To study the expression and function of the brain-specific proteinase deficient disintegrins, ADAM11 and ADAM22 (a disintegrin and metalloproteinase).
Specimens of low- and high-grade gliomas and normal brain were analyzed for ADAM11 and ADAM22 expression using Western blotting. The effects of overexpression of ADAM11 and ADAM22 in glioma cells on growth were analyzed using bromodeoxyuridine incorporation linked to immunocytochemistry. Similarly analyzed were the effects on cell proliferation of bacterially expressed glutathione S-transferase fusion proteins with the disintegrin domain of ADAM11 and ADAM22.
ADAM22 is expressed in normal brain and some low-grade gliomas, but not in high-grade gliomas, whereas ADAM11 is expressed in all low- and high-grade gliomas. In vitro, ADAM22 inhibits cellular proliferation of glioma derived astrocytes. The growth inhibition appears to be mediated by interactions between the disintegrin domain of ADAM22 and specific integrins expressed on the cell surface. This growth inhibition can be avoided by over-expression of integrin linked kinase.
ADAM22, a brain-specific cell surface protein, mediates growth inhibition using an integrin dependent pathway. It is expressed in normal brain but not in high-grade gliomas. A related protein, ADAM11, has only a minor effect on cell growth, and its expression is unchanged in low- and high-grade gliomas.
ADAM (a disintegrin and metalloproteinase) proteins are transmembrane proteins with an extracellular disintegrin and metalloproteinase domain (1,28). The N-terminus of the proteins represents the prodomain, between 176 and 285 amino acids long, which is enzymatically removed. All ADAM proteins have a furin cleavage site that is the cleavage point between the prodomain and the metalloproteinase domain. Approximately half of all members of the family of ADAM proteins lack metalloproteinase activity because of an amino acid sequence, which renders them incapable of binding a Zn2+ ion required for catalytic activity. Several catalytically inactive ADAM proteins play a vital role in fertilization (9). Catalytically active ADAM proteins, especially ADAM10 and ADAM17, have a role as sheddases and as activators of growth factors (13,15). The disintegrin loop, a short sequence within the disintegrin domain of ADAM proteins, can interact with integrins and act as ligand similar to matrix proteins, the main ligands of integrins (18,34). Disintegrin domains were first found to be functional components of snake venoms responsible for the inhibition of platelet aggregation (14,31).
ADAM11, ADAM22, and ADAM23 are all expressed predominantly, if not exclusively, in the brain (24). They all lack catalytic function. The complementary deoxyribonucleic acid (cDNA) encoding these ADAM proteins were cloned some years ago (2,24,25). Although there are several ADAM11, ADAM22, and ADAM23 messenger ribonucleic acid (mRNA) species found by Northern analysis or reverse-transcription polymerase chain reaction of total brain ribonucleic acid (RNA), there is little or no information on the expression of ADAM proteins or their biological function. Our experiments aim to investigate the protein expression of ADAM11 and ADAM22 in normal brain and in gliomas and their role in cellular proliferation.
MATERIALS AND METHODS
All specimens were collected in accordance with the conditions of our local human ethics committee. Human control brain specimens were derived from temporal lobectomies of epilepsy patients. High- and low-grade glioma specimens were collected, avoiding necrotic tissues and the areas at the periphery of the tumors that may contain normal brain tissue. All tissue specimens were given separate code numbers, snap frozen in liquid nitrogen, and stored at -70°C until use.
Cell Culture and Reagents
Human epithelial kidney cells, 293T, were a gift from Dr. C. Hovens, University of Melbourne. The human glioma derived D645 cells and murine glioma derived SMA-560 cells were a gift from Dr. David Ashley, Royal Children's Hospital, Melbourne. Both cell lines originate from the laboratory of Dr. Bigner, Duke University, Durham, NC. Chinese hamster ovary (CHO) cells and CHO cells expressing different integrin subunits (α2, α3, α4, α5, α6, α9, α10, and β3) were all donated by Dr. Yoshikazu Takada, The Scripps Research Institute, La Jolla, CA (32). All cells were propagated in DMEM (Trace Scientific, Melbourne) in the presence of 10% fetal bovine serum. The growth medium of the CHO cells was further enriched by the addition of nonessential amino acids (Gibco, Invitrogen, Carlsbad, CA). The green fluorescence protein-tagged integrin linked kinase (ILK) expression vector was a gift from Dr. Turner, SUNY Upstate Medical University, Syracuse, New York.
