Alieh Fazeli
Evelyn Ullrich
Toni Cathomen
Tobias Bexte- 1Institute for Transfusion Medicine and Gene Therapy, Medical Center - University of Freiburg, Freiburg, Germany
- 2PhD Program, Faculty of Biology, University of Freiburg, Freiburg, Germany
- 3Department of Pediatrics, Experimental Immunology and Cell Therapy, Goethe University Frankfurt, Frankfurt am Main, Germany
- 4Frankfurt Cancer Institute, Goethe University Frankfurt, Frankfurt am Main, Germany
- 5German Cancer Consortium (DKTK) partner site Frankfurt/Mainz, Frankfurt am Main, Germany
- 6Center for Cell and Gene Therapy, Medical Center - University of Freiburg, Freiburg, Germany
- 7German Cancer Consortium (DKTK) partner site Freiburg, Freiburg, Germany
- 8Faculty of Medicine, University of Freiburg, Freiburg, Germany
- 9German Red Cross Blood Service Baden-Württemberg - Hessen, Institute for Transfusion Medicine and Immunohematology, Frankfurt am Main, Germany
- 10Berlin Center for Advanced Therapies, Charité – Universitätsmedizin Berlin, Berlin, Germany
CRISPR-based gene editing has become a transformative tool to enhance immune cell therapies. In particular, engineering natural killer (NK) cells with CRISPR/Cas systems has gained traction due to their ability to mediate strong anti-tumor responses in an MHC-unrestricted, non-alloreactive manner. Early trials show the feasibility and safety of allogeneic NK cells, paving the way as scalable “off-the-shelf” products. CRISPR/Cas9 edits genomes by inducing DNA double-strand breaks (DSBs), mainly repaired through non-homologous end joining (NHEJ) or homology-directed repair (HDR). While effective, CRISPR carries risks of off-target (OT) activity that may disrupt essential genes, cause chromosomal rearrangements, or trigger oncogenic changes - posing threats to product integrity and patient safety. These concerns intensify with multiplex editing, where multiple loci are modified to improve function, persistence, and immune evasion. Since unmodified NK cells are typically short-lived, many clinical-stage products are engineered to express IL-15 or related constructs, extending their half-life and amplifying risks associated with unintended changes. This underscores the urgent need for robust safety assessments. In this review, we summarize the current landscape of safety assessment platforms for evaluating gene edited NK cells. We highlight predictive in silico tools, biochemical in vitro assays, and emerging cell-based detection systems to identify and quantify CRISPR-induced OT events. Particular attention is given to their suitability, limitations, and practical use in primary NK cells and multiplex editing strategies. Our aim is to support the design of safe, effective editing workflows for NK cell therapies - ensuring rigor as the field advances rapidly toward clinical application.
Introduction
The bacterial clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated protein 9 (Cas9) was engineered for RNA-programmable DNA cutting the first time in 2012 (1, 2). Today this system has undergone several cycles of optimization and been frequently adapted for precise editing in several human cells including therapies based on T and NK cells (3). The concept is simple: target gene recognition by the guide RNA (gRNA), which leads to a double stranded break (DSB) introduced by the Cas9 protein. The resulting DSB is then repaired by nonhomologous end-joining (NHEJ) or homology-directed repairs (HDR). In context of primary immune cells, delivery of a gRNA/Cas9 ribonucleoprotein (RNP) complex was identified as the most stable version for primary cell editing and is therefore most frequently used and entered multiple clinical trials (4–6). Therefore, CRISPR applications have become a powerful tool to enhance the therapeutic efficacy of primary immune cell therapies (7–10).
In recent years the application of Natural Killer (NK) cells came to the focus as an emerging promising approach for a particularly safe cellular immunotherapy for cancer (11–13). Their intrinsic functionality and their capacity to recognize infected or transformed cells that lack major histocompatibility complexes (MHC), position them as a promising strategy for targeted cancer treatment (14). In addition, the lack of rearranged antigen receptors enables their use in allogeneic, “off-the-shelf” applications. In this regard, early clinical data indicate that allogeneic NK cell applications show a very favorable safety profile compared to chimeric-antigen-receptor (CAR)-T cells (15, 16). Incorporating a CAR into NK cells can further improve their tumor-targeting capacity, and early-phase clinical trials have demonstrated encouraging responses with CAR−NK cell therapies (12, 15–17). Today, there are more than 120 registered CAR-NK cell trials, for hematological cancers, solid tumor and autoimmune and infectious diseases (18). All of them do not show Graft-versus-Host Disease (GvHD), neurotoxicity (ICANS) and almost no cytokine-release-syndrome (CRS) (12, 18).
Although the patient’s response rate is promising, the NK cells functionality can be impaired by inhibitory receptors and negative regulation by the tumor microenvironment (TME), restricting their anti-tumor capacity (17, 19, 20). In addition, NK cells are typically short-lived and their killing activity is balanced by different activating and inhibitory receptors, which was shown to limit their clinical benefit, also of CAR-NK cells (11, 20). These biological constraints highlight the need and enormous potential for genetic engineering strategies to enhance NK cell function and persistence while maintaining their favorable safety profile. The recent developments of CRISPR applications opens the advertisements for a broad potential for precise engineering tools as a powerful approach to boost the therapeutic window of NK cell therapy as a safe cell therapy product (11, 12, 19, 21, 22). Therefore, the rationale for genome editing in NK cells stems from the need to overcome the key limitations without hurting NK cell therapy’s impeccable safety records.
Suboptimal activation and cytotoxicity can result from inhibitory signaling and insufficient activating receptor engagement. To address this, CRISPR has been used to disrupt or enhance the expression of genes involved in receptor-mediated recognition, such as KLRC1 (encoding NKG2A), SIGLEC7 (sialic acid-binding Ig-like lectin 7), CD96, KLRK1 (killer cell lectin-like receptor subfamily K, member 1; also known as NKG2D), NCR1 (natural cytotoxicity triggering receptor 1), NKp46 (natural killer cell protein 46), and NCAM1 (encoding CD56) (23–32). Furthermore, editing of immune checkpoint regulators, including CISH (cytokine-inducible SH2-containing protein), PD1 (programmed cell death protein 1), TIGIT (T cell immunoreceptor with Ig and ITIM domains), and A2AR (adenosine A2A receptor), has been shown to restore anti-tumor activity and improve in vivo persistence (33–37). To enhance signaling downstream of activating receptors, genes such as SYK (spleen tyrosine kinase), CD247 (encoding CD3 zeta chain), FCGR1A (Fc receptor gamma chain), and CBLB (Casitas B-lineage lymphoma proto-oncogene B) have been modified to amplify NK cell activation and cytokine production (36, 38). In addition, metabolic and cell cycle regulators such as TGFBR2 (Transforming Growth Factor Beta Receptor II) and HPRT1 (Hypoxanthine Phosphoribosyltransferase 1) have been targeted to enhance NK cell function and fitness. Finally, CRISPR-based genetic enhancement of targeting specificity and tracking has been achieved recently also through the knock-in of CAR constructs and reporter genes (e.g., CD19, CD33, GD2, EGFP, mCherry) (23, 33, 39–43).
Overall and most importantly, the limited in vivo persistence and proliferative capacity of NK cells can reduce the durability of functional NK cell responses for the patients (15). To address this, the field introduced NK cells specific cytokine support elements such as IL-15 to improve survival and expansion kinetics (15, 16, 37, 44–48). Almost all of the recent CAR-NK cells studies include IL-15 elements and confirmed an improved response by longer persisting CAR-NK cells without cytokine related toxicity (15, 16). However, other pre-clinical studies with genetically extended in vivo half-life describe systemic adverse effects in vivo in xenograft models by IL-15-armored NK cells (37, 49, 50).
