Loss of Tumor Suppressor in Lung Cancer-1 (TSLC1) Expression in Meningioma Correlates with Increased Malignancy Grade and Reduced Patient Survival

Transcription

1 Journal of Neuropathology and Experimental Neurology Vol. 63, No. 10 Copyright 2004 by the American Association of Neuropathologists October, 2004 pp Loss of Tumor Suppressor in Lung Cancer-1 (TSLC1) Expression in Meningioma Correlates with Increased Malignancy Grade and Reduced Patient Survival EZEQUIEL I. SURACE, MS,* ERIKS LUSIS, BA,* YOSHINORI MURAKAMI, MD, PHD, BERND W. SCHEITHAUER, MD, ARIE PERRY, MD, AND DAVID H. GUTMANN, MD, PHD Abstract. Meningiomas represent the second most common central nervous system tumor affecting adults. Two of the most frequent early events in meningioma tumorigenesis involve loss of expression of the neurofibromatosis 2 (NF2) and 4.1B genes. Recently, 4.1B was shown to interact with the tumor suppressor in lung cancer-1 (TSLC1) protein, prompting us to examine the expression of TSLC1 in meningiomas. We developed specific anti-tslc1 antibodies to examine TSLC1 expression in normal human leptomeninges, human meningioma cell lines, and human meningiomas of different pathological grades by Western blot (n 10) and immunohistochemistry (n 123). Whereas TSLC1 was expressed in normal human leptomeninges by immunohistochemistry, TSLC1 expression was absent in 3 human malignant meningioma cell lines and markedly reduced or absent in 30% of benign meningiomas by Western blot. Restoration of TSLC1 expression in a TSLC1- deficient human meningioma cell line resulted in reduced cell proliferation. In a series of 123 meningiomas (98 adult and 25 pediatric), TSLC1 expression was absent in 48% of benign (WHO grade I), 69% of atypical (grade II), and 85% of anaplastic (grade III) meningiomas. Moreover, TSLC1 loss was associated with decreased patient survival, within the overall group, and in the atypical meningiomas. Collectively, these results suggest that TSLC1 plays an important role in meningioma pathogenesis. Key Words: Brain tumor; DAL-1; Meningioma; Merlin; NF2; Tumor suppressor. INTRODUCTION The two most common central nervous system tumors in adults are astrocytoma (glioma) and meningioma. Meningiomas are often regarded as benign tumors, but up to 20% may display clinically aggressive features and lead to significant patient morbidity and mortality (1, 2). In contrast to gliomas, relatively little is known about the molecular genetic changes associated with the formation and malignant progression of meningiomas. The most frequently observed genetic changes seen in meningiomas are deletion of chromosome 22q and inactivation of the neurofibromatosis type 2 (NF2) gene (3 7). Since individuals with the NF2 inherited cancer syndrome develop meningiomas at high frequency and there is often a loss of expression of the NF2 protein product (merlin or schwannomin) in both NF2-associated and sporadic meningiomas, inactivation of the NF2 gene is likely a critical initiating event in meningioma pathogenesis (8, 9). The important role of the NF2 gene in meningioma From Department of Neurology (EIS, DHG) and Division of Neuropathology (EL, AP), Washington University School of Medicine, St. Louis, Missouri; Tumor Suppression & Functional Genomics Project (YM), National Cancer Center Research Institute, Tokyo Japan; and Division of Neuropathology (BWS), Mayo Clinic, Rochester, Minnesota. *These authors contributed equally to this work. Correspondence to: David H. Gutmann, MD, PhD, Department of Neurology, Washington University School of Medicine, Box 8111, 660 S. Euclid Avenue, St. Louis, MO gutmannd@neuro. wustl.edu Grant support: This work was supported by funding from the National Institutes of Health (NS41097 to D.H.G.) and The US Department of Defense (DAMD to D.H.G.). formation is underscored by the observation that mice with conditional Nf2 inactivation in leptomeningeal cells develop meningeal cell hyperplasia and subsequent meningioma formation (10), and that re-expression of merlin in meningioma cell lines results in growth suppression (11, 12). Analysis of the predicted protein sequence of the NF2 gene reveals that merlin is structurally related to the Protein 4.1 family of molecules. Recent work from our laboratory has implicated another Protein 4.1 tumor suppressor, Protein 4.1B, in the molecular pathogenesis of sporadic meningioma (9, 13, 14). 4.1B gene deletion and loss of protein expression is also common in meningioma, regardless of histologic grade, suggesting that Protein 4.1B loss, like NF2 inactivation, is an early genetic event in the formation of meningioma. Similar to merlin, reexpression of Protein 4.1B in deficient meningioma cell lines results in reduced cell proliferation (11). The region required for meningioma cell growth suppression is contained within a 503 amino acid fragment, termed DAL- 1 (differentially expressed in adenocarcinoma of the lung-1) (11, 15). Several cytogenetic alterations have been associated with meningioma malignant progression, including losses on chromosomes 1p, 6q, 9p, 10, 14q, and 18q as well as gains on chromosomes 1q, 9q, 12q, 15q, 17q, and 20q (16 20). Some genes identified in these chromosomal regions have been implicated in growth regulation. Losses of the p14, p15, and p16 genes located on 9p21 have been demonstrated in anaplastic meningiomas (21, 22) and mutations in PTEN (10q23) or CDKN2C (1p32) have been described in rare anaplastic meningiomas (22, 23). 1015

