OUP user menu

Metalloproteinase Disintegrins ADAM8 and ADAM19 Are Highly Regulated in Human Primary Brain Tumors and their Expression Levels and Activities Are Associated with Invasiveness

Dirk Wildeboer MSc, Silvia Naus PhD, Qing-Xiang Amy Sang PhD, Jörg W. Bartsch PhD, Axel Pagenstecher MD
DOI: http://dx.doi.org/10.1097/01.jnen.0000229240.51490.d3 516-527 First published online: 1 May 2006


Patients with primary brain tumors have bleak prognoses and there is an urgent desire to identify new markers for sensitive diagnosis and new therapeutic targets for effective treatment. A family of proteins, the disintegrin and metalloproteinases (ADAMs or adamalysins), are cell surface and extracellular multidomain proteins implicated in cell-cell signaling, cell adhesion, and cell migration. Their putative biological and pathological roles make them candidates for promoting tumor growth and malignancy. We investigated the expression levels of 12 cerebrally expressed ADAM genes in human primary brain tumors (astrocytoma WHO grade I-III, glioblastoma WHO grade IV, oligoastrocytoma WHO grade II and III, oligodendroglioma WHO grade II and III, ependymoma WHO grade II and III, and primitive neuroectodermal tumor WHO grade IV) using real-time PCR. The mRNAs of the five ADAMs 8, 12, 15, 17, and 19 were significantly upregulated. The ADAM8 and ADAM19 proteins were mainly located in tumor cells and in some tumors in endothelia of blood vessels. In brain tumor tissue, ADAM8 and ADAM19 undergo activation by prodomain removal resulting in active proteases. By using specific peptide substrates for ADAM8 and ADAM19, respectively, we demonstrated that the proteases exert enhanced proteolytic activity in those tumor specimens with the highest expression levels. In addition, expression levels and the protease activities of ADAM8 and ADAM19 correlated with invasive activity of glioma cells, indicating that ADAM8 and ADAM19 may play a significant role in tumor invasion that may be detrimental to patients survival.

Key Words
  • ADAMs
  • Brain neoplasms
  • Invasive gliomas
  • Matrix metalloproteinases


Despite the progress of neurosurgery, radiation therapy, and chemotherapy over the last century, gliomas, the most common primary brain tumors, still bear a grim prognosis. This fact is predominantly due to the diffuse infiltrative growth of gliomas along white matter tracts that regularly prevents complete resection of the tumor. Most low-grade gliomas develop, over time, into malignant tumors presenting the histological hallmarks cell anaplasia, neovascularization (angiogenic switch), and finally necrosis (1). Moreover, patients suffering from glioma show tumor-induced immunosuppression (2).

Matrix metalloproteinases (MMPs), a family of Zn-binding peptidases, have been implicated in the biological behavior of primary brain tumors and several studies from other and our own group demonstrated the expression of a variety of these genes and their inhibitors in primary brain tumors (3-10). A number of studies subsequently demonstrated that MMPs may mediate the potential of cells to infiltrate along white matter tracts (11), and that MMPs confer to the angiogenic switch of gliomas and other malignancies that develop neovascularization (4, 12, 13). The importance of MMP and TIMP gene expression is further underlined by studies demonstrating that synthetic MMP inhibitors limit tumor growth in a glioma tumor model in vivo (14-16) and in vitro (17-19). The MMPs and the membrane bound MT-MMPs in particular share a number of features with another family of proteases termed ADAMs (i.e. disintegrin and metalloprotease) (20). The common functions of MMPs and ADAMs cause difficulties in the differentiation of MMP and ADAM effects (21). The TNF-α-converting enzyme (TACE, ADAM17), for example, was originally believed to be a member of the MMP family (see below). There are currently more than 30 known ADAMs and a large portion of these are proteolytically active. About half of the ADAMs are testis specifically expressed while the others are expressed in many organs (22). The ADAM proteins have a broad variety of functions reaching from spermatogenesis as well as sperm-egg fusion (ADAMs 1, 2) to muscle development (ADAMs 12, 19) and neurogenesis (ADAM10). Moreover, several ADAMs act as potent sheddases that can degrade membrane bound ligands and receptors such as TNF-α (ADAM17 has been identified as the long searched for TNF-α-converting enzyme [TACE]), TGF-α, heparin binding-epidermal growth factor (HB-EGF), TNF-R1 and -R2 to a soluble form. Additionally, ADAMs can degrade extracellular matrix proteins such as collagens as well as constituents of cerebral white matter such as myelin basic protein (MBP) (22). As a result of these functions, ADAMs might play a role in the particular features of glioma behavior. However, at present the data on ADAM expression in gliomas is very limited and a role for ADAM-mediated proteolysis or cell-cell interactions in glioma biology remains to be defined (21). Recently, Kodama et al reported on the expression of various ADAM genes in different glioma species (23). In this publication the authors focused on the selective overexpression of ADAM12 in glioblastomas and its contribution to the shedding of HB-EGF, which could act as a growth factor in tumorigenesis.

