Clinical trails target validation
1 like721 views
The discussion highlights the critical role of chemistry in drug discovery, particularly in targeting and optimizing molecules for better potency and selectivity. Panelists emphasize the importance of understanding molecular recognition and biological context, advocating for integrated approaches that consider multiple targets and use biophysical methods to improve prediction of drug-like properties. A focus on well-characterized hits and early evaluation of compounds leads to more informed decisions, enhancing the likelihood of success in drug development.
Clinical trails target validation
- 1. 20/20 Pharma ON TARGET
- 2. WHItepAperS Wendy Cornell, Director, Chemistry Modeling and Informatics at Merck, moderates an expert panel discussion on targets in drug discovery. The panelists are: Jamie Baumgartner, Ph.D., 2 Senior Director of in vitro Pharmacology at MDS Pharma Services, Brian Lightbody, Vice Presi- CLINICAL trIALS dent of Business Development of Small Molecule Drug Discovery for MicroCal Products Group, GE Healthcare, and Patrick Zarrinkar, Vice President of Technology Development at Ambit BioSciences WC: What are the implications of target/ are equally drugable, and this target-centric approach really make that as- drug molecular recognition in the drug discovery sumption, which isn’t really true. Any given organization only has access to a small fraction of chemistry space, either due to the physical materials that are process, and how can these interactions be better available and/or due to the intellectual property space that it is free to work in. understood? With the chemistry available to any one organization, certainly not all interest- BL: The understanding of the molecular recognition of a ing targets are equally accessible. So it seems that there’s more efficiency to be target in a potential pharmacophore is a very important part of the drug had by having chemistry feed into the decision about which targets to pursue discovery process, since it’s ultimately this process that will be the basis along with biology. What I mean by that is to look at a number of targets that for some form of pharmacological intervention. establishing a basic are biologically interesting, and then ask which one of these is most easily understanding of the binding mechanism of a small molecule to its target, accessed by compounds that we actually have at our disposal. The way we which then leads to a control of the biological response, can have a significant put this idea into practice at Ambit is by screening compound libraries, not just against individual targets that we decide are interesting, but against the impact in a downstream success of a drug discovery program. So this whole panel of kinases that includes most of the ones, if not all of the ones, that knowledge is really critical to the design of molecules, with the optimum are biologically interesting either today, or will become biologically interesting. structural features leading to better potency selectivity into reducing the Then we can pick and choose and say, “Here are eight targets that we would be likelihood of clinical failures due to factors like off-target effects. happy to pursue from a biological perspective, and three of these we can actu- pZ: The foundation of any successful drug is a compound that has appropri- ally access really well with compounds we have on hand.” So that’s where we ate potency and selectivity and the more you understand about the recognition should put our efforts. So the chemistry plays an almost equal role in deciding process between the compound or a series of compounds and the target, the which targets to pursue along with the biology of those targets, and that makes better chance you have of finding compounds that have appropriate potencies the process more efficient, and gives you a chance to focus your discovery re- and selectivity. That’s only the first part. sources on those projects where you have the best starting points that are most The tricky part is to maintain that appropriate potency and selectivity likely to succeed in the shortest period of time. while you’re optimizing the other properties of the compound that are impor- jB: There are two things that we are actually seeing. One is fragment-based tant in having an actual drug from pharmacokinetics to solubility, formulability discovery, where chemistry is not taking the final molecule, but taking frag- and so on. The more you understand about the molecular recognition event, the ments that may have biological activity and starting to class those. The second more you know about where you can modify the compound and how you can thing is, instead of looking at individual targets in isolation, one of the trends modify the compound and still maintain that that we are seeing is taking integrated cell-based disease models and doing appropriate potency and selectivity. I think an in-depth understanding of “chemical genetics”: It’s where the biologists are meeting the chemist halfway, the molecular recognition event often can come from not only from understand- in terms of looking at more targeted libraries in more disease context and ing the interaction with the target of interest, but also from understanding instead of looking at an individual’s targets, looking at a cell-based readout that how the compound interacts with other related targets and the would be a mimic of that disease. So you can take a look at immune response, effect of changes in the molecule across the pattern of interactions with for example, and peripheral blood mononuclear cells and looking at a specific targets. This actually tells you even more about the importance and the cytokine profile, which are compounds that have a desired profile for what you role of any particular interaction