GAS41), a Highly Conserved Putative Transcription Factor

Oncogene (2001) 20, 4853 ± 4863 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc SHORT REPORTS Expression, cellular distribution and protein binding of the glioma ampli®ed sequence (GAS41), a highly conserved putative transcription factor

A Munnia1,4, N SchuÈ tz1,4, BFM Romeike2, E Maldener1, B Glass1, R Maas1, W Nastainczyk3, W Feiden2, U Fischer1,5 and E Meese*,1,5

1Institut fuÈr Humangenetik, UniversitaÈtskliniken des Saarlandes, 66421 Homburg/Saar, Germany; 2Abteilung fuÈr Neuropathologie, UniversitaÈtskliniken des Saarlandes, 66421 Homburg/Saar, Germany; 3Institut fuÈr Medizinische Biochemie, UniversitaÈtskliniken des Saarlandes, 66421 Homburg/Saar, Germany

The glioma ampli®ed sequence 41 (GAS41) was for 40% of the central nervous system tumors (Sehgal, previously isolated by microdissection mediated cDNA 1998; Kleihues et al., 1995). The Glioma Ampli®ed capture from the glioblastoma multiforme cell line Sequence 41 (GAS41), originally isolated from the TX3868 and shown to be frequently ampli®ed in human glioblastoma multiforme (GBM) cell line TX3868 by gliomas. We determined the complete cDNA sequence of microdissection mediated cDNA capture (Fischer et al., the GAS41 gene, demonstrated that the GAS41 protein 1996), was found to be frequently ampli®ed in human is evolutionarily conserved, speci®cally at the N- gliomas (Fischer et al., 1997). The TX3868 cell line terminus, and identi®ed the yeast transcription factor contains a homogeneously staining region (hsr) with tf2f domain within the GAS41 sequence. A human ampli®ed sequences on chromosome 12q13 ± 15. Ab- multiple-tissue Northern blot revealed ubiquitous expres- normalities like hsr and/or double minutes are sion of GAS41 with the highest expression in human frequently found in human tumors and thought to be brain. After generating polyclonal antibodies we found a manifestation of the intrinsic genomic instability of GAS41 protein expression in the nucleus of the TX3868 cancer cells. In addition, it is presumed that over- cell line by Western blot analysis and immuno¯uores- expression of ampli®ed genes confers a selective cence microscopy. The nuclear localization was con- advantage on such cell clones (Livingstone et al., ®rmed for several human tumors including gliomas of 1992; Fischer et al., 1996). Therefore, over-expressed dierent grades of malignancy. In neuroblastoma proteins could potentially play an important role in however, GAS41 was found in the nucleoli but not in tumorigenesis (Suzuki and Iwaki, 2000; Murakami et the nucleoplasm. Yeast two-hybrid screening of the al., 2000). GAS41 shows an increased copy number in TX3868 cell line identi®ed the nuclear mitotic apparatus pilocytic astrocytomas, an entity of gliomas that protein (NuMA), the KIAA1009 protein, and prefoldin usually occurs in the pediatric population and subunit 1 (PFDN1) as potential interacting partners of corresponds to WHO grade I and in gliomas of the GAS41. We generated a polyclonal antibody against the adult which represent a separate tumor entity (Fischer KIAA1009 protein and we demonstrated that the et al., 1997). KIAA1009 protein is a nuclear protein, which appears To learn more about the possible function of GAS41 to be co-localized with the GAS41 protein and NuMA. and its over-expression in the GBM cell line TX3868, Oncogene (2001) 20, 4853 ± 4863. the deduced structure of the GAS41 protein was compared to other proteins. GAS41 has a potential Keywords: ampli®cation; GAS41; glioma; NuMA; C-terminal a-acidic activation domain common to ENL; KIAA1009; PFDN1; TFIIF eukaryotic transcription factors (Mitchell and Tjian, 1989; Remacle et al., 1997) but no potential DNA binding domain. We reported previously that GAS41 is Introduction homologous to the yeast transcription factor AF-9 and to the human transcription factors AF-9 and ENL Glioma is a collective name for tumors of glial cells (Fischer et al., 1997). Because of these similarities, the including astrocytes. Gliomas of astrocytic origin are potential C-terminal a-acidic activation domain and the the most common primary brain tumors and account lack of a potential DNA-binding domain, we suggested that GAS41 could be an incomplete transcription factor. *Correspondence: E Meese, Department of Human Genetics, In this study, we report the complete nucleotide University of Saarland, Building 60, 66421 Homburg/Saar, sequence of the human GAS41 gene, and compare Germany; E-mail: [email protected] the encoded GAS41 protein sequence to proteins 4 These authors contributed equally to this work from diverse species using the BLAST-P, BLAST-X 5Equal contribution as senior authors Received 18 December 2000; revised 9 May 2001; accepted 23 May and BLAST-N algorithms. We were able to identify a 2001 conserved tf2f domain in GAS41 that is also present Cellular distribution and protein binding of GAS41 A Munnia et al 4854 in other transcription factors of yeast and humans. binding partners of the GAS41 protein by screening The GAS41 gene is expressed in various human the TX3868 GBM cell line. tissues as demonstrated by Northern blot analysis and immunohistochemical staining. Using polyclonal antibodies speci®c for the GAS41 C-terminus, we Results localized GAS41 protein expression in the nucleus by Western blot analysis, immuno¯uorescence micro- The nucleotide and corresponding amino acid sequence scopy and immunohistochemical staining. To further of human GAS41 is shown in Figure 1. The cDNA investigate the function of the GAS41 gene, we used sequence of 1444 basepairs (bp) revealed a single open the yeast two-hybrid system to identify potential reading frame with two ATG codons located 12 bases

