My laboratory has two major research interests. In one set of projects, we use chemistry to design and synthesize small molecules, which we then use to illuminate cellular processes relevant to human disease. We have designed electrophilic molecules that selectively inactivate protein kinases; inspired by natural products, we have discovered a cyclic peptide that prevents the biogenesis of secreted proteins involved in cancer and inflammation. In a second major project area, we study the mechanisms by which membrane-associated signaling proteins (including GTPases, protein/lipid kinases, and scaffolding proteins) create and locally amplify networks of actin filaments. By polymerizing adjacent to the membrane surface, such actin networks generate a local protrusive force, which is essential for cell and organelle motility.
Cysteine-Targeted Kinase Inhibitors
Protein kinases regulate all aspects of cellular and organismal physiology. Because aberrant kinase activity has been implicated in many diseases (especially cancer), kinases comprise a major class of therapeutic targets. The discovery of selective protein kinase inhibitors, usually achieved by high-throughput screening and structure-based design, continues to be a significant challenge due to the highly conserved ATP-binding pocket shared by all 500 human kinases.
Starting with a structure-based sequence alignment of all human kinases, we identified 20 positions likely to be oriented toward the ATP pocket. We then searched for nonconserved cysteines at these positions. This "structural bioinformatics" analysis yielded more than 100 human kinases with cysteines in various regions of the ATP pocket. Our motivating hypothesis was that we could convert a low-affinity ligand with poor selectivity into a selective inhibitor by attaching a weak electrophile onto the ligand at just the right spot. We designed a fluoromethylketone-substituted ligand (fmk) that we predicted would recognize two selectivity filters found in p90 ribosomal protein S6 kinase (RSK), but no other kinases: (1) a cysteine near a flexible loop, and (2) a threonine "gatekeeper" to a deep, hydrophobic pocket. At nanomolar concentrations, fmk irreversibly inactivates RSK1/2 in human cells without inhibiting any other kinases (>130 kinases tested to date). These studies argue against the widely held view that electrophilic inhibitors are inherently nonselective. On the contrary, the electrophilic fluoromethylketone is essential for selectivity. It should be possible to target nonconserved, noncatalytic cysteines in other protein classes, beyond the kinase superfamily. Although our primary motivation was to test hypotheses regarding protein/electrophile recognition and develop useful tools for deciphering RSK-signaling pathways in cells, recent studies in collaboration with other laboratories suggest that inhibiting RSK may be of therapeutic benefit for heart disease and certain cancers.
Our success in developing kinase inhibitors that derive their potency and selectivity via covalent bond formation with a nonconserved cysteine suggests that embedded within the druggable genome is a vast "cysteinome" that can be exploited pharmacologically with properly tuned, electrophilic small molecules. We know little about the molecular recognition principles that govern thiol-electrophile interactions in proteins. How does the microenvironment of a given Cys (e.g., proximal general acids/bases) facilitate the formation of covalent bonds with different electrophiles? Which electrophiles are "chemically tuned" for a given Cys microenvironment? A major thrust in my lab will be to develop new experimental approaches for defining the druggable cysteinome. In addition to revealing those Cys that preferentially form covalent adducts with attenuated electrophilic functional groups, this approach will ideally provide electrophilic leads. Using structure-based design and chemical synthesis, our goal is to develop these initial leads into potent, selective inhibitors against therapeutic targets for which no useful inhibitors are known.
New Chemical Probes for Exploring the Kinomes of Disease-Causing Parasites
Eukaryotic parasites cause many of the world's "neglected" diseases, including malaria, leishmaniasis, sleeping sickness, and Chagas disease. Given recent success in the development of kinase inhibitors as safe drugs, parasitic kinases have emerged as potential therapeutic targets for the treatment of neglected tropical diseases. One challenge is that parasitic kinomes are vast, and with a few notable exceptions, it is not known which kinases are essential for growth. For example, the Trypanosoma brucei genome encodes 156 protein kinases, only a few of which are clearly orthologous to essential human or yeast kinases. Given a complete inventory of a parasite's kinome (as deduced from genomic sequence), which kinases are most likely to be good drug targets?
