Genetic modifiers: Why drug developers should care

Sebastian Nijman, Chief Scientific Officer, Scenic Biotech, explains why harnessing genetic modifiers holds great potential to address the present drug target vacuum and why every drug developer should care about them.

According to the National Institute of Health (NIH), approximately 500 of the many thousands of human diseases have an approved treatment. This disappointing result is worsened by the fact that many of these approved treatments have limited efficacy, only work in a minor subset of patients and induce severe side-effects. Despite the tremendous unmet need for new therapeutics, the discovery of high-quality, innovative new drug targets remains challenging. The prospect of harnessing genetic modifiers for drug discovery provides a different perspective and holds potential to address the present drug target vacuum, providing new impetus for the development of novel, transformative therapeutics

The challenge of drug discovery starts with the right target

Even if the painstaking process of target discovery and drug discovery yields an Investigational New Drug (IND), subsequent success rates in the clinic remain low. Historically, almost 90% of all compounds that enter clinical trials fail to gain approval, and this increases to 94% for orphan drugs (those that treat rare diseases and thus likely represent more innovative drugs) ¹. The most common reason for failure of a drug in clinical trials is not safety but efficacy, with the largest attrition rate occurring in clinical Phase II ^2,3. By the time new drugs reach the clinic, most have been shown to engage with the target, displayed efficacy in preclinical models, and have been scrutinised for potential toxicity. Yet most fail to result in sufficient therapeutic benefit in patients. Our limited understanding of the target in the context of the disease explains the staggering drug attrition, and this problem has been noted by industry and academia.

The importance of genetics for drug target discovery

What constitutes a good target? Using drugs that have been approved and those that failed, it is possible to work back to find out what separates the failures from the successes. Based on this, it has been estimated that drugs with targets that have robust genetic evidence, linking them to a specific disease or patient subset, have double the success rate in clinical development ⁴. A clear example of this is the recent success in the treatment of cystic fibrosis (CF) with the drug ivacaftor developed by Vertex Pharmaceuticals⁵. CF is caused by a series of different mutations in the CFTRgene that disrupt the function of the CFTR protein, leading to pulmonary dysfunction. Ivacaftor was developed to stabilise the protein encoded by a specific CFTRmutation found in a relatively minor (4-5%) subset of CF patients. This solid mechanistic link between the disease-causing gene and the drug that directly acts on the faulty protein encoded by this gene was a key success factor in the clinical development. Similarly, the strong genetic evidence that high frequency mutations in certain key cancer genes, such as HER2 and EGFR, are required for tumorigenesis has been a key factor in the development of new oncology drugs. These examples and others have reinforced the notion that a good starting point for the discovery of novel drug targets is genetics.

Genetic modifiers as drug targets: Why is this exciting?

The majority of (inherited) disease is caused by defective, loss of function gene mutations. Indeed, restoring protein function with small molecules, as in the case of the CFTR drug, is notoriously hard and in most cases simply not an option. However, we know from studies that all diseases are modified(ie, affected) by genes that – on their own – do not cause disease. For instance, we recognise that the genetic make-up of each individual determines to a substantial extent their susceptibility to diseases such as cancer and heart disease. Furthermore, clinicians that treat patients suffering from (rare) genetic disorders frequently encounter discordant phenotypes (ie, differing between individuals) of patients with the exact same underlying mutations.

Inherent differences in disease susceptibility and disease severity between individuals carrying the same pathogenic mutations are due to genetic modifiers: genes seemingly hiding in the background that influence disease-causing processes. Genetic modifier is a general term used to describe a gene (variant) that affects the function of another gene. This effect can be enhancing the function, reducing it, inhibiting it, or even changing it to another function. In the context of disease and their relevance for drug discovery, genetic modifiers are of interest as they can reduce the impact of disease-causing mutation and are therefore, sometimes referred to as suppressor genes.

