Using Cooperatively Folded Peptides to Measure Interaction Energies and Conformational Propensities

Conspectus

The rates and equilibria of the folding of biopolymers are determined by the conformational preferences of the subunits that make up the sequence of the biopolymer and by the interactions that are formed in the folded state. Because of the centrality of these processes to life, quantifying conformational propensities and interaction strengths is vitally important to understanding biology. In this Account, we describe our use of peptide model systems that fold cooperatively, yet are small enough to be chemically synthesized to measure such quantities. The necessary measurements are made by perturbing an interaction or conformation of interest by mutation and measuring the difference between the folding free energies of the wild type (in which the interaction or conformation is undisturbed) and the mutant model peptides (in which the interaction has been eliminated or the conformational propensities modified). With the proper controls, and provided that the peptide model system in question folds via a two-state process, these folding free energy differences can be accurate measures of interaction strengths or conformational propensities. This method has the advantage of having high sensitivity and high dynamic range because the energies of interest are coupled to folding free energies, which can be measured with precisions on the order of a few tenths of a kilocalorie by well-established biophysical methods, like chaotrope or thermal denaturation studies monitored by fluorescence or circular dichroism. In addition, because the model peptides can be chemically synthesized, the full arsenal of natural and unnatural amino acids can be used to tune perturbations to be as drastic or subtle as desired. This feature is particularly noteworthy because it enables the use of analytical tools developed for physical organic chemistry, especially linear free energy relationships, to decompose interaction energies into their component parts to obtain a deeper understanding of the forces that drive interactions in biopolymers. We have used this approach, primarily with the WW domain derived from the human Pin1 protein as our model system, to assess hydrogen bond strengths (especially those formed by backbone amides); the dependence of hydrogen bond strengths on the environment in which they form; β-turn propensities of both natural sequences and small molecule β-turn mimics; and the energetics of carbohydrate–protein interactions. In each case, the combination of synthetic accessibility, the ease of measuring folding energies, and the robustness of the structure of the Pin1 WW domain to mutation enabled us to obtain incisive measurements of quantities that have been challenging to measure by other methods.

Graphical abstract

1. Introduction

Life requires the constant making and breaking of non-covalent interactions within proteins or between proteins and other macromolecules during folding/unfolding and/or association/dissociation processes. The rates and equilibria of these processes hinge on the conformational propensities of the molecules involved and the strengths and multiplicity of the interactions that form in the folded or bound states. These are defined by the properties of the molecular entities involved and the medium in which the process takes place. The centrality of these processes to biology has led to an intense and long-standing interest in measuring their energetics. These measurements lie at the intersection of biophysics and physical organic chemistry and methodology from both fields is required to make accurate measurements.

Perhaps the conceptually most obvious way to measure interaction strengths between functional groups in biomolecules is to use simple model small molecules that interact intermolecularly in solution. Thus, our initial understanding of the contribution of the hydrophobic effect to protein folding originated from solubility studies of non-polar solutes in water and the thermodynamics of transfer of models of amino acid side chains between non-polar and aqueous media.^1–5 Similarly, the first estimates of the contribution of hydrogen bonds (H-bonds) to protein stability came from Schellman’s studies of aqueous urea solutions,⁶ which were later expanded and refined by Klotz and others.^7,8

The difficulties associated with using small molecule models in solution to measure interaction strengths are that (1) detecting the product complexes can be difficult because small molecule association constants are usually small (Schellman calculated an association constant of 4.1×10⁻² M⁻¹ for the urea dimer), requiring high concentrations of solutes to be used, which can in turn affect the properties of the solvent; and (2) the geometries of the interactions are unconstrained, leading to ambiguity in the assignment of a measured energy to a particular interaction mode.

The conformational propensities of amino acids have been studied similarly in small model peptides.^9–11 However, such peptides do not fold cooperatively. Instead, they exist as an ensemble of conformations with varying degrees of structure, which complicates the relationship between observables (e.g., the circular dichroism signal at 222 nm, which is proportional to helicity¹²) and conformational propensities.¹³

One way in which this issue has been addressed is through the use of “molecular balances”, small molecules that can exist in two (or more) states, one of which enforces an interaction of interest that is absent in the other conformation(s). The perturbation of the equilibrium by the interaction, typically measured by NMR, reflects the interaction’s strength.^14–18 Molecular balances have also been used to measure helical propensities of amino acids via template-induced helix initiation.¹⁹ They can sensitively measure weak interactions and, because they are synthetic, the interacting functional groups can be varied to probe the nuances of the interaction in question. However, strong interactions can be difficult to measure with molecular balances because the population of the non-interacting state can be difficult to detect and quantify. Moreover, the microenvironmental properties, especially the effective concentration, of functional groups in a small molecule in solution are likely to be different than they are in a folded protein.