between the molecule and the particular think a therapeutic would look like, screen a focus set of compounds and then target of interest. So I think the more you know, the better your chance is of start choosing your hits out of there, and then work backwards using a plat- getting an appropriate molecule with the properties you want. form like Ambit’s to find out where that compound actually interacts. In that particular situation, you’re screening against the whole kinome or the whole WC: I agree with both of you. I think selectivity is becoming genome in that cell-based context instead of looking at individual targets. One more of an issue for everyone — the more tools that of the trends that we are actually seeing is more of a retreat from the traditional we have to get at that early, the better. Given that not high throughput screening. So instead of looking at a million compounds or all biological targets can be reasonably accessed, what 10 million compounds against one particular target, we’re seeing more focused role should chemistry play in target choice? sets and trying to get more biological context early with those particular hits. pZ: traditionally, the way drug discovery is done is biology-driven. Decisions BL: We’re seeing an increasing interest in trying to screen things in the most about which targets to pursue are more or less based on the biology of biologically relevant way possible, as well as trying to be rigorous about deter- those targets, assuming that you are working in a protein class that is considered mining the mechanism of action for some kind of pharmacological interven- drugable. But, of course, even within a drugable class, not all targets tion using things like more endogenous reagents, enzymes, substrates and so 20/20 Pharma
- 3. WHItepAperS on. The purpose is just to make sure that the model system represents the most tion analysis of native molecules that more accurately represent a true picture biologically relevant system possible, so that the knowledge and understanding of the biological system. Secondly, in relation to the overall binding affinity, the gained is relevant, and that’s one big area we’re seeing, especially in the label- information that these technologies can provide, such as the thermodynamics 3 free world. of binding and the kinetics of binding, yield more important insights into the CLINICAL trIALS WC: There definitely seems to be more of a movement to go back to biologi- binding mechanism and residence time of a small molecule to its target part- cal relevance and not have the system be so isolated that it becomes simple ner. Knowledge of the binding mechanism for example, is crucial in selecting to analyze, but is not as relevant as it should be. molecules with the best drug-like features for optimum potency and selectiv- BL: It gives you more confidence in the data. ity. This, of course, results in more confident decisions in the drug discovery process. One trend that we are seeing relative to the previous question is that WC: What advice can you give on how to achieve success once a hit is people are more hesitant these days to make their decisions about which hits to identified? move forward based on binding affinity alone getting more rigorous about try- jB: We’re actually seeing two different modalities in terms of characterizing ing to peel back the underlining factors in the binding affinity. In some cases, hits. One is a classical approach to where we’re seeing iterations to where may even choose molecules with a lower binding affinity, because they really companies are trying to optimize potency initially against a particular target have the desirable structural features for a potential drug candidate. to try to drive down an nanomolar or picomolar type of potencies of a com- WC: I think when pfizer came out with the rule of five, however many years pound against a particular target, and then what we start to see is exploration ago that was, it really “codified” the way a lot of people were thinking anyway of the selectivity of that particular compound. So that’s a more traditional without realizing it. I think it’s clear we need to go well beyond just counting approach. Some of the more successful projects that we’ve interfaced with have molecular weight and individual donors and acceptors, and so on, and really taken a different approach of doing a lot more parallelism having small iterative characterize these molecules. pharmacological tests to test for selectivity among subclasses earlyon. That pZ: I would definitely agree with Brian that a greater understanding of how actually leads to better decisions in terms of how that compound or how that compounds interact with the targets, in addition to how potently they interact lead series is actually characterized, and also puts in place some of the ADMe really helps one understand the compounds and helps guide which hits to assays earlier on, so by the time you start looking for more potency, you have a actually focus on. One thing that we’ve done at Ambit along those lines is better idea of whether that is going to be drugable molecule, and you also will to build assays that are specific for a particular activation state of a kinase. be able to take a look at what you want to have downstream in your develop- So we have separate assays for an activated and an unactivated kinase, and ment pathway that would actually address some the adverse events or toxicities we can understand whether the compounds prefer one state or another, or that may hang up or kill that particular compound or project later on in the whether they don’t care. For example, there are two classes of kinase inhibitors development phase. that are broadly known as type I and type II. type II inhibitors prefer an On target and adding more context to that particular molecule or that