Engineered minimal type I CRISPR-Cas system for transcriptional activation and base editing in human cells

Guo, Jing; Gong, Luyao; Yu, Haiying; Li, Ming; An, Qiaohui; Liu, Zhenquan; Fan, Shuru; Yang, Changjialian; Zhao, Dahe; Han, Jing; Xiang, Hua

doi:10.1038/s41467-024-51695-x

Download PDF

Article
Open access
Published: 23 August 2024

Engineered minimal type I CRISPR-Cas system for transcriptional activation and base editing in human cells

Nature Communications volume 15, Article number: 7277 (2024) Cite this article

2222 Accesses
1 Citations
11 Altmetric
Metrics details

Subjects

Abstract

Type I CRISPR-Cas systems are widespread and have exhibited high versatility and efficiency in genome editing and gene regulation in prokaryotes. However, due to the multi-subunit composition and large size, their application in eukaryotes has not been thoroughly investigated. Here, we demonstrate that the type I-F2 Cascade, the most compact among type I systems, with a total gene size smaller than that of SpCas9, can be developed for transcriptional activation in human cells. The efficiency of the engineered I-F2 tool can match or surpass that of dCas9. Additionally, we create a base editor using the I-F2 Cascade, which induces a considerably wide editing window (~30 nt) with a bimodal distribution. It can expand targetable sites, which is useful for disrupting functional sequences and genetic screening. This research underscores the application of compact type I systems in eukaryotes, particularly in the development of a base editor with a wide editing window.

HMGN1 enhances CRISPR-directed dual-function A-to-G and C-to-G base editing

Article Open access 27 April 2023

Repurposing type I–F CRISPR–Cas system as a transcriptional activation tool in human cells

Article Open access 19 June 2020

Harnessing noncanonical crRNA for highly efficient genome editing

Article Open access 07 May 2024

Introduction

CRISPR-Cas systems are adaptive immune systems that are widely distributed in prokaryotes. They can be classified into two classes, six types, and more than 30 subtypes¹. Class 1 encompasses types I, III, and IV, with effector modules composed of multiple proteins. Class 2 comprises types II, V, and VI, characterized by a single, multi-domain effector protein. The CRISPR immunity mechanism typically involves three processes: adaptation, crRNA biogenesis, and interference².

The type I system is the most prevalent type of CRISPR-Cas and is divided into seven subtypes, I-A to I-G¹. The crRNA-binding complex of type I is called Cascade (CRISPR-associated complex for antiviral defense), which binds to the crRNA for target recognition, facilitates duplex formation between the crRNA and its complementary target DNA for R-loop formation, and recruits Cas3 nuclease to degrade DNA processively³. Type I systems have been developed into various microbial genome manipulation tools, such as tools for genome editing with homologous repair templates⁴, transcriptional regulation⁵, large fragment deletion⁶, large fragment integration without requiring homologous recombination⁷, and simultaneous genome editing and gene regulation with Cascade-Cas3⁸. Since 2019, several type I subtypes have been successfully used in eukaryotic genome manipulation. When Cascade and Cas3 were expressed simultaneously, Cas3 exhibited exonuclease activity, resulting in large fragment deletions of up to 200 kb^{9,10,11,12,13,14,15}. Similar activity was reported for type I-D, of which Cas10 is a functional nuclease, while short indels were detected as well^16,17. Fusing Cas3 with the cytidine deaminase enables wide-ranging random mutagenesis, covering up to 55 kb, which provides an efficient tool for optimizing complex biosynthetic pathways¹⁸. Moreover, Cascade can be fused with different domains to perform different functions. When Cascade was fused to the dimerization-dependent, non-specific FokI nuclease domain, the editing efficiency was comparable to that of dCas9-FokI⁹. When Cascade was fused to the transcriptional activation or repression domain, the expression levels of targeted endogenous genes could be modulated^19,20,21.

However, the Cascades used in eukaryotes comprise four or five subunits, counting in Cas6, which processes the precursor crRNA (pre-crRNA) into mature crRNA. For example, the Cascade of Escherichia coli (E. coli) K12 type I-E system is composed of five subunits, with a total gene size of ~4.4 kb^11,21. The Cascade of Neisseria lactamica ATCC 23970 type I-C system is composed of four subunits, with a total gene size of ~3.8 kb, including the hidden small subunit¹³. The large size hinders applications with cargo size constraints, such as the widely applied adeno-associated virus (AAV) vector with a ~4.7 kb packing limit, which hampers most CRISPR tools’ clinical applications²². Thus, mining more compact type I systems will be helpful for applying type I genome manipulation tools to eukaryotes.

On the other hand, R-loop formation is an important feature of CRISPR systems. In the R-loop structure, the non-target DNA strand displaced by the spacer of crRNA is exposed and accessible to other molecules, which is exploited by base editing technologies^23,24,25. By fusing single-stranded DNA (ssDNA) deaminases or glycosylases to catalytically impaired Cas nuclease, base editors can install various base conversions without the need for double-stranded DNA (dsDNA) breaks or donor DNA templates^25,26. However, base editors are developed mainly with catalytically impaired Cas9 or Cas12 (including their presumed compact ancestors)^27,28. It is noteworthy that the R-loop formed by type I systems is considerably wider, spanning around 30–40 nt^3,12,23, compared to that in Cas9 or Cas12, which has a narrower R-loop of approximately 20 nt^23,29. This extended R-loop in type I systems may provide more nucleotide substrates accessible to ssDNA deaminases or glycosylases, thus expanding targetable sites. Therefore, exploring the potential of type I systems in base editing could be valuable for developing base editors with wide editing windows.

In this work, we study compact type I-F2 systems for applications in human cells. Type I-F2 systems feature the most compact Cascade identified to date, consisting of Cas5, Cas6, and Cas7 without the large subunit (Cas8) or small subunit (Cas11)^30,31,32. The total gene size is significantly smaller than that of SpCas9³³, making it a promising candidate for easier delivery in eukaryotes. However, previous attempts to apply this system to eukaryotes were unsuccessful. For instance, the Cas3 and Cas5–7 derived from the Shewanella putrefaciens CN32 (Spu) I-F2 did not exhibit DNA cleavage activity in human cells¹¹. Similarly, another study reported that the Cas5–7 derived from Spu I-F2 could not activate the expression of the target gene, after fusing to a transcriptional activator¹⁹. Here, we present a compact type I-F2 system for manipulating the human genome. The Cascade is derived from the Moraxella osloensis CCUG 350 (Mos350), with a total gene size of approximately 2.7 kb, and prefers a simple 5′-CC protospacer adjacent motif (PAM). By fusing the transcriptional activation domain with the Cascade, we can modulate gene expression in human cells and achieve robust activity through crRNA and Cascade engineering. In addition, by fusing deoxyadenosine deaminase with the Cascade, we develop a base editor with the type I-F2 system. Interestingly, this adenine base editor (ABE) induces a wide editing window (~30 nt) with a bimodal distribution and can achieve an editing efficiency exceeding 50%. The I-F2 ABE with a wide editing window can be useful for genetic screening and disrupting the functional sequences. These results highlight the potential of compact type I-F2 systems in eukaryotes and expand the base editing toolbox.

Results

Identification of miniaturized type I-F2 CRISPR-Cas systems

Type I-F2 systems consist of five Cas proteins: Cas1, Cas3, Cas5, Cas6, and Cas7, without the large and small subunits (Fig. 1a)^30,31,32. Cas1, Cas3, and Cas6 of type I-F2 have sequence similarities to those of type I-F1 or I-F3 systems, while Cas5 and Cas7 of type I-F2 have no significant sequence homology to any other type I subtypes^31,32. Previously, studies of type I-F2 focused on the Spu I-F2 system, while this system is proven disabled in gene activation¹⁹ or genome editing in human cells¹¹. To develop genetic tools with compact type I systems, we searched public databases for miniaturized type I-F2 systems and identified 93 such systems (Supplementary Fig. 1). We analyzed all 93 I-F2 systems and found no other conserved proteins nearby. Recently, Cas11 was found within the large subunit Cas8/Cas10d^13,34. Since the I-F2 system lacks a large subunit, we analyzed Cas5/Cas6/Cas7 and found no conserved internal translation start site, indicating internal translation initiation regions are not conserved in I-F2 systems. In addition, the structural analysis results of the I-F2 system showed that Cas5 and Cas7 replace the functions of the large and small subunits, respectively³⁰. These results support that the I-F2 system encodes neither the small subunit (independently or within Cascade elements) nor the large subunit.

**Fig. 1: The minimal type I-F2 CRISPR-Cas system.**

Phylogenetic analysis of Cas7 showed that Cas7 of types I-F1 and I-F3 clustered together, whereas all 93 systems of type I-F2 clustered in a separate branch (Fig. 1b), confirming strong sequence deviations in the Cas7 homologs. The gene sizes of these I-F2 Cascades are approximately 2.4–2.8 kb. The repeats are 28 bp long, and the sequences are highly conserved according to multiple sequence alignments (Supplementary Fig. 2a). The spacer of 32 bp accounted for 96.6% of all the spacers (Supplementary Fig. 2b), and the wild-type spacers used in subsequent experiments are of 32 bp. Interestingly, the average number of spacers is 55, ranging from 3 to 180, indicating that I-F2 systems are highly active.

