ANTICANCER RESEARCH 24: 1507-1518 (2004)

Differential Expression of the -like Doppel (PRND) in Astrocytomas: A New Molecular Marker Potentially Involved in Tumor Progression

SERGIO COMINCINI1, ANGELICA FACOETTI2, IGOR DEL VECCHIO1, KATELL PEOC’H3, JEAN-LOUIS LAPLANCHE3, LORENZO MAGRASSI4, MAURO CERONI5, LUCA FERRETTI1 and ROSANNA NANO2

1Dipartimento di Genetica e Microbiologia, Università di Pavia, Pavia; 2Dipartimento di Biologia Animale, Università di Pavia and IGM-CNR, Italy; 3Service de Biochimie et de Biologie Moléculaire, Hôpital Lariboisèere, Paris, France; 4I.R.C.C.S. Policlinico S. Matteo, Università di Pavia, Pavia; 5I.R.C.C.S. Istituto C. Mondino, Dipartimento di Scienze Neurologiche, Università di Pavia and Policlinico di Monza, Monza, Italy

Abstract. The expression of the prion (PRNP) and prion like- and membrane localised staining. Our findings suggest that the doppel (PRND) and the presence of the prion PRND gene might be a useful molecular marker in astrocytoma (PrP) and doppel (Dpl) were investigated in human gliomas. progression and in tumor grade definition. Understanding of the The PRNP and PRND expression profiles were evaluated by mechanisms of PRND increased expression might provide real-time reverse transcription-quantitative PCR in low- and insight into the regulatory pathways of glioma development. high-grade astrocytomas, in glioblastoma-derived cell lines and in non-glial tumor specimens. The presence of PrP and Dpl Astrocytomas are the most common primary central nervous proteins and their cellular localization were evaluated by Western system (CNS) tumors in humans. These tumors are blot and immunohistochemistry. High levels of PRNP expression histopathologically divided into three grades of malignancy: were found in all tumoral samples studied. Unlike the non- low-grade astrocytoma (LGA, WHO grade II), anaplastic tumoral controls, PRND was aberrantly expressed in astrocytoma (AA, grade III) and glioblastoma multiforme glioblastoma multiforme and in two glioblastoma multiforme- (GBM, grade IV) (1). Low-grade astrocytoma comprises an derived cell lines, even in the absence of the PRND gene heterogeneous class of non-infiltrating tumors. However, the amplification. PRND expression was directly related to extensive heterogeneity of glial tumors has made their malignancy of the tumor: highest in glioblastoma multiforme, histological classification rather difficult (2). This fact is lower in anaplastic astrocytoma and even lower in the low-grade relevant on clinical grounds since the prognosis for patients astrocytoma samples. High levels of PRND were also found in with gliomas correlates with tumor grade and patient age. In non-glial malignant tumor samples, such as gastric contrast to the long-standing and well-defined adenocarcinoma and anaplastic meningioma. Western blot histopathological criteria, the underlying molecular and analysis confirmed the PrP and Dpl expression, displaying genetic basis for gliomas is only just emerging (3). In variability in the electrophoretic patterns. Immunohistochemical particular, several genes have been identified as being analysis revealed a diffuse cytoplasmatic Dpl distribution in associated with tumorigenesis and anaplastic progression, but different astrocytic neoplastic cells, in infiltrating their contribution to the tumor classification is limited (4). In and in blood vessel endothelial cells. Of note, Dpl reactivity was general, most of these genes affect either different from that of the PrP, since PrP showed typical Golgi pathways activated by tyrosine receptors, or arrest pathways involving different regulators (5). Specifically for GBM, differential expression markers have been identified, such as the mutually exclusive TP53 and EGFR Correspondence to: Sergio Comincini, Dipartimento di Genetica e gene alterations (6,7). Furthermore, most GBMs express at Microbiologia, via Ferrata 1, Università di Pavia, 27100 Pavia, Italy. Fax: +39-382-528496, e-mail: [email protected] different levels astrocytic differentiation markers such as the GFAP and S-100B proteins (8). New methodologies, mainly Key Words: Prion, doppel , glioblastoma multiforme. based on microarrays analysis on glioma expression libraries,