unactive-like confirmation of the kinase, and it manifests itself with much series is very important, and actually leads to better decisions and downstream higher affinity to the unactivated state relative to the activated state, and we success. can see that in those binding assays. So actually starting to build assays where You do get earlier triage or earlier failures with those compounds, but the end you screen compounds against a panel of those assays, you start to understand, result is more robust. We’re seeing more success with our clients and their abil- not just which targets compounds hit, which is usually what you’re interested ity to achieve IND filings. in when you’re profiling a compound against kinases but how each compound pZ: The broader you look, the earlier and the better you understand the hits might interact with those kinases — is it a type I inhibitor or type II inhibi- that you have and how to move forward with those hits, the better the chance tor? What’s the binding mode? That allows you to start to ask structural ques- you have of actually ending up with a drug. Again, going back to the theme tions in biochemical assays without necessarily having to do a lot of structural of chemistry helping to decide which programs to work on, one approach work up front. is to start with hits against several different targets and to do some exploratory jB: One of the things that we actually come up against is: How do you chemistry to see where you get the best traction, and then focus your resources interpret that early testing data in a biological context? What we’re concerned on that. So you don’t assume that you have a 100 percent success rate on the about is how those assays or those early determinations, or binding mecha- programs you start; you start three programs and then narrow that to one, nisms are influenced by protein. All the drugs that we are talking about for based on which one looks the most promising, focusing your resources on the small molecules have to reach target with respect to binding up to human programs that are most likely to succeed in the shortest period of time. serum, for instance, and how does that effect these early interpretations? It’s something that we’re continuing to have vigorous debate around. WC: I would agree with both of you, and I think having as many well- characterized hits to choose from at the start is really important. Cer- WC: What must be considered when adding more biological context to tainly once chemistry begins — and a lot of SAR is developed around a compounds early? particular lead series — then there can be reluctance to switch to another jB: In essence, I think the protein binding pieces is one. For anybody offering series and possibly lose a lot of that information. I think the choices that pharmacological testing whether its kinase testing or broad pharmacological we make at the beginning are very important. How do you see the role places, the question is: How can you take a look at an early test and make a of biophysical testing relative to evaluating small molecule inhibi- prediction downstream about what that drug would do if it was administered tors or activators of drugable targets? in man? BL: Biophysical methods like microcalorimetry and Spr really play an im- It’s about adding more context earlier and using that to compare to portant role in the selection and design of small molecule pharmacophores for marketed drugs or drugs in the clinic, as well as pre-clinical models and trying two important reasons. First, these methods are label-free, allowing for interac- to squeeze out or trying to get more information from those in vitro tests to 20/20 Pharma
- 4. WHItepAperS predict in vivo activities. WENDY CORNELL BL: It’s obviously very difficult to predict downstream effects of potential 4 DIRECT OR AND TEAM LEAD OF pharmacophore. What we’re seeing with biophysical techniques is increasing CHEMIS -TRY MODELI NG AND CLINICAL trIALS interest in factors like the hydroferbiscity of a drug or drug candidate and how INFORMATICS GROUP MERCK that relates to factors like non-specificity. So that bit of information is some- thing that can be gleemed from something like a calorimeter where you can The Chemistry Modeling and Informatics Group supports the relate the entropy to the hydrophobicity and the intipisis is factors like relating drug discovery efforts at the Rahway, NJ site. In addition, the the degree of hydrogen bonding and much more ionic and more specific bind- group is active in methodology development in areas such as ing events, which is additional information that can be provided by biophysical virtual screening, kinase similarity prediction, ADME modeling, structural bioinformatics, and target assessment. She is a member techniques to more specific binding to places like a cadalidic domain or it may of Merck’s New Technologies Research Licensing Committee be an allosteric site. It’s interesting with things like kinases — the fact that and is active in the Computers in Chemistry (COMP) Division you can do some really interesting multifactorial experiments with things like of the American Chemical Society (ACS) where she is a past non-activated and activated kinases in the kinases cascade in addition to the Program Chair and currently holds the position of Immediate Past Division Chair. binding mechanism — the mechanism of action can definitely give you more confidence in what’s going on and what factors underlay the binding mecha- nisms, and they can be related downstream to potential secondary effects like off-target effects For example, if a potential drug candidate has a high degree of hydrophobic interaction underlying its total binding affinity, there’s a good JAMIE BAUMGARTNER, PH.D. SENIOR DIRECT OR OF IN VIT RO chance that it’s going to be a very non-specific binder because hydrophobicity