The drug discovery process can be long and gruelling, but when backed by genetic evidence, drugs are twice as likely to make it to the clinic. And genomics isn’t the only kind of data that contributes to a more efficient pipeline; multi-omics data such as transcriptomics and proteomics is transforming the drug discovery landscape.
In our latest webinar, our expert speakers discussed the use of multi-dimensional data in drug discovery and showcased how integrating diverse datasets throughout the process can lead to more informed pharmaceutical decision-making and accelerated drug development. Here, we summarise the talk given by Tom Lanz (Senior Director of Multi-omics and Biomarkers, Pfizer), titled ‘Integrating Omics into Drug Discovery: from Target ID to Phase 3 and Beyond’. To hear Tom’s talk in full, as well as the other talks from this webinar, please follow this link.
The role of omics in target discovery
Various omics can be used throughout the drug discovery process. These technologies can address problems related to target identification/validation and preclinical development (including probing efficacy and safety), and can be used at multiple points throughout the clinical trials phase. Tom discussed a few examples of the use of multi-omics and multi-dimensional data in his talk, where he detailed the integration of these methods from target ID, phase 3 and beyond.
Figure 1. Image detailing the steps of the drug discovery process where multi-omics can be used. Screenshot taken directly from FLG webinar.
Tom first discussed the use of multi-omics in target validation, specifically in finding a target for schizophrenia. The team were interested in parvalbumin interneurons, as the dysfunction of these cells can cause impairment of cognitive circuitry in schizophrenia patients. However, this sub-population of neurons is rare, and it is difficult to gain much information using traditional RNA-seq. To get around this problem, the group used laser-capture microdissection prior to RNA-seq to first identify this limited subpopulation of cells and characterise their druggable transcriptome.
This method allowed the team to pinpoint drug targets related to the condition, notably GluN2C/D. These are two subunits of the glutamate receptor, and the team had to figure out which was relevant to schizophrenia in humans. The LCM data was integral in this, and finally identified GluN2D as a potential drug target.
Multi-omics in biomarker discovery
Biologic therapies can elicit cell-mediated immune responses. This response can cause tissue injury and inhibit the activity of therapeutic molecules. Therefore, it is important to find relevant biomarkers to predict and prevent this response. The team established an in vitro method to enrich for T cells soon after their activation, in order to identify the cells that respond to antigen exposure. They did this in two different ways; enriching for T cells, and by enriching for activation markers.
Simply taking a profile of circulating blood cells can lead to a diverse array of cells that is hard to interrogate, and as such, single-cell omics is a useful way forward. The team used RNA-seq with VDJ capture in order to sequence the T cell receptors, and successfully identified T cell clones after exposure. They compared bulk RNA and DNA and single-cell data, and concluded that the single-cell clonal expansion experiment reflected that of a bigger population – addressing one of the concerns around using single cells. They concluded that enriching for T cells or isolating an activation marker, or focussing on expanded clones within biomarker data, were all efficient options for biomarker identification. This resulted in an early identification of an immune response biomarker, for which the next step is to validate the marker at the protein level.
Gene editing vectors and genotoxicity – phase 3 and beyond
Finally, Tom talked about the use of the gene therapy vector AAV (adeno-associated virus), which is canonically non-integrating. However, despite this label, a very small percent of the vector DNA can in fact integrate, specifically in mice. The underlying mechanism for this is currently unknown. The epigenetic state of one’s cells appears to be a big factor in determining where AAV can be integrated.
In the early days of gene therapy, some vectors were linked to cancer causing mutations. It is not thought that this is the case with AAV, but integration and subsequent malignancy has been seen in mice in a particular cancer-associated locus. This locus, however, is not found in humans or other mammals. That said, it is still vital to explore this to ensure the safety of the method.
The team compared three different methods to measure integration – target enrichment sequencing, whole genome sequencing and shearing extension primer tag selection. They were looking for a junction between vector DNA and host DNA, to identify hotspots for integration within the host genome. Whole genome sequencing is very computationally expensive, and so is not necessarily the best method. To combat this, they enriched for vector DNA, sequenced this and assumed that additional host DNA was an integration site.