0250-7005/2004 $2.00+.40 1507 ANTICANCER RESEARCH 24: 1507-1518 (2004) have enabled the characterization of several genes, showing in formalin and embedded in paraffin (for immunohistochemistry), differential expression patterns (9,10). At the cellular level, frozen in liquid nitrogen and stored at -80ÆC for DNA, RNA and during tumorigenesis, newly formed tumor cells expand and proteins extraction. Classification of astrocytomas into GBM infiltrate the surrounding tissues by changing their (grade IV, 12 samples), AA (grade III, 6 samples) and LGA (grade II, 4 samples) was performed according to the guidelines published intracellular/extracellular architecture (11). There is evidence by the WHO (1). The analyzed biopsies also comprised 3 that alteration of adhesion molecule expression could peritumoral GBM specimens and 2 normal brain tissues from the modulate the capacity for cell-cell or cell-matrix interaction, frontal cortex without pathological findings. Further controls which mediates the invasion and growth of neoplastic tissue consisted of total RNA from healthy fetal and adult human brain (12). For this reason, a multitude of membrane glycoproteins, (pooled RNA from tissues of 70 trauma patients), purchased from including cell-cell adhesion molecules (CAMs) and Clontech (Palo Alto, CA, USA). Non-glial highly malignant tumor substratum adhesion molecules (SAMs), have been identified specimens comprised 6 gastric adenocarcinoma and 6 anaplastic meningioma, both WHO classified after histopathological analyses. and hypothesised to be important functional molecules in neoplasia. Among these, CD70/CD27(13), tenascin C (14), Cell cultures. Human GBM cell lines PRT-HU2 and GBL-HU12, CD44 (15), B1-integrin (16), TAX-1(17) and (18) have previously described (32), were cultivated in RPMI-1640 medium recently received much interest in tumorigenesis. Kichuki et supplemented with 10% FBS, 100 units/ml penicillin, 0.1 mg/ml al. (19) have reported that the human glioblastoma cell line streptomycin and 1% L-glutamine (GIBCO, Paisley, UK). The human astrocytoma cell line D384 (provided by M. Ceroni) was maintained T98G expressed a high level of the prion protein (PrP), a with D-MEM F12 medium (GIBCO) supplemented with 10% FBS, membrane glycoprotein anchored by a GPI moiety, highly 100 units/ml penicillin, 0.1 mg/ml streptomycin and 1% L-glutamine. expressed in the CNS, that plays a crucial role in the pathogenesis of prion diseases (20). Recently, the first prion- Real-time quantitative expression analyses (RT-QPCR). Total RNA like protein gene doppel, PRND (doppel=downstream prion from homogenized specimens and cultured cells was isolated by means of the Trizol reagent (Invitrogen, Paisley, UK) using the like protein), was discovered in mammals (21-23) mapping to manufacturer’s specifications and accurately quantified by means of syntenic chromosomal locations (24). While PrP has been spectrophotometric analyses. Ten Ìg of RNA of each sample was widely investigated, given its relevance in prion diseases, for reverse-transcribed using the High Capacity cDNA Archive the doppel protein (Dpl) limited knowledge is available so far (Applied Biosystems, Foster City, CA, USA) according to the (25). Dpl, similarly to PrP, is a glycoprotein attached to the supplier’s specifications. PRND and PRNP Taqman primers and cell surface via a GPI anchor (26) that resembles an N- probes were designed by means of the Primer Express -version 3.1- terminally truncated prion protein lacking the octamer software (Applied Biosystems) with identical annealing temperatures (70ÆC and 59ÆC, for primers and probes, respectively). Primers were repeats and the neurotoxic 106-126 region. In spite of a low designed on adjacent exons of the genes in order to avoid DNA- sequence identity (about 25%), PrP and Dpl show derived amplicons, as follows: PRND (Genbank accession number comparable molecular and biochemical features (27). These AF106918): forward primer: 5'-CGTTTCTCTGGCAGGTTCTGA; similarities have suggested that Dpl might, under special reverse: 5'-GCCAGCCACCACCAGCT; probe: 5'-6-FAM- circumstances, act as a substitute/competitor of PrP, albeit CGATGAGGAAGCACC; PRNP (XM 086682): forward: 5'-GCGC with deleterious cellular effects (28). This was mainly CGCGAGCTTCT; reverse: 5'-AAGGTTCGCCATAATGACTGCT; documented by the induced in Prnp0/0 mice due to probe: 5’-6-FAM-CTCTCCTCACGACCGAG; a GAPDH Taqman probe for expression assays was purchased from Applied Biosystems PRND ectopic expression (21) and by the phenotypic rescue and used as reference. Real-time was performed on a Perkin-Elmer after Prnp reintroduction (29). While the cellular function of 5700 model using three replicas (1500 ng of RNA) for each of the PrP remains elusive, very recently different groups have investigated samples. Pools of identical RNA amounts, representing suggested a physiological role for Dpl in the regulation of the different tumor types and grades, were also prepared. Samples male fertility in mice and humans (30,31). were subjected to 40 amplification cycles of 95ÆC x 15 sec and 59ÆC x In the present study, we started to investigate, for the first 1 min. For each sample, amplifications were performed in 50 Ìl time, the behavior of the PRNP and PRND genes in human volumes containing the primers (900 nM each) and the probe (200 nM) – specifically for PRND, PRNP and GAPDH, 1 x Universal PCR gliomas, reporting differential PRND expression profiles in Master mix No Amperase UNG (Applied Biosystems). The Ct astrocytomas that seem to correlate with their clinical- averages of the replicas performed for the PRND, PRNP and GAPDH pathological classification. Thus, we propose for the prion- genes were determined; then, PRND ¢Ct (CtPRND - CtGAPDH) and like doppel gene a different new cellular and functional PRNP ¢Ct (CtPRNP - CtGAPDH) were calculated for each sample and scenario within human astrocytic tumors for the RNA pools. As a validation test, plots of the log RNA dilutions versus ¢Ct were made, showing that the amplification efficiencies of the targets (PRND, PRNP) and reference (GAPDH) were similar Materials and Methods (slopes: 0.0472 and 0.0587, respectively). These validation experiments enabled the use of the "¢¢Ct comparative method" (33) for relative Tumor specimens, classification and controls. The investigated quantitation of PRND and PRNP expression. To implement this tumor tissues consisted of astrocytoma surgical resection specimens approach, the PRNP average ¢Ct in non-tumor brain-cortex pooled provided after informed consent. Portions of the tumors were fixed specimens (NT-pool) was used as the internal calibrator.