Even though CRISPR/Cas is a very efficient and precise gene editing tool it can lead to DNA cuts at sites other than the intended target site, primarily when the mismatch-tolerating CRISPR/Cas complex mediates binding to DNA sequences that resemble the target site closely. Furthermore, unintended structural variants, such as large on-target site deletions, chromosomal rearrangements, or chromosomal instability have been reported (51–53). Such unintended editing outcomes can – in principle – disrupt vital genes, activate/inactivate undesired genes, cause genomic instability or even oncogenic transformation, ultimately endangering patient safety (52–58). To address some of these limitations, new CRISPR/Cas types like base editors (BE) and prime editors (PE) open new potential for introducing point-mutations and thereby gain- or loss-of-function without depending on DSBs or on an HDR template for precise repair (59). However, the application in NK cells is still limited (35, 37). One study recently reported the first CAR transgene integration in combination with multiplex BE (using up to six guides) in primary NK cells (37). The multiplex edited IL15-expressing CAR-NK cells showed enhanced in−vitro potency. However, although the xenograft mice models confirmed an increased antitumor activity, they also revealed major organ toxicity in the multiplex edited CAR-IL15 NK cell treatment group, limiting the overall survival (37, 60). To assess genomic safety, the authors analyzed editing at the top computationally predicted OT sites using rhAmpSeq (RNase H2-dependent multiplexed PCR Sequencing). While they did not identify OT effects and negligible rates of translocations between on-target sites using this approach, it should be noted that only genome-wide techniques ensure sensitive detection of OT activity and therefore safe clinical translation (52, 61, 62).
In the last years several detection techniques have been developed to profile and mitigate unintended edits, so minimizing OT activity in therapeutic genome editing (63, 64). These include in silico prediction tools, in vitro biochemical tests, and cell-based methods; each of them have their unique benefits and drawbacks (Figure 1). From a translational point of view, it is important to note that regulatory agencies increasingly require orthogonal OT validation, i.e. a combination of in silico predictions, in vitro cleavage assays, and cell-based detection platforms to ensure comprehensive detection of potential OT events (65, 66). The most suitable combination of assays may depend on the cell type, the editing tool, and the intended use of the cell therapy. In particular for multiplex editing strategies in NK cells (19), an urgent need for rigorous and comprehensive safety analysis is necessary to not hurting NK cell therapy’s impeccable safety records in the early clinical trials.
Figure 1. Off-target detection strategies across the CRISPR-NK cell editing workflow. Illustration of key stages in NK cell genome editing and corresponding safety assessment tools. In silico tools (e.g., CRISPOR, DeepCRISPR) assist with gRNA design before editing. In vitro digestion-based methods (e.g., CIRCLE-seq, SITE-seq, Digenome-seq) find cleavage sites in cell-free genomic DNA. Cell-based assay (e.g., GUIDE-seq, DISCOVER-seq) detects DSB-associated off-targets under chromatin context. Structural variants are captured post-expansion using CAST-seq and validated by high-throughput sequencing (HTS). In vivo post-infusion studies assess clonal expansion and tumorigenesis. Created with Biorender.com. AI, Artificial intelligence; DSB, Double-strand break; gRNA, Guide RNA; HTS, High-throughput sequencing; RNP, Ribonucleoprotein; SNV, Single-nucleotide variant; PDX, Patient-derived xenograft model.
In silico prediction tools
In silico tools serve as an essential initial step by computationally predicting potential OT sites based on gRNA sequence similarity across the genome (Figure 1). Traditional tools such as CRISPOR, CCTop (CRISPR/Cas9 Target online predictor), and COSMID (CRISPR OT Sites with Mismatches, Insertions, and Deletions) use sequence alignment and mismatch tolerance models derived from Cas9 activity data to generate ranked lists of potential OT sites (67–70). More recently, deep learning–based tools such as DeepCRISPR, DeepCBE, DeepABE and DeepBE have been developed to predict the efficiencies of BEs and their OT propensities (71, 72). These tools are especially useful in the early design phase of CRISPR editing experiments. Although in silico tools are fast and cost-efficient, their predictions require empirical validation as chromatin accessibility and DNA repair pathway choice have a substantial impact on editing outcomes that cannot be predicted yet (67, 71).
Before performing CRISPR/Cas9 editing in primary NK cells or NK-92 lines, authors in a number of NK cell CRISPR studies first designed gRNAs using in silico tools like Benchling, giving priority to high predicted on-target scores and minimal OT potential (23, 25, 33, 36). In the past years, some studies used next-generation high-throughput sequencing (HTS) to confirm a subset of in silico predicted OTs, while others reported no OT analysis at all (Table 1), possibly as a result of financial constraints, technological limitations, or presumptions regarding the CRISPR/Cas9 system’s high specificity (30, 40, 73, 74). The absence of such analysis raises concerns about the completeness of safety assessments, especially in therapeutic settings.
Table 1. Off-target detection methods applied in CRISPR-edited NK cells.
In vitro detection assays
For in vitro detection assays, purified genomic DNA is treated with nucleases such as CRISPR/Cas9 in a cell-free environment using biochemical techniques like CIRCLE-Seq, SITE-Seq, CHANGE-Seq and Digenome-Seq (75–78). HTS is then used to map cleavage sites (Figure 1). While disregarding chromatin context or cellular repair, these assays offer objective, sensitive mapping of DNA breaks (75–77). For instance, OT sites with high or low cleavage frequencies can be found using SITE-Seq (Selective Enrichment and Identification of Tagged Genomic DNA Ends by Sequencing), which works by tagging DSBs with biotinylated adapters, enriching these fragments, and mapping them via high-throughput sequencing (76). CIRCLE-Seq (Circularization for in vitro reporting of cleavage effects by sequencing) and its successor CHANGE-seq (Circularization for High-Throughput Analysis of Nuclease Genome-wide Effects by sequencing) use circularized fragments of purified genomic DNA, which are then incubated with CRISPR/Cas9 to identify cleaved sites (75, 78, 79). Digenome-Seq uses whole-genome sequencing to identify cleavage sites based on sequence break points after treating naked genomic DNA with Cas9 in vitro (80).
However, because these assays are conducted outside of the cellular environment, they might cut DNA sequences where chromatin compaction or epigenetic regulation prevent Cas9 from cleaving in cells (75, 81), often leading to an overestimation of OT sites. The sensitivity and specificity of any of these assays have not yet been evaluated in the context of NK cells, despite the fact that they are frequently utilized for in vitro OT profiling (Table 1).
Cell-based detection approaches
Cell-based methods such as GUIDE-seq and DISCOVER-seq detect OT cleavage sites within living cells, offering valuable insight into genome editing specificity under physiologically relevant conditions (Figure 1) (80, 84–86), especially if performed in the clinically relevant cell type. GUIDE-seq (Genome-wide Unbiased Identification of DSBs Enabled by Sequencing) introduces double-stranded oligodeoxynucleotides (dsODNs) along with the CRISPR/Cas9 nucleases. As these dsODNs will be incorporated at DNA double-strand breaks (DSBs), GUIDE-seq allows for genome-wide mapping of Cas9-induced OT sites (80). However, dsODNs can be toxic to primary cells, including NK cells, which makes GUIDE-seq less useful in cell types that are important for clinical use. DISCOVER-seq (Discovery of In Situ Cas OTs and Verification by End-Ligation and Repair Sequencing) is based on chromatin immunoprecipitation followed by sequencing (ChIP-seq) targeted to the DNA repair factor MRE11, which is rapidly recruited to DSBs. With the help of this technique, Cas9-induced DNA breaks in living cells can be tracked genome-wide, offering dynamic insights into editing activity in a chromatin-dependent setting (84). While this method offers chromatin-aware OT detection, its application in NK cells remains limited. One publication demonstrating the application of cell-based detection assay for primary NK cells comes from a study targeting the CD38 locus (55). In order to find potential OT sites, GUIDE-seq was first performed in HEK293 cells using a CD38-targeted gRNA. These nominated sites were then evaluated in CRISPR/Cas9-edited primary NK cells using rhAmpSeq (82). The resulting data revealed a high on-target editing efficiency and no detectable OT editing (82).
