ADAM-9 is a novel mediator of tenascin-C-stimulated invasiveness of brain tumor–initiating cells

Abstract

Background

Tenascin-C (TNC), an extracellular matrix protein overexpressed in malignant gliomas, stimulates invasion of conventional glioma cell lines (U251, U87). However, there is a dearth of such information on glioma stemlike cells. Here, we have addressed whether and how TNC may regulate the invasiveness of brain tumor–initiating cells (BTICs) that give rise to glioma progenies.

Methods

Transwell inserts coated with extracellular matrix proteins were used to determine the role of TNC in BTIC invasion. Microarray analysis, lentiviral constructs, RNA interference-mediated knockdown, and activity assay ascertained the role of proteases in TNC-stimulated BTIC invasion in culture. Involvement of proteases was validated using orthotopic brain xenografts in mice.

Results

TNC stimulated BTIC invasiveness in a metalloproteinase-dependent manner. A global gene expression screen identified the metalloproteinase ADAM-9 as a potential regulator of TNC-stimulated BTIC invasiveness, and this was corroborated by an increase of ADAM-9 protein in 4 glioma patient–derived BTIC lines. Notably, RNA interference to ADAM-9, as well as inhibition of mitogen-activated protein kinase 8 (c-Jun NH2-terminal kinase), attenuated TNC-stimulated ADAM-9 expression, proteolytic activity, and BTIC invasiveness. The relevance of ADAM-9 to tumor invasiveness was validated using resected human glioblastoma specimens and orthotopic xenografts where elevation of ADAM-9 and TNC expression was prominent at the invasive front of the tumor.

Conclusions

This study has identified TNC as a promoter of the invasiveness of BTICs through a mechanism involving ADAM-9 proteolysis via the c-Jun NH2-terminal kinase pathway.

Malignant gliomas are intracranial tumors that account for 3%–8% of all cancer-related death. Despite several clinical advances, malignant gliomas remain largely untreatable and median survival time for the high-grade glioblastoma does not exceed 15 months. A major reason for the poor prognosis is the invasive nature of gliomas within the CNS, thereby generating multiple foci of transformed growth. The neoplastic growth of malignant gliomas is thought to be maintained by a rare population of cells that proliferate and undergo self-renewal and that seed and generate more differentiated glioma progenies. These self-renewing transformed precursors have been variously referred to as glioma stem cells or brain tumor–initiating cells (BTICs),^1–3 and are thought to be important in mediating treatment resistance and recurrence. Notably, as few as 10 BTICs deposited into the striatum of mice are sufficient to form intracranial tumors.⁴ BTICs are highly invasive cells, and the multiple foci of new transformed growth is a major factor for the poor prognosis of patients with high-grade gliomas. Thus, mechanisms that regulate invasiveness are reasonable targets for improved therapeutics.

Glioma invasiveness is mediated in part by the interaction of glioma cells with the extracellular matrix (ECM), followed by the proteolytic cleavage of ECM by tumor cell–derived proteases.^5–7 The family of matrix metalloproteinases (MMPs) is implicated in glioma invasiveness, with high expression of several MMP members in resected glioma specimens inversely correlating with the survival of patients. Further, inhibitors of MMPs reduce the invasiveness of glioma cell lines in vitro and attenuate the growth of glioma xenografts in mice.^4–7 While MMP members are implicated in glioma invasiveness, the role of the related a disintegrin and metalloproteinase (ADAM) family of metzincin metalloproteinases in invasiveness of differentiated glioma cells has only recently emerged.^8–10 Currently, there is only one report¹¹ of ADAMs (ADAM-10 and -17) in BTIC biology, principally in regulating their self-renewal.