Targeted gene repression by I-F2 Cascades in E. coli

Since the Cascade-crRNA complex is responsible for target recognition and R-loop formation, it can be a useful component for gene regulation or base editing tools. We firstly tested the targetable DNA-binding ability of type I-F2 Cascades in E. coli. We randomly selected eight candidate type I-F2 systems from different strains (Supplementary Table 1) to analyze the PAM preference by aligning spacer sequences to the virus database and seven type I-F2 systems had corresponding hits in viruses. Except for one system with only one hit in the database, conserved CC sequences were found upstream of most hits for the other six systems (Supplementary Fig. 3). In addition, many related I-F1¹⁹ and I-F3³⁵ systems have a 5′-CC PAM preference on the non-target strand, and the Spu I-F2 system also has a 5′-CC PAM preference³¹. Thus, we synthesized the eight type I-F2 Cascades and designed the 5′-NCC PAM at the 5′ end of the protospacer (in the promoter region of YFP) (Fig. 2a). After inducing the expression of Cascade and crRNA, significant decreases in fluorescence intensity were observed for all eight I-F2 systems, by recognizing 5′-CC PAM, except for the system derived from Acinetobacter sp. WCHAc010052 (AspWC), which prefers 5′-DCC PAM (Fig. 2b). These results indicate that all eight I-F2 Cascades have programmable DNA-binding abilities in E. coli.

Fig. 2: Targeted gene repression by I-F2 Cascades from eight different microbes in *E. coli.*

Expression of mature crRNA and I-F2 Cascade subunits in human cells

To repurpose miniaturized I-F2 Cascades for application in human cells (HEK293T cells), we used the cytomegalovirus (CMV) promoter to express the three subunits of Cascade (human codon-optimized cas5/cas6/cas7), which were linked by self-cleaving 2A peptides. Nuclear localization signals (NLS) derived from the simian virus 40 (SV40) large T antigen were fused to each N-terminus of the three subunits. The RNA polymerase III U6 promoter (hU6) was used to transcribe the pre-crRNA, which could be processed by Cas6 into a mature crRNA (Fig. 3a).

**Fig. 3: Expression of crRNA and Cascade subunits of type I-F2 systems in human cells.**

We synthesized 11 type I-F2 systems derived from different strains (Supplementary Table 1). Northern blot assays were performed to determine whether these crRNAs could be processed and stabilized in human cells. The result showed that only five systems generated mature crRNAs (Fig. 3b). The mature crRNAs of the Spu I-F2 system were not detected. These five systems were used for further studies. Then, we performed small RNA sequencing to determine the mature crRNA transcripts. One of the five systems was randomly selected (Mos350 I-F2 system). The result showed that the pre-crRNA was processed into mature crRNAs of 60 nt, which were composed of a full-length spacer flanked by a repeat-derived 5′ handle of 8 nt and a repeat-derived 3′ handle of 20 nt (Supplementary Fig. 4), consistent with the result of the Spu I-F2 crRNA extracted from Spu strain³¹.

Next, we examined the expression and nuclear localization of Cascade subunits in human cells using the randomly selected Mos350 system. The plasmid expressing the Cascade components and the plasmid expressing crRNAs were co-transfected into HEK293T cells. Western blot results showed that all three Cas proteins were expressed (Fig. 3c). Co-immunoprecipitation by pull-down of the Flag epitope on Cas7 demonstrated the interaction between Cas7 and Cas5/Cas6. Moreover, in the absence of the crRNA, the binding affinity between Cas6 and Cas7 was significantly reduced, suggesting that Cas6 could bind to Cascade through interaction with the crRNA (Fig. 3c). In contrast, the binding affinity between Cas5 and Cas7 did not change regardless of the presence or absence of the crRNA. In addition, immunofluorescence analysis confirmed that each of the three subunits could enter the nucleus via the SV40 NLS (Fig. 3d).

Programmable transcriptional activation by Mos350 Cascade-VPR

The I-F2 Cascade consists of three Cas subunits assembled at various stoichiometries (inferred to be Cas5₁Cas7₆Cas6₁) when bound to the wild-type crRNA³⁰, providing multiple options for synthetic fusion with regulatory domains. Since Cas7 has multiple copies in the Cascade complex, we firstly repurposed the compact I-F2 Cascade for the transcriptional activation of endogenous genes in human cells by fusing Cas7 with the VP64-p65-Rta (VPR) transcriptional activation domains^19,36. We targeted a site within the promoter region of the IL1B gene. Among the five I-F2 systems capable of producing mature crRNAs in human cells, only Mos350 Cascade-VPR significantly activated the expression of the IL1B gene, as determined by quantitative real-time PCR (qPCR) (Fig. 4a). Thus, the Mos350 I-F2 system was studied further.

**Fig. 4: Programmable transcription activation by Mos350 Cascade-VPR.**

To improve the efficiency of Mos350 system, PAM depletion assay was performed to determine the most effective PAM. A plasmid library with four randomized nucleotides at the 5′ end of the protospacer was constructed. The results revealed that 5′-CC PAM is just the best choice (Supplementary Fig. 5).

To test the activation effects of VPRs tethering to different subunits, we expressed the Mos350 Cascade with VPR fused to the C-terminus of Cas5, Cas6, or Cas7 (Fig. 4b). As expected, qPCR results showed that only the VPR fused to Cas7 significantly activated the expression of IL1B (Fig. 4b). The following experiments were performed using the Cas7-VPR fusion strategy. Furthermore, we tried to optimize the linker between Cas7 and VPR and attempted to fuse VPR to the N-terminus of Cas7. None of these attempts significantly enhanced the transcriptional activity (Supplementary Fig. 6). Therefore, we retained the original design of Mos350 Cascade-VPR in subsequent studies.

To determine the programmable endogenous gene activation of the Cascade-VPR complex at different loci in the human genome, a series of crRNAs were designed to target different promoter regions of three genes (ACTC1, ASCL1, and NEUROD1). Target sites were selected for every 100 bp upstream of the transcriptional start site (TSS). Mos350 Cascade-VPR activated the expression of all three endogenous genes to varying degrees (Fig. 4c). The highest activities were mostly achieved when targeting the region within 200 bp upstream of the TSS (Fig. 4d), which is consistent with other Cascade-VPR vehicles^19,21. Next, we designed crRNAs targeting the region within 200 bp upstream of the TSS in two other genes (HBB and HBG1) and observed gene activation as well (Fig. 4e).

Stronger transcription by crRNA and protein engineering

To investigate the potential of the Mos350 Cascade-VPR as a more efficient transcriptional activation tool, we optimized both the crRNA and the protein components. With each 6 nt increment in the spacer length of the crRNA, an additional Cas7 subunit can be incorporated into the Cascade complex^30,37,38 (Fig. 5a), which may increase the copies of VPR and lead to stronger activation. We explored the impact of spacer length on Mos350 Cascade-VPR activity at the IL1B and ASCL1 loci, with spacer lengths varying by multiples of six. As the spacer length increased, the level of transcriptional activation gradually increased at both loci. The maximum transcriptional activation effects were achieved when the spacer length reached 50 nt, resulting in increased activity of up to 5.8-fold and 8.6-fold, compared to that using the wild-type spacer of 32 nt, at the ASCL1 and IL1B loci, respectively (Fig. 5b). We also conducted a Northern blot experiment for samples with varied spacer length (targeting ASCL1 locus) (Supplementary Fig. 7). When the spacer length was shortened to 26 nt, the mature crRNA band was significantly weaker. Extending the spacer length to 56 nt resulted in the disappearance of both the pre-crRNA and mature crRNA bands. These results indicate that, for the I-F2 system, changes in spacer length beyond a certain range may be detrimental to the stability of crRNA, consistent with transcriptional activation results targeting the ASCL1 locus (Fig. 5b).

**Fig. 5: Enhancing transcription activation through crRNA engineering and protein engineering.**

Subsequently, we sought to increase the activity by introducing specific amino acid mutations based on protein structure information. Compared to the Spu Cascade subunits, the Mos350 Cascade subunits have similar three-dimensional structures predicted by AlphaFold (Supplementary Fig. 8). In the DNA-bound structure of the Spu Cascade complex, the aromatic residues emanating from the Cas7 thumb (Y149, F160, and F161) form a stacking force that stabilizes the interaction between the crRNA and the target protospacer³⁰. The results of sequence alignment and structure prediction showed that the Cas7 of Mos350 I-F2 possesses only one aromatic residue at the corresponding positions (L154, L175, and F176) (Fig. 5c and Supplementary Fig. 9). To increase the DNA-binding activity of the Mos350 Cascade-VPR, we introduced L175F into the Mos350 Cas7 protein, named as L175F-Cascade-VPR. As expected, when targeting the ASCL1 and IL1B loci, L175F-Cascade-VPR exhibited improved activities of up to 1.7-fold and 4.7-fold, respectively, compared to the wild-type Mos350 Cascade-VPR. However, adding the L154Y mutation to L175F-Cascade-VPR did not further enhance transcriptional activation (Fig. 5d).