Figure 1 The nucleotide and corresponding amino acid sequence of human `glioma ampli®ed sequence' (GAS41). The sequence data is available in the NCBI database under the accession number U61384. The corresponding amino acids are placed under the coding nucleotides. The 3' polyadenylation signal, AATAAA, is underlined. The peptide sequence used for generating polyclonal antibodies are shaded: K36KREEDGHTHQWT48 for the N-terminal antibody GAS41-N and K212NEIRKLEEDDQAKDI227 for the C-terminal antibody GAS41-C. The positions of the used primers are marked as GAS41 for and GAS41rev

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4855 apart at positions 222 and 234. A 481 bp 3' Table 1. High identities and similarities are found for untranslated region ¯anks the coding sequence with a vertebrates such as Rattus norvegicus (AI031024 and polyadenylation sequence (AATAAA) located at posi- AI012982), Mus musculus (AA529582 and AA086873), tion 1362. The human GAS41 gene sequence that Sus scrofa (AW314206), Bos taurus (AV597623), Gallus contains seven exons potentially encodes a protein with gallus (AJ396415), and Xenopus laevis (AW767025) 227 amino acids and a calculated molecular mass of ranging from 85.2% for Bos taurus to 100% for Sus 26.5 kDa. Using the BLAST-P, BLAST-X, and scrofa. Using the BLAST-P algorithm, we furthermore BLAST-N algorithms, we examined the conservation identi®ed proteins similar to the human GAS41 protein of the human GAS41. As demonstrated in Figure 2, in Drosophila melanogaster (AAF52462), Caenorhabdi- the GAS41 protein is highly conserved among species, tis elegans (CAB1234), Saccharomyces cerevisiae especially at the N-terminus (position R15-L120). For (P53930) and Schizosaccharomyces pombe (Q10319). some species no complete protein sequence was However, their protein sequences have lower identities available in the databases. In these cases, we used the and similarities compared to the vertebrate species. The EST (expressed sequence tag) database to identify protein encoded by the CG92907 gene of Drosophila nucleotide sequences similar to the GAS41 nucleotide melanogaster (AAF52462) has a similarity of 62% and sequence. an identity of 52% to the complete human GAS41 and Using ClustalW program we demonstrated high is predicted to be a 195 amino acid transcription factor similarities and identities between the human GAS41 (FlyBase, 1999; Adams et al., 2000). The most similar protein and proteins of other species as summarized in protein from Caenorhabditis elegans (CAB01234) is 211

Figure 2 Alignment of the GAS41 protein (U61384) with homologous database sequences demonstrates conservation of GAS41 in 11 dierent species. Identical amino acids (aa) are inverted, similar aa are shaded. Gaps inserted by the sequence alignment software (ClustalW) are indicated by horizontal strokes (7). The used open reading frame (ORF's) and expressed sequences taqs (EST's) are all traceable in the database. For Rattus norvegicus (AI031024 and AI012982), Mus musculus (AA529582 and AA 086873), Sus scrofa (AW314206), Bos taurus (AV314206), Gallus gallus (AJ396415), and Xenopus laevis (AW767025) only partial sequences are in the database. Missing sequence parts are indicated by horizontal strokes (7). The complete ORF's for Drosophila melanogaster (AAF5242), Caenorhabditis elegans (CAB01234), Saccharomyces cerevisiae (P53930), and Schizosaccharomyces pombe (Q10319) are known

Table 1 Summary of the results of Figure 2 which demonstrates the conservation of the GAS41 protein from human to yeast aa ± aa Identity to Similarity to Organism Accession number of GAS41 GAS41 (%) GAS41 (%) Function of the protein

R. norvegicus (EST) AI031024/AI012982 1 ± 133 99.2 99.2 Unknown M. musculus (EST) AA529582/AA086873 4 ± 227 97.4 99.6 Unknown S. scrofa (EST) AW314206 1 ± 103 100 100 Unknown B. taurus (EST) AV597623 86 ± 206 85.2 85.2 Unknown G. gallus (EST) AJ396415 108 ± 227 94.1 100 Unknown X. laevis (EST) AW767025 24 ± 118 87.2 93.6 Unknown D. melanogaster AAF52462 5 ± 227 52 62 Transcription factor C. elegans CAB01234 9 ± 227 53.8 67.8 Probably a transcription factor S. cerevisiae P53930 8 ± 227 25.1 45.8 Unknown S. pombe Q10319 10 ± 227 30.8 46.7 Probably a transcription factor

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4856 amino acids in length with similarity to the AF-9 recombinantly expressed as a GST-GAS41 fusion protein, a transcription factor as well (Figure 2). The protein detectable with both antibodies (data not most homologous Saccharomyces cerevisiae protein shown). (P53930) to human GAS41 is 226 amino acids in To further test the speci®city of the GAS41-C length with a calculated molecular mass of 26.0 kDa antibody for GAS41 protein, we performed a Western and like the Schizosaccharomyces pombe protein blot analysis using the protein lysate prepared from the (Q10319) shares similarities with the human transcrip- GBM cell line TX3868 (Figure 5A). The GAS41-C tion factors AF-9 and ENL and the small subunit of antibody reacts speci®cally with a protein of 26.5 kDa, the yeast transcription initiation factor TFIIF. which is the expected size of the GAS41 protein. When To identify the presence of conserved domains the peptide used to generate the GAS41-C antibody is within the GAS41 protein sequence, we utilized the incubated with the puri®ed GAS41-C serum, it NCBI conserved domain database which de®nes eectively blocks the binding to GAS41 demonstrating conserved domains based upon recurring sequence the speci®city of the puri®ed GAS41-C polyclonal patterns or motifs (Altschul et al., 1997). We identi®ed antibody. To con®rm that the 26.5 kDa signal is not the tf2f domain conserved for the human GAS41 generated by a non-speci®c interaction with the protein sequence with an expected value of 26e726 as secondary antibody, we incubated the Western blot shown in Figure 3. A tf2f domain is also present in the with the anti-rabbit secondary antibody only. No human transcription factor ENL, and the yeast ACN1 signal was revealed with the anti-rabbit secondary protein, which is identical to the transcription initiation antibody. To initially examine the subcellular distribu- factor TFIIF small subunit. tion of the GAS41 protein, Western blot analysis was We performed a human multiple-tissue Northern performed on both nuclear and cytoplasmic fractions blot analysis using RNA isolated from human heart, of the TX3868 GBM cell line using the GAS41-C brain, placenta, lung, liver, skeletal muscle, kidney and polyclonal antibody. As shown in Figure 5B, GAS41 is pancreas as shown in Figure 4. Using a 220 bp located predominantly in the nucleus as only the fragment from the GAS41 gene as probe, a *1.4 kb transcript was detected in all of the eight tissues tested. It appears that the GAS41 protein is expressed in many human tissues. After normalization with a GAPDH probe GAS41 appeared to show elevated expression in brain. We generated polyclonal antibodies speci®c for GAS41, and chose amino acids 36 ± 48 with the peptide sequence, KKREEDGHTHQWT, to generate the polyclonal antibody GAS41-N against the amino- terminus (Figure 1). A second peptide sequence, KNEIRKLEEDDQAKDI, corresponding to the amino acids 212 ± 227 was used to generate the GAS41-C polyclonal antibody against the carboxy-terminus (Figure 1). After peptide synthesis, rabbits were immunized, and blood was obtained for preparation of antibody-containing serum. Before and after puri- ®cation of the sera on peptide binding columns, both Figure 4 Human multiple-tissue Northern blot containing 2 mg polyclonal antibodies, GAS41-N and GAS41-C, were of RNA per lane (Clontech) probed with a 32P-labelled GAS41 tested for their speci®city. To this end, a GAS41 and a GAPDH probe revealed a 1.4 kb GAS41 transcript and a sequence was cloned into the pGEX4T1 vector and 1.3 kb GAPDH transcript respectively