As summarized in Table 1, most studies that applied GUIDE-seq in the context of NK cell engineering, including the example discussed above, performed the initial GUIDE-seq screen in surrogate HEK293 cell lines, only the nominated sites were validated in primary NK cells (34, 82). This leaves open the possibility that additional OT events unique to NK cells may have gone undetected. Therefore, for clinical applications, safety evaluations of NK cell-based therapies must be carried out directly in NK cells rather than extrapolated from surrogate cell lines because OT profiles are impacted by cell-type-specific factors like chromatin accessibility, DNA repair mechanisms, and delivery efficiency (87).
Apart from cleavage-based tests, CAST-Seq (Chromosomal Aberration analysis by Single Targeted linker-mediated PCR followed by sequencing) makes it possible to identify structural changes like translocations and large deletions and inversions after CRISPR/Cas9 editing, especially in therapeutic applications where chromosomal stability is crucial (88–93). Notably, CAST-Seq can also nominate potential OT sites that can be further validated by HTS, a strategy that has been employed in several recent studies (88–90, 94). Another advantage of this method is, in contrast to many other genome-wide assays, its compatibility with low cell input, making it especially useful in settings where only limited numbers of edited cells or genomic DNA are available, such as for primary NK cells in early-stage clinical manufacturing or cells isolated from patients or preclinical animal models (Figure 1). Despite not having been applied to NK cells in published studies yet, CAST-Seq is a promising tool for evaluating genomic integrity in next-generation NK-based therapies (Table 1).
As mentioned above, while researchers have applied various OT detection methods in edited (CAR) T cells and hematopoietic stem/progenitor cells (HSPCs) (88, 95–97). They have been applied only to a limited extent to NK cells (Table 1). The application of comprehensive, genome-wide OT detection techniques to NK cells is however vital in order to assure the safe clinical application of this cell type. Given the therapeutic potential of genome-edited NK cells, particularly as readily accessible off-the-shelf cell therapy, applying next-generation OT detection platforms to primary NK cells is imperative (98). Combining optimized in silico prediction pipelines with empirical, genome-wide methods in a standardized processes will ultimately improve the safety profile of this emerging therapy (99).
Pre-clinical in vivo models for immunotoxicity
An additional dimension for pre-clinical safety evaluation of engineered NK cells lies in the appropriate choice and use of animal or alternative in vivo or in vivo-like models, including zebrafish xenografts, chorioallantois membrane (CAM) assays, and micro physiological organ-on-chip systems.
In addition to classical animal and humanized mouse models, several alternative in vivo-like systems are emerging that could expand the safety-testing toolbox for engineered NK cells. The CAM assay, long used for tumor xenografts and angiogenesis studies, provides a rapid and cost-efficient intermediate model between in vitro and murine experiments. While it has already been applied for CAR T-cell efficacy testing, its use with NK cells remains largely unexplored and may offer a convenient platform for short-term assessment of cytotoxicity and tissue interactions (100). Similarly, organ-on-chip and 3D perfused tissue constructs recreate vascularized, microfluidic environments that enable real-time monitoring of NK-cell migration, infiltration, and tumor killing under near-physiological flow conditions. These systems can help dissect cell-intrinsic behavior and immune-microenvironment interactions before moving to full animal studies (101). To limited extent, alternative in vivo systems such as zebrafish larval xenografts are used as rapid, cost-effective platforms to visualize NK-cell cytotoxicity and tissue interactions before advancing to murine studies (102). These in vivo platforms can permit assessment of persistence, biodistribution, cytokine release, off-target toxicity and immune-host relationships in conditions closer to the human situation, yet they also have limitations (103, 104).
So far human-cell derived NK products are typically tested in immunodeficient mouse (or humanized mouse) xenograft models, allowing evaluation of both anti-tumor efficacy and safety (e.g., in murine lymphoma or solid tumor models) (Figure 1) (103, 105). In this regard, one study reported about the cause of severe toxicity in glioblastoma xenograft mouse models after different dose treatments with NK cells engineered to express IL-15 (weight loss and extensive NK cells infiltration) (49). Interestingly, it was not observed with transgenic IL-21 in the same in vivo model (49). Another study reported that armoring NK cells with secreted IL-15 enhanced the anti-AML functionality in vitro significantly, but in the in vivo model, the constitutive IL-15 expressing NK cells showed improved persistency but also revealed unexpected lethal toxicity (50). These highlights the need for such rigors pre-clinical in vivo platforms. However, translation from in vivo findings to humans also shows limitations, as to date none of the ongoing clinical trials evaluating CD19-CAR NK cells co-expressing cytokines like IL-15 have reported such adverse effects (15, 16). Currently, xenograft mouse models are among the most well-studied models, but they also not fully recapitulate the human immune microenvironment and do not reflect the entire human immunology complexity (106, 107).
Therefore, standardized reporting of safety endpoints (dose, persistence, cytokine/chemokine panels, histopathology of host organs, biodistribution) and use of more physiologically relevant models (e.g., humanized immune system mice, canine models) are thus recommended (103). Incorporating these pre-clinical in vivo and in vivo-like platforms into the safety assessment pipeline for NK cell therapies could therefore reduce risk, improve translational predictiveness, and ultimately increase the likelihood of clinical success (Figure 1).
Clinical application of CRISPR-edited NK cells
First genome-edited NK cell therapies are now in clinical or advanced preclinical development. Their current goals are to make them better at fighting tumors, lasting longer, and if quipped with CARs, also to avoid them from targeting themselves (fratricide). In this regard, FT538 (Fate Therapeutics) is one of the most advanced candidates. The iPSC-derived NK cell product is based on multiplexed genome editing using CRISPR/Cas9. It includes knocking out CD38 to stop fratricide and adding a high-affinity, non-cleavable CD16 receptor (hnCD16) and an IL-15 receptor fusion (IL15RF) to help improve their in vivo persistence and boost their specific antibody-dependent cellular cytotoxicity (ADCC) (108). Although they entered first clinical trials, there was no detailed genome-wide OT profiling reported. However, by choosing single-cell iPSC clones based on karyotype analysis and targeted sequencing to make sure that there were no unintended genomic changes or integrations the risk of OT effects was lowered (108). Another candidate, FT596 that expresses a CD19-targeted CAR to eliminate hematologic malignancies, is also made from genetically modified iPSCs harboring hnCD16 and IL15RF (109). Interestingly, there is no public reports of OT analysis methods for this product so far (109).
Pitfalls of today’s safety assessments: homologous receptor gene families and chromosomal instability
Editing at loci with high sequence homology to pseudogenes or paralogs can lead to misinterpretation of on-target activity and an underestimation of OT risks. An example is the locus coding for neutrophil cytosolic factor 1 (NCF1), a gene linked to chronic granulomatous disease. NCF1 is flanked by two closely related pseudogenes with nearly identical sequences. Not surprisingly, it has been shown that CRISPR-editing of NCF1 affects these pseudogenes, leading to large chromosomal rearrangements spanning several megabases (93). Furthermore, OT effects can be mistaken for successful on-target editing unless validated with allele-specific deep sequencing approaches (63). These events not only obscure functional interpretation but also pose a risk for chromosomal stability. Similar challenges occur when editing highly homologous NK cell receptor gene families, such as the NKG2 familiy (KLRC1–4), killer-cell immunoglobulin-like receptor members (KIR2DS, KIR3DS, KIR2DL, KIR3DL), natural cytotoxicity receptors (NCR1–3), SIGLEC, C-type lectin-like receptors (KLRK1, KLRD1, KLRI1/2), and leukocyte immunoglobulin-like receptors (LILRB1, LILRB2). Most of these gene families encode both activating and inhibitory receptors that share >85% sequence similarity in key exons but have distinct roles in NK cell regulation. Several members, including KLRC1/NKG2A, KLRK1/NKG2D, KIR2DL, NCR1, SIGLEC7, SIGLEC9, and LILRB1, have been targeted with CRISPR/Cas9 in primary NK cells or NK-92 lines to enhance cytotoxicity or study receptor function (11, 24, 29–32, 110). While one NKG2A study, using deep sequencing, reported minimal OT activity in CRISPR-modified knock-in NK cells integrating a GD2-CAR into the KLRC1 locus, most other reports did not assess off-target effects, leaving open the possibility of inadvertently altering related family members, with potential consequences for NK cell phenotype and function (42). In this regard, CAST-Seq provides a valuable strategy to detect structural variants such as large deletions or translocations that may arise from editing genes in gene clusters, providing an additional safety layer for clinical-grade genome editing (93).
