Characterization of gRNA-dependent and gRNA-independent off-target binding sites of PspCas13b and RfxCas13d in mammalian cells

Abstract

CRISPR–Cas13 systems, harnessed for RNA-guided transcriptome editing, hold significant promise for clinical and in vivo therapeutic applications. However, understanding their in vivo target specificity and recognition rules remains a challenge. In this study, we employed the uSpyCLIP method, which enhances sensitivity and specificity for identifying RNA-binding protein (RBP) binding sites, to map the transcriptome-wide binding sites of catalytically inactive PspCas13b (dPspCas13b) and RfxCas13d (dRfxCas13d) in HEK293T cells, using a variety of single guide RNAs (gRNAs). Surprisingly, we identified both gRNA-dependent and gRNA-independent off-target binding sites for both dCas13 complexes. These gRNA-independent off-target sites exhibited distinct RNA structural and sequence signatures: dPspCas13b’s gRNA-independent binding was associated with specific RNA structural features, while dRfxCas13d’s was linked to unique sequence motifs. Analysis of gRNA-dependent off-target sites revealed the crucial role of the DR-distal and middle regions of the gRNA in determining binding specificity. Further analysis demonstrated that some off-target binding events led to changes in gene expression at the messenger RNA and/or protein level. Collectively, our findings provide important insights into the characteristics of gRNA-dependent and gRNA-independent off-target binding for PspCas13b and RfxCas13d, offering valuable guidance for optimizing Cas13 and gRNA design in future applications.

Graphical Abstract

Graphical Abstract.

Introduction

Transcriptome-editing technologies adapted from the type VI CRISPR–Cas systems have gained significant attention due to their simplicity and programmable RNA-targeting capabilities [1–3]. Type VI CRISPR–Cas systems consist of a single-subunit Cas13 nuclease and a CRISPR RNA (crRNA), which together form a crRNA-guided RNA-targeting effector complex [4]. These systems are classified into six subtypes, with Cas13b from Prevotella sp. P5-125 (PspCas13b) and Cas13d from Ruminococcus flavefaciens XPD3002 (RfxCas13d) being two of the most widely utilized effectors [5–7]. Both Cas13 nucleases achieve site-specific RNA targeting via a guide RNA (gRNA), which includes a short cognate direct repeat (DR) sequence and a 20–30 bp target-matching spacer sequence [4]. The Cas13 systems have been successfully applied for messenger RNA (mRNA) and long non-coding RNA (lncRNA) knockdown in various cell lines and model organisms [6–17]. Furthermore, catalytically inactive dPspCas13b and dRfxCas13d variants have been employed in additional applications, including RNA editing, RNA imaging, and post-transcriptional regulation [6, 18–27].

As the clinical and in vivo applications of Cas13 systems advance, ensuring their target specificity becomes increasingly critical [5, 28]. However, the in vivo target specificity of these systems remains incompletely understood. Most prior studies have focused on testing a limited number of candidate off-target sites for in vitro cleavage activity [29–31]. Mismatch intolerance analysis using a designed plasmid library revealed that PspCas13b has a region extending from base pairs 4–18 in the spacer (for a 30-nucleotide spacer) that is particularly intolerant to mismatches [6]. A more recent study, using comprehensive spacer-target mutagenesis, showed that PspCas13b can tolerate up to four mismatches, but requires ~26 nucleotides of base pairing (for a 30-nucleotide spacer) with the target to activate its nuclease domains [30]. Although the gRNA spacer of Cas13d was initially thought to lack a defined seed region, it has since been reported that the efficacy of RfxCas13d knockdown is influenced by a seed region located between nucleotides 15 and 21 of the gRNA (for a 27-nucleotide spacer), with its center at nucleotide 18 relative to the gRNA 5′ end [29, 32]. Moreover, RNA sequencing has also been used to detect transcriptome-wide off-target cleavage in vivo, and machine learning models have been applied to predict off-target activity for Cas13d [6, 7, 33, 34]. However, these approaches are unable to provide an unbiased, whole-transcriptome view of the direct off-target binding of the Cas13-gRNA complex.

UV crosslinking and immunoprecipitation (CLIP) is a powerful technique for identifying direct RNA targets and specific binding sites of RNA-binding proteins (RBPs) and ribonucleoprotein (RNP) complexes [35, 36]. Current CLIP methodologies can be classified into two categories based on RNP purification strategies. The first category includes traditional CLIP techniques such as individual-nucleotide resolution CLIP (iCLIP), infrared CLIP (irCLIP), and enhanced CLIP (eCLIP), which rely on antibody-based purification and sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) separation followed by nitrocellulose membrane transfer [37–39]. The second category of CLIP methods, such as SpyCLIP and GoldCLIP, relies on the novel tags fused with RBPs to catalyze the formation of covalent interactions between the tag and its ligand for RNP purification [40, 41]. The covalent system enables denaturing wash conditions and significantly reduces nonspecific binding signals. However, the sensitivity of the current covalent CLIP method could be improved, as suboptimal reaction conditions during library construction may result in significant loss in the capture of transient binding of RBPs to their target RNAs.

In this study, we optimized the SpyCLIP protocol to develop a highly sensitive pipeline while ensuring excellent specificity, which we term ultra SpyCLIP (uSpyCLIP). Using uSpyCLIP, we profiled the binding sites of dPspCas13b-gRNA and dRfxCas13d-gRNA complexes across the transcriptome of HEK293T cells. Our data demonstrated that, in addition to the gRNA-dependent off-target binding sites, dCas13-gRNA complexes exhibit a broad range of gRNA-independent off-target binding sites. Analysis of these off-target sites revealed distinct characteristics of gRNA-independent binding sites, with the DR-distal and middle regions of the gRNA being key determinants of binding specificity. These findings provide valuable insights into the target recognition of dPspCas13b- and dRfxCas13d-gRNA complexes, which may guide future Cas13 engineering and gRNA design for improved specificity in therapeutic applications.

Materials and methods

Plasmid construction

To construct plasmids expressing HA- and Spy-tagged RBPs, the doxycycline (Dox)-inducible lentiviral vector pTRIPZ (Dharmacon) was modified to incorporate N-terminal sequences encoding an HA tag (YPYDVPDYA), a PreScission protease site (LEVLFQGP), and a Spy tag (AHIVMVDAYKPTK). The coding regions of PTBP1, RBFOX2, PspCas13b with C-terminal fused HIV nuclear export signal (NES), dPspCas13b with C-terminal fused HIV NES, RfxCas13d with N- and C-terminal fused nucleoplasmin nuclear localization signal (NLS), and dRfxCas13d with N- and C-terminal fused nucleoplasmin NLS were inserted into the modified vector pTRIPZ between BamH I and Pac I restriction sites.

For PspCas13b gRNA constructs, individual gRNA spacers were cloned into the PspCas13b crRNA backbone (Addgene #103854, a gift from Feng Zhang), which contains a PspCas13b crRNA direct repeat sequence and two BbsI restriction sites for cloning of the spacer sequence. PspCas13b gRNA spacers were designed and synthesized (GenScript) as single-stranded forward and reverse DNA oligos containing CACC and CAAC overhangs, respectively. The forward and reverse DNA oligos were annealed and ligated into the digested PspCas13b crRNA backbone using T4 DNA ligase (New England Biolabs). Similarly, RfxCas13d gRNA spacers were designed and synthesized as single-stranded forward and reverse DNA oligos containing AAAC and AAAA overhangs, respectively, allowing for ligation into the RfxCas13d crRNA backbone (a gift from Lingling Chen). All gRNA spacer sequences are listed in Supplementary Table S1.

To generate a plasmid for SpyCatcher V2.0 expression and purification, the full-length open reading frame (ORF) of SpyCatcher V2.0 was obtained and modified from a previous publication [42] and chemically synthesized from GenScript. Like SpyCatcher V1.0 used in SpyCLIP, an 8-lysine repeat and a 3×(GGGGS) linker were attached to the C-terminus of SpyCatcher V2.0 ORF, and then cloned into the pET-28a expression vector containing a 6× His tag at the N-terminus of the expression cassette through Nco I and Xho I restriction sites. To construct plasmids for the expression and purification of PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d, the corresponding coding regions were inserted into the pET-28a expression vector, which contains an N-terminus 6× His tag, using Nco I and Xho I restriction sites.

Cell culture and construction of stable cell lines

HEK293T and Lenti-X 293T cells were maintained in Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% fetal bovine serum (Ausbian) at 37°C with 5% CO₂. To produce the lentiviruses, HEK293T cells were transfected with a virus vector encoding the expression cassette as well as the VSVG and ∆R8.91 plasmids. Viruses were harvested 48 h post-transfection. RBP-expressing Lenti-X 293T stable cell lines were generated by transduction with lentiviruses for 24 h in the presence of 8 µg/ml polybrene (Sigma–Aldrich), followed by selection with 1 µg/ml puromycin (Gibco) for 1 week. To induce HA- and Spy-tagged RBP expression, 1.5 or 10 ng/ml Dox (BBI LIFE SCIENCES) was added to the cell culture medium, and the cells were harvested 48 h after induction (for induction of HA- and Spy-tagged PTBP1 or RBFOX2, 1.5 ng/ml Dox was used, and for induction of HA- and Spy-tagged PspCas13b, dPspCas13b, RfxCas13d, or dRfxCas13d, 10 ng/ml Dox was used).

