BSI Birmingham, March 1995

The following text and graphics provides information based on my presentation entitled "Clinical use and in-vitro evaluation of humanised CD52 antibodies" within the session chaired by Prof. R. Jefferis on "Protective effector functions mediated by humoral factors". You are welcome to use and to refer to the images for teaching and research but please remember that they remain copyright of Mike Clark. If you do find the information of use or if you would like to make comments I would be happy to hear from you by e-mail (mrc7@cam.ac.uk).

Abstract

A major question concerning the in-vivo effectiveness of antibodies directed to cell surface antigens is why some appear to be able to deplete the target cells whereas others are not. Antibodies to the CD52 antigen present on human lymphocytes have been shown to be highly effective both in-vivo and in-vitro at killing human targets and this therefor provides a good system for study (Reviewed Greenwood & Clark 1993). However for some applications it may be desirable to have antibodies which bind but do not activate effector functions eg in the blockade of T-cell receptor molecules for immunosuppression or tolerance induction (Bolt et al 1993). Studies carried out with conventional and recombinant monoclonal antibodies have provided some but not all of the answers.

Studies with chimaeric antibodies to the hapten NP (Bruggemann et al 1987) and also to the CD52 antigen (Riechmann et al 1987) have shown that the human IgG1 (allotype G1m(1,17)) is very effective in killing human targets in-vitro by complement lysis or by ADCC through activated K-cell effectors.

Using chimaeric rat antibodies to the hapten NP (Bruggemann et al 1989) it was shown that the rat IgG2b was also effective in both complement lysis and ADCC but that the rat IgG2a was only effective in complement lysis. In-vivo in the treatment of lymphoma using antibodies directed to CD52 the rat IgG2a was found to be much less effective than the rat IgG2b (Dyer et al 1989).

Humanisation of the rat IgG2a specific for CD52 was carried out and the four human isotypes IgG1, IgG2, IgG3 and IgG4 were made (Riechmann et al 1988). The IgG1 antibody called CAMPATH-1H proved most effective in-vitro in complement mediated lysis and ADCC and was chosen for therapy (Hale et al 1988). Therapy with CAMPATH-1H has shown it to be effective at depleting lymphocytes in a number of situations (Hale et al 1988, Dyer et al 1989, Mathieson et al 1990, Isaacs et al 1992b). Whereas the rat IgG2b antibody CAMPATH-1G was frequently immunogenic on a first course of treatment, the humanised antibody CAMPATH-1H has been found to be much less immunogenic although repeated administration can lead to anti-idiotypic responses in some patients (Isaacs et al 1992b).

The availability of good in-vitro assays for effector functions means that the CAMPATH-1 (CD52) antibodies provide a good system for studying factors to do with IgG effector functions (Greenwood, Clark & Waldmann 1993, Greenwood & Clark 1993). Human IgG1 and IgG4 are very homologous and differ in only a few residues in the constant region domains (see models) and they also have different hinge region sequences. In previous studies on mouse IgG, Duncan and Winter, identified a motif Glu 318, Lys 320, Lys 322 as being important for complement lysis, however all four human IgG antibodies are identical in sequence at this point. Using domain swap mutants of the CD52 antibodies we managed to identify that the residues responsible for the differences in function between IgG1 and IgG4 in complement lysis and ADCC were in the C-terminal half of the CH2 domain (Greenwood, Clark & Waldmann 1993). The residues concerned are Ala 330 Pro 331 in IgG1 and Ser 330 Ser 331 in IgG4 (see models) .

However during these studies we discovered a polymorphism in ADCC with human isotypes (Greenwood, Clark & Waldmann 1993). Some individual donors of effector cells mediate ADCC with all four human IgG subclasses whilst others only mediate ADCC with human IgG1 CD52 antibodies. This polymorphism remains to be resolved but is obviously of importance to those who have implied that IgG4 would be a good choice for a non-depleting use of a recombinant antibody. In a separate series of experiments in collaboration with Inez-Anne Haagen and Jan van de Winkel we discovered that the polymorphism in human Fc gamma RIIa which affects human IgG2 binding and mouse IgG1 binding, also effects rat IgG2b binding (Haagen et al 1995).

We have also studied IgG1 allotypes and aglycosyl allotypes for their effector functions. In human IgG1 there are allotypic residues in the CH1 and the CH3 domains (see models) . In comparisons in complement mediated lysis with CD52 antibodies of different IgG1 allotypes there is very little difference between them. We then compared antibodies in which the conserved N-linked glycosylation site Asn 297 was mutated to Ala. We expected the mutation to knock out all activity in complement lysis. To our surprise the allotypes seemed to differ in their residual activity (Greenwood & Clark 1993). The allotype G1m(3) has more activity than G1m(1,17) and domain swap mutants giving allotypes G1m(1,3) and G1m(17) showed that it was the Arg 214 G1m(3) for Lys 214 G1m(17) in the CH1 domain which was more important ie a residue outside of the CH2 domain.

In-vivo models using recombinant antibodies to murine CD8 have shown that depletion occurs with wild type antibodies but that aglycosyl mutants fail to deplete. The depletion is unaffected by cobra venom factor treatment and thus presumably does not depend on complement lysis (Isaacs et al 1992).

Thus it is still not clear which effector functions are absolutely required for in-vivo depletion. FcR binding rather than complement would seem to be most important but we don't as yet know which FcR or which cells which express them are most important. The results are likely to be complicated by polymorphisms in the antibodies and in the FcR.

References

This page is from (Sig Picture Not Displayed) Mike Clark
"An antibody engineer who also enjoys the mountains."
mrc7@cam.ac.uk
Mike's home-page
11th June 1997