Beyond the promise: evaluating and mitigating off-target effects in CRISPR gene editing for safer therapeutics

Over the last decade, CRISPR has revolutionized drug development due to its potential to cure genetic diseases that currently do not have any treatment. CRISPR was adapted from bacteria for gene editing in human cells in 2012 and, remarkably, only 11 years later has seen it's very first approval as a medicine for the treatment of sickle cell disease and transfusion-dependent beta-thalassemia. However, the application of CRISPR systems is associated with unintended off-target and on-target alterations (including small indels, and structural variations such as translocations, inversions and large deletions), which are a source of risk for patients and a vital concern for the development of safe therapies. In recent years, a wide range of methods has been developed to detect unwanted effects of CRISPR-Cas nuclease activity. In this review, we summarize the different methods for off-target assessment, discuss their strengths and limitations, and highlight strategies to improve the safety of CRISPR systems. Finally, we discuss their relevance and application for the pre-clinical risk assessment of CRISPR therapeutics within the current regulatory context.

CRISPR/Cas9 as a Mutagenic Factor

The discovery of the CRISPR/Cas9 microbial adaptive immune system has revolutionized the field of genetics, by greatly enhancing the capacity for genome editing. CRISPR/Cas9-based editing starts with DNA breaks (or other lesions) predominantly at target sites and, unfortunately, at off-target genome sites. DNA repair systems differing in accuracy participate in establishing desired genetic changes but also introduce unwanted mutations, that may lead to hereditary, oncological, and other diseases. New approaches to alleviate the risks associated with genome editing include attenuating the off-target activity of editing complex through the use of modified forms of Cas9 nuclease and single guide RNA (sgRNA), improving delivery methods for sgRNA/Cas9 complex, and directing DNA lesions caused by the sgRNA/Cas9 to non-mutagenic repair pathways. Here, we have described CRISPR/Cas9 as a new powerful mutagenic factor, discussed its mutagenic properties, and reviewed factors influencing the mutagenic activity of CRISPR/Cas9.

Deep learning in CRISPR-Cas systems: a review of recent studies

In genetic engineering, the revolutionary CRISPR-Cas system has proven to be a vital tool for precise genome editing. Simultaneously, the emergence and rapid evolution of deep learning methodologies has provided an impetus to the scientific exploration of genomic data. These concurrent advancements mandate regular investigation of the state-of-the-art, particularly given the pace of recent developments. This review focuses on the significant progress achieved during 2019-2023 in the utilization of deep learning for predicting guide RNA (gRNA) activity in the CRISPR-Cas system, a key element determining the effectiveness and specificity of genome editing procedures. In this paper, an analytical overview of contemporary research is provided, with emphasis placed on the amalgamation of artificial intelligence and genetic engineering. The importance of our review is underscored by the necessity to comprehend the rapidly evolving deep learning methodologies and their potential impact on the effectiveness of the CRISPR-Cas system. By analyzing recent literature, this review highlights the achievements and emerging trends in the integration of deep learning with the CRISPR-Cas systems, thus contributing to the future direction of this essential interdisciplinary research area.

Transformer-based anti-noise models for CRISPR-Cas9 off-target activities prediction

The off-target effect occurring in the CRISPR-Cas9 system has been a challenging problem for the practical application of this gene editing technology. In recent years, various prediction models have been proposed to predict potential off-target activities. However, most of the existing prediction methods do not fully exploit guide RNA (gRNA) and DNA sequence pair information effectively. In addition, available prediction methods usually ignore the noise effect in original off-target datasets. To address these issues, we design a novel coding scheme, which considers the key features of mismatch type, mismatch location and the gRNA-DNA sequence pair information. Furthermore, a transformer-based anti-noise model called CrisprDNT is developed to solve the noise problem that exists in the off-target data. Experimental results of eight existing datasets demonstrate that the method with the inclusion of the anti-noise loss functions is superior to available state-of-the-art prediction methods. CrisprDNT is available at https://github.com/gzrgzx/CrisprDNT.