In order to determine the expression pattern of ADAM genes and the proteolytic activities and biological effects of ADAMs in human primary brain tumors, we determined the expression of 12 ADAM genes by quantitative RT-PCR (LightCycler™ analysis). We then focussed on the detailed analysis of ADAM8 and ADAM19 protein in brain tumors by immunoblotting, immunohistochemistry, and activity assays. Finally, we assigned these proteins a function in invasive activities.

Our study revealed distinct patterns of ADAM gene expression in different primary brain tumors. Moreover, a strong proteolytic and invasive activity of both ADAM8 and ADAM19 suggest a significant role of these ADAMs for the biological behavior of primary brain tumors.

Materials and Methods

Tissue Samples

Brain tumor specimens were obtained from archival material of patients who underwent neurosurgery for tumor resection. Histologically normal temporal cortex with adjacent white matter served as control. This material was removed in the course of surgical hippocampectomy for therapy of medical refractory epilepsy. Following histological examination, RNA was extracted and tested for MMP expression by RNase protection assay. Samples that proved no alterations of MMP-expression as compared to normal brain were used in this study. The tissue specimens were dissected in two portions, one of which was immediately snap frozen in liquid nitrogen and stored at −80°C pending RNA isolation or protein extraction or preparation of cryostat sections. The remaining material was fixed in 4% formaldehyde and embedded in paraffin. The histological diagnosis was made according to the World Health Organization criteria for the histological typing of brain tumors. Nine specimens of pilocytic astrocytoma WHO grade I (AI), 8 low grade astrocytomas WHO grade II (AII), 8 anaplastic astrocytomas WHO grade III (AIII), 8 glioblastomas WHO grade IV (GBM), 6 oligoastrocytomas WHO grade II (OAII), 5 anaplastic oligoastrocytomas WHO grade III (OAIII), 11 oligodendrogliomas WHO grade II (OII), 9 anaplastic oligodendrogliomas WHO grade III (OIII), 10 ependymomas WHO grade II (EII), 7 anaplastic ependymomas WHO grade III (EIII) and 8 primitive neuroectodermal tumors WHO grade IV (PNET, 1 supratentorial and 7 infratentorial, i.e. medulloblastomas), and 8 normal brain samples were included in this study. Two of the patients bearing an anaplastic astrocytoma and an anaplastic oligodendroglioma, respectively, had been treated with chemotherapy and radiotherapy before removal of the specimens used in this study. These specimens were included in the study since these are rare tumors. Subsequent analysis revealed that the results of real-time PCR were in the same range as that of tumor samples that had not been treated before. This study was approved by the local ethics committee.

RT-PCR and Real-Time PCR

Total RNA was prepared with Trizol reagent (Gibco BRL, Grand Island, NY) according to the manufacturer's instructions. The RNA samples were stored at −80°C pending RT-PCR. Three μg of total RNA were transcribed with Expand Reverse Transcriptase™ (Roche, Mannheim, Germany) in a volume of 20 μl and afterwards diluted to 60 μl. Subsequently, 2 μl of the RT-product was used for PCR in a 20-μl-reaction volume. Table 1 shows the primer sequences that were used for RT-PCR. The amplification product for hL7 mRNA was used as a reference and the hADAM2 product served as a negative control since it is not expressed in brain. The PCR products were separated on 2% Seakem agarose gels. Following staining with ethidium bromide, gels were scanned and band densities determined using the QuantityOne (BioRad, Munich, Germany) software.

View this table:

Real-time PCR was performed to quantify the amounts of those mRNA species that showed the most robust regulation in this semiquantitative assay. Real-time PCR consisted of a 20-μl reaction with QuantiTect™ SYBR Green PCR Kit (Qiagen, Hilden, Germany) and between 2 μl undiluted and 1 μl of a 1:10 dilution of the RT product was used as template. Table 2 shows the primer sequences that were used for quantitative RT-PCR. The LightCycler™ was programmed as described in Table 3. Each sample was measured in duplicate in three independent experiments. Differences between the tumor groups and normal brain controls were analyzed by Mann-Whitney U test. In order to verify the identity of the PCR products, one representative sample of each tumor was cloned into pCRII-TOPO (Invitrogen, Groningen, Netherlands) and sequenced (IIT Biotech, Bielefeld, Germany). The identity of all products was confirmed by their size and fragments of restriction digests.

