Oral bioavailability, access to intracellular targets, and cost and ease of production are a few advantages that small molecule drugs generally have over large molecule drugs such as biologics, natural products, and peptides. These properties make small molecule drugs highly coveted in the industry.
But despite huge technological advances in genomics, proteomics, and chemistry, the discovery and development of small molecules doesn't seem to be getting any easier.
That was one hot topic of conversation at recent conferences that explored current issues in small molecule drug discovery, including Cambridge Healthtech's "Mastering Medicinal Chemistry" in San Francisco, and IBC's "High Throughput Chemistry" in San Diego.
"Over the last ten to twenty years, lots of scientists have done clever things, for example, with combinatorial chemistry and high throughput screening, but the number of new small molecule drugs developed per year for the whole world has not gone up significantly," says David J. Lockhart, Ph.D., president and CSO of Ambit Biosciences (San Diego).
Jumping the Safety Hurdle
Joseph P. Vacca, Ph.D., executive director, medicinal chemistry at Merck (Whitehouse Station, NJ), presented on the company's decade-long effort to advance HIV integrase inhibitors. Although current marketed AIDS drugs inhibit HIV-encoded reverse transcriptase and protease, there are no approved therapies against HIV integrase, the virus' third enzyme.
Since integrase has historically proved to be a difficult drug target, Merck biologists worked to improve strand transfer screening assays before successfully identifying a new series of integrase inhibitors, the diketobutanoic acids. Medicinal chemists worked on this series to improve potency, decrease protein binding issues, and increase solubility.
The resulting carboxamide lead compound, L-870,810, met Merck's criteria for clinical development, and a Phase I study was initiated. Welcome news followed: in a short, ten-day antiviral study in humans, there was good antiviral activity (a 1.7 log reduction) and no drug-related adverse events.
But Merck halted the development of L-870,810 after observing unexpected toxicity in long-term dog studies. This was "very disappointing," says Dr. Vacca, but Merck is continuing the program.
"We don't think it's a class effect. We think it's compound-specific. We will keep pushing, trying to develop backup compounds until we get something out for patients. With the first compound, we saw that integrase inhibitors work."
The story of L-870,810 illustrates a well-known fact: it takes a long time to find a potential drug, and not all candidates succeed. One way companies seek to improve their chances of finding approvable small molecule drugs is by careful and creative construction of their physical and/or virtual libraries.
"Because chemical space is so huge, companies take great care in how they compile their internal compound collections. Chemists view them as the heart and soul of the company," says Richard Wagner, Ph.D., executive vp, discovery research, Praecis Pharmaceuticals (Waltham, MA).
The trend is also to learn more about ADME and toxicology properties earlier in development to be able to eliminate eventual failures sooner. "You want to develop methods that allow you to throw away molecules. If you throw them away today, you save time and money," says Dr. Lockhart.
Tapping Druggable Space
The construction and evaluation of compound libraries is now largely driven by sets of rules, such as those developed by Lipinski and Veber.
"One thousand and forty compounds are thought to be druggable with molecular weights under 600. If most companies have 106 compounds in their database, there's a lot of space left," says Barry Morgan, Ph.D., vp, chemistry at Praecis.
"People are trying to mine the space but keep within their set of rules. The better and quicker you can explore the druggable chemical space, the more likely you are to find new families of compounds against broadly classed hard-to-drug targets. The more compounds you can make, the more likely you are to succeed."
"Industry has hit a wall at the 1 million compound type level," adds Dr. Wagner. "If you make a big compound library individually, high throughput screening could take months.
"So, for bigger libraries, you have to go back to mixtures, but when you screen from a one billion compound library, any individual compound is only present at low concentration. You need a new way to detect that molecule. Conventional analytical methods, such as mass spectrometry, may not work because of sensitivity limits."
To overcome these challenges, Praecis uses combinatorial chemistry with a twist. With the addition of each specific chemical group, the company also adds a specific 712 base pair duplex DNA tag to the molecule in a separate but linked reaction. The result is that each compound in the library is labeled with a unique 80-base pair tag that follows the history of its synthesis.
Praecis uses the DNA tag to overcome the challenge of "being able to identify hits out of a library of this complexity," says Dr. Wagner. In one reaction vessel, the compound library can be added to a protein target in a needle-in-a-haystack experiment. The molecules that bind are isolated, and PCR is performed to identify them using their unique DNA tags.
PCR has an additional advantage. "We have an idea of relative potency of a species by its frequency in sequencing," explains Dr. Morgan. "We've been encouraged by the data that the system is working," adds Dr. Wagner. Their intent is to create more billion-component libraries "to place other galaxies in chemical space."