2 1016 SURACE ET AL Some insights into merlin and Protein 4.1 function have also derived from the identification of interacting proteins. Merlin interacts with the CD44 transmembrane hyaluronidate receptor and the hepatocyte growth factor regulated tyrosine kinase substrate (HRS/HGS), both of which are required for merlin growth regulation (24, 25). Protein 4.1B also interacts with CD44 (26), but recently has been shown to bind to another transmembrane receptor involved in mediating cell adhesion and attachment, termed tumor suppressor in lung cancer-1 or TSLC1 (27 30). TSLC1 was originally cloned as a tumor suppressor on chromosome 11q23, which is inactivated in non-small cell lung cancer (31), but several studies have now shown that it may be involved in the pathogenesis of diverse types of human cancers (32 35). TSLC1 contains a large, heavily glycosylated extracellular domain and a short cytoplasmic tail containing 47 amino acids, which is required for TSLC1 growth suppression (36). In normal human and rodent tissues, TSLC1 mrna is highly expressed in brain, raising the possibility that TSLC1 might also be important in the molecular pathogenesis of brain tumors. In light of our previous studies on Protein 4.1B in meningioma pathogenesis, we sought to determine the expression of TSLC1 in normal human leptomeninges and meningiomas. In this report, we demonstrate that TSLC1 is robustly expressed in human leptomeninges, but lost or markedly reduced in 30% of WHO grade I meningiomas and lost in 3 human meningioma cell lines by Western blot. Re-expression of TSLC1 in deficient human meningioma cells resulted in reduced cell growth. In a series of 98 adult and 25 pediatric human meningiomas, we found that loss of TSLC1 expression was higher in those tumors with high proliferation indices (groups 3 and 5). Moreover, loss of TSLC1 expression was associated with decreased overall patient survival, both in the entire cohort and, to a lesser extent, in atypical meningiomas, which exhibit the greatest variability in clinical behavior. Collectively, these results suggest that TSLC1 plays an important role in meningioma growth regulation. MATERIALS AND METHODS Patients and Human Tissues Human specimens (normal leptomeninges and meningiomas) were used in strict accordance with approved Human Studies protocols at the Washington University School of Medicine and the Mayo Clinic Foundation. A total of 133 meningiomas were used in this study. Tumors were classified and graded according to the 2000 World Health Organization scheme (37). Clinicopathologic criteria and data for patient follow-up were obtained from previously published studies (1, 2). Briefly, meningioma group 1 tumors were WHO grade I meningiomas with no recurrence after at least 10 years of clinical follow-up. Group 2 meningiomas were also WHO grade I tumors, but patients had tumor recurrence despite gross total resection. Group 3 atypical WHO grade II meningiomas had no evidence of brain invasion, but exhibited 4 mitoses/high power field (HPF), while group 4 atypical WHO grade II meningiomas exhibited brain invasion and had low mitotic indices ( 4 mitoses/hpf). Group 5 represented WHO grade III (anaplastic) meningiomas. Antibodies and cdna Constructs A polypeptide containing 18 amino acids of the TSLC1 cytoplasmic domain (N- INAEGGQNNSEEKKEYFI-C) was synthesized, fused to keyhole limpet hemocyanin (KLH), and used as an immunogen to raise rabbit polyclonal antibodies using standard immunization protocols. Antisera from each of the 3 rabbits immunized were tested against both brain and RT4 rat schwannoma cell lysates as well as a glutathione-s-transferase fusion protein containing TSLC1 sequences (amino acid residues ). Each of the 3 antisera (ES1, ES2, and ES3) recognized TSLC1, and two (ES1, ES2) were affinity-purified on peptide columns containing Sepharose A conjugated to the immunizing peptide. One purified antibody, ES1, was used for all of these studies. To determine the specificity of the ES1 (anti-tslc1) antibody, the ES1 antibody was incubated with Sepharose-conjugated TSLC1 or 4.1B (DAL-1) peptide overnight at 4 C. The beads were then removed and the supernatant used for Western blot and immunohistochemistry on paraffin sections. A previously generated anti-tslc1 (CC2) was used to verify the specificity of the ES1 antibody (27). The V5 epitope (Invitrogen, Carlsbad, CA) and tubulin-specific (Sigma, St. Louis, MO) antibodies were used for Western blots. A previously reported rabbit polyclonal antibody against Protein 4.1B (3A1) was used for immunohistochemistry (14). Full-length TSLC1 cdna was originally cloned in the pcdna3.1.hygro ( ) vector (Invitrogen). Full-length TSLL1 and TSLL2 (two functionally related but antigenically distinct proteins) were originally cloned in the pcdna3.1.v5.his TOPO vector (Invitrogen). The TSLC1, TSLL1, and TSLL2 cdnas used in these experiments were of human origin. In vitro transcription and translation of full-length TSLC1, TSLL1 and TSLL2 were performed using the TNT protocol (Promega, Madison, WI) according to the manufacturer s instructions. A GST fusion peptide containing TSLC1 amino acid residues was generated by digesting the pcdna3.hygro( ) TSLC1 construct with EcoRV and XhoI and cloning the resulting fragment into the pgex-4t3 vector (Amersham Biosciences, Piscataway, NJ). The GST-TSLC1 fusion protein was prepared by transforming DE3 cells (Stratagene, La Jolla, CA) with the pgex-4t3.tslc1 construct. Bacteria were induced overnight with 0.5 mm IPTG at room temperature, and the fusion protein was collected on glutathione-agarose beads (Sigma) and detected by Coomassie blue staining. Generation of Stable Cell Lines and Analysis of Cell Proliferation Stable IOMM-Lee cell lines were established by transfection of 8 g of pcdna3.hygro ( ) (vector) or pcdna3.hygro ( ).TSLC1. Cells were selected in 200 g/ml hygromycin for 2 weeks before individual clones were selected. Twelve individual clones for each construct were isolated and expanded. Expression of TSLC1 was assessed by Western blot using the ES1 antibody (dilution 1:20,000). Although multiple such