Corrigendum: CRISPR nuclease off-target activity and mitigation strategies

[This corrects the article DOI: 10.3389/fgeed.2022.1050507.].

CRISPR nuclease off-target activity and mitigation strategies

The discovery of CRISPR has allowed site-specific genomic modification to become a reality and this technology is now being applied in a number of human clinical trials. While this technology has demonstrated impressive efficacy in the clinic to date, there remains the potential for unintended on- and off-target effects of CRISPR nuclease activity. A variety of in silico-based prediction tools and empirically derived experimental methods have been developed to identify the most common unintended effect-small insertions and deletions at genomic sites with homology to the guide RNA. However, large-scale aberrations have recently been reported such as translocations, inversions, deletions, and even chromothripsis. These are more difficult to detect using current workflows indicating a major unmet need in the field. In this review we summarize potential sequencing-based solutions that may be able to detect these large-scale effects even at low frequencies of occurrence. In addition, many of the current clinical trials using CRISPR involve ex vivo isolation of a patient's own stem cells, modification, and re-transplantation. However, there is growing interest in direct, in vivo delivery of genome editing tools. While this strategy has the potential to address disease in cell types that are not amenable to ex vivo manipulation, in vivo editing has only one desired outcome-on-target editing in the cell type of interest. CRISPR activity in unintended cell types (both on- and off-target) is therefore a major safety as well as ethical concern in tissues that could enable germline transmission. In this review, we have summarized the strengths and weaknesses of current editing and delivery tools and potential improvements to off-target and off-tissue CRISPR activity detection. We have also outlined potential mitigation strategies that will ensure that the safety of CRISPR keeps pace with efficacy, a necessary requirement if this technology is to realize its full translational potential.

Chemokine receptor antagonists with α1-adrenergic receptor blocker activity

OBJECTIVES

Chemokine receptor antagonists are being explored for their therapeutic potential in various disease processes. As the chemokine (C-C motif) receptor 2 (CCR2) antagonist RS504393 is known to compete with ligand binding to α₁-adrenoceptors, we tested a panel of 10 CCR antagonists for interactions with α₁-adrenoceptors to evaluate potential cardiovascular activities and side-effect profiles.

METHODS

The PRESTO-Tango β-arrestin recruitment assay was utilized to test whether the CCR antagonists interfere with α_1b-AR activation upon stimulation with phenylephrine. Pressure myography with isolated rat resistance arteries was employed to assess their effects on phenylephrine-induced vasoconstriction. The following antagonists were tested: CCR1-BX471, BX513, BI639667; CCR2-RS504393, INCB3284; CCR3-SB328437; and CCR4-AZD2098, and C021; CCR5-Maraviroc; CCR10-BI6901. The pan-α₁-adrenoceptor antagonist prazosin was used as control.

RESULTS

Among the CCR antagonists tested, RS504393, BX513, and C021 inhibited phenylephrine-induced β-arrestin recruitment to α_1b-adrenoceptor and phenylephrine-induced vasoconstriction. While RS504393 functioned as a competitive α₁-adrenoceptor blocker, BX513 and C021 functioned as noncompetitive α₁-adrenoceptor antagonists in both assay systems. Furthermore, RS504393, BX513, and C021 dose-dependently dilated arteries that were fully preconstricted with phenylephrine.

CONCLUSIONS

Our data suggest that CCR antagonists should be screened for cross-reactivity with α₁-adrenoceptors to exclude potential adverse cardiovascular effects when used as anti inflammatory drugs.