Blocking the activity of proteins encoded by genetic modifiers holds promise as a new way of suppressing disease. Although the concept of genetic modifiers is by no means new, it has hitherto garnered relatively little interest outside academia. This is not due to the appeal of the concept but because of the lack of suitable technologies to find genetic modifiers, which has meant that very few examples (see below) have successfully reached the clinic. However, our ability to systematically unlock the science of genetic modifiers holds promise as it represents an untapped reservoir of potentially high-quality drug targets.

How to find genetic modifiers?

The approaches that are currently proven to identify genetic modifiers come in three broad categories: human/population genetics, model organism genetics, and functional genomics. This short article cannot do justice to a full overview of these three extensive areas of research and I will not attempt to do so. Below, I will briefly explain how these approaches can be used, including some examples, and point out their strengths and limitations.

 Human (population) genetics

Thousands of genome-wide association studies (GWAS) in healthy individuals and patients have shown that the genetic basis for common diseases is multifactorial and complex [ref]. For instance, over 140 gene variants have been reported to influence asthma susceptibility and at least 108 genes have been associated with schizophrenia [refs]. In other words, a plethora of genetic modifiers play an important role in explaining the variability of common disorder. However, most of these genes contribute only a small proportion to the disease (risk), questioning their suitability as drug targets. Yet, in some isolated cases these type of population genetics studies for common disorders have identified strong genetic modifiers as drug targets. One of these studies analysed an existing cohort of 3,500 multiethnic individuals in the Dallas Heart Study to identify a genetic basis for very low plasma LDL cholesterol-levels ⁶. They gave themselves a head start by placing their bets on one gene that they hypothesised may be a genetic modifier, the PCSK9 gene. PCSK9encodes a serine protease that had already been directly linked to hypercholesterolemia. They got lucky, and identified two loss-of-function (LOF) mutations in PCSK9. Inspired by this, antibodies blocking PCSK9 were developed and the first two (alirocumab and evolocumab) were approved by the US FDA in 2015 for the treatment of high LDL cholesterol.

A recent and more extensive population genetics study further demonstrates the power of genetic modifiers as potential drug targets for treating disease. A group led by Stephen Friend, Eric Schadt and Rong Chen searched in the genomes of over half a million healthy adults for one of 874 known mutations that should have given rise to a severe childhood disease, but somehow had been genetically repressed ⁷. This ambitious and original study identified at least 13 healthy adults carrying mutations for eight debilitating diseases including CF and Pfeiffer syndrome, which causes premature fusion of bones in the skull that can lead to early death. The landmark conclusion of this study is that disease suppression by means of genetic modifiers may be more common than anticipated, and can result in severe diseases becoming subclinical.

However, in this study they were unable to uncover the identity of these valuable genetic modifiers. It would have required extensive access to the individuals, their families, and their medical records, making it challenging to pursue for practical and ethical reasons. Even with extensive follow-up, teasing out the genetic modifier(s) at play may prove to be impossible. The take-home from this study has a mixed message for drug developers: It demonstrates the potential power of exploiting genetic modifiers as drug targets for treating diseases. However, to ultimately identify genetic modifiers of disease that could be exploited as drug targets using these human genetics approaches, access to massive datasets coupled with detailed health records on a global scale is required.

Regardless of the challenges of human genetics to identify high confidence drug targets, especially for rare disease for which only small cohorts are available, these and other studies have sparked a number of large-scale initiatives (eg. FinnGen and Genomics England). In the coming years we will learn if these initiatives will deliver on their promise of identifying novel genetic modifier drug targets.

Genetics in model organism

Genetic modifiers exist along the spectrum of inherited rare mono-genetic diseases to common poly-genetic disorders. However, analysing patients and their genomes, that have been shaped in complex ways by the forces of evolution, is challenging for finding strong monogenetic modifiers or rare genetic modifier variants for disease. Instead, identifying genetic modifiers that correct disease-associated phenotypes is much more straightforward in model organisms, including mammals ⁸. Two key advantages of using model organisms over studies in humans are that they are typically inbred, meaning that they are genetically very similar, and they are amenable to selected crosses, meaning that the underlying gene variant that causes the (modified) phenotype can be isolated.