View this table:
View this table:


Tissue specimens (at least three different specimens of each tumor entity and normal brain) were homogenized (1/10, w/v) on ice in Tris buffer (pH 7.6) containing 1% Nonidet P-40, 5 μg/ml Pepstatin A, and a protease inhibitor cocktail (Complete™, Roche). Protein concentration was determined using Bradford's reagent. The protein lysates were stored at −80°C pending immunoblotting. Immunoblot analysis was performed as described previously with modifications (9). Briefly, 50 μg of each protein lysate were electrophoretically fractionated on 12% Tris/glycine gels at 160 V. Following electrophoresis, the samples were transferred to a PVDF membrane (Immobilon, Millipore, Bedford, MA) and blocked with 5% skim milk powder in TBS. The membranes were incubated with polyclonal antibodies against ADAM8 cytoplasmic domain or ADAM19 (both from Chemicon, Temecula, CA). A subsequent incubation with an antibody against GAPDH (Chemicon) was performed to determine equal protein loading of each lane. Band density was determined with the NIH Image 1.62 software and each value was normalized to the band density of the respective GAPDH band.


Immunohistochemistry was performed on sections of paraffin-embedded material as described previously (9) with antibodies against ADAM8 (Triple Point Biologics, Forest Grove, OR) and against the prodomain, the metalloprotease domain, and disintegrin domain-specific peptides of ADAM19 (24). The Vectastain ABC kit (Vector Laboratories, Burlingame, CA) was used according to the manufacturer's protocol to visualize antibody-binding sites.

Peptide Cleavage Analysis

Human brain tissue extracts were lysed with pestles in 10 μl HEPES buffer (20 mM HEPES pH 7.4, 1% NP-40, 0.5% sodium deoxycholate) per mg fresh weight. Samples were spun and supernatants were purified with ConA sepharose (Amersham Biosciences, Freiburg, Germany) to enrich glycosylated ADAM proteins in the tissue extracts; for each assay 20 μl ConA sepharose was mixed with 50 μl sample. After incubation for 2 hours at 4°C, ConA sepharose pellets were washed twice with HEPES, pH 7.4 containing 0.015 % Brij-35 and subsequently used for activity assays.

For activity assays, 20 μl ConA sepharose pellets were mixed with 100 μl reaction batches containing 1 × Complete™ EDTA-free Inhibitor Cocktail (Roche), or metalloprotease inhibitors as indicated (EDTA 10 mM, TIMP1 40 nM, and TIMP3 40 nM, respectively), and 20 mM HEPES, pH 7.4 with 0.015 % Brij-35. After preincubation with inhibitors for 1.5 hours at 37°C, 500 μM CD23 peptide (DNP-SHHGDQMAQKSQSTQI) or 1 mM CRDA19 peptide (Ac-RPLESNAV) were added.

At time point 0 and after 1, 2 and 3 hours incubation time at 37°C, 20 μl of the reaction batch were removed and mixed with 75 μl 20 mM HEPES, pH 7.4 with 0.015 % Brij-35 and 5 μl 1% Fluorescamine (Sigma, Taufkirchen, Germany). Fluorescence was measured at excitation wavelength 355 nm and emission wavelength 460 nm using the fluorometer Twinkle LB970 (Berthold Technologies, Bad Wildbad, Germany).

Matrigel™ Invasion Assay

Rat glioma C6 cells were transfected with human and mouse ADAM8 (hA8, kind gift of Dr. S. Yamamoto; mA8 25) and human ADAM19 (hA19 26) expression plasmids, mock vector and catalytically inactive mutants of ADAM8 (EQmA8 25) and ADAM19 (EAhA19 26) as controls. 2 × 105 cells were seeded on 35-mm tissue culture dishes and were transfected two hours later using Lipofect-Amine/PLUS reagent (Invitrogen) according to the supplier's instructions. Twenty-four hours after transfection, cells were detached and resuspended to a density of 50,000 cells per ml in culture medium. Transfection efficiency was assessed by Western blotting and immunostaining and 500 μl of the suspension was filled into Matrigel™ invasion chambers (BD Biosciences, Heidelberg, Germany), which were handled according to the supplier's manual. After 22 hours incubation time, invaded cells were fixed with methanol, stained with toluidine blue, and counted. Differences between the ADAM- and mock-transfected cells were analyzed by Mann-Whitney U test.


ADAM Gene Expression in Primary Brain Tumors

In order to screen for ADAM genes regulated in human brain tumors the expression patterns of ADAM mRNAs in normal human brain and in primary tumor samples were analyzed by conventional RT-PCR (data not shown). Of 12 ADAM genes expressed that are not testis specific, eight mRNA species revealed altered expression in brain tumors, ADAM8, 9, 11, 12, 15, 17, 19, and ADAM28. ADAM genes that showed expression changes (i.e. up- or downregulation) of less than 2-fold or that were expressed at very low levels were not included in the quantitative study.