3 TSLC1 EXPRESSION IN MENINGIOMA 1017 clones were generated, 2 vector-transfected clones and 2 TSLC1-expressing clones were selected for further analysis. For the thymidine incorporation experiments, 10 5 cells from each cell line were plated in 6 wells of a 24-well plate. After 24 hours, the media was changed to serum-free DMEM and incubated at 37 C for 24 hours. The media was aspirated and replaced with serum-free DMEM containing 1 mci/ml 3 H-thymidine (Amersham Biosciences) and incubated at 37 C for 4 hours. Cells were washed twice with PBS and solubilized in 200 mm NaOH. Disintegrations per minute were determined in a Liquid Scintillation Beta Analyzer (Packard Instruments, Meridan, CT), with the mean and standard deviation determined for each cell line. Each experiment was repeated 3 times with similar results. Immunohistochemistry (IHC) Immunohistochemistry was performed on paraffin-embedded sections according to established protocols in our laboratory (9, 13). Paraffin sections (mouse tissues, normal human rolled leptomeninges, or meningioma specimens) were deparaffinized and endogenous peroxidase activity was quenched with 3% hydrogen peroxide. Antigen retrieval was accomplished by boiling slides for 10 min in 10 mm sodium citrate. Sections were then blocked with 1% BSA, and incubated overnight with the primary antibody (rabbit polyclonal ES1, 1:20,000 dilution; or 3A1, 1:500 dilution) at 4 C. Secondary anti-rabbit biotinylated antibodies (Sigma) were used at a 1:200 dilution and slides were developed using DAB as the chromogen. Omission of the primary antibody was used as a negative control. Slides were photographed at 200 or 400 magnification using a Nikon Eclipse 660 microscope (Japan) equipped with a CCD camera. Western Blotting Organs from adult mice, as well as confluent rat schwannoma RT4 and human CH157-MN, IOMM-Lee and F5 meningioma cells, human meningioma specimens, and stable IOMM cell lines were homogenized in NP40 lysis buffer containing protease inhibitors. Protein concentration was determined by the Bio-Rad method (Bio-Rad Laboratories, Hercules, CA). Electrophoresis was performed using either 100 g or50 g of each sample separated on 10% SDS-PAGE gels. Proteins were transferred onto Immobilon membranes (Millipore, Bedford, MA) for Western blotting with the ES1 antibody (dilution 1:20,000). Western blots were developed using horseradish peroxidaseconjugated secondary antibodies (1:20,000) and ECL chemiluminescence (Amersham Biosciences). Statistical Analysis All statistics were calculated using SigmaStat 3.0 (SPSS, Chicago, IL). Associations between TSLC1 or Protein 4.1B immunoreactivity and tumor grade were evaluated based on the Fisher exact test or chi-square test. Overall survival rates were estimated using Kaplan-Meier methods and their associations were analyzed using the log rank test. All reported p values were 2-sided and a value of p 0.05 was considered statistically significant. RESULTS TSLC1 Is Expressed in Mouse Brain In order to analyze TSLC1 expression in human brain tumors, we first generated a rabbit polyclonal antibody raised against 18 amino acids contained within the cytoplasmic tail of human TSLC1. Lots of crude sera obtained from 3 independently hyperimmunized rabbits were tested by Western blot. Western blot analysis of RT4 rat schwannoma cells and mouse brain using 2 lots of anti-tslc1 antibody (ES1, ES2) produced identical TSLC1 expression patterns (data not shown). These 2 lots were affinity-purified using peptide column chromatography, and the ES1 antibody was chosen for all experiments. ES1 antibody recognized different TSLC1 peptides by Western blotting, including the full-length TSLC1 molecule as well as a GST fusion peptide consisting of a fragment of TSLC1 that contains the cytoplasmic tail (amino acids ) (Fig. 1A). This antibody was specific to TSLC1 and did not recognize the other 2 related members of the TSLC1 family, TSLL1 and TSLL2 (Fig. 1B). Identical results were obtained with a previously generated anti-tslc1 antibody, CC2 (27; data not shown). To evaluate the specificity of the ES1 anti-tslc1 antibody, we performed absorption experiments using relevant Sepharose-conjugated TSLC1 peptide or irrelevant (control) Sepharose-conjugated Protein 4.1B peptide. Absorbed ES1 antiserum was used for Western blotting (Fig. 1C) and immunohistochemistry on mouse brain sections (Fig. 1D). Absorption with the TSLC1 peptide completely removed the TSLC1 immunoreactivity, whereas no effect was observed when incubating with the Protein 4.1B peptide. The ES1 antibody was then used to analyze expression of TSLC1 in different organs of adult mouse by Western blot and immunohistochemistry. Western blot analysis demonstrated robust expression of TSLC1 in brain and lung (Fig. 2A), as previously reported on the RNA level (31). Interestingly, brain exhibited a pattern of multiple bands ranging from 50 to 100 kda. Sequence analysis predicts the existence of 6 putative N-glycosylation sites in TSLC1, which may account for these multiple protein species (28). A similar pattern was also observed in human brain (data not shown). While only brain exhibited robust TSLC1 expression, lung, and to a lesser degree, spleen, also expressed TSLC1. In these experiments, we consistently noted that the TSLC1 protein expressed in spleen migrates slower than that seen in lung or brain. The expression profile of TSLC1 obtained by Western blot was also confirmed by immunohistochemistry on mouse paraffin sections (Fig. 2B). We observed robust expression in brain, including corpus callosum and hippocampus, where high levels of TSLC1 expression were

