Study of disseminated high-risk neuroblastoma in the bone marrow niche; from microenvironmental modelling to therapeutic targeting

Neuroblastoma (NB) is the most common extracraneal solid tumor diagnosed in the first 5 years of childhood and accounts for approximately 15% of all pediatric cancer-related deaths. At the time of diagnosis, about half of NB patients present disseminated disease, being the bone marrow (BM) the most common site of dissemination. The persistence of infiltrated BM during treatment or relapse is predictive of patient poor outcome. The BM microenvironment has unique biologic properties that favor progression of disseminated NB tumor cells. In this environment, the receptor CXCR4 has a pivotal role for BM homeostasis and is involved in metastatic dissemination in several cancers. In NB, CXCR4 is expressed in tumor cells; however, its oncogenic role in relation with its ligand CXCL12 has shown contradictory results. Using an in vitro model that recapitulates low oxygen levels and chemokine signaling present in the BM environment, we explored whether CXCR4 together with MIF, a second ligand, is critical for NB survival and proliferation in the niche. We also evaluated MIF inhibition as a therapeutic option for NB. To develop a BM-based in vitro model, NB cells were cultured in different conditioned media (CMs) derived from supernatants of patient-derived BM primary cells (CM-BM) and NB cell lines (CM-NB) cultured alone, or in combination (CM-BM/NB). To mimic BM oxygen levels, NB cells were cultured under hypoxia (1% O2), as compared to cell culture normoxia (21% O2). Expanded BM cultures used to generate CMs contained a heterogenic population with a predominant cellularity positive for mesenchymal stromal markers measured by flow cytometry. Cytokine arrays and ELISA assays of CMs revealed MIF as the highest NB released cytokine, whereas CXCL12 was not detected. The expression of MIF and CXCL12 signaling pathways was analyzed with different public NB databases. Among the analyzed genes, the high expression of CXCR4 and MIF was associated with patient poor outcome and high-risk disseminated staging. To further explore the role of CXCR4/MIF axis in high- risk disseminated NB, in vitro and in vivo functional studies were performed with or without the covalent MIF inhibitor 4-IPP and the CXCR4 antagonist AMD3100. When exposed to BM-derived CMs and hypoxia, BM-derived NB cell lines showed increased surface expression of CXCR4 by flow cytometry. The same culture condition increased phosphorylated levels of AKT/PI3K and ERK/MAPK measured by western blot. Cell viability assays showed that hypoxic conditions and BM-derived CMs enhanced NB cell proliferation at different time points. Similarly, in vivo, the co-injection of BM cells favored NB tumor progression by reducing engraftment times in contrast to NB injection alone. Using wound healing assays and Matrigel-coated Transwells, CM-BM/NB chemoattracted and enhanced migration and invasion of LAN-1 cells cultured under hypoxic conditions. These aggressive phenotypes promoted by the BM-based model were reverted by adding sub-lethal concentrations of the MIF inhibitor 4-IPP. We also explored whether MIF present in our CMs affected response to chemotherapy. After treatment with doxorubicin and etoposide at IC50 values LAN-1 cell viability increased in CM-BM/NB compared to control media. In both cases, chemo-sensitivity was restored when 4-IPP was added to CM-BM/NB. Finally, the administration of 4-IPP delayed the tumor progression and increased mice survival in a LAN- 1 subcutaneous xenograft model. In conclusion, our findings provide new understanding of the contribution of BM microenvironment to NB progression. Based on our BM-model, the relationship between the BM microenvironment and NB cells appears mediate, in part, by the autocrine CXCR4/MIF signaling axis. Furthermore, our results suggested that MIF could represent a therapeutic target for the treatment of patients with high-risk NB