Research Interests

Collaborators / Institutions / Kinase Inhibitors

1. Development of small molecule kinase inhibitors

Reversible protein phosphorylation is the major mechanism by which cellular systems transduce signals. The transfer of phosphate from ATP to protein hydroxyl groups is catalyzed by over 500 distinct protein kinases in mammalian systems. Kinase activity is usually strictly regulated by additional phosphorylation events, cellular localization, inhibitory or activating protein partners, protein degradation or gene transcription. Not surprisingly, deregulation of protein kinase activity through mutation to constitutively active alleles, loss of negative regulators, and chromosomal rearrangements that lead to the formation of oncogenic fusion proteins are associated with disorders ranging from leukemias to diabetes. As a consequence, there is currently tremendous interest in the development of selective kinase inhibitors both as potential pharmaceuticals and as biological tools.

We have taken three different approaches to synthesize new kinase inhibitors. First, we have employed combinatorial diversification of privileged natural product scaffolds. One successful example of this strategy has been the development of Purvalanol A as a nanomolar inhibitor of Cdk1/CycB, Cdk2/cycA, and Cdk5/p25. More recently, we synthetically elaborated the marine sponge derived compound Hymenialdisine to create analogs with distinct specificities and biological activities (Figure 1). Second, we have prepared a diverse collection of 150,000 heterocyclic compounds that were designed based on a general ATP-site pharmacophore model using directed-sorting technology (Figure 2, 3). Third, we have used structure-based design of new scaffolds to compound that bind preferentially to inactive kinase conformations and are capable of potently inhibiting a variety of tyrosine and serine/threonine kinases such as Abl, PDGFR?, FLT3, c-kit, CSK, b-raf, Src, and p38.

Following synthesis of new compounds, we use two distinct but complementary approaches to discover and optimize biological function: (1) target-based screening (enzymatic or cellular) or (2) cellular phenotypic screening (Figure 4). Target-based screening supported by cellular assays that precisely monitor the activity of interest and that can guide chemical optimization is the most direct means to obtain functional inhibitors. We have used target-based screening to develop sulfotransferase inhibitors, tubulin depolymerizers (myoseverins), inhibitors of malarial cyclin-dependent kinases, selective agonists of sphingosine-1-phosphate receptors (SEW02871), selective inhibitors of mutant arginine methyltransferases, and Cdk/GSK3? inhibitors such as Hymenialdisine analogs (Figure 5). In contrast, phenotypic screening provides a means to interrogate a pathway in an unbiased fashion with small molecules (Figure 6). Provided that the molecular target(s) of the compound can be identified (usually by affinity chromatography, genetic complementation, or expression profiling), phenotypic screening can deliver new biological insight in addition to yielding useful small molecules. We have used phenotypic screening to identify a small molecule (diminutol) capable of inducing a shrunken mitotic spindle phenotype in Xenopus extracts and used the compound to identify the potential target as a ubiquinone oxidoreductase (Nqo1). We have also identified small molecules capable of inducing cardiomyogenesis and another compound (purmorphamine) that is able to activate the hedgehog signaling pathway.

Directed-Sorting Technology

Our general approach towards the construction of kinase-focused combinatorial libraries is to generate additional functional diversity around heterocyclic templates that are known or possess similar pharmacophoric features to known kinase inhibitors. In order to enable efficient library productions we use directed-sorting technology. This approach involves encoding each compound (using optical bar-codes or radio-frequency tags) through the synthetic process so that compounds can be pooled during synthetic steps without losing track of their identity (Figure 7). For example if you wanted to prepare a 1000-membered library on a scaffold with three sites for diversification in a spatially separate fashion you would perform the first diversification step (10 reactions), then the second step (additional 100 reactions) and finally the third step (additional 1000 reactions) for a total of 1110 reactions. Directing-sorting allows you to pool the compounds thereby allowing 1000-distinct compounds to be prepared using only 30 reactions (3 x 10).