4 1018 SURACE ET AL

5 TSLC1 EXPRESSION IN MENINGIOMA 1019 observed in white matter tracts containing glia. No staining of neuronal cell bodies was seen. In support of glial TSLC1 expression, Western blot analyses of cultured murine primary astrocyte lysates demonstrate TSLC1 expression (data not shown). In the cerebellum, both the molecular and granular layers exhibited strong staining, while Purkinje cells demonstrated no expression. In addition to brain expression, TSLC1 was also found in lung tissues, including bronchiolar epithelium and the alveolar lining. Spleen exhibited diffuse light staining throughout the red pulp, correlating with the results obtained by Western blot. Liver and heart did not express TSLC1. TSLC1 is Expressed in Human Leptomeninges and Is Lost in Low-Grade Human Meningiomas and Meningioma Cell Lines In order to study the possible involvement of TSLC1 in the pathogenesis of meningiomas, we first analyzed its expression in normal human meninges by immunohistochemistry (Fig. 3A). IHC of paraffin sections revealed robust staining of human leptomeningeal cap cells. Next, a series of 10 benign WHO grade I meningiomas were analyzed by Western blot. Three of 10 tumors (samples 1, 5, and 8) exhibited markedly reduced or absent TSLC1 expression (Fig. 3B). We then sought to determine whether TSLC1 was expressed in 3 well-characterized human malignant meningioma cell lines (IOMM-Lee, CH157- MN, and F5). We found no expression of TSLC1 in any of these human meningioma cell lines (Fig. 3C). In contrast, robust TSLC1 expression was detected in the RT4 rat schwannoma cell line. Re-Expression of TSLC1 in Deficient Human Meningioma Cell Lines Reduces Cell Proliferation Consistent with the role of TSLC1 as a tumor suppressor, previous studies have shown that TSLC1 re-expression in deficient lung carcinoma cell lines results in decreased cell proliferation (31). To determine whether TSLC1 negatively regulates growth in meningioma cells, we generated TSLC1-deficient IOMM-Lee human meningioma cell lines that stably express TSLC1. Western blot confirmed constitutive expression in several independent clones. Two clones with similar levels of TSLC1 expression and 2 vector-transfected clones were chosen for further study (Fig. 4A). Cell proliferation was studied by thymidine incorporation. Cells stably expressing TSLC1 showed a statistically significant reduction in cell growth compared to vector-transfected clones (p 0.01; Fig. 4B). These results suggest that TSLC1 also functions as a negative growth regulator in meningioma. TSLC1 Loss Is More Frequent in High-Grade Meningioma Since Western blotting might underestimate the number of TSLC1-negative tumors, due to the presence of contaminating normal stroma and blood vessels that express TSLC1, we expanded our initial analysis to include 123 paraffin-embedded human meningioma specimens (98 adult and 25 pediatric tumors). Using immunohistochemistry, we could specifically examine TSLC1 expression in the tumoral tissue. We observed 3 distinct patterns of TSLC1 immunoreactivity, including positive tumoral staining (Fig. 5A), focal tumoral staining (Fig. 5B), and no tumoral staining (Fig. 5C, D). Tumors exhibiting focal staining were scored as positive for TSLC1 expression. Normal human leptomeninges as well as 4 cases of meningeal hyperplasia (non-neoplastic leptomeningeal growth greater than 10 cell layers thick) were included as controls. As it would be predicted for non-neoplastic tissue, none of the 4 cases of meningeal hyperplasia exhibited loss of TSLC1 expression (Fig. 5E). We found a higher percentage of TSLC1 loss with increasing grade of meningioma tumors (Table 1), which was statistically significant (p 0.01; chi-square test). No obvious differences were seen in the adult versus pediatric cases and the combined frequencies for losses of TSLC1 expression were 48%, 69%, and 85% for WHO grade I, II, and III tumors, respectively. Adult tumors were further stratified by clinicopathologic criteria as described in the Materials and Methods section. Since we wanted to assess which feature, if any, correlated with TSLC1 loss, we chose to stratify atypical WHO grade II meningiomas into a group with no brain invasion but Fig. 1. Characterization of the anti-tslc1 (ES1) antibody. A: ES1 antibody was tested for its ability to recognize different TSLC1 peptides by Western blot. In vitro transcribed and translated full-length TSLC1 (lane 1) and a GST fusion protein containing TSLC1 amino acids (lane 2) were both recognized by ES1 antibodies. B: The specificity of the ES1 antibody was determined by Western blot using in vitro transcribed and translated full-length TSLC1 and related proteins TSLL1 and TSLL2 (left panel). The same blot was stripped and probed with anti-v5 antibody (right panel) to confirm expression of TSLL1 and TSLL2. Neither TSLL1 nor TSLL2 were detected by the anti-tslc1 ES1 antibody. The specificity of the anti-tslc1 ES1 antibody was evaluated by absorption with both relevant (TSLC1) and irrelevant (Protein 4.1B) Sepharose-conjugated beads. Absorption with TSLC1, but not Protein 4.1B, peptide Sepharose-conjugated beads resulted in loss of TSLC1 expression in mouse brain by Western blot (C). Similarly, absorption with TSLC1 peptide Sepharose-conjugated beads (right panel), but not Protein 4.1B peptide Sepharose-conjugated beads (left panel), resulted in loss of TSLC1 immunoreactivity in mouse brain by immunohistochemistry (D). Photomicrographs were taken at 400. Scale bars: 50 m.