Protein expression and purification

pET-28a plasmids expressing His-tagged SpyCatcher V2.0, PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d were individually transformed into Escherichia coli Rosetta (DE3) pLysS competent cells. A single colony was picked and inoculated into 15 ml of Luria-Bertani (LB) medium supplemented with kanamycin overnight at 37°C. Overnight cultures were diluted 100-fold to 1 l of LB medium and grown with shaking at 37°C until the optical density (OD)₆₀₀ reached 0.5. SpyCatcher V2.0 expression was induced with 1 mM isopropyl β-d-thiogalactoside (IPTG) at 30°C for 4 h, and PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d were induced with 0.5 mM IPTG at 16°C overnight. After induction, the cells were collected by centrifugation at 5000 ×  g for 15 min at 4°C for later purification and stored at −80°C.

For SpyCatcher V2.0 protein purification, the cell pellet was resuspended in Ni-NTA binding buffer (20 mM phosphate buffer pH 7.8, 500 mM NaCl, 20 mM imidazole, 1 mM PMSF) and lysed using a high-pressure homogenizer (Uion-Biotech). The lysates were centrifuged at 18 000 × g for 1 h at 4°C and the clarified supernatant was loaded onto a column packed with Ni Sepharose (GE Healthcare). The column was washed sequentially with 10 column volumes (CVs) of Ni-NTA binding buffer, 5 CVs of low-salt Ni-NTA wash buffer (20 mM phosphate buffer pH 7.8, 150 mM NaCl), 10 CVs of high-salt Ni-NTA wash buffer (20 mM phosphate buffer pH 7.8, 2 M NaCl) and another 5 CVs of low-salt Ni-NTA wash buffer, and finally eluted with 7 CVs of Ni-NTA elution buffer (20 mM phosphate buffer pH 7.8, 500 mM NaCl, 150 mM imidazole). Protein was concentrated using an Amicon 15-kDa centrifugal filter (Millipore) and buffer-exchanged into PBS (Thermo Fisher).

To purify His-tagged PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d, cell pellets were resuspended in lysis buffer (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 12 mM β-mercaptoethanol, 0.5 mM PMSF) with 1 mg/ml lysozyme and lysed using a high-pressure homogenizer (Uion-Biotech). The lysates were centrifuged at 18 000 × g for 1 h at 4°C and the clarified supernatant was incubated with Ni Sepharose (GE Healthcare) for 2 h at 4°C. The Sepharose beads were washed with wash buffer (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 20 mM imidazole, 0.5 mM PMSF), and bound proteins were eluted with elution buffer (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 250 mM imidazole, 0.5 mM PMSF) twice. Proteins were concentrated using an Amicon 50-kDa centrifugal filter (Millipore) and buffer-exchanged into storage buffer (20 mM Tris–HCl, pH 7.5, 500 mM NaCl, 5% glycerol).

The purified proteins were quantified using a BCA protein assay kit (Beyotime) and SDS–PAGE, followed by Coomassie Blue staining. The quantified proteins were aliquoted and frozen at −80°C for storage.

Western blot

Collected cells were lysed on ice in ice-cold lysis buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100) supplemented with protease inhibitor cocktail (Sigma–Aldrich) for 15 min. The protein supernatant was collected by centrifuging at 16 000 × g for 10 min at 4°C and mixed with 4× NuPAGE™ LDS buffer with 100 mM DTT. After heating at 70°C for 10 min, the lysed protein samples were resolved by SDS–PAGE and transferred onto a nitrocellulose membrane (Millipore). The target proteins were detected by the indicated antibodies and visualized by enhanced chemiluminescence (ECL) reagent (Thermo Fisher). The following antibodies were used for western blot analysis in this study: PTBP1 polyclonal antibody (Proteintech, 12582-1-AP), RHOA polyclonal antibody (Proteintech, 10749-1-AP), CTSH polyclonal antibody (Proteintech, 10315-1-AP), CDK19 polyclonal antibody (Proteintech, 13761-1-AP), HMGN2 polyclonal antibody (Proteintech, 10953-1-AP), ACVR2B antibody (Abmart, T58048), NUCKS1 polyclonal antibody (Proteintech, 12023-2-AP), HMGA1 polyclonal antibody (Proteintech, 29895-1-AP), EZH2 polyclonal antibody (Proteintech, 21800-1-AP), EFNA3 polyclonal antibody (Proteintech, 12480-1-AP), HPSE polyclonal antibody (Proteintech, 24529-1-AP), SPG7 polyclonal antibody (Proteintech, 27801-1-AP), TRAF3 polyclonal antibody (Proteintech, 18099-1-AP), MDK polyclonal antibody (Proteintech, 11009-1-AP), SCRN1 polyclonal antibody (Proteintech, 14303-1-AP), HA-Tag (26D11) mAb (Abmart, M20003L), GAPDH (3B3) mAb (Abmart, M20006L), ACSL1 polyclonal antibody (Proteintech, 13989-1-AP), FASN polyclonal antibody (Proteintech, 10624-2-AP), RPL14 polyclonal antibody (Proteintech, 14991-1-AP), ACOX1 polyclonal antibody (Proteintech, 10957-1-AP), ACO2 polyclonal antibody (Proteintech, 11134-1-AP), EIF2D polyclonal antibody (Proteintech, 12840-1-AP).

Immunofluorescence analysis

Wild-type or Dox-induced Lenti-X 293T cells were fixed in 4% paraformaldehyde diluted in PBS for 15 min, permeabilized with 0.2% Triton X-100 in PBS for 15 min at 37°C, and then blocked with PBS containing 1% bovine serum albumin and 2% normal donkey serum for 1 h at 25°C. The cells were incubated with anti-PTBP1 (Proteintech, 12582-1-AP) or anti-HA (Abmart, M20003L) antibodies overnight at 4°C. After washing three times with PBST, the cells were incubated with Alexa 488-conjugated donkey anti-mouse IgG (Molecular Probes, A-21202) or Cy3-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch, 711-165-152) for 1 h at 25°C. Nuclei were stained with DAPI for 10 min at 25°C. Fluorescent images were captured using the Olympus SpinSR spinning disk confocal super-resolution microscope.

qPCR analysis

To assess the knockdown activity of gRNAs on target mRNAs and to validate gRNA-dependent off-target effects on essential genes, cells were harvested at 48 h post-transfection, and total RNA was isolated by standard Trizol-chloroform extraction according to the manufacturer’s instructions (Invitrogen). Total RNA (1 µg) was then reverse-transcribed into complementary DNA (cDNA) using the HiScript III RT SuperMix for qPCR (+gDNA wiper) kit (Vazyme) following the manufacturer’s instructions. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed on a QuantStudio 3 Real-Time PCR System (Thermo Fisher) with ChamQ Universal SYBR qPCR Master Mix (Vazyme). The housekeeping gene GAPDH served as an internal control and fold change was calculated relative to GAPDH control using the ∆∆Ct method. The primers for qPCR analysis are listed in Supplementary Table S2.

Electrophoretic mobility shift assay

Cy5-labeled RNA oligonucleotides were designed and synthesized (GenScript), and their sequences are listed in Supplementary Table S3. To validate the binding activity of dPspCas13b and PspCas13b at gRNA-independent off-target sites, the RNA oligonucleotides were heated to 65°C for 5 min and then gradually cooled to 25°C. For dRfxCas13d and RfxCas13d, the RNA oligonucleotides were heated to 65°C for 5 min and then quickly placed on ice. His-tagged PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d were serially diluted into binding buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 1 mM DTT). Twenty nanomolar RNA was mixed with increasing concentrations (25 nM, 50 nM, 100 nM, 200 nM, 300 nM, and 400 nM) of these proteins in 1× binding buffer and incubated at 25°C for 1 h. For validation of gRNA-dependent binding, 200 nM dPspCas13b or dRfxCas13d was mixed with a 1:2 dilution series of gRNA (from 12.5 to 200 nM) in 1× binding buffer at 25°C for 30 min, and then 20 nM Cy5-labeled oligonucleotides were added and incubated for 1 h at 37°C in a 10 µl reaction. Samples were resolved on a 6% native PAGE and run at 160 V for 1 h at room temperature. Cy5 signals on the gel were scanned using an Amersham Typhoon™ Laser Scanning Imager (Cytiva).

In vitro cleavage assay

To validate the cleavage activity of PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d at gRNA-independent off-target sites, RNA oligonucleotides were treated as in the electrophoretic mobility shift assay (EMSA). His-tagged PspCas13b, dPspCas13b, RfxCas13d, and dRfxCas13d were serially diluted into cleavage buffer (50 mM Tris–HCl, pH 7.5, 200 mM NaCl, 10 mM MgCl₂, 1 mM DTT). Twenty nanomolar RNA oligonucleotides were mixed with 200 nM or 400 nM of these proteins in 1× cleavage buffer and incubated at 37°C for 1 h. For validation of gRNA-dependent cleavage, 200 nM PspCas13b, dPspCas13b, RfxCas13d, or dRfxCas13d was mixed with 100 or 200 nM of gRNA in 1× cleavage buffer at 25°C for 30 min, and then 20 nM Cy5-labeled oligonucleotides were added and incubated for 1 h at 37°C in a 10 µl reaction. Reactions were stopped by adding 2× loading buffer and quenched at 75°C for 5 min. Samples were analyzed on a 20% urea denaturing polyacrylamide gel with 1× TBE buffer. Cy5 signals on the gel were scanned using an Amersham Typhoon™ Laser Scanning Imager (Cytiva).

Cell proliferation assay

Cell proliferation was analyzed using CCK-8 assay (MCE) following the manufacturer’s instructions. Briefly, Lenti-X 293T cells expressing catalytically active PspCas13b or RfxCas13d were transfected individually with either 2 µg of targeting gRNAs or non-targeting (NT) controls in six-well plates. At 48 h post-transfection, cells were seeded in 96-well plates at a density of 3 × 10³ cells per well. Subsequently, 10 µl of CCK-8 was added to the wells at 1, 2, 3, 4, and 5 days post-seeding. The absorbance at 450 nm was measured after incubation at 37°C for 2 h.