The microenvironment of tumors includes the ECM around cells. For most tumor types, the availability of ECM proteins is a critical factor for growth^12,13 because they bind and/or activate integrins on the cell surface to trigger survival and proliferative signaling, and because the ECM is a rich source of growth factors. The ECM in malignant gliomas includes vitronectin, proteoglycans, collagens I and IV, osteopontin, and tenascin-C (TNC).^14,15 Of these, perhaps the most prominent glioma ECM component is TNC, with expression correlated with glioma grades.^16,17 Further, antibodies that block TNC reduce the invasiveness of differentiated glioma cells in culture¹⁸ and suppress the growth of gliomas in mice.¹⁹ These preclinical results have evolved to clinical trials of glioma patients using locally introduced small interfering (si)RNA to TNC,²⁰ or iodine-131–labeled anti-TNC antibody,²¹ where the survival results have been encouraging. We reported that TNC is a permissive substrate for the invasiveness of conventional glioma cell lines (U251 or U87) in culture, through regulating MMP-12²² and protein kinase C.²³ Herein, we have addressed the impact and mechanisms of TNC on invasiveness of BTICs.

Materials and Methods

Culture of Human Brain Tumor–Initiating Cells

BTICs were isolated from resected specimens of patients with glioblastoma.^4,24 We used 7 lines, designated BT012, BT025, BT048, and BT069—which were referred to previously as 12EF, 25EF, 48EF, and 69EF, respectively²⁴—and BT143, BT147, and BT157. Lines BT025 and BT048 were employed in many experiments in the current study, since they were characterized extensively in our previous study.²⁴ To propagate the lines, BTICs were dissociated and plated into T25 flasks at regular intervals and grown in serum-free culture medium supplemented with epidermal growth factor and fibroblast growth factor 2 in 5% CO₂ as described elsewhere.^4,25 The lines with higher passage numbers were checked for stemness property periodically and by confirming their self-renewal property (data not shown).

Brain Tumor–Initiating Cell Invasion Assay

Invasion by BTICs was performed using transwell migration chambers (Costar 3422, polycarbonate membrane in 24-well format), with 8 μm pore size. A collagen gel solution was prepared on ice^22,26 and supplemented with 10 μg/mL TNC when the latter was tested. BTICs were suspended in the gel solution and plated onto a transwell insert (2 × 10⁵ cells/70 μL/insert) to polymerize at 37°C for 1 h. Following polymerization, 100 μL of Dulbecco's modified Eagle's medium (DMEM)/F-12 with N2 supplement was added to the upper chamber and 1 mL of BTIC medium supplemented with 10% fetal bovine serum was applied to the lower well. Cells were then allowed to invade out of the 3D matrix of collagen I, across the membrane, at 37°C for 48 h. Noninvasive cells were then removed from the top compartment of the transwell with a cotton swab and the invasive cells present on the underside of the membrane were fixed and stained with hematoxylin. The number of invasive cells was counted per field from 4 random fields of each membrane. In some experiments, the collagen gel was supplemented with metalloproteinase inhibitors BB94 (500 nM; British Biotech) or GM6001 (10 μM; Calbiochem), with the transforming growth factor–β (TGF-β) inhibitor SD208 (1 μM; Sigma-Aldrich), or with both. Inhibitors were added to the gel 1 h before the addition of stimulators, such as TNC.

Microarray and Bioinformatics

To determine the effect of TNC on BTIC gene expression, BT048 cells were seeded in 12-well plates in serum-free BTIC medium (100 000 cells/mL of medium) and were treated with 10 μg/mL TNC for 6 h. Control or TNC-exposed cells were pooled to obtain 1 million cells from which RNA was extracted and subjected to microarray analysis as described elsewhere.²⁴

Western Blot Analyses of Cell Lysates

BTICs were treated with or without 10 μg/mL TNC for 6 h. In some wells, cells were exposed to neutralizing antibodies to TNC receptors (rabbit anti-α2β1 or -α9β1 antibodies; Abcam) for 1 h prior to TNC treatment. Cell lysates were prepared using lysis buffer on ice, and total protein was measured. Equal amounts of proteins were electrophoresed on 3%–8% Tris acetate gel under reducing conditions and transferred to a polyvinylidene difluoride membrane. The latter was then blocked overnight with 5% milk and probed with rabbit anti-human TNC antibody (1:1000; Novus Biologicals), rabbit anti-human ADAM-9 antibody (1:1000; Abcam), phosphorylated stress-activated protein kinase (SAPK)/c-Jun NH2-terminal kinase (JNK) (T183/Y185) rabbit antibody (1:1000; Cell Signaling), or total SAPK/JNK (5668) antibody (1:1000; Cell Signaling). A secondary antibody (anti-rabbit horseradish peroxidase; 1:10 000) was added for 1 h, and signals were detected by an enhanced chemiluminescence detection kit (Amersham Bioscience). Resected samples of glioblastoma multiforme (GBM) were processed as described elsewhere²² and probed for TNC and ADAM-9 as mentioned.

