Reinventing medicinal chemistry

 

 

 

 

On any given day, the media streams are filled with companies announcing their latest biologic achievements in medicine, and given the high profile of these therapeutics in disease areas such as oncology and autoimmunity, it is hardly surprising if some people have forgotten the role of medicinal chemistry in the drug discovery process. Yet, when you take an objective look at the pharmaceutical industry, you’ll see that the workhorse division still holds sway over the entire process, despite the lack of fanfare. Of the 39 new molecular entities approved by the U.S. Food and Drug Administration (FDA) in 2012, only about a third of them would be classified as biologics or polypeptides. The remainder were smaller chemical compounds ranging from classical small molecules to macrocyclics.

 

 

Part of the reason for medicinal chemistry’s disappearing act may be due to attempts by the pharmaceutical industry itself to move medicinal chemists to the back burner by automating molecular discovery and synthesis through methods such as combinatorial chemistry and high-throughput screening (HTS).

 

With an eye to filling those drying pipelines with an endless supply of potentially therapeutic compounds, companies heavily invested in ways to randomly generate tens of thousands of “drug-like” compounds from the chemical building blocks of known drugs (see sidebar, “Leaving Lipinski?” below this story).

 

As the American Chemical Society’s Mark Lesney glibly wrote back in 2002: “A large enough library with the right high-throughput screening assay promised the ultimate brute-force method for obtaining bioactive compounds without biologists and organic chemists getting in the way.”

 

As Jun Xu and colleagues of China’s Sun Yat-Sen University recently pointed out, however, this effort proved largely fruitless with regard to producing approvable therapeutics. In the 25 years from 1981 to 2006, they suggest, only one de-novo compound arising from combinatorial chemistry achieved regulatory approval, that being Bayer’s antitumor compound sorafenib.

 

In an overview of natural products in drug discovery published last year in Metabolites, Daniel Dias and colleagues at the University of Melbourne and RMIT University were even more pointed:

 

“Though the pharmaceutical industry has expended a considerable amount of resources to both HTS and combinatorial chemistry overall, of the 1,184 new chemical entities covering all diseases/countries/sources between the years 1981 to 2006, 30 percent were found to be synthetic,” they wrote. “It is also worth noting that 52 percent (total) of these compounds are either a natural product, a mimic or a chemical modification of an existing natural product pharmacophore.”

 

For Lawson Macartney, CEO of Ambrx, part of the failure was in attempting to remove the art and expertise of medicinal chemistry from the science.

 

“When I was at GSK and also when I was with Shire, one of the things I really had antibodies to was this whole notion of industrializing drug discovery—at the end of the day, that’s really what combinatorial chemistry and HTS is or was—because I really believe that drug discovery of new medicines is truly a creative process,” he says. “The human brain has to intervene much more than just random chance and happenstance, which is really what it’s all about when it comes to combinatorial chemistry and random screening.”

 

For Macartney, the failure of this industrialization effort is borne out by the numbers.

 

“Discovery pipelines have been horribly inefficient, and if you look at the major pharmas, what they’re doing typically is dismantling that industrial framework and moving much more toward either internal-focused discovery with very small discovery units—and that’s certainly how GSK’s done it—or partnering with academic units or highly creative companies like Ambrx to bring forward those creative molecules as rational drug design from different disciplines converges,” he adds.

 

 

Macartney isn’t about to fall into an “us versus them” discussion, however. For him, it is more about the evolution of the role of medicinal chemistry within the pharmaceutical endeavor.

 

“There can be a polarizing view of small molecule versus biologic, but I see convergence of these fields to actually be much more, where the real action and the real benefit could be seen to take—the convergence of the best small-molecule payload with the best targeting biologic,” Macartney offers. “Medicinal chemistry, which is all about rational drug design and rational drug optimization, continues to be fundamental. The reason for that is that so much of what we do in terms of drugability is totally dependent on rational modification of existing chemical moieties or chemical backbones.”

 

But with the advent of new technologies and approaches, the role of the medicinal chemist has dramatically expanded and evolved.

