The recent high-profile translational failures in mouse models have highlighted the need for more relevant animal models. Advances in gene editing tools, including the CRISPR/Cas9 system, have enabled the modification of highly translational organisms such as rats and rabbits, and have also greatly reduced model development timelines.
The ‘translational gap’ has received much recent attention and certainly represents a tremendous concern for the biotech and pharmaceutical industries. It has been well documented that the expanding R&D expenditures have not resulted in new drugs.
One area that has received particular attention has been the dependency on animal models. And, indeed, models that poorly represent human disease or produce results that cannot be translated to human studies represent a significant challenge to drug discovery. In particular, mouse models have come under scrutiny, as mice have been the dominant species used for efficacy due to the widespread availability of transgenic and knockout mice. With increasing failures in mice, researchers are now beginning to call the utility of mice as models into question and are beginning to pursue alternative models. Rat models, in particular, have been given a fresh look for drug discovery due to several advantages over mice as well as recent advances in genome editing technologies that now enables genetic modification in the rat.
History’s chosen animal model
Historically, it is rats that have been the preferred animal model for biomedical research. In fact, rats led mice in the number of publications as recently as 2001. The reason for the switch to mice goes back 25 years. The late 1980s saw the development of targeted gene editing. For the first time, researchers were able to knockout, knock-in or otherwise edit the genome, creating animal models with the exact mutations they wanted. The technique relied on embryonic stem (ES) cells, and only ES cells from a few strains of mice were amenable to targeting. Gene editing in the rat was simply not possible, and engineered mice represented an exciting new tool for drug discovery.
Most in vivo assays, particularly behavioural and cardiovascular assays, however, were developed and validated in the rat. Researchers were now forced to adapt these assays to the mouse, often with compromises. And it is perhaps these compromises that have contributed to the recent translational failures.
Some very exciting new gene editing tools have emerged just over the past five years, including zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs) and clustered regularly interspaced palindromic repeats (CRISPRs). These three technologies are so efficient at genome modification, that they have enabled gene editing directly in embryos, eliminating the need for ES cells and permitting the modification of potentially any species.
The conventional technology to create genetic modifications in mice has been reliant on homologous recombination (HR) in embryonic stem (ES) cells. The rate of HR in ES cells is remarkably low, however, and requires selection and screening of many clones. New gene editing tools, such as CRISPR/Cas9 and zinc finger nucleases (ZFNs), are able to produce double-stranded breaks at predefined, highly specific and targeted loci. The resulting generation of a double-stranded break markedly increases the rate of HR by orders of magnitude. This resulting massive increase in efficiency provides several advantages, including faster model generation, reduction of risk and cost and strain and species flexibility.
Faster model generation
Using traditional ES cell technology, it can typically take a year or more to generate a genetically modified mouse. In contrast, using CRISPR/Cas9 enables the creation of a founder animal with even a complicated conditional knock-in or humanisation in as few as five months. The time savings occurs at several steps in the process. First, while proper design of CRISPR reagents requires significant expertise, the assembly of the reagents is very simple, initiated by the production of an oligonucleotide. The simple assembly allows quick generation of the reagents. Secondly, because CRISPR/Cas9 is so efficient, as mentioned, the gene editing occurs directly in the embryo allowing the researcher to skip all of the ES cell work and move straight into microinjection.
Thirdly, ES-cell-based gene editing produces chimeric mice as the initial offspring and these chimeric mice must be backcrossed to wild type mice to segregate the genotypes. Chimeric mice are composed of both modified and unmodified cells. Chimera generation is inherent in the ES-cell based method – modified ES cells are injected into blastocysts and the embryo begins as a mix of modified and unmodified cells. In contrast, CRISPR/Cas9 can be injected directly into embryos at the singlecell stage, so the resulting offspring consist 100% of modified cells and no backcrossing is required. As a result of these combined increased efficiencies, model generation timelines are drastically reduced – by 50% or greater. In the drug discovery world where getting answers and making decisions as quickly as possible is paramount, the potential to shave off as much as a year in model development time is invaluable. CRISPR/Cas9 technology not only enables researchers to get their answers more quickly, however, it also broadens the applications of animal models. A researcher can now ask a question in vivo that they might have otherwise answered in vitro, as the animal model generation timelines were previously too prohibitive.
Risk and cost reduction
One of the biggest risks in the production of modified mice using ES cells is to spend the nearly one year in generation time (and associated costs and labour) only to find out that the modified allele does not get transmitted to offspring. While advancements in technique have reduced this risk, no ES cell-based generation method can produce modified mice with 100% germline transmission, so this gamble must always be considered. Conversely, CRISPR/Cas9 and ZFN technologies deliver modified alleles with 100% germline transmission. This is again a result of being able to modify embryos directly – this ensures that the genetic modification is present in the germline.
