Research Overview

Our research focuses on characterizing the molecular interactions that drive protein folding and association, and mediate compartmentalization in biological membranes. We develop and apply new optical tools based on ultrafast two-dimensional infrared (2D IR) spectroscopy to directly visualize protein-lipid and lipid-lipid interactions in heterogeneous multi-component membranes. Our current research builds upon our expertise in ultrafast optics, non-linear spectroscopy and microscopy, molecular dynamics simulations, semiclassical models of vibrational spectra, and protein biophysics.

Hydrogen-bond dynamics in complex environments

Intracellular environments are highly crowded and heterogeneous containing proteins, lipids, and osmolytes, which are small organic molecules. In fact, as much as 30% of intracellular water molecules are part of the inner solvation shells of these molecules and exhibit slower hydrogen-bond dynamics. While the effects of crowding are not well understood, it is known that excluded-volumes can provide a certain degree of entropic stabilization of native protein structures. Ultrafast spectroscopy has provided an atomistic description of hydrogen-bond dynamics in bulk water, however probing the heterogeneous environments remains challenging. Our group is applying the 2D IR “toolkit” towards understanding ultrafast hydrogen-bond dynamics at the protein-water interface in crowded environments.

In a similar vein, we are investigating the effect of small molecules known as “cryoprotectants” which disrupt the H-bond donor/acceptor balance and can prevent intracellular ice formation. Such compounds are routinely added to cell cultures prior to low-temperature storage. The thermodynamic effects of cryoprotectants on the proteins and membranes are not well understood. Our group is investigating the specific mechanisms by which DMSO, a common cryoprotectant, disrupts the water-water interactions, and leads to protein denaturation.

Heterogeneous Surfactant Interfaces

Surfactants are used daily in an extremely diverse array of applications, ranging from food and cosmetics to oil recovery and organic synthesis. However, current models of these molecules are extremely reductive and little is known about how interactions between surfactants and their solution environment gives rise to well-characterized bulk behaviors. This forces researchers to use trial-and-error methods to find the compound best-suited for a given application. Furthermore, highly heterogeneous blends of detergents are commonly used in industrial applications. We combine advanced spectroscopic techniques, molecular dynamics simulations, and synthetic labeling protocols to observe the surfactant environment as a function of solution composition and detergent heterogeneity to begin to understand the connection between molecular interactions and useful properties.

Peptide folding and membrane translocation

MembraneInsertionThree-dimensional protein structures are the end result of a complex, multi-step folding process. Spontaneous folding depends on a subtle enthalpic and entropic balance between contact formation, desolvation, and electrostatics. Lipid membranes add an extra layer of complexity as the environment changes significantly between the hydrophilic, solvent-exposed environment of the disordered peptide and the hydrophobic, membrane-embedded folded state. The figure shows possible pathways for the coupled folding-translocation of a small peptide into a lipid membrane. Our group uses a combination of pH-jump nonequilibrium two-dimensional infrared spectroscopy, isotope labeling of the peptide backbone, and molecular dynamics simulations to map protein-membrane interactions and track the folding pathways from the initial disordered state to the final folded structure.

Biophysics of lipid membranes

Biological membranes are highly heterogeneous, containing thousands of lipid species and crowded by hundreds of different proteins. However, to date most biophysical studies are carried out on model bilayers containing a single lipid species. The interplay between interfacial environments, lipid-lipid interactions, local water structure and dynamics is not understood, particularly, for multi-component membranes. Our group uses 2D IR spectroscopy, isotope labels on the lipid ester carbonyls, and molecular dynamics simulations (in collaboration with Prof. Ron Elber) to investigate the local water dynamics (H-bond lifetimes) in heterogeneous membranes. Our experiments are aimed at measuring how membrane composition affects interfacial water environments. Initial studies on this project have shown that ions can significantly slow down the interfacial water dynamics in the presence of negatively-charged lipids. We have also shown that Calcium ions tend to induce clustering in phosphatidylserine lipids.

Biological Ion-binding

Calmodulin (CaM) is a calcium-binding protein that modulates the behavior of a wide range of proteins in response to local calcium concentration. The CaM structure is composed of two lobes, each containing a pair of EF-hand ion-binding motifs, despite the extensive studies, it is not known how ion binding at the different sites leads to conformational changes in CaM, and how those changes are translated into a biological response. In collaboration with the group of Prof. Richard Aldrich (UT Neuroscience), we are investigating the ultrafast conformational dynamics of the calcium binding site in CaM using ultrafast 2D IR spectroscopy in the carboxylate asymmetric stretching region. Specifically, we measure the effect of ion size and charge on the conformation of the negatively-charged carboxylate groups. Recently, using a series of lanthanide ions, we showed that 3+ ions can significantly disorder the binding sites, and distort the configuration of highly-conserved aspartate residues. Future studies will investigate binding mechanisms of CaM to ion channels and its effect on the conductance states of the channel.

Ultrafast 2D IR spectroscopy

Two-dimensional infrared spectroscopy is an ultrafast optical technique that reveals molecular structure and fast dynamics by using a sequence of femtosecond (10-15 s)  laser pulses to measure frequency correlations between (and within) vibrational modes in a sample. In brief, a pair of pulses excites specific vibrational modes in the sample, then, following a short waiting time, a third laser pulse measures the response of the sample. Analogous to 2D NMR, a 2D IR spectrum is, in essence, a two-dimensional excitation-detection frequency correlation map. Cross peaks are observed when two vibrational modes are able to exchange vibrational energy, for instance, two normal modes involving the same atoms, or molecules able to chemically interconvert within the timescales measurement, such as two rapidly-interconverting molecular conformations. The degree of diagonal elongation (frequency correlation) is a direct measure of the frequency fluctuations of a vibrational mode, which in turn reports on the environment around a molecule. In general, the diagonal and off-diagonal lineshapes and peak intensities provide a detailed, bond-specific, view of molecular structure and dynamics.


Protein structure is reflected on the backbone C=O stretching vibrations, known as amide I modes. The periodic arrangement of residues along a protein backbone, gives rise to characteristic vibrational modes which appear in specific regions of the IR spectrum, but the broad, often featureless, peaks that arise from structural disorder and solvent exposure complicate the interpretation of traditional FTIR spectra. Two-dimensional infrared spectroscopy spreads the spectral information onto two frequency axes, thus providing an additional degree of structural characterization of the protein ensemble. In addition, the excitation-waiting-detection sequence of interactions serves to map the fast hydrogen-bonding dynamics of the backbone.

Computational modeling of IR spectra

Vibrational frequencies are particularly sensitive to the environment of a molecule, but lineshapes are often broad and difficult to interpret. Computational models are useful in helping us interpret the measured spectra (frequencies, linewidths, timescales). We develop models that produce accurate results at little computational expense. Particularly, semiclassical models take the advantage of the sensitivity of vibrational frequencies to the electrostatic environment. Our group is currently extending these models to understand how the spectra of lipids report on the overall structure and hydrogen-bonding environment, specifically the degree of hydration inside lipid bilayers.

To learn more about protein 2D IR spectroscopy please refer to:

C. Baiz, M. Reppert and A. Tokmakoff, An Introduction to Protein 2D IR Spectroscopy, in Ultrafast Infrared Vibrational Spectroscopy, ed. by Michael D. Fayer, pp. 361-404 [PDF]