Cell apoptosis assay

Lenti-X 293T cells expressing catalytically active PspCas13b or RfxCas13d were transfected individually with either 0.5 µg of targeting gRNAs or NT controls in 24-well plates. At 48 h post-transfection, cells were stained with Annexin V/PI (absin) according to the manufacturer’s instructions. Apoptosis was analyzed by flow cytometry (Bio-Rad).

uSpyCLIP method

Preparation of cell pellets

RBP-expressing Lenti-X 293T stable cells were grown to 80% confluency in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum. Cells were rinsed with ice-cold PBS and irradiated on ice with UV-C light at 400 mJ/cm² in a UV crosslinker (UVP, CL-1000). Cells were then collected by scraping, counted, and pelleted by centrifugation. The crosslinked cell pellets were stored at −80°C until use or directly proceeded to the next step.

A. For large amounts of cells (>10⁶ cells)

Cell lysis and RNA fragmentation

Cell pellets were resuspended in 1 ml of lysis buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100, 1× protease inhibitor cocktail) and lysed on ice for 10 min. Cell lysates were clarified by centrifugation at 16 000 × g for 10 min, and soluble fractions were transferred to 1.5-ml LoBind microcentrifuge tubes. Next, 5 µl of RiboLock RNase inhibitor (Thermo Fisher, EO0381) and 2 µl of Turbo DNase (Invitrogen, AM2238) were applied to the lysate. After dilution of RNase I (Invitrogen, AM2295) in PBS at 1:25 on ice, 10 µl of diluted RNase I was added to the lysate and incubated in a thermomixer at 37°C for exactly 5 min at 1200 rpm, followed by 5 min on ice.

HA immunoprecipitation, RNA dephosphorylation, and 3′ adapter ligation

Fifty microliters of HA antibody-coupled magnetic beads (Cell Signaling Technology, 11846) were added to the lysate and rotated at 4°C for 1 h. The rotated mixture was subsequently placed on a magnetic stand and the supernatant was discarded. The beads were washed three times with 1 ml of lysis buffer and twice with 1 ml of PNK wash buffer (20 mM Tris–HCl, pH 7.4, 10 mM MgCl₂, 0.2% Tween 20). Dephosphorylation of 3′ ends of RNase I-treated RNAs was performed in 30 µl of PNK mixture [3 µl 10× PNK buffer, 1.5 µl T4 Polynucleotide Kinase (NEB, M0201), 1 µl RiboLock RNase inhibitor, 1 µl Turbo DNase, 23.5 µl water] at 37°C for 20 min. After washing twice with 1 ml of lysis buffer and once with 1 ml of PNK wash buffer, the beads were resuspended in 30 µl of ligation mixture [3 µl 10× T4 RNA ligase buffer, 2.5 µl 3′ adapter (20 µM), 1 µl RiboLock RNase inhibitor, 1.5 µl T4 RNA Ligase 2, truncated KQ (NEB, M0373), 6 µl 50% PEG8000, 16 µl water] and incubated in a thermomixer at 25°C for 2.5 h with an intermittent vortex for 15 s at 1200 rpm every 2 min. The tube was placed on a magnetic stand and the supernatant was discarded. The beads were washed twice with 1 ml of lysis buffer and once with 1 ml of PSP buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 0.1% Triton X-100, 1 mM ethylenediaminetetraacetic acid (EDTA), 1 mM DTT).

Release RNP from HA beads and SpyCatcher V2.0 pull-down

The beads were resuspended with 50 µl of Elution I mix [48 µl PSP buffer, 1 µl PreScission Protease (GE Healthcare, 27-0843-01), 0.5 µl protease inhibitor cocktail, 0.5 µl RiboLock RNase inhibitor] and incubated in a thermomixer at 22°C for 2 h with a vortex at 1200 rpm every 15 s. The tube was placed on a magnetic stand, the supernatant was transferred to a new 1.5-ml tube, and stored on ice. The beads were subsequently resuspended with 50 µl of Elution II mix [49 µl lysis buffer, 0.5 µl protease inhibitor cocktail, 0.5 µl RiboLock RNase inhibitor] and incubated in a thermomixer at 22°C for 20 min with a vortex at 1200 rpm every 15 s. The Elution II supernatant was collected with a magnetic stand and combined with the Elution I supernatant. The elution mixture was diluted with 170 µl of Spy pull-down buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100) supplemented with 2 µl of protease inhibitor cocktail and 1 µl of RiboLock RNase inhibitor. Next, 30 µl of pre-washed SpyCatcher V2.0 beads were added to the elution mixture and rotated at 25°C for 20 min.

Stringent washes and proteinase K digestion

The beads were washed sequentially with 1 ml of urea wash buffer (100 mM Tris–HCl, pH 7.4, 150 mM NaCl, 8 M urea, 0.2% Triton X-100) twice, 1 ml of Spy pull-down buffer once, 1 ml of SDS wash buffer (100 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1% SDS, 0.2% Triton X-100) twice, 1 ml of Spy pull-down buffer once, 1 ml of high-salt wash buffer (50 mM Tris–HCl, pH 7.4, 2 M NaCl, 0.2% Triton X-100) twice, and 1 ml of low-salt wash buffer (50 mM Tris–HCl, pH 7.4, 0.2% Triton X-100) twice. After washing, the beads were resuspended with 150 µl of proteinase K reaction buffer (50 mM Tris–HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.2% SDS) containing 15 µl of proteinase K (NEB, P8107S) and incubated in a thermomixer at 50°C for 30 min with a vortex at 1200 rpm every 15 s.

RNA recovery by Streptavidin beads and reverse transcription on beads

The tube was placed on a magnetic stand and the supernatant was transferred to a new 1.5-ml tube. Next, 150 µl of Streptavidin pull-down buffer (10 mM Tris–HCl, pH 7.4, 1 M NaCl, 1 mM EDTA, 0.2% Tween 20) containing 30 µl of pre-washed Streptavidin beads (BioMag, BMH1000-1) was mixed with the supernatant and rotated at 25°C for 20 min. After washing twice with 1 ml of Streptavidin pull-down buffer and once with 1 ml of Streptavidin wash buffer (10 mM Tris–HCl, pH 7.4, 0.2% Tween 20), the beads were resuspended in 11 µl of DEPC water, 0.5 µl of RT primer (100 µM), and 2 µl of dNTP mix (10 mM each). After denaturation at 65°C for 5 min and snap-chilling on ice for 3 min, the reverse transcription mixture [4 µl 5× reaction buffer, 2 µl BeyoRT II M-MLV reverse transcriptase (Beyotime), 0.5 µl RiboLock RNase inhibitor] was added to the tube and rotated at 44°C for 1 h.

RNase H digestion and UMI adapter ligation on beads

The beads were washed twice with 1 ml of Streptavidin pull-down buffer and once with 1 ml of Streptavidin wash buffer, resuspended with 20 µl of RNase H mixture [2 µl 10× RNase H buffer, 1 µl RNase H (NEB, M0297S), 17 µl DEPC water], and rotated at 37°C for 20 min. After washing the beads twice with 0.5 ml of Streptavidin wash buffer, the beads were resuspended with 30 µl of ligation mixture [9.2 µl DEPC water, 0.5 µl UMI adapter (100 µM), 2 µl 100% DMSO, 0.3 µl ATP (100 mM), 3 µl 10× T4 RNA ligase buffer, 3 µl T4 RNA Ligase 1, high concentration (NEB, M0437), 12 µl 50% PEG8000] and mixed by slowly pipetting up and down until the beads were resuspended. The reaction was rotated at 25°C overnight.

PCR amplification of the cDNA library and size-selection of the PCR products

The beads were washed twice with 0.5 ml of Streptavidin wash buffer and resuspended in 30 µl of nuclease-free water. The cDNA library was first amplified with the following mixture: 15 µl of beads resuspended in nuclease-free water, 2 µl of P5 primer (10 µM), 2 µl of P7 primer (10 µM), 5 µl of 10× buffer for KOD-plus-neo, 5 µl of dNTPs (2 mM), 3 µl of MgSO₄ (25 mM), 1 µl of KOD-plus-neo (TOYOBO), and 17 µl of nuclease-free water. The amplification was performed with the following program: 94°C for 2 min; 98°C for 10 s, 60°C for 30 s, and 68°C for 30 s (6 cycles); 68°C for 5 min, then held at 12°C. The PCR tube was subsequently placed on a magnetic stand for 3 min and the supernatant was transferred to a new 0.2-ml PCR tube for purification with 90 µl of homemade (HM) beads (1.8:1 ratio). After incubation for 5 min, the beads were washed twice with 80% ethanol, dried for 5 min, and eluted in 25 µl of water. The PCR tube was subsequently placed on a magnetic stand, 24.6 µl of eluate was transferred to a new 0.2-ml PCR tube, and 15.4 µl of PCR master mix [2 µl P5 primer (10 µM), 2 µl P7 primer (10 µM), 4 µl 10× buffer for KOD-plus-neo, 4 µl dNTPs (2 mM), 2.4 µl MgSO₄ (25 mM), 1 µl KOD-plus-neo] was added to the tube for final amplification. The PCR program was set as follows: 94°C for 2 min; 98°C for 10 s, 60°C for 30 s, and 68°C for 30 s (2–12 cycles); 68°C for 5 min, then held at 12°C. The PCR products were size-selected, PCR products between 165 and 250 bp were excised from a 6% polyacrylamide gel and purified using ethanol precipitation. Quality control was performed using the Agilent High Sensitivity DNA Kit (Agilent) and libraries were quantified using the Qubit Fluorometer (Invitrogen, Q32866). The library was sequenced using Illumina NovaSeq 6000 at WuXi NextCODE.

