[CANCER RESEARCH 51. 1553-1560. March 1. 1991]
Human Glial Fibrillary Acidic Protein: Complementary DNA Cloning, Chromosome Localization, and Messenger RNA Expression in Human Glioma Cell Lines of Various Phenotypes1 Erik Bongcam-Rudloff, Monica Nister, Christer Betsholtz, Jia-Lun Wang, Goran Stenman, Kay Huebner, Carlo M. Croce, and Bengt W estermark2 Department of Pathology, I'nirersity Hospital, S-751 85 Uppsala, Sweden fE. B-R., M. .V, C B., J-L. H'., B. H'.];Department ofOral Pathology, I'nirersity of GÃ¶teborg, Sweden f(j. S./: and Fels Institute for Cancer Research and Molecular Biology, Temple I'nirersity School oj Medicine. Philadelphia, Pennsylvania 19140 [K. H., C. M. C.]
ABSTRACT Glial fibrillar) acidic protein (GFAP) is a constituent of intermediate filaments of glial cells of the astrocyte lineage. \Vecloncd a human GFAP complementary DINA,deduced the amino acid sequence, and established the chromosomal location (17q21) of the GFAP gene by Southern blot hybridization of somatic cell hybrids and by in situ hybridization. The authenticity of the complementary DNA was proven by expressing it in glioma cells lacking endogenous GFAP; after microinjection of the com plementary DNA, such cells became positive for staining with GFAP antibodies. The levels of fibronectin (FN) and GFAP mRNA of ten human glioblastoma cell lines, determined by Northern blot hybridization of RNA. were related to other phenotypic characteristics (cell morphology and expression of the genes encoding platelet-derived growth factor (PDGF) receptors). A high expression of GFAP mRNA was found only in cells lacking fibronectin mRNA and protein. Glioma cells with a fibroblastic phenotype (bipolar, FN*/GFAP~) were found to express both types of PDGF receptors (a and 0). Relatively high levels of PDGF areceptor mRNA, in the absence of /3-receptor expression, were found in cell lines that express GFAP and lack detectable levels of fibronectin mRNA. The findings are compatible with the idea that the genes encoding PDGF receptors in glioma cells are regulated in concert with other genes, the expression of which may reflect the developmental program of normal glia cell lineages.
INTRODUCTION Intermediate filaments are 8-10-nm fibers that form part of the cytoskeleton of virtually all cell types in higher eukaryotes (1-4). On the basis of their tissue-restricted distribution, inter mediate filament proteins have been divided into five main classes: desmin; vimcntin; GFAP,1 neurofilament proteins; and keratins. All intermediate filament proteins have the same structural organization with a central Â»-helical rod domain flanked by amino and carboxy termini of variable sizes. GFAP is a constituent of glial cells of the astrocytic lineage (5) and is generally also present in primary glial brain tumors (6). However, a high proportion of cell lines established from malignant gliomas do not express GFAP (7). Immunohistochemical analyses of glioma cell lines have shown that the expression of GFAP and the extracellular matrix protein fibro nectin (8) is mutually exclusive (9), even in cells derived from the same tumor (10). In order to facilitate further studies on the expression of GFAP in human glial cells of different pheReceivcd 8/27/90; accepted 12/17/90. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 I'.S.C. Section 1734 solely to indicate this fact. 1This work was supported by grants from the Swedish Cancer Society and the N1H (CA 25875 and C'A .19860). 2To whom requests for reprints should be addressed. ' The abbreviations used are: GFAP. glial fibrillar}- acidic protein: cDNA. complementary DNA: PDGF. platelet-derived growth factor: SSC. standard saline-citrate; SDS. sodium dodecyl sulfate; MEM, Eagle's minimum essential medium: PBS. phosphate-buffered saline: poly(A)" RNA. polyadenylated RNA: kbp. kilobase pairs: FN. fibronectin.
notypes, we have in the present investigation cloned a full length human GFAP cDNA and determined the corresponding amino acid sequence. The chromosomal location of the GFAP gene was also established. Recent studies have suggested that PDGF has a physiological role in the development of glial cells in the central nervous system. Thus, the bipotential O-2A progenitor cells of the rat optic nerve respond by proliferation to PDGF produced by type 1 astrocytes (11, 12). Given the fact that the \-sis oncogene encodes a PDGF-like growth factor (13-15), the finding that intracerebral injection of simian sarcoma virus in newborn marmosets gives rise to brain tumors of the glioblastoma phe notype (16) suggests that an autocrine activation of the PDGF receptor in glial cells is a causal event in tumorigenesis. Whether a similar autocrine mechanism also operates in the genesis of human "spontaneous" glioblastoma is unclear. There are several examples of human glioma cell lines that express one or both of the PDGF chains (17-19) concomitantly with receptors for PDGF (18-20), but it is not known if activation of these genes is a cause or consequence of malignant transfor mation. PDGF synthesis and expression of the cognate receptor may occur in certain normal glia cells (cf., Refs. 12 and 21), and this phenotype may be retained after malignant transfor mation. We have reasoned that if this were the case, one should expect that expression of PDGF and its receptors be related to other phenotypic markers. This is a difficult problem to study since, particularly in the human, very few differentiation mark ers for glial cells are available, and the progenitor cell for glioblastoma has not been identified. In the present investigation we have used the best character ized astrocyte marker, namely GFAP, and studied its expression in relation to cell morphology and expression of PDGF Â«-and /^-receptor and fibronectin. MATERIALS AND METHODS cDNA Cloning and Nucleotide Sequence Analysis. Standard molecu lar biology techniques (22) were used unless otherwise indicated. The mouse GFAP cDNA clone Gl (23) was radiolabeled using ["PJdeoxynucleotidc triphosphate and the Klenow fragment of DNA polymerase I. The radiolabeled plasmid was used to screen a cDNA library made from the human clonal glioma cell line U-342 MGa C12:6 (24) which expresses GFAP (10). The recombinant phages were plated on Escherichia coli C600 hfl (25). Duplicate nitrocellulose filter lifts were hybridized with 32P-labeled Gl at 42Â°Cin 20% formamide-5 x SSC (1 x SSC is ISO m,M sodium chloride-15 HIM sodium citrate)-50 mivi sodium phosphate. pH 7.0-5 x Denhardt's solution-0.1% SDS-200^g/
ml sonicated single stranded DNA and washed in 0.5 x SSC-0.1 % SDS at the same temperature. Positively hybridizing clones were plaque purified and XDNA was isolated. The cDNA insert was isolated by EcoRI digestion and subcloned in pUC-13. The nucleotide sequence of the human cDNA restriction fragments was determined by the dideox1553