Real-time PCR was used to quantify mRNA levels of ADAM8, 12, 15, 17, and 19 with significantly affected expression in primary brain tumors. Primers were designed to amplify comparable PCR products with a length in the range of 121 to 154 bp (Table 2). Using the normal brain specimens, PCR reaction conditions were optimized for the different primers with regard to template dilution and number of cycles to obtain comparable second derivative maxima for analysis (Table 3). Considerable upregulation of ADAM8 gene expression was observed in the majority of tumor specimens leading to an increased mean of gene expression in all tumor entities (Fig. 1A). The highest average ADAM8 inductions were detected in glioblastoma multiforme (up to 75-fold, mean 28-fold) and in anaplastic ependymoma (up to 10-fold, mean 4-fold). The differences to normal brain were significant with p < 0.0001 (Mann-Whitney U test). Two samples of oligoastrocytoma grade II revealed very high mRNA expression for ADAM8 (up to 73-fold, median 5-fold).


Relative expression levels of ADAM mRNAs in primary brain tumors determined by LightCycler™ analysis. Values for ADAM8 (A), ADAM19 (B), ADAM12 (C), ADAM15 (D), and ADAM17 (E) were normalized relative to the expression levels of normal brain tissue (n = 1). Results are shown as mean values of three independent LightCycler™ runs in doublets with n tumor samples of each specimen (n values given below each figure). Thick lines represent the total mean values. For tumor specimens with outlier values, the median is shown (black handle). Note the logarithmic scale of the ordinate. NB, normal brain; AI, pilocytic astrocytoma; AII, low grade astrocytoma; AIII, anaplastic astrocytoma; GBM, glioblastoma; OAII, oligoastrocytoma; OAIII, oligoastrocytoma; OII, oligodendroglioma; OIII, anaplastic oligodendroglioma; EII, ependymoma; EIII, anaplastic ependymoma; PNET, primitive neuroectodermal tumor. Statistical significance of the differences between the tumor groups and normal brain samples were analyzed by Mann-Whitney U test (*, p < 0.01; **, p < 0.001; ***, p < 0.0001).

Compared to normal brain, the strongest upregulation of ADAM19 mRNA was observed in AIII (up to 90-fold, median 5-fold), GBM (up to 140-fold, mean 24-fold), EIII (up to 140-fold, median 3-fold) and in PNET (up to 150-fold, median 4-fold) (Fig. 1B). The differences in GBM samples were highly significant with p < 0.0001 but less significant with p < 0.01 in the other tumor specimens. Some specimens of astrocytoma grade II, oligoastrocytoma grade II, and oligodendroglioma grade III also showed mRNA induction of ADAM19, but at lower levels. ADAM12 gene expression was upregulated in many tumors (Fig. 1C). The most significant upregulations (p < 0.0001, Mann-Whitney U test) were observed in glioblastoma (up to 27-fold, mean 19-fold), and in anaplastic ependymoma (up to 19-fold, mean 9-fold). In anaplastic astrocytoma and low-grade oligoastrocytoma the inductions were more heterogeneous. In agreement with the findings of Kodama et al, glioblastoma was the only tumor that showed significant upregulation of this gene in all samples (Fig. 1C). Less pronounced changes of mRNA expression were observed for ADAM15 and ADAM17 (TACE) (Fig. 1D, E), with not more than 5-fold upregulation for ADAM15 in glioblastoma and 3-fold upregulation in anaplastic oligodendroglioma and anaplastic ependymoma. ADAM17 mRNA was more than 3-fold upregulated in AII, AIII, GBM, and OAII, and the strongest upregulation was revealed in anaplastic ependymoma (up to 5-fold). Although the mRNA levels were heterogeneous between different individual tumor samples, there was an overall correlation between expression levels of ADAM8/ADAM19 and tumor malignancy.

Immunoblotting for ADAM8 and ADAM19

Immunoblotting was performed to detect protein concentrations and the processing states of ADAM8 and ADAM19, respectively (Fig. 2). The antibody against ADAM8 revealed three bands of 90, 50, and 20 kDa molecular weight (Fig. 2A), representing the processed protein after prodomain removal, subsequent metalloprotease domain removal and further processing (25). In all normal brain samples, ADAM8 expression was very low, and only a band migrating with an apparent molecular weight of 50 kDa but not the 90-kDa band was detected. Significant increase of ADAM8 protein is observed for AI, AIII, GBM, OAII, OAIII, EII, EIII, and PNET. All these tumors revealed expression of the 90-kDa band representing the catalytic activity of ADAM8. Ependymomas showed the strongest intensity of this band (Fig. 2A, D). The degree of processing, however, was different in the tumor entities, judged by the relative proportions of the 90-kDa band and the bands migrating at 50 and 20 kDa (Fig. 2D). Interestingly, the processing of ADAM8 was significantly less in the ependymomas that showed the strongest expression of ADAM8. Among the three samples for each tumor entity, protein levels were heterogeneous, but showed a good overall correlation with the results obtained in the LightCycler™ experiments. Discrepancies were observed for AII, EII and PNET samples.