To directly test whether L175F mutation increases transcriptional activation by enhancing binding affinity between crRNA and the target DNA, we conducted electrophoretic mobility shift assays (EMSA). The results showed that the L175F Cascade-crRNA complex exhibited an increased binding affinity to the target DNA, compared to the wild-type complex (Fig. 5e and Supplementary Fig. 10), providing direct evidence for our hypothesis.

We combined the two strategies for further improvement, including extending the spacer length to 50 nt and introducing the L175F mutation in Cas7. The activity of the L175F-Cascade-VPR improved to 16.7-fold and 5.3-fold at the ASCL1 and IL1B loci, respectively, after extending the spacer length to 50 nt (Fig. 5f). We also compared the activation activity with that of the dCas9-VPR (SpCas9). The engineered Mos350 Cascade-VPR and the dCas9-VPR efficiently activated gene expression, with the engineered Mos350 performing even better at the ASCL1 locus (Fig. 5f).

Given the critical importance of binding specificity for transcriptional activators, we conducted a Cascade-dependent DNA off-target analysis of the L175F-Cascade-VPR. Previous research on several type I systems has demonstrated that mismatches in the PAM-proximal region have greater impacts on the activity of CRISPR systems than those in the PAM-distal region, especially those in the region downstream of the position 24^9,11,17,19, counting the nucleotide adjacent to the PAM as position 1 of the protospacer. In addition, structural features contribute to the mismatch tolerance at every sixth position of the guide sequence, which flips out of the RNA-DNA heteroduplex structure³⁰. Therefore, we searched potential off-target sites with ≤4 mismatches to the crRNA except for mismatches in positions 6, 12, 18, 24−32, located on the promoter regions (≤2 kb upstream of the TSS). The target site was selected at the ASCL1 locus, which had overlapping target regions of the Mos350 L175F-Cascade-VPR and dCas9-VPR. Its predicted three off-target sites were measured (Supplementary Fig. 11). For the Mos350 L175F-Cascade-VPR, off-target activation activity was not observed at any of the three potential off-target sites. Importantly, no off-target activation occurred when the spacer length was extended to 50 nt either. However, the dCas9-VPR exhibited activation activity at an off-target site in the PRPH promoter region (Fig. 5g). These results indicate that the engineered Mos350 L175F-Cascade-VPR possesses robust target gene activation activity with high specificity.

I-F2 Cascade-ABE with a wide bimodal editing window

Since the R-loop formed by Cascade is considerably wider than that by Cas9 or Cas12, we investigated the potential of repurposing the Mos350 Cascade as a base editor. As TadA-8e facilitates rapid and efficient deoxyadenosine deamination and is compatible with various Cas proteins³⁹, we created four ABEs named 5CABE, 5NABE, 7CABE, and 7NABE by attaching TadA-8e to either the C- or N-terminus of Cas5 or Cas7 in the Mos350 Cascade (L175F) (Fig. 6a). The base editing window was evaluated by targeting a region within the NIBAN1 gene, featuring an alternating 5′-A-N-A-N-3′ sequence⁴⁰. The region could be targeted by either of two crRNAs, resulting in adenines positioned nearly at every even (site 1) or odd (site 2) position, collectively covering positions 4–27 and 30–32 within the protospacer (Fig. 6b). Remarkably, all four ABEs exhibited A⋅T-to-G⋅C base editing efficiency, with wide bimodal base editing windows spanning nearly the entire protospacer region (~30 nt). When TadA-8e was fused to different termini of Cas proteins, the peak in the PAM-proximal end shifted from position 6 to 13. In contrast, the peak in the PAM-distal end (at position 26 or 27) and the valley (at position 16 or 17) remained consistent. The highest base editing efficiency reaching 30.5%, was achieved at position 27 using 5NABE, in which TadA-8e was fused to the N-terminus of Cas5 (Fig. 6b and Supplementary Fig. 12). The 5NABE was investigated in subsequent experiments.

**Fig. 6: Cascade-ABE8e has a wide editing window with a bimodal distribution.**

We tested the base editing efficiency of 5NABE by targeting different endogenous loci in the genome. We chose four targets (sites 3–6) with multiple adenines and five targets (sites 7–11) with only a single adenine. For targets with multiple adenines, there were varying degrees of A⋅T-to-G⋅C conversion in target positions, with the highest editing efficiency ranging from 7.4% to 22.6% (Fig. 6c, d). For targets with a single adenine, significant A⋅T-to-G⋅C conversion was observed with the highest editing efficiencies ranging from 7.4% to 24.3% (Fig. 6d).

We further investigated the applications at disease-related targets. Upregulating fetal hemoglobin expression is a promising therapeutic option for β-thalassemia and sickle-cell disease^41,42. BCL11A is a transcriptional repressor of fetal hemoglobin. It has been demonstrated that mutations in the core sequence of +58 BCL11A erythroid enhancer, especially at the GATA1 binding site, resulted in reduced BCL11A expression and concomitant induction of fetal hemoglobin^43,44,45. We designed two crRNAs (targeting sites 12 and 13) to install A⋅T-to-G⋅C conversion in the core sequence of +58 BCL11A erythroid enhancer, with the adenines of the GATA1 binding site located around the two peaks (position 9 or position 27) of the editing window, respectively. The three adenines in GATA1 were edited significantly by the two crRNAs, with a maximum editing efficiency of 46.2% at position 7 of site 12 (Fig. 6e and Supplementary Figs. 13 and 14). Besides, multi-adenines outside the GATA1 binding site, which are also located in the core sequence, were efficiently edited, with a maximum efficiency of 49.6% at position 27 of site 12.

Next, we assessed the off-target activity of 5NABE. Potential Cascade-dependent off-target sites were selected with ≤3 mismatches at all positions except mismatches in positions 6, 12, 18, 24–32 of the protospacer. Two targets (sites 5 and 10) with high editing efficiency were selected, as well as seven corresponding potential off-target sites (randomly selected). Most predicted off-target positions exhibited low off-target activity (Fig. 7). The highest off-target base editing efficiency reaching 4.2%, was achieved at OT2 for site 5. The only two mismatches of OT2 were in positions 21 and 22 in the PAM-distal end, which might be the reason for its high editing efficiency (Fig. 7).

**Fig. 7: Off-target effects of 5NABE.**

On the other hand, we analyzed at least 30 nt upstream and downstream of each target site and observed several adenines outside the protospacer were slightly edited (A⋅T-to-G⋅C) (Supplementary Fig. 15). In addition, several adenines on the target strand were edited outside the protospacer, whereas none of the adenines were edited on the target strand within the protospacer region.

Interestingly, when we raised the fluorescence threshold for EGFP-positive cells in fluorescence-activated cell sorting (FACS) (Supplementary Fig. 16), we found that the editing efficiency was significantly improved. Since EGFP was linked by the self-cleaving 2A peptide to the I-F2 ABE system, the results suggested that the editing efficiency could be improved by increasing the gene expression level of I-F2 ABE system. Thus, we selected sites 6–11 to test the editing efficiency after raising the fluorescence threshold from 500 to 2000. When cells with stronger fluorescence signal were collected, the A⋅T-to-G⋅C conversion efficiencies were significantly improved, with the highest editing efficiencies ranging from 36.4% to 52.9%. Notably, the editing efficiencies for three sites (sites 7–9) were close to or above 50% (Fig. 6d). These results revealed that the I-F2 ABE system could be very efficient. We also tested the editing efficiency of the predicted off-target sites for site 10 with the fluorescence threshold of 2000. The editing efficiency of position A8 of off-target site 1 (OT1) was improved from 0.9% to 5.0% (Fig. 7a and Supplementary Fig. 17). These results suggested that the higher editing efficiency might be accompanied with higher off-target activity. If low off-target activity is required, the protein expression level can be reduced or the delivery of protein-RNA complex can be employed^39,46.

Finally, we conducted off-target analysis using whole-genome sequencing (WGS). The results revealed 19 Cas-dependent off-target sites when 5NABE targeted site 10 (Supplementary Fig. 18a). Positions 1–11 of the off-target sequences were highly conserved (excluding position 6), possibly representing the seed region of the I-F2 system. In addition, the I-F2 ABE showed no significant Cas-independent off-target effects (Supplementary Fig. 18b).

Discussion

In this study, we tested 11 type I-F2 Cascades and obtained one system derived from the Mos350 strain, which can be repurposed as a transcriptional activator or base editor in human cells by fusing different domains. The Mos350 Cascade prefers a simple 5′-CC PAM and has a total gene size of ~2.7 kb, which is significantly smaller than SpCas9 (~4.1 kb). The repeat-spacer-repeat unit is 88 nt in length. Multiplexed targeting can be achieved by constructing a CRISPR array containing tandem spacers linked by repeats, which can be processed by Cas6 to target different sites simultaneously. Since the size-minimized AAV backbone for ABE delivery has been developed with small-Cas orthologues like Nme2Cas9 (3.24 kb)⁴⁷, delivering the complete I-F2 Cascade-ABE using a single AAV vector is feasible. Moreover, recent research has found that type I Cascades fused with HNH nuclease domains enable precise dsDNA cleavage, with a total gene size of over 3.3 kb⁴⁸. Compact genome editing tools may be constructed by fusing the type I-F2 Cascade with an HNH domain.