HUMAN GLIAL FIBRILLARE Ai II) PROTEIN
after hybridization were at 40Â°C.Slides were exposed for 8-16 days at 4Â°Cand subsequently developed and G-banded using 0.25% Wright's
ynucleotide chain termination method (26) after subcloning into M13 phage derivatives (27). Analysis of the Expression of Microinjected GFAP cDNA. To deter mine the expression of cloned GFAP cDNA. the clone E2C2 (see below) was inserted in sense orientation into the EcoR\ site of the retroviral vector RD2 (kindly provided by Dr. L. Hellman. Department of Immunology, University of Uppsala). The expression plasmid was designated phGFAP-RD2. The human U-343 MG glioma cell line, which does not express GFAP ( 10), was grow n to subconfluence on 18x 18-mm glass coverslips in Falcon dishes in MEM supplemented with 10% fetal calf serum (Gibco Ltd., United Kingdom) and antibiotics. phGFAP-RD2 DNA, at a concentration of 0.5 ^g/ml, was microinjected into the cell nucleus using an Olympus IMT2-SYF injectoscope. The cells were then incubated at the same culture conditions for 48 h. Cells were fixed in 3% paraformaldehyde in PBS for 10 min at room temperature and in 100% acetone for 1 min at -20Â°C.Immunostaining of GFAP was performed using a rabbit immunoglobulin to cow glial fibrillar) acidic protein (Dakopatts, Denmark) following a rhodamineconjugated swine anti-rabbit IgG antibody (Dakopatts) at room tem perature. Cells were rinsed with phosphate-buffered saline and exam ined by fluorescence microscopy. Cells not treated with injections of GFAP cDNA growing on the same coverslip were used as a control. Chromosomal Localization of the Human GFAP Gene. The GFAP cDNA clone pCIA-5 described in this report (see below) was used to determine the chromosomal localization of the human GFAP gene. Cloned DNA probes for genes which have previously been regionally localized on chromosome 17 were used for comparison and are de scribed in detail in Table I. Probes (entire plasmid including insert and vector) were radiolabeled by nick-translation using [l:P]deox>nucleotide triphosphates to a specific activity of 1 x IO9 cpm/Mg: 1 x 10* cpm were used for each Southern blot hybridization. Isolation, propagation, and characterization of parental cells and somatic cell hybrids have been described (28). For regional localization of the GFAP gene on chromosome 17, a series of hybrid cell lines (Table 1). each retaining only a partial chromosome 17. were used (29. 30). Cellular DNA was isolated as described (31). digested with an excess of restriction endonuclease Hind\\\ or Sstl, size fractionated in 0.8% agarose gels, transferred to nylon membrane using 0.4 M NaOH (32), and dried under vacuum. Prehybridization and hybridization were carried out in 1.5 x saline-sodium phosphate-EDTA ( 1 x saline-sodium phosphate-EDTA is 0.15 M NaCI-0.01 M NaH2PO4-l mM disodium EDTA, pH 7.4), i% SDS, 0.5% nonfat powdered milk, and 0.5 mg/ml carrier salmon sperm DNA at 42Â°C.Filters were washed with 0.1 x SSC-0.1% SDS at 68Â°Cand exposed to Kodak XAR-5 films at -80Â°C with intensifying screens for 12-16 h. For in situ chromosomal hybridization. pCIA-5 was nick-translated with [3H]dTTP to a specific activity of 2.4 x IO7 cpm//ig. Normal human melaphase chromosomes from one male and two females were used for hybridization. The methods for in situ hybridization and autoradiogruphy were as described (33, 34). Hybridization was carried out at a concentration of 100 ng labeled probe/ml hybridization solu tion, and with a 1000-fold excess of carrier herring sperm DNA. Washes
stain in 0.6 M phosphate buffer (pH 6.8) for 5-15 min. The chromo somal positions of grains located on chromosomes or touching chro mosomes were plotted onto ideograms of G-banded human chromo somes at the 550-band stage (35). Cell Lines and Culture Conditions. The establishment of cell lines from human malignant gliomas has been described (36). The cell lines used in the present investigation originated from tumors that were diagnosed as glioblastoma multiforme. All cell lines were grown as monolayer cultures in Eagle's MEM supplemented with 10% newborn calf serum (GIBCO) and antibiotics (100 units of penicillin and 50 Â¿ig of streptomycin per ml). Cultures were maintained at 37Â°Cin humidi fied air containing 5% CO;. By using phase contrast microscopy we assigned the cell lines to either of five morphology groups (19): epithelioid; polygonal; pleomorphic/astrocytoid: fascicular; or bipolar/ fibroblastic. Northern Blot Analysis of Human Glioma Cell Lines. Cells were grown to confluency in routine medium on 100 mm plastic Petri dishes or 850-cnr Falcon roller bottles. Total cellular RNA was isolated using the LiCI/urea method. mRNA was enriched by poly(A)* selection on oligodeoxythymidylate cellulose (Pharmacia). Ten /jg of poly(A)* RNA were denatured for 5 min at 65Â°Cin 50% formamide. 2.2 M formalde hyde, 20 niM morpholinepropanesulfonic acid (pH 7.0), 5 mivi sodium acetate, l mM EDTA, and 10% glycerol; run on 0.8% agarose-formaldehyde gels: transferred to nitrocellulose filters: and baked at 80Â°Cfor 4 h. cDNA probes were ':P-labeled by random priming. Hybridization (lO^cpm/ml) was performed in 50% formamidc-5 x SSC-5 m\i EDTA0.5% SDS, 1 x Denhardt's. 100 f/g/ml salmon sperm DNA at 42Â°C for 12 h and washing in 0.5% SDS-0.1 x SSC at 50Â°Cfor 2 h. Kodak XAR-5 film was used for exposure at -70Â°C. The following human cDNA probes were used for Northern blot hybridization analyses: (a) the GFAP cDNA probe pCIA-5 as described in this communication: h) the PDGF /^-receptor cDNA probe pHPDGFR-2A3 (37): (c) the PDGF Â«-receptor cDNA probe pHPDGFR Al (38); (il) the fibronectin probe pHFNl consisting of a 380-base pair Pst\ insert in pBR322 and encoding the active cell attachment domain of fibronectin (39). Indirect Immunofluorescencc Staining. Glioma cells were suspended in Eagle's MEM-10% fetal calf serum, seeded sparsely on coverslips contained in 35-mm Petri dishes, and allowed to grow for 2-3 days. Before the first incubation, as well as between each subsequent incu bation, cells were washed 3 times in PBS. Incubations were made at room temperature in a moist chamber. Rabbit anti-fibronectin (1:50) (kindly provided by Dr. Antti Vaheri) was allowed to react for 30 min. and bound antibodies were visualized with rhodamine-conjugated goat anti-rabbit immunoglobulin (1:50. 30 min) (Dakopatts). Cells were fixed with 3% paraformaldehyde in PBS for 20 min at room tempera ture and with 100% acetone for 1 min at -20Â°C and then incubated with the monoclonal GFAP-3 antibody (1:100) (kindly provided by Dr. Peter Collins) for 30 min. The antibody was visualized by a fluorescein isothiocyanate-conjugated anti mouse immunoglobulin (1:20. 30 min) (Dakopatts).