Glioblastoma specimens and nontumor control human specimens were also used for western blot analyses of ADAM-9 and TNC levels. These specimens were collected and described in previous work.²²

Lentiviral-Mediated Tenascin-C Knockdown in Brain Tumor–Initiating Cells

Stable knockdown of TNC was established using TNC short hairpin (sh)RNAs (5′-GTA CCG GAG GCT ACT GAA TAC GAA ATT GCT CGA GCA ATT TCG TAT TCA GTA GCC TTT TTT TG-3′ and 5′-CCG GCC AGT GAC AAC ATC GCA ATA GCT CGA GCT ATT GCG ATG TTG TCA CTG GTT TTT G-3′) or Scrambled controls in pLKO.1-puro vector. Briefly, the shRNA vector was cotransfected using Lipofectamine 2000 (Invitrogen) into human embryonic kidney 293 cells with pMD2.G (VSV.G env) and pCMV-ΔR8.91. After a 12-h transfection in DMEM, viral containing media were collected over 2 days in BTIC media. Collected media were then syringe filtered with a 0.22-μm filter and then ultracentrifuged at 26 000 rpm for 90 min at 4°C. Viral pellets were resuspended in BTIC media and added to cultures overnight. One microgram per milliliter of puromycin (Invitrogen) was added to cultures 3 days after infection.

RNA Interference Approach to Knock Down ADAM-9 and JNK Genes in BTICs

Two silencer select predesigned siRNAs (21 oligonucleotides in length) (Life Technologies) were used to target human ADAM-9 or JNK. The annealed siRNAs were analyzed by nondenaturing polyacrylamide gel electrophoresis. A negative control siRNA, composed of a 19-bp scrambled sequence with 3 deoxythymidine overhangs, was used. The sequences have no significant homology to any known gene sequences from mouse, rat, or human. For transfection with siRNAs, BTICs were plated in 12-well plates and were incubated with 20 nM siRNA and Lipofectamine (Invitrogen). After 24 h, cells were harvested for western blotting and invasion assay as described above, with the exception that fewer cells were used for economy reasons.

Animals and Implantation of Brain Tumor–Initiating Cells

Spheres from BTIC lines were dissociated into single-cell suspensions using Accumax solution. Cells were washed with serum-free culture medium. Ten thousand viable cells of BT025 or BT048 were resuspended in 2 µL saline and stereotactically implanted into the right striatum of 6- to 8-week-old female nonobese diabetic severe combined immunodeficient (NOD-SCID) mice (Charles River) as described elsewhere.²⁴ Animals were returned to their cages and allowed free access to food and water. Mice were weighed every other day and observed for symptoms of neurological deficits; they were sacrificed 7 weeks after implantation while still asymptomatic.

Other animals were implanted with BT143 cells, where mice were killed at day 77 post-implant, BT147 at day 46, and BT157 at day 125. These intervals were at time periods when mice were demonstrating outcomes from the implanted tumor (significant loss of weight, lethargy). The whole brain was removed, cut into blocks, fixed in 10% buffered formalin, and embedded in paraffin. Sections of 6 μm were taken every 120 μm apart, through the entire brain. The sections were deparaffinized, rehydrated, and stained with hematoxylin and eosin. All protocols were approved by the Animal Care Committee at the University of Calgary in accordance with research guidelines from the Canadian Council for Animal Care.

Immunohistochemical Analyses of ADAM-9 and Tenascin-C in Intracranial Tumor

For immunostainings, deparaffinized sections were subjected to endogenous peroxidase inactivation using 1% H₂O₂ in methanol. The sections were incubated with 4% horse serum to block nonspecific binding, then incubated with rabbit anti-TNC (1:1000; Novus Biologicals) or rabbit anti–ADAM-9 (1:500; Abcam) overnight at 4°C, followed successively by the biotinylated secondary antibody, avidin-biotin complex reagent (Vectastain ABC kit, Vector Laboratories), and diaminobenzidine. The slides were then lightly counterstained with hematoxylin, dehydrated, and mounted.