Another method to quantify intramolecular interactions and conformational propensities that does not suffer from these drawbacks is to perturb interactions that exist in cooperatively folded biomolecules, usually proteins, and use the resulting change in the folding free energy (ΔG_f) as a measure of the strength of the interaction or the change in conformational propensity.^20,21 For example, a serine-to-alanine mutation in a given protein will eliminate any H-bonds formed by the serine side chain hydroxyl. Then the difference between the folding free energies (ΔΔG_f) of the serine-containing wild type protein and the alanine mutant is partly a measure of the strength of the H-bonds formed by the serine side chain. However, mutations can alter the local structure, leave unsatisfied H-bonding partners, introduce cavities into the protein core, etc. Thus, factors other than H-bond strength can contribute to ΔΔG_f in the serine-to-alanine example. Fersht and co-workers have analyzed mutational effects on folding free energies²² and have shown that “double mutant cycles” can be an effective method to extract interaction energies from ΔΔG_f measurements.^22–24 In a double mutant cycle, two interacting residues (denoted X and Y in Figure 1) are mutated individually and as a pair. The differences in folding free energies between each mutant and the wild type are then calculated, yielding three ΔΔG_f values: one for each of the single mutants (ΔΔG_f,X and ΔΔG_f,Y) and one for the double mutant (ΔΔG_f,XY). The contribution of the interaction of interest to the folding free energy can then generally be isolated by subtracting the sum of ΔΔG_f,X and ΔΔG_f,Y from ΔΔG_f,XY (Figure 1)²².

Figure 1.

Schematic of the double-mutant cycle for assessing interaction strengths (red dashed line) between side chains X and Y, where these side chains are mutated to residues that cause minimal structural perturbation. The black dashed lines represent the interaction of X and Y with the rest of the protein.

Protein folding free energies can be measured over a wide range of protein stabilities by thermal or chemical denaturation with precisions of tenths of kilocalories.²⁵ Thus, interactions with a wide range of strengths can be characterized by analyzing the effects of mutations on folding energetics. However, with the exception of some elegant work using unnatural amino acid incorporation,^26,27 the main drawback of mutational methods is that the ways in which interactions in proteins can be perturbed is limited by the side chains of the naturally occurring amino acids.

A “middle ground” between small-molecule-based molecular balances and biopolymer-based mutational methods for measuring interaction strengths and conformational propensities is to use mutational methods in peptides that fold (or self-assemble²⁸) cooperatively to a well-defined structure, yet are small enough to be chemically synthesized. This approach enables the measurement of quantities of interest via folding free energy perturbations while retaining the flexibility to introduce such perturbations without being limited by ribosomal protein biosynthesis. In this Account, we describe our use of this approach to probe conformational propensities and interaction strengths.

2. Cooperatively Folded Peptide Model Systems

A peptide must possess several features to be suitable for studying intramolecular interactions and conformational propensities. Foremost, it and its mutants should fold and unfold cooperatively so that the folding free energy can be measured accurately. The unfolding transition of cooperatively folded peptides, analogous to a first order phase transition, happens in a narrow window of a thermodynamic variable (e.g., denaturant concentration or temperature). Two-state cooperative unfolding, which should be experimentally verified by, for example, ensuring that denaturation curves measured with different techniques (e.g., circular dichroism and fluorescence) are coincident, rules out any potential contributions from intermediates to the folding free energy, allowing accurate quantification. The model peptide should also have a well-defined structure to facilitate a structure-based interpretation of the energetic contributions from individual residues. Furthermore, the peptide should be short enough to permit an easy chemical synthesis of the desired variants. However, since the cooperativity of folding is directly proportional to polypeptide length²⁹, the model peptide cannot be too short.