Table 1 Regional localisation oj the GFAP %eneon chromosome 1? The SKBR3 cell line carries an amplified HKR-2/nrÂ« locus (59) indicated by ++++: the amplification unit involves several other 17-linked genes" bul not the GFAP gene (not shown). The region of chromosome 17 carried by hybrids cl 31 and 275S has been defined cytogenelicallv (60) and confirmed using DNA probes for chromosome 17-linked genes (28. 31. 60-62) as shown here and elsewhere (28. 30. 63). The region of chromosome 17 retained in hybrids N9 and cl9 has been defined using the DNA markers listed in here and by others (60. 63). The GFAP cognate sequences segregate with the Hox-2 gene in these hybrids: thus the GFAP cognate locus is distal to the NiiFR and proximal to the PKCâ€”locus. probesHer-2/Neu 17-linkcd DNA
lineSKBR Cell 3cl3l
' K. Unebner, unpublished results. 1554
17 region retained17plcrâ€” .17qter I7pterâ€”I7qier 17ql2^17q21I7q21-.l7qter
HUMAN GLIAI. I IBRIU.ARY l Met Glu Arg AroAra IleIhrSerAlaAla ArgArg SerTyrVal Ser l ATG GAG AGGAGA CGCATCACC TCCGCTGC!CGCCGC TCCTACGTC TCC
17 SerGlyGlu MetMet ValGlyGly LeuAla ProGlyArgArg LeuGly 49 TCAGGG GAG ATGATGGTGGGGGGCCTGGCTCCTGGCCGCCGT CTGGGT
33 Pro Gly ThrArg Leu SerLeuAlaArgMet ProProProLeu ProThr 97 CCT GGC ACCCGC CTC TCCCTGGCTCGAATGCCCCCTCCACTCCCG ACC head *â€”Ââ€” ¡ Â»rod 49 Arg valAsp Phe SerLeuAlaGlyAlaLeuAsnAlaGly PheLysGlu 145 CGGGTG GAT TTCTCCCTclcCTGGGGCACTCAATGCTGGC TTCAAGGAG
65 ThrArgAla SerGluArgAlaGluMetMet GluLeuAsnAspArg Phe 193 ACC CGGGCC AGTGAG CGGGCAGAGATGATGGAGCTCAATGAC CGCTT
81 Ala Ser Tyr IleGluLysValArgPheLeuGluGinGinAsn LysAla 241 GCC AGC TACATCGAGAAGGTTCGC TTCCTGGAACAGCAAAACAAG GCG
97 Leu AlaAla GluLeuAsn GinLeuArgAla LysGluProThr LysLeu 289 CTG GCT GCTGAG CTGAACCAGCTGCGGGCCAAGGAGCCCACCAAGCTG
113 AlaAsp ValTyrGin AlaGluLeuArgGlu LeuArgLeuArgLeuAsp 337 GCAGAC GTC TACCAGGCTGAGCTGCGAGAGCTGCGGCTGCGGCTCGAT
129 Gin Leu ThrAlaAsn SerAlaArg LeuGlu ValGluArgAsp AsnLeu 385 CAA CTCACC GCCAACAGCGCCCGGCTGGAGGTTGAGAGGGACAATTTG
145 Ala HisAsp LeuAlaThrLeuArgGin LysLeuGinAspGlu ThrLys 433 GCA CACGAC CTGGCCACTCTGAGGCAGAAGCTCCAGGATGAAACCAAG
161 LeuAro LeuGluAla AspAsnAsn LeuAlaAla TyrArgGinGlu Ala 481 CTG AGG CTGGAAGCAGACAACAACCTGGCTGCCTATAGACAGGAGGCA
177 Asp GluAla ThrLeuAlaArgLeuAsp LeuGluArgLys IleGlu Ser 529 GAT GAA GCCACCCTGGCCCGTCTGGATCTGGAGAGGAAGATT GAG TCG 193 LeuGluGluGlu IleArgPheLeuArgLys IleHisGluGluGlu Val 577 CTGGAG GAGGAGATCCGG TTCTTGAGGAAGATCCACGAGGAG GAGGTT Glu LeuGin GluGinLeuAlaArgGinGinValHisValGlu Leu 209 Arg 625 CGGGAA CTCCAGGAGCAGCTGGCCCGACAGCAGGTCCAT GTGGAGCTT
ValAla LysProAspLeuThrAlaAlaLeuLysGlu IleArg Thr 225 Asp 673 GACGTG GCCAAGCCAGACCTCACCGCAGCCCTGAAAGAG ATCCGCACG 241 Gin TyrGluAlaMetAla SerSerAsnMet HisGluAlaGluGlu Trp 721 CAG TATGAGGCA ATGGCGTCCAGCAACATGCATGAAGCCGAAGAA TGG
257 TvrHis SerLysPheAlaAspLeuThrAspAlaAlaAlaArgAsnAla 769 TACCAC TCCAAG TTTGCAGACCTGACÃ•GACGCTGCTGCCCGCAACGCA 273 Glu LeuLeuArgGinAla LysHisGluAlaAsnAsp TyrArgArgGin 817 GAG CTGCTCCGC CAGGCCAAGCACGAAGCCAACGAC TACCGACGCCAG
Leu Gin SerLeuThrCysAspLeuGlu SerLeuArgGlyThrAsnGlu TTG CAG TCCTTGACC TGCGACCTGGAGTCTCTGCGCGGC ACGAACGAG
SerLeuGlu ArgGinMet ArgGluGinGluGluArgHis ValArg Glu TCCCTGGAG AGGCAGATGCGCGAACAGGAGGAGCGGCACGTG CGGGAG
321 AlaAla SerTyrGinGluAlaLeuAlaArg LeuGluGluAsp GlyGin 961 GCGGCCAGT TATCAGGAGGCACTGGCTCGGCTGGAGGAAGAT GGCCAA
337 SerLeuLysAspGluMetAlaArgHisLeuGinGlu TyrGin AspLeu 1009 AGCCTC AAGGACGAGATGGCCCGCCACTTGCAGGAG TACCAG GACCTG
353 Leu AsnVal LysLeuAla LeuAsp IleGlu IleAla ThrTyrArgLys 1057 CTC AAT GTCAAG CTGGCCCTGGACATCGAGATCGCCACC TACAGGAAG rod Â«â€”.â€”Â» tail 369 Leu LeuGluGly GluGlu AsnArg IleThr IleProValGinThrPhe 1105 CTG CTA GAGGGCGAG GAGlAACCGGATCACCATTCCCGTG CAGACCTTC
385 SerAsn LeuGin IleArgGlu ThrSerLeuAspThrLys SerVal Ser 1153 TCCAAC CTGCAGATT CGAGAAACCAGCCTGGACACC AAG TCTGTG TCA
401 GluGly HisLeu LysArgAsn IleValValLysThrValGluMetArg 1201 GAA GGCCACCTCAAG AGGAACATCGTGGTGAAGACCGTGGAGATG CGG
417 Asp GlyGlu Val IleLysGlu SerLysGinGluHis LysAsp ValMet 1249 GAT GGAGAG GTCATTAAGGAG TCCAAGCAGGAGCACAAGGAT GTGATG
433 â€” 1297 TGAGGC AGGACCCACCTGGTGGCCTCTGCCCCG TCTCATGAGGGGCCC
1345 GAG CAGAAG CAGGATAGTTCG TCCGCCTCTGCTGGC ACA TTTCCCCAG
289 865 305 913
ACID PROTEIN GLATORRLGPGT]RTSLARMPPPLP(TJRVDFSLAGALNÂ¿h
GFKETRASERAEMMELNDRFASYIEK GFKETRASERAEMMELNDRFASYIEKVRFLEQQNKALAAELNQLRAKEPTKLADVYQAEL Ã•RELI RELRLRLDQLTANSARLEVERON
EftNLAAYRQEADEAT RELRLRLDO.LTANSARLEVERO^ RELt rLROKLQDEff_RLEADNNIAAYRQÃ•ADEAT
GTNESLEROMREQEERHyEAASYQEALARLEEgGQSLKDEMARHLQEYQDLLNVKlALD GTNESLERQHREQEERHAREAASYOEAl[t)iLEEEGQSLKDEMARHL(JEYCgLLNVKLALD GTHESLERQHREQEERHARElUASYQEALARlEEEGQSLKlElEHARHLQEYQOLLNVKLALD
IKESKQEHKD-VM-J h IKESKQEHKD-\JE
Fig. 2. Comparison of the amino acid sequences (single-letter code) of the human (A), murine (m). and porcine (p) GFAP. The human sequence was deduced from the GFAP cDNA clone E2C2 described in this work. The murine sequence was deduced from genomic DNA as described (41 ). The amino acid sequences of two stretches, including the NH; terminus, of porcine GFAP were obtained from Ref. 40.
