas13a in the inactive state. Notably, a single dose of AcrVIA1 delivered by an individual virion completely dismantles type VI-A CRISPR-mediated immunity⁵⁶, unlike inhibitors of DNA-cleaving Cas nucleases, which caused limited immunosuppression and require multiple infections to bypass CRISPR defenses. Additional research identifying a proposed subset of AcrVIA proteins that reduced Cas13a nuclease activity and prevented Cas13-mediated RNA targeting and editing in bacteria and human cells⁷³ has been challenged⁷⁵ and remains controversial.

Biotechnological applications based on active Cas13

Type VI systems engendered tremendous interest given that the Cas13–crRNA effector acts as a scaffold necessary for targeted cleavage of ssRNA, expanding the programmable manipulation of transcripts at the RNA level without heritable genome editing. There are opportunities for the application of active Cas13 in biotechnological applications ranging from programmable RNA knockdown to nucleic acid detection.

Programmable RNA knockdown or degradation

Cas13a, Cas13b, Cas13bt and Cas13d have been used for programmable RNA knockdown in prokaryotic and eukaryotic cells and animal models^{34,36,44,62,63,76–82}, in which RfxCas13d and Cas13bt mediated robust and either comparable or higher knockdown efficiency of the endogenous transcript^34,44. Moreover, the size of Cas13d and Cas13bt is more compact than Cas13a–Cas13c, which provides potential advantages in adeno-associated virus (AAV)-mediated delivery.

Cas13-mediated RNA knockdowns have been explored for their therapeutic potential in repressing pathogen infection by degradation of pathogenic RNA. The elimination efficiency of viral and bacterial RNA was investigated in mammalian and plant cell lines and indicated that Cas13 could significantly reduce infection^79,83–89. Further studies performed in animal models by the mRNA-encoded Cas13 approach suggested efficient reduction of influenza RNA in mice and mitigation of severe acute respiratory syndrome coronavirus 2 infection in hamsters⁹⁰.

LwaCas13a-, RfxCas13d- and PspCas13b-mediated off-target activities have been observed in Drosophila^91,92 and some human cell lines^93–95. Structural-based mutation design has been used to engineer Cas13d and Cas13X with reduced off-target effects in human cell lines, mice and somatic cells⁹⁶. Machine learning and deep learning as well as a large-scale Cas13d-based screening dataset have been used to systematically evaluate on-target activity and off-target effects. These efforts have permitted prediction of on-target activity of a crRNA^97,98, which sheds light on the systematic and comprehensive evaluation of the correlation between on- and off-target effects and different factors such as the expression level of target RNA, cell types and target accessibility, among others. Some caution is needed in interpretation of the available literature given concerns about discrepancies based on different controls, inconsistent vectors, variable protein constructs and guide designs, which could result in inconsistent observation of how Cas13 behaves in a test tube, how it behaves in bacteria and how it behaves over varying cell types and make it very difficult to determine what drives on- and off-target activity.