Multi-subunit effector complexes of type I systems provide a broad and flexible platform for integrating different functional domains. First, functional components can be fused to various subunits individually or simultaneously across multiple subunits to improve their effects^9,19,20,21. Second, the Cascade complex is very flexible. Modifying the spacer length can alter the Cas protein stoichiometry within the Cascade complex and influence the extent of the R-loop region, potentially impacting its functionality^{8,19,37,38,49}. For transcriptional activation, significant activation was detected with VPR fused to Cas7 (Fig. 4b), which could be attributed to the multi-copy form of Cas7-VPR in Cascade. Moreover, we engineered spacer lengths ranging from −6 nt to +24 nt, resulting in varying degrees of target gene expression (Fig. 5b). This inherent flexibility of the type I system enables delicate regulation of target genes, which is useful for investigating gene functions and biological mechanisms. However, for base editing, the highest editing efficiency was obtained with TadA-8e fused to Cas5 (Fig. 6b and Supplementary Fig. 12). Besides, in immunofluorescence analysis, we could only detect the signals of all three Cas proteins in partial cells. Thus, higher activation efficiency may be achieved by optimizing the NLS.

Current base editors mainly contain catalytically impaired Cas9 or Cas12 orthologs fused to enzymes that alter the DNA bases, which can help study the effects of mutations within genes and treat genetic diseases^25,28,39,50. The editing window is limited by the R-loop size of Cas9 or Cas12, which is hard to expand over 20 nt^27,51. However, the potential of base editing with Cascade, which induces a large R-loop, has not yet been explored. We constructed ABEs by fusing the I-F2 Cascade with TadA-8e. The editing window with a bimodal distribution covered almost the entire R-loop region, from positions 4 to 32, whereas the Cas9- or Cas12-derived ABE8e typically had a unimodal window of less than 12 nt³⁹. The I-F2 Cascade-ABE system may have more targetable sites and can induce a greater variety of mutation types, rather than efficiently editing the majority of adenines simultaneously. The wide editing window is useful for disruption of functional sequences, like disruption of infection-related genes or inhibitory genes for genetic therapy. The wide editing window is also useful for in situ saturating mutagenesis of important genes, and multiplex base editing. For example, expanding the editing window is an important aspect for crop breeding⁵¹.

On the other hand, among the four ABEs we created, 5NABE is the most efficient design. With 5NABE, the editing efficiency for several targets could be close to or above 50% (Fig. 6d, e). It is also more efficient than most ABEs based on dCas12, such as enAsABE8e derived from dead engineered AsCas12a³⁹, dCasMINI-ABE derived from dead engineered Cas12f⁵⁰, and dCas12Pro-ABE derived from dead Cas12n⁵². Besides, without the nickase activity, the editor may be safer for application⁵³.

Off-target activation efficiency was not observed for Mos350 L175F-Cascade-VPR, even with an extended spacer of 50 nt. In contrast, a significant off-target site was detected for SpCas9 (Fig. 5g). For base editing, with the increase of protein expression, the editing efficiency of target sites improved significantly (Fig. 6d). It should be noted that higher off-target editing efficiency may be induced at the same time (Supplementary Fig. 17). On the other hand, WGS results showed that the PAM-distal mismatches may have less impacts on the activity of CRISPR systems than those in the PAM-proximal region, consistent with previous results of other CRISPR-Cas systems^{9,11,17,19,54}. However, there are multiple factors that determine the editing efficiency of off-target sites, like the number, location, and type of mismatches⁵⁴. Large numbers of gRNAs need to be tested to understand the specificity of the system. In addition, off-target efficiency can be reduced by engineering proteins³⁹ and gRNAs⁵⁵, as well as delivery of protein-RNA complex^39,46.

In summary, we demonstrate that the miniaturized I-F2 Cascade, which is significantly smaller than SpCas9 in gene size, can be developed into programmable tools for use in human cells. By engineering the crRNA and Cascade, the activation efficiency of the Cascade-VPR can match or surpass that of the dCas9-VPR. Moreover, the inherent flexibility of the type I system enables delicate regulation of target genes. We also explored the potential of base editing with type I Cascade and successfully created an ABE with type I-F2 Cascade (5NABE). The 5NABE induces a significantly wide editing window of ~30 nt, which can be useful for genetic screening and disrupting the functional sequences. Furthermore, systematic engineering of the crRNA and Cascade subunits, as well as optimizing the architecture and NLS, may further enhance the on-target efficiency and reduce the off-target activity.

Methods

Identification of type I-F2 systems

We downloaded genomic data of bacteria and archaea from NCBI in 2020, constructing a representative dataset by selecting one genome per species. In addition, we obtained human gut metagenomic data from several articles^56,57,58. CRISPR array was predicted by MinCED (version 0.2.0) software. Cas7, Cas5, and Cas6 proteins were identified by the following three methods with HMMER (version 3.3.2) software: (1) Using Pfam domain (Cas7: PF05107, Cas6: PF09559, PF10040, PF01881, PF17262, PF17952, PF17955, PF19308, Cas5: PF09704, e-value cutoff of 1E-5); (2) Using the previously described CRISPR-Cas protein^1,35; (3) From the CRISPRCasdb database. To find out more I-F2 systems, profiles of I-F2 Cas5/6/7 sequences were generated using MAFFT (version 7.487) and hmmbuild software. Proteins from constructed representative dataset and metagenome were searched against the HMM profiles using hmmsearch. The cluster simultaneously containing proteins Cas5, Cas6, and Cas7 of I-F2 was selected as a candidate I-F2 system.

Phylogenetic tree analysis

After being manually confirmed, 507 Cas7 proteins from different subtypes were aligned using MAFFT (version 7.487) software. The maximum likelihood (ML) phylogenetic tree was constructed using IQ-TREE (version 2.0.3) software⁵⁹, with automatic model selection and 1000 bootstraps and then visualized by the iTOL website. The protein sequences of Cas7 used to construct the phylogenetic tree are listed in Source data. For the phylogenetic tree analysis of the I-F2 systems, we concatenated Cas7, Cas5, and Cas6 from I-F2 CRISPR-Cas systems and constructed the phylogenetic tree with the same method. Multiple sequence alignment of the repeats was conducted using Geneious Prime (version 2023.2.1).

PAM prediction

For each of the eight I-F2 systems, all spacer sequences were aligned to the virus database (IMG/VR: https://genome.jgi.doe.gov/portal/IMG_VR/IMG_VR.home.html) using BLASTN software and the results with alignment length of 32 nt were subjected to further analysis. Then the aligned 32 nt along with its upstream 10 nt were extracted and redundancy was removed when multiple results were found. Ultimately, upstream 10 nt sequences were selected for visualization using WebLogo (https://weblogo.threeplusone.com/), of which only systems with over 10 hits were displayed using WebLogo. Detailed information is listed in Source data.

Plasmid construction

For the expression of the I-F2 CRISPR-Cas system in E. coli, cas5–7 from eight candidate systems were E. coli codon-optimized, synthesized (GeneScript), and ligated to the pETDuet vector (predigested with NdeI and KpnI) using T4 DNA ligase. The mini-CRISPR fragments derived from the primer annealing extensions were ligated to the pETDuet vector (predigested with XbaI and BamHI) using T4 DNA ligase. The Mos350 cas3 gene was E. coli codon-optimized and synthesized into the pCOLADuet vector within NdeI and KpnI (GeneScript).

Cascade fragments with His-tag fusing to Cas7 or Cas7 (L175F) amplified by PCR were ligated to the pETDuet vector (predigested with NdeI and KpnI) using T4 DNA ligase. The CRISPR fragments containing five spacers were synthesized and ligated to the pETDuet vector already containing Cascade-His fragments (predigested with XbaI and BamHI) using T4 DNA ligase. The CRISPR fragments were ligated to pACYC184 (predigested with SphI and BamHI) using T4 DNA ligase.

For transcriptional activation in human cells, dcas9, cas5–7 from 11 candidate systems and VPR transcriptional activator¹⁹ were human codon-optimized, synthesized (GeneScript), and integrated into the pcDNA3.1 vector by Gibson assembly. For base editing in human cells, the protein-coding sequences and the linearized pcDNA3.1-P2A-EGFP were ligated by Gibson assembly. Mini-CRISPRs (repeat-BbsI-BbsI-repeat) and the hU6 promoter were inserted into the linearized pcDNA3.1 vector by Gibson assembly. Oligonucleotides encoding spacers were annealed and ligated into linearized vectors (predigested with BbsI) using T₄ DNA ligase. The Gibson assembly was conducted using the Hieff Clone® Plus Multi One Step Cloning Kit (YESEN). The restriction enzymes and T4 DNA ligase were ordered from New England BioLabs. The full sequences of the plasmids used in the paper are listed in Supplementary Data 1.

Measurement of bacterial growth and fluorescence

For fluorescence measurements, the plasmid expressing the Cascade and mini-CRISPR and the plasmid expressing the YFP reporter were co-transformed into E. coli BL21 (DE3). The transformant was inoculated into 3 mL of Luria-Bertani (LB) medium with antibiotics (100 µg/mL carbenicillin, 34 µg/mL chloramphenicol) and grown overnight at 200 r.p.m. and 37 °C. Then, the seed broth was inoculated at 1:100 into M9 medium (1× M9 salt, 0.1 mM CaCl₂, 2 mM MgSO₄, 10 g/mL thiamine hydrochloride, and 1% (w/v) tyrosine acid) containing antibiotics of the same concentrations and 0.05 mM isopropyl-β-D-thiogalactoside (IPTG). After cultivation at 30 °C for 12 h, the OD₆₀₀ and fluorescence values were measured using the Synergy H4 Hybrid Reader (BioTek). The fluorescence measurement wavelength is set as follows: 500 nm for YFP excitation, 550 nm for YFP emission. Fluorescence values were normalized to OD and the relative YFP fluorescence values of cultures were calculated as the ratio between cultures induced by 50 μM IPTG and cultures induced by 0 μM IPTG.