6 1020 SURACE ET AL

7 TSLC1 EXPRESSION IN MENINGIOMA 1021 high proliferation indices (group 3) and a group with brain invasion but low proliferation indices. We observed fairly equal percentages of TSLC1 expression in group 1 and group 2 meningiomas (WHO grade I), regardless of whether or not the tumors subsequently recurred (Table 2). In contrast, we observed significantly higher percentages of immunonegative tumors in atypical meningiomas with high proliferative indices (87% of group 3 meningiomas lacked TSLC1 expression) and/or anaplastic features (82% loss in group 5). Group 3 and group 5 meningiomas exhibited the highest proliferative indices in this clinicopathologic set. On the other hand, WHO grade II meningiomas with brain invasion, but low proliferative indices (group 4) exhibited 50% TSLC1 loss, similar to that seen in WHO grade I meningioma. The difference between TSLC1 losses encountered in atypical meningiomas with increased proliferation (87%) versus those with low proliferation (50%) was statistically significant (p 0.01) and thus suggests that TSLC1 loss might be specifically associated with abnormalities of cell cycle regulation. Protein 4.1B immunohistochemistry was performed on 88 of the meningiomas included in this study (Table 2). As previously reported (14), Protein 4.1B (DAL-1) loss was not statistically different across meningioma grades (p 0.37, chi-square test). In contrast, TSLC1 loss was observed more frequently in high-grade meningioma (p 0.013, chi-square test). In addition, we did not observe any correlation between TSLC1 loss and Protein 4.1B in this meningioma series. Collectively, these results suggest that loss of 4.1B (DAL-1) is an early molecular alteration in meningioma pathogenesis, while TSLC1 loss represents a late progression-associated event. Patients with Meningiomas Exhibiting TSLC1 Loss Have Decreased Survival In light of our finding that loss of TSLC1 expression correlated with increased meningioma proliferation, we next examined the association between TSLC1 expression status and patient survival. Data had been previously collected for 91 adult patients who underwent surgery for resection of meningioma. This cohort of patients had been followed for a range of 1 week to 23 years, with a median follow up of 9.5 years. There was no statistical association between TSLC1 status and recurrence-free survival (data not shown). However, as shown in Figure 6A, the overall survival for patients across all grades of meningiomas revealed significantly decreased overall survival in patients with TSLC1-immunonegative tumors (p 0.016; log rank test). Since we have already shown that TSLC1 expression is lost in a greater percentage of high-grade tumors and those with high proliferative indices, we wished to determine whether TSLC1 expression was associated with survival, independent of meningioma grade. No survival differences were encountered in benign or anaplastic meningiomas, but when we divided the WHO grade II meningiomas according to TSLC1 expression, we found that in this subset, survival was decreased in patients with TSLC1 negative tumors (Fig. 6B). While this trend did not reach statistical significance (p 0.198; log rank test), these findings suggest that TSLC1 expression might be an independent prognostic marker for meningioma survival within this intermediate grade category that is generally associated with the greatest variability in biologic behavior overall. Previous studies have shown that WHO II meningiomas stratified into clinicopathologic groups 3 and 4 exhibit no differences in survival (1, 2). To confirm these findings on our cohort, we performed the same survival analysis on these 2 clinicopathologic subgroups (data not shown). In contrast to the results obtained when stratifying WHO grade II meningiomas by TSLC1 expression, there was no difference in overall survival between WHO grade II meningiomas with high proliferative indices versus brain invasion (p 0.911; log rank test). DISCUSSION TSLC1 was originally identified as a transmembrane protein involved in specifying cell adhesion. Overexpression of TSLC1 in cell lines in vitro is associated with increased homophilic adhesion (28 30, 38). The 442 amino acid TSLC1 N-linked glycoprotein contains 3 immunoglobulin-like C2 type fragments, a transmembrane region, and a short intracytoplasmic tail. In this regard, the TSLC1 cytoplasmic tail exhibits strong sequence similarly to the cytoplasmic tail of glycophorin C (GPC), which binds Protein 4.1 molecules and maintains cell shape. The similarity between GPC and TSLC1 suggested that it Fig. 2. TSLC1 expression in normal mouse tissues. A: Organs from adult mice were homogenized in NP40 lysis buffer as described in the Materials and Methods section. Total protein concentration was determined and 50 g were loaded onto 10% polyacrylamide gels. Western blot was performed using ES1 antibody. After 1 min exposure (top panel) only lung and brain show a band at 100 kda. In addition, brain shows multiple bands ranging from 50 to 100 kda. After a 5-min exposure (bottom panel), a slower migrating band was observed in spleen. B: Immunohistochemistry on paraffin sections of mouse tissues was performed as described in the Materials and Methods section. In agreement with the results obtained by Western blot, positive staining was observed in brain and lung. Spleen shows diffuse low level staining in red pulp, while heart and liver exhibit negative staining. All photomicrographs were taken at 400 except spleen ( 200). Scale bars denote 50 m in photomicrographs taken at 400, and 100 m for 200 photomicrographs.

8 1022 SURACE ET AL Fig. 3. Expression of TSLC1 in normal human leptomeninges, meningioma surgical specimens, and meningioma cell lines. A: Immunohistochemistry analysis of rolled normal human leptomeninges embedded in paraffin using the ES1 antibody. Predominant staining of leptomeningeal cap cells was observed. Both photomicrographs were taken at 200. Bars denote 100 m. B: Ten meningioma samples were homogenized in NP40 lysis buffer as described in the Materials and Methods section and 100 g of total protein were loaded per lane. Western blot using ES1 antibody was performed. Three of the 10 grade I meningiomas exhibited reduced TSLC1 expression. C: Fifty g of the total protein supernatant from subconfluent rat schwannoma RT4 cells and human meningioma cells (IOMM-Lee, CH157-MN, and F5) were loaded onto a 10% polyacrylamide gel. Western blot using the ES1 antibody was performed. Only RT4 cells expressed TSLC1, while neither human meningioma cell line demonstrated TSLC1 expression. Tubulin was used as a loading control.

