CRISPR: uncovering clues with off-target therapeutic implications

By Jill Granger, Technical Writer, Product Management at Sartorius

Bacteria put up a good fight: ask anyone who’s ever had a strep infection.  Yet bacteria can be vulnerable too, especially when they encounter their nemesis of the micro world; the bacteriophage, which hijacks bacterial machinery for propagation.  Fortunately, bacteria are armed with their own brand of superhero powers, their CRISPR/Cas immune system, which they wield to fend off pesky invaders.

Inspired by bacteria, enter CRISPR technology:  a way to “Keep your friends close and your enemies closer.”

Exploiting bacterial defense mechanisms has led to the development of CRISPR technology for gene editing and identification. CRISPR has become a popular research tool in the fight against cancer, neurological disease, heart failure, and other rare conditions in desperate need of new therapeutic options. With new technology, however, comes new questions: one being how to best to monitor possible off-target effects that might arise from experimental manipulations and determine the factors that contribute? This is not an easy question to answer.

Dr. David Segal of the Genome Center at the University of California Davis describes the complexity of this issue:  “First, if we are editing cells in vitro (that is, in a culture dish), we are usually trying to cause some change in the cell's function resulting from the edit. Changing expression of one gene usually has downstream consequences on other genes, and it is often a technical challenge to sort out what gene expression changes are caused by the edit and which happen later as a result of the first gene change. Second, some studies have found that the cells respond to the introduction of the editing tools themselves (i.e., in the absence of cleavage, such as dCas9 or Cas9 with no gRNA). Often these responses are short-term, though that might depend on the cell type. Third, what we would really like to know is do off-target or other unintended events lead to bad things. Let's consider that the worst thing that could happen is to cause a cell to turn into a tumor. That is because it is probably not a big deal if, say, one neuron starts to produce a liver enzyme. However, if that unintended event causes the cell to grow into a cancer, now the consequences of that event are amplified and become a big deal. But that might be hard to see in cells growing in a culture dish”

There are many factors to consider, such as the method selected, delivery of CRISPR machinery (e.g. guide RNAs, nucleases) and cell types.  Then, there are issues surrounding epigenetic influences. To further complicate matters, the risk level will also depending on the context; whether it be an in vitro disease model, an in vivo animal model, or for the expansion and screening of cells for potential therapeutic use. While there have been many successful efforts to greatly reduce the frequency of off-target effects, they may still occur, which can be a problem when many cells are treated and expanded.  As Dr. Segal points out “The conversation needs to evolve from IF there will be off-targets to what are the CONSEQUENCES of the off-targets. Currently, that's a harder question to answer, especially in cell culture.”  Reliance on in vivo models presents another set of problems as well.  As David Segal points out “Animal models, such as mice or rats, don't completely solve the problem either. Mice and rats generally don't live long enough to develop cancer spontaneously.”

It’s complicated, but doing some additional footwork to understand how models and methods perform in a given context, incorporate the proper controls, and run some additional checks with functional outputs can be very informative. As a case in point, a recent Nature Medicine publication by Dr. Robert Ihry and colleagues at Novartis Institutes for Biomedical Research investigated the reduced efficiencies of genetically engineered hPSCs (human pluripotent stem cells), Although the hPSC lines they produced had stable integration of Cas9 or transient delivery of Cas9-ribonucleoproteins (RNPs), with an average indel efficiency greater than 80%, few of the hPSCs survived.  They next investigated the source of this toxicity using a variety of techniques, but also incorporated the functional measure of cell confluence using the IncuCyte® Live-Cell Imaging and Analysis System. They ultimately determined that P53 was required for the toxic response to double strand breaks (DSBs) that were induced by Cas9.  Moreover, P53 mutant cells still continued to grow, even with the DSB. This toxicity could hypothetically pose a problem for the high-throughput use of CRISPR/Cas 9 for genome engineering and screening of hPSC, and could help explain why the genomic engineering of hPSCs can be inefficient.  If inhibition of p53 was performed to counteract this toxic effect in this scenario using a plasmid, this could hypothetically give rise to an increase in off-target mutations with an associated increase in cancer risk.  This led to the authors’ caution on the use of CRISPR/Cas9-engineered hPSC for cell replacement therapies under the recommendation that the engineered cells be monitored for p53.

As pre-clinical research using CRISPR and related technologies continues to expand, answering questions about the origin, nature, and long-term implications of potential off-target effects could prove challenging, especially against the backdrop of ethical considerations. Unraveling this puzzle, piece-by-piece, will take time, incorporation of the appropriate outputs and analysis techniques, as well as interdisciplinary collaboration to bring new treatments to patients while maintaining tolerable limits of risk, which have yet to be defined.