Proteolytic Activity of ADAM-9

Cell lysates were incubated with the fluorogenic peptide III (R&D Systems), containing the ADAM-9 cleavage sequence A-V (Mca-PLAQA2V-Dpa-RSSSR-NH2). Lysates (10 μL) were suspended in 100 μL of the final activity buffer (25 mM Tris/HCl, pH 8.0); peptide III was diluted in the activity buffer to the final concentration of 10 μM. Lysates were mixed with the substrate in a final volume of 100 μL at room temperature to initiate reaction. In some wells, BB94 (100 nM) was premixed with lysates at 4°C for 15 min. The mixtures were immediately transferred to a 96-well black plate and read in a microplate fluorescence reader (320 nm excitation and 405 nm emission wavelength) for 1 h. Background control (lysis buffer and substrate) was subtracted from sample measurements for calculations.

Statistical Analyses

The 1-way ANOVA with post-hoc Tukey comparisons was used for multiple groups, while the t-test was used for comparisons of 2 groups.

Results

Tenascin-C Stimulates BTIC Invasion and Requires Metalloproteinase Activity

The BTIC cells used for this study were obtained from samples resected from GBM patients, and their stemlike origin has been characterized elsewhere.^4,24,25 To test the importance of ECM components in invasion, we used a 3D model in which BTICs were encased within a collagen gel and their invasion through a transwell insert was determined 48 h later.²² Compared with control, TNC enhanced the invasiveness of both BT025 and BT048 cells (Fig. 1A and B). Further, we found that this TNC-stimulated BTIC invasiveness involved metalloproteinases and TGF-β expression, as BB94 or GM6001 (metalloproteinase inhibitors), or the TGF-β–specific inhibitor SD208, alone or in combination, reduced the TNC-stimulated BTIC invasion (Fig. 1A and B and Supplementary Fig. S1).

Fig. 1.

TNC stimulates BTIC invasiveness, and the process is metalloproteinase dependent. (A) When BT025 and BT048 cells were invaded across a 3D matrix of collagen I gel supplemented with TNC, their invasiveness was increased compared with collagen I control. (B) The metalloproteinase inhibitor BB94 (500 nM) or GM6001 (10 μM) attenuated TNC-stimulated BTIC invasiveness, as depicted (for BT048); ***P < .001 with ANOVA comparison with TNC (n = 4 for all groups). (C) Lentiviral transfection of small hairpin loop RNA (shRNAs) construct was used to knock down TNC in 2 BTIC lines, with confirmation by western blot of lysates of shTNC1 and shTNC2 clones compared with shControl. (D) When shRNA-transfected cells were allowed to invade through 3D matrix of collagen I gel, invasiveness of both lines were attenuated; this was rescued by recombinant TNC; ***P < .001 with ANOVA comparison with control shRNA, n = 4 for all groups. ^aSignificantly different from control shRNA and their respective shTNC-treated groups. Error bars represent SEM.

Since TNC can be produced by glioma cells at the advancing edge of the tumor,^16,18 we determined whether invasiveness of BTICs in vitro is regulated by autocrine signaling of TNC. We stably knocked down TNC in BT025 and BT048 cells using lentiviral constructs; Fig. 1C shows TNC protein level in 2 stable transfectants, shTNC1 and shTNC2, for each BTIC line. When shTNC1 and shTNC2 stably transfected cells were allowed to invade through the 3D matrix of collagen I, the basal invasiveness of both BTIC lines was prominently reduced compared with control shRNA–treated cells (Fig. 1D). Notably, addition of exogenous TNC restored shTNC1- and shTNC2-attenuated BTIC invasiveness (Fig. 1D).

