Journal of Neuroimmunology, 36 (1992) 157-169
© 1992 Elsevier Science Publishers B.V. All rights reserved 0165-5728/92/$05.00 JNI 02114
Transforming growth factor-beta 1 (TGF-/31) expression and regulation in rat cortical astrocytes A n n a da Cunha and Ljubi~a Vitkovid Laboratory of Irnmunoregulation, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20892, USA (Received 5 August 1991) (Accepted 3 September 1991)
Key words: Transforming growth factor-/31; Astrocyte; Interleukin-1
Summary Transforming growth factor-beta 1 (TGF-/31) is a potent modulator of immune and glial cells' functions and thus, could play an important role in neuro-immune interaction. However, published reports disagree on whether or not TGF-/31 is expressed in normal brain. We demonstrate here the constitutive expression of TGF-/31 mRNA but not protein in both cerebral cortex and primary rat cortical astrocytes. Steady-state TGF-/31 mRNA level increased 2-fold in adult compared to neonatal cortex and during proliferation and differentiation of astrocytes in primary culture. This response was not accompanied by the appearance of detectable TGF-/3 protein either in vivo or in vitro. However, both intracellular immunoreactive TGF-/3 and extracellular TGF-/31 activity were detected upon in vitro stimulation of astrocytes with interleukin-1 (IL-1). The extracellular TGF-/31 increased with time of exposure to and concentration of IL-1. In contrast, the amount of TGF-/31 mRNA remained unchanged during stimulation of astrocytes with IL-1. These results suggest that the production of TGF-/31 in astrocytes is regulated at both mRNA and protein levels. The former may occur during astrocytic development, and the latter during astrocytic response to injury in association with elevation of IL-1.
Introduction TGF-/31 is a multifunctional cytokine produced by and acting on various cell types (Rob-
Correspondence to: Dr. L. Vitkovi~, National Institutes of Health, Building 10, Room llB-13, Bethesda, MD 20892, USA. Tel. (301) 496-7296, Fax (301) 402-0070. Abbreviations: FCS, fetal calf serum; TGF-/3, transforming growth factor-beta; GAP-43, growth-associated protein 43; GFAP, glial fibrillary acidic protein; IL-1, interleukin-1; LPS, lipopolysaccharide; TNF-a, tumor necrosis factor-alpha.
erts and Sporn, 1990). It is a potent attractant and deactivator of monocytes and, suppressor of T- and B-cell proliferation (Kehrl et al., 1986a, b; Wahl et al., 1987; Tsunawaki et al., 1988). TGF-fll acts on neural cells by stimulating phosphatidylinositol turnover and translocation of protein kinase C (Robertson et al., 1988), suppressing antigen presentation (Schluesener, 1990), decreasing DNA synthesis and inhibiting glutamine synthetase activity (Toru-Delbauffe et al., 1990) in astrocytes and inducing mitogenesis in Schwann cells (Ridley et al., 1989). Thus, TGF-fll could be
an important mediator of neuro-immune interactions, if present in brain. TGF-/3 protein and mRNA were not detected in sections of adult mouse brain by immunocytochemistry (Thompson et al., 1989) or in situ hybridization (Wilcox and Derynck, 1988a). These results contrast with the more recent report of an unidentified isoform of TGF-/3 mRNA detected by DNA-polymerase chain reaction in mouse brain and astrocytes derived therefrom (Wesselingh et al, 1990). TGF-/31 was found in a wide variety of normal and malignant cells (Roberts and Sporn, 1990) including gliomas, metastatic brain tumors and meningiomas (Takahashi et al., 1990). TGF-/31 mRNA was detected in the glioblastoma cell line A172 (Derynck et al., 1985). However, TGF-/31 expression by transformed glial cells is not a reliable indicator of expression in their normal counterparts, astrocytes or oligodendrocytes. For example, the HEP-G2 hepatoma cell line expresses TGF-/31 mRNA at a high level (Derynck et al., 1985) but its counterpart, the hepatocyte, does not express any (Nakatsukasa et al., 1990). Thus, whether or not TGF-/31 gene is expressed in normal brain and astrocytes derived therefrom remains uncertain. We have been characterizing astrocytic phenotype at the protein level in vitro and in vivo. Towards this end we have recently demonstrated the presence of two proteins not previously reported to be produced by astrocytes: growth-associated protein 43 (GAP-43) and endothelin-3 (Vitkovic et al., 1988; da Cunha and Vitkovic, 1990; Ehrenreich et al., 1991). We describe here expression of another protein, TGF-/31 in nontransformed, primary rat cortical astrocytes. The amount of steady-state TGF-/31 mRNA increased during proliferation and differentiation of astrocytes in culture and in cerebral cortex. It appears, therefore, that TGF-/31 gene expression is regulated at the level of mRNA synthesis. If this message is translated into protein, then the amounts were too low to be detected in vivo and in vitro. TGF-/31 message was, however, translated into detectable amounts of TGF-/31 protein in the presence of interleukin-1 (IL-1). The amount of steady-state TGF-/31 message itself was unaltered while the protein was being induced with IL-1. Thus, in the presence of IL-1,
TGF-/31 expression was apparently regulated at the level of mRNA translation, protein release or processing.
