Join MySLAS Social

Tackling Chemical Space: All Hands on Deck

Different angles. Varying approaches. Deep-thinking. Exploring any and all possibilities. The selection of valuable chemotypes from the expanse of known chemical space is a primary activity for all pharmaceutical institutions. Many strategies have been used to accomplish this task.


"The possibilities can seem overwhelming, but chemical structure rules exist to point people in right direction," relates Christopher A. Lipinski, Ph.D., scientific advisor, Melior Discovery. "The biology world is incredibly complex and to be really certain about activity, one has to run an experimental assay. The chemistry world is far simpler with rules and principles and decades of lore relating chemistry structure to chemical reactivity."

One important tool derives from an examination of the chemical characteristics of successful chemical entities, captured within Lipinski's Rule of Five. These rules have proved useful in informing on what compounds might be derived from hits obtained from high-throughput compound screening campaigns either through no hits or hits with properties that immediately bang up against the Rule of Five. However, not all screening campaigns result in identification of a successful chemotype. Another technique that has moved in many cases from a "second choice if HTS fails" to a position of strength is fragment-based screening – a method that builds on multiple small binding fragments to create a single unique chemotype, much in the way that a jigsaw puzzle might be solved.

To strategize their best possibilities for success, today's biochemists, biologists and the analytical teams responsible for screening and assays must understand the history behind any lead structure they employ.

Chemistry is Simpler….

While college students in the throes of organic chemistry may not agree, Lipinski says "biology is far more complicated than chemistry; we understand chemistry better than we do biology."

Lipinski's Rule of Five, formulated in 1995 after some "scientific fooling around" is a simple algorithm for predicting drug compounds likely to show oral activity. The Rule of Five is so named because a five appears often in the cut-off values; there are actually only four rules. Poor absorption or permeation are more likely when there are:

1. More than 5 H-bond donors

2. The MWT is over 500

3. The CLog P is over 5 (or MLOGP is over 4.15)

4. The sum of Ns and Os is over 10

(Substrates for transporters and natural products are exceptions.)

Even though chemistry criteria are simpler, Lipinski says a lot can still go wrong purely based on chemistry – this is where the Rule of Five as well as filters for problematic chemistry structure come in. "Rules and filters based on chemical structures are important," he says. Lipinski indicates it is one way for medicinal chemists to use existing information to narrow down their options. Chemical space is so enormous that choices have to be made. So the narrowing down of options should be thought of in the sense of improving drug discovery efficiency based on our success history rather than as a narrowing of opportunities. Applying the Rule of Five and compound quality filters became standard operating procedure in most pharmaceutical companies in the late 1990s.

Lipinski says that "for chemical biologists the goal is knowledge; what you require from a compound is just enough to satisfy that goal which usually means selectivity. You need to have reliable mechanistic knowledge on your tool or probe compound. Does it affect just a single mechanism or a panel of mechanisms? Unfortunately as we learn more and more about a family of targets, we discover that a probe compound is seldom really selective. When you experiment with a compound, you would like to be able to know that the assay is valid – that the compound is specifically doing what you would like it to do. This also means there are whole classes of compounds that may not work for you as they may not give you specificity."

Most chemical biologists don't have to be as stringent as those doing true drug discovery experiments, notes Lipinski. "However, although you might be doing a chemical biology experiment, many times in the background is lurking the idea that if I discover something in a chemical biology experiment could it be good enough to be the basis for drug discovery?"

"In drug discovery, we have big pharma, biotechs, and more recently, academic organizations doing drug discovery," Lipinski notes. "With a compound clinical failure rate averaging 91%, we need to go where there is great biological expertise – academia," states Lipinski. "There are very large numbers of collaborations with prominent academic institutions especially in the Boston, San Diego and San Francisco areas."

A Fragment Success Story

The blog, Practical Fragments, reported August 17:

"Today marks history with the first FDA approval of a drug to come out of fragment-based screening. The drug is branded as Zelboraf (vemurafenib), but readers of this blog are probably more familiar with its previous name of PLX4032. Although widely expected to be approved, the FDA acted more than two months ahead of schedule. The drug targets a mutant form of BRAF and has received widespread media coverage because of dramatic clinical results showing that it extends life for patients with a particularly deadly form of skin cancer."

"To those of us working in fragment-based drug discovery, this approval provides validation of the field," expresses Daniel A. Erlanson, Ph.D., co-founder of Carmot Therapeutics, Inc. and editor-in-chief of the blog. "In fact, Zelboraf was developed rather rapidly; the project began in early 2005 and six years later they have a drug that's approved."

Fragment-based drug discovery, or fragment-based lead discovery, takes a drastically different approach to finding hits than HTS. Instead of searching first for a large molecule that binds with its biological target, they look for very small molecular pieces, or fragments.

"Think about high-throughput screening," Erlanson offers. "You need to screen millions of compounds. There are estimates that the number of possible small drug-like molecules that could be made is on the order of 1063, which is just an astronomical number. If you look at fragments, Raymond and colleagues have computationally enumerated all possible small molecules containing up to 11 carbon, nitrogen, oxygen and fluorine atoms. The numbers are big, but they are approachable. It's like a hundred million. If you screen fragments, because there are fewer number of them out there, you can cover a larger portion of chemical diversity space more efficiently."

Erlanson says the late William Jencks of Brandeis University first introduced the concept of fragment-based drug discovery in his 1981 paper in the Proceedings of the National Academy of Science.