1508 Comincini et al: PRND Overexpression in Astrocytomas

PRND gene amplification and sequence analysis. Genomic DNA from 2 x reducing buffer (100 mM Tris-HCl pH 6.8, 10% (v/v) glycerol, all samples was isolated using the DNeasy kit (Qiagen, Hilden, 0.2% (w/v) bromophenol-blue, 4% (w/v) SDS and 200 mM ‚- Germany) and accurately quantified. PRND gene amplification in the mercaptoethanol). Proteins were separated by SDS-PAGE (15% GBM specimens was verified by means of a real-time analysis using a gel) in an electrophoresis chamber at 150 V for 1.5 h, transferred Taqman system, designed on a portion of the PRND gene (forward: onto a nitrocellulose membrane (Hybond-C Extra, Amersham, 5’-TCTCTGCGGTCCAGACGAG; reverse: 5’-GGGCCTTCCGG Freiburgh, Germany) and the membrane was then incubated for 20 TTCCAC; probe: 5’-6-FAM-ATCAAGCACAGAATCA); min in a solution of 10% (w/v) non-fat milk PBS-T and 0.1% (w/v) amplifications were performed using three replicas of 1 ng Tween to avoid non-specific binding. The blots were then genomic DNA for each GBM sample and using an equal amount of incubated with the rabbit polyclonal anti-human Dpl DDC39 a human diploid genomic DNA (Clontech) as an internal calibrator. antibody (a kind gift of Prof. S Lehmann, IGH, Montpellier, Real-time amplification conditions were as above. In addition, a PRNP France) diluted at 1:10000 and, after stripping off the membrane, gene specific Taqman set was designed (forward primer: 5’- with the mouse polyclonal anti-human PrP Pri917 antibody CGTTTCTCTGGCAGGTTCTGA; reverse: 5’-GCCAGCCACC (generously provided by Dr Grassi, SPI, CEA, Saclay, France) ACCAGCT; probe: 5’-6-FAM-CGAATGAGGAAGCACC) and used diluted at 1:10000 and with the mouse monoclonal anti-human Á- to amplify the above mentioned genomic DNA. PRND and PRNP tubulin antibody (Sigma, St.Louis, MI, USA) diluted at 1:10000, in relative gene/copy quantification was performed using the "¢¢Ct the presence of 1% (w/v) non-fat milk PBS-T; the blots were then comparative method" as already described (33). The PRNP average treated with horseradish peroxidase-labeled species-specific anti- ¢Ct in a human genomic DNA was used as the internal calibrator. IgG. The blots were finally revealed by chemiluminescence using The entire PRND coding sequence was then amplified from the the ECL system (Amersham). The stripping method was DNA samples (10 ng) with the following primers: forward 5’- conveniently used to demonstrate the absence of antibody cross- ATGAGGAAGCACCTGAGCTGGTG and reverse 5’-TTATTT reactivity. rDpl was used to confirm the specificity of the CACCATGAGCCAGATCAAA (30 cycles of 94ÆC x 15 sec, 62ÆC antibodies. Negative control reactions were also performed by x 30 sec, 72ÆC x 45 sec) and sequenced on both strands using a omitting the primary antibodies. ABI310 automated sequencer (Applied Biosystems). Sequence analyses were performed with the BLAST network service Immunohistochemistry. The human PRT-HU2, GBM-HU12 and (http://www.ncbi.nlm.nih.gov/cgi-bin/BLAST/). D384 cell lines were fixed with ethanol 70% for 10 min at 4ÆC. rDpl synthesis and expression. A DNA fragment coding for the Indirect immunoperoxidase staining of cultured cells was performed human mature Dpl (amino acids 26-153) was amplified by PCR using the EnVision kit (DAKO, Carpinteria, CA, USA) with Fast directly from genomic DNA using primers adapted from the Red as chromogen. Samples were counterstained with hematoxylin deposited sequence (AF106918), forward primer: 5’-GGAATT for 30 sec. Negative control staining reactions were performed using CCATATGACGAGGGGCATCAAGCA including a NdeI site, the secondary antibody alone. Formalin-fixed, paraffin-embedded and reverse: 5’-CCGGAATTCTTATGCGCCCCTCTCCAACC tissues were stained by the EnVision kit using Diaminobenzidine as including a EcoRI site. The PCR product was cut with NdeI-EcoRI chromogen. The stainings were performed according to the and joined to a pET-23b expression vector (Novagen, Darmstadt, manufacturer’s instructions. Briefly, 6-Ìm-thick paraffin sections Germany), thereby obtaining the rDpl1 plasmid. Fifteen Ìg of were deparaffinized with xylene, rehydrated by passages through purified rDpl1 plasmid was introduced into the Rapid Translation decreasing concentrations of ethanol (100 to 70%) and pretreated System (RTS500, Roche, Mannheim, Germany). The protein for 5 cycles, 2 min each, at 900 W in a microwave oven with synthesis by the RTS500 system was carried out according to the Unmasking Solution (Vector Laboratories, Burlingame, CA, USA) procedure recommended by the supplier (incubation at 30ÆC for and cooled at room temperature for 20 min. The primary antibodies 20 h, stirred at 120 rpm). After the indicated period of incubation, were incubated (1:1000 for both of the antibodies, DDC39 and the reaction mixture was supplemented with 1 mM Pefablock Pri917) at room temperature for 30 min. Samples were (Roche) and stored at -80ÆC until analyzed. counterstained with hematoxylin for 45 sec. Negative control staining reactions were performed by omitting the primary antibodies. Western blot analysis. PRT-HU2, GBL-HU12 and D384 cells were For Dic contrast optic, rehydratated paraffin GBM sections cultured in their media as described above. After removing the were submitted to microwave treatment and immunostained with media, cells were washed with PBS, trypsinized and pelleted (1200g the mouse monoclonal anti-human Dpl Dop151 antibody (a kind for 10 min). Equal amounts of cell pellets (100 mg) were gift of Dr J. Grassi): in this case the antibody dilution was 1:100. solubilized in 100 Ìl of lysis buffer (0.5% Triton X-100, 150 mM NaCl, 0.5% sodium deoxycholate, 20 mM Tris-HCl pH 7.5) in the Results presence of 1 mM Pefablock. The lysates were incubated for 15 min in ice and centrifuged (4000g for 10 min at 4ÆC); the A panel of various histopathological grades of primary and supernatants were removed and stored at -80ÆC. For specimens, secondary astrocytomas was analysed for PRNP and PRND tissues were washed with ice-cold PBS and sonicated in ice. As expression. These included 12 GBM, 6 AA, 4 LGA and 3 GBM above, 100 mg of each homogenized sample were solubilized in 100 peritumoral specimens. Negative controls were represented by Ìl of ice-cold lysis buffer; after centrifugation (4000g for 10 min at 2 brain cortex non-tumor specimens and total RNA from 4ÆC) the supernatants were collected, supplemented with 1 mM Pefablock and stored at -80ÆC. After proteins determination by human adult and fetal brains. In addition, the astrocytoma cell Coomassie blue binding (Bradford method), samples (50 mg for lines, PRT-HU2, GBL-HU12 (32) and D384, were assayed for specimens, 10 mg for cell lines and one microliter of the reaction PRNP and PRND expression. The expression of these genes mixture containing rDpl) were boiled at 95ÆC for 4 min in 4 Ìl of was then determined in 6 gastric adenocarcinoma and 6

1509 ANTICANCER RESEARCH 24: 1507-1518 (2004)

Figure 1. RT-QPCR analyses of PRND and PRNP expression. Amplification curves represent the PRND and PRNP expression profiles of pools representative of the investigated tumor types and grades, controls and cell lines, with corresponding colors or labels; each of the following pools (GBM=glioblastoma multiforme; AA=anaplastic astrocytoma; LGA=low-grade astrocytoma) was compared to identical RNA amounts for controls (Fb=human fetal brain; Ab=human adult brain; PT=peritumoral GBM specimens; NT=brain-cortex non-tumor specimens; NTC=not-template controls), non-glial tumors (KG=gastric adenocarcinoma; M=anaplastic meningioma) and astrocytoma cell lines (PRT-HU2, GBL-HU12 and D384). Threshold baselines (Ct=0.120) were indicated and used to calculate the Ct of PRND, PRNP and GAPDH, as reported in Table I. Rn indicates the magnitude of the fluorescence signals in arbitrary units, generated by the given set of PCR conditions.