PAM library construction

The protospacer with four randomized nucleotides upstream (5′ end) was achieved by PCR with primers PRL1 and PRL2 using the pTemplate vector as the template and ligated to the linearized plasmid pACYC184 fragment (PCR by primers PRL3 and PRL4) by Gibson assembly. The DH5α-competent cells were transformed with the ligation products. More than 10⁴ cells were collected, and library plasmids were extracted using the TIANpure Midi Plasmid Kit (TIANGEN). Primer sequences are listed in Supplementary Data 2.

PAM depletion assay

The plasmid expressing the Cascade and mini-CRISPR and the plasmid expressing Cas3 were co-transformed into E. coli BL21 (DE3) cells. Subsequently, electrocompetent cells were prepared through ice-cold H₂O and 10% glycerol washing. The competent cells were electroporated with 200 ng of library plasmids containing four randomized nucleotides at the upstream (5′ end) of the target sequence. After 2 h of recovery, 10 μL of the culture was inoculated into 1 mL of LB and spread on the selective medium, and the colony-forming units were measured to ensure proper coverage of all possible combinations of four randomized nucleotides. Recovery cultures were transferred to 10 mL of liquid medium containing appropriate antibiotics (100 µg/mL carbenicillin, 34 µg/mL chloramphenicol, and 50 µg/mL kanamycin) and 0.05 mM IPTG (or 0 mM IPTG as a negative control), which ensured plasmid propagation and Cascade-Cas3 effector production, and grew at 25 °C for 48 h. The propagated plasmids were extracted using the TIANpure Midi Plasmid Kit (TIANGEN) and analyzed relative to the non-induced control. The target region was amplified by PCR with primers containing different barcodes and deep-sequenced with the Illumina NovoSeq 6000 platform with the PE150 strategy. Sequencing reads were aligned to the corresponding plasmids and PAM randomized regions were extracted. The abundance of each possible four-nucleotide combination was counted from the mapped reads and normalized to the total reads for each sample. The enrichment score of each PAM was the ratio compared to the abundance in the control group without IPTG induction. PAMs with enrichment scores greater than 2.5 were selected and visualized by WebLogo.

Plasmid interference assay

The plasmid expressing the Cascade and mini-CRISPR (the plasmid expressing the Cascade without the mini-CRISPR was used in the negative control) and plasmids expressing Cas3 were co-transformed into E. coli BL21 (DE3). Each transformant was inoculated into 3 mL of LB medium containing antibiotics (100 µg/mL carbenicillin, 34 µg/mL chloramphenicol, and 50 µg/mL kanamycin), cultured overnight and then inoculated into 50 mL of LB medium at a ratio of 1:100 with the same antibiotic concentrations. Cultures were grown to an OD₆₀₀ of 0.3 and then added with different concentrations of IPTG (0.05 mM, 0.1 mM and 0.2 mM) until the OD₆₀₀ was between 0.6 and 0.7. Electrocompetent cells were prepared by washing with ice-cold H₂O and 10% glycerol. The target plasmid (100 ng) was electroporated. The LB medium (1 mL) was added to recover the culture for 1 h. A series of gradient dilutions were performed, and 5 µL was added dropwise to the LB plate with antibiotics (100 µg/mL carbenicillin, 34 µg/mL chloramphenicol, and 50 µg/mL kanamycin). The plates were observed after overnight incubation at 37 °C.

Cell culture and transfection

HEK293T cells (ATCC CRL-3216) were grown in Dulbecco’s Modified Eagle’s Medium (Invitrogen) with 10% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin, and the cells were cultured at 37 °C in an incubator with 5% CO₂. A medium (1 ml) containing 3 × 10⁵ cells was added to each well of a 12-well plate. When the cell fusion was 70–85% (~24 h), 1.2 µg of the Cascade expression plasmid, 0.8 µg of the mini-CRISPR expression plasmid and 2 μL of PEI were added to 200 μL of DMEM medium, and the mixture was placed at room temperature for 15 min. In a comparison control, only 1.2 µg of the Cascade expression plasmid or no plasmid was transfected, respectively. Then, the mixture was added with 550 μL of 2% FBS DMEM medium and mixed well. The medium in the well plate was replaced with the mixture. After 6 h, the mixture was replaced with 10% FBS DMEM medium.

All target sites are 32 nt in length and have a 5′-CC PAM. For transcriptional activation experiments, we firstly selected a target site within the promoter region of the IL1B gene. Because, by using the related type I-F1 Cascade-VPR system targeting the promoter region, the IL1B gene had a strong transcriptional activation level¹⁹. To determine the gene activation effect at different loci, we selected target sites within promoter regions of five genes, which could be transcriptionally upregulated by dCas9-VPR or I-F1 Cascade-VPR^19,36. For base editing experiments targeting multiple adenines, target sites were selected around the regions, which were proved targetable for editing^39,40. For base editing experiments targeting only a single adenine, target sites were randomly selected, of which the single adenine is located near the two peaks (position 9 or position 27).

Immunofluorescence staining

HEK293T cells were transfected with 1.2 µg of the Cascade expression plasmid and 0.8 µg of the mini-CRISPR expression plasmid in 12-well plates. After 1 d, the cells were digested, and 5 × 10⁴ cells were added to a 35 mm glass bottom. After incubation for 1 d, the cells were washed with PBS and fixed in 4% paraformaldehyde. Cells were incubated with blocking solution (1% BSA PBS, 0.1% Triton X-100 PBS) and then incubated with mouse anti-Flag RBITC (1:300 dilution, Bioss, bs-33346M-RBITC), rabbit anti-Myc FITC (1:300 dilution, Bioss, bs-0842R-FITC), and rabbit anti-HA AF647 (1:300 dilution, Bioss, bs-0966R-A647). The cells were then incubated with DAPI for nucleic acid staining and imaged using a SP8 fluorescence microscope (Leica).

Co-immunoprecipitation and Western blot

HEK293T cells were co-transfected with 2.4 μg of the Cascade expression plasmid and 1.6 μg of the crRNA expression plasmid in 6-well plates. After 1 d, the cells were lysed with 500 μL of the Cell lysis buffer for Western and IP (Beyotime) on ice for 30 min, with protease inhibitor cocktail (Roche) and 1 mM phenylmethanesulfonylfluoride added. Samples were centrifuged at 13,523 × g for 5 min at 4 °C. For co-immunoprecipitation analysis, 400 μL of the supernatant were immunoprecipitated using mouse anti-flag-agarose conjugate (Sigma-Aldrich) at 4 °C for 3 h. The immunoprecipitation products were washed four times with IP Lysis Buffer, then mixed with 5× loading buffer (10% SDS, 0.5% bromophenol blue, 50% glycerol, 5% β-mercaptoethanol, 0.25 M pH 6.8 Tris-HCl), and heated at 100 °C for 10 min as output. The rest 100 μL of the supernatant were mixed with 5× loading buffer and heated at 100 °C for 10 min as input. Samples were loaded onto 12% SDS-PAGE and electrophoresed for 1.5 h at 100 V in the running buffer (0.1% SDS, 25 mM Tris, 192 mM glycine) using the Mini-Protean Tetra system (Bio-Rad). Proteins were transferred to PVDF membranes in the transfer buffer (20% methanol, 25 mM Tris, 190 mM Glycine) at 300 mA for 1.25 h at 4 °C. The blot was blocked with 5% milk-TBST (10 mM Tris, 150 mM NaCl, and 0.1% Tween-20) for 2 h at room temperature, followed by incubation with mouse anti-Flag (1:5000 dilution, Sigma-Aldrich, M2 clone F3165) or rabbit anti-Myc (1:5000 dilution, CapitalBio, P1007t) in 5% milk-TBST at 4 °C overnight. Then the blots were washed with TBST, followed by incubation with goat anti-mouse conjugated horseradish peroxidase (HRP) (1:5000 dilution, Zhongshan Golden Bridge, ZB-2305) or goat anti-rabbit conjugated HRP (1:5000 dilution, Zhongshan Golden Bridge, ZB-2301) in 5% milk-TBST for 2 h at room temperature. The blots were washed in TBST and then visualized using Western-ECL (Thermo Fisher) substrates on a Tanon 5200 Multi Chemiluminescent Imaging System (Tanon).

