AMRI COMPOUD LIBRARY COSORTIUM: A OVEL WAY TO FILL YOUR DRUG PIPELIE Muralikrishna Valluri, PhD & Douglas B. Kitchen, PhD Summary The creation of high-quality, innovative small molecule leads is a continual challenge in drug discovery. Against a background of increasing cost, the rate of drug discovery has remained stagnant for over 60 years. Improving research and development productivity is a considerable obstacle for the pharmaceutical sector. To address these challenges, innovative methods are crucial. High-throughput screening (HTS) is a major cornerstone of drug discovery; however, the availability of an innovative, relevant and high-quality compound collection for screening often dictates the fate of a drug discovery campaign. Given that the chemical space to be sampled in research programs is vast and sparsely populated, significant resources need to be invested to generate and maintain a competitive compound collection. By leveraging diverse experience and industrial expertise in synthetic and medicinal chemistry, AMRI addresses this challenge in a novel way through the Compound Library Consortium (CLC). Inadequacies of Combinatorial Chemistry and Small Molecule Libraries While revolutionary in the 1990s, combinatorial chemistry as a means for rapidly increasing the rate of new drugs in development has proven largely ineffective. However, instead of being abandoned, combinatorial chemistry has evolved, and large compound libraries are being enhanced with smaller, targeted libraries. Even so, large combinatorial libraries that were cost-efficient supplements to collections cannot increase the diversity of chemotypes to address new biological mechanisms, and existing small molecule collections are not adequate to generate viable leads. AMRI has evolved its chemical library synthesis offerings to address both the cost and diversity of libraries. Discovering the Best Candidate Rather than taking an indiscriminate approach to producing large numbers of compounds to high-throughput screening collections, the emphasis of the AMRI-led CLC is on synthesizing the right compounds, purifying them to >90% purity, and designing them for diverse targets. Leads are optimized by making small changes in the chemical structure that affect its physicochemical properties, toxicology, potency and binding interactions all key elements in finding a compound with the best chance of providing quality leads suitable for lead optimization activities. Figure 1. Consortium process Scaffold Ideas (AMRI & CLC Members) Analysis, Data Transfer & Shipment by AMRI Distribution to CLC Members for Selection Library Chemistry Validation, Production & Purification by AMRI Scaffold Selection by CLC Members & Chemistry Validation/Synthesis by AMRI Template Design by AMRI & Distribution to CLC Members for Selection Template Synthesis, Virtual Library Enumeration & Distribution of Libraries to CLC Members for Selection
About the CLC AMRI will prepare and deliver 70,000 compounds in approximately equal sets quarterly over 2.5 to 3 years. Cost-sharing among multiple members reduces the cost of compounds well below typical market prices. Members receive 2 mg of each lead-like compound, 25 mg of each fragment-like compound with each having >90% purity (LC/MS and 10% by 1H MR data provided). Scaffold libraries range in size from ca. 100-200 members/scaffold to maximize diversity. CLC members may file patent applications on compounds in CLC and additional analogs derived from them. Role of the Joint Consortium Committee (JCC) and the Consortium Process Each member company there are currently three has a representative on the Joint Consortium Committee, chaired by AMRI. The JCC is responsible for the general review and monitoring of the CLC s activities. The JCC oversees the designation of scaffolds, templates and final compounds to be synthesized for the CLC. The JCC is involved in proposing, reviewing, approving and amending CLC strategies for the upcoming delivery periods. Quarterly JCC meetings are held and additional teleconferences scheduled as needed. Synthetic Process Ensures Diversity AMRI has designed a set of distinct scaffold cores for the CLC library, starting from interesting scaffolds that are underrepresented in patent and medicinal chemistry literature. Approximately 1,000 distinct scaffold ideas have been designed by AMRI medicinal chemists and are available to the CLC. Every scaffold contains 2-3 points of diversity where the scaffold structure becomes the starting material for 5-10 chemical structures, which we term templates. See Figure 2 for a few representative structures of scaffolds. Templates need to be diverse within a scaffold to achieve a truly diverse parallel library, and scaffold intermediates must be synthesized at the 100 gm scale. Each template elaborates one or two points