Achieving the genome payoff in the age of CRISPR/Cas9
CRISPR/Cas9 remains at forefront of media attention, following the announcement last week that UK research funders will continue to support human genome editing. In this article, from issue 3 of Front Line Genomics magazine, Michele Cleary, Executive Director for Genetics and Pharmacogenomics at Merck & Co. Inc., explores the significant impact the technology has already had on drug development.
Genomic editing is a field dominated by ethical debates and IP battles. How are people applying CRISPR/Cas9 technology? With everything that’s been going on, it’s easy to miss the positive impact it is having on drug development.
In 2010, for the 10th anniversary of the completion of the first draft of the human genome, I was asked by a reporter for a prominent newspaper why our knowledge of the human genome had not yet enabled new medicines. At the time, our team was using RNA interference (RNAi) to identify new drug targets across diverse disease areas. Informed by genomic sequence information, we had built a world class genome-scale RNAi screening platform. The reporter asked if any new drugs had been facilitated by our approach. I replied that it was still early days, and that we would see the fruits of this type of genomics-based drug discovery in years to come. The reporter scoffed and used my statement as support for his thesis that the investment in the human genome had not really paid off.
I often reflected afterwards on my disappointing response. As a post-genome drug discovery scientist, genomic information was woven into every research plan that my teams pursued. At the time of the interview, we had already conducted many successful screens and assembled comprehensive lists of potential new targets. Unfortunately, this was usually as far as we could go. While we could confirm our screening hits in additional assays by re-targeting those genes via RNAi, we lacked tools that were free from appreciable off-target effects and that could more precisely mimic in relevant models genomic changes that are causal for disease. The targets on our lists that gained the most enthusiasm were those for which there was already literature support, putting in question the value of looking for novel genes in the first place. In the years to follow, the genomic and genetic information available to scientists has burgeoned substantially. There has also been a surge in additional ‘omics and clinical data that can be intersected with genome data to empower hypotheses about disease biology and potential novel drug targets. Until recently, however, rigorous experimental follow up and validation of these remained elusive due to a sustained dearth of precise and effective tools to get at the heart of gene function.
Over the past few years, a breakthrough for target validation has been realized in the area of genome editing. Technologies involving engineered zinc finger nucleases and TALENs debuted as options for creating genome modifications that could inform on gene function. But for reasons both technical and practical, the adoption of these approaches has been somewhat limited (for review, see Sander and Joung 2014). In 2013, adaptation of the bacterial CRISPR/Cas9 system emerged for genome editing in mammalian cells, and, since then, it has been evolving at an unprecedented pace (for review, see Hsu, et al. 2014).
CRISPR/Cas9 has many advantages including impressive efficiency, overall flexibility in generating gene knockouts or sequence substitutions, and easily designed and synthesized reagents, which allow for iteration and optimization. The ease and efficiency of CRISPR/Cas9 also enables the engineering of changes at multiple genomic loci simultaneously. Most importantly, however, researchers can use it to mimic disease-relevant human genetic alterations more precisely in cellbased and animal models.
To say that the CRISPR/Cas9 technology has revolutionised drug discovery is an overstatement. There are still some limitations to robust target validation. But CRISPR/Cas9 has enabled the pursuit of those novel targets that typically got left by the wayside after comprehensive target identification campaigns due to the lack of tools for probing the associated biology in depth. Perhaps the biggest advantage, however, is that its speed and ease of use allow for rapid experimental iteration in a wide variety of biological systems. It is this characteristic of CRISPR that has started a new era of functional genomics, in which forward genetic screens benefit from high throughput gene knockouts at the DNA level and a potential to pursue modulation (repression and activation) of RNA transcription (for review, see Shalem et al. 2015).
Our current research is dedicated to providing functional validation of potential new drug targets that are identified through human genetics. In essence, we allow the genetic crapshoot of human reproduction to do the target identification for us (sometimes referred to as experiments of nature). By sifting through genotypes of large numbers of patients and controls, or looking at families with a high incidence of a disease, we arrive at hypotheses about genome alterations that cause pathologies as well as those that may be protective. With information in hand and the robust genome engineering afforded by CRISPR/ Cas9, we can study the biological significance of these events in multiple ways.