The semi-conserved WW domains, present in many proteins involved in regulatory or signaling processes, satisfy all these requirements. WW domains are 30 to 65 amino acids long and comprise three anti-parallel β-strands. The most extensively characterized WW domain is that from the human peptidyl-prolyl cis/trans-isomerase Pin1 (Pin WW), for which extensive kinetic, thermodynamic and structural data are available³⁰. Pin WW is an ideal mini-protein for structure-folding studies because it is an independently folding domain with adequate thermodynamic stability and fast folding kinetics.^31–33 We have used the 34-residue Pin WW domain and, to a lesser extent, other small protein domains to study the energetics of backbone H-bonding, kinetics of beta-sheet folding, turn formation propensities, and the thermodynamics of carbohydrate-aromatic interactions in glycosylated proteins.

3. Energetics of protein main chain H-bonding by backbone mutagenesis

H-bonds, especially those formed between backbone amides, are among the most prevalent non-covalent interactions in proteins. The energetics of backbone H-bonding, however, has been difficult to study. This is mainly because traditional site-directed mutagenesis cannot be used for this purpose. Pin WW, due to its amenability to chemical synthesis, provided us with a platform to experimentally scrutinize the nature of backbone amide H-bonds.

Among the simplest backbone alterations that can be chemically introduced into peptide structure are amide-to-ester and amide-to-olefin mutations. We have used perturbations caused by these mutations to assess the context-dependent contribution of backbone H-bonding to β-sheet and α-helix folding.^34,35 The effect of an amide-to-ester mutation on protein folding is often deduced from the ΔΔG_f values between the wild-type protein and amide-to-ester mutants. However, there are issues involved with the extraction of H-bond energies from amide-to-ester mutations that must be corrected in the thermodynamic analysis.

Amide-to-ester mutations eliminate amide NH groups, which can leave amide carbonyls without H-bonding partners. This is especially thermodynamically unfavorable if the unsatisfied amide carbonyl is buried. Therefore, a correction inversely proportional to the accessible surface area of the amide should be implemented. Amide-to-ester mutations also can introduce Coulombic repulsion between the ester O_e (which replaces the amide NH) and neighboring amide carbonyl oxygens. Although this repulsion does not usually cause significant structural changes, energetic correction for this effect should be included in the thermodynamic analysis of amide-to-ester mutations. These corrections, however, usually are not large enough to change the qualitative interpretation of the data. For example, by comparing the ΔΔG_f of the amide-to-ester and the amide-to-olefin mutants in a β-sheet context, taking into account the transfer free energy differences, we estimated that the O–O repulsion term is only +0.3 kcal/mole.³⁶

We used double-mutant cycle analysis, employing a combination of amide-to-ester backbone and traditional side chain mutagenesis, to further understand the effect of microenvironment polarity on backbone H-bond strength. The destabilization energies of amide-to-ester mutations (which reflect backbone H-bonds strengths) in Pin WW ranges from 1.4 to 4.9 kcal/mole (Figure 2a), which is in agreement with results in other proteins (Figure 2b) and comparable to the strength of sidechain H-bonds.³⁷ The protein is most destabilized by perturbation of backbone H-bonds that are surrounded by hydrophobic sidechains.³⁸ Further studies confirmed that H-bonds could be more favorable by up to −1.2 kcal/mole when they are buried in a hydrophobic environment compared to when they are solvent exposed.³⁹ Because N-H···O H-bonds are largely driven by electrostatic and polarization effects, which are sensitive to the local electric field and the dielectric constant, the context-dependence of H-bonds in proteins is not unexpected.⁴⁰

Figure 2.

The strength of backbone H-bonding is context dependent in Pin WW. a) The color-coding of the perturbed H-bonds depicts the experimentally quantified destabilization effect of amide-to-ester mutations. The adjusted ΔΔG_f from ref.³⁴ is used. Large, positive ΔΔG_f values indicate strong H-bonds. b) Histogram of ΔΔG_f/b (ΔΔG of folding or binding) values from mutants in which backbone-backbone H-bonds were perturbed by amide-to-ester mutations.³⁷

We also measured the folding kinetics of amide-to-ester mutants to examine the formation of secondary structure during the folding of the Pin WW domain. We found that the H-bonds of loop 1 amides are almost fully formed in the folding transition state and that this loop assumes a native-like secondary structure in the transition state.³⁸ These results also showed that the β-strands and loop 2 are only partly structured in the folding transition state. It should be noted that, since the backbone H-bonds in loop 1 are among the weakest H-bonds in Pin WW in the ground state or in the transition state, factors other than H-bond strength—such as conformational propensity and side chain interactions—must drive structure formation in the transition state.