While unintentional cuts in oncogenes or tumor suppressors pose a well-known safety concern (52), chromosomal translocations between loci that are targeted at the same time (111), as well as the disruption of noncoding regulatory regions, such as enhancers and long non-coding RNAs (lncRNAs), may pose additional safety risks and could disrupt important immune functions without even changing the coding sequence (112). Because of these vulnerabilities, safety tests for NK cell editing should not only look for point mutations or small insertions/deletions, but must use methods to identify complex genomic aberrations, and take into account cell type-specific factors that affect the frequency, the nature and the consequences of OT editing outcomes (113).
Future perspectives of today’s safety assessments
The rapid evolution of new editing technologies like BE offers an advantage over traditional CRISPR/Cas9 for multiplex gene editing, as they avoid the simultaneous induction of multiple DSBs and could advance precise multiplex editing of different genes (37, 114). But it also challenges the currently established tools for safety assessments as they mostly rely on the detection of unintended OT effects by DSBs (114). Unlike nuclease-based editing, BEs can introduce OT effects through multiple mechanisms, including gRNA-dependent DNA editing, gRNA-independent deamination of nucleotides, and OT RNA editing (115). Available assays such as Digenome-Seq and EndoV-Seq can detect base editor off-targets, but their reliance on whole-genome sequencing (WGS) (30–40x read depth) limits sensitivity to events above ~2–3% (116–119). Currently, safety assessments for base editing OT analysis include WGS with deep coverage to identify unintended single-nucleotide variants, as well as targeted deep sequencing of in silico predicted OT sites (Figure 1).
Seeking to identify further cell specific alternatives, RNA sequencing (RNA-seq) has been employed to detect OT RNA edits caused by some BEs. However, this approach has limited sensitivity for low-frequency edits and requires high read depth to detect subtle transcriptome-wide changes. To our knowledge they have not be applied in NK cells yet (120).
Finally, the recent advances in artificial intelligence (AI) and machine learning (ML) have the potential also to be implemented as forecast tools for advanced gene editing tools (121). Currently, they are most frequently used to improve gRNA design and OT predictions by scoring algorithms and ML models and recently as CRISPR-GPT, as an agentic for automation of gene editing experiments and data analaysis (121, 122). Today, one limitation for AI and ML models for OT detection, is the limited availability and small size of datasets for training validated deep learning models. However, the recent developments in the field will potentially overcome such limitations and can make AI and ML a promising tool for prediction of OT and chromosomal changes in the future.
Overall, the increasing frequency of CRISPR-editing in primary cells and its promising application for precisely designing and enhancing the inherent killing capacity of NK cells highlight the need for rigorous and comprehensive safety assessments. In addition to molecular and computational approaches, in vivo studies are gaining importance for providing a more complete picture of safety. They allow researchers to observe the persistence and distribution of engineered NK cells, as well as possible cytokine-related or off-target effects under more physiological conditions. Combining these in vivo insights with genomic analyses and the development of artificial prediction models, it can provide a clear comprehensive and precise framework for assessing the next-generation of safe off-the-shelf NK-cell treatments. Although the use of such tools is still scarce in NK cells, there is a strong need to apply and further develop these approaches in this emerging immune cell type to avoid compromising the highly promising safety profile of these cells. The parallel developments of multiplex editing and rigorous safety assessments tools will pave the way for the strong clinical translation of safe and effective NK cell products in the future.
Author contributions
AF: Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization. EU: Writing – review & editing. TC: Supervision, Validation, Writing – review & editing, Conceptualization. TB: Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – review & editing, Writing – original draft.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
TB and EU are inventors on patent applications for genetic editing and cell therapies including non-viral CAR-NK cells approaches. TC is a consultant for Aavigen, AstraZeneca, Excision BioTherapeutics, and T-knife Therapeutics, as well as is an inventor of CAST-Seq US11319580B2. EU has a sponsored research project with Gilead and BMS and acts as medical advisor of Phialogics and CRIION. A.F. declares no competing interests. AF is a recipient of a PhD fellowship of the German Academic Exchange Service DAAD fellowship grant no. 91861217. EU and TC have received support by the Stiftung Deutsche Krebshilfe German Cancer Aid preCDD program/”CAR FACTORY” #70115200. EU, TC and TB have received funding from the German Federal Ministry of Research, Technology and Space BMFTR in the context of the German National Strategy for Gene and Cell-Based Therapies neClectAML–01BIHTP2406 and AzaCAR4AML–01BIHTP2513 to EU and TC; CRISPoptiCARNK–01BIHTP2503 to TB and EU; BEACON-MM–01BIHTP2520 to TB. TB was funded by the Deutsche Forschungsgemeinschaft DFG, German Reserach Foundation - 493624332 INDEEP Clinician Scientist Program and by the MSNZ Frankfurt German Cancer Aid, Funding ID: 70113301.
The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.
Generative AI statement
The author(s) declare that Generative AI was used in the creation of this manuscript. Generative AI was used exclusively for language processing.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, and Charpentier E. A programmable dual-RNA–guided DNA endonuclease in adaptive bacterial immunity. science. (2012) 337:816–21. doi: 10.1126/science.1225829
2. Gasiunas G, Barrangou R, Horvath P, and Siksnys V. Cas9–crRNA ribonucleoprotein complex mediates specific DNA cleavage for adaptive immunity in bacteria. Proc Natl Acad Sci. (2012) 109:E2579–E86. doi: 10.1073/pnas.1208507109
3. Doudna JA. The promise and challenge of therapeutic genome editing. Nature. (2020) 578:229–36. doi: 10.1038/s41586-020-1978-5