of diversity such that it can then be used as starting material for relatively small parallel libraries. We continually investigate additional conditions to achieve diverse parallel library chemistry. Figure 2. Representative structures The current parallel chemistry available: Amide formation Sulfonamide Reductive amination Alkylations Mitsunobu Urea formation Suzuki Buchwald SAr Click chemistry Physical Properties of the Compound Library Two factors create unique diversity of the CLC scaffold and template selection and the limitation of the number of analogs to a maximum of ca. 200 per scaffold. In addition: All libraries are designed to produce lead-like hits as exemplified in the acceptable ranges of physical and constitutive properties (see Table 1). Approximately 1-2 percent of compounds are expected to be fragment-like, obeying the rule of 3. In all cases, reactive compounds are excluded and compounds will pass mutually agreed pan assay interference compounds (PAIS) structural alerts. Libraries are designed to consist of compounds with a high degree of three-dimensional structure. H H OH R Table 1. Key properties monitored Property Lead-Like Acceptable Range Fragment-Like Acceptable Range MW <450 <350 clogp 0.5-4.5 <3.5 tpsa 30-120 umber of rings 5 <3 umber of aromatic rings 3 <2 Rotatable bonds 8
AMRI maximizes 3-D structure in two ways. First, a high fraction of sp3 carbons centers (Fsp3) is desirable in each compound. An average target is set at 0.4. By using scaffolds and templates with high Fsp3, the parallel library step can use aromatic R-groups, which can simplify the library chemistry. Second, AMRI calculates the principal moments of inertia of each compound. Plotting the relative components of the PMI of a compound allows it to be characterized as sphere-like, rod-like and disk-like. Spheres have three nearly equal dimensions, a disk has two nearly equal dimensions and a rod has effectively only one dimension. 2 The introduction of a large number of sp3 carbon centers can introduce stereochemistry into a molecule. Therefore, we have determined that racemic mixtures are acceptable while mixtures of diastereoisomers are avoided. Only single enantiomers are catalogued if there are two or more stereo centers. In Figure 3, we illustrate the distributions of these properties on approximately 27,000 compounds completed as part of the current CLC. In general, we have been able to keep all properties within acceptable ranges with few exceptions. Figure 4 illustrates that the compounds are generally diverse amongst themselves. Half of all compounds have no nearest neighbor with a Tanimoto coefficient greater than 0.7 in spite of the fact that all compounds share common templates and scaffolds with many other compounds (often 40-50 analogs per template). The relatively large number of identical compounds is due to the fact that we have been able to provide enantiomerically pure example of two stereoisomers. Finally, Figure 5 illustrates that the synthesized compounds have a reasonable 3D diversity as well. The shapes primarily span rod-like and disk-like molecules however, a significant number are best described as spheres. Figure 3: Distribution of several key properties in 27,000 synthesized compounds from CLC X Axis: Range for the property Y Axis: Percent of 27,000 compounds Molecular Weight clogp tpsa
Figure 4: Distribution of maximum pairwise similarities, demonstrating compounds diversity X Axis: Value of maximum pairwise Tanimoto coefficient to each compound (similarity calculated using Morgan fingerprints as implemented in RdKit; see RdKit.org) Y Axis: Percentage composition of library Distribution of Max Similarity Pairs PMI: ormalized Components Figure 5: ormalized principal moments of inertia: A measure of 3D diversity PR1 and PR2: Principal moment of inertia normalized for each molecule by dividing each axis by the longest axis of the molecule. Shapes primarily span rod-like and disk-like molecules; however, a significant number are sphere-like.
About AMRI AMRI, a global contract research and manufacturing organization, partners with the pharmaceutical and biotechnology industries to improve patient outcomes and quality of life. With locations in orth America, Europe and Asia, AMRI s team combines scientific expertise and market-leading technology to provide a complete suite of solutions in Discovery, Development, Analytical and Solid State Services, API Manufacturing and Drug Product. For more information about AMRI, visit www.amriglobal.com. 1 Fsp3 = sp3 carbons / total carbons 2 Sauer, WHB, Schwarz, MK; Molecular shape diversity of combinatorial libraries: A prerequisite for broad bioactivity. J Chem Inf Comput Sci 2003, 43, 987 1003. www.amriglobal.com Discovery Development Analytical and Solid State Services API Manufacturing Drug Product