1393 ACC TGAGCTCCC CACCG
Fig. 1. Nuclcotidc sequence of the human GFAP cDNA clone E2C2 and the deduced amino acid sequence. The position of the (Â»-helicalrod domain was determined according to procedures described in Ref. 64.
After being washed with PBS, coverslips were mounted in glyeerolparaphenylcnediamine (Sigma) and were viewed in a Lett/, epifluoreseence microscope with oil immersion objectives. The relative intensity of staining on each cell line was estimated (- to +++++).
RESULTS Isolation of Human GFAP cDNA. A human glioma cell line (U-343 MGa Cl 2:6) cDNA library was screened with the mouse GFAP cDNA probe Gl. Several positive clones were found. One clone, CIA-5, containing a 1.1-kilobase insert was chosen for further characterization. Since this cDNA apparently was
truncated in the 5' end, we rescreened the cDNA library with a 5' 120-base pair Sstl-Accl fragment from pCIA-5. The clone E2C2 was isolated and found to have a 1.4-kilobase insert, the nucleotide sequence of which is shown in Fig. 1. The insert is 1409 base pairs long. The putative translation-initiating codon at the 5' end specifies a 1296-base pair open reading frame which is flanked by a 110-base pair 3'-untranslated sequence. The NH:-terminal sequence of porcine GFAP has been deter mined previously by amino acid sequence analysis of purified protein (40). As can be seen in Fig. 2, the deduced sequence of the human GFAP amino terminus is virtually identical to that of porcine GFAP. Thus, the assumed open reading frame of the human GFAP cDNA shown in Fig. 1 is likely to be authentic. An alignment with the deduced amino acid sequence of mouse GFAP (41) shows an almost perfect match from residue 29 of the human sequence (Fig. 2). The NH;>-terminal of this residue, the mouse sequence (41), diverges entirely from the porcine and human GFAP sequences. Inspection of the nucleotide sequence of the presumed 5'-untranslated segment of the mouse gene (40), however, shows the presence of an out of frame ATG which corresponds to the initiating methionine in the human sequence and an 80% identity of the following 45 downstream nucleotides. All intermediate filament proteins share the same basic struc ture with a central Â«-helicalrod domain of some 310 amino acids that are flanked by NH2-terminal and COOH-terminal head and tail segments which differ in length (1-4). The loca tion of these domain structures in GFAP is shown in Fig. 1. The authenticity of the human GFAP cDNA was further examined by subeloning into the expression vector RD2 and
HUMAN GLIAL F1BRILLARY ACID PROTEIN
123456789 kbp -20.0 -
Fig. 3. Immunofluorescencc staining of recombinant GFAP expressed in microinjected lJ-343 MG cells. A fibrillary network is decorated by the GFAP antibodies only in areas containing cells microinjected with the GFAP expression vector.
introduction by microinjection into U-343 MG cells, which lack endogenous GFAP expression (see below). Indirect immunofluorescence staining with GFAP antibodies revealed a typical network of GFAP fibers in a fraction of the microin jected cells (Fig. 3); cells in adjacent areas served as controls and were devoid of immunostainable fibers. Chromosomal I.ocali/ation of the Human GFAP Gene. A panel of cellular DNAs derived from 20 well-characterized rodent-human cell hybrids retaining overlapping subsets of human chromosome regions (see Fig. 4) was tested for retention of the GFAP gene by Southern blot analysis. An example of such a Southern blot, after Hind\\\ digestion of control and hybrid DNAs, is shown in Fig. 5. Lane 1 contains mouse DNA and shows only one obvious band (approximately 7.2 kbp). Hind\\\ cleaved human DNA (Fig. 5, Lane 2) shows one major Human Chromosomes
G5 Be 3a cl31 GL5 S3 77-31 DSK20 PBSa C4 77-28 AB3 CI21 2S5 N9 EF3 c/2 275S 77-3012345678910n|i2[i3141516|17|18I19â€¢â€¢1â€¢Jâ€¢â€¢202122X
Fig. 4. Presence of the human GFAP locus in a panel of 20 rodent-human cell hybrids. â€¢hybrid named in the left column contains the chromosome indicated in the upper row: B. presence of the long arm (or part of the long arm. indicated by a smaller fraction of stippling) of the chromosome shown above the column: B. presence of the short arm (or partial short arm) of the chromosome listed above the column. D. absence of the chromosome shown above the column. The column for chromosome 17 is boldly outlined and stippled to highlight correlation of the presence of this chromosome (or region of this chromosome) with the presence of the GFAP gene. The pattern of retention of the GFAP gene in the panel is shown in the column to the right of the figure where the presence of the gene in the hybrid is indicated by a stippled box Â»itha plus sign and absence of the gene is indicated by an open bo.\ enclosing a minus sign.
Fig. 5. Southern blot analysis of ///m/Ill-digested DNA from mouse (Lane I). human T-cell line (Lane 2), hybrid cl9 (Lane 3). hybrid GL5 (Lane 4). hybrid N9 (Lane 5). hybrid 275s (LaÃ±e6). hybrid c4 (Lane 7), hybrid G5 (Lane 8), and hybrid BD3 (Lane V) hybridized to the human GFAP probe. The 20-kbp fragment present in DNA from Lanes 2 to 6 is diagnostic for (he human GFAP gene. Hybrid DNAs in Lanes 3-6 contain whole or parts of human chromosome 17. while hybrids in Lanes 7-9 do not retain any chromosome 17 material (cf. Fig. 4 and Table I ).
