RESEARCH

Computer calculates polypeptide structures Cornell method verifies known polyglycine structure, for example, and eventually may be used to predict complex protein structure

HELIX. Dr. Harold A. Scheraga explains to chemists at the IUPAC symposium in Tokyo that the different orientations of the side-chain esters with respect to the peptide backbone (with the resulting difference in dipole-dipole interactions) accounts for the different helical forms for poly-/3-methyl-L-aspartate (left-handed) and poly-7-L-glutamate (right-handed)

Complete three-dimensional structures of a number of homopolyamino acids and a cyclic decapeptide have been calculated by a group of Cornell Uni­versity chemists using energy calcula­tions with computer methods. The computed structures jibe in all cases where there are experimental data.

So far, the method has been used to verify structures already known ex­perimentally. But the Cornell group hopes to perfect the procedure and eventually use it as a tool to predict complex structures of protein. Dr. Harold A. Scheraga, head of the Cornell group, revealed the work at the IUPAC International Symposium on Macromolecular Chemistry, held in Tokyo ( and later in Kyoto ).

The six polyamino acids used in the Cornell work belong to a series of homopolymers that have increasingly complex side chains. The acids are polyglycine, poly-L-alanine, poly-L-valine, poly-L-tyrosine, poly-/?-methyl-L-aspartate, and poly-y-methyl-L-glu-tamate.

The calculation, although complex, is reasonably straightforward. First, the internal energy is set up as a function of the coordinates of all the atoms of the molecule. This involves accurately establishing the compo­nents of the energy. Once the rela­tionship between all the energy com­ponents and spatial coordinates has been worked out, the problem is fed to a computer. The computer varies the coordinates on the energy surface, seeking the point of least energy and thus highest stability. Then the co­ordinates corresponding to this point are read out.

The work verifies that for the poly­amino acids studied (except polygly­cine), the alpha-helical form is the most stable. And for each polyamino acid, the calculations reveal the screw sense (direction of wind, left or right) of the backbone helix. Furthermore, the Cornell team, which also includes Dr. Tatsuo Ooi (now in Japan), Dr. Roy A. Scott, Dr. Garrett Vanderkooi, and Raquel F. Epand, has been able to separate the contributing factors that determine what form the structure takes.

Polyglycine, simplest of the group, shows no preference for left- or right-handed conformation. "In other

words, the alpha-helical forms are equally stable," Dr. Scheraga says. In contrast to the other polyamino acids, the calculations for polyglycine cannot be verified experimentally, since the calculations pertain to single chains and polyglycine exists as a multiple-stranded chain. However, we are now carrying out calculations for interacting or multistranded struc­tures."

But in poly-L-alanine, the methyl side chain destroys symmetry of the backbone, he explains. The right-and left-handed forms are no longer mirror images. The interaction ener­gies of the side-chain methyl groups with atoms of the backbone are such that the right-hand, alpha-helical form has the least internal energy and thus is the more stable form.

Poly-L-valine with its propyl side chain starts to give considerable steric hindrance. "Nevertheless," Dr. Sche­raga says, "we find the right-hand alpha helix again is more stable, ex­ceeding that of the left-hand alpha helix by 0.6 kcal. per residue." This conflicts with experimental results of others, he notes. But he points out that in other work, a solvent was used that itself disrupted the helix.

In preliminary experiments, how­ever, the Cornell group has been able to show that poly-L-valine can form

the alpha helix. Poly-L-valine is in­soluble in most solvents. So the chemists solubilized the compound by forming a water-soluble block co­polymer with soluble poly-DL-lysine, sandwiching the valine blocks be­tween lysine blocks.

They find that the poly-L-valine portion is randomly coiled in water at room temperature, but alpha-helical in 90% aqueous methanol at room temperature. As evidence, they cite the location (233 m^) and magnitude (m', the mean residue rotation is —12,-000 ± 500) of the trough of the Cot­ton effect and the value of the Moffitt-Yang parameter b0 , which equals - 6 0 0 ± 100, calculated with λ0 = 212 m/x.

The phenolic side chains of poly-L-tyrosine further complicate the pic­ture. "The backbone is in the form of a helix and the phenolic side chains are distributed about this backbone helix in another helix concentric with it. Since the phenolic groups outside the backbone helix absorb in the UV, they make difficult interpretation of optical rotation data normally used to determine the screw sense of the backbone helix. For this reason, there has been a lot of ambiguity about the screw sense of this polymer. How­ever, our calculations show pretty clearly that the most stable poly-L-

70 C&EN OCT. 17, 1966

Pure, dry styrene polymerizes reproducibly with gamma rays via an ionic mechanism

tyrosine helix is right-handed," Dr. Scheraga told the Tokyo symposium.

Although poly-/3-methyl-L-aspartate and poly-y-methyl-L-glutamate differ by only one methyl group in their side chains, they differ in screw sense. Previous experiments have shown that the aspartate polymer is left-handed while the glutamate is right-handed. The Cornell study now shows why.

"Our calculations show that one particular interaction accounts for this difference. If this one interaction were missing, both polymers would be right-handed," Dr. Scheraga says. In the aspartate polymer, he continues, the side-chain ester dipole is oriented with respect to the backbone dipole such that the left-handed alpha helix is favored. This dipole-dipole inter­action favors the left-handed form, overcoming the other energies which favor right-handedness. But in the glutamate polymer, the additional methyl group changes the distance and orientation of the side-chain ester di­pole with respect to the backbone di­pole. And this dipole-dipole inter­action reinforces the preference for right-handedness arising from the other energies.