Figure 3 Similarity of the GAS41 protein to the tf2f domain of the yeast transcription factor TFIIF. Identical amino acids (aa) are inverted, similar aa are shaded. Gaps inserted by the sequence alignment software are indicated by horizontal strokes (7)

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4857 dierent neoplastic tissues like human malignant gliomas, medulloblastomas, and in rat C6 gliomas. The ®ndings of GAS41 expression in normal tissues con®rm our Northern blot experiments. In rats with stereotactically implanted C6 gliomas GAS41-immuno- reactivity of normal brain tissue and tumor tissue can be compared within the same tissue section. The expression of GAS41 in the tumor is higher than in the directly adjacent pre-existing normal brain tissue. All tested gliomas contained at least some GAS41-immunopositive cells, whereas the intensity of staining was independent of the grade of malignancy (data not shown). To examine the subcellular distribution of GAS41 on a more sensitive level we employed immunocytochem- ical staining. As shown in Table 2 the following cell cultures and cell lines respectively express the GAS41 protein in the nucleoplasm but not in the nucleoli: A cell culture derived from an astrocytoma grade I (T5498), six cell cultures derived from astrocytomas grade II (T4747, T4761, T4772, T4951, T5184 and T5473), the cell line TX3868, the GBM cell culture T5135, a meningioma cell culture, the prostate Figure 5 (A) Speci®city of the GAS41-C polyclonal antibody for GAS41 expressed in the glioblastoma multiform cell line TX3868 carcinoma cell line PC 3, the breast carcinoma cell by Western blot analysis. Lanes 1 ± 3 contain protein lysates of line MCF-7, the HeLa cells, the cell culture of primary TX3868. Lane 1 was incubated ®rst with the GAS41-C polyclonal astrocytes NHA, and the cell culture of ®broblasts antibody and subsequently with a secondary anti-rabbit antibody, KH5436. All these cells show the same localization of revealing the GAS41 protein with the expected size of 26.5 kDa. Lane 2 was incubated with the GAS41-C peptide and the GAS41- the GAS41 protein exemplarily documented for T4761 C polyclonal antibody ®rst and subsequently with the secondary and TX3868 (Figure 6A, e ± h). In contrast cells of the anti-rabbit antibody. Lane 3 was incubated with only the neuroblastoma cell lines SK-N-SH, Neuro 2a, and secondary anti-rabbit antibody (B) Localization of the GAS41 NS20Y express GAS41 speci®cally in the nucleoli and protein by Western blot analysis of nuclear and cytoplasmic not in the nucleoplasm as con®rmed by the double fractions of the TX3868 cells using the GAS41-C polyclonal antibody. Lane 1 contains total protein lysate, Lane 2 the staining with the Nucleolin-speci®c antibody (data not cytoplasmic fraction, and Lane 3 the nuclear fraction of the shown). This subcellular distribution of the GAS41 TX3868 cells. (C) Speci®city of the KIAA1009 polyclonal protein in neuroblastoma is exemplarily documented antibody for KIAA1009 expressed in the glioblastoma multiform for SK-N-SH and Neuro 2a (Figure 6A, a ± d). cell line TX3868 by Western blot analysis. Lanes 1 and 2 contain protein lysates of TX3868. Lane 1 was incubated ®rst with the Because GAS41 contains a C-terminal acidic activat- KIAA1009 polyclonal antibody and subsequently with a ing domain frequently found in eukaryotic transcrip- secondary anti-rabbit antibody, revealing the KIAA1009 protein tion factors (Mitchell and Tjian, 1989; Remacle et al., with the expected size of 153 kDa. Lane 2 was incubated with the 1997) and lacks a typical DNA-binding domain for KIAA1009 peptide and the KIAA1009 polyclonal antibody ®rst transcriptional activation, we ®rst determined its and subsequently with the secondary anti-rabbit antibody (D) Localization of the KIAA1009 protein by Western blot analysis potential capability of transactivating yeast transcrip- of nuclear and cytoplasmic fractions of the TX3868 cells using the tion using the yeast-two hybrid system. As shown in KIAA1009 polyclonal antibody. Lane 1 contains total protein Figure 7A, the hybrid protein of GAS41 and the DNA lysate, Lane 2 the nuclear fraction, and Lane 3 the cytoplasmic binding domain of the arti®cial yeast system expressed fraction of the TX3868 cells from the pBD/GAS41 construct (GAS41 DNA from position 300 ± 929 cloned into the pBD-Cam vector) does not activate transcription of the LacZ gene. 26.5 kDa protein is clearly revealed in the nuclear Because the GAS41 activating domain together with fraction and in the total protein lysate. To verify the the DNA-binding domain of the yeast two-hybrid separation of nuclear and cytoplasmic fractions, the system does not activate the LacZ repressor gene, the fractions were incubated with an antibody against the system could be used to search for protein interacting human Ki67 protein that is located in the nucleus only partners of the GAS41 protein. (data not shown). We cloned cDNA from the GBM cell line TX3868 Using the GAS41-C antibody for immunohistochem- into the pAD vector of the yeast two-hybrid system to ical staining and in immuno¯uorescence microscopy identify interacting partners of GAS41. The resulting experiments, we determined the GAS41 protein expres- construct was called pAD/TX3868 and was co- sion pattern in the nucleus of various tissue sections and transformed with the pBD/GAS41 construct into the cell types. As shown in Figure 6B, GAS41 protein is YRG-2 yeast strain. The resulting fusion protein expressed in the nucleus of dierent normal tissues like GAL4-BD/GAS41 was used as `bait' to screen the human heart, placenta and brain, as well as in normal human GBM cell line TX3868. Screening 300 000 yeast rat brain. Furthermore it is expressed in the nucleus of clones, we observed b-galactosidase expression in 13