"Dr. Jencks proposed the idea, but it wasn't truly feasible until technology caught up with it," explains Erlanson. That first technology was nuclear magnetic resonance (NMR), which is sensitive enough to detect the small fragments with low affinity. Today, there are roughly a dozen ways to find fragments.

"It was the Abbott Laboratories article in Science in 1996 that truly launched many companies into fragment-based drug discovery," Erlanson adds. "Dr. Stephen Fesik and his Science co-authors convinced people that fragment-based drug discovery is a viable method to discover novel chemical matter for challenging targets and as a way to query chemical space more efficiently than high-throughput screening."

No matter what fragment-finding technology is employed, Erlanson concludes, the advantage is that more chemical space can be explored more rapidly with fewer compounds.

Building on Existing Strategies

"Since Chris Lipinski published his seminal paper on the Rule of Five in 1997, there have been many further studies leading to the emergence of further rules of thumb around drug-like and lead-like space and parameters that are linked to, for instance, toxicity," states Mike Hann, Ph.D., director of bio-molecular structure at GlaxoSmithKline Medicines Research Centre.

Hann states that efforts have been centered on the importance of physical properties and descriptions of molecules for helping to find compounds that have good developability prospects.

"When looking for early leads and you start with a lead molecule that's already 400/450 molecular weight, by the time you've done the lead optimization you have a molecule that's already pushing against the limit," he explains. "Chris [Lipinski] in some ways started the ball rolling but there's a whole spectrum of different types or rules of thumb that people have introduced. For example, the typical fragment leads have a molecular weight of less than 300 – very different types of compounds and nothing like a drug molecule."

Hann and his team published on a model of molecular complexity in 2001. "Molecular complexity describes the amount of functionality in a molecule. If you have too little of it, you won't have any chance of getting a reasonable interaction with the target. And if you have too much, i.e., it is a very complex molecule, it is very difficult to get the combination right such that you find a molecule that can interact with a protein – which itself is, of course, very complex."

The team recently published a follow-up paper and in it coined the phrase molecular obesity. "It seems to be a quite nice analogy to draw between human biology and molecular structures," Hann states. "We wanted to make the point that you cannot add potency to compounds by adding molecular fat to them. Much like in human biology, obesity causes illness or death. If you go too far with the amount of lipophilicity in the molecule, you are decreasing the chances that the molecule will survive through preclinical studies or clinical trials. We thought the term would resonate with people."

Erlanson agrees that there are many ways of approaching the same goal – to be at the best possible starting point for successful drug discovery. "In practice, people tend to use multiple techniques in parallel or in series," he notes. "You may use a biochemical screen and then follow up with NMR or crystallography to confirm your hit and get some indication of where your fragment might be binding."

Erlanson acknowledges that fragment-based drug discovery requires a different mindset and many traditional medicinal chemists are not comfortable with it. "They don't like working with molecules with such low affinities and this has created a barrier in some disciplines. Also, it's interdisciplinary so it doesn't lend itself to handing off from one department to another."

But there has been progress. "Fragment-based drug discovery used to be thought of mostly as the second option," Erlanson adds. "If a target went through HTS and did not come out with any hits, they'd say ‘let's try fragments.' Now, however, more people are starting to try fragment approaches at the beginning."

Learn More!

In recognition of the importance of this aspect of drug discovery, SLAS presents a four-part virtual course: "Interrogating Chemical Space – Rules, Filters, Fragment-Based Screening and Beyond." The course modules offer various strategies for exploring chemical space:


October 25, 2011

Module One: Rules and Filters and Their Impact on Success in Chemical Biology and Drug Discovery

Chris Lipinski, Scientific Advisor, Melior Discovery


November 1, 2011

Module Two: From Molecular Complexity to Molecular Obesity — Lessons From Hind Sight

Mike Hann, Ph.D., Director of Bio-Molecular Structure, GlaxoSmithKline Medicines Research Centre


November 8, 2011

Module Three: Practical Approaches to Fragment-Based Lead Discovery

Daniel A. Erlanson, Ph.D., Co-founder, Carmot Therapeutics, Inc.


November 15, 2011

Module Four: Combining NMR and X-ray Crystallography in Fragment-Based Drug Discovery: From NMR-detected Fragment Hits to Clinical Candidate for BACE-1

Daniel F. Wyss, Ph.D., BioNMR Lead, Merck Research Laboratories


Each module is approximately 90 minutes long and begins at 11:30 a.m. EDT. Register today to take advantage of discounts for members, academics and those registering for all four modules at once. This SLAS virtual course is presented as a live, real-time event as well as a streaming online video recording or on CD.

"Fragment-based drug discovery is an approach that has really proven itself but it is still not mainstream," Erlanson states. "It's a tool that people may have heard of but they may not have really dug into the details. This SLAS virtual course provides a way to see a bigger picture."

"Attendees will get a very good overview of the area from four people with a lot of experience in their chosen topics and who have been practitioners using the concepts and a part of their evolution," Hann adds. "We all need to find ways to use our past experience and knowledge – to be cognizant of our knowledge and yet to be free and unlimited to come up with new ideas. Finding the right balance between those two is the challenge in drug discovery. We need to make use of new technology, but we also shouldn't forget what we do know and what we have learned."

"With things changing regularly, many people do not understand the principles behind these rules," says Lipinski. "A lot of people are confused – they say the Rule of Five works so you should be able to predict biological activity, right? That is absolutely not the case. This SLAS virtual course will help sort this all out."

October 7, 2011