CD44v6 Is Required For In Vivo Tumorigenesis Of Human AML and MM Cells: Role Of Microenvironmental Signals and Therapeutic Implications

Targeting the interactions between tumor cells and their microenvironment is an exciting new frontier in cancer therapy. The biology of acute myeloid leukemia (AML) and multiple myeloma (MM) is characterized by addiction to specific signals uniquely provided within the bone marrow (BM), where tumor cells preferentially home and locally thrive. The hyaluronate receptor CD44 was shown to be required for retroviral-induced leukemogeneis in syngeneic mouse models. Conversely, CD44 mAbs interfere with human leukemia initiation in immunocompromised mice by inhibiting leukemia stem cell homing to the bone marrow (BM). The therapeutic potential of CD44 mAbs is also under clinical investigation in humans. Much less is known on the role of the differently spliced CD44 variant isoforms. The expression of exon 6 (CD44v6) conveys additional properties to standard CD44, like binding to osteopontin and cooperation with different tyrosine kinase receptors (RTKs), like VEGF receptor type II and c-Met. Interestingly, CD44v6 is the most abundantly expressed CD44 isoform in both AML and MM, where it correlates with a bad prognosis. Since CD44v6 expression is much more tumor-restricted than CD44, targeting this isoform may have a better efficacy/toxicity profile than targeting the standard molecule.

To preclinically validate CD44v6 as a therapeutic target in AML and MM

By FACS analysis and RT-qPCR, we established CD44v6 over expression in a relevant fraction of leukemic blasts from AML pts (15/25, 60%) with preference for the M4-5 FAB subtypes, and in the majority of malignant plasmacells from MM pts (13/15, 87%). CD44v6 was also over expressed on THP-1, Kasumi and U937 human AML cells, and on MM.1S, XG-6 and XG-7 MM cells. To address the specific role of CD44v6 in BM homing, we pretreated MM1.S cells with either a CD44 mAb (SFF-2) or a CD44v6 mAb (VFF-18) and infused them i.v. in NSG mice. Unexpectedly, while SFF-2 almost completely inhibited early (18hrs) homing to the BM compared with an irrelevant mAb, VFF-18 had no effect. To rule out confounding variables associated with specific mAb clones, we silenced CD44v6 expression in MM1.S cells by lentiviral-mediated shRNA transduction and confirmed no difference in BM homing compared with control LV-transduced cells. Longer follow-ups (4-6 weeks) however revealed that, despite unaltered rates of in vitro proliferation, CD44v6-silenced MM1.S cells were severely hampered in their tumorigenic capacity in vivo (P<0.001). These results were confirmed by using THP-1 cells (P<0.001) and primary leukemic blasts (P<0.01). Hypothesizing that CD44v6 may be crucial for in vivo tumorigenesis by cooperating with RTKs, we set-up a co-culture system with BM-derived mesenchymal stromal cells (MSCs), which are producers of VEGF and the c-Met ligand HGF. MSCs protected a wide range of tumor cells, including primary leukemic blasts, from spontaneous apoptosis (P<0.05) and from apoptosis induced by Ara-C or daunorubicin (P<0.01), or bortezomib in the case of MM cells (P<0.001). Comparable results were obtained by using MSC supernatants, hinting to a causative soluble factor, which was neither VEGF nor HGF, as demonstrated by inhibition experiments with bevacizumab and crizotinib, respectively. Noteworthy, MSCs or their supernatants prompted a significant up-regulation of CD44v6 expression levels (P<0.01). Most importantly, preventing CD44v6 up-regulation on tumor cells by shRNA silencing restored their sensitivity to spontaneous and drug-induced apoptosis (P<0.01).

These results clearly indicate that CD44v6 is dispensable for BM homing, but responsible for AML and MM addiction to microenvironmental signals. Combining CD44v6 targeting with cytotoxic chemotherapy might interfere with this vicious circle and result in higher and/or more durable response rates.

No relevant conflicts of interest to declare.