A preliminary investigation of the hydrodynamic properties of two novel monoclonal antibodies.

B72.3 is a murine monoclonal antibody which reacts with a range o f ovarian, breast and colon carcinomas. Its immunoscintigraphic applications have been well reported in the literature and i t has now been administered to over 300 patients 11 I . A problem with mouse antibodies is the H A M A (human anti-mouse antibody) response which prevents further therapy. To negate the H A M A response, chimeric B72.3 (cB72.3) has been made and is in clinical trials [2]. T h e effectiveness o f radioimmunoscintigraphy a l s o depends on the tumour penetration properties of the administered antibody and the rate at which it is cleared from the body factors favourably affected by the use of antibody fragments. T h e stability and tumour-binding efficiency of such fragments may be predictably linked to their solution conformations. A s a precursor to the investigation of novel antibody fragments. the solution conformations of B72.3 and cB72.3 are under investigation. with a view t o developing a protocol for macromolecular modelling of this kind. X-ray crystallographic maps of appropriate hinge-deleted immunoglobulins provide a starting estimate for the overall dimensions of a similar antibody. To hydrodynamically model the protein, the TRV program developed by Garcia de la Torre and coworkers 131 is employed. In its most basic form the routine requires only the coordinates and radii o f the bead assembly comprising the model. Among its output are parameters directly related to the diffusion and sedimentation coefficients, intrinsic viscosity and radius o f gyration of the bead model. Consequently it is possible to iteratively refine the model t o match the predicted values for these parameters with their experimentally determined counterparts. Aside from the obvious problem of uniqueness of the model obtained ( a problem which can be alleviated to some extent by increasing the number o f experimentally determined parameters) there is the problem of hydration. Tanford 141 suggests an average protein hydration of 0.2 g water/g of protein. However, the work of Squire & Himmel I S ] obliges one to consider higher values (of the order of 0.5 g of water/g of protein). Clearly solvation to this extent must play an important part in the hydrodynamic behaviour o f protein macromolecules and in an attempt to estimate this cffect for antibodies the Squire & Himmel 151 mcthod has been employed by us. T h e amino acid sequences o f murine IgG1 and human lgG4 are very similar, even in the hinge region. Consequently. a conformational similarity between €372.3 (which is murine IgG 1 ) and cB72.3 (which is human IgG4 with substituted €372.3 variable regions) would not be unexpected. Sedimentation coefficients ( . Y ! ~ , , ~ ) for the two antibodies were determined via sedimentation velocity experiments performed on an MSE Centriscan analytical ultracentrifuge equipped with scanning absorption optics and a monochromator. Within the bounds o f experimental error cB72.3 would appear to exhibit the same sedimentation behaviour as native human IgG4. B72.3 and cB72.3 have values of (6.5 1 f 0.3) S and (6.80 f 0.3) S, respectively, compared with a value o f (6.73 f 0.3) S determined for human lgG4 by K. G. Davis (personal communication). However, the positive