Moving Forward with CRISPR

CRISPR-Cas9 is a powerful and flexible gene-editing tool, providing the means to induce specific and targeted changes to gene expression in a wide range of cell lines and models. CRISPR is an inherently simple system, needing only a short guide RNA to program the Cas9 endonuclease just about anywhere in a genome. Unlike previous gene-editing technologies, such as ZFN and TALENs, reprogramming CRISPR only needs a change to the sgRNA sequence rather than time-consuming and costly protein engineering. This has no doubt contributed to its rapid adoption as a gene-editing tool in a myriad of applications and research areas.

But despite its widespread success, off-target activity and on-target efficiency remain a challenge—and there are also technical considerations for the safe use of CRISPR technology in the context of gene therapy, such as increasing efficiency of delivery and eliminating the potential to cause autoimmune disease. This article discusses the current limitations of CRISPR and future improvements that will allow CRISPR to be taken to its full potential.

Specificity and efficiency of the CRISPR system

Like all gene-editing tools, CRISPR has a potential for off-target activity and editing at sites other than the desired loci, which adds an element of uncertainty to any scientific discoveries. This is of particular concern for therapeutic applications, as even a single off-target effect could have a major effect. Targeting Cas9 relies on a 20-nucleotide sequence of the sgRNA matching the genomic sequence immediately adjacent to the Cas9 binding site, called the PAM site (Protospacer Adjacent Motif). So, if the sgRNA shares sequence similarity with sites elsewhere in the genome, this may result in off-target activity. In particular, mismatches at the proximal end of the sgRNA are more tolerated and so contribute to increased levels of off-target cleavage.

Deep learning to keep CRISPR on target

Scientists working with CRISPR have to balance guide specificity with guide efficiency—efficient cleavage at the target site will also depend on accessibility and DNA binding affinity of the sgRNA. An ideal sgRNA for a particular CRISPR experiment will provide efficient cleavage at the target site, with little or no cleavage elsewhere in the genome—but finding such a guide remains a significant challenge in the application of CRISPR.

The development of computational tools to help design optimal guides has been a major focus in recent years, allowing researchers to select the best target sites and omit sgRNA that are predicted to have low efficiency. Several prediction models have been developed using whole-genome screening data to predict the best guides to use for a particular experiment—and recent advances have seen the application of a new generation of machine learning methods with artificial neural networks.¹ Xiang et al instead trained their deep learning tool CRISPRon using screening data they generated themselves using 10,592 guides—and were able to significantly improve prediction performance compared to existing tools.²

Multi-guide design for improved efficiency

Gene knockouts with CRISPR rely on the introduction of a double-strand break that activates endogenous DNA repair pathways leading to the introduction of insertion or deletion mutations—but this may not always result in a functional gene knockout that can introduce uncertainties in CRISPR experiments. The likelihood of success can be increased by targeting a single gene with multiple guides—for example, Synthego’s multi-guide system uses up to three sgRNAs to target the early exons of a gene.³ The addition of multiple double-strand breaks therefore significantly increases the probability of a functional knockout.

Increasing specificity of Cas9

An alternative way to reduce off-target activity is to modify the Cas9 itself, such as the use of a mutated Cas9 with only a single active nuclease domain, termed a Cas9 nickase, which only introduces a single-strand break at the target site. Gene editing with Cas9 nickases will require a pair of offset nickases to generate the required double-strand break—which reduces the probability of off-target cleavage while maintaining target efficiency.⁴

Alternatively, a high-fidelity version of Cas9 (SpCas9-HF) has been created that improves on-target specificity while maintaining on-target cleavage by disrupting non-specific interactions between the Cas9-sgRNA complex and DNA. Substituting key amino acids in the Cas9 molecule reduces the DNA binding energy of the sgRNA-Cas9 complex, therefore decreasing the ability to cleave target sequences that are mismatched.⁵

Immunogenicity of the CRISPR machinery

When looking to apply CRISPR in a clinical setting, ensuring low immunogenicity of the CRISPR machinery is a consideration. The most commonly used Cas9 protein is derived from S. pyogenes, recognizing the 3’-NTT-5’ PAM site, which is ubiquitous throughout the genome and so offers a wide range of targets for CRISPR experiments. However, the S. pyogenes bacterium is a common pathogen in humans and so the Cas9 protein will be recognized by the immune system—therefore generating an immune response and rapidly degrading the protein and preventing gene editing from occurring. Several strategies can be utilized to minimize immunogenicity, including editing in children before immunity to Cas9 has developed and targeting immune-privileged organs such as the brain and eyes. For example, CRISPR was used to delete CGG repeats in hiPSCs from fragile X syndrome patients and the edit was still present after differentiation into mature neurons.⁶

