The wish to evaluate antibodies in ferrets without having to dose at frequent intervals and without having ferret ADA immune responses interfere with antibody efficacy prompted our effort to test whether these issues could be minimized by redesigning the test antibodies. As we were not aware of published reports that explicitly described the pharmacokinetics of human mAbs in ferrets, we first determined the half-life of a human mAb. For this, a human IgG mAb not thought to bind endogenous ferret antigens (the mAb is specific for the F glycoprotein of RSV) was used as test antibody in order to avoid any antigen-mediated effects on antibody pharmacokinetics. When the PK of the original human IgG1 mAb was studied in ferrets, we observed a half-life of 31 ± 10 hours, longer than the 14 hours we estimated from the published work of Bossart et al., but still much shorter than what has typically been observed in other common models such as mice and cynomolgus monkeys. The introduction of two amino acid substitutions in the human IgG1 Fc, H433L, and N434S, extended its half-life to 72 ± 21 hours, a 2·3-fold improvement over the original 31 hours. Although this met our initial goal of a 3-day half-life, some ferrets showed indications of an ADA response. The presence of ADAs was confirmed in two plasma samples, and a third showed a rapid drop in plasma mAb concentrations indicative of an immune response. This sample had a sixfold higher plasma concentration of residual human mAb, which may have blocked detection of ADAs in our assay. Antibody immune responses against heterologous test antibodies in model species are a common issue, and the incidence increases when more than one dose is given. Because ADAs usually develop 7–10 days after exposure, studies which extend beyond this point risk the effects of ADA-mediated clearance on efficacy. Avoiding ADA responses would be more likely if the antibody test articles contained ferret constant regions. Therefore, we cloned the ferret HC and LC constant regions and used them to make human/ferret chimeric antibodies. Having seen that substitutions in the human Fc could increase half-life in the ferret, several substitutions which had been shown to increase the half-life of human antibodies were introduced into the ferret Fc. The chimeric mAb and variants were screened in ferret FcRn binding assays. Because the sequence of ferret FcRn was only partially known, it also had to first be cloned from ferret tissues. It should be noted that while FcRn binding at pH 6 has been used to compare antibodies engineered for increased half-life, many antibodies which display tighter binding at pH 6 also bind more tightly at higher pH and are not efficiently released. Therefore, FcRn binding at pH 6 can be useful as an initial screening tool, but must be confirmed by in vivo testing. FcRn binding analyzed by an AlphaScreen-based assay revealed the following rank order of affinities for the variants: S252Y=YTE>S252M>N434S>WT. The finding that the serine at position 252 naturally found in the ferret antibody sequence results in decreased binding to ferret FcRn compared with the methionine found at this position in humans is interesting and gives support to the idea that the intrinsic clearance of endogenous antibodies in ferrets is more rapid than in other species. The WT chimeric mAb and three variants, S252M, S252Y, and YTE, were then included in a ferret PK study. The half-lives of the chimeric mAbs in vivo were consistent with the FcRn binding assays, showing a rank order of half-lives of S252Y=YTE>S252M>WT. The strongest FcRn binders S252Y and YTE had half-lives of ~9 days. However, perhaps the most important observation from this study was the lack of precipitous drops in antibody plasma concentration which would indicate strong ADA responses. Others have described dramatic decreases in injected antibody concentrations as pharmacokinetic indications of an ADA response. Preferably, this could have been supported by the direct detection of ADAs; however, residual test article in the plasma samples partially inhibited detection. Therefore, while the observations presented here make a strong case that chimeric human/ferret antibodies are less immunogenic in ferrets, the possibility exists that weak immune responses went undetected. Future studies might include a wash-out period to allow injected antibody to be fully cleared.
Overall, these results show that a chimeric mAb with ferret constant regions can be used to minimize immunogenicity in ferrets and can be engineered for a long half-life. The chimeric mAbs were properly assembled, were biophysically well-behaved, and retained full antigen binding. It should be straightforward to prepare such chimeric antibodies from any antibody variable regions and thereby gain the advantages of lower dosage and less frequent dosing, as well as reduced variability between animals, especially in long-term multidose studies.