Shi-Jie Chen Research Group

 HOME 

 RESEARCH 

 PUBLICATIONS 

 LAB MEMBERS 

 SOFTWARE 

 VFOLD_SERVER 

 TBI_SERVER  

Research Interests:

1. RNA Folding: structure & energy landscape -- "Vfold" and other models

RNAs fold into specific tertiary structures and undergo structural transitions for their biological function. Despite the importance of RNA structures, current theories have only limited abilities to deduce RNA tertiary structure from sequence, to compute the stability of a tertiary structure, or to discern alternative RNA conformations. We are developing ab initio atomistic models to predict RNA tertiary structural folding and structural transitions.

Central to the predictions of RNA folding and conformational changes is the free energy landscape. Accurate computation of RNA free energy landscape from nucleotide sequence remains an unsolved problem in RNA science and human health. The difficulty arises from two related problems: the complex conformational statistics of tertiary folds and lack of knowledge about the detailed energetics of RNA tertiary interactions.

Previous Slide

Next Slide


RNA conformational space is particularly large because each RNA nucleotide has many (seven) torsional angles. An attractive option is to use a reduced representation for the atomic coordinates. We use the C4-P, P-C4 and C4-N1/N9 virtual bonds to reduce the original 7 torsions for a nucleotide to 3 effective bonds (the Vfold model). The Vfold model enables exact conformational enumeration and accurate entropy calculation for RNA folds. Applications of the model to secondary structures and pseudoknots have led to good theory-experiment agreements. In addition, the entropy parameters predicted from the Vfold model, combined with thermodynamic data, would allow us to extract tertiary interaction (such as base triple) energy parameters from experiments (see the image slide show).

Furthermore, the Vfold model can provide a scaffold for 3D tertiary structure. Therefore, the Vfold predictions, in combination with molecular dynamics simulations, can give the all-atom free energy landscape and 3D structures. In the image slide show, we present the predicted structure (purpleblue) for the Beet Western Yellow Virus (BWYV) pseudoknot. By comparing with the experimental structure (color sand), the Root Mean Square Deviation (RMSD) over all heavy atoms is 2.7 Å.

2. RNA Folding: ion electrostatics -- the "TBI" model

Because RNAs are highly charged polyanions, RNA folding involves build-up of negative charges, which induces significant ion binding to RNA. Therefore, RNA structure and stability are strongly coupled to the ion effects in the solution. Many modeling studies on the ion effects in RNA folding are based on the Poisson-Boltzmann (PB) approaches. These approaches consider neither correlations nor fluctuations or finite size of the ions. However, a variety of experiments have pointed to the potential importance of ion correlation, fluctuations and the finite size effects for multivalent ions such as Mg2+ ions. Mg2+ ions are essential for RNA tertiary structure folding. The inability to treat the correlation/fluctuation and ion size effects has greatly limited our ability to understand, predict, and design RNAs and nucleic acids-based therapeutic molecules. Therefore, a new model is urgently needed. We are developing such a model - the Tightly Bound Ion (TBI) model.

The key ideas of the TBI model: Ion correlation effects highlight the importance to consider the different distributions of the ions instead of an average ion distribution. This requires the enumeration of many-ion distribution and the calculation of the energy for each distribution. The Boltzmann average over the different ion distributions gives the partition function and free energy of the system. A complete exact treatment for ion correlations and fluctuations for a realistic system is highly challenging due to the large number of possible ion distributions. What is well achievable, however, is to separate out the strongly correlated ions so that we can focus our computational resources on these relatively small number of ions while employing approximations such as PB to treat the vast number of the weakly correlated ions. The theory then parlays the two parts into the full partition function/free energy of the system.

Applications of the TBI model to a broad range of nucleic acids related problems indicate that the model has the potential to provide answers to many fundamental questions from the quantification of the electrostatic stabilizing forces in RNA folding, prediction of RNA stability and ion-induced conformational changes (such as three-way junctions) to quantitative understanding of the mechanisms of ribozymes, riboswitches, microRNAs and therapeutic RNA aptamers. Furthermore, the model may lead to useful tools that can be significant for rational design and analysis of nucleic acid-based biosensors and nanomachines technologies.

3. RNA Folding: kinetics -- "Kinetic Cluster" and "Seed Growth" models

RNA function is often kinetically controlled. Depending on the nucleotide sequence and the solution condition (ion concentration, temperature), RNA molecules show remarkably complex kinetic behavior. Inspired by the extensive experimental findings on RNA folding kinetics, we develop a folding kinetics theory to uncover the kinetic mechanism for large RNAs at the tertiary structural level. The project is designed to answer a set of key questions for RNA hairpin folding kinetics: How to identify the transition states for RNA folding kinetics? How to treat the astronomically large conformational ensemble of the molecule in the folding kinetics? How to relate the rugged RNA folding energy landscapes through analytical calculations to the rates, pathways, and kinetic traps in RNA folding.

One of the key issues is the reduction of the conformational ensemble using the kinetically essential states. We are developing several effective conformational reduction methods, including the "kinetic cluster method", where a reduced conformational ensemble is constructed from the preequilibrated conformational clusters, and the "seed growth method", where a reduced ensemble is constructed from the conformations that are kinetically connected to the low free energy states. These methods have led to many predictions that are consistent with experimental findings.

In collaboration with Professor Alan Van Orden's lab at the Colorado State University, we are developing a biophysical theory for intramolecular base-pairing and base-stacking in RNA secondary structure formation. These steps are the most fundamental rate processes in RNA folding. Van Orden's lab uses advanced fluorescence fluctuation spectroscopy (FFS) techniques to measure the temperature dependence for the kinetic rates for the designed sequences. Through theory-experiment comparisons, we aim to establish a valid rate model for RNA folding.