 

“Historically, of course, it has always been small molecules, because that’s what chemists understand,” says Macartney. “It is only in the last decade that we’ve had the molecular tools to be able to rationally design proteins as they’re expressed or secreted from a living cell.”

 

For Macartney’s colleague, Ambrx Chief Technology Officer Ho Cho, it is about choosing the best technology or approach for a therapeutic need.

 

“Another way you could think about the evolution or the future of medicine is there are going to be cases where just a small-molecule drug will be useful in treating a certain disease state,” he says. “There will be cases when just the wild-type protein or antibody, without further elaboration, may be suitable for treating a disease state. Examples of that include things like insulin, growth hormones and monoclonal antibodies like Humira. What is emerging, however, are certain targets and disease states that benefit from a convergence of the two modalities, such that you can do the exquisite targeting via the biologic component, like the monoclonal antibodies, and then via site-specific conjugation and rational design, you can deliver a potent payload to the target tissues or cells that you want to modify.”

 

For Meeuwis van Arkel, vice president of product development at Elsevier, the technological expansion is coming at a time when more is expected of medicinal chemists within the drug discovery process.

 

“I think medicinal chemistry is still core, and will always be core, to the discipline of the drug discovery process,” he notes. “But at the same time, I am seeing that the researchers are increasingly challenged by two things. One is the overwhelming amount of data. The second one goes into the art form again.

 

“[Medicinal chemists] are expected more and more to see chemistry in the context of other things such as biology, safety and efficacy, and have a contextual knowledge on a continuous basis about that,” van Arkel continues. “Where you used to see the process far more phased with one department handing over to the next department, that is all being blurred into one process, where each discipline continues to be very domain-specific science, but each discipline is also expected to understand more and more of the related disciplines.”

 

 

One of the major challenges to the pharmaceutical industry is late-stage failure of drug candidates, but van Arkel sees an opportunity for this new contextualized approach to medicinal chemistry to help researchers weed out faulty candidates much earlier in the process—what he and others have described as the “fail early, fail cheaply” approach.

 

To facilitate this approach, Elsevier has developed a suite of online database products that offer researchers—whether medicinal chemists or elsewhere along in the chain—access to the latest information from thousands of journals and repositories. The most recent addition to this suite is Reaxys Medicinal Chemistry.

 

“Reaxys Medicinal Chemistry is very much around identifying, prioritizing and optimizing the selection of your compounds far earlier in the drug discovery process than I would say traditionally that was done,” van Arkel explains. “Reaxys Medicinal Chemistry is there for basically identifying how to fail early, and therefore, how to fail cheaper.”

 

Reaxys Medicinal Chemistry is an offshoot of the company’s Reaxys database, a broader chemistry platform that addresses questions like, does a compound even exist, what types of attributes are known about the compound and is the compound commercially available? It also goes into more business-intelligence types of questions, such as who else is working on this compound and what is known about this area of research?

 

Reaxys Medicinal Chemistry, by comparison, is far more focused on bioactivity and the association of compounds and their known bioactivity, and their known behavior toward potential targets.

 

From van Arkel’s perspective, the platform replaces a significant role in the medicinal chemist’s day that took time away from active research and development. Rather than having the researcher manually identify and analyze dozens or hundreds of sources for tidbits of information, Reaxys and Reaxys Medicinal Chemistry is a one-stop shop for that information.

 

“We plough through 16,000 chemistry-related journals, and have hundreds and hundreds of people literally manually extracting the relevant facts out of this journals and patents content, and provide the backfile of this database and put layers on top of the database that supports them in the decision-making process,” van Arkel explains.

 

A good example of a decision-support resource in Reaxys, he says, is the autosynthesis planner that basically suggests potential synthesis routes, which are pulled from many different sources out of many different articles and collated together.

 

Likewise, Reaxys Medicinal Chemistry offers a heat map, a dynamic table researchers use to display compounds on one axis and known targets on the other, giving the scientist a quick overview of what is known about the scientific landscape.

