Tissue-specific versus cell type-specific expression of the glial fibrillary acidic protein

Transcription

1 Proc. Natl. Acad. Sci. USA Vol. 90, pp , May 1993 Neurobiology Tissue-specific versus cell type-specific expression of the glial fibrillary acidic protein RYUTA KANEKO* AND NOBORU SUEOKAt Department of Molecular, Cellular, and Developmental Biology, University of Colordo, Boulder, CO Communicated by David M. Prescott, March 8, 1993 (receivedfor review January 12, 1993) ABSTRACT Expression of the glial cell-speciflc gene encoding glial fibriwlary acidic protein (GFAP) is regulated in a tissue-specific (neural tissue versus other tissues) as well as a cell type-specific (glial cell versus neuron) manner. Using a family of rat neurotumor RT4 cell lines in which neuronal/glial differentiation occurs in vitro, along with cell lines of different tissue origins, we identified by transient- and permanenttransfection assays two negative regulatory regions, GFAP downstream regulators 1 and 2 (GDR1 and GDR2). Both regions lie 3' of the transcription start site; GDR1 is in a 2.7-kb region extending from the first intron through the fifth exon, and GDR2 is within 1.7 kb 3' of the polyadenylylation site. GDR1 alone is responsible for tissue-specific expression (suppression in nonneural tissues), while both GDR1 and GDR2 are necessary for cell type-specific expression (suppression in neuronal cells). During development, nervous tissue is differentiated at a very early stage from ectoderm to form the neural tube. This event is followed by the formation of the neural crest, which in turn gives rise to the peripheral nervous system through migration of cells. At the neural crest stage, nerve stem cells have potential to become both neuronal and glial cell types (1-3). Multipotential neural crest cells migrate to their correct destination and generate glial and neuronal cell types. There is a strong possibility that these migrating cells are still at the stem cell stage and retain the potential to generate neuroblasts and glioblasts as well as to renew stem cells (2-5). Thus, there must be at least two major regulatory mechanisms for neural system development: one for the determination of neural tissue lineage (tissue-specific regulation) and another for the determination of neurons and glial cells from common neural stem cells (cell type-specific regulation). Previous work established clonal cell lines from an ethylnitrosourea-induced rat peripheral neurotumor, RT4 (6, 7). The RT4 cell-line family has a stem cell type, RT4-AC, where daughter cells have been repeatedly shown to spontaneously and irreversibly differentiate into one of three derivative cell types at a frequency of 10-5: a glial type, RT4-D, and two neuronal types, RT4-E and RT4-B. The glial type, RT4-D, expresses only glial-specific gene products, including glial fibrillary acidic protein (GFAP), S100(3, P0, suppressed camp-inducible POU protein, a low level of myelin basic protein, and 2',3'-cyclic 3'-nucleotide phosphodiesterase (refs. 8-10; N. Hagiwara, S. mada, and N.S., unpublished work). The two neuronal types, RT4-B and RT4-E, exhibit only neuronal characteristics, including the expression of voltage-dependent Na+ and K+ channels and extension of neurite-like processes (7, 11, 12). The parental stem-cell type, RT4-AC, expresses both glial and neuronal phenotypes (7, 12). Therefore, differential suppression of neuronal and glial genes occurs upon conversion of the stem cell type to the glial or neuronal cell type. These properties of the RT4 cell lines The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact. reflect the in vivo situation in both the central (13-16) and the peripheral (4, 5) nervous system. RT4-AC and its derivative cell types grow indefinitely in culture, so they provide an excellent system for the molecular study of regulatory mechanisms of neuronal- and glial-specific gene expression (17), in particular (a) suppression of both glial and neuronal gene expressions in non-neural cells (tissue-specific regulation) and (b) suppression of glial genes in the neuronal cell types, RT4-B and RT4-E, and suppression of neuronal genes in the glial cell type, RT4-D (cell type-specific regulation). GFAP is an intermediate-filament protein expressed primarily in astrocytes and Schwann cells (18) and is well