Western Blot analysis of ADAM8 (A) and ADAM19 (B) in brain tumor tissue extractions. Three representative samples of each tumor specimen were analyzed using specific antibodies against the cytoplasmic domains. A subsequent staining with an antibody against GAPDH was performed to determine equal protein loading (C). Immunoblotting was performed as described in Materials and Methods. (A) 90-kDa and 50-kDa bands represent ADAM8 with the pro- and metalloprotease domain subsequently removed. A further processed form migrates below 20 kDa. (B) The 110-kDa band represents pro-ADAM19 and prodomain removal is reflected by a 90-kDa band. The relative abundance of the bands in (A) and (B) is shown in (D) and (E), respectively. Abbreviations of the tumor specimens are identical to those in the legend to Figure 1.

ADAM19 immunoblotting resulted in two major bands of 110 kDa and 90 kDa (Fig. 2B), representing the latent and the active protease, respectively. A significant increase of ADAM19 protein was observed in at least one sample of each tumor type. Highest protein levels were observed in OAIII, OII, OIII, EIII, and PNET. In few samples, particularly in glioblastoma, smaller fragments of ∼65 kDa were detected (data not shown). These fragments were probably derived by autolytic processing and could account for higher catalytic activity of ADAM19 compared to the 90-kDa form generated by prodomain removal (26). In general, ADAM19 protein levels as judged by quantification of the 90-kDa band (Fig. 2E), reflected gene regulation as determined by LightCycler™, with the exceptions of OAIII and OII, that revealed high amounts of ADAM19 protein but only low mRNA induction compared to normal brain. The antibody against GAPDH was used to normalize for protein loading (Fig. 2C).

Cellular Localization of ADAM8 and ADAM19 Proteins

Immunohistochemistry was performed to characterize the cell types expressing ADAM8 (Fig. 3A-H) and ADAM19 (Fig. 3I-P). In normal brains, ADAM8 and ADAM19 were hardly detectable (Fig. 3A, I). In accordance with RNA analysis, ADAM8 immunoreactivity was detectable in all astrocytic tumors and mixed gliomas (AI, AII, AIII, GBM, OAII, and OAIII). In these tumors, ADAM8 immunoreactivity was mainly localized to gemistocytic astrocytes or to minigemistocytes in the oligodendroglial portion of mixed gliomas (Fig. 3). In glioblastoma and anaplastic oligodendroglioma, staining was additionally localized to foamy macrophages. In ependymomas the staining localized to the cytoplasm of tumor cells. The PNETs were the only tumors that revealed ADAM8 immunoreactivity in blood vessel proliferations (Fig. 3H). Interestingly, in oligodendrogliomas that infiltrated the cortex, ADAM8 immunoreactivity was observed as nuclear and perinuclear staining of neurons. In previous work, a similar staining pattern was observed in motoneurons undergoing degeneration, suggesting that neuronal ADAM8 is indicative for neurodegeneration (Fig. 3F) (27). Omission of the primary antibody resulted in no staining.


Immunolocalization of ADAM8 (A-H) and ADAM19 (I-P) in normal brain and in human primary brain tumors. One representative image was displayed for tumor specimens with significant mRNA/protein inductions. (F) Arrows show immunoreactivity in neurons; arrowheads show tumor cells negative for ADAM8. Arrowheads in (H) and (P) show vessel walls immunoreactive for ADAM8 and ADAM19, respectively. Immunohistochemistry was performed as described in Materials and Methods. Omission of either primary antibody resulted in no staining. Abbreviations of the tumor specimens are identical to those in the legend to Figure 1. Original magnifications: A: 50×; B-P: 100×.

Immunoreactivity for ADAM19 was found in gemistocytic tumor cells in most of the specimens investigated (Fig. 3K-O). The strongest ADAM19 immunoreactivity and the largest portion of labeled tumor cells were observed in anaplastic astrocytoma and glioblastoma (Fig. 3L-N). In glioblastoma, ADAM19 was observed in vital as well as in perinecrotic areas. Comparable to the staining for ADAM8 in PNETs we observed ADAM19 immunoreactivity in endothelia of blood vessel proliferations (Fig. 3P). Low-grade astrocytoma AII, OAII, and EII revealed virtually no immunoreactive cells. Omission of either primary antibody revealed no staining (data not shown). For ADAM8 and ADAM19, the observed staining patterns correlated closely to the mRNA levels determined by quantitative RT-PCR (Fig. 1A, B).

