| The pieces seem to be falling in place for fragment-based drug design (FBDD) companies. An Astex Therapeutics drug derived from this approach achieved a big first last year, reaching Phase I trials. Soon after that, Novartis inked two major deals worth potentially more than $500 million each with Astex and SGX Pharmaceuticals, another FBDD
pioneer.
|
|
Those deals signal not just good fortune for Astex and SGX, but an important step forward for a growing field.
"Not long ago, fragment-based drug discovery was regarded almost as an eccentricity of small companies. Now it's becoming mainstream," says Rod Hubbard, professor of structural biology at Britain's York
University.
Hubbard works with an FBDD company (Vernalis), which may explain some of his enthusiasm. But there are new signs that he's onto something: Big Pharma seems to be showing
interest.
Abbott is believed to be close to starting clinical trials of a fragment-based compound (Abbott 737) aimed at Bcl-2 as well as an LFA-1 inhibitor. The Novartis/Astex deal involves cell cycle inhibitors AT7519 and AT9311. The deal with SGX is around BCR-ABL
inhibitors.
Astex is hitting the key milestones earliest. Last year it launched AT7519
-- a cell cycle kinase inhibitor for cancer. The company took only three years to progress from the initial screening
"hit" against its CDK-2 target to the first treatment of a patient. According to Astex's chief scientific officer Harren Jhoti, that is about twice as fast as current metrics in Big Pharma
-- suggesting that the method could be one solution to the industry's productivity
problem.
In FBDD, researchers start with a library of a few thousand low-molecular-weight compounds, typically 100-300 Daltons each. They screen this small library against the target of interest to identify any fragments that bind, however weakly. Then they use structural information from X-ray crystallographic or NMR analyses to suggest ways of chemically improving these early-stage leads. After a few iterations of modifying the leads, boosting their affinity to the target and adding other drug-like properties, they hope to have found at least one compound that is worth taking forward.
"Until now, many medicinal chemists didn't believe this would work," says Hubbard.
"But in the last two years it has been demonstrated that you can go from a hit fragment with millimolar affinity to a compound with nanomolar affinity." Moreover, according to Jhoti, this optimization can be done much more quickly and efficiently on fragments than on higher-molecular-weight
candidates.
Besides U.K.-based Astex and SGX, other players include U.S.-based Plexxikon and Sunesis as well as U.K.-based Vernalis (with whom Hubbard works). Some of these latter companies have signed prestigious deals but have not yet reached the clinic. If all their plans go well, several FBDD-derived compounds will reach Phase I this
year.
Wolfgang Jahnke, head of fragment screening at Novartis' Biomedical Research Institutes (Basel, Switzerland), says almost all the Big Pharma companies are now trying to develop compounds from FBDD. Often this means trying to adapt their expensive high-throughput assay labs to detect fragment hits. The jury is still out on whether that approach will work. Biochemical assays aren't good at finding low-affinity ligands, although FBDD companies such as Plexxikon and Sareum routinely use them for initial hit
selection.
Most research is being focused on one key problem: how to improve the potency of a hit fragment as quickly as possible. That requires knowing just how the fragment binds to its pocket in the target
-- in particular, its orientation.
X-ray crystallography is still the favored way despite its low throughput. That issue is being eased as robotic sample handling equipment becomes more affordable, says Jhoti. Systematic co-crystallization of bound complexes has become increasingly recognized as vital:
"It is essential to form crystal co-complexes one at a time, as screening mixtures of ligands does not give high enough ligand concentrations at the crystal to ensure good occupancy of the active site," says Tim Mitchell, chief executive of structure-based contract research house
Sareum.
This means FBDD research is now narrowing onto crystallizable protein targets such as kinases, proteases, phosphodiesterases, phosphatases, and ATPases. Some companies (e.g., Sunesis) are entirely focused on
kinases.
Often, though, X-ray structures aren't available. Luckily, says Jahnke, there has in the past year been significant progress in finding orientation information using a new form of NMR called INPHARMA, developed at Germany's Max Planck Institute.
"INPHARMA is not a full structure determination, but it is a way to get the critical binding information very rapidly," says Jahnke.
"We have found it very successful."
The
other key to optimization is getting the fragment
chemistry right. First when selecting compounds to go
into a fragment library for screening, and second when
elaborating or combining the hit fragments. It's
important to keep the library diverse but limited in
size -- unlike combinatorial libraries, which have
often included compounds that can be made rather than
those that should be made, points out Tim Mitchell.
The Rule of
Three
To guide research in this area, Jhoti and colleagues have proposed a heuristic that he calls the Rule of Three (after Lipinski's famous Rule of Five). The rule says fragments should have molecular weights less than 300 Daltons; three or fewer hydrogen bond donors or acceptors; and a solubility coefficient
(logP) less than 3. Several other companies have now taken it up, even some chemistry houses that are offering fragment libraries commercially, says
Jhoti.
But many other factors matter
-- for example, ease of synthesis. The knowledge base here is largely proprietary, and companies are playing their cards close.
"It's still a black art," says Hubbard, who dismisses the Rule of Three as
"not particularly discriminatory." |
|
Moreover, the industry is divided into two distinct camps with vying elaboration strategies. One camp believes in finding two or more fragments that bind to adjacent pockets on the target, and combining them. The other groups concentrate on finding a single hit fragment and improving its affinity by adding functional groups. Abbot and Vernalis do the first; Astex, Sareum, and Plexxikon do the
second.
Astex's Jhoti says the advantage of the single-fragment approach is that chemists can build up fragments in a very focused way:
"We can make just a handful of compounds that test specific questions, such as improved binding." Hubbard in the other camp points to the vast range of possibilities for
"morphing" of highly diverse fragments: "You might see 10 different fragments bound into a protein's active sites with different orientations and different relative positions. Think about how you could exploit all those different interactions within one molecule by combining the chemistries."
The next three years should resolve two controversies: first, whether HTS assays can be adapted to deal with high-concentration fragment screening, and second, whether fragments can be transformed into good leads without having full structure information. The answers could determine who scoops the FBDD pot: small innovator companies, or Big
Pharma.
Email
this page to a friend
|