Nucleic acid detection

Cas13 collateral trans-cleavage activity as monitored by fluorescent RNA has resulted in the development of emerging technologies for rapid, sensitive and specific nucleic acid detection in molecular diagnostics. Usually, the nucleic acid sample is amplified, which is then followed by transcription, thereby generating ssRNA targets (Fig. 7a). When the ssRNA target base pairs with the crRNA spacer, Cas13 is activated and trans-cleaves the fluorophore-quenched ssRNA reporter for fluorescence signal release (Fig. 7b). Isothermal amplification methods, which require lower temperature than traditional PCR amplification, have been integrated into Cas13-based detection applications for template amplification (Fig. 7a). The established methods, such as specific high-sensitivity enzymatic reporter unlocking (SHERLOCK), provide a limit of detection (LOD) at the attomolar level (~6.5 copies per μl)⁹⁹, which is comparable to that of reverse transcription–quantitative PCR (qRT–PCR), and furthermore have proved to be reliable in a clinical study^100–103. Multiplexing using Cas13 and Cas12a orthologs in SHERLOCK enables multiplexed detection of four target sequences in one sample¹⁰⁴. One way for sensitivity improvement involves incorporation with other protein components for signal amplification. A successful method, termed SHERLOCKv2, involves combination with the type III nuclease Csm6 and chemically modified ssRNA activators (Fig. 7c)^104,105. Cas13 trans-cleaves the ssRNA activator to release activator cyclic oligoadenylates, which activate Csm6 RNase activity on the fluorophore-quenched ssRNA reporter and provide zeptomolar sensitivity^104,105. Development of the amplification-free detection method with an LOD in the attomolar range is necessary for low concentration of DNA or RNA samples from clinical and environmental specimens. The fast integrated nuclease detection in tandem (FIND-IT) method uses the Cas13–Csm6 system and the combination of eight designed crRNA species to simultaneously target eight different positions of one single ssRNA, thereby allowing for an LOD of ~30 copies per μl without amplification¹⁰⁶. Another approach involves protein engineering. By inserting the RNA-binding domain (RBD) into a loop region proximal to the active site of the HEPN domains, RBP-LwaCas13a variants achieve an LOD at the attomolar level without amplification¹⁰⁷. Recent efforts have also been focused on optimizing reaction conditions¹⁰¹, developing a field-deployable sample preparation platform¹⁰⁸ and integrating electrochemical microfluidic chips and sensors for quantitative and multiplexed signal readout^{102,109–111}. Additional approaches for development of an amplification-free and ultra-sensitive detection method await development.

Fig. 7 |. Active Cas13-based RNA-targeting technologies.

a, DNA or RNA samples are amplified by isothermal amplification and then transcribed into ssRNA. b, Binding of the crRNA to the complementary target sequence activates Cas13 collateral cleavage of quenched fluorescent ssRNA reporters, thereby releasing fluorescence signal. c, Schematic of Cas13–Csm6-based nucleic acid detection.

Biotechnological applications based on inactive Cas13

Catalytically inactive Cas13 (dCas13) can act as a programmable RNA-binding platform to bring modules close to the RNA of interest to change RNA characteristics, thereby impacting on base editing, modification editing and splicing modulation or to study RNA localization and interaction with partner proteins.

RNA base editing

Site-specific RNA-based editing relies on the concept of dCas13-fused deaminase (Fig. 8a). By generating a single mismatch at the editing site, the flipped-out nucleotide is captured by deaminase. The first version of RNA base editor has been developed by fusing dCas13b with deaminase ADAR2 (adenosine deaminase acting on RNA type 2), termed REPAIR, and allows for adenosine-to-inosine (A-to-I) conversion^63,112. The produced inosine mimics guanosine in translation and splicing reactions and achieves adenosine-to-guanosine exchange. One ADAR2 variant fused with Cas13b (named RESCUE), has served as a cytosine deaminase and has been used for cytosine-to-uridine (C-to-U) replacement¹¹³. However, the base editors REPAIR and RESCUE exhibited high off-target effects. The follow-up optimized methods, namely REPAIRv2 and RESCUE-S, showed reduced off-target rate following rational mutagenesis of ADAR2 (refs. 63,113). Later, other methods have been developed by fusing deaminases with other Cas13 orthologs including Cas13d and Cas13bt3 (also known as Cas13X.1), and these exhibited higher editing efficiency and/or lower off-target effects^{44,78,114,115}. An ADAR2-dependent cytosine-to-uridine editor has retained activity on A-to-I conversion, resulting in off-target effects on adenosine. The cytidine deaminase APOBEC3A together with Cas13 exhibited higher efficiency than RESCUE-S¹¹⁶.

Fig. 8 |. dCas13-based RNA manipulation technologies.

a, RNA base editing. The dCas13–deaminase fusion can be used for site-specific nucleotide conversion. b, Programmable alternative splicing modulation. dCas13 fused to splicing factors can be used for pre-mRNA alternative splicing modulation to generate a customized splicing product. c, Programmable RNA m⁶A modification regulation. dCas13 fused to methyltransferase complex (left) or demethylase (right) can mediate the m⁶A methylation levels of target RNA. d, Endogenous RNA imaging and tracking. dCas13 fused to a fluorescent protein enables the visualization of RNA in live cells. By fusing different localization signals with Cas13, target RNA can be transported to the desired cellular location, such as the nucleus and the cytoplasm. e, Cas13-based proximity labeling. dCas13 fused with ligase or peroxidase can be applied to label proximal proteins with biotin and facilitate the discovery of new proteins interacting with target RNA. f, dCas13 recruited to mRNA may physically abrogate ribosome binding and movement, thereby impeding translation initiation and elongation. dCas13 fused with a translational repressor provides an alternative suppression mechanism upon m⁷G cap binding, increasing translational repression efficiency. dCas13 fused with a translational initiation factor targeting the 5′-UTR of mRNA enables the recruitment of ribosomal subunits and thus presumably increases translation initiation. Through ribosome recruitment by the SINEB2 element, the dCasRx–SINEB2 system allows for translation activation by targeting the initiation region around the AUG start codon. eIF4, eukaryotic initiation factor 4.