B. For low input cells (1000–10 000 cells)

Cell lysis and RNA fragmentation

To unbiasedly evaluate the performance of the uSpyCLIP method, 10 µl of cell lysates were diluted with 40 µl of dilution buffer (50 mM Tris–HCl, pH 7.4, 1% Triton X-100) containing 1 µl of Turbo DNase, 0.5 µl of protease inhibitor cocktail, and 0.5 µl of RiboLock RNase inhibitor in a 0.2-ml PCR tube to bring the number of input cells to 10 000 or 1000. After dilution of RNase I in PBS at 1:2000 (10 000 cells) or 1:5000 (1 000 cells) on ice, 10 µl of diluted RNase I was added to the lysate and incubated in a PCR amplifier at 37°C for exactly 5 min. Next, 1.4 µl of 5 M NaCl, 0.46 µl of 10% SDS, and 4.6 µl of 5% sodium deoxycholate were added to the lysate to optimize the buffer for immunoprecipitation.

HA immunoprecipitation, RNA dephosphorylation, and 3′ adapter ligation

Ten microliters of HA antibody-coupled magnetic beads were added to the lysate and rotated at 4°C for 1 h. The rotated mixture was subsequently placed on a magnetic stand and the supernatant was discarded. The beads were washed twice with 0.2 ml of lysis buffer and once with 0.2 ml of PNK wash buffer (20 mM Tris–HCl, pH 7.4, 10 mM MgCl₂, 0.2% Tween 20). Dephosphorylation of 3′ ends of RNase I-treated RNAs was performed in 20 µl of PNK mixture [2 µl 10× PNK buffer, 1 µl T4 Polynucleotide Kinase, 0.5 µl RiboLock RNase inhibitor, 0.5 µl Turbo DNase, 16 µl water] at 37°C for 20 min. After washing twice with 0.2 ml of lysis buffer and once with 0.2 ml of PNK wash buffer, the beads were resuspended in 20 µl of ligation mixture [2 µl 10× T4 RNA ligase buffer, 1 µl 3′ adapter (20 µM), 0.5 µl RiboLock RNase inhibitor, 1 µl T4 RNA Ligase 2, truncated KQ, 4 µl 50% PEG8000, 11.5 µl water] and incubated in a thermomixer at 25°C for 2.5 h with an intermittent vortex for 15 s at 1 200 rpm every 2 min. The tube was placed on a magnetic stand and the supernatant was discarded. The beads were washed twice with 0.2 ml of lysis buffer and once with 0.2 ml of PSP buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 0.1% Triton X-100, 1 mM EDTA, 1 mM DTT).

Release RNP from HA beads and SpyCatcher V2.0 pull-down

The beads were resuspended with 50 µl of Elution I mix [48 µl PSP buffer, 1 µl PreScission Protease, 0.5 µl protease inhibitor cocktail, 0.5 µl RiboLock RNase inhibitor] and incubated in a thermomixer at 22°C for 1 h with a vortex at 1200 rpm every 15 s. The tube was placed on a magnetic stand, the supernatant was transferred to a new 1.5-ml tube, and stored on ice. The beads were subsequently resuspended with 50 µl of Elution II mix [49 µl lysis buffer, 0.5 µl protease inhibitor cocktail, 0.5 µl RiboLock RNase inhibitor] and incubated in a thermomixer at 22°C for 20 min with a vortex at 1200 rpm every 15 s. The Elution II supernatant was collected with a magnetic stand and combined with the Elution I supernatant. The elution mixture was diluted with 80 µl of Spy pull-down buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100) supplemented with 2 µl of protease inhibitor cocktail and 1 µl of RiboLock RNase inhibitor. Next, 20 µl of pre-washed SpyCatcher V2.0 beads were added to the elution mixture and rotated at 25°C for 20 min.

Stringent washes and proteinase K digestion

The beads were washed sequentially with 0.2 ml of urea wash buffer (100 mM Tris–HCl, pH 7.4, 150 mM NaCl, 8 M urea, 0.2% Triton X-100) twice, 0.2 ml of Spy pull-down buffer once, 0.2 ml of SDS wash buffer (100 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1% SDS, 0.2% Triton X-100) twice, 0.2 ml of Spy pull-down buffer once, 0.2 ml of high-salt wash buffer (50 mM Tris–HCl, pH 7.4, 2 M NaCl, 0.2% Triton X-100) twice, and 0.2 ml of low-salt wash buffer (50 mM Tris–HCl, pH 7.4, 0.2% Triton X-100) twice. After washing, the beads were resuspended with 50 µl of proteinase K reaction buffer (50 mM Tris–HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.2% SDS) containing 5 µl of proteinase K and incubated in a thermomixer at 50°C for 30 min with a vortex at 1200 rpm every 15 s.

RNA recovery by streptavidin beads and reverse transcription on beads

The tube was placed on a magnetic stand and the supernatant was transferred to a new 1.5-ml tube. Next, 100 µl of Streptavidin pull-down buffer (10 mM Tris–HCl, pH 7.4, 1 M NaCl, 1 mM EDTA, 0.2% Tween 20) containing 15 µl of pre-washed Streptavidin beads was mixed with the supernatant and rotated at 25°C for 20 min. After washing twice with 0.2 ml of Streptavidin pull-down buffer and once with 0.2 ml of Streptavidin wash buffer (10 mM Tris–HCl, pH 7.4, 0.2% Tween 20), the beads were resuspended in 11 µl of DEPC water, 0.5 µl of RT primer (100 µM), and 2 µl of dNTP mix (10 mM each). After denaturation at 65°C for 5 min and snap-chilling on ice for 3 min, the reverse transcription mixture [4 µl 5× reaction buffer, 2 µl BeyoRT II M-MLV reverse transcriptase, 0.5 µl RiboLock RNase inhibitor) was added to the tube and rotated at 44°C for 1 h.

RNase H digestion and UMI adapter ligation on beads

The beads were washed twice with 0.2 ml of Streptavidin pull-down buffer and once with 0.2 ml of Streptavidin wash buffer, resuspended with 20 µl of RNase H mixture [2 µl 10× RNase H buffer, 1 µl RNase H, 17 µl DEPC water], and rotated at 37°C for 20 min. After washing the beads twice with 0.2 ml of Streptavidin wash buffer, the beads were resuspended with 30 µl of ligation mixture [9.2 µl DEPC water, 0.5 µl UMI adapter (100 µM), 2 µl 100% DMSO, 0.3 µl ATP (100 mM), 3 µl 10× T4 RNA ligase buffer, 3 µl T4 RNA Ligase 1, high concentration, 12 µl 50% PEG8000] and mixed by slowly pipetting up and down until the beads were resuspended. The reaction was rotated at 25°C overnight.

PCR amplification of the cDNA library and size-selection of the PCR products

The beads were washed twice with 0.2 ml of Streptavidin wash buffer and resuspended in 20 µl of nuclease-free water. The cDNA library was first amplified with the following mixture: 20 µl of beads resuspended in nuclease-free water, 2 µl of P5 primer (10 µM), 2 µl of P7 primer (10 µM), 5 µl of 10× buffer for KOD-plus-neo, 5 µl of dNTPs (2 mM), 3 µl of MgSO₄ (25 mM), 1 µl of KOD-plus-neo (TOYOBO), and 12 µl of nuclease-free water. The amplification was performed with the following program: 94°C for 2 min; 98°C for 10 s, 60°C for 30 s, and 68°C for 30 s (6 cycles); 68°C for 5 min, then held at 12°C. The PCR tube was subsequently placed on a magnetic stand for 3 min and the supernatant was transferred to a new 0.2-ml PCR tube for purification with 90 µl of HM beads (1.8:1 ratio). After incubation for 5 min, the beads were washed twice with 80% ethanol, dried for 5 min, and eluted in 25 µl of water. The PCR tube was subsequently placed on a magnetic stand, 24.6 µl of eluate was transferred to a new 0.2-ml PCR tube, and 15.4 µl of PCR master mix [2 µl P5 primer (10 µM), 2 µl P7 primer (10 µM), 4 µl 10× buffer for KOD-plus-neo, 4 µl dNTPs (2 mM), 2.4 µl MgSO₄ (25 mM), 1 µl KOD-plus-neo] was added to the tube for final amplification. The PCR program was set as follows: 94°C for 2 min; 98°C for 10 s, 60°C for 30 s, and 68°C for 30 s (9 or 11 cycles); 68°C for 5 min, then held at 12°C. The PCR products were size-selected, PCR products between 165 and 250 bp were excised from a 6% polyacrylamide gel and purified using ethanol precipitation. Quality control was performed using the Agilent High Sensitivity DNA Kit (Agilent) and libraries were quantified using the Qubit Fluorometer (Invitrogen, Q32866). The library was sequenced using Illumina NovaSeq 6000 at WuXi NextCODE. All oligos used in the uSpyCLIP protocol are listed in Supplementary Table S4.