ADAM-9 Expression Is Increased by Tenascin-C in Brain Tumor–Initiating Cells

We next sought to identify whether MMPs or ADAM members were involved in TNC-stimulated BTIC invasion. BT048 cells were treated with TNC for 6 h and RNA was subjected to microarray analysis for 39 000 genes in order to determine the changes in gene expression following treatment. The microarray data identified 1292 genes in BT048 that were differentially expressed (fold change [FC] ≥1.5) at 6 h with TNC treatment compared with control (Fig. 2A and Supplementary Table S1). Analysis of the array data using gene-ontology criteria identified a cluster of genes that are involved in migration/invasion (Supplementary Fig. S2). Interestingly, the metalloproteinases ADAM-8 (FC = 1.5), ADAM-9 (FC = 2.3), and ADAM-10 (FC = 1.7) were amongst the most upregulated genes by TNC. We focused on ADAM-9, since our mining of the Repository of Molecular Brain Neoplasia Data (REMBRANDT; http://rembrandt.nci.nih.gov) also found a very high level of ADAM-9 (Fig. 2B), but not ADAM-8 or -10, in all tumor samples based on median expression intensity (Supplementary Fig. S3). Moreover, ADAM-9 expression was negatively correlated with survival of all gliomas (Fig. 2C). A recent report implicates ADAM-8 in glioma invasion,²⁷ but the role of ADAM-9 in glioma invasion is unknown. Interrogation of other brain datasets available through the Oncomine (Compendia Biosciences, http://www.oncomine.org/) database further corroborated elevated levels of ADAM-9 expression in glioma specimens (Supplementary Fig. S4).

Fig. 2.

ADAM-9 is upregulated by TNC in GBM patient–derived BTICs. (A) Hierarchical clustering of differentially expressed genes in BT048 exposed to TNC for 6 h, compared with control, as detected by microarray; the adjacent segment of the heat map highlights several genes elevated by TNC. (B) Mining of the microarray data publicly available from caintegrator.nci.nih.gov/rembrandt revealed that glioma patients with higher ADAM-9 expression have reduced survival compared with (C) patients with intermediate ADAM-9 levels. (D) ADAM-9 protein is elevated in 4 BTIC lines after 6 h of TNC treatment. (E) Addition of neutralizing antibodies to TNC receptors α9β1 and α2β1 attenuated TNC-stimulated ADAM-9 protein in BT048 cells.

The potential role of ADAM-9 in TNC-mediated BTIC invasion was examined further. We found that treatment with TNC elevated ADAM-9 protein in BT025, BT048, and 2 more GBM patient–derived BTICs (Fig. 2D). The conditioned media of BT048 and BT025 lines contained TNC, which suggests that it has the potential to stimulate BTICs in an autocrine manner (Supplementary Fig. S5).

Further, as TNC exerts its effect upon binding integrins on the cell surface, mainly α2β1 and α9β1,²⁸ we tested and found that the addition of neutralizing antibodies to α2β1 and α9β1 integrins reduced TNC-elevated ADAM-9 protein levels in BT048 (Fig. 2E).

While TNC promoted the expression of ADAM-9 in BTICs (Fig. 2), this stimulatory effect of TNC was not observed in the U251 differentiated/conventional glioma cell line (Supplementary Fig. S6), suggesting specificity of TNC-elevated ADAM-9 in stemlike glioma cells only.

GBM Specimens Have Elevated Levels of ADAM-9 and TNC

We sought to confirm the results of the data mining by determining ADAM-9 and TNC expression in resected glioma specimens. Figure 3A shows an elevated level of ADAM-9 expression in the majority of GBM specimens compared with nontumor controls. Notably, GBM specimens with elevated ADAM-9 expression also had very high level of TNC expression, as examined by western blot (Fig. 3A) or immunoreactivity of GBM tissue sections (Fig. 3B). Moreover, immunohistochemical analysis of brain sections from human GBM patients showed an elevated level of ADAM-9 expression compared with no-tumor control sections, and ADAM-9 expression was more evident at the edge (presumed invasive front) of the tumor than in the tumor mass (Fig. 3C and D).

Fig. 3.

