Bispecific Anti-CD20/Anti-CD22 Antibodies with Potent Lymphoma Cytotoxicity.

Abstract

BACKGROUND: Multivalent bispecific antibodies (bsMAbs) have shown improved efficacy in a number of preclinical studies. We made a pair of hexavalent bsMAbs based on epratuzumab (hLL2, anti-CD22) and veltuzumab (hA20, anti-CD20) by the Dock-and-Lock (DNL) method, and compared their properties to each other and their parental MAbs.METHODS AND RESULTS: DNL is used for the site-specific and covalent assembly of modular components, and was utilized to generate stably tethered hexavalent bispecific complexes, each composed of 4 Fab fragments conjugated to an IgG at the latter's carboxyl termini of the heavy chain. DNL-1 has 4 veltuzumab Fabs tethered to epratuzumab IgG; DNL-2 has 4 epratuzumab Fabs bound to veltuzumab IgG. DNL made each construct as a single, defined, homogeneous structure that is stable in serum. All of the constituent Fab fragments are functional, with binding affinities similar to the parental MAbs. In vitro analyses using human lymphoma cell lines demonstrated that, unlike the parental MAbs, the bsMAbs each induced translocation of both CD22 and CD20 into lipid rafts and also strong cell-cell adhesion. DNL1 or DNL2 treatment resulted in increased apoptosis vs. the parental MAbs alone or combined. DNL1 and DNL2 inhibited the growth of Ramos, Raji and Daudi Burkitt lymphoma cell lines without the requirement of crosslinking, and more potently than the combination of the parental MAbs. For Daudi cells, DNL1 and DNL2 showed similar activity, which was approximately 50-fold more potent than the combination of the parental MAbs. For Raji and Ramos, DNL1 was 8–10-fold more potent than DNL2, which in turn was 8–10-fold more potent than the combined parental MAbs. The results suggest that crosslinking of CD20 and CD22 at the cell surface is required for enhanced cytotoxicity. Veltuzumab, but neither epratuzumab nor either bsMAb, displayed CDC activity. Veltuzumab and DNL2 induced a similarly high degree of ADCC. DNL1 induced ADCC to an intermediate level between veltuzumab/DNL2 and epratuzumab. In an ex-vivo assay using fresh whole blood mixed with either Daudi or Raji, DNL1 and DNL2 each demonstrated selective killing of lymphoma cells over normal B-cells compared to veltuzumab or rituximab; the latter depleted normal B-cells with greater efficiency than the bsMAbs. PK studies in mice demonstrated that despite their large size, the bsMAbs have a significantly shorter serum half-life than the IgGs. Even without CDC activity and with a considerably shorter serum half-life, DNL2 had anti-lymphoma efficacy in the Daudi Burkitt lymphoma model in mice that was equivalent to veltuzumab. DNL1 was less potent than DNL2 in vivo, but more effective than epratuzumab and control bsMAbs comprising either epratuzumab-IgG-AD2 (22-14) or 4 veltuzumab-Fab-DDD2 groups (734-22) combined with 4 non-binding Fab-DDD2 groups or a non-binding IgG-AD2, respectively. The anti-tumor efficacy of both DNL1 and DNL2 was abolished in tumor-bearing mice in which their ADCC potential was diminished by depletion of neutrophils and NK cells. These findings suggest that ADCC is the most critical mechanism of action for lymphoma killing in these murine models. DNL1 is more potent than DNL2 in vitro yet DNL2, having the stronger ADCC activity, is more potent than DNL1 in vivo.CONCLUSIONS: These findings suggest that the DNL method can be used to make a variety of multivalent bsMAbs with potent anti-tumor activity, and having distinct properties dependent on their arrangement and composition. CD20/CD22 bsMAbs appear to have different functions than their parental MAbs, even when these were combined, and appear to be potent anti-B-cell lymphoma therapeutics.