Cell Attachment Assays
The cell attachment assay was a modification of a protocol described previously (8), incorporating an 3-(4,5-dimethylthiazol- 2-yl)-2,5-diphenyltetrazoliumbromide (MTT) assay step to quantitate cell adhesion. Ninety-six well Nunclon dishes were coated with recombinant glutathione S-transferase (GST) proteins containing the disintegrin domain of ADAM11, 22, or 23 at (10 μg/ml) overnight at 4°C. Positive control wells were coated with fibronectin (10 μg/ml), whereas negative control wells were coated with bovine serum albumin (BSA, 10 μg/ml) or GST (10 μg/ml). The wells were then blocked with 1% BSA in DMEM for 1 hour at 37°C. Before plating, the integrin-expressing CHO cells were serum-starved in DMEM containing 1% BSA. Cells were disaggregated in 0.2 g/l EDTA, washed three times in phosphate-buffered saline (PBS), and the number of viable cells was determined by trypan blue exclusion. A single-cell suspension was prepared in DMEM supplemented with 1% BSA, and a total of 100,000 viable cells was plated into each well. The cells were allowed to adhere to the substrates for 1.5 hours at 37°C. Nonadherent cells were then removed by three gentle washes in PBS. After the final wash, the attached cells were quantitated using an MTT assay as described below.
The MTT assay is a colorimetric assay that measures reduction of MTT by mitochondrial succinate dehydrogenase (16). Water-soluble MTT is absorbed into the cells and into the mitochondria, where it is then reduced to an insoluble formazan product. The formazan is then solubilized in an acidified detergent solution comprising triton X-100/HCl and the concentration determined by optical density at 590 nm with optical density at 690 nm subtracted as a background control.
Generation of Antibodies to ADAM11 and ADAM22
The expression plasmids encoding FLAG-tagged ADAM11 isoform 1 and ADAM22 isoform 4 and HA-tagged ADAM22 isoform 1 were a gift from Dr. K. Sagane, Eisai Co., Ibaraki, Japan. The metalloproteinase domain encoding regions of ADAM11 was cloned as PvuII fragment (750-1295) into pGEX-3×. The disintegrin domain of ADAM22 (SspI fragment 1334-1648 of cDNA) was cloned into the SmaI site of pGEX-3×. Purified GST-fusion proteins were used for the generation of antibodies in rabbits. Specific antibodies were obtained and purified by affinity chromatography.
Generation of Purified ADAM-GST Fusion Protein for Cell Culture
The disintegrin domain of ADAM11 was cloned as 622 bp PvuII cDNA fragment (bp 1296-1917) into pGEX-2TK. GST and GST fusion proteins containing the disintegrin domain of ADAM11 or of ADAM22 as described above were eluted off the glutathione sepharose beads using 5 mmol/L reduced glutathione in PBS. The eluted peptides were then concentrated by centrifugation through an Ultrafree-CL size exclusion membrane (10,000 NMWL, Amicon, Millipore, Bedford, MA) washed with PBS to reduce the glutathione levels and recentrifuged to further concentrate. This purification step was required because peptides in 5 mmol/L glutathione were too toxic to the cells. The peptides were aliquoted, snap frozen, and stored at -70°C until use.
Immunohistochemical Detection of ADAM11 and ADAM22 in Control Brain
Frozen sections of brain tissue derived from temporal lobectomies of epilepsy patients were air dried for 1 hour, fixed in cold acetone for 10 minutes, and blocked in 10% serum. Then they were incubated with either antibodies to ADAM11 or to ADAM22 for 1 hour, followed by several washes in PBS and incubation with horseradish peroxidase-conjugated anti-rabbit antibodies for 1 hour and several washes in PBS. Finally, the bound antibodies were detected using a DAB immuno detection kit from Vector Laboratories (Burlingame, CA). Control sections were processed using antibodies that have been preincubated for 30 minutes with the GST-fusion proteins that were used to generate the antibodies.