One lesson learned over the last ten years of combinatorial chemistry is that iterative synthesis and biological evaluation of a smaller number of compounds is almost always superior to the synthesis of large libraries in order to identify compounds with new biological activity. In recognition of this observation, we tend to make libraries with approximately 100-500 members around a scaffold with two-diversification points and 500-1000 members for three-diversification points. Once biologically active compounds are identified, a combination of small-focused libraries and “one-off” synthesis is used to optimize compound properties. Integrating the traditional medicinal-chemistry approaches of one-by-one synthesis with libraries is key to insure that SAR that can not be investigated by the combinatorial scheme is investigated early in a compound optimization program. For example it is often more efficient to alter the fundamental scaffold that is being worked on (“scaffold-morphing”) than to continue preparing more derivatives the same scaffold to solve a particular problem.

“Inactive”-kinase conformation binders

Currently all known protein kinase inhibitors can be grouped into three classes: (1) type I inhibitors, which bind exclusively in the ATP-binding site of the kinase; (2) type II inhibitors, which utilize an adjacent allosteric site made accessible when the “activation loop” is folded away from the active site cleft; and (3) type III inhibitors, which bind to a site remote from the ATP binding cleft. To date the vast majority of kinase inhibitors fall in the first class while only three compounds have been biochemically and crystallographically proven to function as type II inhibitors: Gleevec (STI571), an inhibitor of Bcr-abl, PDGFR?, and c-Kit; BAY43-9006 and its congeners, inhibitors of Raf kinase; and BIRB796, an inhibitor of p38 kinase (Figure 8, 9). For an inhibitor to achieve a type II binding conformation typically requires the kinase adopt a putative “inactive conformation”; therefore, type II inhibitors function in part by preventing the conformational change that leads to kinase activation whereas type I inhibitors usually bind the active conformation directly to inhibit kinase activity. Due to the highly conserved nature of the kinase active site combined with the fact that all active kinases must assume a conformation conducive to the catalysis of ATP phosphate transfer, type I inhibitors tend to be less selective than type II inhibitors which can recognize more diverse inactive kinase conformations (although there are clearly some exceptionally selective type I kinase inhibitors). Another advantage of type II inhibitors is that although their binding partially overlaps with the ATP binding site, the inactive conformation that they recognize has a significantly lower affinity for ATP. To date all type II inhibitors have been discovered serendipitously through random screening and medicinal chemical optimization and only later confirmed to function as inhibitors of kinase activation by inspection of co-crystal structures and measurement of relative affinities to active and inactive kinase. From inspection of the co-crystal structures, a general pharmacophore for a type-II inhibitor can be proposed that involves a series of hydrophobic interaction sites and two pairs of H-bonds that anchor the binding mode: one or two H-bonds to the backbone amide of the “hinge-region” amino acid and a second pair to the side-chain of a conserved aspartate in the ?C helix and the backbone NH of the conserved aspartate of the “DFG”-motif (Figure 10). This pharmacophore model can be readily used to design diverse libraries of potential new type-II kinase inhibitors using solid-phase or solution chemistry as appropriate (Figure 10).

top

2. Exploiting Kinase-Directed Libraries

The kinase-directed libraries are screened for inhibitors in both a targeted and pathway-based fashion. The targeted approach will involve screening the library against an individual kinase in a biochemical format (typically scintillation proximity assay (SPA) or time-resolved fluorescence (HTRF)) or in cellular assay where survival and proliferation is dependent on the activity of the kinase of interest(typically Ba/F3 transformed to cytokine independence (IL3) with oncogenic tyrosine kinase fusion protein: bcr-abl, tel-PDGFR?, etc) (Figure 11). The targeted approach will also involve screening a particular compound of interest against a panel of kinase enzymatic or cellular screens in an effort to define kinase selectivity (Figure 12). The purpose is to annotate the kinase selectivity of a given library compound such that a target hypotheses can be generated when the compounds are subsequently identified in pathway-based screens. The pathway approach will usually involve cellular screens with reporter gene, viability, or microscopic visualization read-outs (Figure 13). Systems of particular interest are (1) antiplasmodial kinase inhibitors (screens of plasmodia infected erythrocytes), (2) antiviral kinase inhibitors (hepatitis C virus using replicon-based systems or Dengue using image-based screening), (3) developmental pathways (Wnt, Bmi, hedgehog using reporter gene readouts), and (4) cellular localization screens (forkhead, Gli, beta-catenin using high-content imaging based screening).

