Jay Bradner on open source drug discovery, cancer, and targeting the epigenome

Discussion |

Jay Bradner on open source drug discovery, cancer, and targeting the epigenome

Lisa Girard

 

James “Jay “ Bradner, M.D. is an Assistant Professor of Medicine at Harvard Medical School, Investigator and Staff Physician at the Dana-Farber Cancer Institute, Associate Member of the Chemical Biology Program at the Broad Institute and Affiliate Faculty member of the Harvard Stem Cell Institute. His lab is focused on chemical biological approaches to studying chromatin- associated complexes with a focus on oncology. His achievements in promoting open source drug discovery have the potential to change how cancer and disease therapies are developed.  Jay spoke with the StemBook editor, Lisa Girard, about his work creating chemical tools to fight cancer, his innovative approach to distributing these tools as a means of accelerating discovery, and how this just might change how pharma thinks about making medicines.

 

Can you tell me a little bit about the chemical biology approach your lab takes toward identifying potential cancer therapeutic targets?

 The focus of my laboratory concerns the discovery, chemical optimization, mechanistic characterization, broad dissemination, and ultimately clinical translation of small molecules that modulate gene regulatory pathways. Every cell in the human organism possesses almost identical copies of the same genome, yet of course there is dramatic phenotypic diversity on a cell-by-cell basis. This is because cell state is governed by hierarchies of coordinated transcriptional programs. Cancers which emerge from these tissues overlay transcriptional programs allowing indefinite growth and survival. We are therefore interested in mining malignant gene regulatory pathways for both mechanistic and therapeutic insights. The chemical biology approach leverages unique attributes of small molecules. Experimental use of chemical tools allows temporal, and even spatial, control of cell state by interacting with protein targets and, in the case of our research, controlling a gene regulatory pathway output. A nice feature of the chemical biology approach as regards to research such as ours is that if you are lucky enough to find this holy trinity of chemical biology: a small molecule, a relevant phenotype, and a protein target of interest, insights generated in translational models can on occasion prompt immediate human clinical investigation.

 

 Can you tell me about some targets that you have been able to attack with this approach?

 My laboratory was established only in 2008, so please consider this a work in progress. I organized the laboratory to integrate the disciplines of organic chemistry, protein biochemistry, chromatin biology, and cancer translational science – all around one group meeting conference table. In that regard, concepts or targets enter into the laboratory and can rapidly elaborate small molecule probes through a discovery and iterative optimization cycle. Though we have only just started our research, already we have created chemical tools for several very interesting targets in chromatin biology.  Included among these are the first direct inhibitor of the Notch complex, the first bromodomain inhibitor targeting the BET family of human bromodomains, open-source tools for the study of lysine methyltransferases and numerous chemical tools for histone deacetylases.

 

And which is the Notch pathway inhibitor?

 In 2009, we reported a direct-acting inhibitor of the Notch transactivation complex, called SAHM1, which is a constrained alpha-helical peptide designed and characterized in collaboration with the laboratory of Greg Verdine.   

 

Wasn’t there also a compound that you identified that could interact with BRD4? Could you tell us a little bit about that and what you were able to do with it?

 Let me start by saying that the early sequencing data from human tumors is sort of scary. There are already over a million somatic alterations identified in cancer genomes affecting more than 400 gene products that may contribute to the pathogenesis of cancer. Cancer cells are therefore impossibly complex and tumors even more so, exhibiting significant intra-tumoral heterogeneity causing rapid drug resistance.  As scientists, we must consider that the present paradigm of therapies targeting somatically-altered gene products may be insufficient to cure this disease. We have, therefore, taken a slightly different angle. A glass half-full approach is to consider that if 5-130 genes are somatically altered in a given cancer cell, that means 24,000 genes are normal.  Could we target cellular memory, or cellular state, directly with new types of small molecules, that can relieve the cancer of this memory of being cancer, perhaps inciting natural, embedded transcriptional pathways to prompt cellular differentiation? With this thesis in mind, we have assembled around the challenge of targeting “molecular memory”, an exciting and emerging opportunity in cancer drug discovery born from the mature fields of transcriptional and chromatin biology. At the molecular level, cells remember what type of cells they are by chemically modifying the epigenome.  Insights from the literature prompted our laboratory to become very excited about one family of memory proteins called bromodomains. Bromodomains may be conceptualized as post-it notes placed throughout the epigenome, in particular, at active and open regions of the genome loaded with expressed genes. Conceptually, as the cell passes through mitotic cell division and it unwinds its DNA and divides into two daughter cells it remembers it’s cancer. Molecularly, chromatin-bound BRD4 directs the re-boot of transcriptional programs that re-establish cell state.  We therefore became interested in developing direct-acting inhibitors of BRD4 that could displace this factor from the epigenome.

