Protein-Protein and Protein-Substrate Specificity in Two Unique Membrane Bound Proteins
This researcher is working on the characterization of protein-protein and protein-substrate recognition/specificity by sweet taste receptors and organic anion transporting polypeptides (Oatps), respectively. In each case, the protein structures have been constructed by homology modeling for use in structure based hypothesis generation. Of interest for the sweet taste receptor is the extracellular hydrophilic domain and its interaction with both small ligands and, more importantly, with the sweet-tasting protein brazzein. Protein-protein docking and molecular dynamics are used to understand how brazzein binds and elicits its sweet taste, with interest in surface complimentarity and backbone flexibility. The computational work is being used to drive mutagenesis studies on brazzein. The Oatp project is concerned with small molecule transport selectivity across the cell membrane. Oatps appear to be members of the large Major Facilitator Superfamily (MFS) of transporters which this researcher has used for modeling. This project uses small molecule docking and dynamics, pharmacophore modeling, and 3D-QSAR.
The Oatp project is concerned with small-molecule transport selectivity across the cell membrane. Oatps appear to be members of the large Major Facilitator Superfamily (MFS) of transporters, three of which have known three-dimensional structures. All three structures have been used to generate a homology model of Oatp member 1c1, which is being used to guide mutagenesis studies. In addition, the researcher is using small molecule docking and dynamics, pharmacophore modeling, and 3D-QSAR to understand and predict substrate selectivity for transport across various tissue types. The MSI tools/software used for these studies include homology modeling, molecular dynamics, sequence analysis, pharmacophore modeling, comparative molecular field analysis (CoMFA), and DFT and semi-empirical small molecule modeling.
Cluster Lensing With Hubble Frontier Fields
The primary goal of this project is the reconstruction of sky-projected density distribution within clusters of galaxies. Most of the mass is comprised of dark matter whose distribution cannot be seen directly, so has to be inferred. The inference is done using gravitationally lensed background sources. A typical rich cluster has about 10-100 lensed images. Because this number is relatively small compared to the spatial detail needed for adequate mapping of the clusters, the problem of cluster mass reconstruction is very under-constrained. There are several methods in use that do mass reconstruction, each with its own assumptions and priors.
This group's method is based on a genetic algorithm, called GRALE, which is a free-form, adaptive grid method that uses a genetic algorithm to iteratively refine the mass map solution. It efficiently explores the model space, does not get stuck in local minima, and explores the range of mass uncertainties quite differently from other existing methods. The reconstructed maps are very accurate, but the computation time is large, and only possible on supercomputers.
Given maps of reconstructed mass distribution with accompanying uncertainties, they can be used for two main purposes. One is to discover and examine the very first generation of galaxies that formed in the universe. The light travel time to them is nearly the age of the universe, making them faint and difficult to observe. Most of these are so faint that existing and future telescopes are not adequate. To solve the problem astronomers use "nature's telescopes," clusters of galaxies that act as gravitational lenses to amplify distant sources, by typical factors of between 10 and 50. However, these "telescopes" have very uneven "optics," i.e. mass distributions, hence the need for accurate mass reconstruction.
This research was featured in a Research Spotlight on the MSI website in July 2015.
Return to this PI’s main page.
Mental Maze Solving: A Study of the Visuospatial Processes in the Human Brain Using Functional Magnetic Resonance Imaging
Previous studies have investigated the processes that take place in the nervous system during visuospatial tasks using maze solving as a paradigm. Specifically, psychophysical experiments in humans and primates have shown that mental transversing of the maze path occurs when subjects are required to find the exit of a maze displayed on a computer screen. Neural recordings from the parietal lobes of monkeys have shown that neural activity during this task is related to various features of the maze. A commonly used estimation method, termed the population vector, uses neural activity to predict the direction of the maze that the monkey is tracing.
In this research, functional Magnetic Resonance Imaging (fMRI) is used to record the brain activity of healthy human subjects during maze solving. Preliminary results have shown that this activity correlates with this task, similar to the findings in other primates. The current analysis extends these results by studying the interactions of the activities in volumetric pixels (voxels) recorded in the brain. Specifically, the fMRI signal from each voxel will be prewhitened using an ARIMA model to avoid spurious results. The prewhitened time series will be cross-correlated to discover patterns of significant functional connectivities. The significant connections will then be used to define the edges of a network within the brain, while the voxels will be its vertices. Various measures of this network, such as the degree of each vertex, the centrality, and others will be analyzed in relation to the experimental covariates.
Return to this PI's main page.
Molecular Dynamics Simulation of Chemical and Biochemical Processes
The Gao group is continuing their investigations in several areas, including the interplay between protein dynamics and mechanism of enzymatic reactions; the development of novel quantum mechanical methods for studying energy and charge transfer processes in chemical and biological macromolecular systems; the simulation and modeling of vibrational Stoke shifts of probe molecules in proteins; and solvent effects on chemical reactions and interactions in condensed phases.