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4858

Figure 6 (A) Immuno¯uorescence microscopy of GAS41 with the GAS41-C polyclonal antibody. In human neuroblastoma cell line SK-N-SH (a and b) and the mouse neuroblastoma cell line Neuro 2a (c and d) GAS41 is localized in the nucleoli. In the cell culture T4761, derived from a astrocytoma grade II (e and f) and in the TX3868 cell line (g and h) the GAS41 protein is expressed in the nucleoplasm only sparing the nucleoli. Double staining with the GAS41-C antibody (j) and a monoclonal antibody speci®c for NuMA (k) in the TX3868 cell line indicates the same subcellular localization of the GAS41 protein and NuMA (i ± k). Double staining with the GAS41-C antibody (m) and the KIAA1009-speci®c antibody (n) in TX3868 cells (I-n). (B) Immunohistochemical staining with the GAS41-C antibody in human heart (a), human placenta (b), human brain (c) human GBM (d), human medulloblastoma (e), rat C6 glioma (f). The ®gure of the rat C6 glioma shows normal brain tissue (top) and tumor tissue (bottom)

clones. We isolated the DNAs from these clones, with the pBD/GAS41 construct into the yeast strain sequenced the DNA, and re-transformed each DNA YRG-2 to verify the LacZ and HIS3 interactions for

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4859 each clone. After this veri®cation, nine clones remained were able to transactivate transcription of the LacZ b-galactosidase positive. One of the positive clones is reporter gene by themselves. identical to a part of NuMA (nuclear mitotic apparatus The NuMA cDNA with a length of 7339 bp protein) and a second clone is identical to a part of the contains an ORF with a start at position 259 bp and PFDN1 (subunit 1 of prefoldin). Seven clones are a stop at position 6562 bp (Figure 8A). The GAS41 identical to each other and are part of the KIAA1009 protein interacts with the protein part of NuMA protein (Figure 7B). None of the three binding partners encoded by bp 3386 ± 6562. The NuMA protein contains an essential domain for spindle association (aa 1687 ± 1814), a DNA binding domain (aa 1910 ± 1976) and a ligand binding domain (aa 2022 ± 2242) Table 2 Subcellular distribution of GAS41 in dierent cell types (Figure 8A) (Richard et al., 1997). Double staining Localisation with a monoclonal antibody speci®c for NuMA Name Description of GAS41 Organism (Transduction Laboratories) and GAS41-C antibody Astrocytomas indicates the same subcellular localization of the T5498 + Astrocytoma grade I Nucleus Human GAS41 protein and NuMA (Figure 6A, i ± k). T4747 + Astrocytoma grade II Nucleus Human The PFDN1 clone is composed of 1322 bp with the T4761 + Astrocytoma grade II Nucleus Human ORF between position 48 bp and 414 bp (Figure 8C). T4772 + Astrocytoma grade II Nucleus Human T4951 + Astrocytoma grade II Nucleus Human The portion of the PFDN1 protein that interacts with T5184 + Astrocytoma grade II Nucleus Human GAS41 is encoded by the sequence between positions T5473 + Astrocytoma grade II Nucleus Human 333 bp and 414 bp. PFDN1 is a subunit of prefoldin, TX3868 * GBM Nucleus Human that plays an essential role in the folding of native T5135 + GBM Nucleus Human proteins like actin and/or tubulin, functioning as a Neuroblastomas chaperone (Vainberg et al., 1998; Hansen et al., 1999). SK-N-SH * Neuroblastoma Nucleoli Human The sequence of the KIAA1009 clone that was Neuro 2a * Neuroblastoma Nucleoli Mouse originally isolated in the Kazusa cDNA project NS20Y * Neublastoma Nucleoli Mouse (Nagase et al., 1999) is 5103 bp long with an ORF Other tumor types from position 251 bp to position 4234 bp (Figure 8B). + Meningioma Nucleus Human The portion of the KIAA1009 protein that interacts PC 3 * Prostate carcinoma Nucleus Human with GAS41 is encoded by the sequence from position MCF-7 * Breast carcinoma Nucleus Human 690 bp to 2510 bp. To generate a polyclonal antibody HeLa * Adenocarcinoma, Nucleus Human cervix we chose amino acids 252 ± 265 with the peptide sequence CENEENSKNISTMES, corresponding to Normal cells the KIAA1009 protein sequence with the accession NHA + Primary astrocytes Nucleus Human number XM 004364. To test the antibody for its KH5436 + Fibroblast cells Nucleus Human speci®city a KIAA1009 sequence from position 694 ± Nucleus: staining of the necleoplasm sparing the nucleoli; Nucleoli: 2468 bp (XM 004364) was cloned into the pQE-30 staining of the nucleoli only; * Cell line; + cell culture vector and recombinantly expressed. The correspond-