anaplastic meningioma specimens. Because of the limited compared to GBM) and LGA (~140-fold), respectively. The amount of tumor tissues available, we used a sensitive LGA samples exhibited the lowest PRND expression (LGA- Taqman RT-QPCR fluorescence assay (34) to quantitate the pool, RQ=0.06±0.01), but showed expression profiles higher expression of PRND and PRNP transcripts in astrocytomas. than the GBM peritumoral analysed samples (PT-pool, The human GAPDH housekeeping gene was adopted as RQ=0.03±0.01). A marked PRND expression was also endogenous control for RNA normalisation. PRNP and highlighted in the non-glial tumors analysed, namely gastric PRND RT-QPCR were performed on three replicas of the adenocarcinoma and anaplastic meningioma; in the gastric above-mentioned samples (1500 ng) and on identical amounts adenocarcinoma, as in the GBM, the PRND expression was of pooled RNA samples, according to their tumor type and higher than the corresponding PRNP profiles, as evidenced by histopathology grade. Figure 1 reports the amplification their respective Ct values. Among controls, the fetal brain curves obtained from the RNA pools (1500 ng), compared to showed the highest PRND expression (Fb, RQ=0.16), nearly equal amounts of RNA from the GBM and astrocytoma cell overlapping with that of AA (AA-pool, RQ=0.20±0.09), lines and controls. After a validation test, described in while other samples (the non-tumoral brain cortex and the Materials and Methods, the expression levels of the genes human adult brain) exhibited faint or almost undetectable were calculated using the ¢¢Ct comparative method (33), (D384) PRND expression. PRT-HU2 and GBL-HU12 showed taking the PRNP normalised expression level in the non- high and similar PRND expression (RQ=1.18 and 0.74, tumoral control pooled tissues as the internal calibrator (i.e. respectively). On the other hand, PRNP showed high but less PRNP NT-pool relative quantitation, RQ=100): these variable expression levels than PRND in all the samples analyses are reported in Table I. From the relative examined, with the exception of the D384 cell line that quantitative analyses, PRND exhibited variable expression exhibited a barely detectable and non-quantifiable expression profiles, with in particular the highest expression in the GBM profile. Among tumor samples, the highest PRNP expression samples compared to the control (GBM-pool, RQ=8.36±2.38 was scored in AA (AA-pool, RQ=6.04±2.10) and in LGA SD); furthermore, within astrocytomas, much lower PRND (LGA-pool, RQ=5.79±0.72); within the astrocytoma samples, expression values were detected in AA (by ~70-fold, GBM showed the lowest PRNP expression (GBM-pool,

1510 Comincini et al: PRND Overexpression in Astrocytomas

Table I. PRND and PRNP relative expression quantitation in astrocytomas biopsies and cell lines.

PRND PRNP GAPDH PRND PRNP SAMPLE Ct Avg Ct Avg Ct Avg ¢ Ct1 ¢¢ Ct2 RQ3 ¢ Ct ¢¢ Ct RQ

GBM-1 23.42 26.15 25.15 -1.73 4.42 5.44 1.00 6.93 0.82 GBM-2 23.86 26.17 25.45 -1.59 4.34 4.94 0.72 6.65 0.99 GBM-3 22.89 25.94 25.41 -2.52 3.41 9.41 0.53 5.40 2.37 GBM-4 23.57 25.01 25.66 -2.09 3.84 6.98 -0.65 5.28 2.57 GBM-5 23.11 26.18 25.71 -2.60 3.33 9.94 0.47 6.24 1.18 GBM-6 23.07 26.05 25.75 -2.68 3.25 10.51 0.30 6.23 1.33 GBM-7 23.01 25.99 25.63 -2.62 3.31 10.08 0.36 6.29 1.28 GBM-8 23.07 26.47 25.22 -2.15 3.78 7.28 1.25 7.18 0.70 GBM-9 23.55 26.18 25.37 -1.82 4.11 5.79 0.81 6.74 0.94 GBM-10 23.48 26.54 25.39 -1.91 4.02 6.16 1.15 7.08 0.74 GBM-11 23.10 25.91 25.40 -2.30 3.00 12.5 0.51 6.44 1.15 GBM-12 23.15 26.38 25.39 -2.24 3.69 7.75 0.99 6.92 0.83 GBM-pool 23.28 25.94 25.63 -2.35 3.58 8.36±2.38 0.31 6.24 1.32±0.61 AA-1 29.42 23.37 25.67 3.75 9.68 0.12 -2.30 3.63 9.74 AA-2 28.99 23.39 25.82 3.17 9.10 0.18 -1.83 4.01 5.83 AA-3 29.11 24.11 25.60 3.51 9.44 0.14 -1.49 4.44 4.61 AA-4 29.05 24.17 25.43 3.62 8.13 0.36 -1.26 4.67 3.93 AA-5 28.67 24.19 25.70 2.97 8.90 0.21 -1.51 4.42 4.67 AA-6 28.91 24.05 25.68 3.23 9.16 0.17 -1.63 4.30 5.08 AA-pool 29.11 23.41 25.29 3.82 9.75 0.20±0.09 -1.88 4.05 6.04±2.10 LGA-1 30.28 23.75 25.24 5.04 10.97 0.05 -1.49 4.44 4.61 LGA-2 30.33 23.42 25.35 4.98 10.91 0.05 -1.93 4.00 6.25 LGA-3 30.01 23.68 25.25 4.76 10.69 0.06 -1.57 4.36 4.87 LGA-4 30.15 23.62 25.27 4.88 10.81 0.06 -1.65 4.28 5.15 LGA-pool 30.01 23.37 25.19 4.82 10.75 0.06±0.01 -1.82 4.11 5.79±0.72 PRT-HU2 26.32 24.08 25.84 0.48 6.41 1.18 -1.76 4.17 5.56 GBL-HU12 26.78 23.74 25.64 1.14 7.07 0.74 -1.90 4.03 6.12 D384 37.18 38.73 25.28 11.90 17.83 << 0.01 8.73 9.08 << 0.01 KG-1 27.65 29.18 25.34 2.31 8.24 0.33 3.84 9.77 0.11 KG-2 26.94 28.65 25.31 1.63 7.26 0.65 3.34 9.97 0.16 KG-3 27.02 28.72 25.46 1.56 7.49 0.56 3.26 9.19 0.17 KG-4 26.75 28.08 25.40 1.35 7.28 0.64 2.68 8.61 0.26 KG-5 26.88 28.55 25.67 1.21 7.14 0.71 2.88 8.81 0.22 KG-6 27.11 28.47 25.39 1.72 7.65 0.50 3.08 9.01 0.19 KG-pool 26.99 28.68 25.31 1.68 7.61 0.51±0.14 3.37 9.30 0.16±0.05 M-1 25.15 24.19 25.23 -0.08 5.85 1.72 -1.04 4.89 3.37 M-2 24.85 25.11 25.65 -0.80 5.13 2.86 -0.54 5.39 2.38 M-3 24.05 25.45 25.64 -1.59 4.34 4.93 -0.19 5.74 1.87 M-4 24.06 25.13 25.37 -1.31 4.62 4.06 -0.24 5.69 1.94 M-5 24.22 25.25 25.43 -1.21 4.72 3.79 -0.18 5.75 1.86 M-6 24.21 25.56 25.38 -1.17 4.76 3.69 0.18 6.11 1.45 M-pool 24.79 25.01 25.44 -0.65 5.28 2.57±1.10 -0.43 5.50 2.21±0.67 Fb 28.98 20.91 25.66 3.32 9.25 0.16 -4.75 1.18 44.14 Ab 35.51 21.18 25.47 10.04 15.97 < 0.01 -4.29 1.64 32.09 PT-1 31.85 25.52 25.33 6.52 12.45 0.02 0.19 6.12 1.44 PT-2 31.66 25.87 25.30 6.36 12.29 0.02 0.57 6.50 1.10 PT-3 31.12 24.72 25.46 5.66 11.59 0.03 -0.74 5.19 2.73 PT-pool 31.51 25.07 25.72 5.79 11.72 0.03±0.01 -0.65 5.28 2.57±0.86 NT-1 33.95 19.79 25.55 8.40 14.33 < 0.01 -5.76 0.17 88.89 NT-2 34.45 19.07 25.58 8.87 14.80 < 0.01 -6.51 -0.58 149.48 NT-pool 34.18 19.41 25.34 8.84 14.77 < 0.01 -5.93 0.00 100±42.84