characterized histocytochemically (19, 20). Modulation of GFAP gene expression during development, wound healing, and disease has been observed (21-23). n this report, we describe the existence of two regulatory regions of the GFAP gene that are located downstream of the transcription start site; one is responsible for the tissue-specific regulation (nervous tissue versus other tissues), and both regions are necessary for the regulation of cell type-specific expression (neuronal cells versus glial cells). MATERALS AND METHODS Cell Lines and Culture Conditions. RT4 clonal lines used in these studies were RT4-AC36A, RT4-B8, RT4-D6, and RT4-E5 (6). Rat hepatocytoma cell line HTC was originally obtained from the late G. M. Tomkins. Rat mammary carcinoma cell line was obtained from T. Kano-Sueoka (24). Rat pheochromocytoma cell line PC12, rat astrocytoma cell line C6, and rat kidney fibroblast cell line NRK-49F were purchased from the American Type Culture Collection. RT4 cells and NRK-49F cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum. C6 cells were cultured in DMEM supplemented with 15% horse serum and 2.5% fetal bovine serum. HTC cells and cells were cultured in DMEM supplemented with 8% horse serum and 8% fetal bovine serum. Permanent and Transient Transfection. Permanent transfection of RT4-E was carried out with puc19 or AEMBL3 vectors carrying various portions of the GFAP gene, and expression of fibrous GFAP was examined by fluorescent immunostaining. RT4-E cells were transfected with 75,ug of px14 plasmid or 100 pg of A GFAP vectors plus 1 ug of prsvneo plasmid (Fig. 1) by the calcium phosphate method (28). DNAs used for transfection were purified by two rounds of CsCl/ethidium bromide equilibrium ultracentrifugation. Two days after transfection, G418 (400,ug/ml) was added, cells were incubated for 2 weeks, G418-resistant colonies were selected, and individual colonies were isolated. mmunostaining was performed with mouse anti-human GFAP Abbreviations: GFAP, glial fibrillary acidic protein; GDR, GFAP downstream regulator. *Present address: Nippon Petrochemicals Company, Uchisaiwaicho, Chiyoda-ku, Tokyo 100, Japan. tto whom reprint requests should be addressed. 4698

2 Neurobiology: Kaneko and Sueoka A -10Kb. s..a 0 transciption exonl start "-, 2k. jtron1 *xon9l +1OKb * * polyadenylylation Proc. Natl. Acad. Sci. USA 90 (1993) 4699 Relative Activity of Luciferase (Transient Expression Amoy; SV40 u100) RT4 AC36A BS E5 06 neural nouronsl gilal stem CS PC12 HTC 6424 NRK- 49F gial neuronal liver mamma- ry gbnd kddney fibrobi '3.8k ' pgl k pgl k, B. :2k:. BP' S sob. e pgl2+b101 pgl2+bs'+sb k. 'S pgl2+bs+sb' k. 2k 'BS pgl2+bs+sb pg 2XX S 51 pgl2+bs GDR1 4 GDR2 B 2kb SV40 t antigen sequence fupa (for splicing and polyadenylylation) pgl2+bs+sb pbr322 vector FG. 1. Transient-transfection experiments with luciferase vectors carrying various regions of the rat GFAP gene. (A) Composition of the luciferase expression vectors and relative activities of transiently expressed luciferase. Genomic organization of the GFAP gene is diagrammed at the top left. Black boxes, white boxes, and the hatched box indicate exons, introns, and the 3' untranslated region (3'-UTR), respectively. The termination site and the exon-intron arrangements were assumed to be similar to those of the mouse GFAP gene (25) by the rationale presented in the text. The restriction enzyme sites shown on the map are those of the rat GFAP gene. Various segments of the GFAP genomic sequence were inserted into the luciferase vector pxp1 (26, 27), and the constructs were named as shown in tlhe center column. Transiently expressed luciferase activity was assayed enzymatically for each vector. The table indicates the levels of luciferase expression of each construct in different cell lines. The values in the table are luciferase activities for each vector normalized to the activity of a control plasmid, psluc2 [a luciferase vector containing only the simian virus 40 (SV40) ]. The activity of psluc2 activity is equated to 100. The values are the average of three to eight independent transfection experiments except for those marked with a star, which are the results of single transfection experiments. GFAP downstream regulators 1 and 2 (GDR1 and GDR2), indicated at the bottom of the diagram, are the regions of putative