Neoplastic Cells are the Major Source of MT-MMPs in IDH1-Mutant Glioma, Thus Enhancing Tumor-Cell Intrinsic Brain Infiltration

Abstract

Tumor-cell infiltration is a major obstacle to successful therapy for brain tumors. Membrane-type matrix metalloproteinases (MT-MMPs), a metzincin subfamily of six proteases, are important mediators of infiltration. The cellular source of MT-MMPs and their role in glioma biology, however, remain controversial. Thus, we comprehensively analyzed the expression of MT-MMPs in primary brain tumors. All MT-MMPs were differentially expressed in primary brain tumors. In diffuse gliomas, MT-MMP1, -3, and -4 were predominantly expressed by IDH1^mutated tumor cells, while macrophages/microglia contributed significantly less to MT-MMP expression. For functional analyses, individual MT-MMPs were expressed in primary mouse p53^-/- astrocytes. Invasion and migration potential of MT-MMP-transduced astrocytes was determined via scratch, matrigel invasion, and novel organotypic porcine spinal slice migration (OPoSSM) and invasion assays. Overall, MT-MMP-transduced astrocytes showed enhanced migration compared to controls. MMP14 was the strongest mediator of migration in scratch assays. However, in the OPoSSM assays, the glycosylphosphatidylinositol (GPI)-anchored MT-MMPs MMP17 and MMP25, not MMP14, mediated the highest infiltration rates of astrocytes. Our data unequivocally demonstrate for the first time that glioma cells, not microglia, are the predominant producers of MT-MMPs in glioma and can act as potent mediators of tumor-cell infiltration into CNS tissue. These proteases are therefore promising targets for therapeutic interventions.

Keywords: glioma; invasion; isocitrate dehydrogenase; matrix metalloproteases; organotypic cell invasion assay.

Figure 1

Gene expression of MT-MMP14, -15, -16, -17, -24 and -25 in 67 primary brain tumors. (A) A multiprobe RPA was established to determine the RNA amounts of the six hMT-MMP genes. Ribosomal protein L32 (RPL32) served as a loading control. Representative samples of normal brain (NB), piloctic astrocytoma (AI), diffuse astrocytoma (AII), anaplastic astrocytoma (AIII), and glioblastoma (GBM) are shown. The IDH mutation status of diffuse gliomas is indicated below the RPA image: + either IDH1 or IDH2 mutated; − IDH wild-type; ○ IDH status not known. (B) Levels of MT-MMP gene expression in different tumor types expressed relative to the level in normal brain tissue, diffuse oligodendroglioma (OII), anaplastic oligodendroglioma (OIII), ependymoma (EII), anaplastic ependymoma (EIII), and medulloblastoma (MB). (✕ diffuse gliomas or glioblastoma with a mutation in the IDH1 or IDH2 genes, ○ wild-type for either IDH, ● IDH status not determined).

Figure 2

Characterization of MT-MMP-expressing cells in glioma tissue. Antibodies against mutated IDH1 identified tumor cells and anti-CD68 marked microglia/macrophages. Double staining with the respective MMP antibodies allowed for the identification of MMP-producing glioma cells and microglia/macrophages, respectively. DAPI was used to mark all nuclei and was used to calculate the total cell numbers in the slides. Each figure depicts overlays of three individual photographs taken of the stains using the respective filter cubes. (Scale bars for MMP14, -16, -17: 50 µm; scale bars for MMP25: 20 µm). Representative samples of diffuse or anaplastic astrocytoma are shown. Insets in the lower left show the triple labeling at higher magnification.

Figure 3

Cellular composition of IDH1-mutant gliomas (four AII, five AIII, and four OII). (A) IDH1-R132H-immunoreactive (IR) cells were more frequent in the solid portion of the tumor than in the infiltration zone, and presented a significantly higher percentage of cells than CD68-IR microglia/macrophages. (B) The individual MMPs were produced by different proportions of the cells, with MMP14 expressed by about 50% of cells, and MMP17 by 60% of cells in the infiltration zone and over 80% in the solid portion of the tumor. (C) Tumor cells constituted the largest fraction (50–80%) of MMP-expressing cells, while macrophages/microglia accounted for 5–20% of MMP-producers. (D) While nearly all CD68-IR microglia/macrophages produced all three MMPs, only 50%, 80%, and 90% of IDH-mutant tumor cells showed expression of MMP14, MMP16, and MMP17, respectively. (∗ p < 0.05; ∗∗∗ p < 0.001, t-test.).

Figure 4

Cell migration determined by scratch assay. (A) Equal numbers of astrocytes were grown for two days in six-well plates. Twelve hours before scratching, cultures were washed with PBS and serum-free medium was applied. The scratch was applied using a pipette. The scratched area was photodocumented at five positions immediately after the scratch was applied and 24 h later. The number of astrocytes within the borders of the scratch was counted. (B) The graph shows the number of migrated astrocytes per image from two different experiments, each with three separately transduced astrocyte cultures (six biologically different replicates) for every MMP at t = 24 h. (∗∗∗ p ≤ 0.001, three-way ANOVA).

Figure 5

(A) Schematic depiction of the OPoSSM assay. Porcine spinal slices were positioned on a Milliwell membrane and a circular agarose wall was placed over the slices, thereby forming an inner and an outer chamber on the membrane. Astrocytes were placed into the inner chamber. (B) Agar (borders indicated by the yellow hatched lines) was placed over the spinal slices (longitudinal borders indicated by white hatched lines). Astrocytes were placed in front of the cut sections of spinal slices and allowed to migrate for 1 week. (C) Fluorescence of EGFP-expressing astrocytes right after application of the cells onto the membrane. (D) After one week of incubation, migrating astrocytes formed interlacing chains (white and yellow hatched lines: see legend to B). (E) Slices were fixed and immunohistochemical staining for EGFP was performed in order to enhance astrocytic EGFP fluorescence within the slices. Confocal microscopy was performed to allow the visualization of astrocytic chains at different depths of the slices. Original magnification 200 ×.

Figure 6

Light microscopy and ultrastructural analysis of resin embedded material. Slices were fixed in glutaraldehyde and embedded in plastic resin. (A) Semithin sections revealed that control and MMP-transduced astrocytes migrated predominantly along the capillaries (arrows; toluidin blue; scale bar: 50 µm). (B) Electron microscopy revealed astrocytes predominantly between the basal lamina of capillaries and the myelin of the white matter. Astrocytes produced very long, thin processes (arrows; asterisk: pericyte; A: nucleus of the astrocyte; scale bar: 2.5 µm) and (C) formed tight junctions with the processes of other astrocytes (arrows; scale bar: 400 nm).

Figure 7

Migration of astrocytes as determined by the OPoSSM assay. Numbers of astrocyte chains in the spinal slices were counted every millimeter from the cut surface of the slices. The figure represents the results of two independent experiments with two different sets of transductions that were counted by four independent observers blinded to the genotype of astrocytes. (A) Number of astrocyte chains for every millimeter of infiltration depth. For better legibility, the standard error bars have been omitted. All MMP-transduced astrocytes showed increased infiltration into the spinal cord slices. (B) Total number of astrocyte chains per slice. ANOVA test showed significant differences between control astrocytes and astrocytes expressing MMP14, -17, and -25 (∗ p < 0.017, ∗∗∗ p ≤ 0.001).