Quantitative evaluation of chromosomal rearrangements in gene-edited human stem cells by CAST-Seq

Genome editing has shown great promise for clinical translation but also revealed the risk of genotoxicity caused by off-target effects of programmable nucleases. Here we describe chromosomal aberrations analysis by single targeted linker-mediated PCR sequencing (CAST-Seq), a preclinical assay to identify and quantify chromosomal aberrations derived from on-target and off-target activities of CRISPR-Cas nucleases or transcriptional activator-like effector nucleases (TALENs), respectively, in human hematopoietic stem cells (HSCs). Depending on the employed designer nuclease, CAST-Seq detected translocations in 0%-0.5% of gene-edited human CD34+ HSCs, and up to 20% of on-target loci harbored gross rearrangements. Moreover, CAST-Seq detected distinct types of chromosomal aberrations, such as homology-mediated translocations, that are mediated by homologous recombination and not off-target activity. CAST-Seq is a sensitive assay able to identify and quantify unintended chromosomal rearrangements in addition to the more typical mutations at off-target sites. CAST-Seq analyses may be particularly relevant for therapeutic genome editing to enable thorough risk assessment before clinical application of gene-edited products.

Genome-Wide Analysis of Off-Target CRISPR/Cas9 Activity in Single-Cell-Derived Human Hematopoietic Stem and Progenitor Cell Clones

CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9)-mediated genome editing holds remarkable promise for the treatment of human genetic diseases. However, the possibility of off-target Cas9 activity remains a concern. To address this issue using clinically relevant target cells, we electroporated Cas9 ribonucleoprotein (RNP) complexes (independently targeted to two different genomic loci, the CXCR4 locus on chromosome 2 and the AAVS1 locus on chromosome 19) into human mobilized peripheral blood-derived hematopoietic stem and progenitor cells (HSPCs) and assessed the acquisition of somatic mutations in an unbiased, genome-wide manner via whole genome sequencing (WGS) of single-cell-derived HSPC clones. Bioinformatic analysis identified >20,000 total somatic variants (indels, single nucleotide variants, and structural variants) distributed among Cas9-treated and non-Cas9-treated control HSPC clones. Statistical analysis revealed no significant difference in the number of novel non-targeted indels among the samples. Moreover, data analysis showed no evidence of Cas9-mediated indel formation at 623 predicted off-target sites. The median number of novel single nucleotide variants was slightly elevated in Cas9 RNP-recipient sample groups compared to baseline, but did not reach statistical significance. Structural variants were rare and demonstrated no clear causal connection to Cas9-mediated gene editing procedures. We find that the collective somatic mutational burden observed within Cas9 RNP-edited human HSPC clones is indistinguishable from naturally occurring levels of background genetic heterogeneity.

Anticipating and Identifying Collateral Damage in Genome Editing

Genome editing has powerful applications in research, healthcare, and agriculture. However, the range of possible molecular events resulting from genome editing has been underestimated and the technology remains unpredictable on, and away from, the target locus. This has considerable impact in providing a safe approach for therapeutic genome editing, agriculture, and other applications. This opinion article discusses how to anticipate and detect those editing events by a combination of assays to capture all possible genomic changes. It also discusses strategies for preventing unwanted effects, critical to appraise the benefit or risk associated with the use of the technology. Anticipating and verifying the result of genome editing are essential for the success for all applications.

Determinants for Efficient Editing with Cas9-Mediated Recombineering in Escherichia coli

In E. coli, editing efficiency with Cas9-mediated recombineering varies across targets due to differences in the level of Cas9:gRNA-mediated DNA double-strand break (DSB)-induced cell death. We found that editing efficiency with the same gRNA and repair template can also change with target position, cas9 promoter strength, and growth conditions. Incomplete editing, off-target activity, nontargeted mutations, and failure to cleave target DNA even if Cas9 is bound also compromise editing efficiency. These effects on editing efficiency were gRNA-specific. We propose that differences in the efficiency of Cas9:gRNA-mediated DNA DSBs, as well as possible differences in binding of Cas9:gRNA complexes to their target sites, account for the observed variations in editing efficiency between gRNAs. We show that editing behavior using the same gRNA can be modified by mutating the gRNA spacer, which changes the DNA DSB activity. Finally, we discuss how variable editing with different gRNAs could limit high-throughput applications and provide strategies to overcome these limitations.

