Sir

Your News story about the recent clinical-trial disaster, “Can super-antibody drugs be tamed?” (Nature 440, 855–856; 2006), reports TeGenero founder Thomas Hünig as saying that the monkey CD28 receptor — the target of the TGN1412 antibody — is identical to the human one. TeGenero's ‘investigator's brochure’ (available via http://www.mhra.gov.uk) confirms that toxicity testing of TGN1412 was done in rhesus and cynomolgus monkeys.

Although TeGenero says that the amino-acid sequence of the critical portion of monkey CD28 (the C9D loop) is identical to that of human CD28, published amino-acid and DNA sequence data on rhesus CD28 indicate substantial genetic diversity within the species. Differences of up to 4% (9 of 220 amino acids) between rhesus and human proteins have been found (F. Villinger et al. Immunogenetics 53, 315—328; 2001).

No data are available for CD28 in cynomolgus monkeys, but as they are closely related to rhesus monkeys, a similar degree of variation can be expected.

Two variable positions in the rhesus CD28 sequence — amino acid 65 (glutamic acid or glycine) and 104 (asparagine or tyrosine) — are located on the edge of the contact region between human CD28 and TGN1412. Although they may be tolerated, they are likely to substantially affect the strength of antibody–antigen interaction.

Identity of sequence in the C9D loop is also insufficient as a criterion for compatibility, as this segment contains only 6 of the 12 amino acids of CD28 that make contact with TGN1412. Substitutions at any position in this area might be expected to affect the shape of the antigenic site. Binding of the antibody to rhesus CD28 could therefore potentially be much weaker than its binding to human CD28. And differences in the binding affinities could alter the balance between activation of regulatory T cells (bearing relatively high levels of CD28), on the one hand, and T-helper cells and eosinophilic granulocytes (with lower CD28 expression), on the other (G. Woerly et al. J. Exp. Med. 190, 487–495; 1999).

We propose that future preclinical studies of antibody-based drugs should be performed using a primate species in which the antigenic site matches that in humans completely, and that the individual primates used should be genotyped to exclude polymorphisms.

Furthermore, preclinical studies should also include comparative measurements of the binding affinities for both human antigen and primate antigen, to control for unforeseen variations in protein structure.