A classical approach to identify disease modifiers is to perform crosses between (sub)strains that display variability in the severity of a disease. Phenotypic and genetic analysis of the progeny allows researchers to home in on the genetic modifiers. This approach has been successful in identifying genetic modifiers with relevance for human disease. For instance, the APC^minmouse model, which is used for studying familiar predisposition to colon cancer, has increased incidence of colon polyps, which are considered a pre-malignant stage of colon cancer. However, the number of polyps varies strongly depending on the background strain that is used. By performing classical genetics studies, the secretory phospholipase A2 (Pla2g2a). was identified as a genetic modifier ⁹. Subsequent studies provided further insight into how this secreted phospholipase contributes to colon tumorigenesis ¹⁰. Another striking example concerns Duchenne’s muscular dystrophy in Golden Retriever dogs. Whilst breeding a strain with a mutation in the DMD gene that causes Duchenne’s disease, two dogs were identified with a surprising mild phenotype ¹¹. By crossing these dogs further, and performing detailed genetics analysis, a single point mutation in the Jaggedgene was identified. Interestingly, this indicates that increasing muscle regeneration can at least partially offset the degeneration observed in the disease.

However, the approach in animal models is still highly challenging, precluding it as a systematic method for the hunt for genetic modifiers. Crossing animals is slow, costly and does not guarantee success as the disease variability may itself be multifactorial and hence hard to genetically dissect. Also, these examples still rely on the spontaneous emergence of genetic modifiers, and although one can use experimental agents such as chemical mutagenesis to increase genetic variability, many of the same drawbacks in terms of cost and throughput remain. Indeed, to take advantage of the full spectrum of potential genetic modifiers that is locked into our genome, we need to use methods that lift the genome variability out from its evolutionary constraints.

Functional genomics

So how can we identify genetic modifiers in a more systematic – and importantly – a commercially viable way? At Scenic Biotech we believe that the answer is to use functional genetics approaches. This means using genetic screens in human cells to systematically interrogate all the genes in the genome to see if any of them may act as high-value genetic modifiers for a disease of interest. This can be done for many different diseases. Better yet, the approach opens up the entire genome for interrogation, in contrast to the population genetics approaches that rely on genes that have evolved by natural selection to be modifiers within a population. Every gene has the potential to be a modifier of another gene in ways that the textbooks cannot predict. This means that functional genetics approaches can identify high-value genetic modifiers for all diseases, not just the common ones.

We make us of a proprietary screening technology, that we call Cell-Seq, to identify genetic modifiers of disease ¹². Cell-Seq employs human ‘HAP1’ cells that have a single copy of each of the chromosomes, rather than the usual two ¹³. This is referred to as haploid or monoploid. These cells can be easily genetically engineered into cellular disease models by changing the status of disease-causing genes. For instance, by inactivating genes that cause metabolic disorders (eg. lysosomal storage disorders), normal cells are turned into cells that display many of the hallmarks of patient cells. Such cellular phenotypes can be quantified using fluorescent antibodies or other markers that measure the status of cellular processes that are closely linked to the disease.

Thanks to the haploid status of HAP1 cells, a powerful method of introducing mutations across the genome, called retroviral gene-trap mutagenesis, can be employed. By making use of a virus to introduce mutations, not only is the normal gene function disrupted, but also a ‘flag sequence’ is inserted into the genome. This allows for subsequent identification of the putative genetic modifier using off-the-shelve high throughput DNA sequencing methods. Although each HAP1 cell will have only a single gene inactivated, by using millions of cells in a single experiment, every gene of the 20 thousand or so in the human genome is hit dozens or even up to thousands of times. Each hit has the potential to uncover a genetic modifier of the disease phenotype under study. Thus, cells that carry a mutation in a genetic modifier that affects the cellular phenotype can be isolated.  By determining the mutations of 10s of millions of isolated mutants, a high resolution ‘snapshot’ of all the modifier genes for the particular cellular/disease readout is created.