regulatory elements (see text). (B) Arrangement of the components in pgl2+bs+sb as an example of regulatory expression vectors. Various regions of the GFAP genomic sequence (BS and SB fragments are shown) are inserted upstream of the 2-kb 5' fragment in their original genomic orientation. All expression vectors in A consist of fragments of the GFAP gene inserted into pgl2, which consists of the 2-kb upstream region of the GFAP gene, luciferase cdna, and pbr322 vector sequence. monoclonal antibody (Boehringer Mannheim) followed by fluorescein-conjugated goat anti-mouse gg secondary antibody (Boehringer Mannheim); confluent cultures were used, since GFAP expression reaches its maximum when cultures become confluent (8, 9). Transient transfection experiments were performed with the RT4 cell lines and with PC12, C6, HTC, 64-24, and NRK-49F. Cells were transfected with various luciferase expression vectors by the calcium phosphate method as for permanent transfections and harvested 72 hr after transfection (at which time the cultures were confluent) and luciferase was assayed enzymatically (Promega). RESULTS To isolate clones of the GFAP gene, we initially screened a genomic library from rat liver (Clontech) with a mouse GFAP cdna probe (29). The entire rat GFAP coding sequence and the flanking regions were cloned, including 14 kb downstream from the transcription start site and 11 kb upstream of it. The transcription start site was determined by RNase protection and primer extension assays. The results indicate that there is a single transcription start site corresponding exactly to that of the mouse GFAP gene (25, 30). Although the nucleotide sequence of the entire rat GFAP gene is not available at present, there is the following supportive evidence indicating that the structure ofthe rat GFAP gene is similar to that of the mouse GFAP gene: (i) the sequence of 2 kb of the upstream region closest to the transcription start site of the rat GFAP gene (R.K., N. Hagiwara, and N.S., unpublished) shows 94% identity to the corresponding region of the mouse GFAP gene (25, 30); (ii) we have also sequenced a 200 bp of a rat clone positioned 6 kb downstream of the transcription start site, which showed 98% identity to the mouse sequence at a comparable position (the end of the sixth exon) of the mouse gene (25); and (iii) expression vector px14 (see below

3 4700 Neurobiology: Kaneko and Sueoka and Fig. 2), which contains the entire coding region (including the polyadenylylation site predicted from the mouse sequence), allowed expression of GFAP in transfected cells (this work), indicating that the rat polyadenylylation site is most likely in a similar region to that of the mouse. For analysis of the regions of the rat GFAP gene downstream of the transcription start site, we therefore assume that the lengths and arrangements of the exons and introns and the positions of the polyadenylylation site are similar in mouse and rat (Figs. 1 and 2). To define whether specific regions of the gene were necessary for glial-specific expression of GFAP, expression vectors were constructed using the luciferase expression vector pxp1 for transient transfections (26, 27) and using puc19 and AEMBL3 for permanent transfections. We initially sought to define whether any regulatory sequences were located in the 2-kb region immediately 5' to the transcription start site. The luciferase expression vector pgl2 (Fig. 1A) contains 1958 bp of the 5' flanking region and 25 bp of the transcribed region of the rat GFAP gene. n all RT4 cell lines and nonneural cell lines tested by transient transfection assays, this construct was expressed at nearly the same level as the positive control plasmids psluc2 (SV40 driven; ref. 27) and pt109 (thymidine kinase driven; ref. 27) (data not shown). Construct pgl10.8, containing an additional 8.8 kb of the 5' flanking region, was then tested. This construct was expressed in all cell lines (Fig. 1A). The 5' flanking region (up to 10.8 kb) of the rat GFAP gene is apparently not sufficient to suppress GFAP expression in any rat cell types tested. These results suggest that tissueand cell type-specific regulatory elements lie elsewhere, either further upstream or downstream of the transcription start site. To determine whether regions downstream of the transcription start site were