Materials and methods
Astrocyte cultures Brain cortices of 1-day-old Sprague-Dawley rats were the source of astrocytes. Astrocyte cultures were prepared as previously described (Sensenbrenner et al., 1980; Vitkovic et al., 1988). The number of astrocytes per area was determined by detaching cells with trypsin (0.25%, w/v), EDTA (0.02%, w/v) at 37 °C for 2 rain, diluting (1:1, v/v) in trypan blue (0.4%, w/v) and counting in a hemocytometer.
Character&ation of astrocyte cultures Astrocytes and other cell types in primary cultures were identified by immunolabeling as previously described (Vitkovic et al., 1988; da Cunha and Vitkovic, 1990).
Qualitative analysis of TGF-/31 mRNA by Northern blotting Total RNA was isolated with RNAzol (Cinna/ Biotecx, Friendswood, TX, USA) following the manufacturer's protocol, and denatured thermally and chemically as follows: RNA samples were suspended in 1.6 × MOPS buffer containing 12.5% (v/v) glycerol, 10% (v/v) formaldehyde, 72% (v/v) formamide and heated at 95 °C for 5 rain. 20/xg RNA per sample was electrophoresed through a 1.0% (w/v) agarose-formaldehyde gel in the presence of ethidium bromide and then blotted onto nitrocellulose. An EcoRI insert from AC1 TGF-/31 cDNA was labeled with [32p]dCTP (3000 Ci/mmol, NEN, CA, USA) to an approximate specific activity of 4.4 × 109 dpm//~g using random sequence hexanucleotides (Amersham, Arlington Heights, IL, USA) and hybridized to the RNA on nitrocellulose as previously described (Derynck et al., 1985). The human TGF/31 cDNA sequence used here has a remarkably high degree of homology with its murine counterpart (Derynck et al., 1986; Roberts and Sporn, 1990) and has been previously used for probing rodent RNA for TGF-/31 mRNA (Derynck et al.,
159 1988). The blot was washed 3 times, 20 min each, in 0.2 × SSC, 0.1% sodium dodecyl sulfate at 55 ° C and autoradiographed using Kodak X-Omat XAR-5 film plus DuPont Cronex intensifying screens at - 7 0 ° C overnight. TGF-/31 mRNA was assessed with laser scanning densitometry (Ultrascan XL densitometer, LKB, Bromma, Sweden) of autoradiograms. RNA from the T-cell leukemia cell line, Jurkat, known to constitutively express TGF-/31 mRNA (Kehrl et al., 1986a) served as a positive control, whereas RNA from adult rat liver served as a negative control (Nakatsukasa et al., 1990).
Quantitative analysis of TGF-/31 mRNA The procedures used to quantify TGF-/31 mRNA were carried out as previously described (Hoyt and Lazo, 1989). Briefly, extracted total RNA was dissolved in 11 × SSC and 13% formaldehyde, heated at 60 °C for 15 min, serially diluted and slotted onto nitrocellulose using the manifold apparatus according to the manufacturer's instructions (BRL, Gaithersburg, MD, USA). Amounts of RNA slot-blotted were 1, 2, 4, 6, 8, 10, 15 and 20 /xg. Jurkat T cell RNA was also slot-blotted as positive (Kehrl et al., 1986a) and adult rat liver RNA as negative (Natatsukasa et al., 1990) controls for TGF-/31 mRNA. The hybridization and washing protocols described above for Northern blotting were also utilized here. The RNA on slot blots was hybridized with the TGF-/31 cDNA, stripped and re-hybridized with either 32p-labeled oligo(dT)19_22 to estimate the amount of poly(A) + RNA as described by Harley (1987) or actin cDNA (Oncor, Gaithersburg, MD, USA) according to the manufacturer's instructions. The hybridization signals were quantified by scanning densitometry of autoradiograms. Several autoradiograms exposed for different times were scanned to ensure that densities of bands were in the linear range of the film. The amount of mRNA was calculated from a minimum of three linear data points obtained for three amounts of total RNA. The optical density profile of each slot was recorded on a chart and the area under the curve calculated from the peak height and peak width at one-half the height. A linear signal response was obtained with 1-4 /zg of RNA; larger amounts of RNA did not yield
greater signals presumably due to saturation of the nitrocellulose. The ratios of mRNA signals to the poly(A) + RNA signal in a sample were calculated as representing that fraction of the total mRNA that is TGF-/31 mRNA.