"Furthermore," Dr. Scheraga said in Tokyo, "it was interesting to learn that Japanese work presented at this meeting (by Dr. Masahisa Hashimoto, Dr. S. Arakawa, and Dr. K. Nakamura of Dainippon Pharmaceutical Co.) shows experimental data that agree with our conclusions."

Taking their work further along toward proteins, Dr. Scheraga and his coworkers have also carried out energy calculations for gramicidin-S, a cyclic decapeptide [Biochemistry, 5, 2991 (1966)]. "We have arrived," says Dr. Scheraga, "at what we think may be a structure of lowest energy. And this seems to be consistent with pre­liminary x-ray data." He notes that Dr. Robert Hughes, also at Cornell, is now studying the crystal structure of the compound to confirm whether the calculated configuration is correct. A check here would provide additional confidence in the validity of this ap­proach and would signal a go-ahead for calculating more complicated structures such as proteins, Dr. Sche­raga feels.

The next stage of the work at Cornell will be to compute the struc­ture of a real protein. For this phase of the project, the chemists have picked lysozyme. The three-dimen­sional structure of this compound was recently determined, using x-ray crys­tallography, by Dr. David Phillips of the Royal Institution in London. Now, Dr. Scheraga and his coworkers want to demonstrate that it's possible to calculate the structure using compu­ter methods.

Dr. Donald J. Metz of Brookhaven National Laboratory has developed a technique for polymerizing styrene with gamma rays which gives repro­ducible results. All previous work in this relatively new field has been plagued by a lack of reproducibility caused by an extreme reaction sensi­tivity to impurities, especially traces of water.

Dr. Metz's method was developed at the Upton, N.Y., laboratory with Dr. Richard C. Potter, Carl Johnson, and Dr. Randolph H. Bretton (of Yale University). The method con­sists of four basic steps:

• Vacuum distilling the monomer. • Vacuum baking the glassware. • Degassing the monomer. • Drying the monomer over acti­

vated silica gel. The rate of polymerization of sty­

rene prepared this way is as much as 4000 times the rate of a classical free radical polymerization of styrene [/. Polymer Scl, 4, 2295 (1966)].

This has academic interest, Dr. Metz notes, in that it supports the belief that ionic radiation reactions are highly probable in organic liquids, even though those liquids have low dielectric constants. Ionic reactions, he says, may turn out to be the normal reaction modes. They may not have been observed previously because of impurities.

The Brookhaven results could have practical commercial application as well, according to Dr. Potter. Indus­try has shunned radiation in the past because radiation rates were low and radiation sources expensive. But the situation has reversed. Radiation rates have soared and costs have dropped. Modifications, such as re­placing the vacuum environment with dry nitrogen, would have to be made for a commercial process, he admits.

But radiation has advantages over catalysts. Radiation works best at low temperatures. This eliminates the danger of exceeding a critical

temperature above which depolymeri-zation or decomposition occurs. Also, catalysts require neutralization or removal after reaction. Radiation doesn't.

In the early days of radiation chem­istry, the late Dr. S. C. Lind pre­dicted that ionic species would domi­nate the mechanisms of organic reac­tions initiated by ionizing radiation. But subsequent work seemed to dis­prove his theoryr especially in poly­merization. Styrene and other mate­rials seemed to polymerize through free radical intermediates only.

In a typical free radical polymeri­zation:

• The polymerization rate is propor­tional to the square root of the radia­tion dose rate.

• Free radical scavengers inhibit or retard polymerization.

The first evidence that a radiation-induced polymerization could be ionic appeared in the late fifties. Independ­ent studies in England by Dr. R. Worrall and his coworkers (Dr. W. H. Davison and Dr. S. H. Pinner) and by Dr. F. S. Dainton and his coworkers showed that isobutylene forms high-molecular-weight polymers at —78° C. when irradiated. Until then, isobutyl­ene had done this only in the presence of ionic catalysts. Diisobutene, which inhibits the catalytic ionic polymeri­zation, also inhibited the low-temper­ature radiation process. These work­ers also found that the polymerization rate has a linear dependence on dose rate, rather than the classical square-root dependence. All this pointed to an ionic mechanism.

Styrene behaves the same way. At —78° C , it polymerizes rapidly in halogenated hydrocarbon solvents. Again, the reaction rate is directly proportional to the dose rate. Radi­cal scavengers such as oxygen or ben-zoquinone do not affect the reaction, but ionic scavengers such as acetone, methanol, and methyl acetate retard the reaction.

OCT. 17, 1966 C&EN 71

Relative water content

High

Low

Very low

Relative overall rate of polymer­

ization

Low

High

Very high

Dose-rate depend­ence of rate of

polymerization (n)

n = % η ~ 1

V2<n<l

Kinetic scheme Prop agating Term ination

species

R· R«+R· -^ Product

R+ ( > > R · ) R + +?->Product

R+ and R- R + ? ) R"+? >-> Product R+R i

At very low water content, carbanions may contribute to the termination of styrene polymerization