 

“You can literally drill into that heat map and see where the pockets of activity are, or it might be interesting to see the pockets where there is little known about it, which could be a potential indication of an area where there is still freedom to operate,” he adds.

 

At the same time, the researcher can pick and choose the information he or she wishes to pursue, maintaining a degree of autonomy and intuition in the process.

 

“These types of tools don’t take away the art of the science—what I would describe as the serendipitous process of trying to understand and putting things into context—but rather, help you to make sure that you only apply that art form to areas that somebody else has not already fully followed through on,” van Arkel says.

 

But what does this expanded knowledge base and contextual facility mean for the pharmaceutical company? According to van Arkel, a recent independent study looked at what this capability meant for the productivity of the individual researchers, and showed that it had two effects.

 

“One is that the process moves about 30 percent faster, so you compress a very expensive process. The other one is that it becomes about 23 percent less costly,” he says.

 

These numbers were echoed in July by Maria Sjöberg, chief scientific officer at Karo Bio, which signed an agreement to use Reaxys and Reaxys Medicinal Chemistry. In announcing the deal, she said: “Not only do our chemists have access to more data, including biological as well as chemical information, but the content is more applicable to their research. We expect the availability of the Reaxys Medicinal Chemistry information alone will reduce research and development time by approximately 20 percent.”

 

How desperate are companies to find efficiencies in their medicinal chemistry pipelines? Desperate enough for major players to share data.

 

In June, Roche and AstraZeneca PLC announced a collaboration that will see them share proprietary medicinal chemistry information from their own research efforts, without sharing the actual molecular structures, which they hope will improve factors such as molecular metabolism, pharmacokinetics or safety in potential drugs. Working through an intermediate company, MedChemica, the companies will give each other access to the know-how embedded in their databases of experimental results to help them identify new candidates with fewer rounds of design, synthesis and testing—a case of learning from each other’s successes and mistakes. Further, the companies have committed to making this data available to the wider research community, including research foundations and academia.

 

“It is unique in the history of our industry that two major players are sharing their know-how at such an early stage of research,” says Luca Santarelli, head of Neuroscience and Small Molecule Research at Roche. “We believe this transparency of small-molecule optimization knowledge, in a smart and thoughtful way, could profoundly enhance our ability to design drugs, be of benefit for all parties involved and ultimately help bring better medicines to patients.”

 

For van Arkel, another key aspect of the Elsevier suite is the interoperability of the various platforms, which include not only Reaxys and Reaxys Medicinal Chemistry, but also products like TargetInsights, PharmaPendium and Pathway Studio.

 

“The solutions themselves will continue to remain very domain-specific for one specific end-user,” he says. “But what we are doing in parallel is we are making them more interoperable toward other applications. If, once you’ve found a compound and you want to know what’s known in the FDA files about this specific drug, you can directly click to all of the FDA records that sit in PharmaPendium.”

 

He draws an analogy with the Microsoft Office Suite of products, although he is quick to point out that Elsevier’s platforms have nothing to do with Microsoft.

 

“Word, Excel, PowerPoint have very specific uses, yet they are very interoperable and you can easily jump from one application to the other or even copy-paste facts from one to the other,” van Arkel says. “It is creating an ecosystem that gives the researcher, from a domain-specific lens, the possibility to meander into other contextual areas.”

 

 

With expanded knowledge come expanded capabilities, as Macartney indicates, and it’s these new capabilities that are opening the door to the future evolution of medicinal chemistry and the role of medicinal chemists within drug companies.

 

“There is tremendous opportunity in this new kind of modality for medicinal chemists to contribute their strong chemical thinking, their strong proclivity to rational structure-guided design,” says Ambrx’s Cho. “This is exemplified in some of the collaborations we have with our larger pharma partners.”

