- Email: [email protected]
Proteomics-based identification of differentially expressed genes in human gliomas: down-regulation of SIRT2 gene
BBRC Biochemical and Biophysical Research Communications 309 (2003) 558–566 www.elsevier.com/locate/ybbrc
Proteomics-based identification of differentially expressed genes in human gliomas: down-regulation of SIRT2 gene Masaharu Hiratsuka,a Toshiaki Inoue,b Tosifusa Toda,c Narimichi Kimura,d Yasuaki Shirayoshi,a Hideki Kamitani,e Takashi Watanabe,e Eisaku Ohama,f Candice G.T. Tahimic,b Akihiro Kurimasa,b and Mitsuo Oshimurag,* a
Department of Molecular and Cell Genetics, Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan b Department of Human Genome Science (Kirin Brewery), Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan c Proteomics Collaboration Research Group, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakaecho, Itabashiku, Tokyo 173-0015, Japan d Cellular Signaling Research Group, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakaecho, Itabashiku, Tokyo 173-0015, Japan e Department of Neurosurgery, Institute of Neurological Sciences, Tottori University, Nishimachi 36-1, Yonago, Tottori 683-8504, Japan f Department of Neuropathology, Institute of Neurological Sciences, Tottori University, Nishimachi 36-1, Yonago, Tottori 683-8504, Japan g Department of Biomedical Science, Regenerative Medicine and Biofunction, Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan Received 4 August 2003
Abstract A number of chromosomal abnormalities including 19q deletions have been associated with the formation of human gliomas. In this study, we employed a proteomics-based approach to identify possible genes involved in glioma tumorigenesis which may serve as potential diagnostic molecular markers for this type of cancer. By comparing protein spots from gliomas and non-tumor tissues using two-dimensional (2D) gel electrophoresis, we identified 11 up-regulated proteins and four down-regulated proteins in gliomas. Interestingly, we also discovered that a group of cytoskeleton-related proteins are differentially regulated in gliomas, suggesting the involvement of cytoskeleton modulation in glioma pathogenesis. We then focused on the cytoskeleton-related protein, SIRT2 (sirtuin homologue 2) tubulin deacetylase, which was down-regulated in gliomas. SIRT2 is located at 19q13.2, a region known to be frequently deleted in human gliomas. Subsequent Northern blot analysis revealed that RNA expression of SIRT2 was dramatically diminished in 12 out of 17 gliomas and glioma cell lines, in agreement with proteomic data. Furthermore, ectopic expression of SIRT2 in glioma cell lines led to the perturbation of the microtubule network and caused a remarkable reduction in the number of stable clones expressing SIRT2 as compared to that of a control vector in colony formation assays. These results suggest that SIRT2 may act as a tumor suppressor gene in human gliomas possibly through the regulation of microtubule network and may serve as a novel molecular marker for gliomas. Additional proteins were also identified, whose function in gliomas was previously unsuspected. Ó 2003 Elsevier Inc. All rights reserved. Keywords: Glioma; Proteomics; SIRT2; Cytoskeleton; 19q13; Growth inhibition
Gliomas are the most common type of primary malignant brain tumor and various genetic alterations have been identified in human glioma tumorigenesis. Gene amplification has been observed in the EGF receptor, CDK4, and MDM2 genes, while deletion or inactivating *
Corresponding author. Fax: +81-859-34-8134. E-mail address: [email protected] (M. Oshimura). 0006-291X/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2003.08.029
mutations have been reported for p53 (17p), RB (13q), p16 (9p), PTEN (10q), and DMBT1 (10q) [1]. Moreover, frequent losses of heterozygosity (LOH) for other loci like 1p, 11p, 19q, or 22q in gliomas have been reported [2,3]. In order to provide more insight into the biological behavior of gliomas and identify new targets for therapy, a number of studies have been conducted to identify genes in regions that are genetically altered in this type of cancer. Experiments were also conducted to
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
further identify upstream and downstream target genes that are differentially regulated in gliomas. For example, by using serial analysis of gene expression (SAGE), GOA, a ring finger-B box-coiled coil protein, was identified as an overexpressed protein with no observable gene amplification in human gliomas [4]. ZIC1 and OTX2 genes were also identified to be highly expressed in gliomas using the same method [5]. In this study, we used a proteomics-based technique [6,7] for the identification of differentially regulated proteins in human gliomas. We successfully identified 11 proteins that were up-regulated and four proteins that were down-regulated in gliomas. Interestingly, a group of cytoskeleton-related proteins were observed to be differentially regulated in gliomas. Among these cytoskeleton-related proteins, SIRT2 (sirtuin homologue 2) was observed to be down-regulated. Because SIRT2 is located on 19q13.2, a region known to frequently undergo LOH in human gliomas, we suspected that SIRT2 may be involved in glioma tumorigenesis.
Materials and methods Brain tissues. Brain tumor tissues and non-tumor tissues were obtained from Tottori University Hospital with informed consent and approved by an Institutional Review Committee. Tissues were flashfrozen in liquid nitrogen immediately after surgical removal and stored at )80 °C. Diagnosis of glioma was confirmed histopathologically. Gliomas were then classified according to the WHO grading scheme for central nervous system tumors [8]. Cell culture, transfection of EGFP-SIRT2 expression plasmid, and colony-formation assays. Human glioma cell lines were obtained from ATCC and cultured in DMEM containing 10% calf serum (Invitrogen). A construct encoding fusions of SIRT2 with EGFP was generated by connecting the C-terminus of the fluorescent protein moiety, in-frame, to the N-terminus of full length SIRT2 cDNA using the BamHI/EcoRI sites of pEGFP-C1 (Clontech). SIRT2 cDNA was purchased from ResGen. For transient transfection, sub-confluent cultures of cells in 15 mmdishes (Matsunami, Japan) were transfected with a total of 0.5 lg DNA using Lipofectamine 2000 (Invitrogen) following manufacturer’s protocol. For colony-formation assays, cells in 35-mm dishes were replated at 1:40 after transfection and cultured in medium containing 800 g/ml of G418 (Invitrogen). The medium was changed every 3 days. Twelve days after transfection, the number of colonies was scored. EGFP signals were then visualized using ECLIPSE TE300 (Nikon). Immunohistochemistry. For immunofluorescence detection of tubulin network in transfected cells, cells were fixed in 4% paraformaldehyde/PBS, and immunostained using anti-a-tubulin antibody (1:2000 in 3% skim milk/PBS, Sigma, B-5-1-2) followed by detection with Cy5-labeled secondary antibody (1:100 in 3% skim milk/PBS, Chemicon, AP192F). Simultaneous detection of EGFP-tagged SIRT2 and tubulin network was performed using Bio-Rad Radiance 2000 (Bio-Rad). Protein separation by two-dimensional gel electrophoresis and image analysis. Protein extraction and two-dimensional (2D) electrophoresis were performed as previously reported [7, see also the URL, http:// proteome.tmig.or.jp/2D/2DE_method.html] with minor modifications. Briefly, each 100 mg of tissue sample was crushed in liquid nitrogen
559
and lysed with 300 ll of 5 M urea, 2 M thiourea, 2% CHAPS, 2% SB310, 1% Pharmalyte (pH 3–10, Amersham Biosciences), 1% DTT, and 1% proteinase inhibitor cocktail (Sigma) with sonication. One hundred micrograms of protein was loaded onto gel strips with immobilized pH gradient (pH 4–7, 18 cm, Amersham Biosciences) and isoelectric focusing was performed on a CoolPhoreStar model 3610 (Anatech, Japan). After isoelectric focusing, strips were equilibrated with 50 mM Tris–HCl (pH 6.8), 6 M urea, 2 M thiourea, 2% SDS, 30% glycerol, and 1% DTT for 30 min and then with the same solution containing 4% iodeacetamide instead of DTT for 30 min. Each equilibrated strip was mounted on a 7.5% SDS–polyacrylamide gel and fixed with sharktooth combs. SDS–PAGE was performed on a vertical format using a Tricine buffer system. After second dimension electrophoresis, gels were fixed in 10% methanol and 6% acetic acid for 30 min and protein spots were visualized by SYPRO Ruby (Molecular Probes) staining following the manufacturer’s recommendations. The SYPRO Ruby-stained images were scanned at an excitation wavelength of 488 nm and emission wavelengths of over 550 nm were collected on a Molecular Imager FX (Bio-Rad). Statistical analysis and quantitation of protein expression were conducted using PDQuest software ver7.1 (Bio-Rad). To minimize computational errors resulting from spot dispersion among gels, the volumes of the spots were normalized to the volumes of landmark proteins used for internal standardization [S1; antioxidant protein 2, S2; Rho GDP dissociation inhibitor a, S3; glyoxalase I, and S4; ‘expressed ubiquitously with strong expression in brain’ (Accession No. gi/20521842)]. Tumors from each of the five patients were analyzed in triplicate. On the other hand, non-tumor samples from each of the six patients were analyzed in duplicate (for three samples) or triplicate (for the remaining three samples). Only spots having over twofold changes in density after normalization between two populations were defined as altered. Protein identification. Protein spots were excised by ProteomeWorks Spot Cutter (Bio-Rad) followed by in-gel digestion with trypsin (Promega) according to manufacturer’s specifications. The digested peptide was directly mixed with an equal volume of 10 mg/ml a-cyano4-hydroxycinnamic acid and peptide mass spectra were obtained on an AXIMA-CFR MALDI-TOF-MASS (Shimadzu, Japan) platform. Peptide mass mapping was performed using Mascot Search (Matrix Science). RNA preparation and Northern blotting. Total RNA was prepared using Isogen (Nippongene) from frozen tissues or cultured cells. Electrophoresis was conducted using 5 lg of total RNA loaded per well of a 1.5% agarose/formaldehyde gel. The resolved RNA was then blotted onto a Hybond-N membrane (Amersham Biosciences) and then hybridized with SIRT2 cDNA (NCBI Accession No. NM_030593) probe spanning nucleotides 713–1408. Prior to hybridization, the probe was radioactively labeled using Random Primed DNA Labeling Kit (Roche) in the presence of [a-32 P]dCTP (Amersham Biosciences). RNA loading was verified by ethidium bromide staining of the gel before blotting. Hybridization was performed with 50% formamide, 6 SSC, 0.5% SDS, and 100 lg/ml denatured salmon sperm DNA, and bands were visualized by autoradiography with Xray films (Kodak).