Figure 7 Yeast two-hybrid screen. (A) YRG-2 yeast cells transformed with the GAS41/pBD construct were screened for HIS3 and LacZ expression. These cells are able to produce histidine and unable to activate the transcription of the LacZ reporter gene (no blue staining). (B) Identi®cation of potential binding partners of the GAS41 protein. The bait protein (GAS41 construct) was fused to the GAL4 DNA-binding domain (pBD-Cam phagemid vector). The TX3868 cDNA was cloned in the pAD phagemid vector and interacting proteins identi®ed by HIS3 and LacZ screening. Every identi®ed clone was extracted, sequenced and retransformed together with the GAS41/pBD construct into the YRG-2 yeast cells to verify the interaction. Veri®cation by HIS3 and LacZ screening is demonstrated in this ®gure for the three binding partners: KIAA1009 protein, NuMA protein, and PFDN1 protein

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4860 (Albrecht et al., 2000). Speci®cally in human gliomas, gene ampli®cation is proposed to be a decisive event that contributes to tumorigenesis (Heiskanen et al., 2000). The ampli®cation of the GAS41 gene in grade II diuse gliomas provided the ®rst evidence of an ampli®cation present in low-grade human gliomas, suggesting that this gene is involved in the early process of tumorigenesis (Fischer et al., 1997). The function of the GAS41 protein is suggested, in part, by the primary amino acid sequence and by comparisons with other proteins in the databases. An enrichment of acidic amino acids (27%) within a stretch of 60 amino acids in the C-terminus suggests that GAS41 may be a transcription factor as this kind of negatively charged acidic a-helical structure is typical for activation domains in eukaryotic transcription factors (Mitchell and Tjian, 1989; Remacle et al., 1997). Previously, GAS41 was shown to have a high similarity to the yeast AF-9 protein and to the human AF-9 and ENL proteins, which are involved in translocation events and are suggested to belong to a new class of transcription factors (Nakamura et al., Figure 8 Three potential binding partners (NuMA, KIAA1009, 1993). Human GAS41 appears to be an incomplete and PFDN1) of the GAS41 protein, identi®ed by the yeast-two hybrid screen. (A) The ®rst clone identi®ed in the yeast two transcription factor because it lacks a DNA binding hybrid screen is identical to the 7339 bp long NuMA cDNA from domain. position 3386 ± 7339 bp and contains a domain for spindle In this report, we demonstrated that the GAS41 association, a DNA binding domain, and a ligand binding protein is not only highly conserved within the species domain. (B) The second identi®ed clone is identical to the KIAA1009 cDNA clone from position 690 ± 2510 bp and is of mammals. Generally, the highest conservation was located within the ORF of KIAA1009. (C) The third identi®ed found in vertebrates and a lower identity/ similarity clone is identical to the PFDN1 cDNA sequence from position in non-vertebrates like Drosophila melanogaster 333 ± 1322 bp (AAF52462), Caenorhabditis elegans (CAB01234), Sac- charomyces cerevisiae (P53930) and Schizosaccharo- myces pombe (Q10319). The proteins of these four ing HIS-tagged protein has been detected by Western species are partially characterized as probable tran- blotting with the polyclonal antibody for KIAA1009 scription factors in the NCBI database and show (data not shown). similarities to the AF-9 transcription factor as well. Using the protein lysate prepared from the GBM cell GAS41 contains a tf2f domain that is present in the line TX3868, the KIAA1009 antibody reacts with a yeast transcription factor TFIIF and in other tran- 153 kDa protein in the Western blot analysis (Figure scription factors such as the human transcription factor 5C). This size corresponds to the expected size of the ENL and in the yeast ACN1 protein. The tf2f domain KIAA1009 protein. The binding to the KIAA1009 is described as a conserved domain in the transcription protein is eectively blocked using the peptide which factor TFIIF of ®ssion yeast Schizosaccharomyces was used to generate the KIAA1009 antibody (Figure pombe and is predicted to be a chromatin modifying 5C). protein. The yeast ACN1 protein, which contains the To examine the subcellular distribution of the tf2f domain as well, is also identical to the transcrip- KIAA1009 protein we employed Western blot analysis tion initiation factor TFIIF small subunit and is an and immuno¯uorescence microscopy. Using both integral member of two basal transcription factor nuclear and cytoplasmic fractions of the TX3868 complexes, TFIID and TFIIF. ACN1 is an interacting GBM cell line, Western blot analysis shows KIAA1009 component of the SWI/SNF chromatin remodeling protein in the nucleus, but not in the cytoplasmic complex (Cairns et al., 1996). It interacts with the SWI/ fraction (Figure 5D). These results are consistent SNF complex by binding to the SNF5 protein. The with the immuno¯uorescence microscopy experiments Snr1 protein of Drosophila is similar to the SNF5 (Figure 6A, n). In addition, double staining indicates protein and has been shown to be associated in large the same subcellular distribution of the GAS41 protein complexes equivalent to the yeast SWI/SNF (Wang et and the KIAA1009 protein (Figure 6A, l ± n). al., 1996). The human protein INI1 is similar to SNF5 and may be part of an equivalent human complex. These database comparisons also demonstrate that Discussion GAS41 is highly conserved among eukaryotes and is likely to function as a transcription activation domain. Gene ampli®cation in eukaryotes plays an important Several lines of evidence indicate that GAS41 is a role in drug resistance, tumorigenesis, and evolution transcription factor including the negatively charged