9 TSLC1 EXPRESSION IN MENINGIOMA 1023 Fig. 4. Re-expression of TSLC1 in deficient IOMM-Lee meningioma cells results in reduced cell growth in vitro. A: IOMM-Lee cells were transfected with pcdna3.hygro ( ).TSLC1 or empty vector, and independent clones selected for further analysis. Lanes 1 and 2 correspond to the TSLC1- expressing clones #10 (T10) and #11 (T11), while lanes 3 and 4 correspond to the empty vector clones #4 (V4) and #5 (V5). Tubulin was included as a loading control. B: TSLC1 re-expression results in a decrease in IOMM-Lee cell proliferation, as measured by thymidine incorporation (dpm). The mean and standard deviation for the 2 independent clones of vector-transfected cells (V4 and V5) and TSLC1-transfected cells (T10 and T11) are shown. The results shown are representative of 3 independent experiments. Asterisk denotes statistical significance (p 0.01) using the Student t-test. might also bind proteins that associate with the carboxyl terminal cytoplasmic tail of GPC, like Protein 4.1 molecules. The recent report describing the direct binding of the Protein 4.1B fragment (DAL-1) to the cytoplasmic FERM binding region of TSLC1 (27) raised the intriguing possibility that TSLC1 functions as a tumor suppressor in similar tissues as Protein 4.1B (e.g. lung and brain). To study the expression of TSLC1 in human brain tumors, we generated rabbit polyclonal antibodies that specifically recognize the cytoplasmic tail of TSLC1. Using these antibodies, we detected robust expression in mouse brain, in which TSLC1 migrates as multiple protein species on SDS-PAGE, ranging from 50 to 100 kda. Previous studies have shown that TSLC1 is translated as a 50-kDa protein, which is heavily glycosylated at multiple sites in the extracellular domain, to result in a 75-kDa protein (28). While the most plausible explanation for the multiple protein species is TSLC1 glycosylation, two other possibilities exist. First, it is possible that TSLC1 undergoes alternative splicing to generate the multiple protein species; however, we observe the same pattern upon transfection of a full-length human TSLC1 cdna (E. Surace and D. Gutmann, unpublished results). Second, it is possible that our antibodies recognize the related TSLL1 and TSLL2 proteins, which migrate as 45 and 50 kda proteins. However, using TSLL1 and TSLL2 cdna constructs, we have shown that the anti-tslc1 antibodies do not recognize these related proteins. Lastly, the 100-kDa TSLC1 protein observed in brain lysates was also verified using a previously published TSLC1 antibody (CC2; 27, 28). In light of previous studies on Protein 4.1B in meningioma pathogenesis, we chose to analyze expression of the Protein 4.1B-associated protein TSLC1 in meningioma. We demonstrate that TSLC1, like merlin and Protein 4.1B, is robustly expressed in human leptomeningeal tissues but is absent in 30% to 50% of benign meningiomas. This frequency of TSLC1 loss is similar to other previously reported studies in other tumor types. TSLC1 loss was seen in 40% of non-small cell lung cancer (31), 32% of prostate cancer (32), and 50% of esophageal squamous cell cancers (33). Analysis of 3 well-characterized human meningioma cell lines (CH157-MN, IOMM-Lee, and F5) demonstrated no expression of TSLC1. After re-introduction of TSLC1 into deficient meningioma cells, we observed a reduction in cell proliferation in vitro. These results support the assignment of TSLC1 as a negative growth regulator for meningioma, as has been shown for other tumor types. While this was not the focus of the present work, previous studies in a wide variety of different tumor types have suggested that promoter methylation accounts for TSLC1 gene silencing in a large proportion of cases (32, 39 42). In these reports there was excellent concordance between TSLC1 RNA or protein loss and TSLC1 promoter methylation. Studies are presently underway to determine the mechanism of TSLC1 inactivation in meningioma. In this study, we then examined a large series of 123 human meningioma specimens (98 adult and 25 pediatric

10 1024 SURACE ET AL Fig. 5. TSLC1 expression in meningioma and meningeal hyperplasia. Photomicrographs of representative examples of TSLC1 staining patterns, including (A) positive tumoral staining, (B) focal tumoral staining, (C) negative meningioma with an intratumoral blood vessel acting as a positive control, (D) negative brain invasive WHO grade II meningioma with TSLC1 expression in adjacent brain parenchyma, and (E) positive staining in hyperplastic meningothelial nests (non-neoplastic tissue). Magnification 400. tumors) by immunohistochemistry. TSLC1 was absent in 48% of benign meningiomas. We further demonstrated that TSLC1 loss was associated with increasing histological grade. Strikingly, after stratifying the adult tumors by clinicopathologic criteria, we found a strong correlation between loss of TSLC1 expression and atypical meningiomas with high proliferative indices ( 4 mitoses/10 high-power fields). In group 3 (atypical meningiomas

11 TSLC1 EXPRESSION IN MENINGIOMA 1025 TABLE 1 TSLC1 Loss in Adult and Pediatric Meningiomas Adult* Pediatric WHO Grade n TSLC1 loss (%) n TSLC1 loss (%) I II III * p value 0.01 (chi-square test) with high proliferative indices) and WHO grade III anaplastic meningiomas (group 5), we observed loss of TSLC1 expression in 87% and 82% of tumors, respectively. In contrast, atypical meningiomas with brain invasion but low mitotic index (group 4) had a similar frequency of loss to that of the benign meningiomas. These results suggest that TSLC1 loss might be specifically involved in cell cycle regulation, rather than invasiveness. Further studies on Tslc1 knockout mice will be required to determine whether loss of Tslc1 expression alone is sufficient for meningioma formation, as has been previously reported for the product of the Nf 2 tumor suppressor gene (10). Consistent with our previous studies (13, 14), Protein 4.1B (DAL-1) loss was seen with similar frequencies in all groups of meningioma, suggesting that Protein 4.1B loss