1The ¢Ct value is determined by subtracting the average GAPDH Ct from the averages PRND Ct or PRNP Ct; 2The calculation of ¢¢Ct involves subtraction by the ¢Ct calibrator value (PRNP NT-pool). 3Relative expression (RQ), normalized to an endogenous reference and relative to the calibrator, is given by 2e-¢¢Ct. Investigated samples are: GBM=glioblastoma multiforme; AA=anaplastic astrocytoma; LGA=low-grade astrocytoma; cell lines (PRT-HU2, GBL-HU12 and D384); Fb=human fetal brain; Ab=human adult brain; PT=peritumoral GBM biopsies; KG=gastric adenocarcinoma; M=anaplastic meningioma; NT=brain-cortex non-tumor specimens. Standard deviations are calculated for the RQ of the different pooled samples.

1511 ANTICANCER RESEARCH 24: 1507-1518 (2004)

Figure 3. Immunohistochemical analyses of Dpl and PrP in astrocytoma cell lines and specimens with hematoxylin counterstaining. Dpl staining of (a) PRT-HU2 (1200X), (b) GBL-HU12 (800X) and (c) D384 cell lines (800X); Dpl staining of (d) GBM (600X) and (e) GBM peritumoral specimens (600X). Dpl immunoreactivity is found in neoplastic cells (arrowhead) and in some endothelial cells; vascular structures (glomeruli) with immunoreactive endothelial cells are indicated by an asterisk. PrP staining of (f) GBM specimens (400X), (g) PRT-HU2 (800X), (h) GBL-HU12 (800X) and (i) D384 cell lines (800X). PrP immunoreactivity is found in neoplastic cells (arrowhead). Thin arrowheads indicate the Golgi localised PrP immunoreactivity. (l) Immunocytochemical analysis of human glioblastoma by Dic contrast optic. In the inset a small vessel (asterisk) with immunopositive endothelial cells (arrow), inside the lumen and attached to the vessel wall two -like cells can be seen, both are immunoreactive. Both scale bars are 10 Ìm.

1512 Comincini et al: PRND Overexpression in Astrocytomas

Table II. Relative quantitation of PRND and PRNP genes-amplification in glioblastoma multiforme biopsies and cell lines.

PRND PRNP GAPDH PRNP PRNP SAMPLE Ct Avg Ct Avg Ct Avg ¢ Ct1 ¢ Ct2 RQ3 ¢ Ct ¢¢ Ct RQ

PRT-HU2 26.57 26.51 25.97 0.60 -1.11 2.16 0.54 -1.17 2.25 GBL-HU12 26.63 26.64 25.87 0.76 -0.95 1.93 0.77 -0.94 1.92 GBM-1 27.01 27.11 25.55 1.46 -0.25 1.19 1.56 -0.15 1.11 GBM-2 27.49 27.47 25.97 1.52 -0.19 1.14 1.50 -0,21 1.16 GBM-3 27.04 27.07 25.61 1.43 -0.28 1.21 1.46 -0.25 1.19 GBM-4 27.21 27.57 25.87 1.34 -0.37 1.29 1.70 -0.01 1.01 GBM-5 26.44 26.51 25.15 1.29 -0.19 1.14 1.36 -0.35 1.27 GBM-6 27.05 27.13 25.70 1.35 -0.36 1.28 1.43 -0.28 1.21 GBM-7 27.17 27.37 25.82 1.35 -0.36 1.28 1.55 -0.16 1.12 GBM-8 27.50 27.44 25.80 1.70 -0.01 1.01 1.64 -0.07 1.05 GBM-9 27.44 27.62 25.94 1.50 -0.21 1.16 1.68 -0.03 1.02 GBM-10 26.82 26.82 25.41 1.41 -0.30 1.23 1.41 -0.30 1.23 GBM-11 26.93 27.21 25.61 1.32 -0.39 1.31 1.60 -0.11 1.08 GBM-12 26.89 26.92 25.46 1.43 -0.28 1.21 1.46 -0.25 1.19 DNA Hs 27.29 27.49 25.78 1.51 -0.20 1.15 1.71 0.00 1.00

1The ¢Ct value is determined by subtracting the average GAPDH Ct from the averages PRND Ct or PRNP Ct; 2The calculation of ¢¢Ct involves subtraction by the ¢Ct calibrator value (PRNP DNA Hs); 3Relative expression (RQ), normalized to an endogenous reference and relative to the calibrator, is given by 2e-¢¢Ct. PRT-HU2 and GBL-HU12 are two glioblastoma multiforme-derived cell lines. GBM 1-12 designates the different glioblastoma multiforme biopsies. DNA Hs is a diploid human genomic DNA adopted as control.