ADAM-9 and TNC expression in tumor specimens from GBM patients. (A) Tumor samples obtained from human GBM patients were subjected to western blot analysis to detect ADAM-9 protein level. Elevated level of ADAM-9 was detected in the majority of the GBM patients compared with no-tumor control samples. Notably, we also found an elevated level of TNC protein in the same GBM samples. (B) TNC expression in GBM specimen (right) compared with control brain tissue (left). Scale bars represent 50 μm for both. (C) While ADAM-9 immunoreactivity was not apparent in brain samples from nontumor (NT) control subjects, increased ADAM-9 expression (brown) was readily detected in GBM tumor mass (TM). ADAM-9 expression was prominent at the edge of tumor (bottom left panel) in cellular profiles (bottom right). Scale bars represent 50 μm. (D) The intensity of ADAM-9 immunoreactivity as determined by immunohistochemistry (IHC) was quantified and was more pronounced at the edge (presumed invasive front) of the tumor than within the tumor mass, and both areas had increased levels compared with no-tumor controls; **P < .01, ***P < .001, ANOVA, n = 3 for all groups. Error bars represent SEM.

Knockdown of ADAM-9 and Inhibition of MAPK8 (JNK) Signaling Attenuates TNC- Stimulated BTIC Invasiveness

To ascertain that TNC stimulated BTIC invasiveness through ADAM-9, we knocked down ADAM-9 in 2 BTIC lines using RNA interference. Figure 4A shows effective knockdown (>80%) of ADAM-9 gene expression in both BT025 and BT048 lines, as observed with western blot analysis. We then allowed ADAM-9 siRNA-transfected cells to invade through the 3D matrix of collagen I supplemented with TNC. Remarkably, the invasive capacity of ADAM-9 knockdown cells in response to TNC was significantly attenuated compared with control siRNA-treated cells (Fig. 4B and Supplementary Fig. S7A), indicating that TNC-stimulated BTIC invasion involves ADAM-9.

Fig. 4.

TNC stimulates BTIC invasiveness through ADAM-9 via MAPK8 (JNK) signaling. (A) Two silencer select predesigned siRNAs were used to target ADAM-9 gene knockdown, and they reduced ADAM-9 protein in BT025 and BT048 as detected by western blot analysis. (B) When ADAM-9 siRNA transfected cells were allowed to invade through the 3D matrix of collagen I gel supplemented with TNC, the invasiveness of both BT025 and BT048 lines was significantly attenuated compared with control siRNA treated cells; ***P < .001, with unpaired t-test (n = 4). This was also observed with BT012 and BT069 lines (data not shown). (C) TNC increased the phosphorylation of JNK protein as well as of its protein content in BT025 and BT048 lines within 30 min of treatment. (D) Addition of JNK inhibitor (SP600125) reduced TNC-stimulated ADAM-9 expression. (E) TNC-stimulated BTIC invasiveness was strongly reduced by a specific inhibitor to JNK, SP600125, in both lines and was unaffected by the MAPK inhibitor PD98059; ***P < .001 with ANOVA comparison with TNC, n = 4 for all groups for both cell lines. (F) Analysis of cell lysates from TNC knockdown clones showed reduced ADAM-9 expression in BT048 line, which was restored with recombinant TNC. Error bars represent SEM.

JNK has been implicated in invasiveness of conventional glioma cell lines,^29,30 but its role in BTIC invasiveness is not known. Since the microarray data also showed elevation of mitogen-activated protein kinase 8 (MAPK8; JNK) and its target oncogene Fos with TNC treatment (Fig. 2A), we sought to determine involvement of this pathway in TNC-stimulated BTIC invasiveness. We found that TNC treatment elevated total and phosphorylated JNK protein expression in BT025 and BT048 lines (Fig. 4C). Notably, JNK inhibition attenuated TNC-stimulated ADAM-9 expression (Fig. 4D). Interestingly, an inhibitor of MAPK (p42/44) signaling, PD98059, only slightly altered BTIC invasiveness, whereas a specific inhibitor to JNK, SP600125 and siRNAs to JNK, significantly reduced TNC-stimulated BTIC invasiveness (Fig. 4E and Supplementary Fig. S7B); moreover, TNC stable knockdown cells showed reduced ADAM-9 expression (Fig. 4F). Collectively, these results suggest that JNK signaling is involved in TNC-stimulated ADAM-9 expression and BTIC invasiveness.