Detection of Proteins by Western Blotting
Tissue samples of either normal human brain or gliomas were snap frozen, ground in dry ice and lysed using lysis buffer (150 mmol/L NaCl, 10 mmol/L Tris, pH 7.4, 1 mmol/L EDTA, and 1% Triton X-100, containing aprotinin, leupeptin, and vanadate). Nuclei were removed by low-speed centrifugation followed by high-speed centrifugation to remove other insoluble components. Equal amounts of lysates were used for Western blot analysis using anti-ADAM antibodies. Transfected FLAG-tagged ADAM proteins were detected in cell lysates by using ADAM-specific antibodies or by using anti-FLAG antibodies (clone M2, Sigma, Australia). Antibodies to integrin α1, α2, α3, α4, α5, α6, and β1 were purchased from Santa Cruz (Santa Cruz, CA). Secondary horseradish peroxidase-conjugated anti-rabbit or anti-mouse antibodies were used in conjunction with chemiluminescence reagents Western Lightning (Perkin Elmer, Boston, MA) for the detection of proteins.
Analysis of DNA Synthesis
For proliferation analysis, cells were plated on round 12 mm glass coverslips (Menzel-Glaser, Lomb Scientific Australia). After transfection, cells were incubated a further 24 hours and then pulsed with bromodeoxyuridine (BrdU) (Amersham Pharmacia Biotech) for 16 hours. Cells were fixed and stained for immunofluorescence as follows: cells were washed once in PBS, fixed in 4% paraformaldehyde for 10 to 15 minutes at room temperature, and washed twice in PBS. Then, cells were permeabilized in 0.2% Triton X-100/0.1 mmol/L glycine for 10 minutes at room temperature, washed several times in PBS, and blocked by incubation in 10% fetal calf serum (FCS) in PBS for 60 minutes at room temperature. Samples were then washed several times in PBS and incubated in PBS containing RNase free DNase I (Boehringer Mannheim) at 1 to 2 units/μl and 5 mmol/L MgCl2 for 60 minutes at 37°C. After washing three times in PBS, cells were incubated with anti-BrdU mouse monoclonal antibody (BD PharMingen, San Diego, CA) (diluted to 1:1000 in PBS/FCS) to detect incorporated BrdU and anti-Flag rabbit polyclonal antibody (Sigma, Australia) (diluted to 1:1000 in PBS/FCS) to detect Flag-tagged proteins for 1 hour at room temperature. After three washes in PBS, cells were incubated with ALEXA-fluor secondary antibodies (Molecular Probes Inc., Invitrogen, Groningen) (diluted to 1:5000 in PBS containing 10% FCS) for 1 hour at room temperature in a light-proof container. Samples were then washed three times in PBS, once in PBS containing 1 μg/ml DAPI (Molecular Probes Inc) and once more in PBS, followed by a final wash in water. Coverslips were mounted onto glass slides (Menzel-Glaser microscope slides, Lomb Scientific Australia) by inverting onto 5 to 10 μl of Moviol mountant (Calbiochem, Darmstadt) containing 0.1% para-phenylenediamine (Sigma, St. Louis, MO) as an antiquenching agent. Between 50 and 120 cells were counted as staining either for FLAG, BrdU, or both, the ratios were calculated, and the 95% confidence intervals determined. χ2 tests were performed to determine the statistical significance of the observed results. All proliferation assays were performed at least three times.
Expression of ADAM22 and ADAM11 in Brain
To test the specificity of our anti-ADAM antibodies, we incubated Western blots containing lysates of 293T cells, overexpressing ADAM11, ADAM22, and ADAM23 with antibodies raised against ADAM11 and ADAM22. The anti-ADAM11 antibodies were very specific, whereas the anti-ADAM22 antibodies slightly crossreacted with ADAM11 (Fig. 1A). The overexpressed ADAM11 was found predominantly in the proform of 90 kDa. ADAM11 was barely detectable in the processed form of approximately 83 kDa, although the processed form was the predominant form in the control brain lysate. On the other hand, ADAM22 isoform 1 and isoform 4, although predominantly in the proform, were readily detectable in the processed form of approximately 90 and 84 kDa, respectively. The control brain lysates only showed expression of the processed form of the ADAM22 proteins. Immunohistochemical analysis of human brain specimens derived from the cortex and white matter, when using the anti-ADAM11 and anti-ADAM22 antibodies, showed that both ADAM proteins are expressed in neurons and astrocytes (Fig. 1B). Expression in neurons has been described before, whereas the expression in astrocytes is a novel finding, which was only possible using specific antibodies (22,23).