4. Conformational propensities of natural and unnatural β-sheet folding nucleators

To study the conformational propensities of β-turn-containing loops, loop 1 of Pin WW can be replaced with a variety of naturally occurring and rationally designed turns. Side-chain and backbone mutagenesis of Pin WW in combination with laser temperature-jump relaxation experiments determined that the rate-determining step in the folding of Pin WW is the nucleation of loop 1, a six-residue loop harboring an internal type 2 β-turn. Many sequences have much higher turn propensities, potentially accelerating Pin WW folding. We replaced the wild-type loop 1 with shorter sequences that either have a high propensity to fold into a type 1′ β-turn conformation or a type 1 β-turn with a G1 β-bulge (hereafter referred to as a type 1 β-bulge turn), the latter being common in other WW domains.⁴¹ These substitutions accelerate Pin WW folding by almost an order of magnitude and increase its thermodynamic stability. However, structural, thermodynamic, kinetic, and ligand-binding studies on loop 1 Pin WW indicate that folding rate is not the only selection pressure operating on WW domains. Functional factors have contributed to the evolution of this region of the protein as well.⁴¹

Conformational propensities of unnatural β-turn mimics can also be evaluated by placing them in loop 1 of Pin WW. We have shown that loop 1 can be replaced with a number of small molecule β-turn mimics (e.g., Figure 3a).⁴² Some of these β-turn mimics maintain the thermodynamic stability but modulate the folding rate of WW domain.⁴² The solution structure of the variant with an E-olefin-based β-turn mimic revealed that this turn mimic’s ability to strongly nucleate Pin WW folding is likely tied to the fact that it matches the right-handed twist of the Pin WW β-sheet (Figure 3b).⁴³

Figure 3.

A β-turn mimic that strongly nucleates folding of Pin WW loop 1. a) Structure of the dipeptide β-turn mimic. b) Comparison of the structure of Pin WW with a type 1′ β-turn (grey) or the olefin β-turn mimic in loop1 (pink). Only side chains in loop 1 are shown for clarity.

5. Parsing the contributions to carbohydrate-aromatic interactions: CH–π interactions and the hydrophobic effect in enhanced aromatic sequons

N-Glycosylation, glycosylation on the amide side chain of asparagine, can enhance both the thermodynamic and kinetic stability of glycoproteins. It has been speculated that N-glycans diminish the local conformational fluctuations in the unfolded state of proteins to which they are attached through their bulk and the excluded volume effect. If so, then this mechanism of stabilization would have minimal dependence on the site of glycosylation. Using Pin WW to explore the energetic effect of N-glycosylation on protein stability, we found instead that the effect of N-glycosylation was context dependent and frequently was destabilizing. Incorporation of N-linked N-acetylglucosamine (GlcNAc) at various positions of Pin WW destabilized the native state unless the glycan was installed in certain positions of loop 1 and 2.⁴⁴

N-glycans in Phe_i−2-Xxx_i−1-Asn(glycan)_i-Gly_i+1-Thr_i+2 sequences that are in a type 1 β-bulge turn conformation can reliably stabilize proteins.^45,46 The stabilizing effect of this structural module, which we termed the enhanced aromatic sequon (EAS), is largely due to the interaction of the first GlcNAc on Asn_i with the side chain of Phe 46_i−2 (Figure 4). We hypothesized that other reverse turns, such as those in Pin WW loop 1, could host such stabilizing interactions by positioning Phe and Asn(GlcNAc) differently (cf. Figure 4a and 4b).

Figure 4.

Structures of enhanced aromatic sequons in loop 1 of Pin WW. a) The N-linked GlcNAc packs tightly against Phe_i−2 and Thr_i+2 in a five-residue enhanced aromatic sequon. b) The N-linked GlcNAc packs similarly against Phe_i−3 and Thr_i+2 in a six-residue enhanced aromatic sequon.