4. Lu Y, Xue J, Deng T, Zhou X, Yu K, Deng L, et al. Safety and feasibility of CRISPR-edited T cells in patients with refractory non-small-cell lung cancer. Nat Med. (2020) 26:732–40. doi: 10.1038/s41591-020-0840-5
5. Frangoul H, Altshuler D, Cappellini MD, Chen Y-S, Domm J, Eustace BK, et al. CRISPR-Cas9 gene editing for sickle cell disease and β-thalassemia. New Engl J Med. (2021) 384:252–60. doi: 10.1056/NEJMoa2031054
6. Fu B, Liao J, Chen S, Li W, Wang Q, Hu J, et al. CRISPR–Cas9-mediated gene editing of the BCL11A enhancer for pediatric β0/β0 transfusion-dependent β-thalassemia. Nat Med. (2022) 28:1573–80. doi: 10.1038/s41591-022-01906-z
7. Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. (2013) 339:819–23. doi: 10.1126/science.1231143
8. Mali P, Yang L, Esvelt KM, Aach J, Guell M, DiCarlo JE, et al. RNA-guided human genome engineering via Cas9. Science. (2013) 339:823–6. doi: 10.1126/science.1232033
9. Cho SW, Kim S, Kim JM, and Kim J-S. Targeted genome engineering in human cells with the Cas9 RNA-guided endonuclease. Nat Biotechnol. (2013) 31:230–2. doi: 10.1038/nbt.2507
10. Jinek M, East A, Cheng A, Lin S, Ma E, and Doudna J. RNA-programmed genome editing in human cells. elife. (2013) 2:e00471. doi: 10.7554/eLife.00471
11. Vivier E, Rebuffet L, Narni-Mancinelli E, Cornen S, Igarashi RY, and Fantin VR. Natural killer cell therapies. Nature. (2024) 626:727–36. doi: 10.1038/s41586-023-06945-1
12. Laskowski TJ, Biederstädt A, and Rezvani K. Natural killer cells in antitumour adoptive cell immunotherapy. Nat Rev Cancer. (2022) 22:557–75. doi: 10.1038/s41568-022-00491-0
13. Lanier LL. Five decades of natural killer cell discovery. J Exp Med. (2024) 221:e20231222. doi: 10.1084/jem.20231222
14. Wendel P, Reindl LM, Bexte T, Künnemeyer L, Särchen V, Albinger N, et al. Arming immune cells for battle: a brief journey through the advancements of T and NK cell immunotherapy. Cancers. (2021) 13:1481. doi: 10.3390/cancers13061481
15. Marin D, Li Y, Basar R, Rafei H, Daher M, Dou J, et al. Safety, efficacy and determinants of response of allogeneic CD19-specific CAR-NK cells in CD19+ B cell tumors: a phase 1/2 trial. Nat Med. (2024) 30:772–84. doi: 10.1038/s41591-023-02785-8
16. Liu E, Marin D, Banerjee P, Macapinlac HA, Thompson P, Basar R, et al. Use of CAR-transduced natural killer cells in CD19-positive lymphoid tumors. New Engl J Med. (2020) 382:545–53. doi: 10.1056/NEJMoa1910607
17. Biederstädt A and Rezvani K. Engineering the next generation of CAR-NK immunotherapies. Int J hematol. (2021) 114:554–71. doi: 10.1007/s12185-021-03209-4
18. Jørgensen LV, Christensen EB, Barnkob MB, and Barington T. The clinical landscape of CAR NK cells. Exp Hematol Oncol. (2025) 14:46. doi: 10.1186/s40164-025-00633-8
19. Biederstädt A, Manzar GS, and Daher M. Multiplexed engineering and precision gene editing in cellular immunotherapy. Front Immunol. (2022) 13:1063303. doi: 10.3389/fimmu.2022.1063303
20. Reindl LM, Albinger N, Bexte T, Müller S, Hartmann J, and Ullrich E. Immunotherapy with NK cells: recent developments in gene modification open up new avenues. Oncoimmunology. (2020) 9:1777651. doi: 10.1080/2162402X.2020.1777651
21. Park JD, Shin HE, An YS, Jang HJ, Park J, Kim S-N, et al. Advancing natural killer cell therapy: genetic engineering strategies for enhanced cancer immunotherapy. Ann Lab Med. (2025) 45:146–59. doi: 10.3343/alm.2024.0380
22. Bexte T, Reindl LM, and Ullrich E. Nonviral technologies can pave the way for CAR-NK cell therapy. J Leukocyte Biol. (2023) 114:475–86. doi: 10.1093/jleuko/qiad074
23. Huang R-S, Shih H-A, Lai M-C, Chang Y-J, and Lin S. Enhanced NK-92 cytotoxicity by CRISPR genome engineering using Cas9 ribonucleoproteins. Front Immunol. (2020) 11:1008. doi: 10.3389/fimmu.2020.01008
24. Bexte T, Alzubi J, Reindl LM, Wendel P, Schubert R, Salzmann-Manrique E, et al. CRISPR-Cas9 based gene editing of the immune checkpoint NKG2A enhances NK cell mediated cytotoxicity against multiple myeloma. Oncoimmunology. (2022) 11:2081415. doi: 10.1080/2162402X.2022.2081415
25. Huang R-S, Lai M-C, Shih H-A, and Lin S. A robust platform for expansion and genome editing of primary human natural killer cells. J Exp Med. (2021) 218:e20201529. doi: 10.1084/jem.20201529
26. Rautela J, Surgenor E, and Huntington ND. Efficient genome editing of human natural killer cells by CRISPR RNP. BioRxiv. (2018), 406934. doi: 10.1101/406934
27. Berrien-Elliott MM, Pamela W, Neal C, Wagner JA, Becker-Hapak M, Schappe T, et al. Primary human NK cell gene-editing reveals a critical role for NKG2A in cytokine-induced memory-like NK cell responses. Blood. (2019) 134:3237. doi: 10.1182/blood-2019-129162
28. Peng L, Renauer PA, Sferruzza G, Yang L, Zou Y, Fang Z, et al. In vivo AAV–SB-CRISPR screens of tumor-infiltrating primary NK cells identify genetic checkpoints of CAR-NK therapy. Nat Biotechnol. (2024) 43:752–761. doi: 10.1038/s41587-024-02282-4
29. Alves E, McLeish E, Blancafort P, Coudert JD, and Gaudieri S. Manipulating the NKG2D receptor-ligand axis using CRISPR: novel technologies for improved host immunity. Front Immunol. (2021) 12:712722. doi: 10.3389/fimmu.2021.712722
30. Rautela J, Surgenor E, and Huntington ND. Drug target validation in primary human natural killer cells using CRISPR RNP. J Leucocyte Biol. (2020) 108:1397–408. doi: 10.1002/JLB.2MA0620-074R
31. Dick J, Zandstra B, Sangala AJ, Crompton P, and Hart G. Loss of Siglec-7 drives enhanced NK cell function and correlates with protection from malaria symptoms. J Immunol. (2024) 212:1271_4180–1271_4180. doi: 10.4049/jimmunol.212.supp.1271.4180
32. Daly J, Sarkar S, Natoni A, Stark JC, Riley NM, Bertozzi CR, et al. Targeting hypersialylation in multiple myeloma represents a novel approach to enhance NK cell–mediated tumor responses. Blood Advances. (2022) 6:3352–66. doi: 10.1182/bloodadvances.2021006805
33. Pomeroy EJ, Hunzeker JT, Kluesner MG, Lahr WS, Smeester BA, Crosby MR, et al. A genetically engineered primary human natural killer cell platform for cancer immunotherapy. Mol Ther. (2020) 28:52–63. doi: 10.1016/j.ymthe.2019.10.009
34. Daher M, Basar R, Gokdemir E, Baran N, Uprety N, Nunez Cortes AK, et al. Targeting a cytokine checkpoint enhances the fitness of armored cord blood CAR-NK cells. Blood J Am Soc Hematol. (2021) 137:624–36. doi: 10.1182/blood.2020007748
35. Ying H, Yao C, Fang C, Li G, Wang L, Zhou Y, et al. AccuBaseTM (ceBE), a second-generation high-fidelity base editor, can proficiently modify genes in a human primary NK cells with minimizing off-target effect. Cancer Res. (2024) 84:1328–. doi: 10.1158/1538-7445.AM2024-1328
