STAT3 (Signal transducer and activator of transcription 3) is constitutively active in human cancers such as breast cancer, lung cancer, multiple myeloma and others. Upon phosphorylation, STAT3 forms homodimers that are translocated to cell nucleus where they engage in the transcription of anti-apoptotic genes (JAK-STAT pathway). Preventing the dimerization of STAT3 by blocking its SH2 (Src-homology 2) domain with phosphopeptide-based mimics (peptidomimetics) is an attractive strategy to inhibit the activity of STAT3 and is being actively pursued by our collaborator's lab at MD Anderson Cancer Center. We are developing computational modeling algorithms for determining binding interactions between the peptidomimetics and the SH2 domain of STAT3. Since structures of the SH2-peptidomimetics complexes are unavailable experimentally, computational modeling of the complexes could be very useful.
STAT3 structure from Protein Data Bank (PDB ID 1BG1). The SH2 domain is shown in pink color.
Our goal is to compute accurate conformation(s) of the peptidomimetics bound to the SH2 domain and also correctly estimate the binding affinities of the peptidomimetics, i.e., a measure of how tightly they bind to the SH2 domain. We are interested in developing computer-aided docking programs/protocols for modeling the complexes. Finding accurate conformations is challenging because the peptidomimetics, that we are focusing on, are not the typical small drug molecules. Instead they are large compounds with many rotatable internal degrees of freedom (DOFs) which makes the conformation space of such compounds very high-dimensional and thus very difficult to explore for binding conformations. Following are the main highlights of our work so far:
We conducted a benchmark study using docking programs AUTODOCK and VINA that confirmed that these docking programs are insufficient for our purpose. Not only these docking progams computed inaccurate conformations for many peptidomimetics, the computed binding affinity estimates also did not correlate well with experimental binding affinity values. The computational docking time required was also substantial.
We developed a fast and more accurate incremental docking method to compute docked conformations of large compounds. The method was validated in docking experiments where large compounds (with more than 6 rotatable DOFs) from the core set of the PDBbind dataset were docked to their respective receptors.
We docked peptidomimetics to the SH2 domain using our incremental docking method. Using incremental method, we are able to dock the peptidomimetics in a computationally fast manner. We ran molecular dynamics (MD) simulations using AMBER that are providing us some very interesting insights. More work is underway to further improve the accuracy of docked conformations and better estimate the binding affinities.
A peptidomimetic (in green color) is docked to the SH2 domain of STAT3. Cartoon-stick and surface representations are shown.
This work is done in collaboration with Prof. John S McMurray from the University of Texas MD Anderson Cancer Center.