3. Developing and exploiting selective ligands for sphingosine-1-phosphate receptors

FTY720 is an aminoalcohol lipid with immunosuppressive activity that was discovered serendipitously during a medicinal chemistry campaign aimed at reducing the toxicity and enhancing the physiochemical properties of a natural product myriocin (Figure 14). Myriocin had been reported to be an inhibitor of serine palymitoyltransferase and possess both immunosuppressive and antifungal activity. Interestingly, the chemical “simplifications” leading from myriocin to FTY720 resulted in a loss in biological activity in mouse allogenic mixed lymphocyte reactions (MLR) while enhancing the therapeutic window in in vivo rat skin allografts. Further research revealed that this disconnect in the structure-activity relationship between MLR and the skin transplant model was due the ability of FTY720 to induce immunosuppression by a unique mechanism: instead of effecting lymphocyte activation, treatment with FTY720 in vivo induces the reversible sequestration of lymphocytes to secondary lymphoid tissues (lymph nodes and Peyer’s patches) and blocks the egress from the thymus thereby removing lymphocytes from peripheral circulation (Figure 15). FTY720 is stereospecifically phosphorylated by sphingosine kinases in vivo to form the biologically active phosphate ester metabolite (FTY720-P, Figure 16) which is a close chemical mimic of sphingosine-1-phosphate, an important phospholipid secondary messenger. A natural hypothesis was therefore that FTY720-P could function as a S1P-receptor agonists and indeed FTY720-P potently agonizes of four of the five known sphingosine-1-phosphate receptors (S1P1-5), a subset of the seven trans-membrane G-protein coupled receptors.

In order to determine the minimum S1P-receptor selectivity profile required to induce lymphopenia, two separate efforts were employed to find selective agonists. The first approach involved monitoring the structure-activity relationships between S1P receptor selectivity and ability to induce lymphopenia in vivo for amino alcohol analogs of FTY720. Although this approach was complicated by the structures needing to simultaneously be substrates of sphingosine kinase and S1P receptor agonists, it did result in the identification of potent agonists with a range of selectivities against the S1P1 receptors. The second approach involved a high-throughput 384-well screening using Ca2+-flipper assays of various S1P receptors against 60,000-membered commercially available GPCR-priviledged libraries. This approach resulted in the identification of a novel aromatic oxadiazole tetracyclic S1P1-selective agonist SEW02871. SEW02871 was able to induce a complete lymphocyte sequestration. These studies demonstrate that S1P1-selective agonists are able to mimic the ability of FTY720 to induce lymphopenia and induce therapeutically useful levels of immunosuppression. Although only S1P1-selective agonists have been reported to date, other chemical efforts have reported the development of S1P1/S1P4 benzimidazole dual agonists (Figure 17).

We are now turning our attention to the development of selective agonists and antagonists of the other S1P-receptors (S1P2-5). We hope to identify these compounds through a combination of small molecule library screening and SAR-guided chemical diversification strategies starting from pan-S1P agonists. In order to understand how chemical modifications may impact selectivities towards S1P receptors a homology model of S1P1 has been developed based on the crystallographic structure of rhodopsin. This model predicts that S1P and FTY720-P form a series of salt bridges between the phosphate and arginines 120 and 292 and salt bridge between the ammonium group and glutamate 121 as well as extensive hydrophobic contacts to the lipophilic trans-membrane helices. The potency and selectivity of SEW02871 towards the S1P1 receptor demonstrate that even compounds without strong “headgroup” interaction can result in compounds with pharmacologically relevant levels of activity. Work from Merck has revealed that the amino-phosphate headgroups can be replaced by aminocarboxylates (Figure 17). Once we have obtained selective agonists, we hope to use these compounds as reverse-pharmacological agents to decipher the biological function of the S1P2-5 receptors in a variety of cellular and animal systems.