Results 2D gel separation of proteins Glioma tissues derived from five individuals were classified according to the WHO grading system for central nervous system tumors. In this study, gliomas classified as grade II (G201), III (G301 and G302), and IV (G402 and G405) were processed for 2D xelectrophoresis to isolate candidate proteins that are
560
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
differentially regulated in gliomas. We detected about 1000 protein spots on the 2D gels after SYPRO Ruby staining (Fig. 1). The 2D electrophoresis profiles and relative spot intensities obtained from all samples were reproducible when we performed the experiments with duplicates or triplicates. Normalization was also performed using protein internal standards whose corresponding spot intensities remained unchanged between gliomas and normal tissues. Furthermore, only spots
which had over twofold changes in density after normalization were classified as down- or up-regulated. The reproducibility of data from multiple trials, the assignment of a stringent criterion for spot classification, and the use of internal protein standards ensure the reliability of the proteomic data we obtained. Eleven up-regulated proteins and four down-regulated proteins were selected for subsequent analysis by mass spectrometry. Peptide mass-fingerprinting of the
Fig. 1. Two-dimensional protein profiles for human glioma tissue. (A) Proteins were separated on the basis of pI (x-axis) and molecular mass (y-axis). Spots were visualized with SYPRO Ruby gel staining. Numbered spots were defined as altered and then were identified by mass spectrometry. A1-11, B1-4, and S1-4 represent up-regulated, down-regulated, and internal control proteins, respectively. (B) Enlarged portions from 2D gels of human gliomas. (C) Densities of numbered spots in each patient and non-tumor population. Each of the gliomas from five patients was analyzed in triplicate and the density was plotted for each individual (black bars; means standard deviations, n ¼ 3). A total of 15 samples derived from six tumor-free individuals (see Materials and methods) were analyzed and calculated as a population of non-tumor tissues, and the density was plotted in a bar (white bars: means standard deviations, n ¼ 15). *p < 0:05 by two-tailed Student’s t test.
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
561
Fig. 1. (continued)
selected spots and subsequent database search revealed the identity of these proteins as summarized in Table 1. In the succeeding portions, we then attempt to explain the relevance of the listed proteins in gliomas using known information on the nature of these proteins. PEA-15 acts as a modulator of signaling pathways that control cell proliferation and cell death by regulating the subcellular localization of ERK MAP kinase [9]. Our results showing the down-regulation of this protein in gliomas are consistent with the previous report stating that PEA-15 expression is decreased in TRAIL-sensitive glioma cell lines [10]. This further reinforces the reliability of the approach we employed in this study. Albumin, transthyretin, hemopexin, and apolipoprotein A-1 were defined as up-regulated. Since these proteins act as transporters of components in the bloodstream into the brain, our results probably reflect the greater volume of blood that is supplied to gliomas as compared to non-tumor brain tissue. Prohibitin was defined as up-regulated. This stressresponsive mitochondrial gene is known to be downregulated during cellular senescence. Prohibitin was also
reported to be overexpressed in human bladder primary tumors [11], suggesting that up-regulation of this protein may be widely involved in tumorigenesis. SIRT2 showed a significant decrease in expression in gliomas. Recently, it was discovered that SIRT2 is a NAD-dependent tubulin deacetylase and that this protein is involved in the control of mitosis [12]. Gliomas are characterized by marked aneuploidy [13], suggesting that a defective mitotic spindle checkpoint may be one of the causes of chromosomal instability in this type of cancer [14]. Intriguingly, human SIRT2 is located at 19q13.2, which has been shown to be frequently deleted in gliomas [2,3]. These previous insights led us to further examine the possibility that the decreased expression of SIRT2 by some mechanism may play an important role in glioma tumorigenesis. In addition to SIRT2, we found a cluster of cytoskeleton-related proteins that are differentially regulated in gliomas. First, CRMP (collapsin response mediator protein)-4 was up-regulated. CRMP-2, a related protein, was previously reported to be involved in the dynamics of microtubules [15]. Second, profilin 2 and neurocalcin d, which both bind to actin and affect cytoskeletal
562
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
Table 1 List of proteins having altered expression levels in human gliomas Spot
A1 A2 A3 A4 A5 A6
Protein
Probability
Sequence coverage (%)
NCBI Accession Nos.
Theoretical value
Experimental value
pI
kDa
pI
kDa
Fold change
7.1E + 01 2.9E + 02 1.4E + 02 1.0E + 02 7.5E + 01 1.4E + 02
19 58 26 15 36 57
NP_001378 NP_000468 NP_000604 NP_003906 NP_002625 NP_004568
6.0 6.1 6.6 5.5 5.6 5.5
62 71 52 60 30 25
6.2 6.0 5.7 5.6 5.5 5.5
60 67 69 61 30 27
1.2E + 02 2.1E + 02 7.4E + 01
87 67 32
NP_000362 NP_000030 NP_000745
5.3 5.6 5.3
13 31 30
5.5 5.3 5.3
18 27 30
5.5 (5/5) 9.1 (5/5) 9.4 (5/5)
1.4E + 02
69
NP_001437
5.2
15
5.3
16
4.3 (5/5)
A11
CRMP4 Albumin Hemopexin Copine I Prohibitin Phosphoserine phosphatase Transthyretin Apolipoprotein A-I Catechol-Omethyltransferase Fatty acid binding protein 7 Neurocalcin d
6.2E + 01
16
NP_114430
5.4
22
5.2
20
15.0 (5/5)
B1 B2 B3 B4
SIRT2 Profilin 2, isoform a UCH-L1 PEA-15
1.8E + 02 1.0E + 02 7.5E + 01 1.3E + 01
40 34 32 48
NP_085096 NP_444252 NP_004172 Q15121
6.0 6.6 5.3 4.9
40 15 23 15
6.2 6.0 5.3 5.1
42 16 28 17
A7 A8 A9 A10
5.8 8.5 38.7 2.8 6.3 22.0
(5/5) (4/5) (5/5) (4/5) (5/5) (5/5)
0.17 0.30 0.24 0.19
(3/5) (3/5) (3/5) (5/5)
The table shows proteins that were defined as significantly up- or down-regulated in gliomas. Fold differences were calculated using mean values of glioma tissues that exhibited significant change of the proteins and those of non-tumor tissues. The number of glioma tissues that exhibited significant change for each protein is also indicated.