Mice Implanted With BTICs Have ADAM-9 Expression at the Invasive Front of the Tumor

To determine the role of ADAM-9 in BTIC invasiveness in vivo, BTIC cells were implanted into the right striatum of NOD-SCID mice. The tumor growth was monitored in live asymptomatic animals using T2-weighted magnetic resonance imaging.²⁴ We sacrificed a group of asymptomatic mice at 7 weeks post-implantation and found tumor mass in these mice. An elevated level of ADAM-9 expression was observed in brain sections from mice implanted with BT025 and BT048 (Fig. 5A and B). Impressively, ADAM-9 staining for BTICs was observed more prominently at the edge (presumed invasive front) of the tumor than within the tumor mass. Moreover, the cells positive for ADAM-9 at the invasive front of the tumor were positive for stemlike markers such as nestin, Sox2, and Musashi-1 (Supplementary Fig. S8).

Fig. 5.

ADAM-9 expression at the invasive front of the tumor in mice implanted intracranially with BTICs. (A, B) Elevated level of immunoreactivty for ADAM-9 was observed within tumor mass (TM) compared with areas within the same brain of no tumor (NT) and notably, the expression was prominent at the invasive front (edge) of the tumor as determined by quantitation of ADAM-9 immunoreactivity; *P < .05, **P < .01 with ANOVA comparison to NT, and ***P < .001 compared with tumor mass, n = 5 for BT048 and n = 4 for BT025. Error bars represent SEM. (C) Validation of ADAM-9 expression using different BTIC lines implanted into mice brain, showing elevated level of ADAM-9 immunoreactivity at the invasive front (edge) of the tumor. Scale bars represent 50 μm.

We further validated the expression of ADAM-9 in mice implanted with 3 other GBM patient–derived BTICs (BT143, BT147, and BT157) of divergent genetic backgrounds (Fig. 5C). At sacrifice, the BTICs had invaded from the injection site of one hemisphere along the corpus callosum to the opposite hemisphere (Supplementary Fig. S9). Importantly, corresponding ADAM-9 and TNC immunoreactivity were evident in adjacent brain sections of the mice implanted with these BTICs (Supplementary Fig. S10).

TNC Stimulates ADAM-9 Proteolytic Activity

We determined ADAM-9 proteolytic activity of BTIC cells using a fluorogenic substrate-degradation assay. Figure 6A shows that TNC treatment elevated ADAM-9 proteolytic activity in BT025 and BT048 cells, and this occurred in a concentration- and time-dependent manner (Supplementary Fig. S11), which was reduced in the presence of the protease inhibitor BB94. Impressively, TNC-stimulated proteolytic activity of ADAM-9 was substantially attenuated in ADAM-9 knockdown cells. The increased proteolytic activity of ADAM-9 was corroborated in GBM patient–derived tumor cell lysates (Fig. 6B).

Fig. 6.

TNC enhances proteolytic activity of ADAM-9. (A) ADAM-9 proteolytic activity of BT025 and BT048 cells was elevated with TNC treatment and this was attenuated in the presence of protease inhibitors and in ADAM-9 knocked down cells; **P < .01, ***P < .001 with ANOVA comparison with TNC, n = 4 for all groups. (B) ADAM-9 activity of GBM tumor samples; ***P < .001 with unpaired t-test, n = 4 for all groups. Error bars represent SEM. (C) Postulated mechanism of TNC-stimulated BTIC invasion.

Discussion

The invasiveness of malignant gliomas is an important reason for the tumor's poor prognosis; mechanisms that regulate invasiveness are thus reasonable targets for improved therapeutics. The motility of cells requires cell-intrinsic mechanisms such as the integration of signaling molecules and cell-extrinsic mechanisms and environmental cues. Integrins on the surface of cells engage ECM proteins, resulting in the formation of focal adhesion complexes at contact points containing aggregates of kinases and other signaling molecules that then help regulate the actin cytoskeleton. In concert, there is secretion of proteases that remodel the extracellular environment for cells to advance. Many families of proteases are implicated in the invasive process, and amongst these are the metalloproteinases MMPs and ADAMs.^5–7 The role of metalloproteinases in tumor invasiveness is attributed partly to ECM degradation and movement of cells, but other mechanisms also apply as metalloproteinases have a broad range of substrates, including cytokines, chemokines, growth factors and their receptors, and adhesion molecules.⁵ ADAMs have been found to cleave and remodel components of the ECM in this context.^31,32 Their best characterized function is the proteolytic processing of membrane-anchored precursors and the subsequent release of mature proteins. This process is regarded as “protein ectodomain shedding,” and it subsequently alters the activity of the substrate. The ADAMs are membrane-bound enzymes with more domain modules than MMPs and have been implicated in tumor growth and invasiveness of different cancers,³² including gliomas.^11,33–35 However, the roles of the ADAM family of metalloproteineases in BTIC invasiveness are for the most part still unclear.¹¹