36. Ureña-Bailén G, Dobrowolski J-M, Hou Y, Dirlam A, Roig-Merino A, Schleicher S, et al. Preclinical evaluation of CRISPR-edited CAR-NK-92 cells for off-the-shelf treatment of AML and B-ALL. Int J Mol Sci. (2022) 23:12828. doi: 10.3390/ijms232112828
37. Wang M, Krueger JB, Gilkey AK, Stelljes EM, Kluesner MG, Pomeroy EJ, et al. Precision enhancement of CAR-NK cells through non-viral engineering and highly multiplexed base editing. J ImmunoTher Cancer. (2025) 13:e009560. doi: 10.1136/jitc-2024-009560
38. Dahlvang JD, Dick JK, Sangala JA, Kennedy PR, Pomeroy EJ, Snyder KM, et al. Ablation of SYK kinase from expanded primary human NK cells via CRISPR/Cas9 enhances cytotoxicity and cytokine production. J Immunol. (2023) 210:1108–22. doi: 10.4049/jimmunol.2200488
39. Kararoudi MN, Dolatshad H, Trikha P, Hussain S-RA, Elmas E, Foltz JA, et al. Generation of knock-out primary and expanded human NK cells using Cas9 ribonucleoproteins. J visualized experiments: JoVE. (2018) 136):58237. doi: 10.3791/58237
40. Borges CM, Wasko K, Nasser JM, Donahue K, Pfautz A, Antony LP, et al. Preclinical development of edit-201, a multigene edited healthy donor nk cell with enhanced anti-tumor function and superior serial killing activity in an immunosuppressive environment. Blood. (2020) 136:33. doi: 10.1182/blood-2020-139988
41. Kath J, Franke C, Drosdek V, Du W, Glaser V, Fuster-Garcia C, et al. Integration of ζ-deficient CARs into the CD3 ζ gene conveys potent cytotoxicity in T and NK cells. Blood. (2024) 143:2599–611. doi: 10.1182/blood.2023020973
42. Shankar K, Zingler-Hoslet I, Tabima DM, Zima S, Shi L, Gimse K, et al. Virus-free CRISPR knockin of a chimeric antigen receptor into KLRC1 generates potent GD2-specific natural killer cells. Mol Ther. (2025) 33:1014–30. doi: 10.1016/j.ymthe.2025.01.024
43. Bexte T, Albinger N, Al Ajami A, Wendel P, Buchinger L, Gessner A, et al. CRISPR/Cas9 editing of NKG2A improves the efficacy of primary CD33-directed chimeric antigen receptor natural killer cells. Nat Commun. (2024) 15:8439. doi: 10.1038/s41467-024-52388-1
44. Ma S, Caligiuri MA, and Yu J. Harnessing IL-15 signaling to potentiate NK cell-mediated cancer immunotherapy. Trends Immunol. (2022) 43:833–47. doi: 10.1016/j.it.2022.08.004
45. Liu E, Tong Y, Dotti G, Shaim H, Savoldo B, Mukherjee M, et al. Cord blood NK cells engineered to express IL-15 and a CD19-targeted CAR show long-term persistence and potent antitumor activity. Leukemia. (2018) 32:520–31. doi: 10.1038/leu.2017.226
46. Rafei H, Basar R, Acharya S, Hsu Y-S, Liu P, Zhang D, et al. CREM is a regulatory checkpoint of CAR and IL-15 signalling in NK cells. Nature. (2025) 643:1076–1086. doi: 10.1038/s41586-025-09087-8
47. Knisely A, Rafei H, Basar R, Banerjee PP, Metwalli Z, Lito K, et al. Phase I/II study of TROP2 CAR engineered IL15-transduced cord blood-derived NK cells delivered intraperitoneally for the management of platinum resistant ovarian cancer, mesonephric-like adenocarcinoma, and pancreatic cancer. Am Soc Clin Oncol. (2024) 42:16_suppl. doi: 10.1200/JCO.2024.42.16_suppl.TPS5626
48. Li L, Mohanty V, Dou J, Huang Y, Banerjee PP, Miao Q, et al. Loss of metabolic fitness drives tumor resistance after CAR-NK cell therapy and can be overcome by cytokine engineering. Sci Adv. (2023) 9:eadd6997. doi: 10.1126/sciadv.add6997
49. Shanley M, Daher M, Dou J, Li S, Basar R, Rafei H, et al. Interleukin-21 engineering enhances NK cell activity against glioblastoma via CEBPD. Cancer Cell. (2024) 42:1450–66.e11. doi: 10.1016/j.ccell.2024.07.007
50. Christodoulou I, Ho WJ, Marple A, Ravich JW, Tam A, Rahnama R, et al. Engineering CAR-NK cells to secrete IL-15 sustains their anti-AML functionality but is associated with systemic toxicities. J immunother cancer. (2021) 9:e003894. doi: 10.1136/jitc-2021-003894
51. Kim D, Kim J, Hur JK, Been KW, Yoon S-h, and Kim J-S. Genome-wide analysis reveals specificities of Cpf1 endonucleases in human cells. Nat Biotechnol. (2016) 34:863–8. doi: 10.1038/nbt.3609
52. Kosicki M, Tomberg K, and Bradley A. Repair of double-strand breaks induced by CRISPR–Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. (2018) 36:765–71. doi: 10.1038/nbt.4192
53. Cullot G, Boutin J, Toutain J, Prat F, Pennamen P, Rooryck C, et al. CRISPR-Cas9 genome editing induces megabase-scale chromosomal truncations. Nat Commun. (2019) 10:1136. doi: 10.1038/s41467-019-09006-2
54. Fu Y, Foden JA, Khayter C, Maeder ML, Reyon D, Joung JK, et al. High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat Biotechnol. (2013) 31:822–6. doi: 10.1038/nbt.2623
55. Liang P, Xu Y, Zhang X, Ding C, Huang R, Zhang Z, et al. CRISPR/Cas9-mediated gene editing in human tripronuclear zygotes. Protein Cell. (2015) 6:363–72. doi: 10.1007/s13238-015-0153-5
56. Haapaniemi E, Botla S, Persson J, Schmierer B, and Taipale J. CRISPR–Cas9 genome editing induces a p53-mediated DNA damage response. Nat Med. (2018) 24:927–30. doi: 10.1038/s41591-018-0049-z
57. Ihry RJ, Worringer KA, Salick MR, Frias E, Ho D, Theriault K, et al. p53 inhibits CRISPR–Cas9 engineering in human pluripotent stem cells. Nat Med. (2018) 24:939–46. doi: 10.1038/s41591-018-0050-6
58. Zhang X-H, Tee LY, Wang X-G, Huang Q-S, and Yang S-H. Off-target effects in CRISPR/Cas9-mediated genome engineering. Mol Ther Nucleic Acids. (2015) 4:e264. doi: 10.1038/mtna.2015.37
59. Anzalone AV, Koblan LW, and Liu DR. Genome editing with CRISPR–Cas nucleases, base editors, transposases and prime editors. Nat Biotechnol. (2020) 38:824–44. doi: 10.1038/s41587-020-0561-9
60. Bexte T and Wagner DL. Next-generation multiplex-edited CAR-NK cells: more edits, more power? J Immunother Cancer. (2025) 13:e012841. doi: 10.1136/jitc-2025-012841
61. Schubert M, Rettig GR, Kurgan G, McNeill MS, Wang J, Fiedler S, et al. Evaluate CRISPR-Cas9 edits quickly and accurately with rhAmpSeq targeted sequencing. (2024).
62. Turk R. CRISPR Applications in Medicine Depend on Minimizing Off-Target Editing: Integrated DNA Technologies describes methods that can be used to quantify and characterize off-target editing in gene editing experiments. Genet Eng Biotechnol News. (2021) 41:64–6. doi: 10.1089/gen.41.09.24
63. Tycko J, Myer VE, and Hsu PD. Methods for optimizing CRISPR-Cas9 genome editing specificity. Mol Cell. (2016) 63:355–70. doi: 10.1016/j.molcel.2016.07.004
64. Huang TP, Newby GA, and Liu DR. Publisher Correction: Precision genome editing using cytosine and adenine base editors in mammalian cells. Nat Protoc. (2021) 16:5740–. doi: 10.1038/s41596-021-00525-1
65. Harris A and Gilbert F. Human Gene Therapy Products Incorporating Human Genome Editing: Draft Guidance for Industry. (2022), Availability. (Australia: Food and Drug Administration (FDA)).
