(761b) Interstitial Flow Increases Glioma Cell Migration Via CXCR4/CXCL12 Dependent Autologous Chemotaxis

Introduction: Malignant brain tumors have a highly invasive phenotype that makes them difficult to treat with normal therapies. The invasion of gliomas is well documented but still poorly understood and thus difficult to overcome in a clinical setting. Patterns of migration in the brain correlate with cerebrospinal fluid flow patterns. Further, gradients of CXCL12 coincide with the same flow patterns in the brain. Additionally, it has been well studied that the chemokine CXCL12 has a chemoattractant effect on many gliomas and the degree to which this chemokine attracts glioma cells correlates with degree of malignancy.  Here we propose a mechanism for glioma invasion in the brain that is dependent on flow and chemokine gradients. The Swartz lab has previously described a mechanism of autologous chemotaxis by breast cancer cells under interstitial flow conditions via CCR7/CCL21 signaling (Shields, J., et al. 2007 Cancer Cell). This mechanism says that a cell secretes a chemokine that will bind to the extracellular matrix. Under non-flow conditions, the chemokine binds radially out in all directions. However, when interstitial flow occurs around the cell, this chemokine is pushed downstream and creates a gradient around the cell causing chemotaxis. We propose and show evidence that a similar mechanism of flow-induced migration is occurring via the chemokine/receptor pair CXCL12/CXCR4 which is so prevalent in brain tumors.

Methods: Several glioblastoma cell lines were used for experiments (RT2, C6, 9L and U87MG). 100,000 cells were embedded in 3-D matrices consisting of 0.08% hyaluronic acid (Glycosan) and 0.12% collagen I (BD Biosciences) at a total volume of 100ul in a tissue culture insert (Millipore). Flow was applied by adding media to the top of the transwell  flowed overnight by pressure differential. Gels were removed, inserts imaged and quantified. For chemokine detection, gels, cells and media were collected, protein quantified, and CXCL12 measured using ELISA (R&D). For live imaging, cells were plated in gels in a specially designed microfluidic device with pump-controlled flow applied and imaged over 16 hours in an incubated chamber. Cell migration was analyzed using ImageJ tracking programs and MATLAB postprocessing for analysis of migration velocity and distance. Further studies were conducted in radial flow chambers to examine the mechanism by which chemotaxis occurs through immunohistochemical staining. Last, to model chemotaxis, a COMSOL model was developed from a previously used model (Fleury, M, et al. 2006 Biophys. J.) to calculate the gradients formed around cells in our in vitro system.

Results: RT2 migration was enhanced the most at 2.5 fold when exposed to flow in a 3-D matrix. Other cell types also showed increased invasion under flow conditions (1.5-2 fold). Transmigrated cells are expressed as a percent of total cells and this increased from an average of 0.1% of total to 0.3% of total cells (p<0.001). Further, addition of the small molecule CXCR4 inhibitor AMD3100 (Sigma) at a concentration of 10µM inhibited cell migration enhancement under flow back to 0.1% migrated cells (p<0.001).  This data indicates that there is an effect of interstitial flow on the migration of glioma cells and that this enhancement is dependent on the chemokine receptor pair CXCL12-CXCR4.

Results of live imaging revealed two interesting effects: general cell migration was increased under flow conditions (p<0.05), and there was a general change in the trend of cell direction with velocity moving in the direction of flow in a subset of cells. Further, in follow up experiments in radial flow chambers with the chemokine CXCL12 in excess, there was an inhibition of the flow-enhanced invasion effect. These data together indicate that there is directional migration of cells and that blocking the ability of gradients to form in the gel yields a decreased effect. Results of the model indicated that based on the experimental factors involved in the studies, a 9% CXCL12 gradient would form around an individual cell at disease-state flow rates (2µm/s linear velocity).

Conclusions: We show here that flow influences the migration of glioma cells in a 3-D microenvironment and that this enhancement is dependent on CXCL12 and CXCR4. This is significant in that this chemokine and receptor are highly implicated in the malignancy of brain cancer and development of an invasive phenotype. Building upon previous research concerning autologous chemotaxis using CCL21 in tumors of the periphery, we show here that there is a similar mechanism in central nervous system tumor migration with the chemokine CXCL12. Thus, we show the first instance of flow directly affecting brain tumor cells and evidence the possibility of autologous chemotaxis in glioma.