4th Medicinal Chemistry & Protein Degradation Summit: conference highlights

Miss the Medicinal Chemistry and Protein Degradation Summit (28—29 October 2019; London, UK)? Check out these highlights from RxNet Editor, Zoe Campbell.

Like Comment

Day 1
Implementation of artificial intelligence in medicinal chemistry
Medicinal chemistry case studies

Day 2
Protein Degradation

Day 1

Monday 28th November marked the first day of this two-day event — the 4th Medicinal Chemistry and Protein Degradation Summit (28—29 October 2019; London, UK). Even before the day officially began, the atmosphere was full of excitement in anticipation for the series of talks and presentations ahead of us. A new addition to this year’s summit was the focus on protein degradation — an emerging field that many are pursuing in their research efforts.

Track 1 - Implementation of artificial intelligence in medicinal chemistry

Technically not part of this track, but the first of the keynote presentations that began the day, this talk certainly underlined the use of artificial intelligence (AI) in drug discovery. Alexander Hillisch (VP, Head of Computational Molecular Design at Bayer AG; Leverkusen, Germany) focused his presentation discussing Bayer’s in silico ADMET platform, which employs machine learning (ML) to assist in the company’s early phases of drug discovery. In any of the approaches the company took in their preclinical work, Hillisch made sure to highlight how the triad of data, algorithms and descriptors was applied why it was essential for the implementation of their machine learning models.

 Artificial intelligence in drug discovery: an interview with Ola Engkvist

At Recursion (UT, YSA), as described by Mike Genin (VP of Chemistry at Recursion), they have been exploring the application of phenotypic screening in drug discovery — trying to figure out a way to identify successful hits faster. Using phenomics data, along with AI, they are working to build models that can bridge between the low dimensional and sparely filed datasets. Traditional predictive models are limited to the nature of training data and cannot be used to predict anything that falls outside of the chemical space of that training data. Integration of AI at this stage of drug discovery may help reduce the time spent before a new drug product is approved, especially in lead optimization.

With all the hype and excitement around AI and ML, it is important to not devalue real/rational intelligence (i.e., the real data that is used to drive forward AI). “Artificial intelligence is driven by real intelligence,” remarked Johnathan Mason (Senior Research Fellow at Sosei Heptares; Cambridge, UK) during his presentation. Without structural-based studies and the structures obtained from this work, structure-based drug design would not be possible. And without the lessons learnt from this, we would not understand the importance of the factors that provide the real evidence essential for the development and application of AI and ML models, such as how the water network controls GPCR ligand binding and selectivity.

Ask the Experts: artificial intelligence

After reading the abstract in the program for this next talk, I was interested in learning how these researchers, from CAS (OH, USA), successfully improved their predictive accuracy of ligand-biological activity by 31%. Predicting bioactivity from the molecular structure is seen as a challenge in drug discovery, but ML has been tipped as a way to improve the outcomes of this key step. Why? — ML could be utilized to classify the ligand-based protein interactions. Yugal Sharma (Senior Director at CAS) noted that this challenge could technically be overcome in one of two ways, (a) with more data or (b) with better data. Data quality is obviously the better choice of the two, as more data that is still poor quality would not improve the outputs of model prediction. At CAS they substituted the fingerprints used in the common machine learning models (e.g., native bayer, tanimoto, graph convolution and support vector machines) with their own and measured the impact on the accuracy of the predictive data. Here, they saw more than a 30% increase in accuracy. 

Back to top of page

Track 2 - Medicinal chemistry case studies

The last of the morning’s three keynote presentations were all about a novel therapeutic avenue for Type 2 diabetes. Jeyaraj Duraiswamy (Jubilant Biosys; Karnataka, India) began his talk by highlighting to us why there is a need for new a Type 2 Diabetes drug, due to the in issues seen with those approved and on the market. For example, Avandia has been subject to review by the EMA due to concerns of it leading to heart attacks. Metformin, one of the most common drugs used for diabetes intervention, is also known to be associated with gastrointestinal problems. Hence, there is a need for a novel drug with a new mode of action. At Jubilant, they have identified GPCR40 (GP40) as an attractive candidate. They began the designing stage of this work by analoging the natural ligand of GP40. To avoid the issues of a low free drug concentration, a lack of selectivity against PPARs and possible metabolites forming, they worked to optimize the different functional groups present on GP40. The lead compounds developed have shown good selectivity and no toxicity. Preclinically, they demonstrated good insulin secretion and the results indicated a reduction in glucose levels.

