ON THE STRUCTURE OF ANTIBODIES AND OTHER PROTEINS

By Linus Pauling

1937, First Version

[With use of the results of the recent development of our knowledge of the structure of molecules as a guide of the structure of molecules and the forces operating between atoms obtained by the investigation of simple molecules, the results of experimental studies of the properties of proteins, and especially of antibodies, can be made the basis for the formulation of a general system of protein structure.] This system is described briefly in the following paragraphs.

We assume that a protein molecule consists of one or more very long polypeptide chains; in the extended configuration [see original Linus Pauling manuscript for drawing] such a chain containing about three hundred amino acid residues (corresponding to a molecular weight of about 34,000) would be about 10,000 Å long. The ease of dissociation of many of the larger protein molecules to units with molecular weights around 17,000 or 34,000 suggests that the chains in all or most proteins contain around 150 or 300 residues. It is possible that in some cases the terminal amino acid carboxyl groups are condensed, forming a very large ring. The evidence that all the natural amino acids have the same (levo) configuration requires that in the coplanar extended form shown above for the polypeptide backbone the side-chain groups R project alternately above and below the backbone plane.

Because of conjugation of the unshared electron pair of the imino nitrogen atom and the carbonyl double bond, the [see original drawing for carbonyl group] group (including adjacent singly-bonded carbon atoms) tends to remain coplanar. There is essential freedom of rotation about the single bonds to the CHR carbon atom, however, and in consequence the molecule might assume any one of an extremely large number of configurations, corresponding to various orientations about each of these bonds.

It was pointed out by Mirsky and Pauling¹ that the high specificity in properties of many native proteins can be explained only by the assumption that all of the molecules of anyone of these proteins have a single well-defined configuration, and that the process of denaturation of the protein involves the change from this defined configuration to other configurations. The large increase in entropy accompanying denaturation is due to the large number of possible configurations accessible to a molecule of the denatured protein.

Among the questions which we now ask are the following: What are the forces which constrain a protein molecule to a particular configuration? What is the nature of the configuration of a native protein molecule? What is the mechanism which leads to the selection of one configuration from the multitude possible?

The forces effective between parts of a protein molecule may be classified as van der Waals forces and hydrogen-bond forces. Van der Waals attractive forces are not strongly directed; they result in the main from the London dispersion interaction, which operates between all atoms and groups, and to a smaller extent from interactions involving permanent electric dipoles. The van der Waals attraction is balanced by the characteristic repulsion between atoms at distances which can be described in terms of non-bonded or van der Waals atomic radii². These forces confer maximum stability on configurations which show the closest packing of atoms permitted by the non-bonded radii.

The forces which are of the greatest importance in determining the configurations of protein molecules are those corresponding to the formation of hydrogen bonds between carboxyl, amino, carbonyl, imino, and hydroxyl groups. (The force of electrostatic attraction between -NH₃ ⁺ and -COO^- groups leads to their approach and the formation of hydrogen bonds, and is accordingly included in the hydrogen-bond forces.) Since the discovery of the hydrogen bond a large amount of information has been obtained regarding its nature. Hydrogen bonds are formed only between electronegative atoms, the strength of the bond increasing with increase in the electronegativity of the two bonded atoms. The energy of a hydrogen bond is about 5 to 10 kcal/mole; there is no additional activation energy accompanying the rupture or the formation of the bond, which accordingly can be broken or be formed with ease. the bond has certain directional and metrical properties.

The 4.6Å main-chain spacing reported by Astbury³ for the extended polypeptide chains in β-keratin and other proteins corresponds to the formation of carbonyl-imino hydrogen bonds which stabilize sheets, with the side chains extending an average of 5Å on either side. The amino and carboxyl groups of the side chains of the di-amino and dicarboxyl amino residues then form hydrogen bonds which hold the sheets together with Astbury's 10Å side - chain spacing. Other configurations, such as Astbury's α-keratin structure, can also be devised which involve approximately coplanar folded main chains with carbonyl-imino hydrogen bonds, and others still in which well-defined layers are not present.

A globular protein molecule held in a definite configuration by hydrogen bonds would retain this configuration even though some of the hydrogen bonds were broken, so long as the groups involved are constrained by adjacent bonds to remain in their initial neighborhood; if, however, enough bonds are broken, the molecule may rearrange to a new configuration1.

[The very many possible configurations for a protein molecule which are separated from one another by potential barriers differ in free energy. It may happen that for a particular protein configuration which is more stable than any other, and that this is the configuration of molecules of the native protein. Denaturation of such a protein would be reversible; by very slowly decreasing the temperature after thermal denaturation or by slowly changing the environment in other ways the molecules of denatured protein would abandon their various configurations the more stable native configuration. Trypsin and hemoglobin are proteins of this type.

The problem of the mechanism of the synthesis of a native protein of this type in vivo is thus reduced to that of the proper ordering of the amino acid residues in the polypeptide chain, which could then assume its native configuration.

Serum globulin and the antibodies are quite different. The antibody for a particular antigen has specific properties, and its molecules have a definite configuration, or more probably several definite configurations with certain common characteristic features. These definite configurations are determined by the antigen in the course of the synthesis of the antibody from its precursor, and they are not distinguished in stability from other configurations of the molecule. The definite prediction can accordingly be made on the basis of this theory of protein structure that the denaturation of an antibody cannot be reversed.

The valuable experiments made by Landsteiner on the formation of antibodies by synthetic conjugated proteins and on the interaction of these antibodies with the corresponding simple molecules (haptens) used as the prosthetic groups in the antigens provide a great amount of information regarding the structure of proteins. The prosthetic groups which are effective in antibody formation are those containing hydrogen-bond-forming radicals, such as -COOH, -SO₃H, -AsO₃H₂, -NH₂, =C=O, -OH, etc., and haptens containing these radicals combine most strongly with the antibodies. It is clear that the configuration of the antibody must be such as to present a complementary atom or group for each of these radicals. The facts that benzoic acid but not o-chlorobenzoic acid or other ortho or meta substituted benzoic acids combines with the antibody for an antigen obtained by coupling p-aminobenzoic acid with a protein and that both o-chlorobenzoic acid and o-methylbenzoic acid combine with the antibody for the antigen with 2-chloro-4-aminobenzoic acid coupled to a protein show that the attraction between the antibody and hapten (as well as antigen) involves can der Waals forces as well as hydrogen-bond forces; the antibody surrounds the prosthetic group (hapten) nearly completely. The antibody can be described as having a hole into which the hapten or the prosthetic group of the conjugated antigen fits closely, so as to give the maximum van der Waals attraction, and, moreover, the antibody provides suitably located groups to form hydrogen bonds with each active radical of the hapten or antigen.

It is a fundamental property of proteins that their physiological reactions depend upon small differences - their reactions are never violent but always mild. Only by bringing into action in the most effective manner all of the forces which can act between antibody antigen is specific antibody activity achieved.

Many further examples supporting this theory, which need not be quoted here, can be found in the tables of antibody-antigen and antibody-hapten inhibition reactions given by Landsteiner. An antibody molecule must contain holes for the binding of at least two antigen molecules.