The group's approach is based on statistical mechanical Monte Carlo and molecular dynamics simulations, making use of combined quantum mechanical and molecular mechanical (QM/MM) potentials. The first project area involves molecular dynamics simulations of enzymatic reactions including the demethylation reactions catalyzed by an FAD-dependent enzyme and metalloenzymes, the final step in nucleotide UMP biosynthesis by OMP decarboxylase, thiamine-dependent enzymes, and proton-coupled electron transfer (PCET) processes in ribonucleotide reductase and in photosystem II. These studies will provide a deeper understanding of the reaction mechanism and the origin of catalysis. In addition, the group has initiated a study of the ultraviolet-8 activated dimer dissociation of UVR8, which triggers cellular response in plants.
The second project aims at the development of multistate density functional theory for charge transfer, and the explicit polarization (X-Pol) potential as a next-generation and quantum force field for biomolecular and materials simulations. These methods represents novel approaches to describe molecular systems and to determine the potential energy surface, and it goes beyond the so-called combined QM/MM approach, which was awarded by the 2013 Nobel Prize in Chemistry.
The third project is aimed at developing a simulation system to understand the electrostatic environment inside of an enzyme's active site.
The final project area focuses on development of novel computational techniques including mixed molecular orbital and valence bond (MOVB) and an X-Pol based reactive force field and applications to modeling solvent effects on a variety of chemical reactions and reaction networks.
Return to this PI's main page.
Genomics of Host-Microbiome Interactions
With the advent of next generation sequencing, it has become possible to examine entire genomes and transcriptomes of humans and other animals with relative ease. These data have been used to validate previously discovered biological mechanisms as well as to discover new phenomena that are implicated in diseases, such as cancer. This technological advance has also opened up new research possibilities by allowing scientists to survey and quantify the microbiome, the collection of microorganisms that reside in and on the bodies of humans and other animals. Again, researchers have leveraged this new source of information to make breakthroughs in human disease that are caused by alterations in the microbiome, including diabetes, inflammatory bowel disease, and obesity. Although the human microbiome is influenced by environmental factors, bacteria also interact with human cells through immune system and metabolic pathways.
This group's research aims to understand and characterize host-microbiome interactions in a variety of conditions. They have several ongoing projects, aiming to answer the following biological questions:
- What are the molecular mechanisms controlling host-bacteria interactions? Which genes and pathways are involved in both the host and microbiome side?
- How does host genetic variation control interactions with our microbiome? What are the effects of different environments and genetic backgrounds across human populations?
- How did the complex symbiosis between us and our microbiome evolve throughout human history? Can we identify signatures of coevolution in human and microbial genomes?
- How do host-microbiome interactions control susceptibility to complex disease? What are the unique roles of host genetics, bacterial communities, and environmental exposures?
Each of these research projects involves analyses of large genomic and metagenomic datasets using resources from MSI, including installed software, CPU time, storage space, and the parallel computing environment.
This research was featured in a Research Spotlight on the MSI website in October 2015.
Return to this PI's main page.
Neural Mechanisms of Personality and Decision Making
Psychologists are increasingly focused on identifying the biological substrates of personality traits. One promising method for identifying brain systems associated with personality traits is magnetic resonance imaging (MRI). Functional MRI is used to identify personality traits that are associated with brain processes during specific tasks. However, structural MRI can also be used and has the advantage that hypotheses about many personality traits can be examined simultaneously, because associations of traits with brain structure are not dependent on any specific task. In a previous test of the hypothesis that personality traits would be associated with the volume of specific brain structures, these researchers found support for a model specifying brain systems underlying each of the Big Five personality traits (Extraversion, Neuroticism, Agreeableness, Conscientiousness, and Openness/Intellect). However, this study did not differentiate between volume in gray matter versus white matter, an important anatomical distinction in the brain. Nor did it examine more specific personality traits based on subdivisions of the Big Five, which describe personality at a less general level of resolution and would allow testing of more specific biological hypotheses. The current project uses more finely differentiated measurements of personality traits in conjunction with a more sophisticated assessment of brain structure that parcellates gray and white matter and provides an index of gray matter thickness throughout the cortex. Cortical thickness has been shown to have important correlates, from learning of motor tasks to psychopathology. Assessing cortical thickness from MRI data is a computationally intensive process that can be accomplished using the MRI analysis program FreeSurfer. As this program typically takes up to 48 hours to process data for a single subject, and as this project examines associations of cortical thickness with personality in 300 subjects, the resources of MSI are invaluable. The researchers are also doing a more sophisticated DTI analysis for up to 200 individuals.
Return to this PI's main page.