Detection of ADAM Activities in Primary Human Brain Tumors

In order to determine whether the increased expression of ADAM8 mRNA/protein and ADAM19 mRNA/protein correlated to increased proteolytic activity, a peptide cleavage assay was performed with tumor lysates after ConA-enrichment of ADAM protease activities (28). From each tumor entity, two samples representing the mean values in LightCycler™ analysis were chosen. For detection of ADAM8 activities we used a peptide based on the low affinity IgE receptor CD23 target sequence for which proteolytic cleavage has recently been described (29). Briefly, several peptides from the CD23 extracellular domain were screened for cleavage and the peptide with the sequence Dnp-SHHGDQMAQKSQSTQI was efficiently and specifically cleaved by soluble ADAM8 (30) and used to determine the catalytic activity of ADAM8. For this peptide we cannot rule out additional cleavage by ADAM28, but as the levels of ADAM28 expression were extremely low (mRNA as detected by RT-PCR) or not detectable (immunoblotting) in most of the tumor specimens (data not shown), differences in peptide cleavages were interpreted as changed ADAM8 activities. For ADAM19, we used the peptide substrate Ac-RPLESNAV (CRDA19), which represents the autocleavage site in the cysteine-rich domain of human ADAM19 (26). For both peptides we established the assay using ADAM8- and ADAM19-transfected C6 cells respectively, comparing the activity to untransfected cells (data not shown). All assays were performed using crude homogenates of the tumor tissues. To inhibit non-specific cleavage of the peptides, different inhibitors were used. All assays were performed in the presence of a protease inhibitor cocktail to inhibit serine and cysteine protease activities. EDTA was used as a metal chelator to determine metalloprotease activity. To distinguish ADAM activities from those of MMPs, a combination of recombinant human TIMP1 and human TIMP3 was used and preincubated with the respective tumor homogenates (Fig. 4). Cleavage of CD23 peptide was increased in all tested tumor tissue extractions compared to normal brain (Fig. 4A). Preincubation of the homogenates with the non-specific metalloprotease inhibitor EDTA strongly inhibited the peptide cleavage. In addition, preincubation of homogenates with recombinant TIMP1/3 did not affect peptide cleavage in extracts from AI, AIII, and GBM. In contrast, extracts from OAII tumors showed decreased cleavage, indicating the presence of another metalloprotease that can be inhibited by TIMP1/3.


Proteolytic activities of ADAM8 and ADAM19 in tumor extracts. Fluorometric analysis of protease activities for ADAM8 (A) and ADAM19 (B) in native protein extracts of two tumor tissue samples of specimen showing significant mRNA/protein inductions. Assays were performed with either CD23 peptide (ADAM8) or CRDA19 peptide (ADAM19) in the absence or presence of 10 mM EDTA or 40 nm TIMPs 1 and 3, respectively. Peptide cleavage was determined by the fluorescamine method (see Materials and Methods section for details) after 3-hour incubation time. Relative fluorescence values were normalized to the value of normal brain tissue extractions (n = 1,000). Two samples of each specimen were assayed in two independent experiments. Abbreviations of the tumor specimens are identical to those in the legend to Figure 1. Note that the slight increase of activity in some TIMP1/3-treated samples is within the range of the experimental error.

Cleavage results obtained with the CRDA19 peptide revealed maximum processing in anaplastic astrocytomas and glioblastomas and lower levels of activity in pilocytic astrocytoma and oligoastrocytoma (Fig. 4B). This result correlated with increased ADAM19 expression as detected by mRNA/protein analysis in AIII and GBM tumors. Most likely, ADAM19 activity observed in glioblastoma is due to extensive activation of ADAM19 in these tumors. Inhibition by EDTA of metalloprotease activity in general and the inhibitory effect of TIMP1/3 observed in glioblastoma extracts but not in AIII extracts is indicative of additional metalloproteinase activity that is inhibited by TIMP1 and/or by TIMP3 in glioblastomas (9).

Activation of Cellular Migration/Invasive Activity of Glioma Cells by ADAM8 and ADAM19

ADAM proteins contain conserved ectodomains that are required for proteolysis and/or cell adhesion. To demonstrate a functional role for ADAM8 and ADAM19 in glioma cells, we performed Matrigel™ Invasion Assays with C6 glioma cells after transfection of cDNAs encoding either intact ADAM protease or a mutated form lacking enzymatic activity (Fig. 5). Cultured C6 glioma cells did not show invasive activity in Matrigel™ assays. However, when cells were transfected and express either human or mouse ADAM8, respectively, there was a significant invasive activity through Matrigel™ judged by the number of cells migrated to the bottom of the invasion chamber (Fig. 5A). Similarly, human ADAM19 increased the invasive potential of C6 cells (Fig. 5B), whereas transfection of mock vector resulted in very low migration rates through the artificial extracellular matrix. Interestingly, catalytically inactive mutants of ADAM8 (glutamate to glutamine exchange at the glutamate residue 330) and ADAM19 (glutamate to alanine exchange at glutamate residue 346) did not increase the invasive activity of C6 cells. These results clearly demonstrate that ADAM8 and ADAM19 increase the invasive activity of glioma cells and that the catalytic activities of ADAM8 and ADAM19 are required for invasive activity.