structure, decreased and increased, respectively [16,17]. Third, domain analysis using PROSITE (http://kr.expasy.org/prosite/) revealed that PEA-15, a protein downregulated in gliomas, has two putative domains that can bind to microtubule, supporting the hypothesis that the cytoskeleton has some association with the ERK MAP kinase pathway. Taken together, these data suggest that the alteration of the cytoskeletal network may be involved in glioma tumorigenesis. Northern blot analysis of SIRT2 gene We performed Northern blot analysis of the SIRT2 gene on 17 gliomas (grades II–IV) and three normal tissues to determine whether the decrease of expression is observable in the mRNA level. As shown in Fig. 2, a dramatic decrease or even loss of mRNA expression was observed in 12 out of 17 gliomas. On the other hand, non-tumor brain tissues exhibited comparable levels of SIRT2 mRNA transcripts. It should be noted that eight out of 10 grade VI gliomas exhibited barely detectable expression of SIRT2 mRNA, indicating that decreased expression of SIRT2 correlates with the progression of glioma. Furthermore, SIRT2 mRNA significantly decreased in all the human glioma cell lines that we examined, as compared to a glioma sample, G404, whose SIRT2 mRNA expression level was comparable to that of non-tumor brain tissues (Fig. 2, upper and lower panels). This result shows that down-regulation of SIRT2 protein in gliomas is at least partially caused by the decrease of SIRT2 mRNA levels, which in turn may
be a consequence of the high frequency of 9q13.2-4 deletions in human gliomas [2,3]. Ectopic expression of SIRT2 protein in glioma cell lines leads to relocation of microtubule network It has been reported that SIRT2 is a NAD-dependent tubulin deacetylase and that it colocalizes with microtubules in HeLa cells [12]. Thus, we investigated whether this was also the case in gliomas. Because antiSIRT2 antibody was not available, we constructed an EGFP-tagged SIRT2 expression vector and transfected this DNA into HTB14, a glioma cell line which shows barely detectable expression of SIRT2 mRNA (Fig. 2) in order to observe the localization of tubulin network and SIRT2 protein. As shown in Fig. 3, EGFP-SIRT2 was visualized as fine grains predominantly in cytoplasm and to a much lesser degree, in the nucleus as reported in the previous paper [12]. A control expression vector containing only EGFP showed a diffused distribution of green fluorescent signals both in the cytoplasm and nucleus (data not shown). Unlike the previous report stating complete colocalization of a-tubulin and SIRT2 protein in human fibroblast [12], a-tubulin and SIRT2 protein exhibited a partial colocalization in HTB14 cells. Interestingly, in non-transfected HTB14 cells (essentially SIRT2-negative), microtubules were concentrated in the perinuclear compartment, whereas microtubules were rather diffusely localized throughout the cytoplasm in EGFPSIRT2-transfected HTB14 cells (Fig. 3). A similar result
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
563
Fig. 2. Northern blot analysis of SIRT2 mRNA in non-tumor and glioma tissues, and glioma cell lines. Total RNA (5 lg/lane) was electrophoresed in a formaldehyde gel, blotted onto a nylon membrane, and then hybridized with SIRT2 cDNA probe. Ethidium bromide staining of the gel before blotting was undertaken to monitor RNA loading.
Fig. 3. Distribution of EGFP-SIRT2 and the microtubule network. Localization of exogenously expressed EGFP-SIRT2 (left) and immunolocalization of endogenous a-tubulin (right) in HTB14 human glioma cells. EGFP-positive (upper portion) and EGFP-negative (lower portion) cells in the same culture-disc were analyzed. In EGFP-positive cells, the same field was analyzed for the expression of SIRT2 and a-tubulin.
was also observed in U251MG (data not shown), another human glioma cell line which shows barely detectable expression of SIRT2 mRNA (Fig. 2). Taken together with the previous reports that SIRT2 deacetylates tubulin [12], this result shows that SIRT2 may regulate the location of tubulin network possibly through the deacetylation of tubulin.
Inhibition of colony formation by ectopic expression of SIRT2 in glioma cell lines The reduction of SIRT2 expression in gliomas and a previous report that SIRT2 is located on 19q13.2 prompted us to test the possibility that SIRT2 may play a role in tumor suppression. For this, HTB14 cells were
564
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
Discussion
Fig. 4. Effect of SIRT2 on colony formation. HTB14 cells were transfected with pEGFP-C1 or pEGFP-SIRT2 which both carry the neomycin resistance gene. After 12 days in selective medium, the G418resistant colonies were counted and GFP signals were visualized. The number of G418-resistant colonies and that of EGFP-positive colonies are shown as black and white bars, respectively. Data are shown as means (bars, standard deviation) (n ¼ 4). *p < 0:05 by two-tailed Student’s t test.
transfected with EGFP-SIRT2 or EGFP expression vector which carries a neomycin resistance gene, and colony-formation assays were then performed in the presence of G418. As shown in Fig. 4, transfection of EGFP-SIRT2 caused a 42% reduction in the number of G418-resistant colonies observed, relative to that of the EGFP vector control. Eighteen hours after transfection, the percentage of EGFP-positive cells and the intensity of the EGFP signals in EGFP-SIRT2 and EGFP-transfected cells were comparable (T. Inoue and H. Yamada, unpublished observation), demonstrating that the transfection efficiency and expression levels of EGFP-SIRT2 and EGFP proteins were comparable during the early stage of transfection. Intriguingly, 12 days after transfection, only 10% of G418-resistant colonies obtained were EGFP-positive, while 68% of those obtained by transfection of EGFP alone were EGFP-positive. The same trend was observed at 48 and 72 h after the transfection. This striking difference in percentage of G418 resistant cells to EGFP-positive cells after G418selection shows that colony formation in cells containing transfected SIRT2 is compromised. Consistent results were obtained in several experiments using HTB14 and U251MG cells (data not shown). Taken together, SIRT2 may have a role in growth suppression of glioma cells.
In this study, we analyzed differentially expressed proteins in human gliomas using a proteomics-based approach. To our knowledge, this is the first report of the use of a proteomics-based approach for the study of human gliomas. We identified 11 proteins that are overexpressed and four proteins that are down-regulated in this type of cancer. One of the down-regulated proteins was identified as SIRT2 tubulin deacetylase which is located on 19q13.2, a region whose frequent deletion in gliomas was previously reported [2,3]. Importantly, expression levels of SIRT2 mRNA were also significantly decreased in 12 out of 17 gliomas (Fig. 2). The decreased expression of SIRT2 mRNA must have contributed to the decreased expression of SIRT2 protein in gliomas. At present, we are conducting studies to determine what chromosomal alteration (deletion, abnormal methylation or inactivating mutation among others) may have caused the decreased expression of SIRT2. We also observed that ectopic expression of SIRT2 caused growth suppression in glioma cell lines that had barely detectable endogenous expression of SIRT2, raising the possibility that SIRT2 may act as a tumor suppressor in glioma cells. Although it is not clear how SIRT2 suppresses the growth of glioma cell lines, our study and previous reports suggest the possibility that SIRT2 performs this function through the regulation of microtubules. Intriguingly, in HTB14 glioma cells, overexpression of SIRT2 resulted in altered localization of microtubules. It is also known that high-grade gliomas are invasive into the surrounding brain tissues [18]. It is not clear whether acetylation status of tubulin is involved in mitotic checkpoint or cell invasion. However recent reports by Dryden et al. [19] demonstrated that regulation of SIRT2 expression levels is important for the control of mitosis. In their study, the amount of SIRT2 protein increased during mitosis and overexpression of SIRT2 caused a marked extension of the mitotic phase. We also observed a significant accumulation of U251MG cells at G2/M phase after transfection of SIRT2 (H. Yamada, unpublished observation). It is therefore of importance to study whether the absence of SIRT2 will cause a corresponding change on tubulin acetylation. If it is indeed the case, this change in tubulin acetylation will most likely affect mitotic checkpoint in gliomas. This would further clarify the role of SIRT2 as a potential tumor suppressor gene in gliomas. Mutation analysis has suggested that mutations in mitotic spindle checkpoint genes such as hBUB1 and hBUB3 do not significantly contribute to the observed chromosomal instability in glioblastoma [20] unlike in colorectal cancer [21]. Therefore, we speculate that the alteration of cytoskeleton may more directly contribute
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
to chromosome instability and invasiveness in highgrade gliomas. Furthermore, it should be noticed that other cytoskeleton proteins were defined as altered in this study. Employing a microarray-based approach, Rickman et al. [22] also reported that several genes coding for cytoskeletal proteins like tubulin b4, tubulin c1, myosin light polypeptide kinase, caldesmon 1, and filamin Aa were up-regulated in high-grade gliomas as compared to low-grade gliomas. Our finding raises the possibility that the modulation of cytoskeleton may be involved in the pathogenesis of gliomas. To determine whether cytoskeletal proteins indeed play a role in gliomas, our group is currently conducting studies to determine whether ectopic expression of SIRT2 protein affects the anchorage-independent growth and invasiveness of human glioma cells. Apart from cytoskeletal proteins, we found several proteins that are differentially regulated in gliomas. UCH-L1 and catechol-O-methyltransferase have been shown to be involved in neurodegenerative diseases such as Parkinson’s disease [23,24]. However, it is not clear whether these proteins are also involved in the tumorigenesis of gliomas. Our study provides new potential diagnostic markers of gliomas and may shed light on the novel link between glioma and neurodegeneration. Several lines of study have shown that retinoids and glucocorticoids are candidate differentiation-inducing agents for gliomas [25,26]. More detailed studies are necessary for comparing protein profiles of transformed gliomas and gliomas that reverted to a normal phenotype by treatment with therapeutic drugs such as treated retinoids and glucocorticoids, or among gliomas at different stages. This will further strengthen the link between the proteins listed in this study and the pathogenesis of gliomas. Furthermore, such protein profile comparisons may also uncover novel genes that may be directly involved in the pathogenesis of glial tumors.