We have started developing new pharmacological tools to address this challenge. We designed and synthesized multispecific (or, semipromiscuous, depending on your point of view) irreversible kinase inhibitors with three critical elements: (1) a recognition scaffold designed to interact with a small subset of trypanosomal protein kinases, (2) a "warhead" designed to react irreversibly with the strictly conserved catalytic lysine, and (3) an alkyne tag that can be captured with "click chemistry" (copper-catalyzed bioorthogonal conjugation with an alkyl azide) for affinity purification of the covalently attached protein kinase target. These inhibitors kill T. brucei parasites at nanomolar concentrations and covalently inactivate a subset of the T. brucei kinome. They will likely reveal one or more previously uncharacterized therapeutic targets for sleeping sickness and, possibly, Chagas disease (caused by a related parasite). Our kinase-directed probes can be used to explore the kinomes of parasites that are difficult to study by standard genetic methods (RNA interference, targeted knockout), including the causative agents of Chagas disease, leishmaniasis, and malaria.
Cotransins: Substrate-Specific Inhibitors of Cotranslational Translocation
Inspired by a fungal cyclodepsipeptide called HUN-7293, we designed and synthesized a simplified version, which we call cotransin. Like HUN-7293, cotransin potently inhibited the expression of a pro-inflammatory cell adhesion molecule (VCAM1) in stimulated human endothelial cells. Until recently, the molecular mechanism of these compounds' biological activity was a complete mystery.
We followed a trail of cell biological clues that eventually led us to the endoplasmic reticulum (ER) as the site of cotransin's activity. Using a cell-free system, we found that cotransin abolished the cotranslational translocation of nascent VCAM1 into the lumen of ER microsomes, whereas it had no effect on VCAM1 translation. Cotransin was highly specific, inhibiting the translocation (and secretion) of a small subset of proteins in cells and in the cell-free system. Even more remarkable, sensitivity of a given secretory protein to cotransin was determined solely by its N-terminal signal sequence, which targets nascent secretory proteins to the Sec61 translocation channel in the ER membrane. By a mechanism that is still obscure, cotransin binds directly to the large subunit of the Sec61 translocon and prevents it from opening in response to a subset of N-terminal signal sequences. Besides suggesting a new pharmacological strategy for targeting proteins involved in cancer and inflammation, our studies with cotransin provide the first clear evidence that signal sequences are themselves regulatory elements that can be differentially modulated to affect the functional expression of secreted and membrane proteins.
A long-term goal is to elucidate the full set of human proteins that are affected by cotransins and the "rules" that govern the sensitivity or resistance of specific signal sequences. Ultimately, we hope to use our molecular-level understanding of cotransins' effects to design small molecules that block cotranslational translocation with even greater selectivity.
How Soluble Signaling Proteins Self-Assemble into Force-Producing Cytoskeletal Networks on Membrane Surfaces
In eukaryotic cells, actin filaments are nucleated by membrane-associated signaling complexes comprising Rho-family GTPases and their downstream effectors. It has long been appreciated that polymerizing actin filaments generate forces required for membrane protrusion and cell motility. Unknown are the molecular mechanisms by which dynamic, branched actin networks nucleated by the Arp2/3 complex attach to and exert force against membranes. In a recent study, we showed that actin networks nucleated by N-WASP and the Arp2/3 complex dynamically attach to the membrane via direct interactions between actin filament ends and a short peptide motif on N-WASP (WH2 domain).
Fitting together the pieces of this mechanistic puzzle required the reconstitution of actin-powered membrane movement from purified components, which we achieved for the first time. Using a mixture of eight soluble proteins and lipid bilayers supported on silica microspheres, we reconstituted a signal transduction pathway that begins with the activation of geranylgeranylated Cdc42 (initially bound to its negative regulator, RhoGDI) at the membrane surface. This promotes the assembly of a polarized, force-generating actin network and the persistent motility of the lipid-coated beads. This reconstituted system opens the door to quantitative biochemical studies of membrane-associated signaling components (including protein kinases and phosphatases) whose "emergent properties" result in the physical movement of macroscopic objects.