Invasive activity of C6 glioma cells expressing ADAM proteins. Invasiveness of C6 rat glioma cells after transfection with different ADAM expressing plasmids was determined by Matrigel™ Invasion assay. (A) Bright-field microscopy of cells attached to the bottom of the invasion chamber stained with toluidine blue. (B) Relative number of invaded cells determined by cell counting of three viewing fields. "Mock," pcDNA3.1 plasmid; hA8, expression plasmid for human ADAM8; mA8, expression plasmid for mouse ADAM8; EQmA8, expression plasmid for mutated catalytically inactive mouse ADAM8; hA19, expression plasmid for human ADAM19; EAhA19, expression plasmid encoding catalytically inactive ADAM19. Values were obtained from 3 independent experiments in duplicates and are presented as mean values ± SD. Statistical significance of the differences between the ADAM transfected and mock transfected cells were analyzed by Mann-Whitney U test (*, p < 0.01; **, p < 0.001). Scale bar in (A) 50 μM, valid for all images.


Primary brain tumors in general bear a grim prognosis due to diffuse infiltration of tumor cells into the surrounding brain tissue, preventing complete surgical removal of the tumor. Moreover, low-grade gliomas gradually develop into more malignant tumors and ultimately into glioblastoma (1). In addition, glioma patients suffer from a tumor-induced immunosuppression (2). At present, no curative therapy is known for gliomas and there is no therapeutical approach that significantly prolongs the life of glioma patients. Thus, glioma biology is studied thoroughly to define potential targets for therapeutic approaches.

The proteins of the ADAMs family are likely to be involved in different features of glioma behavior, such as migration of tumor cells into the surrounding brain tissue and ectodomain shedding of growth factors and/or immune receptors, which in turn would influence the proliferation of tumor cells or serve immunomodulatory effects within the tumor and in the entire patient. With our analysis we highlighted distinct ADAM proteins with a functional role in human primary brain tumors. So far, overexpression of ADAM12 in glioblastomas has been described (23). Using identical primers and LightCycler™ analysis we found additional upregulation of ADAM12 in AIII, OAII, EIII, and in AII specimens. Our findings are in agreement with the observations of Kodama et al (23) since they also observed occasional expression of ADAM12 mRNA in other tumors than glioblastoma. Our quantitative LightCycler™ analysis revealed the strongest regulation of the genes for ADAM8 and 19 in different tumor entities.

ADAM8 has been detected in many cell types, including myeloid and B cells, macrophages, osteoclasts, and bronchiolar lung epithelial cells (22, 31-34). In the present study, we observed maximal ADAM8 mRNA expression in glioblastoma, low-grade oligoastrocytoma, and anaplastic ependymoma. ADAM8 immunoreactivity was localized mainly to the cytoplasm of tumor cells and to reactive astrocytes. In glioblastoma, ADAM8 immunoreactivity was observed on cell membranes that could belong to foamy macrophages. In oligodendroglioma-infiltrated cortex, immunoreactivity for ADAM8 was observed in the cytoplasm and nuclei of neurons whereas the normal cortex was negative for ADAM8. This finding might indicate neurodegeneration in the infiltrated cortex since strong ADAM8 immunoreactivity was observed in degenerating neurons in the mouse CNS (27). Intriguingly, this localization of ADAM8 immunoreactivity closely resembles the gelatinolytic activity as detected by in situ zymography and underlines the potential of neuronal cells to induce the expression of metalloproteases (9).

Immunoblots and activity assays revealed that most of the ADAM8 protein was processed and enzymatically active. All astrocytic tumors and mixed gliomas tested by Western blotting (n = 18) showed considerable expression of the remnant forms of ADAM8 (∼50 kDa and ∼20 kDa). This indicates strong processing of ADAM8 in astrocytic brain tumors and correlated with high proteolytic activity of ADAM8 in the respective tumors. Strong expression of the 90-kDa form of ADAM8 in all ependymomas detected by immunoblotting was remarkable but did not correlate with immunohistochemistry, where only moderate immunoreactivity in a fraction of the tumor cells was seen. At present we can only hypothesize on this observation. Two different antibodies were employed for immunoblotting and immunohistochemistry because the antibody used for immunohistochemistry did not detect the 90-kDa form with the same sensitivity as the antibody used for immunoblotting (data not shown). The antibody used for immunoblotting did not work for immunohistochemistry. Moreover, ependymoma samples showed a discrepancy between immunoblotting and LightCycler™ analysis. While both grade II and III ependymomas revealed strong protein expression, only anaplastic ependymomas showed a strong upregulation of ADAM8 gene expression. This difference between mRNA and protein levels could be due to stabilization of the protein, for example, by vesicular localization in a protein complex, as demonstrated for MT1-MMP (35).

What could be the role of ADAM8 in glioma biology? ADAM8 expression has been observed in inflammatory states, in particular, under conditions in which the TNF-α system is activated, for example, in neurodegeneration (27). In these pathological states, ADAM8 is induced transcriptionally by TNF-α and is mainly localized to reactive astrocytes. One possible role for upregulation of ADAM8 is that the cells evade apoptotic effects of TNF-α, as active ADAM8 contributes to shedding of TNF-R1 (Bartsch et al, unpublished observations). Functional roles for CHL1 and CD23, additional substrates of ADAM8, have not yet been established in brain tumors.