In the current study, we show that in a 3D matrix of collagen I, TNC stimulates the invasiveness of BTICs through elaboration of ADAM-9, which involves the JNK signaling pathway. Consistent with this observation, we detected an elevated level of ADAM-9 gene expression using a microarray screen that we then validated at the protein level in 4 different BTIC lines and in situ. Moreover, mice implanted with BTICs showed elevated levels of ADAM-9 and TNC expression prominently at the invasive front of the tumor (Supplementary Fig. S10). This is in agreement with the elevated level of ADAM-9 expression in resected human glioma specimens that corroborates the REMBRANDT dataset. Our inability to detect ADAM-9 in normal brain by immunohistochemistry, in contrast to transcript or proteolytic expression, is likely related to the lower sensitivity of immunohistochemistry compared with PCR or the ADAM-9 proteolysis assay. We note that the purported ADAM-9 substrate from the manufacturer may also be cleaved by related proteases such as ADAM-10 and ADAM-17; however, that the knockdown of ADAM-9 by siRNA abolishes the detected proteolytic activity (Fig. 6A) supports the predominance of ADAM-9 over related proteases. Overall, these results identified a new molecule, ADAM-9, in the regulation of BTIC invasiveness in response to TNC.

A recent report shows that a soluble form of ADAM-9 can promote carcinoma cell invasion³⁶; thus, it will be tempting to speculate and determine in future whether or not TNC may also upregulate the soluble form of ADAM-9 for BTIC invasiveness.

Several studies have implicated ADAM family members in malignant gliomas. ADAM-10 and ADAM-17 expressed on the cell surface of glioma-initiating cells have been shown to produce an immunosuppressive phenotype.³⁷ Moreover, ADAM-17 is involved in TGF-β1–stimulated glioma cell motility and invasiveness,³⁸ while ADAM-10 can promote glioblastoma cell migration through a protein kinase C–dependent N-cadherin cleavage.³⁴ ADAM-8 and ADAM-19 have been linked to glioma invasiveness.²⁷ It should be noted that we have not excluded other metalloproteinases from regulating the TNC-promoted invasiveness of BTICs. Indeed, ADAM-8, similar to ADAM-9, shows a relationship of higher expression with lower survival when considering all glioma cases (Supplementary Fig. S3), although this relationship was marginal in statistical significance (P = .045 compared with P = 8E-10 for ADAM-9; Fig. 2). For ADAM-10, although the relationship between high expression level and reduced survival for all gliomas had a P-value for the log-rank test of .508, a subpopulation of patients with high ADAM-10 expression died early (Supplementary Fig. S3). Thus, our results do not exclude the involvement of other proteases in TNC-mediated invasiveness, but they do extend the roles of ADAMs in gliomas by implicating a previously unidentified ADAM, ADAM-9, in glioma invasiveness downstream of TNC. Significantly, ADAM-9 is a negative prognostic factor for patients with gliomas.

In conclusion, our results have implicated TNC in increasing the invasiveness of BTICs. In this context it would be important to investigate not only the cellular source of TNC, but also cells that have binding sites for TNC. Our study indicates differential mechanisms of TNC on invasiveness of BTICs versus differentiated cells. Our future study will seek to understand the causal relationship between TNC and ADAM-9 expression in the context of BTIC invasiveness in vivo through a perturbation experiment. Overall, our data lead us to conclude that TNC promotes the invasiveness of BTICs through a JNK pathway mediated by a new factor in glioma pathophysiology, ADAM-9.

Supplementary Material

Supplementary material is available at Neuro-Oncology Journal online (http://neuro-oncology.oxfordjournals.org/).

Funding

This work was supported by Canadian Institutes of Health Research and Alberta Innovates–Health Solutions/Alberta Cancer Foundation operating grants.