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4861 acidic a-helical activation domain, high similarities to protein as bait in the yeast two-hybrid system, other transcription factors and its expression in the Harborth et al., 2000 have identi®ed the GAS41 nucleus. Since GAS41 is missing a DNA binding protein as an interacting partner. domain, we used the yeast two-hybrid system to test We have demonstrated that, like the NuMA protein, the ability of GAS41 to transactivate yeast transcrip- GAS41 is also expressed in the nucleus (Figures 5B, 6A tion on its own and then to identify protein interacting and 6B). GAS41 protein is localized in the nucleo- partners of GAS41. GAS41 was not able to transact- plasm, but not in the nucleoli, of all tested cell lines ivate yeast transcription on its own probably because and cell cultures exclusive of the tested neuroblastoma this process requires that several components including cell lines. We suggest that the dierent subcellular DNA-binding proteins, transactivation proteins, co- distributions of the GAS41 protein may be dependent activators and accessory proteins must be present upon the cell type. In addition, the expression of (Martin, 1991; NaÈ aÈ r et al., 1999). GAS41 in the nucleus is consistent with the location of It is important to notice that there are described a transcription domain/factor. dierences in transactivation between organisms like The independent ®ndings of the interaction between human and yeast (Carlson and Laurent, 1994). For GAS41 and NuMA by us and by Harborth et al., 2000 example, the yeast transcription factor GCN4 contain- and their common nuclear location make their ing an acidic activation domain needs the co-activators interaction at some point during the cell cycle highly ADA2, ADA3 and CGN5 (Barlev et al., 1995). In probable. We suggest a model for the potential multi- mammalian cells many DNA binding domains need the protein transcription complex in which GAS41 with its coactivator p300/CBP to initiate transcription (Jahn- acidic activation domain has the potential to activate knecht and Hunter, 1996). Using the yeast-two hybrid transcription in human cells in combination with a system, we did identify three potential binding partners second protein containing a DNA-binding domain for GAS41: NuMA (nuclear mitotic apparatus pro- such as NuMA, for example (Figure 9). It is highly tein), KIAA1009 protein and PFDN1 (subunit 1 of probable that GAS41 builds a speci®c transcription prefoldin). complex with supplementary proteins, as in the yeast KIAA1009 was originally identi®ed in the course of SWI-SNF complex (Wang et al., 1996 and Dingwall et sequencing cDNA clones from size-fractionated human al., 1995), that are essential for the function of the brain cDNA libraries (Nagase et al., 1999). The GAS41 protein. Additionally, it is not inconceivable KIAA1009 clone is expressed in heart, brain, lung, that GAS41 may also interact with prefoldin, which liver, skeletal muscle, kidney, pancreas, spleen, testis, plays an essential role in the folding reaction of native ovary and diverse brain regions with the lowest proteins like actin and/or tubulin (Vainberg et al., expression in the spleen, and the highest expression in 1998; Hansen et al., 1999), as NuMA associates with the ovary (Nagase et al., 1999). We found expression of tubulin microtubules and stabilizes the mitotic spindle the KIAA1009 clone in leukocytes and in the during mitosis. Further functional experimentation is glioblastoma cell line TX3868 (data not shown). The required to de®nitively demonstrate the role played by only indication of function for the KIAA1009 clone is GAS41 within the transcription complex. its possible interaction with GAS41. We generated a polyclonal antibody speci®c for the KIAA1009 protein and demonstrated that KIAA1009 is located in the nucleus only. Immuno¯uorescence microscopy indicates co-localization of the GAS41 protein, the KIAA1009 protein and NuMA. These results are consistent with the idea of the interaction of these three proteins. NuMA is a component of the putative mitotic spindle matrix in vertebrates (Dionne et al., 1999) and shows a striking change in location during the cell cycle, moving from the nuclear matrix in interphase to the spindle poles in mitotic cells (Lyderson and Pettijohn, 1980; Wang et al., 1999). This change is accompanied by speci®c changes in phosphorylation of the NuMA protein (Compton and Luo, 1995; Zeng, 2000) and presumably also by changes in the binding partners with which NuMA interacts, one of these is possibly GAS41. In interphase, NuMA is located in the nucleus as a component of the nuclear matrix (Luderus et al., 1994). Immunoelectron microscopy shows that nuclear lattices are built from NuMA (Gueth-Hallonet Figure 9 Model of the potential GAS41 multi-protein transcrip- et al., 1998). These lattices could possibly provide a tion complex. The proposed transcription factor complex contains GAS41 as DNA-activation domain, NuMA as DNA-binding framework on which nuclear processes, such as domain, coactivators and accessory helper proteins (X and Y), transcription, take place. When using the NuMA and basic transcription factors