We observed significant, yet different, levels of native state stabilization when loop 1 (residues 16–21) of Pin WW was converted to EASs with five- (Phe-Ala-Asn-Gly-Thr) or six- (Phe-Arg-Ser-Asn-Gly-Thr) residue loops comprising Asn N-glycosylated with GlcNAc.^46–48 In these EASs, thermodynamic cycle analysis indicated that most of the glycosylation-associated stabilization arises from interactions between the Phe aromatic ring and the GlcNAc.⁴⁷ (A four-residue loop based on a type 1′ β-turn (Phe-Asn-Gly-Thr) was also stabilized by N-glycosylation, but the effect did not appear to be mediated by direct interaction between GlcNAc and Phe.⁴⁷) The solution NMR structures of the glycosylated variants confirmed that the EAS facilitates face-to-face monosaccharide-aromatic interactions (Figure 4).⁴⁸ Taken together, these factors suggested that enhanced aromatic sequons within Pin WW could be used to experimentally parse the energetics of intramolecular carbohydrate-aromatic interactions in aqueous solution.

Since Pin WW is amenable to chemical synthesis, a wide variety of unnatural amino acids can be incorporated at the aromatic site to perturb the carbohydrate-aromatic interaction. Mutation of the Phe in the five- and six-residue EASs to a series of natural and unnatural amino acids, including non-aromatic amino acids as well as electron-rich and -poor analogs of Phe, modulated the extent to which N-glycosylation of the EAS stabilized Pin WW. This stabilization was quantified by ΔΔG_glyc = ΔGfold,glyc − ΔGfold,nonglyc, where ΔGfold,glyc and ΔGfold,nonglyc are the folding free energies of the N-glycosylated and non-glycosylated forms of a Pin WW variant. In a simplistic model, ΔΔG_glyc can be decomposed into three energy components. The first component, ΔΔG_int, describes the intrinsic stabilizing effect of glycosylating the Asn in the EAS. This energy component accounts for the conformational preference of glycosylated versus the non-glycosylated Asn and the interaction of glycan with parts of the protein outside the EAS. The second energy component, ΔΔG_phob, describes the stabilization arising from the hydrophobic burial of the α-face of GlcNAc by the side chain in the i−2 position of the five-residue EAS or the i−3 posiUon of the six-residue EAS (hereafter, we will refer to this as the “interactor position”).⁴⁹ The transfer free energy from non-polar media to water, ΔG_tr, has been found to correlate with hydrophobic burial energies like ΔΔG_phob.^50,51 The correlation between ΔΔG_glyc and ΔG_tr of residues at the interactor position explains about 30 percent of total variation of ΔΔG_glyc in the five-residue EAS series (Figure 5a), but the correlation is much stronger between ΔΔG_glyc and ΔG_tr for non-aromatic residues than for aromatic residues (Figure 5a, dashed lines). Thus, hydrophobic burial is more determinative of ΔΔG_glyc when there are non-aromatic amino acids in the interactor position. In other words, while the hydrophobic effect can be considered as a major contributor to the interaction between GlcNAc and non-aromatic residues in the interactor position, factors other than the hydrophobic effect must be contemplated in order to explain the stronger interaction between GlcNAc and aromatic side chains.

Figure 5.

a) ΔΔG_glyc representing the interaction of GlcNAc with the residue in the interactor position of the five-residue EAS in Pin WW domain correlates weakly (solid line, R²=0.32) with the hydrophobicity (ΔG_tr) of side chains in the interactor position. Hydrophobicity of non-aromatic residues (grey circles) shows a stronger correlation with ΔΔG_glyc (dashed grey line) than that of aromatic residues (red circles and red dashed line). b) ΔΔG_glyc exhibits a strong correlation with the polarizabilities of side chains in the interactor position. Polarizabilities were calculated using the DFT method (B3LYP) as implemented in Gaussian09.

The third contribution to ΔΔG_glyc comes from CH–π interactions. The importance of this interaction is suggested by the solution structures of Pin WW with the five- and six-residue EASs, which show that H5 of GlcNAc and its neighboring axial hydrogens are positioned toward the polarizable π-electron cloud of the aromatic ring in the interactor position (Figure 4).⁴⁸ In addition, CH–π interactions are thought to have substantial electrostatic components.⁵² However, we observed a negligible correlation between Hammett’s σ_m and σ_p values of the side chains and ΔΔGglyc. This result suggests that energy components arising from the partial charges or permanent dipoles (Coulombic, Keesom and Debye interactions) at the interactor position do not contribute to the net interaction energy in this system. We note that this finding does not imply that CH–π interactions do not have electrostatic components. The “net interaction energy” is the difference between the energies of the state in which the GlcNAc and the aromatic in the interactor position are solvated and the state in which they are interacting with each other. Therefore, our result simply suggests that the electrostatic components of the CH– π interaction cancel the electrostatic components of the OH–π interaction when the aromatic in the interactor position is solvated.