GFAP specific fragment of 20 kbp and numerous (at least 10 are clear on longer exposure) more weakly hybridizing bands of various lower molecular weights; these bands most likely represent cross-hybridizing members of a gene family, while the strongly hybridizing band at 20 kbp is the cognate gene. Lanes 3-6 (Fig. 5) contain DNA from hybrids which contain the GFAP cognate gene at 20 kbp. while hybrids in Lanes 7-9 do not retain the cognate gene. The cross-hybridizing bands seen in human DNA do not appear in any of the hybrid DNAs in Lanes 3-9 (or in hybrid cl31, which retains an entire chro mosome 17; not shown) and thus do not segregate with the cognate gene. Since the hybrid in Lane 3 (Fig. 5) retains human chromosome 17 as its sole human chromosome and other positive hybrids (Fig. 5, Lanes 4-6) also retain chromosome 17 or a region of 17 while hybrids in Lanes 7-9 do not retain chromosome 17, the GFAP cognate gene must be linked to chromosome 17 (cf. Fig. 4). Similar analysis of the entire chromosome panel confirmed linkage of GFAP to chromosome 17 (cf. Fig. 4). Hybrid and control DNAs were similarly ana lyzed after cleavage with restriction enzyme Sst\ with similar results. G FA P-specific Ssfl fragments of 2.8, 1.2, and<0.5 kbp, representing the cognate GFAP gene, segregated concordantly (and with a specific region of human chromosome 17) in the hybrid panel, while at least six cross-hybridizing Sstl fragments were not linked to chromosome 17. Analysis of the somatic cell hybrids retaining specific regions of chromosome 17 (see legend to Fig. 4) was also the basis for the regional localization of the cognate GFAP gene on chro mosome 17. as summarized in Table 1. Since the cognate locus segregates with the Hox-2 locus in these hybrids (see Table 1), the GFAP locus is distal to the NGFR locus and proximal to the PKCâ€”locus and located at 17q21. The regional localization of the human GFAP locus was also determined by in situ hybridization. A total of 105 metaphase cells were scored of which 35 (33'V ) had grains on chromosome
17. Specific labeling was observed over the midregion of the long arm with a distinct peak at band 17q21. Forty-six% of 1556
HUMAN (Â¡LIAI-FIBRIU.ARY ACID PROTEIN
total grains on chromosome 17 clustered at this band. A rep resentation of the grain distribution on chromosome 17 is shown in Fig. 6. In conclusion, the in situ hybridization data assign the GFAP locus to band 17q21 and thus confirm the data obtained by Southern blot analysis of cell hybrids. Expression of Fibronectin and GFAP in Human Glioma Cell Lines. The levels of fibronectin and GFAP mRNA were studied by Northern blot analysis of poly(A)+ RNA isolated from ten established human malignant glioma cell lines (Fig. 7; Table 2). This panel of cell lines includes three sublines derived from the U-343 glioma (U-343 MG. U-343 MGa Cl 2:6, U-343 MGa 31L) (18, 42). The fibronectin probe hybridized to a 6-kilobase transcript whereas the GFAP cDNA revealed a 2.7-kilobase mRNA. Fibronectin mRNA was found in four of the cell lines (U-251 MG sp, U-1242 MG, U-343 MG, U-178 MG). GFAP mRNA was found in five of the ten lines analyzed (U-343 MGa 31L, U-343 MGa Cl 2:6, U-251 MG sp, U-373 MG, U-1231 MG). (The 2.7-kilobase band seen on the blot in U-343 MGa 31L hybridized with fibronectin cDNA represents a signal remaining from the previous hybridization with GFAP cDNA). The pattern of mRNA expression of fibronectin and GFAP in the different cell lines was found to conform with the expression of protein, as estimated by indirect immunofluorescence stain ing (Table 2). PDGF a- and ÃŸ-ReceptormRNA in Human Glioma Cells: Relation to Cell Morphology and Levels of Fibronectin and GFAP mRNA. Northern blots of poly(A)+ RNA from the ten glioma cell lines were hybridized with PDGF a- and /i-receptor cDNA probes (Fig. 7). The strongest signal of the 6.5-kilobase PDGF Â«-receptor transcript was found in U-343 MGa 31L; detectable levels were also found in U-373 MG, U-178 MG, U343 MG, and U-1242 MG after longer exposure times. Abun dant levels of the 5.4-kilobase PDGF /i-receptor mRNA were found in U-178 MG, U-343 MG, U-1242 MG, and U-1231 MG; a weak signal was found in U-1240 MG. The weak signal seen with the PDGF ^-receptor probe on Northern blot of U343 MGa 31L cells represents cross-hybridizing to Â«-receptor mRNA (cf. RÃ©f. 38). The data on mRNA expression of fibronectin, GFAP, and PDGF a- and /i-receptor were compiled in Table 2 and related to cell morphology as described (19). A striking feature is the essentially inverse relation between the expression of GFAP and fibronectin genes, as has previously been found with regard to the corresponding proteins (9, 10). One exception was found (U-251 MG sp), in which a fibronectin mRNA signal of medium strength was found together with GFAP mRNA and protein expression. Table 2 also makes it clear that glioma cell lines expressing both types of PDGF receptor mRNA. have a fibroblastic phenotype. i.e., bipolar, fibroblast-like morphology and
Fig. 7. Northern blot analysis of mRNA from human glioma cells. Ten Mgof poly(A)* RNA per lane were electrophoresed in agarose gels and blotted on to nitrocellulose filters which were sequentially hybridized to human cDNA probes for GFAP. fibronectin. and PDGF <>-(PDGFR A) and Â¿-receptors(PDGFR B). Lane.-I. AG 1523: Lane B. U-343 MGa 31L: Lane C. U-251 MG sp: Lane D, U1240 MG: Lane E. U-1242 MG; Lane F. U-343 MG: Lane G. U-178 MG: Lane H. U-563 MG; Lane I. U-373 MG: Lane J. U-1231 MG; Lane K, U-343 MGa Cl 2:6.
high levels of fibronectin mRNA. Relatively high levels of GFAP Â«-receptor mRNA, in the absence of/i-receptor mRNA expression, were found in two cell lines (U-343 MGa 3IL, U373 MG); these belong to the group of cell lines that express GFAP and lack detectable levels of fibronectin mRNA. DISCUSSION
Fig. 6. Representation of the grain distribution on chromosome 17 assigning the GFAP locus to region I7q2l by in situ hybridization.
We have cloned human GFAP cDNA and deduced the pri mary structure of the protein. The translation product appears to be a 432-residue protein with 88 and 93% sequence similar ity, respectively, with the two stretches of porcine GFAP that
HUMAN OLIAI. HHRIt.LARY
Table 2 Phenol) pic characteristics of human glioma cell lines A panel of human glioma cell lines of different morphologies were used. Kor original references and assignment to morphological groups see Refs. IX. 19. and 42. The relative amounts of GFAP, fibronectin, and PDGF <>-and iJ-receptor mRNAs were estimated (- to +++++) from sequentially hybridized Northern blots. When the level of expression was extremely low and the cell line difficult to group as positive or negative it is noted as -+. Relative amounts of GFAP and fibronectin (to +++++) Â»ereestimated from immunofluorescence stainings. Several of the GFAP-positivc cultures contained a minor component of GFAP-negative cells (+/-). PDGFmRNA lineU-1231 Cell MGU-343 12:6U-343 MGa C LU-373MGa 31 MGU-251 MGspU-I240MGU-563
mRNA+++++ Immunofluorescence +++/-++++ â€”+++++ ++++ ++++ +++ 11 i *'/iii
MGU-178 -+ MGU-343 ++++++ + +++ MGU-I242MGMorphologyFpilhelioidPolygonalPleo/Astro"Pleo/AstroFascicularFascicularPolygonalBip/FibrBip/FibrBip/FibrGFAP ++++++ ++++ +++++
Bip/Fibr. bipolar, fibroblastic: n.d.. not determined.