Construction of dPspCas13b and dRfxCas13d uSpyCLIP libraries

Cells were prepared according to the designed research program. Lenti-X 293T cells stably expressing HA- and Spy-tagged dPspCas13b or dRfxCas13d were plated at a density of 5 000 000 cells in a 10-cm plate and transfected at 80% cell confluence using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Twenty micrograms or ten micrograms of indicated gRNA expression plasmids were used for dPspCas13b or dRfxCas13d, respectively. To induce HA- and Spy-tagged dPspCas13b and dRfxCas13d expression, the Opti-MEM serum-reduced medium was replaced with DMEM medium supplemented with 10% fetal bovine serum and 10 ng/ml Dox. The cells were irradiated twice on ice with UV-C light at 400 mJ/cm² in a UV crosslinker at 48 h after induction. The uSpyCLIP libraries were constructed following the protocol for large amounts of cells, with HA antibody-coupled magnetic beads (Cell Signaling Technology, 11846) replaced by anti-HA magnetic beads (MBL, M180-11).

High-throughput sequencing and mapping

High-throughput sequencing of the uSpyCLIP libraries was performed on the Illumina NovaSeq 6000 platform. The adapter sequence at the 3′ end of reads was trimmed using Trim-Galore (v0.0.6), and the 10-nucleotide unique molecular identifier (UMI) at the 5′ end of the reads was removed and used as the read identity for further deduplication. The remaining reads were mapped to the human genome (hg38) using the STAR program (v2.7.9a) [43] with the parameters (--twopassMode Basic --outFilterMultimapNmax 1 --alignEndsType Extend5pofRead1). All mapped reads can be visualized on the Integrative Genomics Viewer (IGV).

Identification of uSpyCLIP clusters

After sorting and indexing the output BAM file from the STAR software by Samtools (v1.9), PCR duplicates (reads that mapped to the same genomic position and had identical UMIs) were collapsed into a single read using the deduplication function of UMI-tools (v1.0.1) [44]. The remaining usable reads were used for cluster identification using the iCount program with a 7-nt clustering window (v2.0.0). Pearson’s correlation coefficients between replicates or between different samples were performed as previously described [40]. In brief, the number of usable reads in each cluster was normalized to RPM (reads per million genome-mapped reads) and considered as the abundance. The correlation was determined by comparing the abundance of all clusters.

Identification of off-target binding sites of the dPspCas13b-gRNA complex and the dRfxCas13d-gRNA complex

Specific clusters were identified for each group of uSpyCLIP libraries designed in Fig. 2A. For the identification of gRNA-independent off-target binding sites, the identified clusters of Group A and Group B were combined, the abundance of each cluster in both groups were counted, and the fold enrichment ratio of the read count in each cluster was calculated. The clusters in Group B that exhibited at least eight-fold enrichment and an adjusted P-value < 0.05 over their counterparts in Group A were defined as gRNA-independent off-target sites. To identify the gRNA-dependent off-target binding sites of the dPspCas13b-gRNA complex, the fold enrichment ratio of the read count in the clusters from Group D versus their counterparts in Group B was calculated, and the clusters from Group D that exhibited at least four-fold enrichment and the adjusted P-value < 0.05 over their counterparts in Group B were defined as gRNA-dependent off-target sites. To identify the gRNA-dependent off-target binding sites of the dRfxCas13d-gRNA complex, the fold enrichment ratio of the read count in the clusters from Group D versus their counterparts in Group C was calculated, and the clusters from Group D that exhibited at least four-fold enrichment and the adjusted P-value < 0.05 over their counterparts in Group C were defined as gRNA-dependent off-target sites.

Figure 2.

Use uSpyCLIP to identify the off-target binding sites of the dPspCas13b-gRNA and dRfxCas13d-gRNA complexes. (A) Schematic representation of the strategy for identifying the off-target binding sites of dCas13-gRNA complex. Take dCas13d as an example. (B) Schematic for constructs expressing Cas13 effectors and corresponding gRNA cassettes. (C) Western blot of protein extracts from Lenti-X 293T cells stably expressing HA- and Spy-tagged dPspCas13b or dRfxCas13d and their catalytically active counterparts. (D) Immunocytochemistry of dPspCas13b and dRfxCas13d proteins showing localization and expression. (E) RNA knockdown activity of endogenous EZH2 or TP53 mRNA in Lenti-X 293T cells expressing HA- and Spy-tagged PspCas13b or dPspCas13b. (F) RNA knockdown activity of endogenous EZH2 or TP53 mRNA in Lenti-X 293T cells expressing HA- and Spy-tagged RfxCas13d or dRfxCas13d. gNT, non-targeting gRNA; EZH2-g1 and EZH2-g2, two independent gRNAs targeting EZH2 mRNA; TP53-g1, gRNA targeting TP53 mRNA. Values are shown as mean ± SD with n = 3.

Annotation and de novo motif identification of RBP-binding clusters

The genomic annotation of human genes and gene elements was downloaded from ENSEMBL. Off-target binding sites were finely categorized into non-coding RNA (including miRNA, rRNA, tRNA, snRNA, snoRNA, and lncRNA), coding RNA elements (including 5′ UTR, CDS, intron, and 3′ UTR), and others. De novo motif identification was performed using HOMER (v5.1) [45]. The region (−50 to +50 nt) around the apex of each cluster was extracted and findMotifsGenome.pl was used with the parameters [-p 8 -S 10 -rna (-len 6 was used for the PTBP1 and RBFOX2)].

RNA secondary structure analysis of gRNA-independent off-target binding sites

The regions (−40 to +40 nt) around the apex of the gRNA-independent off-target binding sites were extracted and RNA secondary structures of these regions were predicted using RNAfold (v2.6.4) [46]. The pairing probability at each position across the region was calculated to create a pairing ratio map. Representative examples of gRNA-independent off-target binding sites were plotted using the Mfold web server (v2.3) [47].

Prediction of gRNA target sites and analysis of gRNA targeting rules

The nucleotide sequences from the extended off-target binding regions were extracted and reverse complementary paired with the corresponding gRNAs. A scoring system was employed as follows: consecutive base pairs were scored as 1, non-consecutive base pairs were scored as 0.5, and unpaired bases were scored as 0. The subregion with the highest score was designated as the gRNA target site. For the identification of gRNA targeting rules, the number of consecutively complementary pairs between gRNA-dependent off-target sites and their corresponding gRNAs or random controls was counted, and the ratio of the number of off-target sites with different numbers of consecutive pairs to the total number of off-target sites was calculated. The positions of base pairing between the top 50 gRNA-dependent off-target sites and their corresponding gRNAs were visualized using a heatmap generated with the pheatmap R package.

mRNA library construction and sequencing

mRNA libraries were constructed using Smart-seq2 as previously described [48]. Briefly, 10 ng of total RNA was reverse transcribed using ProtoScript II Reverse Transcriptase (NEB, M0368). After PCR pre-amplification, cDNA libraries were purified using Agencourt Ampure XP beads (Beckman Coulter, A63881), and 2 ng of purified cDNA was used for the next tagmentation reaction. The final amplified libraries were purified again by Agencourt Ampure XP beads and sequenced at 2 × 150 bp on the Illumina NovaSeq 6000 platform.

Data analysis of RNA sequencing and differential expression analysis

Adapters were trimmed from the raw paired-end fastq reads using Trim-Galore (v0.0.6) with the parameters --quality 20 --stringency = 1 --length = 35 --paired. The remaining reads were mapped to the human genome (hg38) using the STAR program (v2.7.9a) [43]. Gene abundance was normalized as TPM (transcripts per million). Only genes with an average TPM value of at least 5 were retained for differential expression analysis. Differentially expressed genes were identified using the limma package with the thresholds of log₂(fold change) > 1.5 and FDR < 0.01.

Statistical analyses

All experiments were performed in two or three biological replicates. Statistical analyses were carried out using GraphPad software and data are presented as the mean ± SD. Statistical significance was determined using an unpaired two-tailed t-test. P-values are indicated in the figures. A P-value of < 0.05 was considered significant.

Results

Development of uSpyCLIP method with improved sensitivity

We optimized the SpyCLIP protocol by enhancing several key steps (Fig. 1A and Supplementary Fig. S1): (i) the RBP fusion tags were reduced to 36 amino acids (Supplementary Fig. S2A and B), and the library preparation process was streamlined to 2 days (Supplementary Fig. S1); (ii) the RNase I concentration and reaction conditions were optimized to improve RNA fragmentation efficiency; (iii) SpyCatcher was upgraded to SpyCatcher V2.0 [42], enabling faster and more efficient RBP pull-downs, which allowed for the identification of a greater number of binding sites (Supplementary Fig. S2C–F); and (iv) a bead-based, single-tube strategy was implemented for the second adapter ligation, minimizing sample loss and reducing unwanted by-products. To validate the uSpyCLIP method, we compared it to existing techniques, including SpyCLIP, iCLIP, and eCLIP, using CLIP datasets from PTBP1 and RBFOX2. Our results showed that uSpyCLIP yielded significantly higher percentages of uniquely mapped and usable reads, and identified substantially more binding sites, all with a comparable number of cells (Fig. 1B and Supplementary Fig. S3B, D, and E).

Figure 1.

uSpyCLIP enables efficient and time-effective identification of RBP binding sites. (A) Schematic of uSpyCLIP procedure with major changes highlighted in red italics. (B) Mapping rate comparison for uSpyCLIP, SpyCLIP, and iCLIP methods. All CLIP datasets were mapped to the genome using the same pipeline at the same sequencing depth. (C) Saturation analysis of unique fragments in different PTBP1 CLIP datasets. uSpyCLIP generated a higher-complexity library, even with fewer cells. (D) Read density tracks of different PTBP1 CLIP data on the known PTBP1 target DDX17. (E) Top HOMER motifs identified for all PTBP1 clusters from uSpyCLIP or the published iCLIP and SpyCLIP datasets.