involved in glial-specific regulation of GFAP, segments of the gene were cloned into expression vectors and permanent transfectants were established by using the neuronal cell line RT4-E as the recipient. Two GFAP expression vectors were used: a genomic clone, px14 (which contains DNA sequences from -4 kb to +10 kb relative to the transcription start site, including the entire coding region through the polyadenylylation site), and a A GFAP clone (which contains DNA sequences from -7 kb to +13 kb relative to the transcription start site, including the entire coding region through the polyadenylylation site, and 3 kb downstream of the polyadenylylation site) (Fig. 2). GFAP expression was assayed by immunological staining. nterestingly, with the px14 transfectants, 150 out of 150 individual clones showed a GFAP fiber network stained with anti-gfap antibody, whereas in the A GFAP transfectants, -1 OKb 0.. a transcription start site +10Kb.. polyadenylylation site \ GF^P J~~~~~~~~~R (Sau3A) (Sau3A) Xpol Xiol (Sau3A) ~~~~~~ Xhol 14kb rat genomic fragment ~ ~ ~ ~ ' 0 < p~uc19 vector 4pX14 A-EMBL3 vector 20kb rat genomic fragment Lambda GFAP FG. 2. Map of the rat GFAP genomic clones for permanenttransfection experiments. Black boxes are exons, white boxes are introns, and the hatched box is the 3' untranslated region. px14 was constructed by inserting the Xho genomic fragment (14 kb) into puc19, and the A GFAP vector was made by inserting the Sau3Al fragment (20 kb) into AEMBL3. 79 out of 80 individual clones did not show any GFAP staining. This result strongly suggested that there was a neuronal-specific negative element(s) downstream of the polyadenylylation site. To analyze the downstream region further, we constructed and examined a number of luciferase expression vectors containing various downstream regions in addition to the 2-kb upstream region (Fig. 1A). The luciferase construct pgl2+bb10, which carries 2 kb ofthe 5' flanking region plus a 10-kb fragment including the first intron through 2-kb downstream of the polyadenylylation site (Fig. 1A), conferred high levels of expression in the glial cell lines RT4-D and C6 and the stem cell line RT4-AC, but not in neuronal cell lines RT4-E and PC12 or in cell lines from other rat tissues (Fig. 1A). These data confirm the results obtained by permanent transfection described above, and further show that the regions which pgl2+bb10 contains are sufficient to suppress GFAP expression not only in neuronal cell types but also in nonneural cell types. To better define the regulatory region(s), deletion mutants of pgl2+bb10 were made and tested as shown in Fig. 1A and Fig. 3. The results with these constructs revealed that two regions were required for the suppression of the expression of GFAP in neuronal cell lines. One, termed GDR1, is located in a 2.7-kb region extending from the first intron through the fifth exon. The other, termed GDR2, is located in a 1.7-kb region downstream of the polyadenylylation site (Fig. 1A). GDR1 and GDR2 were functional when placed upstream of the 2-kb 5' flanking region (Fig. 1B). Therefore, GDR1 and GDR2 function in a position-independent manner, similar to other "silencers" (31). The presence of both regions was necessary for suppression of GFAP gene expression in neuronal cell types (construct pgl2+bs+sb, Figs. 1 and 3). However, for suppression ofgfap expression in nonneural cells, only GDR1 was necessary (construct pgl2+bs, Figs. 1 and 3). The construct pgl2+x3.2, which carries only GDR2 (Fig. 1A), showed GFAP expression in all A GDR Glial [ RT4-D61 RT4-B8 Neuronal RT4-E5 L PC12 HTC Other tssue NRK-49F B GDR1+GDR2 Neuralstem[R4A Neuralstem [ RT4-AC Glial [ RT4-D6 C6 RT4-B8 Neuronal Other Proc. Natl. Acad Sci. USA 90 (1993) RT4-E5 L PC12 HTC tissue LNRK-49F Relative Lucifere Activity ~~~~~~~~~~~~~~~~~~~~~~~~ 0 50 ioo l FG. 3. Transient expression ofluciferase in various rat tissue cell lines. (A) Luciferase activity of various cell lines with the pgl2+bs expression vector, which carries GDR1. (B) Luciferase activity of various cell lines with the pgl2+bs+sb expression vector, which carries both GDR1 and GDR2 (see Fig. 1 for detail). Values of the luciferase activity of pgl2+bs or pgl2+bs+sb were normalized to the activity of psluc2 (control) transfectants for each of the cell lines shown in Fig. 1.