PHARMACOLOGY is caused by the repulsion of the molecule from the cell and the environment MD S PHARMA SERVICES that it’s in and not so much the attraction to any given specifics on a target molecule. Mr. James (Jamie) Baumgartner is the Senior Director at MDS There’s been several retrospectives looking at this, such as the HIV 1 Pharma Services, a provider of innovative early stage drug devel- protease opment solutions from discovery through Phase IIa. Previously inhibitors and the statins to equate the history of the evolution of different Dr.Baumgartner was a Research Scientist at Amgen and prior to that he worked as a Research Scientist at ZymoGenetics. drug candidates and their subsequent increase in hydrogen bonding and decrease in hydrophobic interactives to more downstream specificity, and lack of drug polymorphism in viral molecules, and how that would effect the drug potency. I think that is interesting work. BRIAN LIGHTBODY pZ: It’s very helpful and important to add biological context early-on to VIC E PRESID ENT OF BUSI NESS DE- understand your compounds and that one has to think carefully in how VELOPM ENT OF SMALL MOLECULE to interpret those results for biological assays and interpret them in light DRUG DISC OVERY FOR MIC ROCAL of what one knows about the properties of the compound. I think it boils PRODUCTS GROUP down to knowing your assays and knowing your compounds and how an GE HEALTHCARE assay is run. The details of an assay can affect the outcome and one has to keep that in mind when looking at the results. For example, if it’s a cellular Mr. Lightbody is responsible for developing and orchestrating assay with cell stimulated, what was the concentration of the stimulant? programs that will facilitate the adoption of microcalorimetry into How long was the stimulant present? Was the compound added before, small molecule drug discovery programs. He has over 25 years of experience in developing and commercializing products for during or after the stimulant was added? What’s the serum concentration biotech and pharmaceutical research. that goes to the protein binding issue that jamie mentioned? All these factors can affect the outcome of an assay. They have to understand the details and the compound. In the oncology space, a lot of biological assays are cellular proliferation assays — they’re neither in sub cultures or in zenographsv, which are PATRICK ZARRINKAR, PH.D. just cell proliferation experiments in animals — and you might be working VIC E PRESID ENT OF with an inhibitor of say, kinase X and you’re working a cell system that is TECHNOLOGY DEVELOPM ENT AMBIT BIOSCIENCES believed to be dependant on kinase X for growth, and you see good activity, but unbeknownst to you, your compound may be hitting other kinases that are also important. For example, we know that a fraction of kinase Dr. Zarrinkar has been at Ambit Biosciences since 2000, and currently is Vice President of Technology Development. He has inhibitors have activity against aurora kinase, which is an important kinase led the development of Ambit’s KinomeScan profiling platform, with a cell cycle. So if you inhibit an aurora kinase, you are going to inhibit pro- and was the biology project team leader for the discovery AC220, liferation of most cell types. So if you get a result with your supposedly kinase a FLT3 inhibitor currently in phase II clinical trials. Prior to join- ing Ambit Dr. Zarrinkar was a Staff Scientist at the Genomics X inhibitor and a proliferation assay, you might ascribe that to inhibition of Institute of the Novartis Research Foundation (GNF) (1999- kinase X, where as you might have also been inhibiting aurora kinase, and that’s 2000). what you’re seeing and the effect of doing that. Of course, aurora is only one of many kinases that are involved in the cell 20/20 Pharma
- 5. WHItepAperS cycle. So to interpret the biological assay, you really have to understand pathway, or whether you’re hitting a multiple kinase where you can do multiple the properties of your compound, because you think it might be an inhibitor mechanisms of action. For example, in the cancer field, what we’re actually see- 5 of a particular target, but your compound doesn’t know that that’s what ing this measure cellular proliferation, which has been the gold standard to add CLINICAL trIALS you expect it to do, and it’s going to do whatever it is going to do. I think reads that would look at aphratosis for phospho sonatri that would measure if you understand your assays and understand your compounds in some that cell cycle block. It may be a specific modal measure, like a phosphor-spe- detail and think hard about how to interpret the results of the assay in light cific substraight inside them, like ATT1 for example. Adding more context just of that knowledge, that’s really where you are going to get some real informa- from the assay itself and getting more reads out of that compound can actually tion about what you have. address many of the questions that patrick is talking about in terms of adding jB: We talked about adding biological context. Certainly the assays them- more context or why that compound is having that effect on that particular cell selves we’re seeing a different type of trend of adding multiplex readouts, line. When we have that type of information, you make much better decisions so in the oncology field for example, the aurora kinase example is fantastic in about which compound, which hit, which lead you actually want to put forth, the sense that if you take a look at only cellular proliferation, you miss the boat. and again increase your probability of success in terms of development and You don’t know what your global mechanism is, whether you’re hitting specifi- clinical candidate. cally on one kinase or whether you’re hitting on a node in terms a signaling 20/20 Pharma