While the ‘sticker price’ of model generation using ES cells has been typically lower than for CRISPR/Cas9 and ZFNs, the rapid advancements in these technologies as well as competition in the marketplace is continually driving these costs down. In fact, by the time this article goes to press, nuclease-based targeting may very well have prices comparable to or even lower than ES-cell based services. In any case, these tools are almost certainly more cost-effective when looking at the project as a whole. Additional breeding expense is surely incurred with the additional backcrossing, chimera segregation, and cassette removal required by ES cell-based techniques, and there is a significant opportunity cost of the long generation and breeding times. Further, there is always the very real possibility that a model may lack germline transmission, setting the researcher back tens of thousands of dollars in generation fees along with delays of several months to even years, as well as the potential loss in competitive advantage over another lab or company. These risks are mitigated with CRISPR/Cas9 and ZFN technologies; not only is the model generation piece faster, the researcher can also skip all of the additional breeding required to overcome the shortcomings of ES cell technology, and, with 100% germline transmission, there is no risk of catastrophic loss of the line.
Additional cost savings are realised with the potential to generate both knockout animals as well as knock-in animals simultaneously. CRISPR/Cas9 and ZFNs both create double stranded breaks at targeted loci, and these breaks are then repaired using either nonhomologous end joining (NHEJ) or homologous recombination (HR). Researchers take advantage of these pathways for genetic engineering, namely NHEJ for the production of knockouts and HR for knock-ins. However, it is important to note that both processes occur simultaneously, and NHEJ events still occur even though a researcher is attempting to insert a donor plasmid via HR. Indeed, during founder selection and screening, we routinely observe the generation of knockout animals as a by-product of knock-in generation. This offers the researcher a unique opportunity to create two models (knockout and knock-in) with little to no additional cost or effort.
Species and strain flexibility
As mentioned, conventional gene editing technologies have relied on ES cells, and only mouse ES cells have been amenable to modification. As a result, genetically modified models have been limited to the mouse, and particularly to just a few mouse strains. Mouse background strain has received much attention recently for its profound effect on phenotypes, and background strain must be considered when developing a new model organism. Unfortunately, only a few mouse strains have been amenable to gene targeting via ES cells. CRISPR/Cas9 removes this background strain limitation allowing the researcher to go directly in their strain of choice. At SAGE Labs, we have successfully modified a number of strains including C57Bl6, FVB, Balb/C and CBA CA mice and Sprague-Dawley, Long Evans Hooded, Wistar, Wistar Han, Brown Norway and Fisher 344 rats, among many others. Because only a few mouse strains are amenable for gene targeting in ES cells, the researcher must often backcross their line on to their strain of interest, typically requiring at least 10 generations and an additional year or longer after generation of a model. Even when ES cellbased modification can be done directly in the desired strain, some backcrossing is still required to eliminate chimeras as well as selection cassettes. Using CRISPR/Cas9, the researcher is able to avoid these issues by generating their model in their strain of choice, without the production of chimeras or need for selection.
One of the most exciting aspects of nuclease-based model development is species flexibility. While EScell- based technologies are limited to mice, CRISPR/Cas9 and ZFNs remove this limitation and researchers have now successfully modified many other model organisms including rats, rabbits, silkworms, cows, zebrafish, mosquitoes and many more. For the first time in 25 years, the researcher can now choose the best model organism for their research rather than being forced to adapt to the mouse.
Importantly, this has allowed researchers the opportunity to return to the rat as their default animal model. This affords many advantages over mice and early results from genetically modified rats have produced some very exciting, highly translational results.
Rats are more than just large mice. Numerous anatomical and physiological differences have now been noted between mice and rats, and rats are consistently more representative of humans. The heart rate of a mouse is ~600 bpm, while the rat is less than half of that and much closer to the 70 bpm of humans. The human genome contains 2.9 billion base pairs, rats have 2.75 billion, more than the 2.6 billion of the mouse.
One of the greatest initial challenges to working with mice is that they are particularly tricky to work with on behavioural assays. Mice are less robust and reliable on behavioural assays than rats and mouse behavioural assays typically require cohort sizes up to 50% larger than those needed for rats due to this increased variability. Rats are smarter than mice and perform far better than mice on assays of learning and memory and addiction. In pain assays, rats are less susceptible to anxiety-induced analgesia and again perform much more reliably; in fact, pain research is one field where mice have never surpassed rats in a number of publications.
Rats are more social than mice and perform behaviours that mice do not. One example is juvenile play – young rats will playfully wrestle with one another much as kids do. Mice do not. In a knockout rat model of autism (lacking neuroligin 3), this play behaviour is disrupted – a very translational endpoint. In addition, rats have recently been demonstrated to exhibit very surprising sophisticated behaviours such as empathy and even processing of human speech.
