While the fashion and music tastes of the late 1980’s was rather questionable, it was a time of historic technological and societal change. Between the breakdown of the Soviet Union and the rise of the global computer network that we now call internet, a small group of researchers succeeded for the first time in generating genetically engineered mammalian cells and organisms.
Tools of the Genome Editing Trade
Since these first steps, the field of genome editing has been evolving with tremendous speed. In the last decade, new tools have been developed that greatly facilitate the generation of targeted genetic manipulations.
These tools come in the form of engineered nucleases that allow researchers to generate site-specific double strand breaks (DSBs) in genomic sequences. These DSBs can than be exploited to introduce the desired genetic alterations by harnessing the cell’s own DNA repair mechanisms:
- Non-homologous end-joining (NHEJ) leads to small insertions or deletions (indels), which can result in gene disruption by causing a frameshift
- Two DSBs on the same chromosome can be used to create deletions of up to several megabases
- Homologous recombination (HR) can be exploited to introduce defined mutations by supplying an exogenous DNA donor that harbours homology to the target site
So far three different families of engineered nucleases have been adapted for genome editing: Zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs) and clustered regularly interspaced short palindromic repeats (CRISPR/Cas) or RNA-guided endonucleases (RGENs).
Discovery of the latter two systems, in particular the CRISPR/Cas system, has led to a gold rush of genome engineering applications due to the ease with which they can be adapted to target specific genomic sequences.
However, while everyone is rushing ahead into the golden age of genome engineering a central question remains: How specific are these tools? Several studies have addressed this and come to the conclusion that ”it depends”. It depends on the type of nuclease used. It also depends on the targeted sequence, its similarity to other sites and its accessibility.
The Problem of Off-target Activity
Genomes are huge and offer a lot of possible binding sites for a nuclease with an intrinsic potential to bind DNA. A certain risk of off-target activity therefore always remains. The only solution to deal with this problem is to predict potential off-target sites, optimize the design of the nuclease and screen the generated cell lines for additional unwanted mutations.
Identifying Off-targets
Several approaches have been developed to predict potential off-target sites of a given engineered nuclease:
in vitro Methods
Systematic evolution of ligands by exponential enrichment (SELEX) has been used to identify the sequences that individual nucleases prefer to bind. SELEX uses a pool of DNA sequences that is submitted to alternate cycles of ligand selection and amplification of the sequences bound by the nuclease. A novel variation of this method is based on the cleavage of potential target sites by the nuclease in a pool of DNA sequences. The cleaved products can than be identified by paired-end high-throughput DNA sequencing.
In vitro techniques give an unbiased overview of all potential off-target sequences for an individual engineered nuclease. However, the obtained results strongly depend on experimental conditions and do not take other factors such as chromatin structure and locus accessibility into account, both of which affect rates of off target cleavage.
in vivo Methods
Integrase-deficient lentiviruses (IDLVs) or adeno-associated viruses (AAVs) integrate at sites of DSBs, which can be mapped to reveal nuclease off-target sites. Alternatively, ChIP-seq can be used to pull down the nuclease and map the sequences it is bound to.
In vivo methods overcome the problems of chromatin structure and locus accessibility associated with in vitro methods, however usually have a lower sensitivity or yield higher false-positive rates.
in silico Methods
Algorithms, based on data collected by in vitro and in vivo studies can be used to predict off-target sites without the need to perform complicated assays for each individual nuclease. Examples of such algorithms include PROGNOS, which can be used for off-target prediction for ZFNs and TALENs; the CRISPR Design Tool, which was specifically designed to predict and score off-targets for the CRISPR/Cas system; and CHOPCHOP, an algorithm suitable for both CRISPR/Cas and TALEN off-target prediction.
Minimizing Off-targets Events
In order to save time and effort it is best to try to minimize the chances for an off-target cleavage to occur. Several measures should be taken into consideration when performing nuclease assisted gene editing:
Choice of Nuclease
Bear in mind of the differences between systems when choosing a nuclease. ZNFs and TALENs are considered to be the most specific but are relatively complex to produce. The CRISPR/Cas system, on the other hand, is simple to us but has been shown to possess a considerable rate of off-target activity.
Target Site Choice
Target site choice is usually offers some flexibility. Using the above mentioned in silico tools, the target site and/or nuclease can be selected to minimize the amount of potential off-target effects.
Nuclease Delivery Method
The way the nuclease is delivered into the cells has been shown to influence the amount of off-target activity. Compared to transfection of expression plasmids, delivery of engineered nucleases as mRNA or purified protein reduces the chances of off-target effects by lowering the protein amount and time of expression.
The point in the cell cycle that the nuclease is delivered also plays a critical role. It has been shown that cell cycle synchronization and timed delivery of the nuclease in G2 increases HR efficiency while reducing unwanted NHEJ events
Analyzing Off-target Events
After everything’s set and done, the cells need to be checked for potential off-target mutations. Here is what you can do:
- PCR amplification of predicted off-target sites followed by either Sanger sequencing, CEL-I, or T7E1 assays.
- Copy number variations of regions larger than 1kb can be identified by array comparative genomic hybridization (aCGH).
- If you can afford it or for clinical purposes whole exome or whole genome sequencing (WGS) can be performed.
Unmet Needs of Analyzing Off-Target Events
Experimental identification of off-target sites by in vitro or in vivo methods is time consuming and, therefore, only useful when the nuclease is created for multiple targetings, for example in clinical applications.
While computational tools do not require experimental effort, they do not necessarily cover all potential off-target sites. Therefore WGS remains the most comprehensive method to assess off-targets produced by engineered nucleases. It is important to mention, however, that especially WGS approaches are, at the moment, still very expensive and require an enormous bioinformatic effort to analyze.
In addition, when working with bulk cell populations this method is not applicable, due to a lack of sensitivity to discover rare off-target events. While the above-mentioned methods address some issues, a strategy that allows fast, sensitive and cost-effective detection of off-target activtiy is still missing
For a detailed summary about on- and off-target analysis take a a look at Trends in Biotechnology, February 2015.