Lysates of human control brain derived from temporal lobectomies of epilepsy patients and of gliomas were analyzed for ADAM22 and ADAM11 expression using Western blotting. Control brain showed high levels of ADAM22 and lower levels of ADAM11 (Fig. 2A). No apparent proforms of ADAM11 or ADAM22 could be detected, but the diffuse nature of the ADAM22 band suggests that it represents a mixture of several isoforms. So far, five isoforms of ADAM22 that vary slightly in molecular weight have been described (25). All glioblastoma (GBM) specimens expressed ADAM11 (Fig. 2 and data not shown), whereas only 1 (specimen 1179) of 22 GBM specimens expressed ADAM22 (Fig. 2B and data not shown). Further analysis of low-grade and anaplastic astrocytomas also showed loss of ADAM22 expression in 7 of 12 of the specimens (Fig. 2C and data not shown). One mixed astrocyte/oligodendrocyte tumor also expressed ADAM22.
Expression of ADAM22 Inhibits Cellular Proliferation
To investigate the biological effects of ADAM22, we intended to make cell lines that stably overexpress these proteins using co-expression of a puromycin resistance gene. Our target cell lines of interest were a murine glioma derived cell line, SMA-560, and the human glioma cell lines, U87MG and D645. Many independent attempts failed to generate cell lines expressing ADAM22, although ADAM11-expressing cell lines could easily be generated. To rule out the possibility that transfected cells during the selection for cotransfected puromycin resistance lifted off the plate and were lost during media changes, we plated the cells in soft agar and isolated colonies for further investigation for ADAM expression. Again, all puromycin resistant clones were found not to express the cotransfected ADAM protein. These results suggested that expression of ADAM22 may interfere with cellular proliferation.
To investigate this hypothesis further, we transfected human D645 glioma cells with C-terminally FLAG-tagged ADAM proteins and 24 hours later incubated the cells with BrdU for 16 hours to monitor DNA synthesis. ADAM22 expression interfered strongly with DNA replication, whereas ADAM11 expression had only a slight inhibitory effect (Fig. 3). Double staining with BrdU and FLAG was only detected in 19% of cells expressing ADAM22 (Fig. 3A). Over 80% of nontransfected (Flag-negative) cells in the same wells underwent DNA replication (Fig. 3, A and B). In additional control experiments, we transfected cells with a green fluorescence protein expression plasmid and could show that overexpression of a foreign protein as such did not inhibit cellular proliferation (Fig. 3A).
GST-disintegrin Fusion Peptides Inhibit Proliferation of Cells Expressing Integrin αvβ3
To assess the possible role of effects mediated by the disintegrin domain of ADAM22 on proliferation, we made GST-fusion proteins containing the disintegrin domain of ADAM22. We first tested which integrins the ADAM22-GST fusion proteins can bind to. A panel of CHO cells over-expressing individual integrin subunits (32) were plated onto GST-disintegrin coated wells, and their adhesion was assessed 1.5 hours later. Our results showed that CHO cells over-expressing integrin α9, α6, and β3 bind to ADAM22- GST fusion protein (Fig. 4A). The related ADAM23 has previously been shown to bind to integrin α9 and to integrin αv, which we confirmed (2,7). In fact, all three related brain-specific ADAM proteins bind to the same integrin subunits. During the 1.5 hour incubation period, all cells bound well to fibronectin coated wells but not to uncoated wells (data not shown). These results show that for binding to integrins the disintegrin domain of ADAM22 is sufficient and does not require the presence of the cysteine rich domain as well. Because the β3 subunit forms a functional integrin dimer with αv in CHO cells, the results suggest that the ADAM proteins bind to αvβ3. Integrin α6 and α9 form functional heterodimers with β1, and α6 can dimerize with β4 as well. CHO cells express endogenous β1 integrin; the status of β4 expression is not known.
D645 cells express integrin subunits α3, α5, β1, and β3, low levels of α6, but no significant amounts of α2 or α4 (Fig. 4B and data not shown). No information is available on the expression of subunits αv and α9. However, in nonhemopoietic cells, the β3 subunit dimerises predominantly with the αv, which suggests this to be the case in D645 cells as well.