They saw that vector DNA integrates randomly throughout the genome, not in one specific locus. This randomness was also seen in regard to which part of the vector itself integrated. This showed that there was no specific cancer-associated locus prone to integration in humans like there is in mice.
Figure 2. Image showing AAV integration sites. Screenshot taken directly from FLG webinar.
Ultimately, the team concluded that there are very few robust hotspots and little enrichment in cancer genes. There was also no evidence of clonal expansion. This goes to show that multi-dimensional data can be used to assess samples and provide context in order to assess the efficacy and toxicity of new therapeutics.
Going forward, Pfizer has programs in Phase 3 and the plan for therapy development is to look for integration in pre-clinical studies. Beyond Phase 3, there are plans to evaluate integration in patients using liver biopsies.
Tom: ‘So far, the data looks promising for AAV. I think this is a novel therapy with life changing benefits and we will continue to monitor the risk. This is one way that omics can help us keep an eye on our compounds post-Phase 3.’
Q&A Highlights
Q: What is the biggest challenge you find using these multi-omics methods to make actionable drug discovery insights?
A: In general, the cost of typical sequencing has been going down over time but some of the newer methods, single-cell and any spatial omics, they are very expensive. You invest a lot of money and time to do these experiments, so finding the right questions to apply these resources is key.
Q: You mentioned your group use different omics, i.e., proteomics/epigenomics etc. Do you have plans to use metabolomics at any point?
A: There are a couple of programs in which we have started to use metabolomics when we have a defined reason to do so. We want to use proteomics or transcriptomics if we are fishing for a biomarker. The problem with metabolomics is it’s harder to build back the path from your target to your biomarker if you don’t have a good view of how that metabolite signature might have been generated. It’s early days compared to the proteomics or transcriptomics space.
References:
All information obtained from Webinar 2: Integrating Omics into Drug Discovery: from Target ID to Phase 3 and Beyond from Front Line Genomics’ Genomics in Drug Discovery ONLINE series.
Gonzalez-Burgos, G., Fish, K.N. and Lewis, D.A., 2011. GABA neuron alterations, cortical circuit dysfunction and cognitive deficits in schizophrenia. Neural plasticity, 2011.
Arion, D., Corradi, J.P., Tang, S., Datta, D., Boothe, F., He, A., Cacace, A.M., Zaczek, R., Albright, C.F., Tseng, G. and Lewis, D.A., 2015. Distinctive transcriptome alterations of prefrontal pyramidal neurons in schizophrenia and schizoaffective disorder. Molecular psychiatry, 20(11), pp.1397-1405.
Garst-Orozco, J., Malik, R., Lanz, T.A., Weber, M.L., Xi, H., Arion, D., Enwright III, J.F., Lewis, D.A., O’Donnell, P., Sohal, V.S. and Buhl, D.L., 2020. GluN2D-mediated excitatory drive onto medial prefrontal cortical PV+ fast-spiking inhibitory interneurons. PLoS One, 15(6), p.e0233895.
Weickert, C.S., Fung, S.J., Catts, V.S., Schofield, P.R., Allen, K.M., Moore, L.T., Newell, K.A., Pellen, D., Huang, X.F., Catts, S.V. and Weickert, T., 2013. Molecular evidence of N-methyl-D-aspartate receptor hypofunction in schizophrenia. Molecular psychiatry, 18(11), pp.1185-1192.
Mingozzi, F. and High, K.A., 2017. Overcoming the host immune response to adeno-associated virus gene delivery vectors: the race between clearance, tolerance, neutralization, and escape. Annual review of virology, 4, pp.511-534.
Chandler, R.J., LaFave, M.C., Varshney, G.K., Trivedi, N.S., Carrillo-Carrasco, N., Senac, J.S., Wu, W., Hoffmann, V., Elkahloun, A.G., Burgess, S.M. and Venditti, C.P., 2015. Vector design influences hepatic genotoxicity after adeno-associated virus gene therapy. The Journal of clinical investigation, 125(2), pp.870-880.