top

4. Developing small molecule ligands for new biological targets

Currently there is only a small fraction of the proteins encoded in human genome for which there are known small molecule effectors. Typically the protein families for which this is true have either been of long-standing interest to the pharmaceutical industry: GPCRs, proteases, kinases, ion channels or the targets of natural products: tubulin, FKBP/Calcinuerin, D-ala-D-ala ligase, DNA, etc. Some major challenges for the chemical biology community are:

1. What other proteins can have their activity modulated by small molecules?
2. What subset of these are therapeutically relevant?
3. How do we find/make and eludicate the mechanism of these new compounds efficiently?

While it is likely that both forward and reverse chemical genetic approaches will play important roles in answering these questions, there is a need for new developments in these approaches that can be applied in a robust and generic fashion. One approach we are interesting in developing further is to use yeast (S. cerevisae) as a model organism to both screen and identify the targets of new compounds. Despite yeast being one of the most utilized model organisms, there is a surprising dearth of compounds with antifungal activity that have a distinct mechanisms of action. Some possible reasons for this are: (1) as yeast are unicellular eukaryotes they have evolved to withstand a wide variety of chemical assault (through efficient drug pumps, detoxification enzymes, and multiple redundant pathways), (2) yeast have primarily been used by biologist doing genetic or biochemical manipulations and less by chemists looking for new bioactive compounds and (3) many good antifungal drugs have already been discovered which has reduced interest from the pharmaceutical industry. We are interested in finding new compounds that exhibit biological activity in wild-type yeast or in yeast that are genetically deficient in specific pathways. While inhibition of yeast growth is the most obvious compound-induced phenotype to screen for, there are a large number of other phenotypes of interest that can be examined. By searching for small molecules that only induce a phenotype in a specific yeast background we will greatly increase our chances of finding active compounds and be directed towards a plausible biological hypothesis regarding mechanism of action. For example if we screen a yeast strain that displays hyper-sensitivity to the microtubule depolymerizing drug benomyl, we are likely to find other compounds that directly target tubulin or modulate tubulin in an indirect fashion. We plan to use three complementary approaches to elucidate the mechanism of action of newly identified compounds: (1) complementation of compound-induced phenotypes with genomic DNA-libraries, (2) sensitivity or resistance screen with a complete collection of viable null-alleles, and (3) biochemical affinity chromatography-based approaches.

Collaborators

Young-Tae Chang (link: http://homepages.nyu.edu/~ytc1/)
Sheng Ding (link: http://www.scripps.edu/chem/ding/)
Laurant Meijer (link: http://www.sb-roscoff.fr/CyCell/)
Rebecca Heald (link: http://mcb.berkeley.edu/labs/heald/)
Luis Schang (link: http://www.biochem.ualberta.ca/faculty_detail.php?id=14)
Pamela Silver (link: http://research.dfci.harvard.edu/silverlab/)
Hugh Rosen (link: http://www.scripps.edu/imm/rosen/)
Elizabeth Winzeler (link: http://www.scripps.edu/cb/winzeler/index.html)
Priscilla Yang (link: http://yanglab.med.harvard.edu/)

Institutions

Dana-Farber Cancer Institute (DFCI) (link: http://www.dfci.harvard.edu/res/)
Harvard Medical School (HMS) (link: http://hms.harvard.edu/hms/home.asp)
Genomics Institute of the Novartis Research Foundation (GNF) (link: http://web.gnf.org/index.shtml)
Novartis Institute for Biomedical Research (NIBR) (link: http://www.nibr.novartis.com/)
Novartis Institute for Tropical Disease (NITD) (link: http://www.nitd.novartis.com/home.shtml)
Burnham Cancer Institute (http://www.burnham.org/FacultyAndResearch/CancerCenter.asp)
The Scripps Research Institute (TSRI) (link: http://www.scripps.edu)
Broad Institute (http://www.broad.mit.edu/)

Kinase Inhibitors

Commercial Protein Kinase Inhibitors :
(http://www.proteinkinase.de/html/protein_kinase_inhibitors.html)