To assess the effects of the ADAM22 disintegrin domain linked to GST on cell growth, we performed cellular proliferation studies using CHO cells or CHO cells over-expressing the β3 subunit and D645 cells expressing endogenous integrin β3 (32). We found no inhibition of proliferation in CHO cells by GST or ADAM22-GST. On the other hand, CHO cells expressing integrin β3 were specifically growth inhibited by the addition of the ADAM22-GST peptide, allowing only 33% of the cells to undergo DNA replication, whereas between 97 and 100% cells incubated with GST alone underwent DNA replication (Fig. 5A). Similarly, the D645 gliomas cells were growth inhibited in the presence of the ADAM22-GST peptide in a dose-dependent manner (Fig. 5B). Neither GST alone nor the disintegrin domain of ADAM11 could mediate growth inhibition even at the highest concentration of 12.5 μg/ml.
Over-expression of ILK Overrides Growth Inhibitory Effect of Disintegrin Domain of ADAM22
ILK is one of the first proteins involved in the cascade of signaling events triggered by ligand binding to integrin β1 and β3 (11,21). Inhibition of ILK using KP-392, a small specific inhibitor, results in inhibition of cellular proliferation (5). If the ADAM22 disintegrin domain binding to integrin αvβ3 interferes with ILK activation, we could expect rescue of the growth effects by over-expression of ILK. We therefore overexpressed ILK in D645 cells and assessed its effect on growth in the presence and absence of the ADAM22-GST peptides. Over-expression of ILK linked to green fluorescence protein had no effect on cell proliferation but completely blocked the growth inhibition observed caused by ADAM22-GST, whereas untransfected cells on the same coverslip that were exposed to ADAM22-GST were growth inhibited (Fig. 5C). It therefore appears that the ILK pathway may be a target of ADAM22 mediated growth suppression.
Studies on the expression of the brain-specific ADAM22 have so far mainly focused on the splice variants, predicting several protein isoforms. ADAM22-deficient mice die soon after birth (23,25). The precise nature of the vital function of ADAM22 is not clear. The structurally related ADAM11 and ADAM23 proteins appear to have different functions because ADAM11-deficient mice lack any discernible phenotype, whereas ADAM23 deficient mice die in utero (19,25).
ADAM10 and ADAM17 induce cell proliferation as they cleave growth factor precursors of epidermal growth factor, betacellulin, epiregulin, transforming growth factor-α, amphiregulin, and heparin-binding epidermal growth factor, respectively (26,29). On the other hand, ADAM12 and ADAM15 have been described to have growth inhibitory function on myoblast cells and on endothelial cells, respectively (3,30). The full growth inhibitory function of ADAM12 depended on the presence of the extracellular domain and the cytoplasmic domain because domain swapping using ADAM9 as donor of alternative domains proved inefficient. However, the prodomain and metalloproteinase domains of ADAM12 were not required for growth inhibition (3). In contrast, growth inhibitory effects of ADAM15 on endothelial cells were elicited using a ADAM15-GST fusion protein, containing the disintegrin domain only (30). Similar to ADAM22, ADAM15 binds to the αvβ3 and α9β1integrin (8,20). This suggests that growth inhibition is mediated by the disintegrin domain binding to cellular integrins.
Our results demonstrate growth inhibition caused by transfection of cells with full-length ADAM22 expression plasmid and after treatment with GST-fusion peptide containing only the disintegrin domain. Growth inhibition using the disintegrin fusion peptide can be mediated in cells expressing the appropriate integrin but not in cells expressing other integrins. A related study showed that the snake venom derived disintegrin, echistatin, reduced insulin-like growth factor-1 mediated cellular proliferation. This effect was dependent on the integrin β3 subunit to which echistatin could bind (17,33). It is intriguing to speculate that the ADAM22 disintegrin domain added to the medium interferes with the same pathway.