Defining the seed sequence of the Cas12b CRISPR-Cas effector complex

Target binding by CRISPR-Cas ribonucleoprotein effectors is initiated by the recognition of double-stranded PAM motifs by the Cas protein moiety followed by destabilization, localized melting, and interrogation of the target by the guide part of CRISPR RNA moiety. The latter process depends on seed sequences, parts of the target that must be strictly complementary to CRISPR RNA guide. Mismatches between the target and CRISPR RNA guide outside the seed have minor effects on target binding, thus contributing to off-target activity of CRISPR-Cas effectors. Here, we define the seed sequence of the Type V Cas12b effector from Bacillus thermoamylovorans. While the Cas12b seed is just five bases long, in contrast to all other effectors characterized to date, the nucleotide base at the site of target cleavage makes a very strong contribution to target binding. The generality of this additional requirement was confirmed during analysis of target recognition by Cas12b effector from Alicyclobacillus acidoterrestris. Thus, while the short seed may contribute to Cas12b promiscuity, the additional specificity determinant at the site of cleavage may have a compensatory effect making Cas12b suitable for specialized genome editing applications.

Defining CRISPR-Cas9 genome-wide nuclease activities with CIRCLE-seq

Circularization for in vitro reporting of cleavage effects by sequencing (CIRCLE-seq) is a sensitive and unbiased method for defining the genome-wide activity (on-target and off-target) of CRISPR-Cas9 nucleases by selective sequencing of nuclease-cleaved genomic DNA (gDNA). Here, we describe a detailed experimental and analytical protocol for CIRCLE-seq. The principle of our method is to generate a library of circularized gDNA with minimized numbers of free ends. Highly purified gDNA circles are treated with CRISPR-Cas9 ribonucleoprotein complexes, and nuclease-linearized DNA fragments are then ligated to adapters for high-throughput sequencing. The primary advantages of CIRCLE-seq as compared with other in vitro methods for defining genome-wide genome editing activity are (i) high enrichment for sequencing nuclease-cleaved gDNA/low background, enabling sensitive detection with low sequencing depth requirements; and (ii) the fact that paired-end reads can contain complete information on individual nuclease cleavage sites, enabling use of CIRCLE-seq in species without high-quality reference genomes. The entire protocol can be completed in 2 weeks, including time for gRNA cloning, sequence verification, in vitro transcription, library preparation, and sequencing.

Data-Driven Exploration of Selectivity and Off-Target Activities of Designated Chemical Probes

Chemical probes are of central relevance for chemical biology. To unambiguously explore the role of target proteins in triggering or mediating biological functions, small molecules used as probes should ideally be target-specific; at least, they should have sufficiently high selectivity for a primary target. We present a thorough analysis of currently available activity data for designated chemical probes to address several key questions: How well defined are chemical probes? What is their level of selectivity? Is there evidence for additional activities? Are some probes "better" than others? Therefore, highly curated chemical probes were collected and their selectivity was analyzed on the basis of publicly available compound activity data. Different selectivity patterns were observed, which distinguished designated high-quality probes.

Reduced seed region-based off-target activity with lentivirus-mediated RNAi

Along with silencing intended target genes, transfected siRNAs regulate numerous unintended transcripts through a mechanism in which the equivalent of a microRNA-like seed region in the siRNA recognizes complementary sequences in transcript 3' UTRs. Amelioration of this off-target silencing would lead to more accurate interpretation of RNA interference (RNAi) experiments and thus greatly enhance their value. We tested whether lentivirus-mediated delivery of shRNA is prone to the sequence-based off-target activity prevalent in siRNA experiments. We compared target gene silencing and overall impact on global gene expression caused by multiple sequences delivered as both transfected siRNAs and lentivirus vector-expressed shRNAs. At equivalent levels of target gene silencing, signatures induced by shRNAs were significantly smaller than those induced by cognate siRNAs and arose less frequently from seed region activity. Interestingly, the low level of seed region-based off-target activity exhibited by shRNAs resulted in down-regulation of transcripts that were largely distinct from those regulated by siRNAs. On the basis of these observations, we recommend lentivirus-mediated RNAi for pathway profiling experiments that measure whole genome transcriptional readouts as well as for large-scale screens when resources for extensive follow up are limited.