My interest in biomolecular modeling and simulation has its origins in my graduate work at Duke University under the direction of Paul M. Gross and Marcus E. Hobbs, and in my year-long courses in quantum mechanics and statistical mechanics with Fritz London. Gross had previously spent a sabbatical leave with Peter Debye in Leipzig, and returned to Duke with an interest in the relation between molecular structure and dipole moments. Shortly before his arrival at Duke, London had formulated a quantum mechanical treatment of van der Waals forces, in which polarizability played an important role. In this atmosphere, I began graduate research using electrical birefringence (Kerr effect) to determine anisotropic polarizabilities of small organic molecules. This research was interrupted by the entry of the US into World War II, and my resulting participation in a war project at Duke.

One day I had a chance encounter in the chemistry library with a then new book by Edwin Cohn and John Edsall, titled Proteins, Amino Acids and Peptides, which contained chapters by several authors besides Cohn and Edsall, namely John Kirkwood, George Scatchard, and Larry Oncley. Edsall described flow birefringence and Oncley described dielectric dispersion of proteins. This appealed to me as a chance to take up the birefringence work that I had to drop at Duke and, as an ACS postdoctoral fellow at Harvard Medical School, I applied flow birefringence to proteins under Edsall's guidance in an atmosphere devoted to the physical chemistry of blood plasma proteins.

Then, at Cornell, I began experimental work on the mechanism of the action of thrombin on fibrinogen to produce the fibrin clot. In a limited proteolytic reaction, thrombin releases peptides from fibrinogen, exposing a polymerization site on the resulting fibrin monomer. I used flow birefringence to elucidate the nature of the staggered-overlapped rod-like polymers formed from fibrin monomer on the pathway to the blood clot.

At the same time, Pauling and Corey had proposed the α and β structures of proteins, focusing on the backbone hydrogen bonds. With my first graduate student, Michael Laskowski, I examined the role of side-chain hydrogen bonds in proteins. Specifically, we demonstrated how side-chain hydrogen bonds are involved in the polymerization of fibrin monomer,¹  and also influence the pK's of ionizable groups²  as well as limited proteolysis³  in which it is necessary to break hydrogen bonds (during the hydrolysis of a peptide bond) to liberate a fragment which had been connected to the rest of the molecule by such hydrogen bonds.

This led to an attempt to determine protein structure by acquisition of distance constraints by location of side-chain hydrogen bonds experimentally. Charles Tanford had used UV titration of ribonuclease A (RNase A) in the pH region near the pK° of tyrosine, viz. ∼10, to demonstrate that three of the six tyrosines had abnormally large pK's⁴  and, with Jan Hermans, we used potentiometric titration to demonstrate that three of the eleven COOH groups had abnormally low pK's.⁵  During my sabbatical leave with Kai Linderstrøm-Lang at the Carlsberg Laboratory in Copenhagen in 1956–57, with UV difference spectroscopy⁶  (see Figure 1.1), I showed that the UV absorption spectrum of tyrosine varied with pH at low pH where COOH groups ionize, suggesting the proximity of COOH group(s) near tyrosine(s). Back at Cornell, I started a long series of experiments with graduate students and postdocs which ultimately located three Tyr–Asp interactions,⁷ viz., Tyr25 with Asp14, Tyr92 with Asp38, and Tyr97 with Asp83. These were subsequently verified by the crystal structure of RNase A.⁸ 

Also, during my sabbatical leave at the Carlsberg Laboratory, Walter Kauzmann arrived in mid-year and, with Linderstrøm-Lang, we had many discussions about hydrophobic interactions. When Walter returned to Princeton, he wrote his famous article on hydrophobic interactions,⁹  and upon my return to Cornell, I started a new graduate student, George Némethy, on a statistical mechanical theory of hydrophobic interactions.¹⁰  Simultaneously, with Izchak Steinberg and George Némethy,¹¹  we discussed the interactions between hydrogen bonds and hydrophobic interactions, and pointed out how the nonpolar portions of so-called polar side chains can be involved in hydrophobic interactions with nearby nonpolar side chains, as shown in Figure 1.2, providing increased strength to the hydrogen bond, and its consequent influence on protein structure and stability.

With George Némethy, we decided to formulate a theoretical approach to determine protein structure^12,13  by making use of the distance constraints implied by the three Try–Asp interactions and the location of the four disulfide bonds in RNase A. This computational work evolved over the years, first by formulation of an all-atom force field, ECEPP,¹⁴  an Empirical Conformational Energy Program for Peptides, and subsequently, by development of UNRES,¹⁵  a united-residue coarse-grained force field.

In addition to the effort to determine the structure of RNase A, we also embarked on experimental work to determine its folding pathways by oxidation with two redox systems, GSSG/GSH¹⁶  and DTT^ox/DTT^red.¹⁷  Later, with UNRES, we expanded our computational approach to simulate folding pathways¹⁸  and folding kinetics.¹⁹ 

The ultimate goal of all our research, beginning with our work on the thrombin–fibrinogen interaction, was to use physical chemistry to elucidate biological structure and function. With our coarse-grained force field, we have recently embarked on studies of protein–protein interactions, e.g., Aβ,²⁰  PICK1²¹  and Hsp 70,²²  and have started to formulate a coarse-grained nucleic acid force field to be able to treat protein–nucleic acid interactions.

All of this work has been an excellent vehicle with which to train undergraduate, graduate, and postdoctoral students whose contributions to this research have been fundamental.

A more detailed description of some of the difficulties encountered and surmounted during the implementation of this research is provided in a prefatory chapter in Annual Reviews of Biophysics.²³