Cas13 base editors allow nucleotide conversion on transcripts without changing DNA sequence, which provides an alternative approach toward site-direct RNA editing treatment for monogenic diseases. Several studies on the editing of pathogenic mutations in different cell types and mouse models have been reported and showed different efficiency in restoring protein expression^63,117–120. As the editing systems have been established recently, a comprehensive evaluation of on-target activity and off-target effects on both RNA editing and related protein products still remains to be done.

Alternative splicing modulation

Alternative splicing of mRNA precursors (pre-mRNA) transcribed from a single gene produces diverse mRNA species that encode proteins with different sequences and activities, thereby increasing proteome diversity in different types of eukaryotic cells. Alternative splicing is modulated by the core splicing spliceosome machinery together with many RNA-binding proteins and regulatory splicing factors. These include serine/arginine-rich (SR) proteins belonging to the RNA recognition motif-containing SR family and SR-related proteins containing one or more S/R-rich low-complexity domains (RS domains)^121,122. dCas13 has been applied for programmable control of alternative splicing using several strategies. By binding with splicing elements of target pre-mRNA, dCas13 alone can alter splicing and attempt to induce exon exclusion³⁴. Given that this approach failed to induce exon inclusion, replacement of the RNA recognition motif of splicing factors with CasRx (named CASFx) enabled programmable control of alternative splicing¹²³ (Fig. 8b). When coupled with the chemical-inducible FKBP (FK506 binding protein)–FRB (FKBP–rapamycin binding) domain (named iCASFx), CASFx comes under the control of a small molecule and allows for spatiotemporal control of alternative splicing. Subsequently optimized CASFx–SR, that is, derived from CASFx by fusing with RS domains, was able to manipulate alternative splicing and regulate SF expression, which acts as a convenient tool for evaluating the functional roles of splicing factors in tumors¹²⁴. These methods provide a potential therapeutic solution for RNA mis-splicing diseases.

Programmable RNA m⁶A modification

The formation, removal and function of N⁶-methyladenosine (m⁶A) modifications are regulated by a series of proteins: the methyltransferase complex METTL3–METTL14 and associated proteins, demethylases FTO and ALKBH5 and readers YTHDF1–YTHDF3. dCas13 can be used as an adaptor for programmable RNA m⁶A modification editing by fusing with different proteins, allowing investigation on the functional role of m⁶A modification at single transcripts of interest in live cells¹²⁵ (Fig. 8c). Linking with methyltransferases or demethylases can be used to edit the status of m⁶A modification at a single site in varied regions such the 3′-untranslated region (UTR) and the 5′-UTR in different cell localization conditions (nucleus or cytoplasm) with limited off-target effects^126–131. Coupling with the m⁶A readers YTHDF1 and YTHDF2 uncovered their functions in translation promotion and RNA degradation, respectively¹³². Two methods have been developed for m⁶A modification editing at multiple sites or transcripts. The dCas13–FTO system with its CRISPR array can simultaneously target multiple sites¹³¹. Another method called targeted RNA demethylation by SunTag system (TRADES) uses dCas13 and the SunTag system¹³³. The SunTag system is composed of the single-chain variable fragment (scFv) and multiple copies of its repeating-epitope GCN4 (an eukaryotic transcriptional activator protein) peptides, which are fused to a demethylase and dCas13 separately. Cas13–GCN4 recruits multiples of scFV–demethylases and achieves multiplexed m⁶A editing. To achieve spatiotemporal control of site-specific m⁶A modification editing, the blue light-sensitive protein CBIN and its adaptor CY2 were fused with dCas13 and methyltransferase or demethylase separately for the recruitment of each other under control of blue light¹³⁴. dCas13-based m⁶A editing tools may be applied in controlling disease progress by customized modifications in transcripts from oncogenes and tumor-suppressor genes. Ongoing research on programmable control of other RNA modifications such as m¹A and m⁵C is being undertaken by fusing dCas13 with the respective methylases, demethylases and readers.

