Introduction:
New research published in Nature Biotechnology demonstrates that artificial intelligence (AI) can accurately predict the on- and off-target activity of CRISPR tools that target RNA instead of DNA. This study, conducted by researchers from New York University (NYU), Columbia Engineering, and the New York Genome Center, combines deep learning models with CRISPR screens to precisely control the expression of human genes. The ability to modulate gene activity in such precise ways holds potential for the development of novel CRISPR-based therapies.
The Role of RNA-Targeting CRISPR:
While CRISPR is widely known as a gene editing technology that targets DNA using the Cas9 enzyme, scientists have also discovered another type of CRISPR that targets RNA using the Cas13 enzyme. RNA-targeting CRISPRs have numerous applications, including RNA editing, gene knockdown, and high-throughput screening for potential drug candidates. Additionally, given that RNA is the primary genetic material in viruses like SARS-CoV-2 and the flu, RNA-targeting CRISPRs offer promise for the prevention and treatment of viral infections. Understanding RNA regulation and the function of non-coding RNAs is another goal of utilizing RNA-targeting CRISPRs.
Improving RNA-Targeting CRISPR Accuracy:
The study aimed to maximize the intended target RNA activity of RNA-targeting CRISPRs while minimizing activity on other RNAs, which could have detrimental effects on cells. Off-target activity includes mismatches, insertion mutations, and deletion mutations. Previous studies on RNA-targeting CRISPRs mainly focused on on-target activity and mismatches, neglecting the prediction of insertion and deletion mutations. Given that about one in five mutations in human populations are insertions or deletions, accurately predicting these types of off-targets is crucial for effective CRISPR design.
The TIGER Model:
To address this challenge, the researchers performed pooled RNA-targeting CRISPR screens in human cells, evaluating the activity of 200,000 guide RNAs targeting essential human genes. They collaborated with machine learning expert David Knowles to develop a deep learning model called TIGER (Targeted Inhibition of Gene Expression via guide RNA design). TIGER was trained using data from the CRISPR screens, and its predictions were compared to laboratory tests in human cells. The model outperformed previous models and successfully predicted both on- and off-target activity, including insertion and deletion mutations.
Benefits of TIGER:
TIGER's predictions provide insights into designing RNA-targeting CRISPRs that strike a balance between on-target knockdown and avoiding off-target effects. The model's off-target predictions can be utilized to precisely modulate gene dosage, enabling partial inhibition of gene expression with mismatch guides. This feature holds potential for addressing conditions where gene copies need to be regulated, such as Down syndrome, certain forms of schizophrenia, Charcot-Marie-Tooth disease, and cancer characterized by abnormal gene expression and uncontrolled tumor growth.
Advancing RNA-Targeting Therapies:
The integration of AI with RNA-targeting CRISPR screens paves the way for RNA-targeting therapies with enhanced precision and reduced off-target activity. By leveraging TIGER's predictions, unwanted CRISPR effects can be avoided, propelling the development of a new generation of RNA-targeting treatments.
Conclusion:
The recent research demonstrates the successful application of AI in predicting the on- and off-target activity of RNA-targeting CRISPR tools. By combining deep learning models with CRISPR screens, researchers have gained insights into gene modulation and the potential for developing precise RNA-targeting therapies. T
he ability to control gene expression in a precise manner has significant implications for various therapeutic applications. The study conducted by researchers from New York University (NYU), Columbia Engineering, and the New York Genome Center showcases the successful integration of artificial intelligence (AI) with CRISPR screens to accurately predict the activity of RNA-targeting CRISPR tools. This advancement opens doors for the development of new therapies that leverage the potential of RNA modulation.
CRISPR technology, known for its DNA-targeting capabilities using the Cas9 enzyme, has expanded to include RNA-targeting CRISPRs that employ the Cas13 enzyme. These RNA-targeting CRISPRs offer versatile applications, such as RNA editing, gene knockdown, and high-throughput screening for drug discovery. Furthermore, targeting RNA is particularly relevant in combating viral infections, given the vital role of RNA in viral genetic material.
The study aimed to optimize the efficacy of RNA-targeting CRISPRs on the intended target RNA while minimizing off-target effects. Off-target activity includes mismatches, insertion mutations, and deletion mutations. Previous studies primarily focused on on-target activity and mismatches, neglecting the importance of accurately predicting insertion and deletion mutations. However, these types of mutations account for a significant proportion of potential off-target effects, emphasizing their relevance in CRISPR design.
To address these challenges, the researchers conducted CRISPR screens in human cells, evaluating the activity of numerous guide RNAs targeting essential human genes. Collaborating with AI expert David Knowles, they developed a deep learning model named TIGER (Targeted Inhibition of Gene Expression via guide RNA design). By training TIGER with data from the CRISPR screens, the researchers were able to predict both on- and off-target activity, surpassing previous models. TIGER's predictions allow for the design of RNA-targeting CRISPRs that achieve the desired gene knockdown while minimizing off-target effects, including insertion and deletion mutations.
The benefits of TIGER extend beyond accurate guide RNA design. The model's off-target predictions enable precise modulation of gene dosage, offering the ability to partially inhibit gene expression using mismatch guides. This feature holds promise in addressing conditions characterized by aberrant gene expression, such as Down syndrome, certain forms of schizophrenia, Charcot-Marie-Tooth disease, and cancers where uncontrolled tumor growth is linked to abnormal gene expression.
The integration of AI and CRISPR screens for RNA targeting represents a significant advancement in the field. By leveraging TIGER's predictions, researchers can mitigate undesired off-target effects and drive the development of a new generation of RNA-targeting therapies. The ability to precisely modulate gene expression using RNA-targeting CRISPRs has the potential to revolutionize biomedical research and therapeutic interventions.
