Contact Information

    Theoretical and Computational Biophysics Group, Beckman Institute

    University of Illinois at Urbana-Champaign

    405 N. Mathews, #3117, Urbana, IL 61801

    Office phone: 217-244-1851       Email:


Research accomplishments

The underlying theme of my research is to characterize ways in which laws of physics yield biomolecular function over a wide range of temporal and spatial scales. Specifically my work focused on developing multiscale theoretical and computational methods to extend all-atom simulations to biologically relevant timescales [1-7]. These methods enabled me to study molecular machines that participate in many key cellular processes, including DNA repair, gene expression, protein transport, protein degradation.

1. Advanced sampling techniques

Molecular dynamics (MD) simulations, though capable of providing atomic details, are limited in the study of molecular machines owing to the challenge that these systems function on a millisecond time scale which, for a long time, could not be easily covered computationally. One of my most important development involves advanced sampling techniques based on the so-called transition path theory, to study the mechanochemical energy transduction in molecular machines (for theoretical details see Supporting Information in Ref.4). To obtain the transition paths as well as the related free energy landscape and kinetics, I have implemented the string method with swarms of trajectories combined with the milestoning technique with NAMD. Recent progress included further development of the milestoning method enabled by a multiplex replica algorithm, which allows very efficient allocation of computational resource among a large number of independent MD trajectories [2*]. The packages can be found here.

2. Mechanochemical energy transduction and allostery in molecular machines

(2a) RNA translocation by Rho helicase. The advanced computational approach described above enabled me to describe the highly coordinated motion of Rho hexameric helicase while it unidirectionally translocates RNA (Fig. 1) [4]. Our simulations show that the release of hydrolysis product (ADP+Pi) triggers the force-generating process of Rho through a 0.1 millisecond-long conformational transition, of which a similar time scale is also seen in experiments. The study not only revealed in new detail the mechanism employed by ring-shaped ATPase motors, for example the use of loosely bound and tightly bound hydrolysis reactant and product states to coordinate motor action, but also provided an effective approach to identify allosteric sites of multimeric enzymes in general. See more information in TCBG news, and the research page, as well as a youtube video made by me.

Fig. 1. Mechanism of RNA translocation by Rho helicase. ADP+Pi release triggers Rho transitioning from an initial state I to a final state F. The initial ligand binding states, after a 60° clockwise rotation around the z-axis, are the same as those of the final state. In going from I to F, a molecule of RNA is propelled through Rho's central pore. The schematic free energy landscape, governing the transition, is shown in the middle.

(2b) Protein unfolding by ClpX unfoldase. Following the footstep of the study on Rho, we applied the pathway sampling techniques to study how an unfoldase ClpX uses its central pore loops to translocate polypeptides during its mechanochemical cycle (in collaboration with Prof. Andreas Martin, UC Berkeley) (Fig. 2) [3*]. The ClpX ATPase belongs to the large family of AAA+ unfoldases, which function in isolation or associated with compartmental peptidases to remove damaged or misfolded polypeptides, and turn over regulatory proteins that control numerous vital processes in all cells. Our collaborator's experiments imply that phosphate release after ATP hydrolysis drives ClpX conformational changes, and that the translocation step size is determined by the number of hydrolyzing ClpX subunits. We have characterized in atomic detail how phosphate release triggers the long-timescale conformational changes representing a basic power stroke for mechanical substrate processing.

Fig. 2. ClpX unfolds and translocates a protein by pulling its terminus through its central pore. Such a mechanical process is powered by ATP hydrolysis via conformational changes of the ClpX hexamer. (a) A schematic representation of ClpX unfolding a protein. (b) An intermediate state during polypeptide translocation by ClpX.

3. Characterize protein sequence-structure-function relationships

An interesting question in biophysics is how to bridge the information from protein sequence with their structure and function. Here we investigated an examplery DNA-processing enzyme UvrD helicase, which plays key roles in DNA replication and repair, by unwinding nucleic acid strands. Collaborating with Prof. Yann Chemla (University of Illinois), who used optical tweezer combined with FRET to study the single-molecule behavior of UvrD (Fig. 3a), we characterized the mechanism of which UvrD changes its conformation to alter its function from unwinding to rezipping dsDNA, by applying path sampling simulations[1*] (movie). Our simualtions were guided by bioinformatic surveys combined with principal component analysis (Fig. 3b), from which we obtained a key metastable state during the functional switch of UvrD (Fig. 3b).

Fig. 3. (a) Population distribution calculated using unwinding velocity and FRET signals (adapted from Fig. 3D of Comstock, ..., Chemla, Science 348: 352 (2015)). The map shows a strong correlation between UvrD activity and UvrD conformation. (b) Projection of crystal structures onto the first two modes from principal component analysis. The structures were obtained through a pdb survey. A key conformation is discovered and labeled as the "tilt" (open) state. The structure in the middle of the panel shows three states labeled in blue (closed state), green (tilt state) and orange (apo state), respectively.

4. Multiscale modeling

My endeavors also involved developing theoretical models to understand simulation or experimental data from my collaborators. In the first year of my PhD study, I formulated a statistical model to study how flagellin monomers are transported to grow the bacterial flagellum [7] (an hours-long process), based on the parameters derived from MD simulations.

Later on, employing the free energy perturbation method, I was able to reveal the impact of cytosine methylation on the recognition of DNA binding proteins, illustrating the importance of free energy landscapes in understanding epigenetic modifications [6].

Recently collaborating with my colleagues, we applied a coarse-grained model to characterize the structural ensembles of monomeric alpha-synuclein, in which a beta-hairpin was formed and expected to be crucial for alpha-synuclein aggregation [5]. This simulation result was later dramatically confirmed by an experimental study (Salveson, JACS, 2016, 138: 4458).

During my undergraduate study, I used computational models (Poisson-Nernst-Planck theory) to help experimentalists improve the energy conversion efficiency of their battery device powered by concentration gradient in an ion-selective nanofluidic channel [8,9].


Awards and Honors

  • Graduate College travel grant for Gordon Research Conference - Proteins, 2015
  • Travel award, Biophysical Society 2015 annual meeting
  • Award for outstanding undergraduate research, Peking University, Nov. 2009
  • Wusi fellowship, Peking University, 2007~2009

Talks and Lectures

Poster Presentations

  • "Force Generation Mechanism of Molecular Motors." Poster section for NSF site visit to CPLC, University of Illinois, April 2013.
  • "RNA Translocation Coupled to Large-Scale Conformational Transitions of a Hexameric Helicase." 59th Biophysical Society annual meeting, Baltimore, February 2015.
  • "Mechanism of Substrate Translocation by a Ring-Shaped ATPase Motor at Millisecond Resolution." Gordon Research Conference: Proteins, Holderness, June 2015.