When we reduced the number of input cells, the correlation between PTBP1 binding sites identified in two biological replicates remained high and closely matched the results obtained using bulk cells (Supplementary Fig. S3A). uSpyCLIP with 100 000 input cells retained a significant proportion of usable reads and identified more binding sites than SpyCLIP with 10 000 000 cells. Remarkably, uSpyCLIP with just 10 000 cells outperformed iCLIP with 10 000 000 cells (Fig. 1B and Supplementary Fig. S3B). Saturation analysis of unique fragments revealed that uSpyCLIP generated a higher-complexity library, even with fewer cells (Fig. 1C). Notably, this increased sensitivity did not come at the cost of reduced specificity, and both binding sites and motifs were accurately identified even with a reduced number of cells (Fig. 1D and E and Supplementary Fig. S3C and F). These results demonstrate that uSpyCLIP is a highly sensitive, time-efficient method for identifying RBP binding sites, with excellent specificity.

Identifying off-target binding sites of dPspCas13b-gRNA and dRfxCas13d-gRNA

To investigate the binding sites of dCas13 and the dCas13-gRNA complex, we used Lenti-X 293T cells stably expressing HA- and Spy-tagged dCas13 as cell models. Experiments were designed with four different sample groups (Fig. 2A). Group A, which was not transfected with gRNA and did not undergo UV crosslinking, served as the background noise control. In this group, RNAs bound to dCas13 are removed under denaturing washing conditions due to the absence of UV crosslinking. Group B, which was also not transfected with gRNA but underwent UV crosslinking, represented background noise as well as gRNA-independent off-target binding sites. Group C, transfected with a non-targeting gRNA and subjected to UV crosslinking, provided a control for identifying gRNA-dependent binding sites. Finally, Group D, transfected with a gRNA targeting a specific gene transcript and subjected to UV crosslinking, included the signals of background noise, gRNA-independent off-target binding sites, gRNA-dependent off-target binding sites, and the known on-target site. To identify gRNA-independent off-target binding sites, we subtracted signals in Group A libraries from those in Group B libraries. Similarly, gRNA-dependent off-target binding sites were identified by subtracting signals in Group B or Group C libraries from those in Group D.

We constructed Lenti-X 293T cell lines stably expressing HA- and Spy-tagged dPspCas13b or dRfxCas13d, along with their catalytically active wild-type counterparts. Since PspCas13b and RfxCas13d have been reported to be active when localized to the cytoplasm and nucleus, respectively [6, 7, 49], we added NES to dPspCas13b and NLS to dRfxCas13d (Fig. 2B). The expression and localization of these proteins were confirmed by western blot and immunocytochemistry (Fig. 2C and D). Next, we selected three validated gRNAs from previous studies: two targeting different sites within the EZH2 transcript (EZH2-g1 and EZH2-g2) and one targeting the TP53 transcript (TP53-g1) [7, 50]. Each of these gRNAs harbors a 30-nt spacer sequence. To verify the knockdown activity of these gRNAs, we transiently transfected a non-targeting gRNA (gNT) or the targeting gRNAs into cells stably expressing either Cas13 or dCas13. RT-qPCR analysis showed that the three gRNAs significantly knocked down their target mRNAs in cells expressing Cas13, but not in cells expressing dCas13, confirming the activity of both Cas13 and the gRNAs (Fig. 2E and F).

We then performed uSpyCLIP-seq analysis of dPspCas13b and dRfxCas13d following the designed strategy (Fig. 2A). For each gRNA tested, specific gRNA-dependent enrichment was observed at the on-target site, demonstrating that uSpyCLIP enables the accurate identification of binding sites for the dCas13-gRNA complex (Fig. 3A). Intriguingly, although target specificity of Cas13 system has been thought to be largely determined by the guide sequence of the gRNA, our bioinformatics analysis revealed that both dPspCas13b and dRfxCas13d intrinsically bind to thousands of sites across the transcriptome, which are defined as gRNA-independent off-target binding sites (Fig. 3B and E). These off-target binding sites showed high Pearson correlation coefficients between biological replicates, indicating the high reproducibility of uSpyCLIP in identifying off-target binding sites (Supplementary Fig. S4). Notably, the number of gRNA-dependent off-target binding sites varied significantly among the three gRNAs tested. For dPspCas13b, EZH2-g2 resulted in 625 off-target sites, while EZH2-g1 yielded only 59. For dRfxCas13d, nearly 2000 off-target sites were identified for both EZH2-g2 and TP53-g1, but only 966 for EZH2-g1 (Fig. 3C and F). The off-target binding sites for both dPspCas13b and dRfxCas13d were primarily located in the coding sequence (CDS) and 3′ untranslated regions (3′ UTR) of mRNAs (Fig. 3D and G). However, dRfxCas13d exhibited a higher number of intronic binding sites compared to dPspCas13b. Gene Ontology (GO) analysis revealed that gRNA-independent off-target sites bound by dPspCas13b and dRfxCas13d were significantly enriched in biological processes such as cytoplasmic translation and RNA localization (Supplementary Figs S5 and S6). In contrast, gRNA-dependent off-target sites were enriched in distinct biological processes. These findings suggest that Cas13-gRNA off-target sites are associated with numerous critical biological functions, which could potentially impact cellular activities.

Figure 3.

Both dPspCas13b-gRNA and dRfxCas13d-gRNA complexes have a wide range of gRNA-independent and gRNA-dependent off-target binding sites. (A) uSpyCLIP signals around on-target sites. On-target sites are indicated by colored rectangles. EZH2 target site 1, EZH2 target site 2, and TP53 target site 1 are indicated by red, green, and blue rectangles, respectively. (B and E) Fold enrichment ratio of the read count in the clusters from dPspCas13b (B) or dRfxCas13d (E) uSpyCLIP with UV crosslinking versus their counterparts without UV crosslinking. The clusters in dCas13 uSpyCLIP with UV crosslinking that exhibited at least eight-fold enrichment and the P-value < 0.05 over their counterparts without UV crosslinking are indicated by black dots and defined as gRNA-independent off-target sites. The number of gRNA-independent off-target sites bound by dPspCas13b and dRfxCas13d is 4701 and 2808, respectively. (C) Fold enrichment ratio of the read count in the clusters from dPspCas13b uSpyCLIP transfected with one of the three gRNAs (EZH2-g1, EZH2-g2, and TP53-g1) versus their counterparts without gRNA transfection. The clusters in dPspCas13b uSpyCLIP transfected with gRNAs that exhibited at least four-fold enrichment and the P-value < 0.05 over their counterparts without gRNA transfection are indicated by blue dots and defined as gRNA-dependent off-target sites. The number of gRNA-dependent off-target sites for the three gRNAs of dPspCas13b is 59, 625, and 123, respectively. (D and G) Genomic distribution of the gRNA-independent, EZH2, and TP53 gRNA-dependent off-target sites for the dPspCas13b-gRNA complex (D) and the dRfxCas13d-gRNA complex (G). (F) Fold enrichment ratio of the read count in the clusters from dRfxCas13d uSpyCLIP transfected with one of the three gRNAs (EZH2-g1, EZH2-g2, and TP53-g1) versus their counterparts transfected with a non-targeting gRNA. The clusters in dRfxCas13d uSpyCLIP transfected with gRNAs that exhibited at least four-fold enrichment and the P-value < 0.05 over their counterparts transfected with the non-targeting gRNA are indicated by blue dots and defined as gRNA-dependent off-target sites. The number of gRNA-dependent off-target sites for the three gRNAs of dRfxCas13d is 966, 1994, and 2138, respectively.

gRNA-independent off-target sites of dPspCas13b and dRfxCas13d exhibit distinct characteristics

To investigate the sequence and structural characteristics of gRNA-independent off-target sites of dPspCas13b and dRfxCas13d, we analyzed the sequences of these off-target sites using HOMER and RNAfold software. Approximately 30% of the gRNA-independent off-target sites of dPspCas13b showed enrichment of U-rich binding motifs. The UUUG motif was prevalent across all binding sites, while the UUUUGAG motif was notably enriched among the top 500 binding sites (Fig. 4A). The base-pairing frequency at these binding sites displayed a distinct pattern, with a central valley flanked by two peaks in the adjacent regions (−40 to +40 nt), resembling the binding behavior of SLBP, which is known to bind stable stem-loop structures in the 3′ UTR of histone mRNA (Fig. 4B). This suggests that dPspCas13b preferentially binds to the loop region of long stem-loop structures. In contrast, ~17% of the gRNA-independent off-target sites of dRfxCas13d exhibited a clear CAACUG binding motif. The base-pairing frequency at these sites was elevated compared to the more distal surrounding regions (Fig. 5A and B).

Figure 4.

dPspCas13b preferentially binds to RNA sites with long stem-loop structures. (A) Top HOMER motif identified for all dPspCas13b binding sites or the top 500 dPspCas13b binding sites. (B) The frequency of base pairing for all dPspCas13b binding sites or the top 500 dPspCas13b binding sites. RBFOX2 and SLBP were used as controls. (C) Schematic of the PspCas13b crRNA direct repeat region. (D) The sequence composition and secondary structure of three representative gRNA-independent off-target sites bound by dPspCas13b. The secondary structure was calculated using RNAfold. (E) EMSA assays examining binding of His-tagged dPspCas13b to Cy5-labeled RNA oligonucleotides corresponding to the identified off-target sites and a scrambled control.

Figure 5.

