We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.


Cutting With CRISPR: Assessing Safety as Technology Moves Into the Clinic

A gloved hand holds a section of DNA in metal tweezers, which has come out of a DNA helix.
Credit: iStock
Listen with
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 7 minutes

CRISPR/Cas is a genome-editing technology, which was recently recognized by a Nobel Prize in Chemistry in 2020 awarded to Emmanuelle Charpentier and Jennifer Doudna. CRISPR/Cas is currently associated with exciting medical applications; however, research into CRISPR/Cas began on its role as a bacterial immune defense system, by cleaving viral DNA. Charpentier and Doudna realized that CRISPR/Cas could be repurposed as a genome editing tool due to its ability to target specific DNA sequences.

In its canonical form, CRISPR/Cas is comprised of CRISPR, an RNA guide molecule that targets the gene of interest by base pairing, complexed with a Cas endonuclease – usually Cas9 – a protein molecule that cleaves the target sequence, i.e., generates a double-stranded break (DSB). Once the target sequence is cleaved, an incomplete repair can lead to deletion or random insertion of DNA bases into the target gene, inactivating it. If a complementary strand to the gene is present, repair may instead lead to targeted gene insertion. The technology has since been modified to perform more controlled modifications by only nicking the target DNA, i.e., breaks to a single strand. Subsequently, CRISPR/Cas-conjugated base editors perform single-base changes to the target gene whereas prime editors lead to more complex changes, such as small insertions, deletions and replacements.

CRISPR/Cas is relatively rapid, cost effective and efficient. It is also extremely versatile and can edit genomes in multiple model organisms, including bacterial, fly, worm, rodent, non-human primates and human. Manipulations can be performed in cells, but increasingly, in vivo applications are also feasible options.

CRISPR/Cas: Safety studies please!

CRISPR/Cas has evolved rapidly since its earliest applications as a genome editing tool. Although it is not in routine clinical use, clinical trials are ongoing and some patients have received CRISPR/Cas-edited cells, e.g., to treat β-thalassemia, sickle cell disease and limited types of cancer. Given its imminent move to the clinic, assessing CRISPR/Cas1 safety2 is an active area of research.

“Previous methods of genome modification, such as viral insertion, randomly integrated into the genome. By contrast, CRISPR/Cas is targeted to a specific locus, by base pairing of the CRISPR guide RNA to the gene of interest. However, the genome has billions of bases and CRISPR/Cas is used to edit billions of cells, so it is still possible for errors to occur,” explained Kyle Cromer, assistant professor at the Department of Surgery, University of California San Francisco. “These errors, can be ‘on-target’, meaning they occur at the desired location, but are incorrectly edited, or they can be ‘off-target’, meaning the error occurred at a locus other than the intended target.”

It is critical to assess these errors because they can result in a myriad of undesirable and potentially dangerous consequences for the patient, such as a truncated or missense protein with toxic loss-of-function or gain-of-function or by disrupting tumor suppressor genes or activating oncogenes.

Off-target errors can arise because the CRISPR guide RNA can make a near-match pairing with a stretch of DNA at a locus other than the target gene due to the billions of bases present in the genome. “There are several methods to assess potential off-target effects, both computational in silico approaches and experimental, such as in vitro cell-free and in cellulo,” explained Roberto Chiarle, professor at the Department of Pathology, Boston Children's Hospital and Harvard Medical School. “These methods vary in their strengths and weaknesses. For example, although cell-free methods are fast, they identify pseudo off-target sites versus in cellulo methods, which introduces Cas directly into cells,” Jianli Tao, postdoctoral fellow with Chiarle at the Department of Pathology, Boston Children's Hospital and Harvard Medical School, elaborated.

Ultimately, as CRISPR/Cas heads to the clinic, it is necessary to comprehensively assess safety. “If CRISPR/Cas is to become the standard of care for genome editing, the bottom line is that we need to comprehensively assess safety,” Cromer said. “At present, only a handful of patients have received CRISPR/Cas-edited cells. But if the technology will go into more people, the occurrence of unintended events could rise.”

CRISPR/Cas: Quality control by deep sequencing

Cromer is interested in assessing the safety and efficacy of CRISPR/Cas, including in patient-derived cells. “We recently performed deep sequencing3 of primary human hematopoietic stem and progenitor cells (HSPCs) from three donors (with consent) following CRISPR/Cas editing to experimentally profile off-target incidents,” Cromer said. This experiment was significant because Cromer used patient-derived cells as a clinically relevant context for testing safety versus immortalized cells, which can influence off-target frequency. Moreover, HSPCs are editable by CRISPR/Cas to potentially treat diseases of hemoglobin mutations in blood, such as β-thalassemia and sickle cell anemia.

The study used ultra-deep sequencing to detect extremely rare mutations from off-target effects at 523 cancer-relevant genes after CRISPR/Cas editing with RNA guides targeted to 3 genes. The CRISPR guide RNA for one of the gene targets also had a predicted off-target site, as a positive control. “We found that CRISPR/Cas did not cause off-target effects at any of the cancer-relevant genes that we tested,” Cromer summarized. “Importantly, a single nucleotide polymorphism (SNP) in the target DNA, meaning a difference in only one base, prevented off-target activity from the guide RNA in the CRISPR/Cas-modified HSPCs.”

In addition to validating the specificity of CRISPR/Cas, the study had important clinical implications for the natural genetic variation present in the human population. “The CRISPR guide RNA is designed using a consensus sequence of the human genome. However, natural genetic variability in different individuals means they might have SNPs, i.e., single DNA base differences, in a gene of interest. Our results show that even a single base difference can alter the likelihood of CRISPR editing in the target gene in a patient. This means that we will need to consider natural genetic variability, if CRISPR/Cas is to make it into the clinic, possibly by sequencing the patient’s gene of interest for SNPs.”