In this study, ADAM19 showed the strongest mRNA upregulation overall and the largest variation in expression between different tumor entities. Maximum levels of ADAM19 mRNA were observed in anaplastic astrocytoma, glioblastoma, anaplastic ependymoma, and PNET. ADAM19-deficient mice exert abnormal heart development (36, 37) due to failure in endothelial mesenchymal transformation, demonstrating that ADAM19 is directly involved in differentiation and proliferation of cells. Immunohistochemistry for ADAM19 revealed that the protein was localized mainly to the cytoplasm of tumor cells. In most of the tumor samples, ADAM19 was detected in the latent and activated form. Although these bands were hardly detectable in glioblastoma lysates, the highest ADAM19 activities were measured in the latter ones. Interestingly, those samples showed an additional band migrating at 65 kDa. This band has been described as a product of autolytic cleavage within the ADAM19 cysteine-rich domain and is required for full activation of the enzyme (26). Candidate substrates of ADAM19 are neuregulin (38) or members of the EGFR family. The latter proteins are of particular interest since in the course of progressive malignancy, the EGFR becomes strongly upregulated in malignant astrocytomas (1). The shedding of neuregulin NRG-β1 by ADAM19 can activate ErbB2/ErbB3 receptors in an autocrine way (39) and ErbB2 was detected in highly malignant gliomas (40). Moreover, a recent study demonstrated that an ADAM inhibitor was able to decrease the processing of heparin binding-EGF in glioblastoma tissue (23).

The genes for ADAM11, 15 and 17 were regulated too, however, only up to seven-fold as compared to normal brain. Although this upregulation was rather modest, it could have significant impact on glioma biology. For ADAM17 it has been shown that even a 2- to 3-fold upregulation of this enzyme may have a significant physiological effects (41) and, like ADAM8, could be involved in immunomodulatory functions, for example, by shedding of TNF receptors (42).

In all tumor samples investigated there was considerable agreement between protein expression levels and proteolytic activities. Taking into account that processing of ADAM proteins in tumors is an important issue, analysis of ADAM protein levels appears to be much more reliable than mRNA expression analysis for defining a functional role of ADAM proteins. A prominent feature of a majority of gliomas, in particular the astrocytic tumors, is the diffuse infiltration of the surrounding brain tissue that renders these neoplasms virtually a systemic disease of the brain. A number of MMPs have been shown to contribute to the behavior of astrocytic tumor cells (4, 11, 16, 43, 44). Considering the similarities between MMPs and ADAMs, the latter are likely candidates to contribute to the infiltrative behavior of glioma cells too. In the present work we demonstrated that expression of ADAM8 and ADAM19, respectively, in C6 glioma cells leads to enhanced cellular migration/invasive activity. More importantly, invasive activity was dependent on proteolytic activity because exchange of the critical glutamate residues in the catalytic sites of ADAM8 and ADAM19 reduced cell migration to a basal level. This highlights proteolytic processing of cell adhesion molecules or ECM components rather than interaction of the disintegrin/cysteine-rich domain with cellular integrins or ECM as instrumental for cell migration. This notion is supported by recent work demonstrating a correlation between ADAM8 expression and cancer prognosis in lung adenocarcinomas. In Matrigel™ assays, transfection of ADAM8 cDNA resulted in increased invasive activity of COS7 and NIH 3T3 cells (45). Taken together, our findings strongly suggest that ADAM8 and ADAM19 might have a significant role in the survival and the infiltrative behavior of astrocytic brain tumors.

In conclusion, our data demonstrate distinct expression patterns and activity states of ADAM8, 12, 15, 17, and 19 in primary brain tumors. For ADAM8 and ADAM19, a correlation between expression levels and activity has been demonstrated and a possible role in invasion of tumor cells in the CNS parenchyma has been demonstrated. Thus, our analysis provides a basis for further investigation of ADAMs as potential targets in the therapy of primary brain tumors.


We thank Dr. S. Yamamoto, Oita Medical School, Japan, for the human ADAM8 expression plasmid, Tina Hagena, Beate Hobmaier and Viviana Sverdlick for their excellent technical assistance and Dr. Stuart Knight for critical reading of the manuscript. This work is dedicated to the 65th birthday of my (AP) academic teacher Dr. Bendedikt Volk.


  • This work was supported by grants of the Deutsche Forschungsgemeinschaft to AP and JWB (Pa 602/3, SFB549/A4, Ba 1606/2-1) and the Wilhelm Sander Stiftung to AP. QXS was supported by grants from D.O.D./U.S. Army Prostate Cancer Research Program, DAMD17-02-1-0238, and the National Institutes of Health (NIH), CA78646.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
View Abstract