"From idea to IND”

Vicky Steadman (GM and Business Line leader at Eurofins Discovery; Essex, UK) gave an interesting talk on the changing and evolving nature of CROs in today’s drug discovery
landscape. Taking Eurofins as an example, she set to show how CROs were no
longer being used solely to carry out certain tasks but now they were being
seen as an integral part of many drug discovery programs, helping drive
innovation. As the shift towards a more equal partnership between all the key
players in this industry, we have seen a further integration of CROs in this
field, moving from a transactional-based view of CROs to a solution-based one.

The landscape of the UK drug discovery industry: an interview with Chris Molloy

Shifting to academic drug discovery, Alexander Dömling (Independent Group Leader at the University of Dortmund; Germany), took us through the discovery of three small molecule protein—protein interaction (PPI) agonists, all based on structural biology. Overall, the approach in the lab was largely computational. The antagonist generated for IL-17a (involved in the pathogenesis of various inflammatory diseases) was very interesting. In this, they employed automation and AI to drive an iterative cycle of synthesis → screening → quality control. IL-17a is a soluble, extracellular homodimer that is very difficult to target with small molecules, due to its flat shape. Their AI-enabled approach led them fragments that served a building block for multiple chemistries. Continuing with the application of AI, they moved this forward to drive fragment expansion with AI, which led them to the discovery of high-affinity 1L-17a agonists.

Back to top of page

Day 2 - Protein degradation

Day 2 of the summit was entirely devoted to the expanded scope of protein degradation. Beginning the session, Chris Yates (Associate Director of Chemistry at Kymera Therapeutics; MA, USA) drew our attention to the fact that with our current methods, we are generally limited to targeting only about 15% of the genome. What Kymera are trying to do is access and drug the remaining 85%, using degraders. Only a small proportion of E3 degraders have been proven to successfully degrade proteins, again this is another thing that Kymera are working to improve. Yates focused his presentation on their work surrounding an IRAK4 degrader – IRAK4 forms a complex with Myd88, which is known to be mutated in about 30% of primary CNS lymphomas. He mentions that balancing key parameters (such as oral pK) allowed them to advance from the first generation of IRAK degraders the second generation.



The Crews Lab at the University of Yale (CT, USA) has been working on Proteolysis Targeting Chimera (PROTACs) for the past two decades. In this talk, George Burslem (Research Fellow a Crew Lab) featured new approaches to PROTAC-mediated protein degradation. In their work, they have extended the degradation approach to also target transmembrane proteins, which was not thought to be possible. With PROTACs they achieved sustained pathway inhibition. They initially began their work on nutlin in 2002 but later abandoned research efforts until recently, where they returned to look at this molecule again and built upon their previous work with the new knowledge they gained over time. Here they discovered that suppressing both MYC and p53 together with a nutlin-based PROTACs leads to positive results – “two heads are better than one”. The talk was concluded mentioning that the first PROTACs to reach the clinic (ARV-110 and -471) have entered the preclinical efficacious range associated with tumor growth inhibitors and displayed no dose-limiting toxicities.

Overcoming challenges in protein degradation was the theme of the panel discussion from day 2. George Burslem, Chris Yates, Michal Plewe (VP at Cullgen; CA, USA ) and panel chair, Peter Ettmayer (Scientific Director at Boehringer Ingelheim; Ingelheim, Germany) set to dive into the issues surrounding the advancement of protein degradation. Highlights from this include discussion on whether protein degradation is delivering the promise of targeting the ‘undruggable’ or if it is all talk. Burslem noted that it was still important to remember that, “Undruggable proteins still need to be ligandable.” Moving forward, the conversation shifted to the question of whether there is a need for new E3 ligands binding to new E3 ligases? During this part of the discussion, the panel deliberated the number of E3 ligases that exist. Many cite that there are an estimated 600 E3 ligases, however, one could argue if (a) this number is accurate and (b) if the unknown ones will also perform the same roles like the ones already found, which are well evolved to execute polyubiquitination. Ultimately, the panel emphasized the need to be creative in our approaches in novel PROTAC development.



Throughout the two-day summit, it was interesting to see that a theme that kept occurring for many researchers, groups and companies was prioritizing unmet needs. Many highlighted that taking into consideration commercial rational (unmet need)/clinical rational was an important aspect of how they decide to pursue a project. In the panel discussion from day 2, all the panelists agreed that one of the first questions you need to ask yourself when beginning any research is to figure out the medical need you want to address – is it an unmet need?

Overall, I found the conference particularly illuminating and interesting event, consisting of a varied selection of talks. I would like to thank the all the presenters for their time, as well as Global Engage for organizing the event – I look forward to the 2020 edition of the summit!

Back to top of page

Go to the profile of RxNet

RxNet

A network for the drug discovery and development community, Future Science Group

Your source for the latest news and views from the fields of drug discovery and medicinal chemistry.

No comments yet.