RNA analysis

The Cascade-VPR expression plasmid (1.2 μg) and mini-CRISPR expression plasmid (0.8 μg) were co-transfected into HEK293T cells in 12-well plates. Total RNA was extracted using a SteadyPure Universal RNA Extraction Kit II (Accurate Biology) after 2 d. For qPCR analysis, the total RNA of each sample (1.25 μg) was used for reverse transcription in a 20 μL reaction using the ABScript III RT Master Mix for qPCR with a gDNA Remover kit (Abclonal). In each reaction, 2 μL of the cDNA was used with the KAPA SYBR FAST Universal qPCR Kit (KAPA) and run on a CFX96 real-time PCR detection system (Bio-Rad). All qPCR data were expressed as fold change in RNA, normalized to GAPDH expression, and analyzed relative to the control group. The relative expression level was determined by the 2^−ΔΔCt method. To normalize data from different TSS between genes (ACTC1, ASCL1, and NEUROD1), data in the same gene were normalized by percentage, and 100% was defined by the sum of all values. The target sites were divided into six regions based on the distance to the TSS (0−100 bp, 100−200 bp, 200−300 bp, 300−400 bp, 400−500 bp, >500 bp). Normalized values for all 3 genes are plotted as box-and-whisker plots with min to max as an option. Each bar chart of qPCR we showed is the results of samples treated in the same batch (basically including three biological replicates per sample and three technical replicates per biological replicate). Moreover, each experiment was conducted at least in two batches with similar differential trends. All primer sequences used in qPCR are listed in Supplementary Data 2.

For Northern blot analysis, the total RNA of each sample (10 μg) and the Century-Plus RNA ladder (Thermo Fisher Scientific) were mixed with an equal volume of RNA loading dye (New England Biolabs) respectively, denatured in a water bath at 65 °C for 10 min, and placed on ice for 2 min. Samples were shortly centrifuged, loaded onto 8% polyacrylamide gel, and electrophoresed at 200 V for 40 min in TBE buffer using the Mini-Protean Tetra system (Bio-Rad). The lane including the RNA marker was excised, stained with ethidium bromide, and imaged using a GenoSens2100 System (CLINX). The RNA samples were transferred onto a Biodyne B nylon membrane (Pall) in TBE buffer and then the membrane was UV cross-linked. The hybridization signal from the biotin-labeled probe was detected using the Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Fisher Scientific), following the manufacturer’s protocol. The membrane was imaged using a Tanon 5200 Multi chemiluminescent imaging system (Tanon). All probe sequences are listed in Supplementary Data 2. Uncropped and unprocessed scans of blot/gel images are provided in Supplementary Fig. 19 or Source data.

For small RNA sequencing, the total RNA of each sample (50 μg) was treated with 1 μL of polynucleotide kinase (New England Biolabs) in 50 μL of total solution at 37 °C for 6 h, followed by incubation at 37 °C for 2 h with 5 μL of 10 mM ATP. Then, the proteins were removed using the phenol-chloroform method. Purified RNA was treated with RNA 5′ pyrophosphohydrolase (New England Biolabs) for 2 h at 37°C. RNA was purified by phenol-chloroform extraction for sequencing. RNA molecules ranging from 50 to 100 nt were selected to construct a small RNA library using the NEXTFLEX Small RNA-Seq Kit (Bioo Scientific) and then subjected to Illumina HiSeq sequencing (paired-end, 150-bp reads). The raw data were processed to remove the adapters using Cutadapt (version 3.5) software⁶⁰. The resulting reads were mapped to the mini-CRISPR sequence using the BLASTN software, and the coverage depth of each site was calculated using custom Perl scripts.

Structure analysis

The superposition of three-dimensional structures of Cas proteins were analyzed using PyMol (version 2.5.5) software. The structure of SpuCascade³⁰ (5O6U) was downloaded from the PDB database. The structure of Mos350 Cas5 (A0A378Q7F6), Mos350 Cas6 (A0A378Q7P1), and Mos350 Cas7 (A0A378Q7G2) were downloaded from AlphaFoldDB^61,62. Sequence alignment of Cas7 proteins from Spu and Mos350 was performed by MUSCLE alignment using Geneious Prime (version 2023.2.1).

Protein expression and purification

The Cascade and CRISPR array (with 5 spacers) expression plasmid and mini-CRISPR expression plasmid were co-transformed into E. coli BL21 (DE3). Cells were induced with 0.1 mM IPTG at OD600 = 0.6. After 12 h of incubation at 16 °C, cells were collected by centrifugation and resuspended using lysis buffer (20 mM HEPES pH 8, 5% glycerol, 500 mM NaCl, 20 mM imidazole). Cells were lysed through the ultrahigh pressure cell disrupter at 4 °C. Proteins were purified by Histrap^TM column (Cytiva) and eluted using 80% lysis buffer containing 500 mM imidazole. The eluate was further purified by a Superdex 200 Increase 10/300 GL (Cytiva) equilibrated with a buffer containing 20 mM HEPES pH 7.5, 150 mM NaCl, 5% glycerol.

EMSA experiments

FAM-labeled dsDNA preparation: the FAM-labeled ssDNA molecule and its complementary unlabeled ssDNA molecule were annealed in a 1:1.5 ratio to obtain FAM-labeled double-stranded DNA.

For DNA binding experiments, gradient concentrations (0, 5, 50, 100, 150, 200, 250, 500, 1000, 2000 nM) of Cascade-crRNA complex (Mos350 Cascade or Mos350 Cascade L175F) and 25 nM FAM-labeled 59 bp dsDNA were incubated at 30 °C for 30 min in a 10 μL reaction system containing 20 mM HEPES pH 7.5, 150 mM NaCl, 5% glycerol. For the competition mobility shifts assays, 500 nM Mos350 Cascade-crRNA complex or 100 nM Mos350 Cascade L175F-crRNA complex and 25 nM FAM-labeled 59 bp dsDNA were incubated with 1250 nM non-specific dsDNA or specific dsDNA without FAM labels (50×) at 30 °C for 30 min in the same reaction system. Reactants were electrophoresed on a 4% polyacrylamide (79:1, acrylamide:bisacrylamide) native gel in 0.5× TBE buffer and visualized by fluorescence imaging on a Tanon 5200 Multi Chemiluminescent Imaging System (Tanon). Uncropped and unprocessed scans of blot/gel images are provided in Supplementary Fig. 19 or Source data.

Off-target analysis

To predict Mos350 Cascade-VPR off-target sites, disregarding the mismatches at positions 6, 12, 18, and 24–32 of the protospacer, we searched for all possible off-targets with ≤4 mismatches to the remaining positions of the protospacer using Cas-OFFinder⁶³. All possible off-target sites located in the promoter region of a given gene (≤2 kb upstream of the TSS) are listed in Source data. To predict Mos350 Cascade-ABE off-target sites, we searched for sites of ≤3 mismatches, disregarding the mismatches at positions 6, 12, 18, and 24–32 of the protospacer, on the whole genome using Cas-OFFinder. Target and off-target site information of base editing experiments are listed in Supplementary Data 3.

Genomic DNA extraction and high-throughput DNA sequencing

At 48 h post-transfection, cells were washed with PBS. Then cells were digested with 0.25% trypsin (Gibco) for FACS. Samples were operated on a flow cytometer (BD influx) using BD FACSDiva™ Software (version 6.1.3) after passing through a 70 μm cell strainer. Linear SSC-A versus linear FSC-A was used to select the live HEK293T cells. Within the gate, FSC-W versus FSC-H and SSC-H versus SSC-W were used to gate on single cells. Then within the single live cell gate, we utilized EGFP-negative control to delineate the boundary between positive and negative cells. The final gate was utilized to sort GPF-positive single live cells. Then, cells that were positive for the EGFP signal (20,000–40,000 cells) were collected and lysed with 30–60 μL lysis solution (per liter, 10 mL of 1 M Tris-HCl, pH 7.5, 5 mL of 10% (wt/vol) SDS solution, adding nuclease-free water to a final volume of 1 L) with 1:1000 (vol/vol) proteinase K (New England Biolabs) at 37 °C for 1 h. After heated at 80 °C for 20 min, the samples were used as PCR templates. PCR products were purified by 2% agarose gel electrophoresis. PCR products with different barcodes were mixed and deep-sequenced using the Illumina NovoSeq 6000 platform. Editing results were analyzed using CRISPResso2 with the following parameters: minimum of 60% homology for alignment to the amplicon sequence, quantification window of 20 bp, quantification_window_center of −15. The base editing values were calculated as the percentage of reads with A⋅T-to-G⋅C conversions to the total aligned reads. Primer sequences for PCR are listed in Supplementary Data 2.

WGS and data analysis

For WGS, NC (negative control) group was wild-type HEK293T cells; TC (treatment control) group was cells transfected with two empty plasmids (pcDNA3.1-P2A-EGFP and pcDNA3.1-hU6); KLF4 group was cells transfected with plasmids expressing 5NABE and crRNA targeting site 10. Each group contained three biological replicates. At 48 h post-transfection, 1 × 10⁶ cells that were positive for the EGFP signal (FITC ≥ 500) were collected and genome DNA was extracted using DNeasy Blood & Tissue Kit (QIAGEN). For library construction, ~1 μg of genomic DNA was used according to the protocol of MGIEasy Universal DNA Library Prep Set (MGI) to obtain the final library. WGS analysis was performed at mean coverage of 50× by MGISEQ-2000.