is an early molecular event in meningioma pathogenesis. In contrast, TSLC1 loss in meningioma was more frequent in the high-grade tumors. While TSLC1 and Protein 4.1B interact, suggesting a common signaling pathway, we did not observe any correlation between TSLC1 loss and Protein 4.1B expression. It is possible that other TSLC1 protein interactions are important for TSLC1 function, including binding to PDZ-containing proteins, such as in the recently reported association with the human homologue of Drosophila tumor suppressor Dlg (43). Studies are in progress to determine whether TSLC1 growth suppression requires Protein 4.1B. Previous reports on genetic changes in high-grade meningioma have not identified 11q23 as a region of deletion. However, this is not surprising given our previous studies on 4.1B loss in meningioma (9, 13, 14). While Fig. 6. Survival rates of patients with meningioma stratified by TSLC1 expression. A: Ninety-one patients representing all grades of meningiomas are plotted by time of patient death measured in years from surgery using the Kaplan-Meier method. Patients with TSLC1-immunonegative tumors had decreased survival rates compared to patients with TSLC1-immunopositive meningiomas (p 0.016). B: The survival rates are plotted for 45 patients with WHO grade II meningiomas. The survival rates in patients with meningiomas of the same grade were also reduced in TSLC1-immunonegative tumors compared to TSLC1-immunopositive tumors, but this trend did not reach statistical significance (p 0.198). TABLE 2 TSLC1 and 4.1B Loss in Adult Meningiomas Stratified by Clinicopathologic Criteria WHO Grade Group TSLC1 loss 4.1B loss I (benign) II (atypical) III (anaplastic) 1 (without recurrence) 2 (with recurrence) 3 (high proliferative index) 4 (brain invasive, low proliferative index) 5 (malignant) 9/20 (45%) 7/16 (44%) 20/23 (87%) 11/22 (50%) 14/17 (82%) 12/20 (60%) 10/16 (63%) 15/23 (65%) 17/22 (77%) 6/7 (86%)

12 1026 SURACE ET AL loss of heterozygosity and fluorescence in situ hybridization analyses demonstrated 4.1B loss in meningioma, chromosome 18p11.3 losses had similarly not been previously reported in meningioma, yet there are now multiple lines of evidence supporting its role as a tumor suppressor. Additional studies are in progress to confirm our TSLC1 protein expression changes in meningioma at the DNA level in an independent series of meningiomas. Other potential mechanisms of inactivation, such as promoter methylation, have already been mentioned. Finally, we examined whether loss of TSLC1 expression in meningiomas correlated with patient survival. We observed that TSLC1 loss was associated with reduced survival when assessing the entire cohort. When we stratified WHO grade II tumors by their TSLC1 expression status, we additionally found that TSLC1 loss correlated with reduced patient survival, irrespective of mitotic index. Collectively, these results raise the intriguing possibility that TSLC1 may be an independent predictor of survival for the subset of patients with atypical meningioma. However, since this association did not quite reach statistical significance, larger series of atypical meningiomas will need to be studied to verify this relationship. Further studies are also required to define the mechanism of TSLC1 growth suppression in these common brain tumors. In summary, we have shown that while TSLC1 is normally expressed in human leptomeninges and non-neoplastic meningothelial hyperplasia, loss of TSLC1 expression is observed in all grades of meningiomas. In keeping with its proposed role as a tumor suppressor for meningioma, restoration of TSLC1 expression in TSLC1- deficient human meningioma cell lines results in reduced cell proliferation. Moreover, loss of TSLC1 expression is associated with meningiomas exhibiting high proliferative indices, suggesting that TSLC1 may be important for leptomeningeal cell cycle growth control. Mechanistic studies are presently in progress to examine this possibility. ACKNOWLEDGMENTS We thank Mark Foster and Erin Winkeler for expert technical assistance and Dr. Victoria Robb for suggestions during the execution of this project. REFERENCES 1. Perry A, Scheithauer BW, Stafford SL, Lohse CM, Wollan PC. Malignancy in meningiomas: A clinicopathologic study of 116 patients, with grading implications. Cancer 1999;85: Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM. Meningioma grading: An analysis of histologic parameters. Am J Surg Pathol 1997;21: Ruttledge MH, Sarrazin J, Rangaratnam S, et al. Evidence for the complete inactivation of the NF2 gene in the majority of sporadic meningiomas. Nat Genet 1994;6: Gutmann DH, Giordano MJ, Fishback AS, Guha A. Loss of merlin expression in sporadic meningiomas, ependymomas and schwannomas. Neurology 1997;49: Huynh DP, Mautner V, Baser ME, Stavrou D, Pulst SM. Immunohistochemical detection of schwannomin and neurofibromin in vestibular schwannomas, ependymomas and meningiomas. J Neuropathol Exp Neurol 1997;56: Leone PE, Bello MJ, de Campos JM, Vaquero J, Sarasa JL, Pestana A, Rey JA. NF2 gene mutations and allelic status of 1p, 14q and 22q in sporadic meningiomas. Oncogene 1999;18: Ueki K, Wen-Bin C, Narita Y, Asai A, Kirino T. Tight association of loss of merlin expression with loss of heterozygosity at chromosome 22q in sporadic meningiomas. Cancer Res 1999;59: Stemmer-Rachamimov AO, Ino Y, Lim ZY, et al. Loss of the NF2 gene and merlin occur by the tumorlet stage of schwannoma development in neurofibromatosis 2. J Neuropathol Exp Neurol 1998; 57: Perry A, Giannini C, Raghavan R, et al. Aggressive phenotypic and genotypic features in pediatric and NF2-associated meningiomas: A clinicopathologic study of 53 cases. J Neuropathol Exp Neurol 2001;60: Kalamarides M, Niwa-Kawakita M, Leblois H, et al. Nf2 gene inactivation in arachnoidal cells is rate-limiting for meningioma development in the mouse. Genes Dev 2002;16: Gutmann DH, Hirbe AC, Huang ZY, Haipek CA. The protein 4.1 tumor suppressor, DAL-1, impairs cell motility, but regulates proliferation in