were previously subjected to amino acid sequence analysis (40). The authenticity of the cDNA was confirmed by its expression in microinjected cells. An immunostainable, fibrillary network was apparent in transfected U-343 MG cells that lack endoge nous GFAP expression. During the progress of the present work. Reeves et al. (43) reported on the cloning of human GFAP cDNA: our present data confirm their coding sequence. An alignment with the assumed coding sequence of the mouse gene showed that the predicted initiating ATG (Met) in the mouse sequence corresponds to ACG at positions 41-43 in the human nucleotide sequence (Fig. 1). Inspection of the assumed 5'-untranslated sequence of the mouse gene (41) shows the presence of a short open reading frame, in which the first 13 codons are almost identical to the 5' end of the human coding sequence. Since the NHi terminus of mouse GFAP has not been established by amino acid sequence analysis, we do not know at present if the assumed initiating ATG of the mouse gene is the authentic one. If this is indeed the case, the conser vation of a sequence, that is untranslated in the mouse and most probably translated in the human, is remarkable. Southern blot analysis of human/mouse hybrid cell lines and in situ hybridization were used to assign the human GFAP gene to chromosome 17, band q21. Interestingly, there are several other genes of interest in relation to GFAP that are located on the proximal half of chromosome 17. These include the erbBI (rat neuro/glioblastoma-derived oncogene homologue) gene, the NGFR (nerve growth factor receptor) gene, and the NF1 (von Recklinghausen neurofibromatosis) gene (44). Several members of the cytokeratin gene family have also been mapped to this region of chromosome 17 (44). Recent mapping studies, assigning the mouse GFAP gene to chromosome 11 (45), indicate that GFAP is part of a large syntenic gene cluster including at least 22 genes located on human chromosome 17 and mouse chromosome 11 (44, 45). Recent studies have shown that there exist two types of PDGF receptors that differ in ligand specificity. The PDGF Â«receptor binds all three isoforms of PDGF (PDGF-AA, -AB, -BB) with high affinity, whereas the /i-receptor binds only PDGF-BB with high affinity (46, 47). The two receptors are structurally homologous (20. 37,48), and both have been show n to transduce mitogenic signals in response to the proper ligand (49, 50). Available information suggests that the Â«-and flreceptors are differentially expressed in normal cells. Thus, the O-2A glial progenitor cell of the rat opticus nerve expresses Â«-receptors only (51), whereas human meningeal cells (52) and rat brain capillary endothelial cells (53) express only /i-recep-
tors. Various kinds of fibroblasts in culture, however, express both types of receptors. The present investigation showed a wide range of Â«-and tfreceptor mRNA levels in human glioma cells, as determined by Northern blot hybridi/.ation analysis. A comparison of PDGF receptor expression with cell morphology and fibronectin/ GFAP phenotype revealed an interesting pattern. One subgroup was formed by cells with a fibroblastic phenotype: such cells displayed the same pattern of PDGF receptor expression as bona fide fibroblasts; i.e., they expressed both types of receptor mRNA. One might argue that these cells are actually derived from a separate mesenchymal component, present in the orig inal tumor. The high frequency of establishment of glioma cell lines with such phenotype argues against this notion. Another argument is the finding of the same unique marker chromosome in the "fibroblastic" cell line U-343 MG, having the GFAP~/ FN* phenotype, and the "glial" line U-343 MGa and clones thereof, which display the GFAP+/FN phenotype. These cell lines were derived from different primary cultures of the same glioma biopsy. Since all derivatives have an inversion in chro mosome 1 [inv(l)pl3q43] (42. 54) they most probably have the same clonal origin. Unfortunately, this finding does not abso lutely prove a monoclonal origin, since we lack data on the karyotype of the host tissue. It is also interesting that other studies have indicated that certain normal glial cells produce fibronectin (55, 56). The possible relationship between these cells and the fibronectin-positive glial cells remains to be elucidated. A relationship between the pattern of PDGF receptor expres sion and glioma cell phenotype was further evidenced by the finding that cell lines expressing GFAP had no detectable, or very low, levels of PDGF /3-receptor mRNA. In this group we found two cell lines (U-343 MGa 31L and U-373 MG) that expressed relatively high levels of PDGF Â«-receptor mRNA. Expression of Â«-receptors in U-343 MGa 31L has been de scribed previously; these cells were used for initial characteriza tion (20) and cDNA cloning (38) of the Â«-receptor. Our study highlights the phenotypic heterogeneity of glioma cell populations, with regard to cells both within a certain population and between different populations. Cellular hetero geneity is not a specific trait of gliomas and is considered an important character of malignant tumor cell populations in general inasmuch as it may form the basis for tumor progression (see Ref. 57 for review). According to Nowell's original hypoth esis (58), phenotypic heterogeneity may be the result of the genetic instability of tumor cells allowing for the generation of a plethora of genotypic and phenotypic variants. Another, not
Hl'MAN GUAI. FIBRILLARE ACID PROTKIN
mutually exclusive hypothesis infers that neoplastic stem cells give rise to variants through more or less normal differentiation processes that do not involve any structural changes in the genome. Our findings that PDGF receptor expression is related to cell morphology and synthesis of GFAP and fihronectin are compatible with the second hypothesis for explaining the dif ferential PDGF receptor expression in glioma cells. Previous studies on the expression of mRNA encoding the subunit chains of PDGF have shown A chain transcripts in most cell lines whereas B chain mRNA was present mainly in clones or cell lines composed of small cells or cells with an epithclioid or pleomorphic/astrocytic morphology (18, 19). As is also shown in Table 1, only such cells were found to express GFAP. These findings, in conjunction with the present report, are compatible with the idea that the expression of the genes encoding PDGF and its receptors in glioma cells is regulated in concert with other gene families which may or may not be developmental!)' regulated in normal glial cell lineages. ACKNOWLEDGMENTS Skillful technical help was provided by Annika Hermansson and Marianne Kastemar.
