CXCR4 mediates the proliferation of glioblastoma progenitor cells

Abstract

Increasing evidence points to a fundamental role for cancer stem cells (CSC) in the initiation and propagation of many tumors. As such, in the context of glioblastoma multiforme (GBM), the development of treatment strategies specifically targeted towards CSC-like populations may hold significant therapeutic promise. To this end, we now report that the cell surface chemokine receptor, CXCR4, a known mediator of cancer cell proliferation and invasion, is overexpressed in primary glioblastoma progenitor cells versus corresponding differentiated tumor cells. Furthermore, administration of CXCL12, the only known ligand for CXCR4, stimulates a specific and significant proliferative response in progenitors but not differentiated tumor cells. Taken together, these results implicate an important role for the CXCR4 signaling mechanism in glioma CSC biology and point to the therapeutic potential of targeting this pathway in patients with GBM.

Fig. 1

Isolation and characterization of glioma progenitor cells. Fresh primary human GBM explants were disassociated and cultured as described (Section 2) in the absence of serum. Within 3–5 days of culture, we detected the presence of neurosphere-like cellular aggregates termed glioma-derived spheres (GDS) (A and B). These spheres expressed the neural precursor markers nestin (C) and CD133 (D). Furthermore, when re-plated into differentiation inducing conditions, these cells could terminally differentiate into Beta-III tubulin expressing neuron-like cells (E) GFAP positive astrocyte-like cells (F) as well as CNPase expressing oligodendroglial-like cells. (G) These cells also expressed CXCR4 (H; DIC-differential interference contrast image). RT-PCR performed on cDNA generated from these cultures indicated that GDS cells exhibited markedly stronger expression of stem cell antigens than corresponding differentiated tumor cells (I). These findings indicate that GDS consist of multipotent progenitor cells.

Fig. 2

Glioma progenitor cells form tumor in vivo and recapitulate the histopathology of human GBM with high fidelity. To confirm the tumorigenicity of GDS, 50,000 GDS cells were implanted into the basal ganglia using an athymic nude mouse model. Brain tissue was harvested, sectioned, and analyzed four weeks after implantation. Representative images are depicted. Panel A demonstrates the presence of a diffuse tumor in the basal ganglia (main tumor mass T, demarcated by arrow-heads). Note that the tumor appears to be infiltrating laterally along a white matter tract. Panel B represents a high power image of the tumor-normal tissue boundary from the same section depicted in (A). The tumor T, is clearly visible in the left half of the panel, with large irregular nuclei whereas the right extreme of the panel is occupied by normal parenchyma N. The boundary of tumor-normal parenchyma is not clearly visible and is populated by diffusely invading clusters of tumor cells (arrows). Panel C demonstrates a non-immunohistochemically stained higher power magnification of the tumor demonstrating irregular, hyperchromatic nuclei with visible mitoses (two examples are indicated by arrow-heads). Panel D represents tumor-bearing tissue that has been immunohistochemically stained with an anti-human pan-HLA specific antibody and developed with the peroxidase substrate Nova-Red (red-brown precipitate) in order to definitively identify human tumor cells within mouse brain. Note that infiltrative tumor cells (labeled red-brown) extend far beyond the boundaries of the main tumor mass. Panel E is a higher power magnification of the boxed area in (D). Note the large number of invasive tumor cells that infiltrate parenchyma adjacent to the principal tumor mass, T. The highly invasive nature of these xenografts is further illustrated in Panel F which demonstrates the presence of human-HLA+ migratory tumor cells in the ipsilateral corpus callosum at significant distance from the primary tumor mass (the intraventricular choroid plexus is non-specifically labeled due to endogenous peroxidase activity). Panel G represents a high power magnification of the boxed area in F. These cells appear to be in the form of a migratory stream traversing towards the genu of the corpus callosum.

Fig. 3

CXCR4 expression and function in glioma progenitor cells. (A) To assess the expression status of CXCR4 in GBM progenitor cells, we immunohistochemically probed histopathologically-verified human GBM tissue sections for the stem cell antigens Mushashi-1 and Sox-2 as well as CXCR4. This revealed significant staining for both of these markers in tumor tissue within cells bearing large pleomorphic nuclei, thereby confirming the presence of cancerous progenitor cells in human GBM. Importantly, we also observed expression of Sox-2 within tumor cell nuclei as opposed to predominantly cytoplasmic localization in adjacent non-neoplastic cortical tissue. This may indicate activation induced nuclear translocation of this transcription factor within GBM progenitor cells. CXCR4 co-expression was visible within a significant portion of Sox-2 expressing tumor cells. (B) To further quantify differences in CXCR4 expression levels between GDS and corresponding differentiated tumor cells, we utilized real-time quantitative PCR to measure expression in cDNA samples generated from these cultures. As depicted in Table 1, we observed an average of 2.6-fold greater CXCR4 expression in GDS cells versus differentiated tumor. (C) To place the overexpression of CXCR4 in GDS in context, we assayed for the functional consequences of CXCR4 activation. GDS cultures were supplemented with recombinant human CXCL12 for 48 h in the presence or absence of AMD3100 (a small molecule antagonist of CXCR4) and then analyzed by means of a WST-1 proliferation assay. We observed a marked proliferative response to CXCL12 in GDS but not in corresponding differentiated tumor cells. This was completely abrogated by AMD3100, indicating the specific role of CXCR4 in mediating this proliferative response.