A distinct CAACUG motif is enriched in gRNA-independent off-target sites bound by dRfxCas13d. (A) Top HOMER motif identified for all dRfxCas13d binding sites or the top 500 dRfxCas13d binding sites. (B) The frequency of base pairing for all dRfxCas13d binding sites or the top 500 dRfxCas13d binding sites. RBFOX2 and SLBP were used as controls. (C) Schematic of the RfxCas13d crRNA direct repeat region. (D) The sequence composition and secondary structure of three representative gRNA-independent off-target sites bound by dRfxCas13d. The secondary structure was calculated using RNAfold. (E) EMSA assays examining binding of His-tagged dRfxCas13d to Cy5-labeled RNA oligonucleotides corresponding to the identified off-target sites and a scrambled control.

Given that Cas13 proteins typically bind to the DR region of the gRNA, we hypothesized that dPspCas13b and dRfxCas13d may recognize off-target sites through sequence or structural features similar to the gRNA DRs. To test this, we analyzed the sequence composition of the DR regions in PspCas13b and RfxCas13d. We found that the DR of PspCas13b forms a long stem-loop structure with a U-rich loop region, containing the UUUUGAG motif. In contrast, the DR of RfxCas13d forms a shorter stem-loop structure, with the loop region corresponding to the CAACUG motif (Figs 4C and 5C). We further selected three top-ranked off-target sites for each Cas13 variant and analyzed the sequence composition and structure of the extended regions (−20 to +20 nt) around the center of these sites. For dPspCas13b, the off-target sites in the CDS of TOMM20, the 3′ UTR of MYL12A, and the 3′ UTR of IARS1 all exhibited long stem-loop structures with U-rich loop regions (Fig. 4D and Supplementary Fig. S7A). Notably, while the loop regions of these stem-loops did not always strictly match the UUUG or UUUUGAG motifs, the presence of a stem-loop structure was a common feature, suggesting that, in addition to U-rich sequences, the overall structure of the stem-loop may be critical for dPspCas13b binding. For dRfxCas13d, the three off-target sites in the mRNAs of ZDHHC6, GNS, and RNGTT formed less stable structures, but all contained the CAACUG motif in the loop region (Fig. 5D and Supplementary Fig. S7B). To assess whether the identified motifs are sufficient to mediate Cas13 binding in vitro, we performed EMSA using purified Cas13 proteins and Cy5-labeled, single-stranded RNA (ssRNA) oligonucleotides. The addition of recombinant dPspCas13b or PspCas13b produced a clear supershift for the RNA oligonucleotides corresponding to the off-target sites in MYL12A and TOMM20 transcripts, but not for a scrambled control oligonucleotide (Fig. 4E and Supplementary Fig. S7C). Likewise, both dRfxCas13d and RfxCas13d generated a supershift with the RNA oligonucleotides corresponding to the off-target sites in ZDHHC6 and RNGTT transcripts, whereas no binding was observed with a scrambled RNA oligonucleotide lacking the CAACUG motif (Fig. 5E and Supplementary Fig. S7D). Together, these findings suggest that PspCas13b and RfxCas13d bind to RNA sites with structural and sequence features resembling their corresponding gRNA DR regions, but exhibit distinct binding characteristics. To determine whether these bound off-target sites could be cleaved by Cas13, we conducted in vitro cleavage assays using catalytically active Cas13 proteins. The results showed that none of the off-target RNAs were cleaved (Supplementary Fig. S8A and B), indicating that motif-specific recognition alone is insufficient to trigger the nuclease activity of PspCas13b and RfxCas13d.

Characteristics of gRNA-dependent off-target sites of dPspCas13b and dRfxCas13d

To investigate the gRNA targeting rules of dPspCas13b and dRfxCas13d, we first analyzed the number of consecutive complementary base pairs between the gRNA-dependent off-target sites and their corresponding gRNAs. More than 50% of the off-target sites for dPspCas13b contained at least 12 consecutive complementary base pairs with their gRNAs, and 80% had at least 10 consecutive complementary pairs. In contrast, only 19% of the off-target sites for dRfxCas13d exhibited at least 12 consecutive complementary pairs, while 63% of dRfxCas13d off-target sites had fewer than 10 consecutive complementary pairs (Fig. 6A–F). We further analyzed the effect of internal mismatches within the 10–12 nt region of sequence homology, and the data suggest that off-target recognition depends on both the length and continuity of gRNA–RNA pairing. Longer pairing regions can tolerate more internal mismatches; i.e. a 12-nt complementary region can accommodate up to four mismatches (Supplementary Fig. S9). Notably, dPspCas13b displays heightened sensitivity to internal mismatches, as increasing mismatch numbers lead to a rapid reduction in detectable off-target sites. In contrast, dRfxCas13d is more tolerant of internal mismatches, maintaining off-target binding despite greater sequence divergence. Together, these findings indicate that internal mismatches within the core homologous region substantially influence off-target binding in a Cas13 variant-dependent manner.

Figure 6.

The DR-distal and middle regions of the gRNA are critical in determining binding specificity. (A–F) Density plots of the number of consecutive complementary pairs between gRNA-dependent off-target sites and their corresponding gRNAs. The horizontal axis indicates the number of consecutively paired bases, and the vertical axis indicates the ratio of the number of off-target sites corresponding to different numbers of consecutively paired bases to the total number of off-target sites. Off-target sites with at least seven bases of consecutive complementary pairing with their corresponding gRNA were considered to be off-target sites with higher confidence. (G–L) Heatmap of base pairing between the top 50 gRNA-dependent off-target sites and their corresponding gRNAs. Nucleotides complementarily paired with the gRNAs of dPspCas13b are highlighted in red, and nucleotides complementarily paired with the gRNAs of dRfxCas13d are highlighted in green.

We also analyzed base pairing for randomly selected sequences from transcripts and found that the number of consecutive complementary pairs between these random sequences and the gRNAs was typically <10, with over 90% being fewer than 7 (Fig. 6A–F). Based on this, we considered off-target sites with at least seven consecutive complementary pairs with gRNAs as high-confidence sites. Using this threshold, the high-confidence off-target sites for the three gRNAs of dPspCas13b were 47, 462, and 77, respectively, while for dRfxCas13d, they were 563, 648, and 1150. There was no overlap among the off-target sites of the three dPspCas13b gRNAs, and only minimal overlap among the off-target sites of the three dRfxCas13d gRNAs, indicating high specificity in the identified off-target sites under the current filtering criteria (Supplementary Fig. S10). These results suggest that dPspCas13b and dRfxCas13d have different requirements for the number of consecutive complementary base pairs for gRNA targeting, with dRfxCas13d requiring fewer base pairs compared to dPspCas13b.

In RNA-guided targeting systems, regulatory RNAs such as microRNAs, siRNAs, and crRNAs/gRNAs typically recognize and bind target RNAs through their seed regions [51]. To identify the gRNA seed regions of Cas13 variants, we further analyzed the base-pairing positions between the gRNAs and the top 50 high-confidence gRNA-dependent off-target sites. The results revealed that the consecutive base-pairing positions for dPspCas13b gRNAs were significantly enriched at the 5′ end (DR-distal region) or in the middle region of the gRNAs (Fig. 6G–L). In contrast, for dRfxCas13d, consecutive base-pairing positions were enriched at the 3′ end (DR-distal region) or in the middle region of the gRNAs. In vitro gel-shift assays further confirmed that 10 consecutive complementary base pairs within the first 20 nucleotides of the PspCas13b gRNA spacer, and 7 consecutive complementary base pairs within nucleotides 11–30 of the RfxCas13d gRNA spacer, are sufficient to mediate binding to target RNAs (Fig. 7A and C and Supplementary Fig. S11A). These findings indicate that the DR-distal and middle regions of the gRNAs are critical in determining the binding specificity of dPspCas13b and dRfxCas13d. In vitro cleavage assays showed that the perfectly matched target sites can be cleaved by the Cas13-gRNA complex, while these gRNA-dependent off-target sites are not cleaved (Fig. 7B and D and Supplementary Fig. S11B and C). This behavior is analogous to the identified gRNA-independent off-target sites that cannot be cleaved by Cas13 nuclease (Supplementary Fig. S8), indicating that substrate binding activity can be uncoupled from the cleavage activity of Cas13.

Figure 7.

Binding and cleavage specificity of Cas13-EZH2-g2 complexes. (A) EMSA analysis of 50-nt ssRNA oligonucleotides corresponding to the on-target site of PspCas13b EZH2-g2 (30 consecutive complementary base pairs), an off-target site containing 10 consecutive complementary base pairs, and a negative control lacking complementarity. (B) Denaturing gel analysis showing efficient cleavage of the on-target RNA substrate by the PspCas13b-EZH2-g2 complex, whereas no detectable cleavage is observed for the off-target or negative-control substrates. (C) EMSA analysis of 50-nt ssRNA substrates corresponding to the on-target site of RfxCas13d EZH2-g2 (30 consecutive complementary base pairs), an off-target site containing seven consecutive complementary base pairs, and a negative control lacking complementarity. (D) Denaturing gel analysis demonstrating specific cleavage of the on-target RNA by the RfxCas13d-EZH2-g2 complex, with no cleavage detected for the off-target or negative-control substrate.

Off-target binding of Cas13 variants could perturb endogenous protein expression

To investigate whether off-target binding of the Cas13-gRNA complexes affects mRNA expression, we performed transcriptome-wide mRNA sequencing to identify differentially expressed genes caused by either the dPspCas13b or dRfxCas13d protein (gRNA-independent effects) or a given gRNA (gRNA-dependent effects) (Supplementary Fig. S12A). We observed only minimal changes in mRNA expression in cells expressing dPspCas13b or dRfxCas13d protein alone (Supplementary Fig. S12B and F). Further transfection of a specific targeting gRNA in these cells did not significantly alter endogenous transcripts when compared to a non-targeting control gRNA (Supplementary Fig. S12C–E and G–I). Additionally, the few differentially expressed genes were not identified as uSpyCLIP off-targets, suggesting that off-target binding of the dPspCas13b-gRNA and dRfxCas13d-gRNA complexes has minimal effect on mRNA expression.