Qualitative analysis of TGF-/3 protein by immunocytochemistry The procedures used for immunohistochemical staining of cells and tissue sections were similar to those described previously (Heine et al., 1987; Flanders et al., 1989). The Jurkat T-cell line known to constitutively secrete TGF-/31 (Kehrl et al., 1986a) and 15-day-old mouse embryos with a known distribution of TGF-/3 immunoreactivity (Heine et al., 1987) were used as positive controls for staining of cells and tissue sections respectively. Briefly, Jurkat T cells were washed in phosphate buffered saline (PBS) and deposited on glass coverslips prior to fixation. These cells and primary astrocytes and tissues were fixed in Bouin's solution for 20 min and 3 days respectively as described by Heine et al. (1987). Cells were rinsed thoroughly in wash buffer (DMEM containing 0.02 M Hepes). The tissues were paraffin embedded, sectioned at 5 ~m and mounted on poly-L-lysine coated slides. After deparaffinization, tissue was passed through graded series of ethanol dilutions, rinsed in water and blocked with 10% goat serum (Vector Laboratories, Burlingame, CA, USA). In preliminary experiments, tissue sections were digested with 1 mg/ml hyaluronidase (Sigma, St. Louis, MO, USA) in 0.1 M sodium acetate buffer (pH 5.5) containing 0.15 M NaC1 for 10 min. Cells and tissue sections were overlaid with affinity-purified rabbit anti-TGF-/3 antibody (R & D Systems, Minneapolis, MN, USA) diluted 1:50 (v/v) in 0.1% gelatin, 1% BSA in wash buffer (diluent), and left to incubate overnight at room temperature. This antibody does not distinguish 131 and /32 isoforms of TGF (R & D Systems product information). Immunoreactivity was visualized with biotinylated sheep anti-rabbit IgG (secondary antibody, Boehringer-Mannheim, Germany, 1:50, v/v) followed by streptavidin alkaline-phosphatase (Amersham, 1:50, v/v) or streptavidin horseradish peroxidase (Amersham, 1:200, v/v). The alkaline phosphatase substrate
solutions used were supplied with the R & D Systems' TGF-/3 detection kit, and those for horseradish peroxidase were from the ABC kit (Vector Laboratories). When the peroxidase detection system was used, cells were incubated with hydrogen peroxide before incubation with primary antibody as described (Heine et al., 1987). All incubations were carried out at room temperature for 30 min, followed by three 5 min washes in wash buffer. Several control experiments were performed: (a) the primary antibody was omitted from the diluent and cells were immunoreacted with secondary antibody; (b) the antibody in diluent (1:50, v/v) was incubated with a 20-fold molar excess of porcine TGF-/31 protein (R & D Systems) for 2 h at room temperature prior to immunocytochemistry; and (c) the antibody in diluent (1:10, v/v) was incubated with TGF-/31 coupled to Sepharose 4B, overnight at 4°C and centrifuged to remove anti-TGF-/3 antibodies. In these experiments no staining of cells nor tissues was observed. The immunostaining was reproduced using the second rabbit anti-TGF-/3 (neutralizing) antibody obtained from the same source (R & D Systems). Cells or tissues were washed in distilled water and mounted in aqueous mounting medium (Chemicon, E1 Segundo, CA, USA). Tissue sections were counterstained with fast red, dehydrated and mounted in Permount (Fisher Scientific, PA, USA).