Reads were separately aligned to the GRCh38 reference genome using BWA-MEM (version 0.7.17-r1188). The aligned BAMs were then sorted and duplicates were marked using Picard tools (version 2.25.4) with default settings. Base quality recalibration was performed using GATK (version 4.2.4.0). Variant calling was conducted independently for each sample using three algorithms: GATK HaplotypeCaller, Lofreq (version 2.1.5)⁶⁴, and Strelka (version 2.9.10). Variants called in wild-type samples (NC group) were considered background mutations and removed from the control group (TC group) and experimental group (KLF4 group). SNVs called by all three algorithms were considered true variants. To minimize false-positive sites due to sequence alignment errors in repetitive genomic regions, variants detected in genomicSuperDups/simpleRepeat/rmsk/chainSelf/microsat from the UCSC database were filtered out using bcftools (version 1.3.1). To assess gRNA-independent off-target edits, the locations of predicted gRNA-dependent off-target edits generated by Cas-OFFinder software were removed. Finally, the number of SNVs in the KLF4 group and TC group were compared, and the P value was calculated using unpaired two-tailed Student’s t-test.

Statistics and reproducibility

All data analyzed was presented as mean ± s.d. (n = 3 biological replicates), unless otherwise stated. All P values were calculated by one-way analysis of variance (ANOVA) with Dunnett’s multiple comparisons tests (α = 0.05) (for data having more than two groups) or unpaired two-tailed Student’s t-test (for data having only two groups). Statistical significance level: n.s., not significant; *P < 0.05; **P < 0.01; ***P < 0.001. All P values are listed in Supplementary Data 4. No data were excluded from the analyses.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The WGS data generated in this study have been deposited in the Genome Sequence Archive⁶⁵ in National Genomics Data Center⁶⁶, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences under accession code HRA008013. The small RNA sequencing data generated in this study have been deposited in the same database under accession code HRA008012. The NGS data for base editing generated in this study have been deposited in the same database under accession code HRA008009, HRA008010, HRA008011, respectively. The NGS data for PAM identification generated in this study have been deposited in the same database under accession code CRA017797. Source data are provided with this paper.

Code availability

Custom scripts designed for splitting Fastq files based on barcodes during base editing analysis and calculating the coverage depth of each site in small RNA sequencing have been deposited in Zenodo [https://doi.org/10.5281/zenodo.12748513]⁶⁷.

References

Makarova, K. S. et al. Evolutionary classification of CRISPR-Cas systems: a burst of class 2 and derived variants. Nat. Rev. Microbiol. 18, 67–83 (2020).
Article CAS PubMed Google Scholar
Hille, F. et al. The biology of CRISPR-Cas: backward and forward. Cell 172, 1239–1259 (2018).
Article CAS PubMed Google Scholar
Xiao, Y. et al. Structure basis for directional R-loop formation and substrate handover mechanisms in Type I CRISPR-Cas system. Cell 170, 48–60.e11 (2017).
Article CAS PubMed PubMed Central Google Scholar
Cheng, F. et al. Harnessing the native type I-B CRISPR-Cas for genome editing in a polyploid archaeon. J. Genet. Genomics 44, 541–548 (2017).
Article CAS PubMed Google Scholar
Luo, M. L., Mullis, A. S., Leenay, R. T. & Beisel, C. L. Repurposing endogenous type I CRISPR-Cas systems for programmable gene repression. Nucleic Acids Res. 43, 674–681 (2015).
Article CAS PubMed Google Scholar
Csorgo, B. et al. A compact Cascade-Cas3 system for targeted genome engineering. Nat. Methods 17, 1183–1190 (2020).
Article CAS PubMed PubMed Central Google Scholar
Klompe, S. E., Vo, P. L. H., Halpin-Healy, T. S. & Sternberg, S. H. Transposon-encoded CRISPR-Cas systems direct RNA-guided DNA integration. Nature 571, 219–225 (2019).
Article CAS PubMed Google Scholar
Du, K., Gong, L., Li, M., Yu, H. & Xiang, H. Reprogramming the endogenous type I CRISPR‐Cas system for simultaneous gene regulation and editing in Haloarcula hispanica. mLife 1, 40–50 (2022).
Article CAS PubMed PubMed Central Google Scholar
Cameron, P. et al. Harnessing type I CRISPR-Cas systems for genome engineering in human cells. Nat. Biotechnol. 37, 1471–1477 (2019).
Article CAS PubMed Google Scholar
Dolan, A. E. et al. Introducing a spectrum of long-range genomic deletions in human embryonic stem cells using Type I CRISPR-Cas. Mol. Cell 74, 936–950.e935 (2019).
Article CAS PubMed PubMed Central Google Scholar
Morisaka, H. et al. CRISPR-Cas3 induces broad and unidirectional genome editing in human cells. Nat. Commun. 10, 5302 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Hu, C. et al. Allosteric control of type I-A CRISPR-Cas3 complexes and establishment as effective nucleic acid detection and human genome editing tools. Mol. Cell 82, 2754–2768.e2755 (2022).
Article CAS PubMed PubMed Central Google Scholar
Tan, R. et al. Cas11 enables genome engineering in human cells with compact CRISPR-Cas3 systems. Mol. Cell 82, 852–867.e855 (2022).
Article CAS PubMed PubMed Central Google Scholar
Li, Y. et al. Targeted large fragment deletion in plants using paired crRNAs with type I CRISPR system. Plant Biotechnol. J. 21, 2196–2208 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kita, Y. et al. Dual CRISPR-Cas3 system for inducing multi-exon skipping in DMD patient-derived iPSCs. Stem Cell Rep. 18, 1753–1765 (2023).
Article CAS Google Scholar
Osakabe, K. et al. Genome editing in plants using CRISPR type I-D nuclease. Commun. Biol. 3, 648 (2020).
Article CAS PubMed PubMed Central Google Scholar
Osakabe, K., Wada, N., Murakami, E., Miyashita, N. & Osakabe, Y. Genome editing in mammalian cells using the CRISPR type I-D nuclease. Nucleic Acids Res. 49, 6347–6363 (2021).
Article CAS PubMed PubMed Central Google Scholar
Zimmermann, A. et al. A Cas3-base editing tool for targetable in vivo mutagenesis. Nat. Commun. 14, 3389 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Chen, Y. et al. Repurposing type I-F CRISPR-Cas system as a transcriptional activation tool in human cells. Nat. Commun. 11, 3136 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Young, J. K. et al. The repurposing of type I-E CRISPR-Cascade for gene activation in plants. Commun. Biol. 2, 383 (2019).
Article PubMed PubMed Central Google Scholar
Pickar-Oliver, A. et al. Targeted transcriptional modulation with type I CRISPR-Cas systems in human cells. Nat. Biotechnol. 37, 1493–1501 (2019).
Article CAS PubMed PubMed Central Google Scholar
Mendell, J. R. et al. Current clinical applications of in vivo gene therapy with AAVs. Mol. Ther. 29, 464–488 (2021).
Article CAS PubMed Google Scholar
Szczelkun, M. D. et al. Direct observation of R-loop formation by single RNA-guided Cas9 and Cascade effector complexes. Proc. Natl Acad. Sci. USA 111, 9798–9803 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Jiang, F. et al. Structures of a CRISPR-Cas9 R-loop complex primed for DNA cleavage. Science 351, 867–871 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A. & Liu, D. R. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533, 420–424 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Ye, L. et al. Glycosylase-based base editors for efficient T-to-G and C-to-G editing in mammalian cells. Nat. Biotechnol. https://doi.org/10.1038/s41587-023-02050-w (2024).
Article PubMed Google Scholar
Anzalone, A. V., Koblan, L. W. & Liu, D. R. Genome editing with CRISPR-Cas nucleases, base editors, transposases and prime editors. Nat. Biotechnol. 38, 824–844 (2020).
Article CAS PubMed Google Scholar
Han, D. et al. Development of miniature base editors using engineered IscB nickase. Nat. Methods 20, 1029–1036 (2023).
Article CAS PubMed Google Scholar
Swarts, D. C., van der Oost, J. & Jinek, M. Structural basis for guide RNA processing and seed-dependent DNA targeting by CRISPR-Cas12a. Mol. Cell 66, 221–233.e224 (2017).
Article CAS PubMed PubMed Central Google Scholar
Pausch, P. et al. Structural variation of Type I-F CRISPR RNA guided DNA surveillance. Mol. Cell 67, 622–632.e624 (2017).
Article CAS PubMed Google Scholar
Dwarakanath, S. et al. Interference activity of a minimal Type I CRISPR-Cas system from Shewanella putrefaciens. Nucleic Acids Res 43, 8913–8923 (2015).
Article CAS PubMed PubMed Central Google Scholar
Makarova, K. S. et al. Evolution and classification of the CRISPR-Cas systems. Nat. Rev. Microbiol. 9, 467–477 (2011).
Article CAS PubMed Google Scholar
Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816–821 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
McBride, T. M. et al. Diverse CRISPR-Cas complexes require independent translation of small and large subunits from a single gene. Mol. Cell 80, 971–979.e977 (2020).
Article CAS PubMed Google Scholar
Klompe, S. E. et al. Evolutionary and mechanistic diversity of Type I-F CRISPR-associated transposons. Mol. Cell 82, 616–628.e615 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chavez, A. et al. Highly efficient Cas9-mediated transcriptional programming. Nat. Methods 12, 326–328 (2015).
Article CAS PubMed PubMed Central Google Scholar
Kuznedelov, K. et al. Altered stoichiometry Escherichia coli Cascade complexes with shortened CRISPR RNA spacers are capable of interference and primed adaptation. Nucleic Acids Res. 44, 10849–10861 (2016).
Article CAS PubMed PubMed Central Google Scholar
Gleditzsch, D. et al. Modulating the Cascade architecture of a minimal Type I-F CRISPR-Cas system. Nucleic Acids Res. 44, 5872–5882 (2016).
Article CAS PubMed PubMed Central Google Scholar
Richter, M. F. et al. Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nat. Biotechnol. 38, 883–891 (2020).
Article CAS PubMed PubMed Central Google Scholar
Gaudelli, N. M. et al. Programmable base editing of A*T to G*C in genomic DNA without DNA cleavage. Nature 551, 464–471 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Lettre, G. & Bauer, D. E. Fetal haemoglobin in sickle-cell disease: from genetic epidemiology to new therapeutic strategies. Lancet 387, 2554–2564 (2016).
Article CAS PubMed Google Scholar
Sankaran, V. G. & Orkin, S. H. The switch from fetal to adult hemoglobin. Cold Spring Harb. Perspect. Med. 3, a011643 (2013).
Article PubMed PubMed Central Google Scholar
Canver, M. C. et al. BCL11A enhancer dissection by Cas9-mediated in situ saturating mutagenesis. Nature 527, 192–197 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Wu, Y. et al. Highly efficient therapeutic gene editing of human hematopoietic stem cells. Nat. Med. 25, 776–783 (2019).
Article CAS PubMed PubMed Central Google Scholar
Fu, B. et al. CRISPR-Cas9-mediated gene editing of the BCL11A enhancer for pediatric beta⁰/beta⁰ transfusion-dependent beta-thalassemia. Nat. Med. 28, 1573–1580 (2022).
Article CAS PubMed Google Scholar
Rees, H. A. et al. Improving the DNA specificity and applicability of base editing through protein engineering and protein delivery. Nat. Commun. 8, 15790 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Davis, J. R. et al. Efficient in vivo base editing via single adeno-associated viruses with size-optimized genomes encoding compact adenine base editors. Nat. Biomed. Eng. 6, 1272–1283 (2022).
Article CAS PubMed PubMed Central Google Scholar
Altae-Tran, H. et al. Uncovering the functional diversity of rare CRISPR-Cas systems with deep terascale clustering. Science 382, eadi1910 (2023).
Article CAS PubMed PubMed Central Google Scholar
Gong, L. et al. Primed adaptation tolerates extensive structural and size variations of the CRISPR RNA guide in Haloarcula hispanica. Nucleic Acids Res. 47, 5880–5891 (2019).
Article CAS PubMed PubMed Central Google Scholar
Xu, X. et al. Engineered miniature CRISPR-Cas system for mammalian genome regulation and editing. Mol. Cell 81, 4333–4345.e4334 (2021).
Article CAS PubMed Google Scholar
Li, J. et al. Plant base editing and prime editing: The current status and future perspectives. J. Integr. Plant Biol. 65, 444–467 (2023).
Article PubMed Google Scholar
Chen, W. et al. Cas12n nucleases, early evolutionary intermediates of type V CRISPR, comprise a distinct family of miniature genome editors. Mol. Cell 83, 2768–2780.e2766 (2023).
Article CAS PubMed Google Scholar
Caldecott, K. W. Causes and consequences of DNA single-strand breaks. Trends Biochem. Sci. 49, 68–78 (2024).
Article CAS PubMed Google Scholar
Tsai, S. Q. et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat. Biotechnol. 33, 187–197 (2015).
Article CAS PubMed Google Scholar
Fu, Y., Sander, J. D., Reyon, D., Cascio, V. M. & Joung, J. K. Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat. Biotechnol. 32, 279–284 (2014).
Article CAS PubMed PubMed Central Google Scholar
Chen, L. et al. Short- and long-read metagenomics expand individualized structural variations in gut microbiomes. Nat. Commun. 13, 3175 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Almeida, A. et al. A new genomic blueprint of the human gut microbiota. Nature 568, 499–504 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Zou, Y. et al. 1520 reference genomes from cultivated human gut bacteria enable functional microbiome analyses. Nat. Biotechnol. 37, 179–185 (2019).
Article CAS PubMed PubMed Central Google Scholar
Minh, B. Q. et al. IQ-TREE 2: New models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).
Article CAS PubMed PubMed Central Google Scholar
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. 17, 10–12 (2011).
Article Google Scholar
Varadi, M. et al. AlphaFold protein structure database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 50, D439–D444 (2022).
Article CAS PubMed Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Bae, S., Park, J. & Kim, J. S. Cas-OFFinder: a fast and versatile algorithm that searches for potential off-target sites of Cas9 RNA-guided endonucleases. Bioinformatics 30, 1473–1475 (2014).
Article CAS PubMed PubMed Central Google Scholar
Wilm, A. et al. LoFreq: a sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. Nucleic Acids Res. 40, 11189–11201 (2012).
Article CAS PubMed PubMed Central Google Scholar
Chen, T. et al. The genome sequence archive family: toward explosive data growth and diverse data types. Genom. Proteom. Bioinforma. 19, 578–583 (2021).
Article Google Scholar
CNCB-NGDC members and partners. Database resources of the National Genomics Data Center, China National Center for Bioinformation in 2022. Nucleic Acids Res. 50, D27–D38 (2022).
Yu, H. Calculate the coverage of sRNA-seq reads from BLAST output and used to split fastq barcode samples. Zenodo. https://doi.org/10.5281/zenodo.12748513 (2024).