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4862 and the pellet was lysed in lysis buer. After an additional Materials and methods centrifugation step of 5 min at 6000 r.p.m. at 48C, the supernatant contains the cytoplasmic fraction and the pellet DNA sequencing contains the nuclei. The pellet was washed in wash buer Sequencing was performed according to the manufacturer's (lysis buer without NP40) and centrifuged 5 min at instructions using the Perkin Elmer ABIPrism Cycle Sequen- 6000 r.p.m. The pellet was lysed in extraction buer cing kit. Clone inserts were sequenced with an automated (25 mM Tris-HCl, 1 mM EDTA, 0.5 mM NaCl, 10 mM b- sequencer, LI-COR dna sequencer model 4000L. Mercaptoethanol, 0.6% (v/v) TritonX100) and incubated on ice for 40 min. The cell debris was collected by centrifugation for 30 min at 40 000 r.p.m. Northern blot A multiple tissue Northern blot obtained from Clontech was Gel electrophoresis and Western blot hybridized with a [a-32P]-dATP labeled 220 bp fragment of GAS41 and with a [a-32P]-dATP labeled 1.3 kb fragment of Proteins were subjected to 12.5% SDS ± PAGE for experi- GAPDH respectively. Hybridization conditions were as ments with GAS41 and to 7.5% SDS ± PAGE for experi- described in the manufacturer's instructions (Clontech). ments with KIAA1009 and transferred to PVDF membranes (Millipore, Bedford). The GAS41-C antibody and the KIAA1009 antibody were used at a 1 : 1000 dilution. Reactive Peptide synthesis proteins were detected by appropriate Horseradish-perox- The GAS41-N peptide composed of amino acids 36 ± 48, the idase conjugated anti-IgG antibodies and visualized with the GAS41-C peptide composed of amino acids 212 ± 227 (Figure ECLTM Western Blotting Detection Reagent (Amersham 1) and KIAA 1009 peptide composed of the amino acids Pharmacia Biotech). CENEENSKNISTMES were synthesized using the Fmoc method of solid phase peptide chemistry (Merri®eld 1963) on Immunofluorescence a Milligen / Biosearch 9050 Pep-Synthesizer. Cleavage of the peptides from the resin and removal of the protecting groups We used the following cell lines and cell cultures respectively: were achieved by treating the peptide with 90% tri¯uoro- a cell culture derived from an astrocytoma grade I (T5498), acetic acid, 2.5% phenol, 2.5% 1,2-ethanedithiol, 1% six cell cultures derived from astrocytomas grade II (T4747, triisopropylsilane and 5% water. T4761, T4772, T4951, T5184, and T5473), the cell line TX3868 and the GBM cell culture T5135, a meningioma cell culture, the prostate carcinoma cell line PC 3, the breast Generation of polyclonal antibodies carcinoma cell line MCF-7, the HeLA cell line, the cell Antibodies were generated against the peptide plus an culture of primary astrocytes NHA, the cell line of ®broblasts additional amino-terminal cysteine. The peptide was coupled KH5436, neuroblastoma cell lines SK-N-SH, Neuro 2a, and to keyhole limpet hemocyanin that had been activated with NS20Y. The cells were grown on coverslips, ®xed with 3.7% N-succinimidyl-3-maleinimido-proprionate. To obtain anti- formaldehyde for 15 min at room temperature, and permea- bodies, rabbits (bastards) were immunized by subcutaneous bilized with 0.1% Triton X-100 in PBS for 15 min at room injections of 300 mg of coupled peptide emulsi®ed in Freund's temperature. The cells were incubated for 1 h with the complete adjuvant (Sigma) (Harlow and Lane 1989). The primary antibody (GAS41-C, 1 : 5 dilution). After washing immunization was repeated twice at two week intervals with three times with PBS they were incubated with ¯uorescein- the same amount of coupled peptide in Freund's incomplete conjugated goat anti-rabbit IgG (Jackson Research) for 1 h. adjuvant. The rabbits were bled 8 ± 10 days after immuniza- The coverslips were then washed three times with PBS and tions. The sera were tested after each bleeding on mounted in Vectashield (Vector Laboratories). To demon- recombinant protein by Western blot. strate the same subcellular distribution of GAS41 protein, KIAA1009 protein and NuMA the TX3868 cell line triple labeling was performed by incubation with three protein- Purification of antibodies speci®c antibodies and the corresponding secondary anti- Pre-tested rabbit sera were puri®ed on peptide binding bodies in the following succession: Incubation with GAS41-C HiTrap TMNHS-activated Sepharose TMHP columns (Amer- (1 : 5), ¯uorescein-conjugated goat anti-rabbit IgG (Jackson sham Pharmacia BiotechAB, Uppsala, Sweden) according to Research), a KIAA1009-speci®c antibody, Alexa Fluor 594 the manufacturer's instructions. (Molecular Probes, 2 mg/ml), a NuMA speci®c antibody (Transduction Laboratories; 1 : 25), and AMCA-conjugated AniPure goat anti-mouse IgM, m chain speci®c antibody Cell culture with minimal cross-reaction to human, bovine, and horse All cells were cultured in DMEM containing 10% FCS, serum proteins (Dianova; 1 : 25). Each antibody incubation exclusive to the cell culture of primary astrocytes NHA. took 1 h and the incubation steps were separated by washing These cells were cultured in Astrocyte Growth Medium three times with PBS. BulletKit (CellSystems). Immunohistochemical staining Cell fractionation Immunohistochemical staining was performed on formalin TX3868 GBM cells were rinsed twice with cold PBS and ®xed and paran embedded tissue sections using the avidin- collected (10 min; 1200 r.p.m.). The cell pellet was washed biotin peroxidase method. Pre-treatment was performed with and then lysed in 2 volumes of lysis buer [10 mM MES microwave and citrate buer for 5 min at 700 W and 5 min (pH 7.9), 10 mM NaCl, 1.5 mM MgCl2, 10% Glycerin, 1 mM at 350 W. The primary antibody, GAS41-C, was applied at a EDTA, 5 mM DTT, 1% (v/v) NP40 3 mg/ml] 5 min at 48C. dilution of 1 : 100 for an incubation period of 20 min at 378C. The lysate was centrifuged at 1500 r.p.m. for 5 min at 48C Negative controls were performed by incubating tissue

Oncogene Cellular distribution and protein binding of GAS41 A Munnia et al 4863 sections with antibody diluent (DAKO) instead of the The YRG-2 yeast strain was simultaneously transformed with primary antibody, GAS41-C. The speci®c reaction of the bait plasmid pBD-GAS41, and the human TX3868 GAS41-C antibody was detected with a secondary biotin cDNA library cloned into the pAD-GAL4 vector, according labeled antibody (swine anti-rabbit /DAKO). Sections were to the manufacturer's instructions. Positive interactions were counterstained with hematoxylin after the DAB reaction. selected by histidine auxotrophy. LacZ activity was determined for all growing colonies using a blue/white ®lter assay according to the manufacturer's instructions. Plasmids from Yeast two-hybrid screening (Stratagene) positive clones were extracted and sequenced. For the yeast-two hybrid screen, we used HybriZAP Two- Hybrid cDNA Gigapack Cloning Kit (Stratagene). A 630 bp fragment of GAS41 corresponding to bases 300 ± 929 was generated by PCR. The primers were synthesized by an Acknowledgments RNA/DNA synthesizer (Applied Biosystems) with the The authors thank Dr WI Steudel and the sta of the following sequences: GAS41 for (5' - GTC GAC TAC Neurosurgical Clinic for providing fresh tumor samples, Dr GGT AAT GTT GCT CGG-3') and GAS41 rev (5'-CTG Norbert Schuster, Department of Anatomy, for the kind CAG CTA CCA AGT TCT CAT GAG-3'). The puri®ed gift of the nucleolin-speci®c antibody and the neuroblas- DNA fragment was PstI/SalI digested at 208C overnight, toma cell lines and Dr Jens Meyer for critical reading of ligated into the pBD-GAL4-Cam phagemid vector (Strata- the manuscript. This work was supported by a Grant from gene) and transformed in competent E. coli XL1/MRF cells. the Deutsche Forschungsgemeinschaft (SFB 399, A1).