4 polyad. site Neural tissue Other GFAP gone Glial Neuronal Tissue GDR1 GDR2 Neurobiology: Kaneko and Sueoka i GDR1 + + GDR FG. 4. Summary of tissue and cell type specificity of GFAP gene expression. polyad., Polyadenylylation. cell lines, and therefore GDR2 alone is not sufficient for either tissue-specific or cell type-specific suppression. DSCUSSON We conclude from these results that GDR1 is responsible for tissue (neural)-specific expression and that GDR1 and GDR2 together are responsible for the cell type (glial)-specific expression of GFAP (Fig. 4). Using permanent transfections of the human glioma cell line U251 and epitheloid carcinoma cell line HeLa with mouse GFAP genomic clones, Sarkar and Cowan (32) reported that a region including the first intron to fourth exon of the mouse GFAP gene has a negative regulatory activity responsible for the suppression of GFAP expression in HeLa cells but not in glioma cells. The regulatory region those workers identified corresponds to the GDR1 region of the rat GFAP gene. Their observation agrees with our finding that a downstream element(s) of the GFAP gene is responsible for the tissue specificity of GFAP expression. Other groups have described the presence of glial-specific positive elements located upstream of the transcription start site in the mouse or human GFAP genes. Miura et al. (30) reported that only a 256-bp region 5' of the transcription start site is necessary for cell type (glial)-specific expression of the mouse GFAP gene. Sarid (33), also using the mouse GFAP gene, reported that only the 5' flanking region is necessary for C6 rat glioma-specific expression. Besnard et al. (34), using the human GFAP gene in human astrocytes and other tissuederived cell lines, reported that the 2-kb 5' flanking region alone allowed expression in glial cells. According to our results, expression vectors containing at least up to 10.8 kb upstream of the transcription start site were expressed not only in glial cells but also in neuronal and nonneural cells. As will be described elsewhere (R.K., N. Hagiwara, K. Leader, and N.S., unpublished work), we have discovered campresponsive enhancer elements upstream of the transcription start site. These enhancer elements may be equivalent to some of the positive regulators described by others. On the basis of our results, we propose the following hypothesis to be tested on the mechanism of GFAP gene expression. During development, those cells that differentiate into nonneural tissues have a negative regulatory factor(s) that binds to GDR1 and inhibits expression of GFAP. This ubiquitous repressor (UbR) may be functional in all nonneural tissues. n contrast, those cells that differentiate toward neural precursor cells (neuronal-glial stem cells) lack UbR, or UbR is inactivated by a neural-specific factor (NSF). When neural stem cells convert to neuronal cell types, neuronal cells synthesize the negative factor (neuronal downstream repressor, NDR) which binds to GDR2, and this complex interacts with GDR1 to stop the transcription. Alternatively, the NDR-GDR2 complex may reactivate UbR to bind to GDR1 and inhibit transcriptional initiation. We do not exclude the possibility that the function of GDR1 is involved in chromatin-mediated suppression of the GFAP gene (35). Proc. Natl. Acad. Sci. USA 90 (1993) 4701 Expression vectors such as pgl2+bb10 and pgl2+bs+ SB are complete regulatory vectors. These vectors include most, if not all, regulatory elements for the expression of the GFAP gene in