Glioma Cell Invasion is Significantly Enhanced in Composite Hydrogel Matrices Composed of Chondroitin 4- and 4,6-Sulfated Glycosaminoglycans

Abstract

Glioblastoma multiforme (GBM) is the most aggressive form of astrocytoma accounting for a majority of primary malignant brain tumors in the United States. Chondroitin sulfate proteoglycans (CSPGs) and their glycosaminoglycan (GAG) side chains are key constituents of the brain extracellular matrix (ECM) implicated in promoting tumor invasion. However, the mechanisms by which sulfated CS-GAGs promote brain tumor invasion are currently unknown. We hypothesize that glioma cell invasion is triggered by the altered sulfation of CS-GAGs in the tumor extracellular environment, and that this is potentially mediated by independent mechanisms involving CXCL12/CXCR4 and LAR signaling respectively. This was tested in vitro by encapsulating the human glioma cell line U87MG-EGFP into monosulfated (4-sulfated; CS-A), composite (4 and 4,6-sulfated; CS-A/E), unsulfated hyaluronic acid (HA), and unsulfated agarose (AG; polysaccharide) hydrogels within microfluidics-based choice assays. Our results demonstrated the enhanced preferential cell invasion into composite hydrogels, when compared to other hydrogel matrices (p<0.05). Haptotaxis assays demonstrated the significantly (p<0.05) faster migration of U87MG-EGFP cells in CXCL12 containing CS-GAG hydrogels when compared to other hydrogel matrices containing the same chemokine concentration. This is likely due to the significantly (p<0.05) greater affinity of composite CS-GAGs to CXCL12 over other hydrogel matrices. Results from qRT-PCR assays further demonstrated the significant (p<0.05) upregulation of the chemokine receptor CXCR4, and the CSPG receptor LAR in glioma cells within CS-GAG hydrogels compared to control hydrogels. Western blot analysis of cell lysates derived from glioma cells encapsulated in different hydrogel matrices further corroborate qRT-PCR results, and indicate the presence of a potential variant of LAR that is selectively expressed only in glioma cells encapsulated in CS-GAG hydrogels. These results suggest that sulfated CS-GAGs may directly induce enhanced invasion and haptotaxis of glioma cells associated with aggressive brain tumors via distinct mechanisms.

Keywords: Glioblastoma; chondroitin sulfate glycosaminoglycans; haptotaxis; invasion.

Fig.1

Evaluation of the biomechanical properties of hydrogel matrices. (A) 500X and 1000X (insert) magnification images of 0.5%AG, 0.5%HA, and 2%CS-A hydrogels showing relative porosity. Scale bars = 100 μm. (B) Pore size measurements done through quantitative ImageJ analysis shows comparable pore sizes in each hydrogel model. (C) Storage modulus measured using a parallel-plate rheometer over a standard 0–100 rad/sec frequency sweep for all hydrogels.

Fig. 2

Viability assessment of cells encapsulated into different hydrogel matrices, as determined by a Calcein Blue AM assay 48h post-encapsulation. Representative 20X GFP/Calcein Blue and 20X GFP/Calcein Blue/Brightfield images of cells encapsulated in 0.5%AG, 0.5%HA, and 2%CS-A hydrogels. Scale bar = 50 μm. No significant differences in % live cells were observed between the three groups, as determined by a one-way ANOVA.

Fig. 3

Microfluidics-based evaluation of cellular preference of hydrogel environment. (A) Schematic of three-channeled PDMS microfluidic devices. Channels are 1000 μm in width, with 100 μm trapezoidal barriers between channels and 4 mm diameter wells. Cells were seeded into center channels, with hydrogel choices placed into right and left channels. Representative 40X oil images of GFP/Hoechst-stained cells from each hydrogel channel shown below. Scale bar = 100 μm. Quantification of cell migration into from each type of choice assay was performed across n=4 for each choice. All choice assays were performed against the monosulfated CS-A. Significant differences were represented by ‘*’ indicating p<0.05. No significance is represented by ‘ns.’ (B) A 10X tiled image showing cell invasion into hydrogel choices. Areas represented by ROIs (red and yellow) indicate hydrogel channel and exclude cell-containing center channel. Scale bar = 1000 μm. (C) Blue ROI represents trapezoidal barriers, with 10X representative images of cells infiltrating into CS-A hydrogels at 6 h post cell-seeding. Scale bar = 100 μm.

Fig. 4

Immunocytochemical staining of encapsulated glioma cells within microfluidics devices 6 h post cell-seeding. (A) FAK (yellow) and vinculin (red) demonstrates evidence of cell migration in hydrogel matrices; (B) Quantification of % F-actin containing cells. Scale bar = 100 μm. (C) Phalloidin staining (red) to visualize F-actin polymerization among cells in each choice assay, with (D) t-test quantifications (p<0.05). Cells were also Hoechst stained (blue) to show cell nuclei. Scale bar 100 μm. Means with ‘*’ (p<0.05) are significantly different, ‘ns’ represents no significant difference.

Fig. 5

Haptotaxis of cells in response to matrix immobilized CXCL12 presence. (A) Proof of establishment of a chemokine gradient performed using Alexa Fluor 488-conjugated bovine aprotinin. Fluorescence was quantified at zero, three and six hour time points (six hour time point data shown in graph compared to zero hours). No significant differences in chemokine gradient diffusion was detected across different hydrogels as evaluated using a one-way ANOVA. Representative images shown from each time point to demonstrate chemokine diffusion through hydrogel matrices after 6 hours. Scale bar 100 μm. (B) Representative 40X GFP/Hoechst images of migrating cells in response to CXCL12 presence in hydrogel matrices after 6 hours. Scale bar = 100 μm.

Fig. 6

Quantification results of haptotaxis in hydrogel matrices with and without 10 ng/mL CXCL12 at three and six hours post-encapsulation within (A) AG, (B) HA, (C) CS-A, and (D) COMP hydrogels. Data are represented as mean ± SD, and means with ‘*’ (p<0.05) are significantly different from other treatments.

Fig. 7

Binding of CXCL12 to immobilized GAGs as quantified by sandwich ELISA assay. (A) Schematic demonstrating ELISA methods to determine amount of bound CXCL12 to biotinylated HA, CS-A and COMP GAGs. (B) Data representing mean OD values obtained across four different CXCL12 concentrations against each GAG analyzed in quadruplicate. Data are represented as mean ± SD, and means with ‘*’ (p<0.05) are significantly different from other treatments.

Fig. 8

Quantitative RT-PCR results demonstrating relative expression levels of (A) CXCL12, (B) CXCR4 and (C) LAR transcripts isolated from encapsulated cells. All fold changes were calculated relative to expression levels in media-only controls, and normalized against expression levels of housekeeping genes GAPDH and HPRT1. Data are represented as mean ± SD, and means with ‘*’ (p<0.05) are significantly different. A label of ‘ns’ demonstrates no significant difference. (D) Western blot results demonstrating the differential expression levels of CXCR4 and LAR receptors in glioma cells encapsulated in CS-GAG hydrogels when compared to control treatments. GAPDH is presented as a loading control.