Abstract
From collected data on variable region sequences of immunoglobulin, H chains, the probability of random associations of any 2 amino-acid residues in the complementarity-determining segments was computed, and pairs of residues occurring significantly more frequently than expected were selected by computer. Significant associations between Phe 32 and Tyr 33, Phe 32 and Glu 35 and Tyr 33 and Glu 35 were found in 6 proteins, all of which were mouse myeloma proteins which bound phosphorylcholine. From the X-ray structure of McPC603, Tyr 33 and Glu 35 are contacting residues; a 7th phosphorylcholine-binding mouse myeloma protein also contained Phe 32 and Tyr 33, but position 35 was only determined as Glx, and this position was not selected. Met 34 occurred in all 7 phosphorylcholine-binding myeloma proteins, but was also present at this position in 29 other proteins and was not selected; it is seen in the X-ray structure not to be a contacting residue. The role of Phe 32 is not obvious, but it could have some conformational influence. A human phosphorylcholine-binding myeloma protein also had Phe, Tyr and Met at positions 32, 33 and 34, but had Asp instead of Glu at position 35 and showed a lower binding constant. The ability to use sequence data to locate residues in complementarity-determining segments making contact with antigenic determinants and those playing essentially a structural role would contribute substantially to the understanding of antibody specificity.