RNA imaging in live cells

dCas13 enables endogenous RNA visualization in live cells by coupling with fluorescent proteins or split fluorophores. The most straightforward method is to use fluorescence-labeled dCas13 for detection of endogenous RNA transcripts⁶² (Fig. 8d). Other methods such as SunTag and split fluorophores have been used to increase fluorescence signal and reduce background noise. By fusing multiple copies of the GCN4 peptide and scFv with Cas13 and fluorescence protein separately, the SunTag system can amplify the fluorescence signal¹³⁵. Furthermore, split fluorophores together with the Cas13a–SunTag system produced a better signal-to-noise ratio than that of the previously established the MS2 and MS2-coat protein (MS2–MCP) system¹³⁶. Despite the preponderance of single-transcript imaging, coupling dCas13 orthologs with different fluorescent proteins can provide simultaneous two-transcript imaging in single cells. Moreover, combination of dCas13 with the MS2–MCP or dCas9 system enables dual-color imaging of two different transcripts or alternately RNA transcripts and genomic DNA, respectively^137–139. Current Cas13-based RNA imaging approaches expand our understanding of nuclear activities but still show limitation for low-abundance transcripts or RNA with secondary structures. Therefore, a versatile multiple-color image toolkit with low background noise needs to be established.

Mapping RNA–protein interactions

dCas13 could be applied to identify proteins interacting with RNA of interest in live cells. These approaches rely on the binding of dCas13 to target specific RNA and in the process capturing surrounding related RNA-binding proteins by UV cross-linking proteins physically bound to Cas13 (named CBRIP)¹⁴⁰ or proximity labeling proteins with biotin by enzymes including ligase and ascorbate peroxidase (Fig. 8e). dCas13 fused to the ligase PafA (named CRUIS) can mediate ligation of surrounding proteins with biotin-tagged PupE¹⁴¹. Similarly, the dCas13-fused ligase BASU (named CARPID) allows for biotinylating adjacent proteins. The dCas13-fused peroxidase APEX2 has been applied to biotinylate proteins within a distance of 25 nm with the addition with H₂O₂ and biotin-phenol¹⁴². To improve efficiency, a double-stranded RBD from human protein kinase R (PKR) was fused to the C terminus of dCas13, which stabilizes the RNA duplex and exhibits an increased enrichment level of proteins. However, these methods have limitations related to low abundance of targeted RNA and transient and unstable RNA–protein interactions.

Translational regulation

As active Cas13 showed toxicity and induced cell dormancy in bacteria due to its collateral activities^36,65, dCas13 has been repurposed for translational repression without affecting mRNA stability (Fig. 8f). dCas13d can be programmed to tightly bind with target mRNA and act as a road blocker to prevent ribosome binding or movement, thereby resulting in translational repression (Fig. 8f, top left). Engineered dCas13-based translational repression platforms have been established by targeting ribosomal binding sites or the 5′ end of the open reading frame and exhibited effective translational repression in cell-free expression systems¹⁴³ and in E. coli cells¹⁴⁴. The dCas13-based strategy (named CRISPRδ) has also been used to repress translation in human HEK293 cells by targeting the 5′-UTR or the start codon¹⁴⁵. By screening different translational repressor proteins, further enhancement of the repression effects can be achieved using the dCas13–4EHP (eIF4E-homologous protein, also termed eIF4E2) fusion protein and allows for expanded targeting sites including the start codon, the 5′-UTR and the 3′-UTR¹⁴⁵ (Fig. 8f, bottom left). Additionally, dCas13 can also be used to activate translation initiation of a fluorescent protein encoded by cis-repressed mRNA in a cell-free expression system and in E. coli cells¹⁴³. dCas13 fused with the translation initiation factor IF3 provided significant enhanced expression of fluorescent protein and a native gene encoding β-galactosidase in E. coli cells¹⁴⁶ (Fig. 8f, top right). In an alternative dCasRx-based approach, crRNA was covalently linked with the SINEB2 (short non-pairing interspersed nuclear element (SINE) B2) element of the long noncoding RNA uchl1, enabling increased expression of fluorescent protein in several cell lines and endogenous tumor-suppressor proteins in bladder cancer cells, respectively¹⁴⁷ (Fig. 8f, bottom right). These strategies provide potent and efficient tools for translational regulation regardless of mRNA degradation.