Stimulation of astrocytes Astrocytes (2.2 × 107) were grown in culture for 12 days and then adapted to a medium containing 1% (v/v) dialysed fetal calf serum (FCS) (Gibco, Grand Island, NY, USA) for 18 h, washed with PBS (pH 7.4), exposed to the fresh medium and either left unperturbed or exposed to stimulants as follows: lipopolysaccharide (LPS) (1 and 10 ~g/ml; Escherichia coli 026:B6; RIBI Immunochemical Research, Hamilton, MT, USA), IL-1 (concentrations indicated in the text; recombinant human IL-la; specific activity 107 units/ mg; Genzyme, Boston, MA, USA) and tumor necrosis factor-alpha (TNF-a) (5 and 50 ng/ml; recombinant murine TNF-c~; specific activity 4 × 107 units/rag; Genzyme). The cells were incubated for the times indicated and stained with the anti-TGF-/3 antibody. Culture fluids were col-
lected, transiently acidified to activate TGF-/3 (Brown et al., 1990), concentrated 10-fold and assayed for TGF-/3 activity. Neither of the stimulants interfered with the TGF-/3 bioassay.
Quantitatit~e analysis' of TGF-~ protein by estimation of biological actit,ity Biological activity of TGF-/3 was assessed by the mink lung epithelial cell (ATCC CCL-64) growth inhibition assay described by Danielpour et al. (1989). Serial dilutions of TGF-/31 protein (to give final concentrations from 1600 to 1.5 pg/ml) were assayed in duplicate and used as a standard. Samples containing an unknown amount of TGF-/3 were divided in half, one half assayed directly and another acidified with 6 N HC1 to pH 2. After 10 min, the acidified samples were neutralized with 6 N N a O H : I M Hepes (1:1, v/v) (Brown et al., 1990) and left for 5 rain at room temperature to allow the pH to stabilize at 7.0. All samples were serially diluted up to 1:1064 (v/v) and added (80 /xl) to the wells containing exponentially growing CCL-64 cells. Each sample was assayed in duplicate. Proliferation of CCL-64 cells was measured in wells to which buffer alone was added. 22 h after addition of samples, 0.25 p~Ci of [3H]deoxythymidine ([3H]TdR, specific activity 60 Ci/mmol; ICN)was added to each well. 4 h later incorporation of the label was stopped by adding 1.0 ml of methanol/acetic acid (3:1, v/v) per well. Radioactivity in the lysates was quantified in the presence of scintillation fluid (Aquasol) in a 2000 CA Tri-Carb liquid scintillation counter (Hewlett-Packard). Growth inhibition was expressed as percent of [3H]TdR incorporation in the absence of inhibitors. A dose-response curve (serial dilution vs. percent inhibition) was plotted, EDs0 calculated from it, compared with the EDs0 of the standard (standard porcine TGF-/31; R & D Systems) and TGF-/3 concentration in a sample was deduced. EDso values for the TGF-/31 standard were 0.8 + 0.3 pM (n = 10) which compares well with the previously published data (Danielpour et al., 1989). The specificity of the assay for TGF-/31 was determined as follows: serial dilutions of the standard were incubated with neutralizing rabbit antiTGF-~I,2 antibody (final concentration of 10
izg/ml IgG, R & D Systems) for 1 h at 22 o C, in a total volume of 80 ~l TBS and 1 mg/ml BSA, pH 7.5. The samples were then assayed for TGF-/3 as described above. A 60% neutralization of porcine TGF-/31 activity equal to 5 times the EDs0 was observed in agreement with data supplied by the manufacturer. A neutralizing turkey anti-TGF-/31 antibody was also used in the above assay in identical manner to the neutralizing antibody described above. This antibody neutralized to 95 + 3% (n = 5) porcine TGF-/31 in the above assay. Neither of the neutralizing antibodies alone or normal turkey serum had any effect on the growth of CCL-64 cells.
Blocking of TGF-fl-fike activity in astrocyte-conditioned medium by anti-TGF-fl antibodies A medium was conditioned for 24 h by astrocytes stimulated with IL-la (250 units/ml) and diluted in the presence or absence of various concentrations of neutralizing antibody (either anti-TGF-/31 or anti-TGF-/31,2 as indicated) into the TGF-/3 assay described above. The control measured incorporation of [3H]TdR into CCL-64 cells in the absence of the conditioned medium and antibodies. The antibodies alone had little if any effect on incorporation of [3H]TdR into CCL-64 cells. The data are expressed as percent of [3H]TdR incorporation not inhibited by the astrocyte-conditioned medium, as previously described (Danielpour et al., 1989).