a cell-type-specific fashion. Neurobiol Dis 2001;8: Ikeda K, Saeki Y, Gonzalez-Agosti C, Ramesh V, Chiocca EA. Inhibition of NF2-negative and NF2-positive primary human meningioma cell proliferation by overexpression of merlin due to vector-mediated gene transfer. J Neurosurg 1999;91: Gutmann DH, Donahoe J, Perry A, et al. Loss of DAL-1, a protein 4.1-related tumor suppressor, is an important early event in the pathogenesis of meningiomas. Hum Mol Genet 2000;9: Perry A, Cai DX, Scheithauer BW, et al. Merlin, DAL-1, and progesterone receptor expression in clinicopathologic subsets of meningioma: A correlative immunohistochemical study of 175 cases. J Neuropathol Exp Neurol 2000;59: Tran YK, Bogler O, Gorse KM, Wieland I, Green MR, Newsham IF. A novel member of the NF2/ERM/4.1 superfamily with growth suppressing properties in lung cancer. Cancer Res 1999;59: Buschges R, Ichimura K, Weber RG, Reifenberger G, Collins VP. Allelic gain and amplification on the long arm of chromosome 17 in anaplastic meningiomas. Brain Pathol 2002;12: Cai DX, Banerjee R, Scheithauer BW, Lohse CM, Kleinschmidt- Demasters BK, Perry A. Chromosome 1p and 14q FISH Analysis in clinicopathologic subsets of meningioma: Diagnostic and prognostic implications. J Neuropathol Exp Neurol 2001;60: Lamszus K, Kluwe L, Matschke J, Meissner H, Laas R, Westphal M. Allelic losses at 1p, 9q, 10q, 14q, and 22q in the progression of aggressive meningiomas and undifferentiated meningeal sarcomas. Cancer Genet Cytogenet 1999;110: Ozaki S, Nishizaki T, Ito H, Sasaki K. Comparative genomic hybridization analysis of genetic alterations associated with malignant progression of meningioma. J Neuro-Oncol 1999;41: Weber RG, Bostrom J, Wolter M, et al. Analysis of genomic alterations in benign, atypical, and anaplastic meningiomas: Toward a genetic model of meningioma progression. Proc Natl Acad Sci USA 1997;94: Perry A, Banerjee R, Lohse CM, Kleinschmidt-DeMasters BK, Scheithauer BW. A role for chromosome 9p21 deletions in the malignant progression of meningiomas and the prognosis of anaplastic meningiomas. Brain Pathol 2002;12: Bostrom J, Meyer-Puttlitz B, Wolter M, et al. Alterations of the tumor suppressor genes CDKN2A ( p16 INK4a ), p14 ARF, CDKN2B

13 TSLC1 EXPRESSION IN MENINGIOMA 1027 ( p15 INK4b ), and CDKN2C ( p18 INK4c ) in atypical and anaplastic meningiomas. Am J Pathol 2001;159: Peters N, Wellenreuther R, Rollbrocker B, et al. Analysis of the PTEN gene in human meningiomas. Neuropathol Appl Neurobiol 1998;24: Morrison H, Sherman LS, Legg J, et al. The NF2 tumor suppressor gene product, merlin, mediates contact inhibition of growth through interactions with CD44. Genes Dev 2001;15: Sun CX, Haipek C, Scoles DR, et al. Functional analysis of the relationship between the neurofibromatosis 2 tumor suppressor and its binding partner, hepatocyte growth factor-regulated tyrosine kinase substrate. Hum Mol Genet 2002;11: Robb VA, Li W, Gascard P, Perry A, Mohandas N, Gutmann DH. Identification of a third Protein 4.1 tumor suppressor, Protein 4.1R, in meningioma pathogenesis. Neurobiol Dis 2003;13: Yageta M, Kuramochi M, Masuda M, et al. Direct association of TSLC1 and DAL-1, two distinct tumor suppressor proteins in lung cancer. Cancer Res 2002;62: Masuda M, Yageta M, Fukuhara H, et al. The tumor suppressor protein TSLC1 is involved in cell-cell adhesion. J Biol Chem 2002; 277: Watabe K, Ito A, Koma YI, Kitamura Y. IGSF4, a new intracellular adhesion molecule that is called by three names, TSLC1, SgIGSF, and SynCAM, by virtue of its diverse function. Histol Histopath 2003;18: Fujita E, Soyama A, Momoi T. RA175, which is the mouse ortholog of TSLC1, a tumor suppressor gene in human lung cancer, is a cell adhesion molecule. Exp Cell Res 2003;287: Kuramochi M, Fukuhara H, Nobukuni T, et al. TSLC1 is a tumorsuppressor gene in human non-small-cell lung cancer. Nat Genet 2001;27: Fukuhara H, Kuramochi M, Fukami T, et al. Promoter methylation of TSLC1 and tumor suppression by its gene product in human prostate cancer. Jpn J Cancer Res 2002;93: Ito T, Shimada Y, Hashimoto Y, et al. Involvement of TSLC1 in progression of esophageal squamous cell carcinoma. Cancer Res 2003;63: Hui AB, Lo KW, Kwong J, et al. Epigenetic inactivation of TSLC1 gene in nasopharyngeal carcinoma. Mol Carcinog 2003;38: Uchino K, Ito A, Wakayama T, et al. Clinical implication and prognostic significance of the tumor suppressor TSLC1 gene deleted in adenocarcinoma of the lung. Cancer 2003;98: Mao X, Seidlitz E, Ghosh K, Murakami Y, Ghosh HP. The cytoplasmic domain is critical to the tumor suppressor activity of TSLC1 in nonsmall cell lung cancer. Cancer Res 2003;15: Louis DN, Scheithauer BW, Budka H, von Deimling A, Kepes JJ. Meningiomas. In: Kleihues P, Cavenee WK, eds. World Health Organization classification of tumours: Pathology and genetics of tumours of the nervous system. Lyon: IARC Press, 2000: Shingai T, Ikeda W, Kakunaga S, et al. Implications of nectin-like molecule-2/igsf4/ra175/sgigsf/tslc1/syncam1 in cell-cell adhesion and transmembrane protein localization in epithelial cells. J Biol Chem 2003;278: Fukami T, Fukuhara H, Kuramochi M, et al. Promoter methylation of the TSLC1 gene in advanced lung tumors and various cancer cell lines. Int J Cancer 2003;107: Honda T, Tamura G, Waki T, et al. Hypermethylation of the TSLC1 gene promoter in primary gastric cancers and gastric cancer cell lines. Jpn J Cancer Res 2002;93: Jansen M, Fukushima N, Rosty C, et al. Aberrant methylation of the 5 CpG island of TSLC1 is common in pancreatic ductal adenocarcinoma and is manifest in high grade Pan1Ns. Cancer Biol Ther 2002;1: Allinen M, Peri L, Kujala S, et al. Analysis of 11q21 24 loss of heterozygosity candidate genes in breast cancer: Indications of TSLC1 promoter methylation. Genes Chromosomes and Cancer 2002;34: Fukuhara H, Masuda M, Yageta M, et al. Association of a lung tumor suppressor TSLC1 with MPP3, a human homologue of Drosophila tumor suppressor Dlg. Oncogene 2003;22: Received March 30, 2004 Revision received May 25, 2004 Accepted May 27, 2004