66. Tavridou A, Rogers D, Farinelli G, Gravanis I, and Jekerle V. Genome-editing medicinal products: the EMA perspective. Nat Rev Drug Discov. (2024) 23:242–243. doi: 10.1038/d41573-024-00050-2
67. Haeussler M, Schönig K, Eckert H, Eschstruth A, Mianné J, Renaud J-B, et al. Evaluation of off-target and on-target scoring algorithms and integration into the guide RNA selection tool CRISPOR. Genome Biol. (2016) 17:148. doi: 10.1186/s13059-016-1012-2
68. Stemmer M, Thumberger T, del Sol Keyer M, Wittbrodt J, and Mateo JL. CCTop: an intuitive, flexible and reliable CRISPR/Cas9 target prediction tool. PloS One. (2015) 10:e0124633. doi: 10.1371/journal.pone.0124633
69. Cradick TJ, Qiu P, Lee CM, Fine EJ, and Bao G. COSMID: a web-based tool for identifying and validating CRISPR/Cas off-target sites. Mol Ther Nucleic Acids. (2014) 3:e214. doi: 10.1038/mtna.2014.64
70. Fine EJ, Cradick TJ, Zhao CL, Lin Y, and Bao G. An online bioinformatics tool predicts zinc finger and TALE nuclease off-target cleavage. Nucleic Acids Res. (2014) 42:e42–e. doi: 10.1093/nar/gkt1326
71. Kim HK, Min S, Song M, Jung S, Choi JW, Kim Y, et al. Deep learning improves prediction of CRISPR–Cpf1 guide RNA activity. Nat Biotechnol. (2018) 36:239–41. doi: 10.1038/nbt.4061
72. Hwang G-H, Park J, Lim K, Kim S, Yu J, Yu E, et al. Web-based design and analysis tools for CRISPR base editing. BMC Bioinf. (2018) 19:542. doi: 10.1186/s12859-018-2585-4
73. Albinger N, Pfeifer R, Nitsche M, Mertlitz S, Campe J, Stein K, et al. Primary CD33-targeting CAR-NK cells for the treatment of acute myeloid leukemia. Blood Cancer J. (2022) 12:61. doi: 10.1038/s41408-022-00660-2
74. Berrien-Elliott MM, Cashen AF, Cubitt CC, Neal CC, Wong P, Wagner JA, et al. Multidimensional analyses of donor memory-like NK cells reveal new associations with response after adoptive immunotherapy for leukemia. Cancer discov. (2020) 10:1854–71. doi: 10.1158/2159-8290.CD-20-0312
75. Lazzarotto CR, Nguyen NT, Tang X, Malagon-Lopez J, Guo JA, Aryee MJ, et al. Defining CRISPR–Cas9 genome-wide nuclease activities with CIRCLE-seq. Nat Protoc. (2018) 13:2615–42. doi: 10.1038/s41596-018-0055-0
76. Cameron P, Fuller CK, Donohoue PD, Jones BN, Thompson MS, Carter MM, et al. Mapping the genomic landscape of CRISPR–Cas9 cleavage. Nat Methods. (2017) 14:600–6. doi: 10.1038/nmeth.4284
77. Kim D and Kim J-S. DIG-seq: a genome-wide CRISPR off-target profiling method using chromatin DNA. Genome Res. (2018) 28:1894–900. doi: 10.1101/gr.236620.118
78. Lazzarotto CR, Malinin NL, Li Y, Zhang R, Yang Y, Lee G, et al. CHANGE-seq reveals genetic and epigenetic effects on CRISPR–Cas9 genome-wide activity. Nat Biotechnol. (2020) 38:1317–27. doi: 10.1038/s41587-020-0555-7
79. Lazzarotto CR, Katta V, Li Y, Urbina E, Lee G, and Tsai SQ. CHANGE-seq-BE enables simultaneously sensitive and unbiased in vitro profiling of base editor genome-wide activity. bioRxiv. (2024). doi: 10.1101/2024.03.28.586621
80. Tsai SQ, Zheng Z, Nguyen NT, Liebers M, Topkar VV, Thapar V, et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat Biotechnol. (2015) 33:187–97. doi: 10.1038/nbt.3117
81. Kim N, Choi S, Kim S, Song M, Seo JH, Min S, et al. Deep learning models to predict the editing efficiencies and outcomes of diverse base editors. Nat Biotechnol. (2024) 42:484–97. doi: 10.1038/s41587-023-01792-x
82. Edri A, Ben-Haim N, Hailu A, Brycman N, Berhani-Zipori O, Rifman J, et al. Nicotinamide-expanded allogeneic natural killer cells with CD38 deletion, expressing an enhanced CD38 chimeric antigen receptor, target multiple myeloma cells. Int J Mol Sci. (2023) 24:17231. doi: 10.3390/ijms242417231
83. Gurney M, Stikvoort A, Nolan E, Kirkham-McCarthy L, Khoruzhenko S, Shivakumar R, et al. CD38 knockout natural killer cells expressing an affinity optimized CD38 chimeric antigen receptor successfully target acute myeloid leukemia with reduced effector cell fratricide. Haematologica. (2020) 107:437. doi: 10.3324/haematol.2020.271908
85. Crosetto N, Mitra A, Silva MJ, Bienko M, Dojer N, Wang Q, et al. Nucleotide-resolution DNA double-strand break mapping by next-generation sequencing. Nat Methods. (2013) 10:361–5. doi: 10.1038/nmeth.2408
86. Frock RL, Hu J, Meyers RM, Ho Y-J, Kii E, and Alt FW. Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat Biotechnol. (2015) 33:179–86. doi: 10.1038/nbt.3101
87. Vicente MM, Chaves-Ferreira M, Jorge JM, Proença JT, and Barreto VM. The off-targets of clustered regularly interspaced short palindromic repeats gene editing. Front Cell Dev Biol. (2021) 9:718466. doi: 10.3389/fcell.2021.718466
88. Turchiano G, Andrieux G, Klermund J, Blattner G, Pennucci V, El Gaz M, et al. Quantitative evaluation of chromosomal rearrangements in gene-edited human stem cells by CAST-Seq. Cell Stem Cell. (2021) 28:1136–47.e5. doi: 10.1016/j.stem.2021.02.002
89. Rhiel M, Geiger K, Andrieux G, Rositzka J, Boerries M, Cathomen T, et al. T-CAST: An optimized CAST-Seq pipeline for TALEN confirms superior safety and efficacy of obligate-heterodimeric scaffolds. Front Genome Editing. (2023) 5:1130736. doi: 10.3389/fgeed.2023.1130736
90. Klermund J, Rhiel M, Kocher T, Chmielewski KO, Bischof J, Andrieux G, et al. On-and off-target effects of paired CRISPR-Cas nickase in primary human cells. Mol Ther. (2024) 32:1298–310. doi: 10.1016/j.ymthe.2024.03.006
91. Dettmer-Monaco V, Weißert K, Ammann S, Monaco G, Lei L, Gräßel L, et al. Gene editing of hematopoietic stem cells restores T-cell response in familial hemophagocytic lymphohistiocytosis. J Allergy Clin Immunol. (2024) 153:243–55.e14. doi: 10.1016/j.jaci.2023.08.003
92. Cullot G, Aird EJ, Schlapansky MF, Yeh CD, van de Venn L, Vykhlyantseva I, et al. Genome editing with the HDR-enhancing DNA-PKcs inhibitor AZD7648 causes large-scale genomic alterations. Nat Biotechnol. (2024) 43:1778–1782. doi: 10.1038/s41587-024-02488-6
93. Raimondi F, Siow KM, Wrona D, Fuster-García C, Pastukhov O, Schmitz M, et al. Gene editing of NCF1 loci is associated with homologous recombination and chromosomal rearrangements. Commun Biol. (2024) 7:1291. doi: 10.1038/s42003-024-06959-z
94. Conti A, Giannetti K, Midena F, Beretta S, Gualandi N, De Marco R, et al. Senescence and inflammation are unintended adverse consequences of CRISPR-Cas9/AAV6-mediated gene editing in hematopoietic stem cells. Cell Rep Med. (2025) 6:102157. doi: 10.1016/j.xcrm.2025.102157
95. Schiroli G, Ferrari S, Conway A, Jacob A, Capo V, Albano L, et al. Preclinical modeling highlights the therapeutic potential of hematopoietic stem cell gene editing for correction of SCID-X1. Sci Trans Med. (2017) 9:eaan0820. doi: 10.1126/scitranslmed.aan0820
