Gene Therapy Development | Industry Spotlights & Insight Articles

Abundant Stochastic Off-Target Activity by Base Editors

Can complementary approaches for off-target quantification demonstrate substantial levels in various state-of-the-art editors?

Presented by Eli Eisenberg, School of Physics and Astronomy, at Tel Aviv University

Transcribed by Tia Byer

Professor Eli Eisenberg of Tel Aviv University's School of Physics and Astronomy works on the application of statistical methods and modelling of large-scale RNA expression data. He specialises in analysing the inconsistencies between observed RNA sequence content and reference genomes.

During our Gene Therapy Development and Manufacturing: In-Person event in June 2022, Eisenberg presented on the utility of base editors for RNA genome editing. In this article, we share an exclusive summary of that presentation and the key academic insights it explored.

Understanding Base Editors

Base editors refer to the programmed introduction of single-nucleotide variants into DNA or RNA in living cells. DNA base editors are a promising genome editing technology and according to Eisenberg “are ideal for correcting deleterious single-point mutations”.

RNA base editors have become popular in recent years and edit the RNA directly instead of the DNA. This technology has many advantages and, in some cases, such as with postmitotic cells, may be the preferable editing tool.

DNA base editors are a promising genome editing technology and according to Eisenberg “are ideal for correcting deleterious single-point mutations”.

For example, RNA is transient, which means it has a better safety read-out than DNA base editors as there is a lower risk of off-target mutations. In addition to this, RNA-based therapeutics are already FDA-approved.

Assessing the Risk of Off-Target Activity

Within gene editing, there can be both guide RNA-dependent off-targets which are localised to regions complementary to the guide RNA and guide RNA-independent editing activity of the deaminases. Eisenberg explained how whilst efforts are being made to identify and quantify target activity, there is still a way to go.

Current approaches are limited to optimising either the enzyme or guide to try and minimise off-target activity. “Optimisation efforts require reliable quantification of the outcome, and most efforts apply single nucleotide variants (SNVs) detection tools to count the number of SNVs,” Eisenberg pointed out. This is effective for specific off-target sites, which are consistently edited in multiple molecules. However, SNV tools remain insensitive to non-specific off-targets.

Quantifying Base Editors Off-Targets

It is possible to quantify base edits by analysing the distribution of the number of events that occur off-target. By calculating the number of off-target events by the level of event per site, it is possible to judge the effectiveness of the base editor optimisation. In most cases, optimisation efforts are shown to suppress preferentially the specifically edited sites, detectable by SNV tools. “In fact, most off-target editing events occur at weakly-edited positions,” Eisenberg commented.

A further strategy for quantification includes using an adaptation of the Alu editing index. Alu repeats are 300bp-long repetitive elements abundant in the human genome. The pairing of two Alu copies creates a preferred Adenosine-to-Inosine (ADAR) target, resulting in millions of editable sites. Although the editing level at most Alu sites is very low, together Alu editing accounts for ~99% of all endogenous ADAR activity.

“The editing index is a robust and useful tool to quantify globally stochastic off-targets load of base-editors.”

Due to the low editing rates, detecting and analysing endogenous Alu editing can be challenging. The Alu-editing index was created to overcome this challenge. It provides a single number that quantifies global ADAR1 editing in human cells, and aims to account for editing activity across all Alu adenosine sites, including low coverage regions, to allow for a reliable comparison across samples.

A similar approach was applied to quantify off-target editing within coding RNA and DNA sequences following the introduction of engineered base-editors.

Characterising Non-Specific Editing and Future Outlooks

The sites where stochastic off-target editing occurs do conform to the motif of the relevant enzymes. However, other than that they are spread evenly, synonymous and non-synonymous sites are equally edited and the effect size does not depend on expression level; essential genes, oncogenes, and clinically relevant mutation sites are all equally targeted.

“The editing index is a robust and useful tool to quantify globally stochastic off-targets load of base-editors,” Eisenberg concluded. Eisenberg continues his research into the role of endogenous ADAR1 editing in protection against false activation of the innate immune system. Future projects include developing a robust and reliable method for profiling recoding sites in humans and other mammals, and understanding the evolution of recoding events features as a main point of investigation.

To read Eisenberg's full report and paper, visit: https://genome.cshlp.org/content/31/12/2354.full

Want to find out more about the latest cell therapy news? Register now for Oxford Global‘s Cell UK: In-Person event to advance your understanding of cell-based products to ensure clinical and commercial success.