RQ=1.32±0.61). The lowest PRNP expression was scored in from the PRT-HU2 and GBL-HU12 cell lines, using equal the gastric adenocarcinoma specimens (KG-pool, amounts of genomic DNA (p<0.001; ANOVA). Similar RQ=0.16±0.05). Differently from PRND, the highest PRNP results in the relative quantitation analysis of PRND and expression was revealed in brain control samples, namely the PRNP expression were obtained analyzing the 0.1 and 10 normal tissues from frontal cortex (NT-pool, RQ=100±42.84) ng DNA concentrations (data not shown). In accordance and in fetal and adult human brain (RQ=44.14 and 32.09, with our results, the DNA content of the above-mentioned respectively). PRNP expression in anaplastic meningioma (M- cell lines had previously been reported to be near-triploid pool, RQ=2.21±0.67) was similar to the GBM peritumoral or tetraploid (32). In order to reveal polymorphisms specimens (PT-pool, RQ=2.57±0.86). Similar results in eventually associated with tumor grades of malignancy, the expression profiles were obtained through the analysis of entire PRND coding sequences of the investigated samples amplifications using different RNA quantities (100 and 1000 were amplified and sequenced on both strands, revealing ng) (data not shown). In summary, the expression profiles of no polymorphisms. PRND alone appeared related to the different astrocytoma Western blots were performed on the above-mentioned grades (p<0.0001; ANOVA). cell lines and specimens with the polyclonal anti-Dpl DDC39 In order to assess whether the PRND high expression in and the monoclonal anti-PrP Pri917 antibodies, as previously GBM was due to amplification or to gene regulation described (31). Á-tubulin was adopted as the internal control. processes, we estimated the PRND gene content in the The results of Western blot analysis on representative genomic DNA of the investigated GBM specimens. This samples are reported in Figure 2. We found Dpl expression was performed using a Taqman probe designed on a DNA in all the investigated samples. Dpl patterns were variable, portion of the PRND gene and amplifying 1 ng of the DNA exhibiting a common band migrating at 30-31 kDa; extracted from the GBM samples. As an additional additional bands (at about 24, 39 and 45 kDa) were experimental control, a Taqman PRNP probe was visualized at different proportions and intensities within the employed to estimate the PRNP gene-copy content. The Ct GBM and LGA analyzed samples. Conversely, the AA average values, reported in Table II, indicated that the examined specimens exhibited fragments of 31 and 39 kDa. different GBM samples had nearly overlapping In addition, a minor band migrating at 14-15 kDa was solely PRND/PRNP-derived amplicons to a diploid human detected in one AA sample: this matched the rDpl E.coli genomic DNA adopted as internal control; on the contrary, synthesized protein band and probably represented the these values were different compared to those obtained unglycosylated . PrP detection on the same

1513 ANTICANCER RESEARCH 24: 1507-1518 (2004)

Figure 2. Western blotting analysis of Dpl and PrP in specimens and cultured astrocytoma cells. SDS-PAGE with protein extracts of rDpl, astrocytoma (WHO grades IV, III, II), gastric adenocarcinoma (KG), anaplastic meningioma biopsies (M) and astrocytoma cell lines (PRT-HU2, GBL-HU12 and D384), probed with the human DDC39 (Dpl) and Pri917 (PrP) antibodies. Á-tubulin was adopted as positive control. Molecular weights (kDa), tumor types and grades are indicated.

membrane exhibited an intense heterodisperse staining in all a few scattered neoplastic cells (panel e). PrP staining was the samples, with electrophoretic signals of approximately 30 revealed in the GBM specimens (panel f) and focused on kDa. PrP exhibited electrophoretic fragments of different Golgi in the PRT-HU2 and GBL-HU12 cell lines. molecular weights and a faint signal in the D384 cell line. Á- Additional PrP immunoreactivity was also visible in the tubulin revealed the expected 48-kDa fragment in all respective cell membranes, as expected (panels g, h). The samples. rDpl blot exhibited a specific band at 14-15 kDa. D384 cell line showed no PrP reactivity (panel i). Similar As expected, no cross reactivity was detected after stripping results in Dpl staining within GBM specimens were obtained and reprobing the filter with anti-PrP Pri917 and anti-Á- by Dic contrast optic analysis: as reported in panel l, Dpl is tubulin antibodies. Because tumor tissue samples contain a found in neoplastic and in some endothelial cells; heterogeneous population of cells, we assessed the specificity immunoreactive endothelial cells within vascular structures of PrP and Dpl expression in the cultured GBM cell lines (glomeruli) were also evidenced, as well as some and specimens by immunohistochemistry. Three cell lines lymphocyte-like cells. and paraffined sections from tumor specimens and from peritumoral GBM tissues, previously analyzed for PRND Discussion expression by RT-QPCR, were analysed (Figure 3). A granular and diffuse Dpl immunoreactivity was mainly The majority of high-grade astrocytomas are highly invasive observed in the cytoplasm of PRT-HU2 (panel a) and GBL- and develop recurrences at the primary site within a short HU12 cell lines (panel b), while D384 showed no Dpl period of time after the primary tumor is removed. reactivity (panel c). In the GBM tissues, neoplastic cells However, tumors with the same histological diagnosis can exhibited a diffuse cytoplasmatic Dpl immunoreactivity vary greatly in terms of their prognosis and response to (panels d). In addition, Dpl reactivity was also detected in therapy (3). Therefore, it is important to identify molecular infiltrating lymphocytes and in endothelial cells of blood markers for glial tumors and to understand the process of vessels. Peritumoral GBM sections, that presented a low transformation to improve the prognostic prediction and level of PRND expression, showed immunoreactivity only in response to treatment strategies (5).