Acknowledgments We thank M. Kameyama, H. Yamada, and I. Kishimoto for technical assistance. We also thank R. Nishigaki for computing assistance. This work was supported in part by grants from the Ministry of Health, Labour and Welfare and the Ministry of Education, Culture, Sports, Science and Technology of Japan. C.G.T.T. is supported by the Japanese Government (Monbukagakusho) Scholarship.
References [1] J. Fueyo, C. Gomez-Manzano, W.K. Yung, A.P. Kyritsis, The functional role of tumor suppressor genes in gliomas: clues for future therapeutic strategies, Neurology 51 (1998) 1250–1255. [2] B.K. Rasheed, R.N. Wiltshire, S.H. Bigner, D.D. Bigner, Molecular pathogenesis of malignant gliomas, Curr. Opin. Oncol. 11 (1999) 162–167.
565
[3] A. von Deimling, J. Nagel, B. Bender, D. Lenartz, J. Schramm, D.N. Louis, O.D. Wiestler, Deletion mapping of chromosome 19 in human gliomas, Int. J. Cancer 57 (1994) 676–680. [4] D.A. Vandeputte, C.B. Meije, M. van Dartel, S. Leenstra, H. IJlst-Keizers, P.K. Das, D. Troost, D.A. Bosch, F. Baas, T.J. Hulsebos, GOA, a novel gene encoding a ring finger B-box coiledcoil protein, is overexpressed in astrocytoma, Biochem. Biophys. Res. Commun. 286 (2001) 574–579. [5] E.M. Michiels, E. Oussoren, M. Van Groenigen, E. Pauws, P.M. Bossuyt, P.A. Voute, F. Baas, Genes differentially expressed in medulloblastoma and fetal brain, Physiol. Genomics 1 (1999) 83– 91. [6] R. Nishigaki, T. Shinohara, T. Toda, A. Omori, S. Ichinose, M. Itoh, Y. Shirayoshi, A. Kurimasa, M. Oshimura, An extra human chromosome 21 reduces mlc-2a expression in chimeric mice and Down syndrome, Biochem. Biophys. Res. Commun. 295 (2002) 112–118. [7] T. Toda, K. Kaji, N. Kimura, TMIG-2DPAGE: a new concept of two-dimensional gel protein database for research on aging, Electrophoresis 19 (1998) 344–348. [8] P. Kleuhues, P.C. Burger, B.W. Scheithauer, The new WHO classification of brain tumors, Brain Pathol. 3 (1993) 255– 268. [9] E. Formstecher, J.W. Ramos, M. Fauquet, D.A. Calderwood, J.C. Hsieh, B. Canton, X.T. Nguyen, J.V. Barnier, J. Camonis, M.H. Ginsberg, H. Chneiweiss, PEA-15 mediates cytoplasmic sequestration of ERK MAP kinase, Dev. Cell 1 (2001) 239–250. [10] C. Hao, F. Beguinot, G. Condorelli, A. Trencia, E.G. Van Meir, V.W. Yong, I.F. Parney, W.H. Roa, K.C. Petruk, Induction and intracellular regulation of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) mediated apoptosis in human malignant glioma cells, Cancer Res. 61 (2001) 1162–1170. [11] M. Asamoto, S.M. Cohen, Prohibitin gene is overexpressed but not mutated in rat bladder carcinomas and cell lines, Cancer Lett. 83 (1994) 201–207. [12] B.J. North, B.L. Marshall, M.T. Borra, J.M. Denu, E. Verdin, The human Sir2 ortholog, SIRT2, is an NAD+-dependent tubulin deacetylase, Mol. Cell 11 (2003) 437–444. [13] S.H. Bigner, B. Vogelstein, Cytogenetics and molecular genetics of malignant gliomas and medulloblastoma, Brain Pathol. 1 (1990) 12–18. [14] D.W. Cleveland, Y. Mao, K.F. Sullivan, Centromeres and kinetochores: from epigenetics to mitotic checkpoint signaling, Cell 112 (2003) 407–421. [15] Y. Gu, Y. Ihara, Evidence that collapsin response mediator protein-2 is involved in the dynamics of microtubules, J. Biol. Chem. 275 (2000) 17917–17920. [16] C.G. dos Remedios, D. Chhabra, M. Kekic, I.V. Dedova, M. Tsubakihara, D.A. Berry, N.J. Nosworthy, Actin binding proteins: regulation of cytoskeletal microfilaments, Physiol. Rev. 83 (2003) 433–473. [17] D. Mornet, A. Bonet-Kerrache, Neurocalcin–actin interaction, Biochim. Biophys. Acta 1549 (2001) 197–203. [18] D.K. Binder, M.S. Berger, Proteases and the biology of glioma invasion, J. Neurooncol. 56 (2002) 149–158. [19] S.C. Dryden, F.A. Nahhas, J.E. Nowak, A.S. Goustin, M.A. Tainsky, Role for human SIRT2 NAD-dependent deacetylase activity in control of mitotic exit in the cell cycle, Mol. Cell. Biol. 23 (2003) 3173–3185. [20] R.M. Reis, M. Nakamura, J. Masuoka, T. Watanabe, S. Colella, Y. Yonekawa, P. Kleihues, H. Ohgaki, Mutation analysis of hBUB1, hBUBR1 and hBUB3 genes in glioblastomas, Acta Neuropathol. 101 (2001) 297–304. [21] R.G. Jaffrey, S.C. Pritchard, C. Clark, G.I. Murray, J. Cassidy, K.M. Kerr, M.C. Nicolson, H.L. McLeod, Genomic instability at
566
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
the BUB1 locus in colorectal cancer, but not in non-small cell lung cancer, Cancer Res. 60 (2000) 4349–4352. [22] D.S. Rickman, M.P. Bobek, D.E. Misek, R. Kuick, M. Blaivas, D.M. Kurnit, J. Taylor, S.M. Hanash, Distinctive molecular profiles of high-grade and low-grade gliomas based on oligonucleotide microarray analysis, Cancer Res. 61 (2001) 6885– 6891. [23] K. Steece-Collier, E. Maries, J.H. Kordower, Etiology of Parkinson’s disease: genetics and environment revisited, Proc. Natl. Acad. Sci. USA 99 (2002) 13972–13974.
[24] A. Gordin, S. Kaakkola, H. Teravainen, Position of COMT inhibition in the treatment of Parkinson’s disease, Adv. Neurol. 91 (2003) 237–250. [25] H. Bouterfa, T. Picht, D. Kess, C. Herbold, E. Noll, P.M. Black, K. Roosen, J.C. Tonn, Retinoids inhibit human glioma cell proliferation and migration in primary cell cultures but not in established cell lines, Neurosurgery 46 (2000) 419–430. [26] C.G. Vedoy, M.C. Sogayar, Isolation and characterization of genes associated with the anti-tumor activity of glucocorticoids, Brain Res. Mol. Brain Res. 106 (2002) 57–69.