Download references

Acknowledgements

This work is supported by the Strategic Priority Research Program of the Chinese Academy of Sciences [XDA24020101] (to H.X.), the National Natural Science Foundation of China [32230061] (to H.X.), the National Key R&D Program of China [2020YFA0906800] (to H.X.), the National Natural Science Foundation of China [32100499] (to L.G.), and the National Natural Science Foundation of China [32150020] (to L.G. and M.L.). We thank Wei Jiang, Xinyu Li, and Dahui Zhao for help with HEK293T cell culture, co-immunoprecipitation, and immunofluorescence assays. We thank Tong Zhao for help with flow cell sorting and Zehua Chen for help with fluorescence reporter system construction in E. coli. We thank Pengju Yu for his help in the qPCR experiments. We thank Qiang Gao, Jiacheng Hu, Shenghan Gao, Yingfeng Luo, and Xinyu Tan for their help in the WGS analysis. We thank Jialin Fan for his help in the off-target site analysis. We thank Zhihua Li and Junyu Chen for their help in the protein purification and EMSA experiments. We thank all members of the Xiang laboratory for helpful advice and discussion.

Author information

These authors contributed equally: Jing Guo, Luyao Gong, Haiying Yu.

Authors and Affiliations

State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
Jing Guo, Luyao Gong, Haiying Yu, Ming Li, Qiaohui An, Zhenquan Liu, Shuru Fan, Changjialian Yang, Dahe Zhao, Jing Han & Hua Xiang
College of Life Science, University of Chinese Academy of Sciences, Beijing, China
Jing Guo, Ming Li, Qiaohui An, Zhenquan Liu, Changjialian Yang & Hua Xiang
Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin, China
Hua Xiang

Authors

Jing Guo
View author publications
You can also search for this author in PubMed Google Scholar
Luyao Gong
View author publications
You can also search for this author in PubMed Google Scholar
Haiying Yu
View author publications
You can also search for this author in PubMed Google Scholar
Ming Li
View author publications
You can also search for this author in PubMed Google Scholar
Qiaohui An
View author publications
You can also search for this author in PubMed Google Scholar
Zhenquan Liu
View author publications
You can also search for this author in PubMed Google Scholar
Shuru Fan
View author publications
You can also search for this author in PubMed Google Scholar
Changjialian Yang
View author publications
You can also search for this author in PubMed Google Scholar
Dahe Zhao
View author publications
You can also search for this author in PubMed Google Scholar
Jing Han
View author publications
You can also search for this author in PubMed Google Scholar
Hua Xiang
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

L.G., J.G., and H.X. designed the project, performed analyses, and wrote the manuscript. J.G. conducted experiments. Q.A., Z.L., S.F., and C.Y. participated in vector construction and cell experiments. H.Y. performed bioinformatic analyses. M.L., J.H., and D.Z. performed analyses. H.X. and L.G. supervised the research. All the authors have proofread the manuscript.

Corresponding authors

Correspondence to Luyao Gong or Hua Xiang.

Ethics declarations

Competing interests

H.X., J.G., and L.G. have submitted a patent application on the use of I-F2 Cascades for transcriptional activation tools (application number PCT/CN2022/143090; 202211702167.3, China). H.X., J.G., L.G., and H.Y. have submitted a patent application on the use of I-F2 Cascades for base editing tools (application number PCT/CN2024/074258). The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Guo, J., Gong, L., Yu, H. et al. Engineered minimal type I CRISPR-Cas system for transcriptional activation and base editing in human cells. Nat Commun 15, 7277 (2024). https://doi.org/10.1038/s41467-024-51695-x

Download citation

Received: 24 January 2024
Accepted: 15 August 2024
Published: 23 August 2024
DOI: https://doi.org/10.1038/s41467-024-51695-x

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.