 In 2010, we reported the first bromodomain inhibitor, which we call JQ1 for Jun Qi, the originating chemist in my laboratory.  In collaboration with Prof. Stefan Knapp at Oxford University, we demonstrated that JQ1 binds to BRD2, 3, and 4, a subset of the bromodomain family of proteins, and functions to remove them from the epigenome. In a genetically defined subset of lung cancer cells in culture, JQ1 prompts a radical change in cell state in which the cells express keratin and turn into a squamous epithelium cells on the bottom of the Petri dish. Effectively, the drug prompts a genetically-defined type of lung cancer addicted to BRD4 to turn into a squamous epithelium that can be peeled away, as a non-malignant tissue. In mice bearing these tumors, JQ1 prompts a dramatic tumor regression accompanied by a significant survival benefit.  As bromodomains function through protein-protein interactions that are traditionally thought of as “undruggable”, or just beyond the reach of conventional drug discovery, this project has been exciting as a declaration of open-season on epigenetic reader proteins.

 

 With as broad a class as bromodomains, how are you able to get specificity that can offset any kind of toxicity?       

 JQ1 is a highly selective ligand for the BET family of human bromodomains. Biochemical studies in our lab, and others, have demonstrated tight binding to only these targets, confirmed later by proteomic analysis.  Very recently we’ve come to understand the specificity of JQ1 for the BET subfamily at the molecular level. The bromodomain is a very rigid protein fold and JQ1 adopts a three dimensional structure that is perfect fit only for the BET bromodomain active site.

 

 Once you identified JQ1, what were your next steps?

 The activity of JQ1 in BRD4-rearranged lung cancer has prompted a real motivation by our group to deliver a drug-like derivative of JQ1 to the clinic. Dr. Qi and I have now completed the characterization of drug-like derivatives and are excited that in the next two months a clinical trial of JQ2 (now called TEN-010 at Tensha Therapeutics), will soon be initiated in cancer. Beyond our own research, we perceived that others in the chromatin biology community would be interested to assess the role of that bromodomain in their disease or biological model of interest. We therefore have adopted a strategy of open innovation to allow and enable a free distribution of JQ1 as a chemical tool to catalyze research in this exciting and raidly emerging area. Through this approach, we have collaboratively gleaned insight into the bromodomain in a number of provocative pathologic contexts, such as HIV latency, Human papillomavirus pathogenesis, and spermatogenesis.

 

 After you identified JQ1, you published its structure, how to make it, and you made it available. Given these kinds of actions, and how successful this effort was out of your lab I’m sure it will inspire other similar efforts. How could you imagine this changing how pharma is going to have to think about drug discovery and their interactions with academics? 

 As you know, pharmaceutical drug discovery is a famously secretive and private process owing to the incredible financial stakes accompanied by success in first developing a new class of drug molecule.  This is also a highly uncertain process, with a significant failure rate.  Commonly, failure to successfully develop a drug leads to shelving projects.  Like JQ1, these prototype drugs could be incredibly informative chemical tools, which if released for broad study could resuscitate pharmaceutical drug development toward a new and unrecognized opportunity. It is a luxury for us, in academia, to experiment with an approach that prioritizes immediate and unrestricted access to chemical probes emerging from our group. The data we are collecting argue that this open-source approach to drug discovery profoundly accelerates discovery horizons, provides early independent pharmacologic target validation, generates vast knowledge and guides drug development on many levels. I sense that open source drug discovery has a transformative potential to deliver sophisticated biology to abandoned drug molecules, breathing new life into a program that was once a source of major financial and creative investment. Beyond this opportunity, the experience of distributing, studying, and translating JQ1 leads me to speculate whether a private or secretive strategy can even be competitive in the future as others adopt open source drug discovery. So rapid are the insights distributed by this process, that those participating in open source drug discovery should have a competitive edge in early drug development where time is of the essence.

 

What would be your idealized view of how pharma companies could catch up and work most effectively?

 It’s interesting, I have been contacted by a number of major pharmaceutical companies who are very taken with our approach and are now going through the process of understanding on what scale they, too, can experiment with this strategy. I advise companies with an interest in exploring open source drug discovery to identify projects where a chemical tool might be made freely and widely available and where they have an actionable drug molecule ready for immediate clinical investigation. In that regard, data is generated freely with the chemical tool, but importantly, patients can benefit by a drug molecule waiting in the wings for human clinical investigation.

 

So, a pharma company could take a compound they already own and have sitting on the shelf and it could be repurposed to greater utility. How could this fit in well with their existing business model?

 I’m very sensitive to the commercial and financial commitments of pharmaceutical companies. As a doctor, I’m very glad that there is a large financial reward for those who make meaningful medicines, as it brings organized, creative, and well-resourced scientists to bear on historic challenges of drug discovery, particularly in cancer. Open-source drug discovery is not really at odds with their financial ambitions. In fact, it is very likely a way of generating a clinical, and therefore perhaps a commercial, trajectory for technologies that have been abandoned.

 Jay, thanks for your time.

 

 Sure