Proteomics-based identification of differentially expressed genes in human gliomas: down-regulation of SIRT2 gene Masaharu Hiratsuka,a Toshiaki Inoue,b Tosifusa Toda,c Narimichi Kimura,d Yasuaki Shirayoshi,a Hideki Kamitani,e Takashi Watanabe,e Eisaku Ohama,f Candice G.T. Tahimic,b Akihiro Kurimasa,b and Mitsuo Oshimurag,* a
Department of Molecular and Cell Genetics, Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan b Department of Human Genome Science (Kirin Brewery), Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan c Proteomics Collaboration Research Group, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakaecho, Itabashiku, Tokyo 173-0015, Japan d Cellular Signaling Research Group, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakaecho, Itabashiku, Tokyo 173-0015, Japan e Department of Neurosurgery, Institute of Neurological Sciences, Tottori University, Nishimachi 36-1, Yonago, Tottori 683-8504, Japan f Department of Neuropathology, Institute of Neurological Sciences, Tottori University, Nishimachi 36-1, Yonago, Tottori 683-8504, Japan g Department of Biomedical Science, Regenerative Medicine and Biofunction, Graduate School of Medical Science, Tottori University, Nishimachi 86, Yonago, Tottori 683-8503, Japan Received 4 August 2003
Abstract A number of chromosomal abnormalities including 19q deletions have been associated with the formation of human gliomas. In this study, we employed a proteomics-based approach to identify possible genes involved in glioma tumorigenesis which may serve as potential diagnostic molecular markers for this type of cancer. By comparing protein spots from gliomas and non-tumor tissues using two-dimensional (2D) gel electrophoresis, we identified 11 up-regulated proteins and four down-regulated proteins in gliomas. Interestingly, we also discovered that a group of cytoskeleton-related proteins are differentially regulated in gliomas, suggesting the involvement of cytoskeleton modulation in glioma pathogenesis. We then focused on the cytoskeleton-related protein, SIRT2 (sirtuin homologue 2) tubulin deacetylase, which was down-regulated in gliomas. SIRT2 is located at 19q13.2, a region known to be frequently deleted in human gliomas. Subsequent Northern blot analysis revealed that RNA expression of SIRT2 was dramatically diminished in 12 out of 17 gliomas and glioma cell lines, in agreement with proteomic data. Furthermore, ectopic expression of SIRT2 in glioma cell lines led to the perturbation of the microtubule network and caused a remarkable reduction in the number of stable clones expressing SIRT2 as compared to that of a control vector in colony formation assays. These results suggest that SIRT2 may act as a tumor suppressor gene in human gliomas possibly through the regulation of microtubule network and may serve as a novel molecular marker for gliomas. Additional proteins were also identified, whose function in gliomas was previously unsuspected. Ó 2003 Elsevier Inc. All rights reserved. Keywords: Glioma; Proteomics; SIRT2; Cytoskeleton; 19q13; Growth inhibition
Gliomas are the most common type of primary malignant brain tumor and various genetic alterations have been identified in human glioma tumorigenesis. Gene amplification has been observed in the EGF receptor, CDK4, and MDM2 genes, while deletion or inactivating *
Corresponding author. Fax: +81-859-34-8134. E-mail address: [email protected] (M. Oshimura). 0006-291X/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2003.08.029
mutations have been reported for p53 (17p), RB (13q), p16 (9p), PTEN (10q), and DMBT1 (10q) [1]. Moreover, frequent losses of heterozygosity (LOH) for other loci like 1p, 11p, 19q, or 22q in gliomas have been reported [2,3]. In order to provide more insight into the biological behavior of gliomas and identify new targets for therapy, a number of studies have been conducted to identify genes in regions that are genetically altered in this type of cancer. Experiments were also conducted to
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
further identify upstream and downstream target genes that are differentially regulated in gliomas. For example, by using serial analysis of gene expression (SAGE), GOA, a ring finger-B box-coiled coil protein, was identified as an overexpressed protein with no observable gene amplification in human gliomas [4]. ZIC1 and OTX2 genes were also identified to be highly expressed in gliomas using the same method [5]. In this study, we used a proteomics-based technique [6,7] for the identification of differentially regulated proteins in human gliomas. We successfully identified 11 proteins that were up-regulated and four proteins that were down-regulated in gliomas. Interestingly, a group of cytoskeleton-related proteins were observed to be differentially regulated in gliomas. Among these cytoskeleton-related proteins, SIRT2 (sirtuin homologue 2) was observed to be down-regulated. Because SIRT2 is located on 19q13.2, a region known to frequently undergo LOH in human gliomas, we suspected that SIRT2 may be involved in glioma tumorigenesis.
Materials and methods Brain tissues. Brain tumor tissues and non-tumor tissues were obtained from Tottori University Hospital with informed consent and approved by an Institutional Review Committee. Tissues were flashfrozen in liquid nitrogen immediately after surgical removal and stored at )80 °C. Diagnosis of glioma was confirmed histopathologically. Gliomas were then classified according to the WHO grading scheme for central nervous system tumors [8]. Cell culture, transfection of EGFP-SIRT2 expression plasmid, and colony-formation assays. Human glioma cell lines were obtained from ATCC and cultured in DMEM containing 10% calf serum (Invitrogen). A construct encoding fusions of SIRT2 with EGFP was generated by connecting the C-terminus of the fluorescent protein moiety, in-frame, to the N-terminus of full length SIRT2 cDNA using the BamHI/EcoRI sites of pEGFP-C1 (Clontech). SIRT2 cDNA was purchased from ResGen. For transient transfection, sub-confluent cultures of cells in 15 mmdishes (Matsunami, Japan) were transfected with a total of 0.5 lg DNA using Lipofectamine 2000 (Invitrogen) following manufacturer’s protocol. For colony-formation assays, cells in 35-mm dishes were replated at 1:40 after transfection and cultured in medium containing 800 g/ml of G418 (Invitrogen). The medium was changed every 3 days. Twelve days after transfection, the number of colonies was scored. EGFP signals were then visualized using ECLIPSE TE300 (Nikon). Immunohistochemistry. For immunofluorescence detection of tubulin network in transfected cells, cells were fixed in 4% paraformaldehyde/PBS, and immunostained using anti-a-tubulin antibody (1:2000 in 3% skim milk/PBS, Sigma, B-5-1-2) followed by detection with Cy5-labeled secondary antibody (1:100 in 3% skim milk/PBS, Chemicon, AP192F). Simultaneous detection of EGFP-tagged SIRT2 and tubulin network was performed using Bio-Rad Radiance 2000 (Bio-Rad). Protein separation by two-dimensional gel electrophoresis and image analysis. Protein extraction and two-dimensional (2D) electrophoresis were performed as previously reported [7, see also the URL, http:// proteome.tmig.or.jp/2D/2DE_method.html] with minor modifications. Briefly, each 100 mg of tissue sample was crushed in liquid nitrogen
559
and lysed with 300 ll of 5 M urea, 2 M thiourea, 2% CHAPS, 2% SB310, 1% Pharmalyte (pH 3–10, Amersham Biosciences), 1% DTT, and 1% proteinase inhibitor cocktail (Sigma) with sonication. One hundred micrograms of protein was loaded onto gel strips with immobilized pH gradient (pH 4–7, 18 cm, Amersham Biosciences) and isoelectric focusing was performed on a CoolPhoreStar model 3610 (Anatech, Japan). After isoelectric focusing, strips were equilibrated with 50 mM Tris–HCl (pH 6.8), 6 M urea, 2 M thiourea, 2% SDS, 30% glycerol, and 1% DTT for 30 min and then with the same solution containing 4% iodeacetamide instead of DTT for 30 min. Each equilibrated strip was mounted on a 7.5% SDS–polyacrylamide gel and fixed with sharktooth combs. SDS–PAGE was performed on a vertical format using a Tricine buffer system. After second dimension electrophoresis, gels were fixed in 10% methanol and 6% acetic acid for 30 min and protein spots were visualized by SYPRO Ruby (Molecular Probes) staining following the manufacturer’s recommendations. The SYPRO Ruby-stained images were scanned at an excitation wavelength of 488 nm and emission wavelengths of over 550 nm were collected on a Molecular Imager FX (Bio-Rad). Statistical analysis and quantitation of protein expression were conducted using PDQuest software ver7.1 (Bio-Rad). To minimize computational errors resulting from spot dispersion among gels, the volumes of the spots were normalized to the volumes of landmark proteins used for internal standardization [S1; antioxidant protein 2, S2; Rho GDP dissociation inhibitor a, S3; glyoxalase I, and S4; ‘expressed ubiquitously with strong expression in brain’ (Accession No. gi/20521842)]. Tumors from each of the five patients were analyzed in triplicate. On the other hand, non-tumor samples from each of the six patients were analyzed in duplicate (for three samples) or triplicate (for the remaining three samples). Only spots having over twofold changes in density after normalization between two populations were defined as altered. Protein identification. Protein spots were excised by ProteomeWorks Spot Cutter (Bio-Rad) followed by in-gel digestion with trypsin (Promega) according to manufacturer’s specifications. The digested peptide was directly mixed with an equal volume of 10 mg/ml a-cyano4-hydroxycinnamic acid and peptide mass spectra were obtained on an AXIMA-CFR MALDI-TOF-MASS (Shimadzu, Japan) platform. Peptide mass mapping was performed using Mascot Search (Matrix Science). RNA preparation and Northern blotting. Total RNA was prepared using Isogen (Nippongene) from frozen tissues or cultured cells. Electrophoresis was conducted using 5 lg of total RNA loaded per well of a 1.5% agarose/formaldehyde gel. The resolved RNA was then blotted onto a Hybond-N membrane (Amersham Biosciences) and then hybridized with SIRT2 cDNA (NCBI Accession No. NM_030593) probe spanning nucleotides 713–1408. Prior to hybridization, the probe was radioactively labeled using Random Primed DNA Labeling Kit (Roche) in the presence of [a-32 P]dCTP (Amersham Biosciences). RNA loading was verified by ethidium bromide staining of the gel before blotting. Hybridization was performed with 50% formamide, 6 SSC, 0.5% SDS, and 100 lg/ml denatured salmon sperm DNA, and bands were visualized by autoradiography with Xray films (Kodak).
