Our group is currently engaged in the development
and application of computational methods to study bio-molecular systems (such
as proteins and nucleic acids), often with explicit all-atom representation
of both solvent and the macromolecules. Typical calculations involve molecular
dynamics simulations of the system to provide an atomic-detail picture of
the behavior of a single molecule.
Protein Folding and Aggregation
The self-assembly of proteins into their native state from random coil state to carry out biological functions has been an extensively studied problem for over a few decades. Understanding the mechanism of folding would enable us to predict protein structures more accurately and to design new proteins. Due to the extreme difficulties in probing the folding process with experimental methods, computer simulations have gained importance in filling this gap. The current challenge lies in the understanding how particular chemical detail in proteins lead to particular protein structures and folding mechanisms.
Protein misfolding
has been linked to a number of human diseases, including cystic fibrosis, Alzheimer’s
disease and other amyloidoses, and prion spongiform encephalopathies such
as Creutzfeldt-Jacob disease. An understanding
of the competition between folding and aggregation is critical to preventing misfolding
and misassembly of proteins. We have carried out series of computational
studies on short peptides to study their aggregation behavior.
Simulation Method Development
Because of large number of particles presented and
fine integration time step required, it has been difficult to study long-time
behavior of bio-molecules at time scales important to the biochemical events.
One area of focus in our group is to develop new algorithms for accurate
and efficient simulations. As part of the
AMBER
development team, we are developing highly efficient simulation algorithms
that can take advantage of modern massively parallel computers and to allow
us to study the long-time dynamics of biomolecules. We are aslo developing
efficient conformational sampling methods that allows us to sample substantially
more conformational space than that dictated by the fundamental dynamics.
Such methods will facilitate applications in structural refinement.
Structure and Dynamics of G-Protein Coupled Receptors
G-protein coupled receptors (GPCR) are membrane proteins.
They typcailly act as the initiators of many signal transduction pathways.
They can be activated/deactivated by external signals and interact
with the G-proteins which act as signal transducer.
Because of their convenient location and their physiological significance,
the GPCR superfamily receptors have been the premier targets of drug development effort.
About 50% of the existing drugs on the market are designed to interact GPCRs.
Despite the effort, it is still rather challenging to obtain high resolution
membrane protein structures. So far, only one GPCR structure has been solved.
Fortunately,
GPCRs are believed to share a common 7 transmembrane archeticture
with relatively flexible extra-cellular and cytoplasmic domains.
Therefore, computational modeling has played important role to provide detailed
and accurate information on the structure and dynamics of GPCRs.
DNA-protein interaction
Computer Aided Drug Design
We are currently collaborating with
Dr. Honggao Yan from Michigan
State University to study 6-Hydroxymethyl-7,8-dihydropterin pyrophosphokinase
(HPPK) as a potential drug target. HPPK is an enzyme that catalyzes the
first reaction in the folate pathway. As mammals obtain folate through dietary
means while microorganisms must synthesize this essential metabolite de
novo, that makes HPPK an ideal target for antimicrobial agents. Initial
studies will focus on molecular dynamics simulations to understand the induced-conformational
changes in the active-site to provide important structural and dynamics
information for structure-based drug design.
HIV integrase is an emzyme responsible for integrating the viral DNA to the host DNA.
Shown on the left is the crystal structure of the catalytic (core) domain.
In this structure, however, the loop overhanging the catalytic site (marked by
Mg2+ ion shown as the
green sphere) is partially disordered, making it rather difficult to interpret
the functional role of the loop. In collaboration with Jim Briggs of University of
Houston, we studied the dynamics of the core domain and found that the loop can
potentially undergo conformational transition between the open and closed forms.
Computing Resources
Major supercomputer resource is provided by National
Partnership for Advanced Computational Infrastructure (NPACI) and Pittsburgh
Supercomputing Center (PSC).
Local clusters:
Elan cluster: 40
dual-CPU Pentium IV Xeon 2.4GHz, compute nodes, 3 storage nodes (1.8TB
RAID 5).