Targeting RANKL for Immunotherapy of Multiple Myeloma

Abstract 2905

The TNF family member RANKL and its receptors RANK and osteoprotegerin are key regulators of bone remodelling, but also largely influence progression of cancers of various origins (Tan et al., Nature 2011; Sordillo et al., Cancer 2003). Recently we reported that multiple myeloma (MM) and chronic lymphocytic leukemia (CLL) cells can express RANKL. Moreover, we generated RANK-Ig fusion proteins with wild type (RANK-Fc-WT) or modified human IgG1 Fc parts displaying highly enhanced (RANK-Fc-ADCC) or abrogated (RANK-Fc-KO) affinity to the NK cell Fc receptor to induce antibody-dependent cellular cytotoxicity (ADCC) against RANKL-expressing targets (Schmiedel et al., Blood 2010 116, 21 :1252–1253). Here we extended our analyses on RANKL expression and release in B cell-derived malignancies and evaluated the effects of RANKL targeting with our fusion proteins in vivo. We found substantial expression of RANKL on plasma cells in 35 of 44 (80%) MM patients. RANKL in soluble form (sRANKL) was detectable in culture supernatants of bone marrow (BM) cells of MM patients (6/10) and MM cell lines (all 7). In contrast, RANKL release was neither observed with blood (n=10) and BM (n=4) cells of healthy donors nor with primary CLL cells (n=10) despite the fact that all CLL cells expressed RANKL on the cell surface. In line, mRNA of the splice variant for sRANKL was expressed only in BM cells of MM patients (7/10) and all 7 MM cell lines, but not in blood and BM samples of CLL patients and healthy donors, respectively (n=10 each). In light of available data that neutralization of RANKL prevented disease progression in a MM mouse model (Sordillo et al., Cancer 2003) we went on to evaluate the anti-tumor efficacy and also potential side effects of our Fc-modified RANK-Fc fusion proteins in vivo. First we took advantage of the fact that human RANK binds both human and mouse RANKL (Bossen et al., JBC 2006) to determine potential in vivo toxicity of the fusion proteins. Neither RANK-Fc-ADCC nor RANK-Fc-KO application altered cell counts or distribution in peripheral blood or BM in DBA/2N mice, and analysis by computed tomography after 10 and 30 days revealed no alterations of bone density. In a syngeneic DBA/2N tumor model, treatment with RANK-Fc-WT and RANK-Fc-ADCC significantly (both p<0.05, Mann-Whitney U test) prolonged survival of mice inoculated with RANKL-expressing P815 cells as compared to treatment with RANK-Fc-KO or isotype controls. No effects of treatment were observed in mice inoculated with RANKL-negative P815 cells confirming the target antigen restriction of the effects induced by RANK-Fc-WT and RANK-Fc-ADCC. Surprisingly, no difference between the effects of RANK-Fc-ADCC and RANK-Fc-WT was observed. Since RANK-Fc-ADCC induced by far higher ADCC against RANKL-expressing targets than RANK-Fc-WT in analyses with human NK cells in vitro we next aimed to elucidate why this was not mirrored in our mouse model. Comparative analyses with human PBMC and splenocytes of two different mouse strains (DBA/2N and C57/BL6) revealed that ADCC of NK cells was generally rather weak in mice as compared to the human system, and in line with the in vivo results no benefit of the Fc-optimization was observed with mouse NK cells. These results indicate that ADCC induction by human Fc parts is less potent in mice compared to humans raising doubt regarding the validity of mouse models to study ADCC in general and immunostimulatory effects of Fc-modifications of therapeutic proteins or antibodies in particular. However, when interpreting the results of our tumor model in light of these findings it seems likely that the therapeutic effects of RANK-Fc-ADCC in humans will exceed that observed in mice. Thus, our data provide a clear rationale for evaluation of RANK-Fc-ADCC treatment in patients with MM and potentially also other B cell-derived hematopoietic malignancies.