ILK is a serine/threonine kinase that is required for the phosphorylation of several substrates such as PKB/Akt and glycogen synthase kinase (GSK)-3 (6). Over-expression of ILK can cause anchorage-independent cell growth and oncogenic transformation of cells (11,21). Cyclic arginine-glycine-aspartic acid (RGD) peptides and integrin-specific blocking antibodies block integrin signaling in response to ligands. They inhibit cell growth, ILK activity, and the phosphorylation of Akt on serine 473 (5). Over-expression of ILK rescued the inhibitory effects of the antibody and of cyclic-RGD (5). These results are the same as the rescue results we obtained using the ADAM22-GST peptides in the presence of ILK. The crucial role of ILK in mediating the ADAM22 effects may explain as well the more dramatic effects of the ADAM22-GST peptides on CHO-β3 cells than on D645 cells. D645 cells like some other glioma cell lines are PTEN deficient (unpublished observations) and may have higher ILK/Akt signaling activity than CHO cells (10). In addition, D645 cells are tumor derived cells whose proliferation may be controlled differently from the untransformed CHO cells.
Our experiments have shown that in addition to neurons, normal astrocytes and some low-grade astrocytomas express ADAM11 and ADAM22. Although ADAM11 and ADAM22 expression in neurons has been shown by in situ hybridization experiments, the finding of their expression in astrocytes is novel (22,23). The use of antibodies allows a more detailed analysis of protein expression than the use of in situ hybridization, and the expression levels of the ADAM proteins in neurons are certainly much higher than in astrocytes (Fig. 1B).
We found further that, although all gliomas retain ADAM11 expression, high-grade astrocytomas lack ADAM22 expression. What could the consequences be in high-grade gliomas for a loss of ADAM22 protein expression? Because some low-grade tumors express ADAM22, loss of expression seems to be a marker for tumor progression. Alternatively, because low-grade gliomas are more infiltrating tumors, they may contain pockets of normal cells expressing ADAM22. Apart from a loss of possible growth inhibitory signals, reduction of ADAM protein expression may reduce the interaction with neighboring cells via their integrins and may increase their invasive potential. It has been shown that the disintegrin snake venom contortrostatin inhibits cellular invasion through matrigel in vitro (27). Another snake venom derived disintegrin, salmosin, was similarly shown to inhibit cell migration through matrigel and B16 cell derived tumor metastasis in vivo; however, it appeared to be a combined effect of migration inhibition and induction of apoptosis (4,12). We are currently investigating how inducible ADAM22 expression affects cell migration in vitro and tumor growth and invasiveness in vivo.
We thank Stephen Cody for assistance with the immunocytochemistry and Dr. Lachlan MacGregor for assistance with statistical analyses of the data. This work was supported by the Royal Melbourne Hospital Neuroscience Foundation and by the Neurosurgical Research Foundation.
ADAM is a disintegrin and metalloproteinase. ADAM11, ADAM22 and ADAM23 are predominantly expressed in the brain. The aim of the authors was to evaluate the protein expression of ADAM11 and 22 in normal brain and in gliomas by Western blot and their role in cellular proliferation in vitro. ADAM22 was found to be expressed in normal brain and low grade gliomas but not in high grade gliomas, it was also found to inhibit cellular proliferation. The authors further demonstrate that the Integrin-linked kinase could be a target for ADAM22 mediated growth suppression.
This is a very well conducted experimental study. It should be considered as preliminary and we are looking forward to further in vivo experiments assessing the role of ADAM22 in invasiveness.
Nicolas de Tribolet
ADAM proteins are a new class of binding sites that may have a role in proliferation of glial tumors. The authors have done a nice job of demonstrating an interaction between these proteins and human tumor cell lines. Growth inhibition is potentially useful in planning future treatment strategies.
Joseph M. Piepmeier
New Haven, Connecticut
ADAM proteins are transmembrane proteins that are of interest because of their surface location which allows for pharmacological interaction and their association with disintegrin and metalloproteinase which suggests a possible role in proliferation and migration. D'Abaco et al. studied the in vitro expression and function of ADAM11 and ADAM22 in a variety of gliomas. They demonstrated that ADAM22 is found in normal brain and low grade gliomas but not in high grade gliomas and has some role in mediating and inhibiting cellular proliferation through specific interaction in its disintegrin domain. ADAM11 was found in all low and high grade gliomas examined but had minimal effect on cell growth.
As a preliminary report, the authors offer a provocative basis for ADAM22 function as a potential anti-tumor target and provide insight into relevant mechanistic pathways. Extrapolation of these findings to an in vivo model will be the next logical step toward validating these results.
Jeffrey N. Bruce
New York, New York