Results 2D gel separation of proteins Glioma tissues derived from five individuals were classified according to the WHO grading system for central nervous system tumors. In this study, gliomas classified as grade II (G201), III (G301 and G302), and IV (G402 and G405) were processed for 2D xelectrophoresis to isolate candidate proteins that are
560
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
differentially regulated in gliomas. We detected about 1000 protein spots on the 2D gels after SYPRO Ruby staining (Fig. 1). The 2D electrophoresis profiles and relative spot intensities obtained from all samples were reproducible when we performed the experiments with duplicates or triplicates. Normalization was also performed using protein internal standards whose corresponding spot intensities remained unchanged between gliomas and normal tissues. Furthermore, only spots
which had over twofold changes in density after normalization were classified as down- or up-regulated. The reproducibility of data from multiple trials, the assignment of a stringent criterion for spot classification, and the use of internal protein standards ensure the reliability of the proteomic data we obtained. Eleven up-regulated proteins and four down-regulated proteins were selected for subsequent analysis by mass spectrometry. Peptide mass-fingerprinting of the
Fig. 1. Two-dimensional protein profiles for human glioma tissue. (A) Proteins were separated on the basis of pI (x-axis) and molecular mass (y-axis). Spots were visualized with SYPRO Ruby gel staining. Numbered spots were defined as altered and then were identified by mass spectrometry. A1-11, B1-4, and S1-4 represent up-regulated, down-regulated, and internal control proteins, respectively. (B) Enlarged portions from 2D gels of human gliomas. (C) Densities of numbered spots in each patient and non-tumor population. Each of the gliomas from five patients was analyzed in triplicate and the density was plotted for each individual (black bars; means standard deviations, n ¼ 3). A total of 15 samples derived from six tumor-free individuals (see Materials and methods) were analyzed and calculated as a population of non-tumor tissues, and the density was plotted in a bar (white bars: means standard deviations, n ¼ 15). *p < 0:05 by two-tailed Student’s t test.
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
561
Fig. 1. (continued)
selected spots and subsequent database search revealed the identity of these proteins as summarized in Table 1. In the succeeding portions, we then attempt to explain the relevance of the listed proteins in gliomas using known information on the nature of these proteins. PEA-15 acts as a modulator of signaling pathways that control cell proliferation and cell death by regulating the subcellular localization of ERK MAP kinase [9]. Our results showing the down-regulation of this protein in gliomas are consistent with the previous report stating that PEA-15 expression is decreased in TRAIL-sensitive glioma cell lines [10]. This further reinforces the reliability of the approach we employed in this study. Albumin, transthyretin, hemopexin, and apolipoprotein A-1 were defined as up-regulated. Since these proteins act as transporters of components in the bloodstream into the brain, our results probably reflect the greater volume of blood that is supplied to gliomas as compared to non-tumor brain tissue. Prohibitin was defined as up-regulated. This stressresponsive mitochondrial gene is known to be downregulated during cellular senescence. Prohibitin was also
reported to be overexpressed in human bladder primary tumors [11], suggesting that up-regulation of this protein may be widely involved in tumorigenesis. SIRT2 showed a significant decrease in expression in gliomas. Recently, it was discovered that SIRT2 is a NAD-dependent tubulin deacetylase and that this protein is involved in the control of mitosis [12]. Gliomas are characterized by marked aneuploidy [13], suggesting that a defective mitotic spindle checkpoint may be one of the causes of chromosomal instability in this type of cancer [14]. Intriguingly, human SIRT2 is located at 19q13.2, which has been shown to be frequently deleted in gliomas [2,3]. These previous insights led us to further examine the possibility that the decreased expression of SIRT2 by some mechanism may play an important role in glioma tumorigenesis. In addition to SIRT2, we found a cluster of cytoskeleton-related proteins that are differentially regulated in gliomas. First, CRMP (collapsin response mediator protein)-4 was up-regulated. CRMP-2, a related protein, was previously reported to be involved in the dynamics of microtubules [15]. Second, profilin 2 and neurocalcin d, which both bind to actin and affect cytoskeletal
562
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
Table 1 List of proteins having altered expression levels in human gliomas Spot
A1 A2 A3 A4 A5 A6
Protein
Probability
Sequence coverage (%)
NCBI Accession Nos.
Theoretical value
Experimental value
pI
kDa
pI
kDa
Fold change
7.1E + 01 2.9E + 02 1.4E + 02 1.0E + 02 7.5E + 01 1.4E + 02
19 58 26 15 36 57
NP_001378 NP_000468 NP_000604 NP_003906 NP_002625 NP_004568
6.0 6.1 6.6 5.5 5.6 5.5
62 71 52 60 30 25
6.2 6.0 5.7 5.6 5.5 5.5
60 67 69 61 30 27
1.2E + 02 2.1E + 02 7.4E + 01
87 67 32
NP_000362 NP_000030 NP_000745
5.3 5.6 5.3
13 31 30
5.5 5.3 5.3
18 27 30
5.5 (5/5) 9.1 (5/5) 9.4 (5/5)
1.4E + 02
69
NP_001437
5.2
15
5.3
16
4.3 (5/5)
A11
CRMP4 Albumin Hemopexin Copine I Prohibitin Phosphoserine phosphatase Transthyretin Apolipoprotein A-I Catechol-Omethyltransferase Fatty acid binding protein 7 Neurocalcin d
6.2E + 01
16
NP_114430
5.4
22
5.2
20
15.0 (5/5)
B1 B2 B3 B4
SIRT2 Profilin 2, isoform a UCH-L1 PEA-15
1.8E + 02 1.0E + 02 7.5E + 01 1.3E + 01
40 34 32 48
NP_085096 NP_444252 NP_004172 Q15121
6.0 6.6 5.3 4.9
40 15 23 15
6.2 6.0 5.3 5.1
42 16 28 17
A7 A8 A9 A10
5.8 8.5 38.7 2.8 6.3 22.0
(5/5) (4/5) (5/5) (4/5) (5/5) (5/5)
0.17 0.30 0.24 0.19
(3/5) (3/5) (3/5) (5/5)
The table shows proteins that were defined as significantly up- or down-regulated in gliomas. Fold differences were calculated using mean values of glioma tissues that exhibited significant change of the proteins and those of non-tumor tissues. The number of glioma tissues that exhibited significant change for each protein is also indicated.