Future challenges

Compared to the Cas9 and Cas12 systems, RNA-guided Cas13 systems intrinsically target RNA instead of DNA, giving more benefit to on-target specificity. Importantly, recently discovered type III-E systems elegantly deploy the single Cas protein gRAMP (also known as Cas7–11) for ssRNA cleavage and are associated with caspase-like TPR-CHAT (also known as Csx29) for potential protein targeting^94,148. Space limitations do not allow us to describe this interesting system. In addition, the relatively compact type V-G Cas12g displays RNA-guided cleavage activity on target ssRNA and collateral RNase and DNase activities¹⁴⁹. Compared with type VI Cas13 systems, these two types are evolutionarily distant, structurally unrelated and rely on distinctly different mechanisms for not only pre-crRNA processing but also sensing and cleavage of target RNA, which expands the toolbox for RNA manipulation tools. Given the simplicity of composition and diverse RNA-centric target properties, Cas13 has attracted attention as a promising RNA detection and manipulation platform.

Other type VI Cas13 subtypes

To date, only the assembly and RNA recognition mechanisms of subtypes VI-A and VI-D have been investigated in depth, leaving a number of challenges that need to be addressed in the future. Thus, (1) whether target RNA and substrate RNA access the catalytic pocket in the same manner among different subtypes remains unknown. (2) How orthologs such as Cas13b and Cas13c scan and load target RNA is still unclear. (3) The pathway to prevent autoimmunity in other subtypes except VI-A remains to be determined. (4) The accurate modulation mechanisms of Acr proteins and intrinsic accessory proteins targeting type VI systems remain to be explored. In the future, we anticipate expansion of new subtypes with new functional features through the screening of ever-growing genomic and metagenomic databases, with the capacity for identification and development of new systems harboring advantageous properties.

Effect of off-target effects

The major concern of currently available Cas13-based RNA-targeting technologies is the complication associated with widespread off-target effects. Both wild-type and dCas13-based technologies such as knockdown or base editors exhibit off-target effects. Thus, minimization of off-target effects remains a major challenge that needs to be overcome for effective clinical applications of this methodology.

Cellular delivery of Cas13–RNA

A second concern relates to effective all-in-one AAV delivery of a CRISPR array or guide RNA, an engineered Cas13 encoding gene, an optional module and required expression or regulatory elements. The development of a smaller, more efficient and less cytotoxic Cas13 toolkit will be of great importance for manipulation of primary cells and clinical treatments. Recent advances associated with the remarkably small size of Cas13bt, Cas13ct and Cas13d should encourage further development of engineered small Cas13 effectors.

Effect of immunogenic and cytotoxic effects

A further concern relates to the potential immunogenic and cytotoxic effects of Cas13. Surprisingly, crRNA alone substantially reduced viral RNA expression and presented stronger suppression on addition of PspCas13b in insect cells, which has not been observed in mammalian and plant cells¹⁵⁰. It remains unclear whether other Cas13 proteins alone or bound to their cognate crRNA cause similar cell growth repression in insect or other eukaryotic cells.

Benefits of Cas13-based therapeutic intervention

Before the development of CRISPR-based technologies, antisense oligonucleotides and RNA interference approaches were effective for RNA knockdown and have been clinically used for RNA-targeting therapies even often with off-target effects. Cas13 systems have exhibited high efficiency and specificity in RNA knockdown in mouse disease models. Importantly, Cas13-based strategies have also expanded beyond knockdown capacities, allowing for base editing, RNA demethylation, splicing modulation and translational regulation. These valuable Cas13-based approaches can be achieved through delivery from AAV vectors, providing strong candidates for future durable RNA-targeting therapies.