96. Cromer MK, Majeti KR, Rettig GR, Murugan K, Kurgan GL, Bode NM, et al. Comparative analysis of CRISPR off-target discovery tools following ex vivo editing of CD34+ hematopoietic stem and progenitor cells. Mol Ther. (2023) 31:1074–87. doi: 10.1016/j.ymthe.2023.02.011
97. Kalter N, Fuster-García C, Silva A, Ronco-Díaz V, Roncelli S, Turchiano G, et al. Off-target effects in CRISPR-Cas genome editing for human therapeutics: Progress and challenges. Mol Ther Nucleic Acids. (2025) 36:102636. doi: 10.1016/j.omtn.2025.102636
98. Elmas E, Saljoughian N, de Souza Fernandes Pereira M, Tullius BP, Sorathia K, Nakkula RJ, et al. CRISPR gene editing of human primary NK and T cells for cancer immunotherapy. Front Oncol. (2022) 12:834002. doi: 10.3389/fonc.2022.834002
99. Kwon J, Kim M, Hwang W, Jo A, Hwang G-H, Jung M, et al. Extru-seq: a method for predicting genome-wide Cas9 off-target sites with advantages of both cell-based and in vitro approaches. Genome Biol. (2023) 24:4. doi: 10.1186/s13059-022-02842-4
100. Nipper AJ, Warren EA, Liao KS, Liu H-C, Michikawa C, Porter CE, et al. Chick embryo chorioallantoic membrane as a platform for assessing the in vivo efficacy of chimeric antigen receptor T-cell therapy in solid tumors. ImmunoHorizons. (2024) 8:598–605. doi: 10.4049/immunohorizons.2400059
101. Marzagalli M, Pelizzoni G, Fedi A, Vitale C, Fontana F, Bruno S, et al. A multi-organ-on-chip to recapitulate the infiltration and the cytotoxic activity of circulating NK cells in 3D matrix-based tumor model. Front bioengineering Biotechnol. (2022) 10:945149. doi: 10.3389/fbioe.2022.945149
102. Murali Shankar N, Ortiz-Montero P, Kurzyukova A, Rackwitz W, Künzel SR, Wels WS, et al. Preclinical assessment of CAR-NK cell-mediated killing efficacy and pharmacokinetics in a rapid zebrafish xenograft model of metastatic breast cancer. Front Immunol. (2023) 14:1254821. doi: 10.3389/fimmu.2023.1254821
103. Parodi M, Astigiano S, Carrega P, Pietra G, Vitale C, Damele L, et al. Murine models to study human NK cells in human solid tumors. Front Immunol. (2023) 14:1209237. doi: 10.3389/fimmu.2023.1209237
104. Park S-J, Yoon H-J, Gu E-Y, Lee B-S, Kim Y, Jung J, et al. A general toxicity and biodistribution study of human natural killer cells by single or repeated intravenous dose in severe combined immune deficient mice. Toxicological Res. (2022) 38:545–55. doi: 10.1007/s43188-022-00138-0
105. Li H, Song W, Li Z, and Zhang M. Preclinical and clinical studies of CAR-NK-cell therapies for Malignancies. Front Immunol. (2022) 13:992232. doi: 10.3389/fimmu.2022.992232
106. Matsuda M, Ono R, Iyoda T, Endo T, Iwasaki M, Tomizawa-Murasawa M, et al. Human NK cell development in hIL-7 and hIL-15 knockin NOD/SCID/IL2rgKO mice. Life Sci alliance. (2019) 2:e201800195. doi: 10.26508/lsa.201800195
107. Katano I, Nishime C, Ito R, Kamisako T, Mizusawa T, Ka Y, et al. Long-term maintenance of peripheral blood derived human NK cells in a novel human IL-15- transgenic NOG mouse. Sci Rep. (2017) 7:17230. doi: 10.1038/s41598-017-17442-7
108. Bachanova V, Cayci Z, Lewis D, Maakaron JE, Janakiram M, Bartz A, et al. Initial clinical activity of FT596, a first-in-class, multi-antigen targeted, off-the-shelf, iPSC-derived CD19 CAR NK cell therapy in relapsed/refractory B-cell lymphoma. Blood. (2020) 136:8. doi: 10.1182/blood-2020-141606
109. Strati P, Castro J, Goodman A, Bachanova V, Kamdar M, Awan FT, et al. Off-the-shelf induced pluripotent stem-cell-derived natural killer-cell therapy in relapsed or refractory B-cell lymphoma: a multicentre, open-label, phase 1 study. Lancet Haematol. (2025) 12:e505–e15. doi: 10.1016/S2352-3026(25)00142-5
110. Yu K, Davidson CE, and Burshtyn DN. LILRB1 intron 1 has a polymorphic regulatory region that enhances transcription in NK cells and recruits YY1. J Immunol. (2020) 204:3030–41. doi: 10.4049/jimmunol.2000164
111. Samuelson C, Radtke S, Zhu H, Llewellyn M, Fields E, Cook S, et al. Multiplex CRISPR/Cas9 genome editing in hematopoietic stem cells for fetal hemoglobin reinduction generates chromosomal translocations. Mol Ther Methods Clin Dev. (2021) 23:507–23. doi: 10.1016/j.omtm.2021.10.008
112. Tycko J, Wainberg M, Marinov GK, Ursu O, Hess GT, Ego BK, et al. Mitigation of off-target toxicity in CRISPR-Cas9 screens for essential non-coding elements. Nat Commun. (2019) 10:4063. doi: 10.1038/s41467-019-11955-7
113. Aussel C, Cathomen T, and Fuster-García C. The hidden risks of CRISPR/Cas: structural variations and genome integrity. Nat Commun. (2025) 16:7208. doi: 10.1038/s41467-025-62606-z
114. Tao J, Bauer DE, and Chiarle R. Assessing and advancing the safety of CRISPR-Cas tools: from DNA to RNA editing. Nat Commun. (2023) 14:212. doi: 10.1038/s41467-023-35886-6
115. Richter MF, Zhao KT, Eton E, Lapinaite A, Newby GA, Thuronyi BW, et al. Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nat Biotechnol. (2020) 38:883–91. doi: 10.1038/s41587-020-0453-z
116. Lopes R and Prasad MK. Beyond the promise: evaluating and mitigating off-target effects in CRISPR gene editing for safer therapeutics. Front Bioeng Biotechnol. (2023) 11:1339189. doi: 10.3389/fbioe.2023.1339189
117. Kim D, Lim K, Kim ST, Yoon SH, Kim K, Ryu SM, et al. Genome-wide target specificities of CRISPR RNA-guided programmable deaminases. Nat Biotechnol. (2017) 35:475–80. doi: 10.1038/nbt.3852
118. Kim D, Kim DE, Lee G, Cho SI, and Kim JS. Genome-wide target specificity of CRISPR RNA-guided adenine base editors. Nat Biotechnol. (2019) 37:430–5. doi: 10.1038/s41587-019-0050-1
119. Liang P, Xie X, Zhi S, Sun H, Zhang X, Chen Y, et al. Genome-wide profiling of adenine base editor specificity by EndoV-seq. Nat Commun. (2019) 10:67. doi: 10.1038/s41467-018-07988-z
120. Grünewald J, Zhou R, Garcia SP, Iyer S, Lareau CA, Aryee MJ, et al. Transcriptome-wide off-target RNA editing induced by CRISPR-guided DNA base editors. Nature. (2019) 569:433–7. doi: 10.1038/s41586-019-1161-z
121. Du W, Zhao L, Diao K, Zheng Y, Yang Q, Zhu Z, et al. A versatile CRISPR/Cas9 system off-target prediction tool using language model. Commun Biol. (2025) 8:882. doi: 10.1038/s42003-025-08275-6
Keywords: CRISPR, NK cells, safety assessment, off-target, allogenic, gene-editing, cell therapy
Citation: Fazeli A, Ullrich E, Cathomen T and Bexte T (2025) Engineering with care: safety assessment platforms for CRISPR-modified natural killer cells. Front. Immunol. 16:1711414. doi: 10.3389/fimmu.2025.1711414
Received: 23 September 2025; Accepted: 10 November 2025; Revised: 02 November 2025;
Published: 05 December 2025.
Edited by:
Karl Petri, University Hospital Würzburg, GermanyReviewed by:
Minjing Wang, MyoKardia, Inc., United StatesCopyright © 2025 Fazeli, Ullrich, Cathomen and Bexte. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Tobias Bexte, YmV4dGVAbWVkLnVuaS1mcmFua2Z1cnQuZGU=
†These authors share senior authorship