structure, decreased and increased, respectively [16,17]. Third, domain analysis using PROSITE (http://kr.expasy.org/prosite/) revealed that PEA-15, a protein downregulated in gliomas, has two putative domains that can bind to microtubule, supporting the hypothesis that the cytoskeleton has some association with the ERK MAP kinase pathway. Taken together, these data suggest that the alteration of the cytoskeletal network may be involved in glioma tumorigenesis. Northern blot analysis of SIRT2 gene We performed Northern blot analysis of the SIRT2 gene on 17 gliomas (grades II–IV) and three normal tissues to determine whether the decrease of expression is observable in the mRNA level. As shown in Fig. 2, a dramatic decrease or even loss of mRNA expression was observed in 12 out of 17 gliomas. On the other hand, non-tumor brain tissues exhibited comparable levels of SIRT2 mRNA transcripts. It should be noted that eight out of 10 grade VI gliomas exhibited barely detectable expression of SIRT2 mRNA, indicating that decreased expression of SIRT2 correlates with the progression of glioma. Furthermore, SIRT2 mRNA significantly decreased in all the human glioma cell lines that we examined, as compared to a glioma sample, G404, whose SIRT2 mRNA expression level was comparable to that of non-tumor brain tissues (Fig. 2, upper and lower panels). This result shows that down-regulation of SIRT2 protein in gliomas is at least partially caused by the decrease of SIRT2 mRNA levels, which in turn may
be a consequence of the high frequency of 9q13.2-4 deletions in human gliomas [2,3]. Ectopic expression of SIRT2 protein in glioma cell lines leads to relocation of microtubule network It has been reported that SIRT2 is a NAD-dependent tubulin deacetylase and that it colocalizes with microtubules in HeLa cells [12]. Thus, we investigated whether this was also the case in gliomas. Because antiSIRT2 antibody was not available, we constructed an EGFP-tagged SIRT2 expression vector and transfected this DNA into HTB14, a glioma cell line which shows barely detectable expression of SIRT2 mRNA (Fig. 2) in order to observe the localization of tubulin network and SIRT2 protein. As shown in Fig. 3, EGFP-SIRT2 was visualized as fine grains predominantly in cytoplasm and to a much lesser degree, in the nucleus as reported in the previous paper [12]. A control expression vector containing only EGFP showed a diffused distribution of green fluorescent signals both in the cytoplasm and nucleus (data not shown). Unlike the previous report stating complete colocalization of a-tubulin and SIRT2 protein in human fibroblast [12], a-tubulin and SIRT2 protein exhibited a partial colocalization in HTB14 cells. Interestingly, in non-transfected HTB14 cells (essentially SIRT2-negative), microtubules were concentrated in the perinuclear compartment, whereas microtubules were rather diffusely localized throughout the cytoplasm in EGFPSIRT2-transfected HTB14 cells (Fig. 3). A similar result
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
563
Fig. 2. Northern blot analysis of SIRT2 mRNA in non-tumor and glioma tissues, and glioma cell lines. Total RNA (5 lg/lane) was electrophoresed in a formaldehyde gel, blotted onto a nylon membrane, and then hybridized with SIRT2 cDNA probe. Ethidium bromide staining of the gel before blotting was undertaken to monitor RNA loading.
Fig. 3. Distribution of EGFP-SIRT2 and the microtubule network. Localization of exogenously expressed EGFP-SIRT2 (left) and immunolocalization of endogenous a-tubulin (right) in HTB14 human glioma cells. EGFP-positive (upper portion) and EGFP-negative (lower portion) cells in the same culture-disc were analyzed. In EGFP-positive cells, the same field was analyzed for the expression of SIRT2 and a-tubulin.
was also observed in U251MG (data not shown), another human glioma cell line which shows barely detectable expression of SIRT2 mRNA (Fig. 2). Taken together with the previous reports that SIRT2 deacetylates tubulin [12], this result shows that SIRT2 may regulate the location of tubulin network possibly through the deacetylation of tubulin.
Inhibition of colony formation by ectopic expression of SIRT2 in glioma cell lines The reduction of SIRT2 expression in gliomas and a previous report that SIRT2 is located on 19q13.2 prompted us to test the possibility that SIRT2 may play a role in tumor suppression. For this, HTB14 cells were
564
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
Discussion
Fig. 4. Effect of SIRT2 on colony formation. HTB14 cells were transfected with pEGFP-C1 or pEGFP-SIRT2 which both carry the neomycin resistance gene. After 12 days in selective medium, the G418resistant colonies were counted and GFP signals were visualized. The number of G418-resistant colonies and that of EGFP-positive colonies are shown as black and white bars, respectively. Data are shown as means (bars, standard deviation) (n ¼ 4). *p < 0:05 by two-tailed Student’s t test.
transfected with EGFP-SIRT2 or EGFP expression vector which carries a neomycin resistance gene, and colony-formation assays were then performed in the presence of G418. As shown in Fig. 4, transfection of EGFP-SIRT2 caused a 42% reduction in the number of G418-resistant colonies observed, relative to that of the EGFP vector control. Eighteen hours after transfection, the percentage of EGFP-positive cells and the intensity of the EGFP signals in EGFP-SIRT2 and EGFP-transfected cells were comparable (T. Inoue and H. Yamada, unpublished observation), demonstrating that the transfection efficiency and expression levels of EGFP-SIRT2 and EGFP proteins were comparable during the early stage of transfection. Intriguingly, 12 days after transfection, only 10% of G418-resistant colonies obtained were EGFP-positive, while 68% of those obtained by transfection of EGFP alone were EGFP-positive. The same trend was observed at 48 and 72 h after the transfection. This striking difference in percentage of G418 resistant cells to EGFP-positive cells after G418selection shows that colony formation in cells containing transfected SIRT2 is compromised. Consistent results were obtained in several experiments using HTB14 and U251MG cells (data not shown). Taken together, SIRT2 may have a role in growth suppression of glioma cells.
In this study, we analyzed differentially expressed proteins in human gliomas using a proteomics-based approach. To our knowledge, this is the first report of the use of a proteomics-based approach for the study of human gliomas. We identified 11 proteins that are overexpressed and four proteins that are down-regulated in this type of cancer. One of the down-regulated proteins was identified as SIRT2 tubulin deacetylase which is located on 19q13.2, a region whose frequent deletion in gliomas was previously reported [2,3]. Importantly, expression levels of SIRT2 mRNA were also significantly decreased in 12 out of 17 gliomas (Fig. 2). The decreased expression of SIRT2 mRNA must have contributed to the decreased expression of SIRT2 protein in gliomas. At present, we are conducting studies to determine what chromosomal alteration (deletion, abnormal methylation or inactivating mutation among others) may have caused the decreased expression of SIRT2. We also observed that ectopic expression of SIRT2 caused growth suppression in glioma cell lines that had barely detectable endogenous expression of SIRT2, raising the possibility that SIRT2 may act as a tumor suppressor in glioma cells. Although it is not clear how SIRT2 suppresses the growth of glioma cell lines, our study and previous reports suggest the possibility that SIRT2 performs this function through the regulation of microtubules. Intriguingly, in HTB14 glioma cells, overexpression of SIRT2 resulted in altered localization of microtubules. It is also known that high-grade gliomas are invasive into the surrounding brain tissues [18]. It is not clear whether acetylation status of tubulin is involved in mitotic checkpoint or cell invasion. However recent reports by Dryden et al. [19] demonstrated that regulation of SIRT2 expression levels is important for the control of mitosis. In their study, the amount of SIRT2 protein increased during mitosis and overexpression of SIRT2 caused a marked extension of the mitotic phase. We also observed a significant accumulation of U251MG cells at G2/M phase after transfection of SIRT2 (H. Yamada, unpublished observation). It is therefore of importance to study whether the absence of SIRT2 will cause a corresponding change on tubulin acetylation. If it is indeed the case, this change in tubulin acetylation will most likely affect mitotic checkpoint in gliomas. This would further clarify the role of SIRT2 as a potential tumor suppressor gene in gliomas. Mutation analysis has suggested that mutations in mitotic spindle checkpoint genes such as hBUB1 and hBUB3 do not significantly contribute to the observed chromosomal instability in glioblastoma [20] unlike in colorectal cancer [21]. Therefore, we speculate that the alteration of cytoskeleton may more directly contribute
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
to chromosome instability and invasiveness in highgrade gliomas. Furthermore, it should be noticed that other cytoskeleton proteins were defined as altered in this study. Employing a microarray-based approach, Rickman et al. [22] also reported that several genes coding for cytoskeletal proteins like tubulin b4, tubulin c1, myosin light polypeptide kinase, caldesmon 1, and filamin Aa were up-regulated in high-grade gliomas as compared to low-grade gliomas. Our finding raises the possibility that the modulation of cytoskeleton may be involved in the pathogenesis of gliomas. To determine whether cytoskeletal proteins indeed play a role in gliomas, our group is currently conducting studies to determine whether ectopic expression of SIRT2 protein affects the anchorage-independent growth and invasiveness of human glioma cells. Apart from cytoskeletal proteins, we found several proteins that are differentially regulated in gliomas. UCH-L1 and catechol-O-methyltransferase have been shown to be involved in neurodegenerative diseases such as Parkinson’s disease [23,24]. However, it is not clear whether these proteins are also involved in the tumorigenesis of gliomas. Our study provides new potential diagnostic markers of gliomas and may shed light on the novel link between glioma and neurodegeneration. Several lines of study have shown that retinoids and glucocorticoids are candidate differentiation-inducing agents for gliomas [25,26]. More detailed studies are necessary for comparing protein profiles of transformed gliomas and gliomas that reverted to a normal phenotype by treatment with therapeutic drugs such as treated retinoids and glucocorticoids, or among gliomas at different stages. This will further strengthen the link between the proteins listed in this study and the pathogenesis of gliomas. Furthermore, such protein profile comparisons may also uncover novel genes that may be directly involved in the pathogenesis of glial tumors.