 

He offers the example of collaborator Robert Garbaccio from Merck & Co.: “Garbaccio is a classic medicinal chemist who, until he started working on our collaboration, had never really done much thinking about antibodies or proteins, but now he’s a strong advocate of this converged modality where you’re using the best aspects of an antibody—let’s say, combining that with the best aspects of a payload, and these payloads can be small-molecule drugs or peptides,” Cho relates. “Now, you have a better targeted drug with the pharmacokinetics and pharmacodynamics that really make it an attractive product. And that targeting gives you the enhanced therapeutic window such that one can not only think about oncology, but also diseases in autoimmunity, inflammation, etc.”

 

“Right now, we know of some very potent immunomodulators or anti-inflammatory molecules, but we also know that if we don’t target these molecules and give them systemically, while they’re efficacious, we actually can’t give them to their full efficacious levels because they have some off-target toxicities,” adds Macartney. “Can we, in a very specific way, take this effective small molecule, but by virtue of conjugating it, adding it to a biologic targeting moiety, leverage its efficacy and minimize its safety issues?”

 

Ambrx isn’t the only company looking at merging the protein or peptide world with small-molecule medicine. In October, Soricimed Biopharma presented results of its proof-of-concept studies of peptide-drug conjugates in both breast and ovarian cancer at the World ADC conference in San Francisco. They demonstrated that a paclitaxel-conjugate targeting TRPV6-rich cancer cells performed better than paclitaxel alone in reducing cancer cell viability.

 

Said Jack Stewart, the company’s chief scientific officer, in announcing the findings: “As calcium ion influx channels, termed TRPV channels, have emerged as potential cancer targets, not only can Soricimed’s SOR peptides inhibit the TRPV6 channel activity directly by triggering cancer cell-death programs, as is being studied in the Phase I clinical trial underway in Canada and the United States, but they can also carry a drug or multiple drugs to the tumor site, thereby decreasing the dose of chemotoxic reagent required and increasing the efficacy and possibly reducing toxicities.”

 

“It’s not just about modifying a small-molecule backbone,” explains Ambrx’s Macartney. “Now it’s the potential to not only do that, but also modify the biological moieties and join them together to optimize the medicine. It’s all about adding two and two and coming up with a medicine that’s worth six.”

 

 

As Elsevier’s van Arkel concludes, medicinal chemistry is still the pivot and the nexus of what drug discovery today is about. The role and the expectations of medicinal chemists, however, have significantly evolved.

 

By being connected to and aware of many more disciplines throughout the drug discovery chain, it is possible that the importance of medicinal chemists has never been greater to the pharmaceutical endeavor and its future success.

 

 

“Another, more forward-thinking concept to consider is that right now, when a small-molecule chemist thinks about designing a medicine, they’re largely restricting their design space to molecules that obey Lipinski’s rules,” says Ambrx Chief Technology Officer Ho Cho.

 

The so-called Rule of Fives, described in 1997 by Christopher Lipinski, set out to define the parameters for orally delivered small molecules that made them “drug-like” based on the database of known drugs at the time.

 

The Rule of Fives states a molecule should have:

There has been much debate in the last two decades over the relative merits of the rule or any of its components, and how reflective of the medicinal space they truly are. There can be no doubting, however, that quite a few of the compounds that get approved each year violate one or more of the Fives.

 

It should be acknowledged, however, that most of these compounds are also not intended for oral delivery, a key component to Lipinski’s rule.

 

Cho’s position is that the new technologies, including conjugation with targeting moieties as in the case of Ambrx, offer medicinal chemists the opportunity to think outside of Lipinski’s box.

 

“I think it’s going to further open up design space such that you could actually contemplate prosecuting targets that would historically be considered undrugable because you couldn’t do it with a molecule that obeyed Lipinski’s rules,” he says.

 

As an indication of that trend, he points to the growing interest in macrocyclic molecules and biosynthetic peptides, a trend reflected in the agenda for the MedChem 2013 annual meeting taking place later this month in Beerse, Belgium.

 

“One of the areas that is really not fully explored in terms of drugable targets is creating molecules that can interfere with protein-protein interactions that require a lot more surface area coverage,” he explains, suggesting that the converging disciplines offer researchers “the ability to then deliver those types of molecules intercellularly through the assistance of the antibody delivery vehicle.”