Generation of matrix metalloproteinase MMP-2 deficiency in tumor and/or host cells and effects on tumor propagation. (A) (a) Gelatin zymography profiles of MMP-2 and MMP-9 activity in two independent wild-type glioblastoma multiforme (WT-GBM) orthotopic tumors. Protein extracts from mouse spleen were loaded alongside as controls for latent and active MMP-2 (72 and 62 kDa, respectively) and MMP-9. (b) Fluorescence-activated cell sorting was used to separate tumor cells (TC), CD31⁻, Gr1⁻ endothelial cells (EC), and CD45⁺ monocytic inflammatory cells (IC). The isolated cell populations were tested for MMP-2 transcription by reverse transcriptase (RT)-PCR. The ribosomal protein L19 was used as an internal control, and reactions were performed without RT as negative controls. (B) Schematic representation of MMP-2 knockout (MMP-2ko) GBM and WT-GBM cell line generation. Four distinct groups were analyzed and compared to each other: WT-GBM in WT mice; WT-GBM in MMP-2ko mice, MMP-2ko GBM in WT mice, and MMP-2ko GBM in MMP-2ko mice. (C) MMP-2 deficiency increases survival of GBM-bearing mice in a dose-dependent manner. (a) WT and MMP-2ko mice bearing an MMP-2ko GBM had significantly increased survival over mice with WT tumors, by 1.9-fold (p = 0.007) and 2.7-fold (p = 0.001), respectively. (b) Kaplan-Meier survival curves for WT mice with WT-GBM, and MMP-2ko mice with MMP-2ko GBM showed a significant difference in survival (p = 0.001). (c) Ki-67 or proliferative indices for four tumor/host combinations showed no difference among the four groups. (d) Apoptotic indices for the four groups exhibited a 2.4-fold increase in apoptosis when MMP-2 activity was completely abolished in the host and tumor cell compartment (p = 0.056). Error bars represent SEM.

The matrix metalloproteinase MMP-2 in glioblastoma multiforme (GBM) regulates tumor angiogenesis by affecting vascular branching. Blood vessels are visualized by fluorescein isothiocyanate–lectin perfusion and CD31 staining (green) of wild-type (WT)-GBM in WT host (A), WT-GBM in MMP-2 knockout (MMP-2ko) host (B), MMP-2ko GBM in WT host (C), and MMP-2ko GBM in MMP-2ko host (D). (E) Quantitative analysis of blood vessel density in GBM revealed a significant increase of blood vessels in both the MMP-2ko GBM in WT host group (p = 0.023) and the MMP-2ko GBM in MMP-2ko host group (p < 0.001). Error bars represent SD.

Matrix metalloproteinase MMP-2–deficient tumors undergo vascular remodeling but are dysfunctional and fail to activate pericytes. (A) Activated tumor blood vessels were detected by the presence of the vascular endothelial growth factor (VEGF)–vascular endothelial growth factor receptor 2 (VEGFR2) complex (brown) in wild-type glioblastoma multiforme (WT-GBM) in WT host (a) and MMP-2 knockout glioblastoma multiforme (MMP-2ko GBM) in MMP-2ko host (d). Both tumor/host groups expressed VEGFR2 (b,e). Activated, angiogenic pericytes on blood vessels were detected by RGS-5 transcription on tumor sections. RGS-5⁺ tumor pericytes were abundant in WT-GBMs (c) but were not present when MMP-2 activity was completely abolished in GBM (f). Scale bars = 83 μm. (B) Western blot analysis revealed that WT-GBM tumors and cell lines, as well as MMP-2ko tumors and cell lines, both produced equivalent levels of VEGF₁₆₄. Equivalent protein concentrations were loaded and α-tubulin was used as an additional protein loading control for tumor and cell lysates. (C) VEGFR2 RNA expression levels were reduced 4.8-fold in MMP-2ko GBMs compared to WT-GBMs, as determined by quantitative reverse transcriptase PCR. (D) Western blot analysis of tumor extracts revealed a reduction of VEGFR2 protein levels in MMP-2ko GBMs compared to WT-GBMs; α-tubulin was used as a protein loading control in tumor lysates.

Matrix metalloproteinase MMP-2 deficiency reduces pericyte coverage on glioblastoma multiforme (GBM) tumor vessels. Pericytes (red) were detected with desmin (A, C) and α-smooth muscle actin (α-SMA) antibodies (B, D). Blood vessels (green) were perfused with fluorescein isothiocyanate (FITC)-lectin and stained with FITC-CD31 antibody. Scale bars = 83 μm. (E, F) Quantification of the endothelial cell:pericyte ratio in GBM revealed a decrease in pericytes in the absence of MMP-2 (desmin, p = 0.031; α-SMA, p = 0.001). Error bars represent SEM. (G) Perfusion of tissues with FITC-lectin (green) before animal euthanization, followed by subsequent immunohistochemical staining for CD31 (red), suggests that the tumor vasculature in MMP-2 knockout (MMP-2ko) tumors is less functional than in wild-type (WT)-GBM tumors as represented by the relative lack of FITC-lectin staining in MMP-2ko GBMs. Vessels that co-stain for both lectin and CD31 are shown in yellow when merged (solid white arrowheads). Note the presence of lectin/CD31⁺ functional vessels in the surrounding normal brain of both WT-GBMs and MMP-2ko tumors (outlined arrowheads). The border between tumor (T) and normal brain (N) is depicted by the dotted line.

Matrix metalloproteinase MMP-2 deficiency increases perivascular tumor invasion in glioblastoma multiforme (GBM). (A and B) (a, b) Tumor cells in the brain parenchyma were detected by SV40 T-antigen staining (brown or red), and vessels were visualized with fluorescein isothiocyanate (FITC)-lectin and an FITC-CD31 antibody (green). (B) In the absence of MMP-2, tumor cells became less infiltrative but became more invasive by moving along blood vessels. Scale bars = 83 μm. (C) Infiltration by tumor cells was measured by the number of single infiltrating cells at the tumor edge that were not associated with a blood vessel (yellow arrows in A,b). Infiltration was decreased in MMP-2 knockout (MMP-2ko) tumors in both wild-type (WT) (p = 0.004) and MMP-2ko host (p = 0.015). (D) Perivascular invasion, defined as tumor cells associated with blood vessels (white arrowheads in B,b), was quantified on a scale from 1 to 3 where 1 = minimal, 2 = moderate, and 3 = extensive perivascular invasion. MMP-2ko GBM were more perivascular invasive than WT-GBM in WT hosts (p = 0.065) or MMP-2ko hosts (p = 0.030). Error bars represent SEM.