REFERENCES 1. Lazarides. F.. Intermediate filaments: a chemically heterogeneous, developmentally regulated elass of proteins. Annu. Re\. Biochem.. 51: 219-250, 1982. 2. Osborn. M.. and \\ eher. K. Intermediate filament proteins: multigene family distinguishing major cell lineages. Trends Biochem. Sci., //:469â€”Mi. 1986. 3. Steinert. P. M., and Roop. D. R. Molecular and cellular biology of interme diate filaments. Annu. Rev. Biochem.. 57: 593-625. 1988. 4. Steinert. P. M., Steven, A. C., and Roop. D. R. The molecular biology of intermediate filaments. Cell, 42:411-419, 1985. 5. Eng, L. F. CÃ¬lial fibrillary acidic protein (GFAP): the major protein of glial intermediate filaments in differentiated astrocytes. J. Neuroimmunol.. 8: 203-214. 1985. 6. Eng. L. F.. and Rubinstein. I,. .1. Contribution of immunohistochemistry to diagnostic problems of human cerebral tumors. J. Histochem. Cytochem.. 26:513-521. 1978. 7. Bigner. D. D.. Bigner. S. H.. PonlÃ©n.J.. Mahley. M. S.. Ruoslahti. E.. Hershman. H.. Eng. L. F.. and \\ikstrand. C. J. Heterogeneity of genotypic and phenotypic characteristics of fifteen permanent cell lines derived from human gliomas. J. Neuropathol. E\p. Neurol.. 40: 201-229, 1981. 8. Yamada. K. Fibronectin domains and receptors. In: D. F. Mosher (ed.). Fibronectin. pp. 47-121. New York: Academic Press. 1989. 9. Paetau. A., Mellstrom, K.. Westermark. B.. Dahl. D.. Haltia. M.. and Vaheri. A. Mutually exclusive expression of fibronectin and glial fibrillary acidic protein in cultured brain cells. Exp. Cell Res., 129: 337-344, 1980. 10. Vvestermark. B.. Magnusson, A., and Heldin. C-H. Effect of epidermal growth factor on membrane motility and cell locomotion in cultures of human clonal glioma cells. J. Neurosci. Res.. 8: 491-507, 1982. 11. Noble, M.. Murray. K.. Stroobant, P.. \\aterfield. M. D.. and Riddle. P. Platelet-derived growth factor promotes division and motility and inhibits premature differentiation of the oligodendrocyte/type-2 astrocyte progenitor cell. Nature (Lond.)..?.?.?: 560-562. 1988. 12. Richardson. W. D., Pringle. N., Mosley. M. J.. Westermark. B.. and DuboisDalcq. M. A role for platelet-derived growth factor in normal gliogenesis in the central nervous system. Cell. 53: 309-319. 1988. 13. Devare. S. G.. Reddy, E. P., Law, J. D.. Robbins. K. C., and Aaronson. S. A. Nucleolide sequence of the simian sarcoma virus genome: demonstration that its acquired cellular gene sequences encode the transforming gene product p28"'. Proc. Nati. Acad. Sci. USA. Aft-731-735. 1983. 14. Doolittle. R. F.. Hunkapiller. M. W.. Hood. L. E.. Dcvare. S. G.. Robbins. K. C.. Aaronson. S. A., and Antoniades. H. N. Simian sarcoma virus oncogene, v-s/s, is derived from the gene (or genes) encoding a plateletderived growth factor. Science (Washington DC). 221: 275-277, 1983. 15. VVaterfield. M. D.. Scarce. G. T.. Whittle. N.. Stroobant. P.. Johnsson, A.. Wastcson. A.. Westermark. B.. Heldin. C.-H.. Huang. J. S.. and Deuel, T. F. Platelet-derived growth factor is structurally related to the putative trans forming protein p28"* of simian sarcoma virus. Nature (Lond.). 304: 35-39, 1983. 16. Deinhardt, F. Biology of primate retroviruses. In: G. Klein (ed.). Viral Oncology, pp. 357-398. New York: Raven Press. 1980. 17. Nister. M.. Heldin. C-H.. Wasteson. A., and Westermark. B. A gliomaderived analog to platelet-derived growth factor: demonstration of receptor
competing activity and immunological crossreactivity. Proc. Nati. Acad. Sci. USA. 81: 926-930. 1984. Nistcr. M.. Heldin. C.-H., and Westermark. B. Clonal variation in the production of platelet-derived growth factor-like protein and expression of corresponding receptors in a human malignant glioma. Cancer Res.. 46:332340. 1986. Nistcr. M.. l.ibermann. T. A.. Betsholtz. C.. Pettersson. M.. Claesson-Welsh, L.. Heldin. C.-H.. Schlessinger. J.. and Westermark. B. Expression of mes senger RNAs for platelet-derived growth factor and transforming growth factor and their receptors in human malignant glioma cell lines. Cancer Res., â€¢/Â«.â€¢39IO-39I8, 1988. Claesson-Welsh. I... Hammacher. A.. Westermark. B.. Heidin, C-H., and NistÃ©r.M. Identification and structural analysis of the A type receptor for PDGF: similarities with the B type receptor. J. Biol. Chem.. 264: 17421747. 1989. Pringle. N.. Collarini. E. J., Mosley, M. J.. Heldin. C-H., Westermark, B., and Richardson. \V. D. PDGF A chain homodimers drive proliferation of bipotential (O-2A) glial progenitor cells in the developing rat optic nerve. EMBO J.. A: 1049-1056. 1989. Maniatis. T.. Fritseh. E. F.. and Sambrook. J. Molecular Cloning. A Labo ratory Manual. C'old Spring Harbor. NY: Cold Spring Harbor Laboratory.
1982! 23. Lewis. S. A.. Balcarek. J. M.. Krek. V.. Shelanski. M.. and Cowan. N. J. Sequence of a cDNA clone encoding mouse glial fibrillary acidic protein: structural conservation of intermediate filaments. Proc. Nati. Acad. Sci. USA. SI: 2743-2746. 1984. 24. Betsholtz. C.. Johnsson. A.. Heldin. C-H.. Westermark. B., Lind. P.. Urdea, M. S.. Eddy. R.. Shows, T. B., Philpott. K., Mellor. A. L.. Knott. T. J., and Scott. J. The human platelet-derived growth factor A-ehain: complementary DNA sequence, chromosomal localization and expression in tumour cell lines. Nature (Lond.). 320: 695-699. 1989. 25. Huynh. T. V., Young. R. A., and Davies, R. W. Constructing and screening cDNA libraries in gtlO and gtll. In: Glower (ed.). Cloning Techniques: A Practical Approach. Oxford. United Kingdom: IRL. pp. 49-78. 1986. Sanger. F".. M Â¡eklen.S., and Coulson. A. R. DNA sequencing with chain26. terminating inhibitors. Proc. Nati. Acad. Sci. USA, 74: 5463-5467. 1977. 27. Messing. J.. Grea, R., and Seeburg. P. H. A system for shotgun DNA sequencing. Nucleic Acids Res.. 9: 309-321, 1981. 28. Cannizzaro. L. A.. Croce. C. M., Griffen. C. A.. Simeonc. A.. Bonicelli. E., and Huebner. K. Human homeobox-containing genes located at chromosome regions 2q31â€”2g37 and 12ql2â€”>12ql3. Am. J. Hum. Genet., 41: 1-15. 1987. 29. Dalla Pavera, R.. (Â¡allo.R. C.. Giallongo. A., and Croce. C. M. Chromosomal localization of the human homolog (c-sis) of the simian sarcoma virus onegene. Science (Washington DC), 218: 686-688, 1982. 30. Dayton. A.. Seiden. J. R., Laws, G.. Dorney. D. J., Finan. J., Tripputi. P., Emanuel. B. S.. Rovera. G., Nowell, P. C., and Croce, C. M. A human cerhA oncogene homolog is closely proximal to the chromosome 17 breakpoint in acute promyeloeytic leukemia. Proc. Nati. Acad. Sci. USA. 81: 44954499. 1984 31. Isobe. M., Huebner. K.. Maddon. P. J.. Littman. D. R.. Axel. R.. and Croce. C. M. The gene encoding the T-cell surface protein T4 is located on human chromosome 12. Proc. Nati. Acad. Sci. USA. Â«J:4399-4402. 1986 32. Reed. K. C., and Mann. D. A. Rapid transfer of DNA from agarose gels to nylon membranes. Nucleic Acids Res.. 13: 7201-7221. 1985. 33. Harper. M. F... and Saunders. G. F. Localization of single copy DNA sequences on G-banded human chromosomes by in silu hybridization. Chromosoma (Beri.), 83: 431-439. 1981. 