HELP LINE-1: Retrotransposons at CRISPR/Cas breaks

Chiarle and Tao are also assessing the safety of CRISPR/Cas. Their interest, among other projects, lies in the integration of retrotransposons at CRISPR/Cas breaks, which they examined in a recent paper.4 Retrotransposons pose a threat because they can insert into critical genes, e.g., tumor suppressors, deactivating them. “We first used canonical CRISPR/Cas, which produces double-stranded breaks to edit relevant genes. We designed CRISPR RNA guides to target genes in three cell models, including MYC (a famous oncogene), CCR5 (the main coreceptor for HIV infection) and RAG1, which contributes to recombination of V (variable), D (diversity) and J (joining) – VDJ – domains of antibodies to increase diversity). We then monitored for insertion of the retrotransposon called long interspersed element-1 (LINE-1),” Tao elaborated on the experimental design. “We found that LINE-1 insertions are an undesirable outcome associated with canonical CRISPR/Cas editing using DSBs. We did not formally demonstrate the biological consequences of these LINE-1 insertions in our paper; however, they might potentially pose a safety risk by inactivating important genes or through ensuing chromosomal modifications.”

The study also tested the frequency of LINE-1 retrotransposon insertions after genome modification using base or prime editors exploiting the CRISPR/Cas systems, which only nick the target DNA to generate a single-stranded break. “We found that LINE-1 insertions are less frequent when target genes are edited using base or prime editors,” Chiarle described. “These results are consistent with the observation that base and prime editors cause fewer plasmid and genomic insertions and suggest that this newer generation of CRISPR/Cas tools are likely much safer for use in patients than canonical CRISPR/Cas.”

Although base and prime editors are routinely used in research labs, they are relatively new and are still in the early stages of the clinical pipeline. “As a matter of fact, clinical trials still more frequently use canonical CRISPR/Cas,” Tao said of the state of the field. “Therefore, our results suggest that an extra layer of the safety check for LINE-1 and other retrotransposon insertions should be added to fully evaluate the safety of CRISPR/Cas editing systems, especially in trials using canonical CRISPR/Cas instead of base and prime editor systems.”

CRISPR/Cas safety: The road ahead

Although research has made significant advances in assessing CRISPR/Cas safety there are many outstanding questions and concerns.

As Cromer pointed out, inadvertent chromosomal changes post CRISPR/Cas editing is a significant issue. “We’ve seen that CRISPR/Cas is very efficient for genome editing, and that it is precise for loci specific changes. However, CRISPR/Cas cuts can cause large disruptions, which are harder to identify, such as large deletions, translocations and chromothripsis (chromosome shattering). These can occur as a consequence of on-target editing or repeated cutting.” Chiarle and Tao expressed similar concerns. “Retrotransposons, such as LINE-1, which inserts into CRISPR/Cas-generated double-stranded breaks, can cause complex genomic rearrangements. And of course, the double-stranded breaks from CRISPR/Cas itself, can induce large genomic alterations as well.”

Additionally, in vivo editing within patients is also on the horizon. “We will need additional safety tests if we are taking CRISPR/Cas in vivo, which is the present trajectory,” Cromer cautioned. “It is already difficult to assess quality control of CRISPR/Cas editing in cells, performing it in vivo is even harder. In addition to considering on-target and off-target events to the genome, in vivo safety assessments will also need to consider on-tissue and off-tissue events, i.e., whether the CRISPR/Cas agents are delivered to intended or unintended tissue.”

Chiarle and Tao raised what future challenges might include: “To take CRISPR/Cas to the clinic, we will need to consider the method of in vivo delivery, therapeutic benefit, and editing efficiency, precision, and specificity, among other obstacles likely to occur. Concerning safety in vivo, precision and specificity will be of particular importance, although these are difficult to address at present due to a lack of models to assess this accurately. Monitoring trial participants will be key, however this is limited by accessibility to some tissues.”

Cromer added that he thinks synthetic biology will take on a very important role in CRISPR/Cas’s future: “We can engineer payloads to make delivery more precise. We can also make CRISPR/Cas editing safer by also introducing inducible safety switches, which may be activated to clear edited cells in the case of an adverse event. CRISPR/Cas has evolved rapidly, and we will need to evolve safety mechanisms to keep pace,” he said.



1.     Wienert B, Cromer MK. CRISPR nuclease off-target activity and mitigation strategies. Front Genome Ed. 2022;4:1050507. doi: 10.3389/fgeed.2022.1050507

2.     Tao J, Bauer DE, Chiarle R. Assessing and advancing the safety of CRISPR-Cas tools: from DNA to RNA editing. Nat Commun. Jan 13 2023;14(1):212. doi: 10.1038/s41467-023-35886-6

3.     Cromer MK, Barsan VV, Jaeger E, et al. Ultra-deep sequencing validates safety of CRISPR/Cas9 genome editing in human hematopoietic stem and progenitor cells. Nat Commun. Aug 11 2022;13(1):4724. doi: 10.1038/s41467-022-32233-z

4.     Tao J, Wang Q, Mendez-Dorantes C, Burns KH, Chiarle R. Frequency and mechanisms of LINE-1 retrotransposon insertions at CRISPR/Cas9 sites. Nat Commun. Jun 27 2022;13(1):3685. doi: 10.1038/s41467-022-31322-3