To explore whether off-target binding of the Cas13-gRNA complexes affects the abundance of endogenous proteins, we randomly selected four gRNA-independent off-target sites bound by dPspCas13b or dRfxCas13d for western blot analysis. Compared to controls, protein abundance of the two off-targets, RHOA and CTSH, was significantly downregulated and upregulated, respectively, in HEK293T cells expressing dPspCas13b (Supplementary Fig. S13A and B), whereas protein abundance of the off-targets ACVR2B and NUCKS1 was significantly downregulated and upregulated, respectively, in HEK293T cells expressing dRfxCas13d (Supplementary Fig. S13E and F). Next, we randomly selected three off-target sites with start codons covered by EZH2-g2 gRNA sequences from the gRNA-dependent off-target sites of the dPspCas13b-gRNA and dRfxCas13d-gRNA complexes for western blot analysis (Supplementary Fig. S13I and N). Comparative analysis under non-targeting and targeting conditions showed that EZH2-g2 gRNAs associated with dPspCas13b or dRfxCas13d did not alter EZH2 protein expression (Supplementary Fig. S13J and O). However, EZH2-g2 from dPspCas13b significantly downregulated protein abundance of EFNA3 (Supplementary Fig. S13K), while EZH2-g2 from dPspCas13b and dRfxCas13d weakly downregulated protein levels of HPSE and TRAF3, respectively (Supplementary Fig. 13L and P). Together, these results indicate that off-target binding of the dPspCas13b-gRNA and dRfxCas13d-gRNA complexes can perturb the expression of certain endogenous proteins without significantly affecting their mRNA levels.

We further analyzed differentially expressed genes resulting from catalytically active Cas13 protein expression alone (gRNA-independent effects) or from the presence of a specific gRNA (gRNA-dependent effects). Both PspCas13b and RfxCas13d induced substantially more differentially expressed genes than their catalytically inactive counterparts (Supplementary Fig. S12J–Q). Notably, only a small subset of these genes was identified by uSpyCLIP, indicating that most Cas13 off-target binding events do not lead to RNA degradation. To further evaluate gRNA-dependent off-target effects of catalytically active Cas13 in the cells, we selected three essential genes harboring off-target motifs for PspCas13b EZH2-g2 and three essential genes harboring off-target motifs for RfxCas13d EZH2-g2. Each group included two metabolic genes and one ribosomal gene. RT-qPCR and western blot analysis confirmed that the expression of EZH2 was significantly suppressed by the Cas13-EZH2-g2 complex (Fig. 8A and E). Modest changes were detected in the mRNA levels of ACSL1 and FASN bound by the PspCas13b-EZH2-g2 complex, and EIF2D bound by the RfxCas13d-EZH2-g2 complex (Fig. 8B, C, and F). At the protein level, ACSL1 and EIF2D expression was reduced by the Cas13-EZH2-g2 complexes, with the decrease in ACSL1 protein being more pronounced than its corresponding mRNA change (Fig. 8B and F). In addition, we assessed the impact of gRNA-dependent off-target binding on cellular proliferation and apoptosis. No significant effects on apoptosis were detected; however, marked alterations in cell proliferation were observed, particularly for EZH2-g2 and TP53-g1, likely reflecting their higher number of off-target binding sites (Fig. 8I and J). These results indicate that specific off-target binding events of catalytically active PspCas13b and RfxCas13d can perturb gene expression at the mRNA and/or protein level, and may disrupt normal cellular functions.

Figure 8.

gRNA-dependent off-target binding alters the expression of essential genes and induces substantial alterations in cell proliferation. (A–D) RT-qPCR and western blot analyses of on-target EZH2 and three gRNA-dependent off-target genes bound by the PspCas13b-EZH2-g2 complex, with comparisons between non-targeting and EZH2-g2 targeting conditions. (E–H) RT-qPCR and western blot analyses of on-target EZH2 and three gRNA-dependent off-target genes bound by the RfxCas13d-EZH2-g2 complex, with comparisons between non-targeting and EZH2-g2 targeting conditions. RT-qPCR data are presented as mean ± SD (n = 3). Relative changes in protein abundance were quantified by densitometric analysis and are indicated as red numbers at the bottom of each panel. (I) Proliferation of cells expressing PspCas13b (left panel) or RfxCas13d (right panel) with either targeting gRNAs or a non-targeting control gRNA (gNT), measured by CCK-8 assay. Proliferation rates were measured by absorbance at days 1–5 and normalized to that at day 1. Data are presented as mean ± SD (n = 4). *P < 0.05; **P < 0.01; ***P < 0.001. ns, not significant. (J) Apoptosis analysis of cells expressing PspCas13b (left panel) or RfxCas13d (right panel) with either targeting gRNAs or non-targeting control gRNA (gNT). Data are presented as mean ± SD (n = 3). ns, not significant.

Discussion

In this study, we introduce the uSpyCLIP method, which exhibits both high sensitivity and exceptional specificity for identifying RBP binding sites. The uSpyCLIP protocol has been optimized to minimize material loss and hands-on time, significantly enhancing its sensitivity compared to existing protocols. This high sensitivity allows for the detection of more binding sites in bulk cells at the same sequencing depth and enables the profiling of RBP binding sites in as few as 1000 cells (Supplementary Fig. S3B and C). Furthermore, uSpyCLIP is fully bead-based until the final size-selection step, facilitating automation and reducing labor. The entire process can be completed in just 2 days, making it one of the most time-efficient CLIP-related techniques available. Additionally, the RBP fusion tag in uSpyCLIP has been minimized to reduce interference with RBP activity.

Using uSpyCLIP, we mapped the binding sites of the dPspCas13b-gRNA and dRfxCas13d-gRNA complexes across the transcriptome. Our findings reveal that dCas13 binding is more promiscuous than previously anticipated. Both complexes exhibit considerable off-target binding events, encompassing both gRNA-dependent and gRNA-independent sites. Notably, some of these events alter the abundance of endogenous protein without inducing significant changes in mRNA level. Accordingly, approaches such as quantitative mass spectrometry may be necessary to characterize the protein-level off-target effects of Cas13. Collectively, these findings indicate that transcriptome analyses relying on RNA-seq may considerably underestimate the off-target effects of dPspCas13b-gRNA and dRfxCas13d-gRNA complexes. Consequently, the applications based on dCas13 or dCas13-effector fusions, such as post-transcriptional regulation, RNA imaging, and RNA editing, may be complicated by substantial off-target binding.

Previous studies have shown that PspCas13b gRNA contains a critical region for RNA cleavage spanning nucleotides 2–18 (for a 30-nucleotide spacer), which is highly sensitive to double mismatches [6]. Additionally, the spacer region between nucleotides 15 and 21 (for a 27-nucleotide spacer) has been identified as a “seed region” of gRNA critical for the efficacy of RfxCas13d knockdown [32]. Our results corroborate these findings, indicating that the DR-proximal region of the gRNA spacer is less important for binding specificity. However, our data demonstrate that a perfect match of ten bases in the first 20 nucleotides of the PspCas13b gRNA spacer appears sufficient to mediate binding, while a nearly perfect match of seven bases in nucleotides 11–30 of the RfxCas13d gRNA spacer is sufficient for RfxCas13d binding. These observations lead us to propose a model of target recognition, where the activator RNA is first primed by the DR-distal or middle region of the spacer, followed by binding to the DR-proximal region.

A recent study demonstrated the intrinsic targeting of host RNA by these Cas13 proteins in mammalian cells using RNA immunoprecipitation coupled with high-throughput sequencing (RIP-seq) to identify Cas13-associated RNA targets [50]. However, the exact binding sites were not determined. In this study, we show that both dPspCas13b and dRfxCas13d bind to thousands of gRNA-independent off-target sites by using uSpyCLIP, which provides precise, location-specific information about the RNA targets bound by dCas13, enabling a more accurate characterization of binding events. Our results reveal novel RNA signatures associated with dPspCas13b and dRfxCas13d, shedding light on their distinct off-target binding mechanisms. Beyond PspCas13b and RfxCas13d, recent discoveries of Cas13X/Y, Cas13bt, Cas7-Cas11, and Cas13j have expanded the RNA-targeting toolbox [52–55] and we anticipate that uSpyCLIP will be valuable in systematically studying the off-target effects of these other Cas systems, which should provide valuable insights for the rational design of specific gRNAs and the engineering of Cas proteins for more accurate RNA-targeting applications.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by the National Key R&D Program of China [2021YFA1100201 and 2022YFA1303301 to L.W.]; the Strategic Priority Research Program of the Chinese Academy of Sciences [XDB0570000 to L.W.]; the National Natural Science Foundation of China [82400181 to H.F.; 82270160 to L.Z.]; the China Postdoctoral Science Foundation [2024M751998 to H.F.]; and the Key projects from the Ministry of Science and Technology [2023YFC2706401 to L.Z.]. Funding to pay the Open Access publication charges for this article was provided by the National Key R&D Program of China [2021YFA1100201 and 2022YFA1303301].

Data availability

Raw sequencing data of uSpyCLIP libraries and RNA-seq have been deposited in the Gene Expression Omnibus under accession numbers GSE291436 and GSE291437, respectively. Previously published SpyCLIP, iCLIP, and eCLIP data that were reanalyzed here are available under accession codes GSE114720, PRJEB15159, and GSE77629, respectively.