34. Stenman. G.. and Sager. R. Genetic analysis of tumorigenesis: a conserved region in the human and Chinese hamster genomes contains genetically identified tumor-suppressor genes. Proc. Nati. Acad. Sci. USA. 84: 90999102. 1987 35. Harnden. D. G., and Klinger. H. P. (eds.). ISCN. An International System for Human Cytogenetic Nomenclature. Published in collaboration with Cytogcnetics and Cellular Genetics, p. 117. Basel: S. Karger AG. 1985. 36. Westermark. B.. PontÃ©n.J.. and Hugosson. R. Determinants for the estab lishment of permanent tissue culture lines from human malignant gliomas. Acta Palhol. Microbiol. Scand. Sect. A, 81: 791-805. 1973. Claesson-Welsh, L.. Eriksson. A.. Moren, A., Severinsson, L., Ek. B.. Ã–st37. man. A.. Betsholtz, C.. and Heldin. C-H. cDNA cloning and expression of a human platelet-derived growth factor (PDGF) receptor specific for B-chain containing PDGF molecules. Mol. Cell. Biol.. 8: 3476-3486, 1988. 38. Claesson-Welsh, L.. Eriksson, A.. Westermark. B., and Heldin. C-H. cDNA cloning and expression of the human A-type platelet-derived growth factor (PDGF) establishes structural similarity to the B-type PDGF recep tor. Proc.receptor .Nati. Acad. Sci. USA, 86: 4917-4921". 1989 39. Oldberg. A.. Linney. E.. and Rouslahti. E. Molecular cloning and nucleotide sequence of a cDNA clone coding for the cell attachment domain in human fibronectin. J. Biol. Chem., 25Â«:10193-10196. 1983 40. Geislcr. N.. and Weber. K. Amino acid sequence data on glial fibrillary acidic prolein (GFA): implications for the subdivision of intermediate filaments into epithelial and non-epithelial members. EMBO J.. 2: 2059-2063. 1983. 41. Balcarek. J. M., and Cowan. N. J. Structure of the mouse glial fibrillary acidic protein: implications for the evolution of the intermediate filament multigene family. Nucleic Acids Res.. 13: 5527-5543, 1985. 42. NistÃ©r.M.. Wedeil. B.. Betsholtz. C., Bywater. M.. Pettersson, M., Wester mark. B.. and Mark. J. Evidence for progressional changes in the human
HUMAN GLIAL FIBRILLARE ACID PROTFIN
malignant glioma line U-343 MGa: analysis of karyotypc and expression of genes encoding the subunit ehains of platelet-derived growth faetor. Cancer Res., 47: 4953-4960, 1987. Reeves, S. A., Helman. L. J.. Allison, A., and Israel. M. Molecular cloning and primary structure of human glial fibrillary acidic protein. Proc. Nati. Acad. Sci. LJSA. 86: 5178-5182. 1989. Solomon, E.. and Barker. D. F. Report of the Committee on the Genetic Constitution of Chromosome 17. Human Gene Mapping 10: Tenth Inter national Workshop on Human Gene Mapping. Cytogenet. Cell Genet.. 51: 319-337, 1989. Lalley. P. A.. Davisson, M. T.. Graves. J. A. M.. O'Brien. S. J., Womack. J. E.. Roderick. T. H.. Cream-Goldberg. N.. Hillyard. A. L.. Dooliltle. D. P., and Rogers. J. A. Report of the Committee on Comparative Mapping. Human Gene Mapping 10: Tenth International Workshop on Human Gene Mapping. Cytogenet. Cell Genet.. 5/: 503-532. 1989. Hart. C. E., Forstrom. J. W.. Kelly, J. D., Seifert. R. A.. Smith. R. A.. Ross. R.. Murray. M.. and Bowen-Pope. D. F. Two classes of PDGF receptors recognize different isoformsof PDGF. Science (Washington DC). 240:15291531. 1988. Heldin. C-H., BackstrÃ¶m.G., Ã–stman. A.. Hammacher. A.. Ronnstrand. L.. Rubin. K.. NistÃ©r,M., and \Vestermark. B. Binding of different dimeric forms of PDGF to human fibroblasts: evidence for two separate receptor types. EMBOJ.. 7: 1387-1393. 1988. Matsu. T.. Heidaran. M.. Toru. M., Popescu. N.. La Rochelle. W.. Kraus. M.. Pierce. J.. and Aaronson. S. A. Isolation of a novel receptor cDNA establishes the existence of two PDGF receptor types. Science (Washington DC), 243: 800-803. 1989. Hammacher, A.. NistÃ©r,M., and Heldin, C-H. The A-type receptor for platelet-derived growth factor mediates protein tyrosinc phosphorylation. receptor transmodulation and a mitogenic response. Biochem. J.. 265: 1520, 1989. Kazlauskas, A., Bowen-Pope. D.. Seifert. R., Hart, C. E.. and Cooper, J. A. Different effects of homo- and heterodimers of platelet-derived growth factor A- and B-chains on human and mouse fibroblasts. EMBO J.. 7: 3727-3735, 1988. Hart. I. K.. Richardson. W. D.. Heldin. C-H.. Westermark. B., and Raff. M. PDGF receptors on cells of the oligodendrocyte-type-2 astrocyte (O-2A) cell lineage. Development. 105: 595-603. 1989. Wang. J-L.. NistÃ©r,M.. Hermansson, M.. Westermark. B., and Pontcn. J.
53. 54. 55. 56. 57. 58. 59.
Expression of PDGF ,i-rcceptors in human mcnmuicim.i cells. Int. J. Cancer. 46:112-11Â». 1990. Smits. A.. Hermansson. M.. NistÃ©r,M.. Karnushina. !.. Heldin. C-H., Wes termark. B.. and FuÃ±a,K. Rat brain capillary cndolhclial cells express functional PDGF B-type receptors. Growth Factors. 2: 1-8. 1989. Mark. .1.. PontÃ©n..1.. and Westermark. B. Cytogenetical studies with Ci-band technique of established cell lines of human malignant gliomas. Hereditas. 78: 304-308. 1974. Kavinsky. C. J.. and Garber. B. B. Fibronectin associated with the glial component of embryonic brain cell cultures. J. Supramol. Struct.. //: 269281. 1979. Price. J.. and Hynes. R. O. Astrocytes in cultures synthesize and release a variant form of fibronectin. J. Neurosei.. 5: 2205-2211. 1985. Heppner. G. H.. and Miller. B. Tumor heterogeneity: biological implications and therapeutic consequences. Cancer Metastasis Rev.. 2: 5-23. 1983. Nowell. P. C. The donai evolution of tumor cell populations. Science (Washington DC'). 194: 23-28. 1976 Kraus. M. H.. Popescu, N. P., Amsbaugh. S. A., and King. C'. R. Overexpression of the EOF receptor-related proto-oncogene erh-H-2 in human mammary tumor cell lines by different molecular mechanisms. EMBO J.. 6: 605-610, 1987. Huebner. K.. Isobe. M.. Chao. M.. Botimeli. M.. Ross. A. H.. Finan, J., Hoxie. J. A.. Seghal. A.. Buck. C. R.. LaÃ±aban.A.. Nowell. P. C.. Koprowski, H., and Croce, C'. M. The nerve growth factor receptor gene is at human chromosome region I7ql2-I7q22. distal to the chromosome 17 breakpoint in acute leukcmias. Proc. Nati. Acad. Sci. l'SA. X3: 1403-1407. 1986. Coussens, L.. Yang-Feng. T. L.. Liao. Y-C.. Chen. E.. Gray, A., McGrath, J., Seeburg. P. H.. l.iberman. T. A.. Schlessinger. .1.. Franckc. L'.. Levinson. A., and I'llrich, A. Tyrosine kinase receptor with extensive homology to EGF receptor shares chromosomal location with neu oncogene. Science (Washington DC). 230: 1132-1139. 1985. Coussens. L.. Parker. P. P.. Rhee. L.. Yang-Feng. T. L.. Chen. E.. Waterfield, M. D.. Francke, I1., and I'llrich. A. Multiple, distinct forms of bovine and human protein kinase C suggest diversity in cellular signaling pathways. Science (Washington DC). 233: 859-866. 1986. Lessin. S. R.. Huebner. K.. Isobe. M.. Croce, C'. M., and Steinert. P. M.
Chromosomal mapping of human keratin genes: evidence of nonlinkage. J. Invest. Dermatol.. 91: 572-578. 1988. 64. Garmier. J.. Osguthorpe. D. J.. and Robson. B. Analysis of accuracy and implications of simple methods for predicting the secondary structure of globular proteins. J. Mol. Biol.. 120: 97-120. 1978.