Bone Marrow Interaction and Multiple Myeloma - Approximating Reality in Novel High-Throughput Multiple Myeloma Coculture Systems

Introduction: In the past decade, substantial progress has been made in the understanding of multiple myeloma (MM) cell biology and its interaction with the bone marrow microenvironment (BMM). Binding of MM cells to BM stroma cells (BMSCs) alters the expression of SDF-1α and its receptor CXCR4, leading to the secretion of anti-apoptotic cytokines, promoting tumor growth, drug resistance and migration. MM cancer stem cells migrate to endosteal BM niches, where they escape therapies in a quiescent state causing relapse in the course of the disease. The development of novel agents that aim to target the MM and BMM interaction includes drugs as promising as 2^nd and 3^rdgeneration IMIDs or proteasome inhibitors. Despite these profound advances, the failure rate of preclinically proven cytotoxic single substances is sizeable, as preclinical models often lack the biological, genetic, etiological and immunological properties of the disease (Schüler, Expert. Opin. Biol. Ther. 2013; Kortüm. CLML 2014; Rongvaux. Annu Rev Immunol 2013).

Methods & Results: We have previously demonstrated that BM interaction and homing to niches, mediated by the adhesion molecules CXCR4, CD49d and CD44, protect MM cell lines (MMCL) and primary plasma cells (PC) from the cytotoxic effect of anti-MM agents, such as bortezomib (Bor), vorinostat (Vor) and pomalidomide (Pom). Our in vitro and in vivo observed cytotoxic effects from Bor, Vor and Pom confirmed their potent cytotoxicity, whereas cocultivation with M2-10B4 substantially reduced apoptosis and induced tumor protective effects. Additional treatment with the CXCR4 inhibitor AMD3100 blocked CXCR4 in coculture, but left CD49d, CD44 and CD11a widely unchanged. Toxic or therapeutic effects from AMD3100 monotherapy were excluded for the doses used. Comparison of the CXCR4 antibody (ab)-clones 12G5, 44717 and 4G10 revealed that AMD3100 treatment of U266 cells reduced CXCR4 expression with use of 12G5 and 44717, whereas binding of both FITC- and PE-coupled 4G10 was not influenced, making the latter the most reliable for CXCR4 analysis. Use of image cytometry (IC) allowed accurate visualization of co-localisation of CXCR4 expression both on the cell surface and within the cytoplasm of MM cells. IC correlated with flow cytometry-determined CXCR4 expression and allowed the detailed assessment of treatment studies with and without anti-MM agents and AMD3100. Of note, AMD3100 resensitized MM cells to Bor, Vor and Pom (Waldschmidt. Blood 2012:2450), whereas carfilzomib (Cfz) reduced CXCR4 expression in MMCL and could not be antagonized by stroma coculture. Cfz sensitivity was not increased by adding AMD3100 (Simon. Blood 2013:3851). These preclinical studies need additional adaptation to the clinical setting in order to surpass prior drug failure rates, and there is a need to develop more broadly available and better predictive preclinical systems. Therefore, we are currently assessing a 3D co-culture MM model composed of agarose matrix interlayers, based on a novel liquid overlay technique. This model has been specifically adapted to MM cell and BM component interactions as described (Udi. BJH 2013; Zlei. Exp Hematol 2007; Schüler. EOBT 2013). MM cells are cultivated in conical microwells of a non-adherent agarose matrix after BMSCs were plated on the bottom of each plate, allowing the diffusion of soluble cytokines but no direct contact between BMSC and MMCL. Therein, we are presently testing novel anti-MM substances in comparison to our standard-coculture system.

Conclusion: Targeting microenvironmental mediators, like SDF-1α and CXCR4, is a promising approach to expand the choice of antimyeloma agents and amplify the effects of established antimyeloma drugs, as previously shown by us and others for the combination of AMD3100 and Bor or Pom. However, as our knowledge on MM and its BMM has dramatically increased a great effort has been made in the preclinical testing of promising new anti-MM agents, and more complex high-throughput in vitro models are urgently needed to better predict the potency of these substances in order to reduce dropouts in clinical trials. We hereby provide a novel approach which better reflects the spatial growth of human MM samples in BMSC coculture, and more closely mimics the growth and proliferation of human MM clones in vivo.