The use of synthetic sgRNA can also help mitigate against potential immunogenic effects of the CRISPR machinery for clinical applications, as the generation of synthetic sgRNA via in vitro transcription (IVT) allows the addition of chemical modifications to reduce the immune response and improve in vivo editing capability. Modifications such as 3’-phosphorothioate internucleotide linkages and 2’-O-methyl modifications at the 5’ and 3’ terminal ends reduce immunogenicity, as well as provide resistance to nucleases, decreasing off-target activity and improving editing efficiency.

Delivery challenges

Selecting a safe and precise delivery method for gene therapy applications is also a challenge that needs to be addressed. There are several approaches used to deliver CRISPR components, including viral methods, physical, or vesicle-based delivery systems, which all come with their own strengths and limitations. The most important considerations when selecting a mechanism are packaging of the necessary components, delivery, and how it can be targeted to the correct site. For example, adeno-associated viruses (AAV) such as adenovirus and lentiviral vectors, which have been successfully used, are safe for use in humans and come with a reduced immune response—but AAV doesn’t have sufficient packaging size for Cas9, and lentiviral vectors randomly integrate into the genome and so can come with a risk of mutations. The continual expression that comes with viral vectors can also increase off-target activity and cytotoxicity. Rather than CRISPR plasmids, an alternative option is recombinant ribonucleoprotein (RNP) complexes delivered by lipofection or electroporation. Because the RNP is fully developed it is immediately active and so can quickly enter the nucleus for immediate gene editing. But importantly, the RNP complex will be naturally turned over by cellular machinery—transient expression of CRISPR components have been shown to reduce off-target activity and minimize cellular effects.⁷

The full potential of CRISPR

As well as the challenges surrounding efficiency and specificity of the CRISPR system, there are also significant ethical considerations to be addressed. Genetic alterations in humans have always been an important ethical debate but given the rapid development of the CRISPR technology and wide range of potential applications, the ethical and societal implications of the technology need to be examined. But despite the limitations, CRISPR-Cas9 remains a major revolution for gene editing, allowing researchers to make targeted genomic changes relatively easily with standard molecular biology techniques. CRISPR has changed the world and work continues to overcome these challenges to take CRISPR to its full potential.

Key Takeaways

• CRISPR-Cas9 technology has revolutionized gene editing, allowing targeted and specific genetic changes to be made without the need for costly protein engineering
• CRISPR technology shows great promise for therapeutic applications, but there are challenges facing CRISPR-based treatments such as immunogenicity and delivery of CRISPR components—however, there are several strategies that can be employed to overcome these limitations
• Reducing off-target activity and maximizing on-target efficiency remain a challenge of the CRISPR technology, however using machine learning to select the best guides, multi-guide design for efficient cleavage, and the use of Cas9 variants such as Cas9 nickase or high-fidelity Cas9 can ensure precise, specific and efficient gene editing and take CRISPR to its full potential

References

1. Konstantakos, Vasileios et al. “CRISPR-Cas9 gRNA efficiency prediction: an overview of predictive tools and the role of deep learning.” Nucleic acids research vol. 50,7 (2022): 3616-3637. doi:10.1093/nar/gkac192

2. Xiang, Xi et al. “Enhancing CRISPR-Cas9 gRNA efficiency prediction by data integration and deep learning.” Nature communications vol. 12,1 3238. 28 May. 2021, doi:10.1038/s41467-021-23576-0

3. https://www.synthego.com/help/synthego-multi-guide-design

4. Ran, F Ann et al. “Double nicking by RNA-guided CRISPR Cas9 for enhanced genome editing specificity.” Cell vol. 154,6 (2013): 1380-9. doi:10.1016/j.cell.2013.08.021

5. Kleinstiver, Benjamin P et al. “High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects.” Nature vol. 529,7587 (2016): 490-5. doi:10.1038/nature16526

6. Park, Chul-Yong et al. “Reversion of FMR1 Methylation and Silencing by Editing the Triplet Repeats in Fragile X iPSC-Derived Neurons.” Cell reports vol. 13,2 (2015): 234-41. doi:10.1016/j.celrep.2015.08.084

7. Kim, Sojung et al. “Highly efficient RNA-guided genome editing in human cells via delivery of purified Cas9 ribonucleoproteins.” Genome research vol. 24,6 (2014): 1012-9. doi:10.1101/gr.171322.113