This highly scalable approach can be employed with endless variation and is generating a browsable phenotype-genotype ‘genetic modifier map’ across dozens of pathways and processes. Using this approach at Scenic aimed at specific pathways in engineered disease models, we have interrogated numerous pathologies with unmet medical needs, and initiated drug discovery programs based on completely new targets.

Looking ahead

The identification of high-quality drug targets using new approaches, including functional genetic screens and human genetics to fully exploit the concept of genetic modifiers is building momentum. Since the inception of Scenic Biotech in 2017, which was founded with the explicit aim to pioneer the genetic modifier space for novel targets, several other companies have entered the scene (eg Maze Therapeutics, Adrestia). These developments are a testament to a shared vision that the era of genetic modifiers as a therapeutic concept is finally coming of age. Together, we have the potential to fill the current ‘drug target vacuum’ for the full range of human diseases.

Volume 22, Issue 1 – Winter 2020/21

About the author

Sebastian Nijman PhD is Founder and Chief Scientific Officer, Scenic Biotech. He co-founded Scenic Biotech in 2017. He previously co-founded and was COO of Haplogen GmbH. He has held faculty positions at Oxford University and the Research Center for Molecular Medicine (CeMM) in Vienna. He was a postdoctoral fellow at the Broad Institute and received his PhD training at The Netherlands Cancer Institute.

References

1. Wong, C. H., Siah, K. W. & Lo, A. W. Estimation of clinical trial success rates and related parameters. Biostatistics(2019). doi:10.1093/biostatistics/kxx069
2. Harrison, R. K. Phase II and phase III failures: 2013-2015. Nat. Rev. Drug Discov.(2016). doi:10.1038/nrd.2016.184
3. Arrowsmith, J. & Miller, P. Trial Watch: Phase II and Phase III attrition rates 2011-2012. Nature Reviews Drug Discovery(2013). doi:10.1038/nrd4090
4. Nelson, M. R. et al.The support of human genetic evidence for approved drug indications. Nat. Genet.(2015). doi:10.1038/ng.3314
5. Davis, P. B., Yasothan, U. & Kirkpatrick, P. Ivacaftor. Nature Reviews Drug Discovery(2012). doi:10.1038/nrd3723
6. Cohen, J. et al.Low LDL cholesterol in individuals of African descent resulting from frequent nonsense mutations in PCSK9. Nat Genet37, 161–165 (2005).
7. Chen, R.et al.Analysis of 589,306 genomes identifies individuals resilient to severe Mendelian childhood diseases. Nat Biotechnol(2016). doi:10.1038/nbt.3514
8. Nadeau, J. H. Modifier genes and protective alleles in humans and mice. Current Opinion in Genetics and Development(2003). doi:10.1016/S0959-437X(03)00061-3
9. Dietrich, W. F. et al.Genetic identification of Mom-1, a major modifier locus affecting Min-induced intestinal neoplasia in the mouse. Cell(1993). doi:10.1016/0092-8674(93)90484-8
10. Schewe, M. et al.Secreted Phospholipases A2 Are Intestinal Stem Cell Niche Factors with Distinct Roles in Homeostasis, Inflammation, and Cancer. Cell Stem Cell(2016). doi:10.1016/j.stem.2016.05.023
11. Vieira, N. M. et al.Jagged 1 Rescues the Duchenne Muscular Dystrophy Phenotype.Cell163, 1204–1213 (2015).
12. Brockmann, M. et al.Genetic wiring maps of single-cell protein states reveal an off-switch for GPCR signalling. Nature(2017). doi:10.1038/nature22376
13. Carette, J. E. et al.Generation of iPSCs from cultured human malignant cells. Blood115, 4039–4042 (2010).