Acknowledgments We thank M. Kameyama, H. Yamada, and I. Kishimoto for technical assistance. We also thank R. Nishigaki for computing assistance. This work was supported in part by grants from the Ministry of Health, Labour and Welfare and the Ministry of Education, Culture, Sports, Science and Technology of Japan. C.G.T.T. is supported by the Japanese Government (Monbukagakusho) Scholarship.
References [1] J. Fueyo, C. Gomez-Manzano, W.K. Yung, A.P. Kyritsis, The functional role of tumor suppressor genes in gliomas: clues for future therapeutic strategies, Neurology 51 (1998) 1250–1255. [2] B.K. Rasheed, R.N. Wiltshire, S.H. Bigner, D.D. Bigner, Molecular pathogenesis of malignant gliomas, Curr. Opin. Oncol. 11 (1999) 162–167.
565
[3] A. von Deimling, J. Nagel, B. Bender, D. Lenartz, J. Schramm, D.N. Louis, O.D. Wiestler, Deletion mapping of chromosome 19 in human gliomas, Int. J. Cancer 57 (1994) 676–680. [4] D.A. Vandeputte, C.B. Meije, M. van Dartel, S. Leenstra, H. IJlst-Keizers, P.K. Das, D. Troost, D.A. Bosch, F. Baas, T.J. Hulsebos, GOA, a novel gene encoding a ring finger B-box coiledcoil protein, is overexpressed in astrocytoma, Biochem. Biophys. Res. Commun. 286 (2001) 574–579. [5] E.M. Michiels, E. Oussoren, M. Van Groenigen, E. Pauws, P.M. Bossuyt, P.A. Voute, F. Baas, Genes differentially expressed in medulloblastoma and fetal brain, Physiol. Genomics 1 (1999) 83– 91. [6] R. Nishigaki, T. Shinohara, T. Toda, A. Omori, S. Ichinose, M. Itoh, Y. Shirayoshi, A. Kurimasa, M. Oshimura, An extra human chromosome 21 reduces mlc-2a expression in chimeric mice and Down syndrome, Biochem. Biophys. Res. Commun. 295 (2002) 112–118. [7] T. Toda, K. Kaji, N. Kimura, TMIG-2DPAGE: a new concept of two-dimensional gel protein database for research on aging, Electrophoresis 19 (1998) 344–348. [8] P. Kleuhues, P.C. Burger, B.W. Scheithauer, The new WHO classification of brain tumors, Brain Pathol. 3 (1993) 255– 268. [9] E. Formstecher, J.W. Ramos, M. Fauquet, D.A. Calderwood, J.C. Hsieh, B. Canton, X.T. Nguyen, J.V. Barnier, J. Camonis, M.H. Ginsberg, H. Chneiweiss, PEA-15 mediates cytoplasmic sequestration of ERK MAP kinase, Dev. Cell 1 (2001) 239–250. [10] C. Hao, F. Beguinot, G. Condorelli, A. Trencia, E.G. Van Meir, V.W. Yong, I.F. Parney, W.H. Roa, K.C. Petruk, Induction and intracellular regulation of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) mediated apoptosis in human malignant glioma cells, Cancer Res. 61 (2001) 1162–1170. [11] M. Asamoto, S.M. Cohen, Prohibitin gene is overexpressed but not mutated in rat bladder carcinomas and cell lines, Cancer Lett. 83 (1994) 201–207. [12] B.J. North, B.L. Marshall, M.T. Borra, J.M. Denu, E. Verdin, The human Sir2 ortholog, SIRT2, is an NAD+-dependent tubulin deacetylase, Mol. Cell 11 (2003) 437–444. [13] S.H. Bigner, B. Vogelstein, Cytogenetics and molecular genetics of malignant gliomas and medulloblastoma, Brain Pathol. 1 (1990) 12–18. [14] D.W. Cleveland, Y. Mao, K.F. Sullivan, Centromeres and kinetochores: from epigenetics to mitotic checkpoint signaling, Cell 112 (2003) 407–421. [15] Y. Gu, Y. Ihara, Evidence that collapsin response mediator protein-2 is involved in the dynamics of microtubules, J. Biol. Chem. 275 (2000) 17917–17920. [16] C.G. dos Remedios, D. Chhabra, M. Kekic, I.V. Dedova, M. Tsubakihara, D.A. Berry, N.J. Nosworthy, Actin binding proteins: regulation of cytoskeletal microfilaments, Physiol. Rev. 83 (2003) 433–473. [17] D. Mornet, A. Bonet-Kerrache, Neurocalcin–actin interaction, Biochim. Biophys. Acta 1549 (2001) 197–203. [18] D.K. Binder, M.S. Berger, Proteases and the biology of glioma invasion, J. Neurooncol. 56 (2002) 149–158. [19] S.C. Dryden, F.A. Nahhas, J.E. Nowak, A.S. Goustin, M.A. Tainsky, Role for human SIRT2 NAD-dependent deacetylase activity in control of mitotic exit in the cell cycle, Mol. Cell. Biol. 23 (2003) 3173–3185. [20] R.M. Reis, M. Nakamura, J. Masuoka, T. Watanabe, S. Colella, Y. Yonekawa, P. Kleihues, H. Ohgaki, Mutation analysis of hBUB1, hBUBR1 and hBUB3 genes in glioblastomas, Acta Neuropathol. 101 (2001) 297–304. [21] R.G. Jaffrey, S.C. Pritchard, C. Clark, G.I. Murray, J. Cassidy, K.M. Kerr, M.C. Nicolson, H.L. McLeod, Genomic instability at
566
M. Hiratsuka et al. / Biochemical and Biophysical Research Communications 309 (2003) 558–566
the BUB1 locus in colorectal cancer, but not in non-small cell lung cancer, Cancer Res. 60 (2000) 4349–4352. [22] D.S. Rickman, M.P. Bobek, D.E. Misek, R. Kuick, M. Blaivas, D.M. Kurnit, J. Taylor, S.M. Hanash, Distinctive molecular profiles of high-grade and low-grade gliomas based on oligonucleotide microarray analysis, Cancer Res. 61 (2001) 6885– 6891. [23] K. Steece-Collier, E. Maries, J.H. Kordower, Etiology of Parkinson’s disease: genetics and environment revisited, Proc. Natl. Acad. Sci. USA 99 (2002) 13972–13974.
[24] A. Gordin, S. Kaakkola, H. Teravainen, Position of COMT inhibition in the treatment of Parkinson’s disease, Adv. Neurol. 91 (2003) 237–250. [25] H. Bouterfa, T. Picht, D. Kess, C. Herbold, E. Noll, P.M. Black, K. Roosen, J.C. Tonn, Retinoids inhibit human glioma cell proliferation and migration in primary cell cultures but not in established cell lines, Neurosurgery 46 (2000) 419–430. [26] C.G. Vedoy, M.C. Sogayar, Isolation and characterization of genes associated with the anti-tumor activity of glucocorticoids, Brain Res. Mol. Brain Res. 106 (2002) 57–69.