1514 Comincini et al: PRND Overexpression in Astrocytomas

In this study, we evaluated the expression of the PRNP stage, direction and fate of tumor progression (40). The gene and its recently discovered paralog, PRND, in Dpl patterns that we found are different from those astrocytic tumor samples. A surprisingly high level of PRND previously described in humans. In brain homogenates, expression and of the corresponding protein, Dpl, in Dpl is undetectable by Western blotting (36) and, in testis, glioblastoma multiforme and in other highly malignant Dpl resulted in a large heterodisperse band of 29-35 kDa tumors was reported. in size (31). PRND is physiologically expressed at high levels only in Our immunohistochemical findings in the GBM investigated testis, at lower levels in other peripheral organs and, tissues suggest that Dpl is produced by a variety of cells, notably, at very low levels in the brain (21,35,36). Moreover, including neoplastic, endothelial and lymphocytes. Li et al. (37) observed a transient expression of Dpl in the Significantly, the GBM cell lines PRT-HU2 and GBL-HU12, brain endothelial cells of neonatal wild-type mice, suggesting as well as the neoplastic cells, exhibited a marked cytoplasmatic a role of Dpl in the blood-brain barrier maturation in the Dpl immunoreactivity: this localization was partly unexpected CNS. Furthermore, Dpl is associated with for Dpl, according to previous studies that indicated a raft- when ectopically expressed in the membrane associated localization for Dpl (21,25,26,41). The due to Prnp ablation (21). It is therefore peculiarity of Dpl localization was reinforced by the expected reasonable to speculate that, although Dpl may play a role PrP localization in both GBM specimens and cell lines on the in the initial development of the brain, its main outer and in correspondence with the Golgi physiological function probably lies within other tissues or apparatus, as previously described (42). As a consequence, might be associated with still uncovered pathological during tumorogenesis an altered Dpl cellular trafficking, conditions. Unlike PRND, PRNP is virtually ubiquitous, probably due to different transcriptional or post-translation showing the highest level of expression within the CNS in uncovered modifications, or to retro-translocation of a all the investigated species (38) and is associated with misfolded-Dpl protein from the ER into the cytosol, could neurodegenerative diseases (20). determine a different behavior of Dpl. Similarly, a We adopted a highly sensitive real-time PCR analysis to cytoplasmatic accumulation of PrP has been attributed to a study PRNP and PRND expression and found that PRND damaged primary quality-control system for cytoplasmatic expression appears to be related to the degree of protein degradation (43). Furthermore, the observation that astrocytoma histological malignancy, with the highest scores the N-terminal region of the Dpl is not well folded (44) would of expression in the high-grade GBM. Expression variations suggest that a fraction of the protein is constantly subjected to were also scored for PRNP but were not related to the various ER quality control: therefore it is reasonable for Dpl to appear astrocytoma grades. Significantly, two out of three of the in the cytoplasm when the folding capacity of the system is investigated astrocytoma cell lines (PRT-HU2 and GBL- compromised, as in environmental stress, aging, biological HU12) exhibited high levels of expression of both PRND and traumas or cancer. In many tumors, including astrocytomas, it PRNP. To our knowledge, these represent the first human cell has been established that macrophages and, to a lesser extent, lines that express detectable levels for both PRND and PRNP. T lymphocytes are the predominant infiltrating cells and that However, the D384 cell line exhibited very low PRND and they may exert cellular immunity (45). Our results clearly PRNP expression profiles, as in the assayed glioma-derived indicate a Dpl-staining in T-cell-like sub-populations, probably (Hu197, GL15, 8MG) and non-glial-derived cell lines (HeLa, CD4+ or CD8+, as previously described in astrocytic tumors COS) (data not shown). Nevertheless, Brunner et al. (39) have (45). Noticeably, in CD4+ lymphocytes from various human described amplification in the chromosomal region 20p11p12 tumor SAGE expression libraries (http://www.ncbi.nlm. in some GBM: the authors stated that this region may contain nih.gov/SAGE/), we found reliable tags for PRND expression tumor candidate genes such as PCNA and PYGB, but (data not shown). Experiments are in progress in order to additionally it contains both PRNP and PRND, separated by identify the T-cells subtypes expressing PRND. In addition, our a 22 kb chromosomal fragment (23). Our analysis revealed results indicate that Dpl is found to be associated with blood that in all the investigated GBM specimens both genes do not tumor vessels, suggesting an endothelial cells immuno- exhibit chromosomal amplifications; consequently, regulation reactivity; moreover, these stainings agreed with those processes could be evoked for PRND overexpression, possibly previously reported for Dpl in endothelial cells of newborn involving the transcriptional regulation or the stability of mice (37). Whether Dpl expression could be related to the PRND mRNA or both. Western blot confirmed a marked Dpl vasogenic brain edema associated with malignant glioma, expression in the investigated samples, which is characterized determining vascular hyperpermeability or promoting tumor- by complex electrophoretic patterns. These might result from associated angiogenesis, needs to be further investigated. changes in the of these proteins. It is known It is likely that Dpl represents a protein whose main that aberrant glycosylation in tumor cells has been physiological expression patterns have been spatially implicated as an essential mechanism in defining the restricted to the male reproductive tract and, possibly, within

1515 ANTICANCER RESEARCH 24: 1507-1518 (2004) sperm-egg interactions processes (30,31). Nevertheless, since 9 Ljubimova JY, Khazenzon N et al: Gene expression tumor cells in gliomas exhibited expression of proteins that abnormalities in human glial tumors identified by gene array. are normally produced at different developmental stages in Int J Oncol 18: 287-295, 2001. 10 Fuller GN, Hess KR, Rhee CH, Yung WK, Sawaya RA, Bruner the astrocytic differentiation pathway, the high expression JM and Zhang W: Molecular classification of human diffuse profiles of Dpl in the first stages of brain development (37) gliomas by multidimensional scaling analysis of gene expression and in GBM (our study) may indicate the neoplastic cell’s profiles parallels morphology-based classification, correlates re-acquisition of a "primitive" expression behavior, typical of with survival, and reveals clinically-relevant novel glioma the CNS development. subsets. Brain Pathol 12: 108-116, 2002. In conclusion, the findings described in the present study 11 Pilkington GJ: Tumour cell migration in the central nervous indicate that: 1) astrocytomas express PRND at different system. Brain Pathol 4: 157-166, 1994. 12 Pignatelli M and Vessey CJ: Adhesion molecules: novel molecular levels which relate to the tumor grades; in addition, PRND tools in tumor pathology. Hum Pathol 25: 849-856, 1994. has high levels of expression in two different highly 13 Feindt JH and Metlein R: CD70/CD27 ligand, a member of the malignant non-glial tumors; 2) two GBM cell lines display TNF family, is expressed in human brain tumors. Int J Cancer high PRND and PRNP expression profiles; 3) Dpl is 98: 352-356, 2002. expressed in different cells within astrocytomas, showing a 14 Adamsky K, Schilling J, Garwood J, Faissner A and Peles E: peculiar cytoplasmic localization in the neoplastic cells; 4) Glial tumor cell adhesion is mediated by binding of the FNIII lastly, a new biological scenario is described in which Dpl domain of receptor protein tyrosine phosphatase beta may play a role in tumor progression. (RPTPbeta) to tenascin C. Oncogene 20: 609-618, 2001. 15 Okamoto I, Tsuiki I et al: Proteolytic cleavage of the CD44 adhesion molecule in multiple human tumors. Am J Pathol 160: Acknowledgements 441-447, 2002. 16 Wu Y, Chen L, Zheng PS and Yang BB: B1-integrin-mediated The authors are grateful to Prof. Giuseppe Nappi, Scientific glioma cell adhesion and free radical-induced apoptosis are Director of the Mondino Institute, Italy, for his support to this regulated by binding to a C-terminal domain of PG-M/Versica. line of research, to Prof. Eugenio Benericetti (Unità di J Biol Chem 277: 12294-12301, 2002. Neurochirurgia, Ospedale di Parma, Italy) and to Dott. Carlo 17 Rickman DS, Tyagi R et al: The gene for the axonal cell Zibera (Fondazione "Maugeri", Clinica del Lavoro, Pavia, Italy) adhesion molecule TAX-1 is amplified and aberrantly expressed for providing tumor specimens and GBM-derived cell lines. SC, in malignant gliomas. Cancer Res 61: 2162-2168, 2001. RN and LF are supported by a grant from the Italian Ministry 18 Senner V, Kismann E, Puttmann S, Hoess N, Baur I and Paulus for Education, University and Research (MIUR), "Progetti di W: L1 expressed by glioma cells promotes adhesion but not Ricerca di Interesse Nazionale (2003)". KP and JLL are migration. Glia 38: 146-154, 2002. supported by a GIS grant "Infections à " (French Ministry 19 Kikuchi Y, Kakeya T et al: G1-dependent prion protein of Research). MC is supported by IRCCS Mondino Institute, expression in human glioblastoma cell line T98G. Biol Pharm Pavia, Italy. Bul 25: 728-733, 2002. 20 Prusiner SB: Prions. Proc Natl Acad Sci USA 95: 13363- References 13383, 1998. 21 Moore RC, Lee I et al: Ataxia in prion protein (PrP)-deficient 1 Kleihues P and Cavenee WK: World Health Organization mice is associated with upregulation of the novel PrP-like Classification of Tumors of the . Lyon: protein doppel. J Mol Biol 292: 797-817, 1999. WHO/IARC, 2000. 22 Peoc'h K, Guerin C, Brandel JP, Launay JM and Laplanche JL: 2 Louis DN, Holland EC and Cairncross JG: Glioma classification: First report of polymorphisms in the prion-like protein gene a molecular reappraisal. Am J Pathol 159: 779-786, 2001. (PRND): implications for human prion diseases. Neurosci Lett 3 Dai C and Holland EC: Glioma models. Bioch Bioph Acta 286: 144-148, 2000. 1551: 19-27, 2001. 23 Comincini S, Foti MG et al: Genomic organization, comparative 4 Comincini S: Searching for molecular markers of human analysis, and genetic of the bovine and ovine prion gliomas. Funct Neurol 16: 291-298, 2001. Doppel genes (PRND). Mamm Genome 12: 729-733, 2001. 5 Holland EC: Gliomagenesis: genetic alterations and mouse 24 Castiglioni B, Comincini S, Drisaldi B, Motta T and Ferretti L: models. Nat Rev Genet 2: 120-129, 2000. Comparative mapping of the prion gene (PRNP) in cattle, 6 Rasheed BK, McLendon R et al: Alterations of the TP53 gene and human with PCR-generated probes. Mamm Genome in human gliomas. Cancer Res 54: 1324-1330, 1994. 9: 853-855, 1998. 7 Nozaki M, Tada M et al: Roles of the functional loss of p53 and 25 Mastrangelo P and Westaway D: The prion gene complex other genes in astrocytoma tumorigenesis and progression. encoding PrP(C) and Doppel: insights from mutational analysis. Neuro-Oncol 1: 124-37, 1999. Gene 275: 1-18, 2002. 8 Kimura T, Budka H and Soler-Federsppiel S: An 26 Silverman GL, Qin K et al: Doppel is an N-glycosylated, immunocytochemical comparison of the glial-associated glycosylphosphatidylinositol-anchored protein. Expression in proteins glial fibrillary acidic protein (GFAP) and S-100 testis and ectopic production in the brains of Prnp(0/0) mice protein (S100P) in human brain tumors. Clin Neuropathol 5: predisposed to loss. J Biol Chem 275: 26834- 21-27, 1986. 26841, 2000.