References

Adams MD, Celniker SE, Gibbs RA, Rubin GM and Venter Kleihues P, Soylemezoglu F, Schauble B, Scheithauer BW CJ. (2000). Science, 287, 2185 ± 2195. and Burger PC. (1995). Glia, 15, 211 ± 221. Albrecht EB, Hunyady AB, Stark GR and Patterson TE. Livingstone LR, White A, Sprouse J, Livanos E, Jacks T and (2000). Mol. Biol. Cell., 11, 873 ± 886. Tlsty TD. (1992). Cell, 70, 923 ± 935. Altschul SF, Madden TL, SchaÈ er AA, Zhang J, Zhang Z, Luderus ME, den Blaauwen JL, de Smit OJ, Compton DA, Miller W and Lipman DJ. (1997). Nuc. Acids. Res., 25, van Driel R. (1994). Mol. Cell. Biol., 14, 6297 ± 6305. 3389 ± 3402. Lyderson BK and Pettijohn DE. (1980). Cell, 22 (Pt 2), 498 ± Barlev NA, Candau R, Wang L, Darpino P, Silverman N and 499. Berger SL. (1995). J. Biol. Chem., 270, 19337 ± 19344. Martin KJ. (1991). BioEssays, 13, 113 ± 123. Cairns BR, Henry NL and Kornberg RD. (1996). Mol. Cell. Merri®eld RB. (1963). J. Am. Chem. Soc., 85, 2149 ± 2158. Biol., 16, 3308 ± 3316. Mitchell PJ and Tjian R. (1989). Science, 245, 371 ± 378. Carlson M and Laurent BC. (1994). Curr. Opin. Cell. Biol., 6, Murakami Y, Tateyama S, Rungsipipat A, Uchida K and 396 ± 402. Yamaguchi R. (2000). J. Vet. Med. Sci., 62, 783 ± 787. Compton DA and Luo C. (1995). J. Cell. Sci., 108 (Pt 2), NaÈ aÈ r AM, Beaurang PA, Zhou S, Abraham S, Solomon W 621 ± 633. and Tjian R. (1999). Nature, 398, 828 ± 832. Dingwall AK, Beek SJ, McCallum CM, Tamkun JW, Nagase T, Ishikawa K, Suyama M, Kikuno R, Hirosawa M, Kalpana GV, Go SP and Scott MP. (1995). Mol. Biol. MiyajimaN,TanakaA,KotaniH,NomuraNandOhara Cell., 6, 777 ± 791. O. (1999). DNA Res., 6, 63 ± 70. Dionne MA, Howard L and Compton DA. (1999). Cell. Nakamura T, Adler H, Gu Y, Prasad R, Canaani O, Motil. Cytoskeleton, 42, 189 ± 203. Kamada N, Gale RP, Lange B, Crist WM, Nowell PC, FischerU,HeckelD,MichelA,JankaM,HulsebosTand Croce CM and Canaani E. (1993). Proc. Natl. Acad. Sci. Meese E. (1997). Hum. Mol. Genet., 6, 1817 ± 1822. USA, 90, 4631 ± 4635. Fischer U, Meltzer P and Meese E. (1996). Hum. Genet., 98, Remacle JE, Albrecht G, Brys R, Braus GH and Huyle- 625 ± 628. broeck D. (1997). EMBO J., 16, 5722 ± 5729. FlyBase. (1999). Nucleic. Acids Res., 27, 85 ± 88. http:// Richard AW, Catzavelos C and Kamel-Reid S. (1997). ¯ybase.bio.indiana.edu/. Nature Genetics, 17, 109 ± 113. Gueth-Hallonet C, Wang J, Harborth J, Weber K and Sehgal A. (1998). Semin. Surg. Oncol., 14, 3 ± 12. Osborn M. (1998). Exp. Cell. Res., 243, 434 ± 452. Suzuki SO and Iwaki T. (2000). Mod. Pathol., 13, 548 ± 553. Hansen WJ, Cowan N and Welch WJ. (1999). J. Cell. Biol., VainbergIE,LewisSA,RommelaereH,AmpeC,Vade- 145, 265 ± 277. kerckhove J, Klein HK and Cowan NJ. (1998). Cell, 93, Harlow E and Lane D. (1989). Antibodies. A Laboratory 863 ± 873. Manual, Cold Spring Harbor Laboratory: USA. Wang ZH, Ding MX, Chew-Cheng SB, Yun JP and Chew Harborth J, Weber K and Osborn M. (2000). J. Biol. Chem., EC. (1999). Anticancer Res., 19, 5445 ± 5449. 275, 31979 ± 31985. Wang W, Cote J, Xue Y, Zhou S, Khavari PA, Biggar SR, HeiskanenMA,BittnerML,ChenY,KhanJ,AdlerKE, Muchardt C, Kalpana GV, Go SP, Yaniv M, Workman Trent JM and Meltzer PS. (2000). Cancer Res., 60, 799 ± JL and Crabtree GRv. (1996). EMBO J., 15, 5370 ± 5382. 802. Zeng C. (2000). Microscopy Res. Technique., 49, 467 ± 477. Jahnknecht R and Hunter T. (1996). Nature, 383, 22 ± 23.

Oncogene