[Show abstract][Hide abstract]ABSTRACT: The provision of the T-cell costimulatory molecule B7 to tumor cells can be an effective way to trigger a tumor-specific cytolytic T-cell response. One way to provide B7 to tumor cells would be to couple an antitumor antibody directly to B7. Such a molecule should target tumors displaying antigen and provide the costimulatory signal to T cells, resulting in the initiation of an antitumor T-cell response. To this end, a fusion protein was designed that incorporates a single-chain antibody fragment (scFv) to erbB-2 (Her2/neu), an oncogene product overexpressed by 30% to 50% of breast carcinomas, and the ECD of B7-2 (CD86). This fusion protein, expressed and purified from Pichia pastoris, was shown to retain binding activity to both counter receptors, erbB-2 and CD28. The fusion protein was also shown to target erbB-2-positive tumor cells and to deliver a CD28-specific T-cell costimulatory signal. These results suggest that a fusion protein engineered to target tumor cells and signal T cells for activation may be an effective means of cancer immunotherapy. Further studies should be performed to characterize the fusion protein in erbB-2 tumor-bearing mice for in vivo tumor targeting, biodistribution, and efficacy.
[Show abstract][Hide abstract]ABSTRACT: Mr 25,000 single-chain Fv (scFv) molecules are rapidly eliminated from the circulation of immunodeficient mice, yielding highly specific retention of small quantities of scFv in human tumor xenografts. We postulated that the specific retention of scFv in tumor could be enhanced by engineering significant increases in the affinity of the scFv for its target antigens. Affinity mutants of the human anti-HER2/neu (c-erbB-2) scFv C6.5 were generated by site-directed mutagenesis, which target the same antigenic epitope with a 320-fold range in affinity (3.2 x 10(-7) to 1.0 x 10(-9) M). In vitro, the Kd of each scFv correlated closely with the duration of its retention on the surface of human ovarian carcinoma SK-OV-3 cells overexpressing HER2/neu. In biodistribution studies performed in scid mice bearing established SK-OV-3 tumors, the degree and specificity of tumor localization increased significantly with increasing affinity. At 24 h after injection, tumor retention of the highest affinity scFv was 7-fold greater than that of a mutant with 320-fold lower affinity for HER2/neu. Because the rapid renal clearance of scFv may blunt the impact of improved affinity on tumor targeting, the distributions were also assayed in the absence of renal clearance (e.g., in mice rendered surgically anephric). In this model, the peak tumor retentions of the two higher affinity scFv approximated that reported previously for IgG targeting the same SK-OV-3 tumors in scid mice with intact kidneys. In contrast, the mutant with the lowest affinity for HER2/neu failed to accumulate in tumor, indicating the presence of an affinity threshold that must be exceeded for active in vivo tumor uptake. These results indicate that affinity can significantly impact the in vivo tumor-specific retention of scFv molecules.
[Show abstract][Hide abstract]ABSTRACT: We determined the extent to which additional binding energy could be achieved by diversifying the complementarity determining regions (CDRs) located in the center of the antibody combining site of C6.5, a human single-chain Fv (scFv) isolated from a non-immune phage library which binds the tumor antigen c-erbB-2. CDR3 of the light (V(L)) and heavy (V(H)) chain variable region of C6.5 were sequentially mutated, the mutant scFv displayed on phage, and higher affinity mutants selected on antigen. Mutation of V(L) CDR3 yielded a scFv (C6ML3-9) with a 16-fold lower Kd (1.0 x 10(-9) M) than C6.5. Due to its length of 20 amino acids, four V(H) CDR3 libraries of C6ML3-9 were constructed. The greatest increase in affinity from a single library was ninefold (Kd = 1.1 x 10(-10) M). Combination of mutations isolated from separate V(H) CDR3 libraries yielded additional ninefold decreases in Kd, resulting in a scFv with a 1230-fold increase in affinity from wild-type C6.5 (Kd = 1.3 x 10(-11) M). The increase in affinity, and its absolute value, are comparable to the largest values observed for antibody affinity maturation in vivo or in vitro and indicate that mutation of V(L) and V(H) CDR3 may be a particularly efficient means to increase antibody affinity. This result, combined with the location of amino acid conservation and substitution, suggests an overall strategy for in vitro antibody affinity maturation. In addition, the affinities and binding kinetics of the single-chain Fv provide reagents with potential tumor targeting abilities not previously available.