1516 Comincini et al: PRND Overexpression in Astrocytomas

27 Lu K, Wang W, Xie Z, Wong BS, Li R, Petersen RB, Sy MS 38 Fournier JG: Nonneuronal cellular prion protein. Int Rev Cytol and Chen SG: Expression and structural characterization of the 208: 121-161, 2001. recombinant human doppel protein. Biochemistry 39: 13575- 39 Brunner C, Jung V, Henn W, Zang KD and Urbschat S: 13583, 2000. Comparative genomic hybridization reveals recurrent 28 Behrens A and Aguzzi A: Small is not beautiful: antagonizing enhancements on 20 and in one case combined functions for the prion protein PrP(C) and its homologue Dpl. amplification sites on 15q24q26 and 20p11p12 in glioblastomas. Trends Neurosci 25: 150-154, 2002. Cancer Genet Cytogenet 121: 124-127, 2000. 29 Moore RC, Mastrangelo P et al: Doppel-induced cerebellar 40 Hakomori S: Tumor malignancy defined by aberrant degeneration in transgenic mice. Proc Natl Acad Sci USA 98: glycosylation and sphingo(glyco)lipid metabolism. Cancer Res 15288-15293, 2001. 56: 5309-5318, 1996. 30 Behrens A, Genoud N et al: Absence of the prion protein 41 Shaked Y, Hijazi N and Gabizon R: Doppel and PrP(C) do not homologue Doppel causes male sterility. EMBO J 21: 3652- share the same membrane microenvironment. FEBS Lett 530: 3658, 2002. 85-88, 2002. 31 Peoc'h K, Serres C et al: The human "prion-like" protein 42 Laine J, Marc ME, Sy MS and Axelrad H: Cellular and Doppel is expressed in both Sertoli cells and spermatozoa. J subcellular morphological localization of normal prion protein Biol Chem 277: 43071-43078, 2002. in rodent cerebellum. Eur J Neurosci 14: 47-56, 2001. 32 Bacciocchi G, Gibelli N et al: Establishment and 43 Ma J and Lindquist S: Wild-type PrP and a mutant associated characterization of two cell lines derived from human with prion disease are subject to retrograde transport and glioblastoma multiforme. Anticancer Res 12: 853-861, 1992. proteasome degradation. Proc Natl Acad Sci USA 98: 14955- 33 Livak K and Schmittgen TD: Analysis of relative gene 14960, 2001. expression data using real-time quantitative PCR and the 2 44 Luhrs T, Riek R, Gunter P and Wuthrich K: NMR structure of –¢¢Ct method. Methods 25: 402-408, 2001. the human doppel protein. J Mol Biol 326: 1549-57, 2003. 34 Heid CA, Stevens J, Livak KJ and Williams PM: Real-time 45 Walker PR and Dietrich PY: Immune escape of gliomas. Prog quantitative PCR. Genome Res 6: 986-994, 1996. Brain Res 132: 685-698, 2001. 35 Makrinou E, Collinge J and Antoniou M: Genomic characterization of the human prion protein (PrP) gene locus. Mamm Genome 13: 696-703, 2002. 36 Peoc'h K, Volland H et al: Prion-like protein Doppel expression is not modified in -infected cells and in the brains of patients with Creutzfeldt-Jakob disease. FEBS Lett 536: 61-65, 2003. 37 Li A, Sakaguchi S, Shigematsu K et al: Physiological expression of the gene for PrP-like protein, PrPLP/Dpl, by brain endothelial cells and its ectopic expression in of PrP- Received December 9, 2003 deficient mice ataxic due to Purkinje cell degeneration. Am J Revised January 20, 2004 Pathol 157: 1447-1452, 2000. Accepted February 24, 2004

1517