Another major contribution to CH–π interactions comes from London dispersion interactions, which are directly proportional to the polarizability of the interacting entities.^53–55 Consistent with other computational and experimental observations,⁵⁶ ΔΔG_glyc correlates strongly (R² = 0.72) with the polarizability of sidechains at the interactor position (Figure 5b). Thus, most of the variation of ΔΔG_glyc in this system is explained by the polarizability of the sidechains at the interactor position. This observation is consistent with London dispersion interactions driving the CH–π interactions in this system rather than interactions due to static charge distributions, possibly because the α-face of GlcNAc is not very strongly electropositive.⁵⁷

Varying the sugar attached to the Asn in the EAS while keeping the aromatic residue in the interactor position constant as Phe enabled us to evaluate how the glycan structure influences the carbohydrate-aromatic interaction energy in Pin WW glycovariants. ⁵⁸ Our thermodynamic and corresponding structural data show that the relative strengths of carbohydrate-aromatic interactions depend on the stereochemistry and identity of the substituents on the sugars’ pyranose rings. The interaction energy of the monosaccharides with the phenyl ring in the interactor positions decreases in the following order: allose > mannose > GlcNAc, xylose > glucose > N-acetylgalactosamine > galactose > L-idose. Peracetylation of the monosaccharides uniformly increases the strength of these interactions, as has been observed previously,⁵⁹ but changes the order only slightly.⁵⁸

The solution NMR structures of the Pin WW glycovariants show that their structures are not significantly altered by changes in the carbohydrate ring. However, the exocyclic hydroxymethyl group protruding from C5 of the pyranose rings of the hexoses populates both gauche⁻ and gauche⁺ rotamers. In the NMR structures of the glycovariants, only one of the two diastereotopic hydrogens attached to C6 is pointed toward the Phe aromatic ring in the gauche⁺ rotamer while almost burying the OH group. In contrast, both of the C6 hydrogens are turned toward the aromatic ring in the gauche⁻ rotamer. The more stabilized glycovariants (e.g., allose) have a higher propensity to populate the gauche⁻ conformation than the less stabilized glycovariants (e.g., galactose). The very small stabilizing effect of L-idose is explained by the unusually poor packing of L-idose against the Phe aromatic ring.⁵⁸

6. Conclusions

Peptides can be used as macromolecular hosts for measuring interaction energies and conformational propensities if they (1) are cooperative two-state folders; (2) have structures that are robust to mutational perturbation; and (3) are easily synthesized. Interaction energies or conformational propensities can then be measured in such peptides by introducing mutations that perturb an interaction (e.g., a H-bond) or a conformational propensity (e.g., β-turn propensity) and measuring the difference in the folding free energies of the wild-type and the mutant peptides using a variety of easily accessed biophysical methods. With the proper controls, this difference can provide an incisive measure of the quantity of interest, showing the power and utility of cooperatively folded peptides. Moreover, this methodology will become even more powerful as more unnatural amino acids become commercially available and peptide synthesis strategies continue to improve.

Biographies

Maziar S. Ardejani graduated from the Sharif University of Technology with a B.Sc. in Chemistry. He received his M.Sc. and Ph.D. from Nanyang Technological University. He has received postdoctoral training at King’s College London and currenly is a research associate at The Scripps Research Institute. His research revolves around protein folding and design.

Evan T. Powers received his B.S. in chemistry from Cornell University (1992) and his Ph.D. in organic chemistry from the Massachusetts Institute of Technology (1999) under the mentorship of Daniel S. Kemp. He was then an NIH post-doctoral fellow at The Scripps Research Institute with Jeffery Kelly (1999–2000), where he is currently an Associate Professor of Chemistry. His research interests are in protein folding and protein homeostasis.

Jeffery W. Kelly received his B.S. in chemistry from the State University of New York College at Fredonia and his Ph.D. in organic chemistry from the University of North Carolina. After receiving postdoctoral training at the Rockefeller University under the mentorship of E. Thomas Kaiser, he joined the faculty of the Chemistry Department of Texas A&M University. In 1997, he moved to the Scripps Research Institute, serving as Dean of Graduate Studies (2000–2008), Vice President of Academic Affairs (2000–2006) and is currently Chair of the Department of Molecular Medicine. His research interests are in chemical biology, protein folding and cellular protein homeostasis.