Priaton (Tutzing, Germany) is another company creatively tapping into the druggable chemical space. Its approach centers around multicomponent reactions, defined as "reactions where you have at least three different reactants, and the product you obtain has to have essential structural elements of the reactants," according to Christoph Burdack, Ph.D., CSO.
Priaton computationally mines its database of approximately 400 multicomponent reactions to identify novel entities and virtually screen them for desirable parameters. Priaton "works through the reactions to determine which products you can really make in the lab," explains Dr. Burdack.
Although other methods of molecular modeling can identify interesting chemical structures, "sometimes it takes twenty synthetic steps to get the molecule. In our case, all molecules can be made by three synthetic steps."
The company works with pharmaceutical partners to create scaffold-focused libraries, class-focused libraries (e.g., for a kinase), or target-specific libraries containing preselected compounds with specific biological information.
Ambit Biosciences presented on its proprietary high throughput technology for profiling the activity and specificity of kinase inhibitors at "Mastering Medicinal Chemistry". Almost all small molecule kinase inhibitors work by binding and interfering with the ATP site.
Screening for Activity and Selectivity
With over 500 kinases in the human genome, it is important to know the specificity profile of development candidates, but this can be time consuming and labor intensive. Ambit's approach is to build ATP-site dependent competition binding assays in an innovative way.
Each kinase or kinase domain is cloned and inserted into a proprietary T7 phage genome. The T7 phage are then allowed to infect E. coli cells. When the phage replicate, the kinase protein "sticks out" from the T7 phage.
"It's like a GFP fusion but instead of GFP, there's a T7 phage," explains Lockhart. In addition to the T7 phage-kinase fusion, the competitive binding assays also employ immobilized bait ligands and the kinase inhibitors to be tested.
Ambit allows the three components to interact in 96-well plates, with separate assays for more than 180 human kinases. After a washing step, the bound T7 phage are eluted and quantified using PCR, which is both sensitive (detection down to 10 molecules of protein per well) and quantitative (proportional to the number of bound molecules).
Ambit has used this technology to profile the selectivity of many kinase inhibitors currently on the market and in development.For example, the company recently discovered that Vertex Pharmaceuticals' (Cambridge, MA) VX-680, currently in Phase I clinical trials as an Aurora inhibitor, also hits Abl and multiple Gleevec-resistant forms of Abl. Information like this provides "clear directions on new uses and new treatment options," says Dr. Lockhart.
Ambit is also using its screening method to identify new kinase inhibitors. "We built a focused collection of small molecules with druglike properties that we think have a high probability of hitting the ATP site of some kinases or a kinase," says Dr. Lockhart.
By screening this library against the panel of human kinases, they came up with a series of potent FLT3 inhibitors with attractive specificity profiles. They are currently selecting a lead to go into formal IND enabling studies.
The question of specificity remains a key issue for small molecule drug discovery in general.
"Some of the activity of small molecules may be based on the fact that they are not selective," says Steve Arkinstall, Ph.D., vp, research, Serono Research Institute (Rockland, MA).
"Companies always strive for selectivity, but if you have really selective molecules they may be less efficacious in some assays modeling certain aspects of disease. The biology is starting to tell us that it's targeting multiple receptors, which that may be a more effective way of controlling angiogenesis. The same is true for tumor cell survival. You may need to target more than one pathway to get results."
According to Brian Healey, Ph.D., head of lead discovery at Serono Research Institute, these considerations have led to a "definite change in how people are focusing their chemistry effort."
As one way of addressing biological complexity, Serono has chosen to pursue high content screening and analysis methods, those where researchers are "measuring multiple endpoints simultaneously and/or where image content or microscopy is used," says Dr. Healey.
His team has developed a new automated high throughput screen for angiogenesis inhibitors. Serono's assay is based on the BD Biosciences (San Jose, CA) endothelial cell tube formation assay system, in which human umbilical vascular endothelial cells or human myometrial vascular endothelial cells are seeded into 96-well plates that contain BD Bioscience's matrigel.
In the presence of appropriate growth factors, the endothelial cells organize into tubular vasculature. Dr. Healey developed an image-processing algorithm using the Molecular Devices' (Sunnyvale, CA) MetaMorph image-analysis software, "compressing 3-D multi-plane images of the microvasculature into a single enhanced image to yield a robust quantitation of tube length."
Combined with a robotic screening system that employs an integrated Discovery-1 automated microscope, the imaging algorithm allows high throughput screening of angiogenesis inhibitors. "Moving to cell-based assays is a way to get closer to the pharmacology," explains Dr. Healey.