A first step in following up on genetic findings is to determine whether perturbations in a gene impact phenotypes relevant to a disease. Achieving this goal requires an assay or panel of assays that are translatable. In other words, with such assays there is a high level of confidence that the knowledge gleaned is truly meaningful with respect to the pathophysiology observed in humans. CRISPR/Cas9 can offer a very quick view into the potential impact of a gene in an assay through a direct knockout of that gene. If a change in phenotype results, the gene could likely be important from a pathway perspective. If there is no change, however, the gene isn’t necessarily irrelevant. More work just needs to be done.
Genetic variations can have multiple impacts on the genes with which they are associated. If a single nucleotide polymorphism in a coding region results in a synonymous change, the encoded amino acid remains the same and the impact on protein sequence will be negligible. Missense variants resulting in amino acid sequence alteration can change the folding of a protein, its function or its ability to bind other molecules in a cell. Nonsense mutations can result in loss of function when the “business end” of a protein is deleted. Getting at the heart of these changes biologically is best done by recreating them in a normal cell or correcting the variant back to wild-type in a cell from a genetic carrier. It is toward this goal that CRISPR/ Cas9 genome engineering offers a significant advantage over RNAi or even a straight gene knockout as it allows for the replacement of deleted sequence with new sequence carrying the desired alteration or variation. In addition to this direct application of CRISPR/Cas9 in recreating genic mutations, it can also be used to recreate intergenic or non-coding variations that also result in disease.
To probe in-depth the role of genetic variation in disease, we are creating both engineered cell lines and animal models that carry these variants to understand the overall physiology that they perturb. With the proper assays, we can determine whether gene changes lead to loss, gain, or change in function, and we can use our engineered cell and animal models to screen for pharmacologic agents that may reverse or ameliorate the disease pathology.
As we have begun to incorporate CRISPR/Cas9 genome engineering into our target validation work flow, we have uncovered several deficiencies that could benefit from optimization. For our research, an important limitation is in the commoditization of CRISPR engineered cell line generation. Our lab workforce is one of our most limited resources. We have a modest-sized team with genome-sized goals. Wherever we are able to do so, we outsource to enhance productivity. Lulled by the hype that CRISPR engineering is fast, efficient, and easy, we mistakenly viewed it as an established commodity. Unfortunately, our experience with outsourcing cell line generation has been disappointing as the turnaround times for delivery of homozygous knockout clones have been lengthy. It is our hope that in the not too distant future, engineered cells can be made externally in less than 2 months, allowing these tools to keep up with the pace of our science.
Another application with room for improvement is CRISPR/Cas9 for sequence substitution. CRISPR results in gene knockouts through the non-homologous end joining (NHEJ) process. NHEJ is a DNA repair process that responds to the double-stranded breaks that CRISPR/Cas9 enzymatic activity induces. The repair is error-prone and coupled with chewing back of the broken DNA strands. When the chewed back sequence is then ligated, sequence deletions result and some will cause frameshift mutations in coding sequence that could lead to early stop codons and protein truncation. CRISPR also induces homologydirected repair (HDR), which requires a template with considerable homology to the site of the double strand breaks. NHEJ competes with HDR and is much more efficient (upwards of 50% success versus HDR’s rate of <10%). Therefore, knockouts are more readily achieved and the hunt for a replacement involves the analysis of many clones. There may be future steps that can be taken to shift the balance to HDR and recent papers point to means for inactivating of the NHEJ machinery (Chu et al. 2015; Maruyama et al. 2015).
We have been successful in using the HDR approach to introduce single or limited size sequence substitutions into cells. Our aspiration, however, has been to use HDR to fully humanize genes in our animal models. This endeavor requires sequence substitutions on the order of tens to hundreds of kilobases. To date, sequences of several kilobases have been successfully inserted or replaced. But creation of the best models of human disease may be ones that carry the human gene and its relevant human variants. Some of our target genes are substantial in size (>100 KB) and we have concerns that making the necessary substitutions to replace their animal counterparts may be too challenging. The rapid advancement of CRISPR/Cas9 and the intense focus of many excellent thought leaders make us optimistic that breakthroughs for this application will be just around the bend.
In 2020, I hope to have the opportunity to redeem the investments that have been made to amass genomic information and fine tune our tools. If I am so privileged to once again address the question asked of me in 2010, I anticipate reciting a list of new drugs empowered by our knowledge of the genome so lengthy that the payoff for human health is clearly a no-brainer.