Initial reports from genetically modified rat models have suggested the existence of phenotypes not present in equivalent mouse models. Mouse models of Parkinson’s disease (PD) present one area in particular where mouse models have been lacking. Unlike the genetic mouse models, early work on knockout rat PD models has been encouraging. Pink1 and DJ-1 knockout rats display a progressive loss of dopaminergic neurons in the substantia nigra, with a ~50% loss by eight months of age (Figure 3). This is the first observation of dopamine cell loss in any genetic model of PD. Coupled with the loss of dopaminergic neurons are moderate to severe motor deficits. Specifically, the Pink1 and DJ-1 knockout rats show impaired performance on the tapered balance beam and hind limb fatigue assays, as well as gait abnormalities as assessed by NeuroCube (Figure 3). A subpopulation (~30%) of Pink1 and DJ-1 knockout rats also show a marked, severe dragging of the hind limbs with an onset of around five months of age.
Initial results have demonstrated that the Pink1 knockout rats also possess non-motor and nondopaminergic characteristics of PD. Anosmia, or the loss of the sense of smell, is frequently reported as an early symptom in human PD patients. Awake Pink1 knockout rats displayed deficits in olfactory circuits in response to almond odour (previously demonstrated as ‘pleasurable’) as assessed by fMRI. The fMRI also revealed lack of activation of downstream centres suggesting anhedonia, or lack of pleasure, another symptom frequently seen in PD. Diffusion tensor imaging (DTI) in Pink1 KO rats further revealed altered DTI signals not only in the expected dopaminergic regions, but in non-dopaminergic regions as well. The rat is particularly well suited for imaging studies such as these, as its increased size offers higher resolution as compared to mouse.
Efficacy and safety in the same species
In the past, researchers carrying out drug discovery would screen for efficacy in genetically-modified mouse models and then switch models and assess safety in the rat, due to the large volume of historical safety data in the rat. This methodology relies on extrapolations of mouse dosing to rat and the assumption that this dose would have similar efficacy in the rat. The uncertainty in this approach is less than ideal, as drug efficacy has been observed to be highly variable in different mouse strains, let alone in different species. Genetically modified rat models now help solve this problem, enabling drug efficacy and safety to be performed not only in the same species (rat) but even in the same background strain. There also exists an immense archive of historical physiological data in the rat, and these data can now again be utilised. Combined, these data suggest that genetically modified rat models, enabled by CRISPR/Cas9 and ZFN technologies, offer the potential to serve as more relevant models than mice. Whether or not these models produce drugs that are relevant for the clinic remains to be seen, however, the results so far are encouraging.
Intellectual property and licensing
Even the most relevant, most translational models are of limited use if the researcher lacks freedom to operate. The intellectual property around these new genome editing technologies ranges from crystal clear to relatively murky. Zinc finger nucleases present the clearest case. Sangamo Biosciences retains the IP around ZFNs, and has licensed SAGE Labs for exclusive use in rats, mice and rabbits for research purposes. Sigma-Aldrich and Horizon Discovery both have licences to use ZFNs in cell lines and Sangamo has retained the IP to use ZFNs for therapeutic applications.
The IP landscape surrounding CRISPR/Cas9 is a bit murkier, however. SAGE Labs was one of the first to license the technology. SAGE Labs procured the licence to generate mice as well as the exclusive licence for rats from Caribou Biosciences in the fall of 2013. Horizon Discovery was next to license CRISPR/Cas9 and obtained a licence for in vitro work from ERS Genomics. On April 15, 2014 the Broad Institute was awarded the first patent for the CRISPR/Cas9 system. SAGE Labs, Taconic and Horizon Discovery have all since procured licences from the Broad Institute. CRISPR/Cas9 reagents and related services are also offered by several other companies in absence of licensing.
Next-generation gene editing tools have revolutionised animal model development and present a tremendous potential to rapidly produce more relevant and more translational models. These models offer the promise to help narrow the translational gap and allow researchers to transition their research back to the rat.
Dr Kevin M. Gamber received his PhD in Pharmacology and Physiology from St Louis University followed by a post-doctoral fellowship at Harvard Medical School. Kevin next joined Sigma-Aldrich developing customer education programmes and, since 2011, has been with SAGE Labs developing and commercialising custom animal model generation services using the latest gene editing tools.
Editing the Human Genome: role in functional genomics and translational medicine. Summer 12 READ MORE
Gene Editing: Re-writing genomes with targetable nucleases. Fall 13 READ MORE
Advancing the promise of therapeutic genomic editing. READ MORE
Precise genome editing: the key to a CRISPR drug discovery pipeline? READ MORE
Synthetic lethality and anti cancer drug discovery READ MORE
With synthetic biology, drug discovery is going virtual. READ MORE
Latest developments in high content screening systems. READ MORE
CRISPR Cas9: transforming gene editing in drug discovery labs READ MORE