various cell types with respect to both tissueand cell type-specific regulation and also for campresponsive expression in stem cell and glial cell types. These regulatory vectors (pgl2+bb10 and pgl2+bs+sb) of the GFAP gene are expected to be useful for the expression of proteins exclusively in glial cells paralleling GFAP expression. Further analysis of GDR1 and GDR2 and the identification of factors that bind to these elements should provide crucial information about the mechanisms of tissue-specific and cell type-specific regulations of gene expression in development of the nervous system. We thank Dr. N. J. Cowan for donating the cdna clone of the mouse GFAP. We thank our colleague Nobuko Hagiwara for providing the 2-kb upstream fragment of the GFAP gene and sequence information, and Adam R. Jenkins for technical assistance. We are grateful to Drs. Tamiko Kano-Sueoka, Joseph Heilig, and John J. Laffan and other members ofour laboratory for critical reading of the manuscript. This project has been funded by a grant (NS21512) from the National nstitutes of Health. R.K. was supported by Nippon Petrochemicals Company. 1. Frank, E. & Sanes, J. R. (1991) Development 111, Fraser, S. E. & Bronner-Fraser, M. (1991) Development 112, LeDouarin, N. C., Durac, C., Dupin, E. & Cameron-Curry, P. (1991) Glia 4, Duff, R. S., Langtimm, C. J., Richardson, M. K. & Sieber- Blum, M. (1991) Dev. Biol. 147, Stemple, D. L. & Anderson, D. J. (1992) Cell 71, mada, M. & Sueoka, N. (1987) Dev. Biol. 66, Tomozawa, Y. & Sueoka, N. (1978) Proc. Natl. Acad. Sci. USA 75, Freeman, M. & Sueoka, N. (1987) Proc. Natl. Acad. Sci. USA 84, Freeman, M., Beckmann, S. L. & Sueoka, N. (1989) Exp. Cell Res. 182, Gandelman, K.-Y., Pfeiffer, S. E. & Carson, J. H. (1989) Development 106, Droms, K. & Sueoka, N. (1987) Proc. Natl. Acad. Sci. USA 84, Donahue, L. M., Schaller, K. & Sueoka, N. (1991) Dev. Biol. 147, De Vitry, F., Picart, R., Jacque, C., Legault, L., Deponey, P. & Tixier-Vidal A. (1980) Proc. Natl. Acad. Sci. USA 77, Turner, D. L. & Cepko, C. L. (1987) Nature (London) 328, Temple, S. (1989) Nature (London) 340, Cattaneo, E. & McKay, R. D. G. (1990) Nature (London) 347, Sueoka, N. & Droms, K. (1986) Curr. Top. Dev. Biol. 20, Eng, L. F. (1980) Proteins ofthe Nervous System (Raven, New York), 2nd Ed., pp Lazarides, E. (1980) Nature (London) 283, Mokuno, K., Kamholz, J., Behrman, T., Black, C., Sessa, M., Feinstein, D., Lee, V. & Pleasure, D. (1989) J. Neurosci. Res. 23, 3% Mason, P., Attema, B., De Vries, G. H. (1991) Neurotoxicology 12, O'Callaghan, J. P. (1991) Biomed. Environ. Sci. 4, Tardy, M., Fages, C., LePrince, G., Rolland, B. & Nunes, J. (1991) Adv. Exp. Med. Biol. 265, Kano-Sueoka & Hsieh, P. (1973) Proc. Natl. Acad. Sci. USA 70, Balcarek, J. B. & Cowan, N. J. (1985) Nucleic Acids Res. 13, de Wet, J. R., Wood, K. V., DeLuca, M., Helinski, D. R. & Subramani, S. (1987) Mol. Cell. Biol. 7, Nordeen, S. K. (1988) BioTechniques 6, Graham, F. & van der Eb, A. (1973) Virology 52,

5 4702 Neurobiology: Kaneko and Sueoka 29. Lewis, S. A., Balcarek, J. M., Krek, V., Sheranski, M. & Cowan, N. J. (1984) Proc. Natl. Acad. Sci. USA 81, Miura, M., Tamura, T. & Mikoshiba, K. (1990) J. Neurochem. 55, Goodboum, S. (1990) Biochim. Biophys. Acta 1032, Proc. Natl. Acad. Sci. USA 90 (1993) 32. Sarkar, S. & Cowan, N. J. (1991) J. Neurochem. 57, Sand, J. (1991) J. Neurosci. Res. 28, Besnard, F., Brenner, M., Nakatani, Y., Chao, R., Purohit, H. J. & Freese, E. (1991) J. Biol. Chem. 266, Elgin, S. C. (1990) Curr. Opin. Cell Biol. 2,