Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Biology (Cell Biology, Microbiology, Molecular Biology)

Major Professor

Sameer Varma, Ph.D.

Committee Member

Kristina Schmidt, Ph.D.

Committee Member

Gary Daughdrill, Ph.D.

Committee Member

Bin Xue, Ph.D.


conformational fluctuations, Molecular Dynamics, Paramyxovirus, PDZ, machine learning, signal propagation


Allostery describes the phenomenon where perturbations in one region of a protein affect protein behavior in another non-overlapping region. Considerable efforts made over decades to understand the molecular basis of allostery, yet an overarching theory that can predict signaling pathways and contributions from chemical components is still lacking. In fact, molecular details in even the most well-studied of model systems, PDZ domains and GPCRs, remain unclear. In this dissertation I use molecular simulation methods to understand the role of allostery in the regulated entry of paramyxovirus into host cells, and also develop a new method to determine time-dependent signaling pathways in proteins.

Paramyxoviruses include notable pathogens like the Parainfluenza, Measles, Nipah, and NDV, which continue to cause significant loss of life in humans and animals. Paramyxoviruses use two membrane proteins to infect, the attachment protein and the fusion protein. Attachment proteins have a stalk embedded in the membrane and four receptor binding domains that are organized as a dimer of dimers at the end of the stalk. The binding signal is relayed from the receptor binding domains to the stalk, which then triggers the fusion protein to activate and fuse the viral and host cell membranes. The details of how the receptor binding signal is transmitted from the binding domain to the stalk remain to be elucidated. To better understand this process, we perform simulations of the Measles dimer as well as the NDV dimer and tetramer. Our key result is that receptor binding has little effect on the structures of individual receptor binding domains, but these small changes in structure coupled with changes in thermal fluctuations, can can alter dimer interfaces — in NDV, dimer interfaces rotate by about 10° , while the interface of the Measles dimer shows no change. In our previous studies of the Nipah dimer, which were validated subsequently by experiment, we had found that receptor binding led to very large changes in dimeric interface. Taken together, our results suggest that the molecular details of allostery are not conserved among the different paramyxoviruses; a result that stands in stark contrast to inferences derived from macroscopic experiments that suggest paramyxoviruses to have a universally conserved fusion mechanism. For NDV, we also provide a detailed signaling model that is consistent with experimental findings on the effect of site-directed mutagenesis on receptor binding and fusion. Our studies on the crystal structure of the tetramer, which is missing ~90 stalk residues, show that it does not make a stable system for simulation – the receptor escapes and the partial stalk segment loses its structural integrity. Investigating receptor-induced changes at the tetramer level, which is necessary for measles, requires inclusion of the entire tetrameric stalk.

The results we discussed above for Paramyxoviruses were obtained from equilibrium simulations, where we simulated proteins in receptor-bound and receptor-free states, and by comprising differences in structure and dynamics between these states, we identified the changes induced by receptors. These simulations, however, do not provide information on how signals propagate between allosteric sites. Understanding signal propagation requires generation and analysis of non-equilibrium data that captures the time-dependent switching between states. Here we develop a rigorous computational approach that quantitatively analyzes the time-evolution of allosteric signals in proteins, and identifies time-dependent signaling paths and allosteric signaling hubs. We use PDZ domains as our model system for which a lot of experimental data is available for testing and validation, and where signaling is known to occur through a combination of changes in structure and dynamics. The challenges here are to track and then analyze time-dependent changes in conformational “ensembles,” so that changes in both structure and dynamics are considered. We demonstrate that simultaneous tracking of changes in structure and dynamics can be accomplished by supervised machine learning, and we perform analysis on event tracking by using directed graphs. In its application to PDZ domains, we note that our method correctly distinguishes between experimentally characterized residues that affect receptor binding and those that do not affect receptor binding.

Overall, these studies provide new insight into the role of protein dynamics in allosteric signaling, and identify the specific roles of allostery in the regulated fusion machinery of paramyxoviruses. Additionally, this work provides a new method to determine time-dependent signaling pathways in proteins, which we anticipate will be applied in future to resolve key mechanistic issues in many biomedically important proteins that are controlled by dynamic allostery.

Included in

Biophysics Commons