Results and d i s c u s s i o n
TGF-[31 mRNA is constitutively expressed in primary astrocytes and cerebral cortex The presence of TGF-/31 mRNA was determined in primary astrocytes isolated from neonatal rat cortices by Northern analyses. Astrocyte cultures were 95-98% homogeneous 2 weeks after seeding, as judged by labeling with antiserum against GFAP, a cell-type specific marker. The small proportion of contaminating cells were fibroblasts. There was no immunocytochemical evidence of either neurons or oligodendrocytes (by day 2) nor microglia (by day 7 after seeding) in these cultures (da Cunha and Vitkovic, 1990; Vitkovic et al., 1990). Astrocytic RNA was ana-
Fig. 1. TGF-131 mRNA is detectable in primary rat cortical astrocytes and in neonatal cortex. Northern blot containing total cellular RNA (20 /xg/lane) from neonatal rat cortex (lane 1) and rat cortical astrocytes cultured for 7 and 31 days (lanes 3 and 4 respectively) and Jurkat T cells (positive control; lane 2) probed with TGF-/31 cDNA revealed a single 2.4 kb mRNA corresponding in size to TGF-/31 mRNA. The data are representative of three independent experiments.
lyzed under conditions used previously for the hybridization of this human TGF-/31 cDNA probe to rodent RNA (Derynck et al., 1988). An RNA hybridizing with the TGF-/31 probe was detected in 7-, 9-, 15- and 31-day-old primary astrocytes (Fig. 1). The size of this RNA was similar to the size of TGF-/31 mRNA from Jurkat T cells (Fig. 1, Kehrl et al., 1986), rodent cell lines (Derynck et al., 1986, 1988) and mouse embryo (Heine et al., 1987). These results suggest that TGF-/31 was constitutively produced by rat astrocytes in culture. All RNA samples assayed contained intact
ribosomal RNA and little low molecular weight RNA as judged by ethidium bromide staining, indicating that little, if any, RNA was degraded (data not shown). Because the hybridization and wash conditions yielded a single m R N A species of appropriate size and specificity, we used them throughout this study. Total RNA from neonatal and adult rat cortex was also assessed for TGF-/31 mRNA by Northern analysis. A single m R N A of the size described for TGF-/31 m R N A (Derynck et al., 1985, 1986, 1988) was detected in neonatal (Fig. 1) and adult (data not shown) rat cortex. No other size RNA hybridized with this probe under these conditions. The probe hybridized to a 2.5 kb RNA in Jurkat T cells (Fig. 1) and did not hybridize to total liver RNA (data not shown) which is known to contain little, if any TGF-/31 m R N A (Derynck et al., 1985; Nakatsukasa et al., 1990). The presence of TGF-/31 m R N A in neonatal cortex and astrocytes derived therefrom indicate that astrocytes in vivo may constitutively express TGF-/31. Thus, TGF-/3 m R N A detected in murine astrocytes by polymerase chain reaction likely encodes TGF-/31 (Wesselingh et al., 1990). These results contrast with the inability to detect TGF/31 mRNA in adult murine brain by in situ hybridization (Wilcox and Derynck, 1988a). Differences in detection methods may be responsible, at least in part, for the discrepancy in these results. TGF-/3 irnmunoreactiuity is undetectable in astrocytes and cortex The presence of TGF-/31 m R N A in astrocytes and cortex suggested the presence of TGF-/3 protein. A procedure previously used to detect TGF/3 immunoreactivity in mouse embryo (Heine et al., 1987) was first attempted with Jurkat T cells (previously shown to constitutively secrete high amounts of TGF-/3 protein). Following exposure to anti-TGF-/3 antibody the majority of Jurkat cells stained positive. Moreover, sagittal sections of 15-day-old mouse embryo were stained with this antibody and the staining pattern was similar, if not identical to that previously described (Heine et al., 1987). In contrast, primary astrocytes exposed to anti-TGF-/3 antibody were rarely stained and the staining, when observed, was of low in-
Fig. 2. TGF-/31 immunoreactivity is not detectable in primary astrocytes nor in neonatal cortex. Photomicrographs of primary astrocytes stained with either anti-TGF-/3 ( A ) or antiG F A P (B) antibodies. Primary astrocytes were grown in culture for 12 days, fixed, incubated with the primary antibodies followed by biotinylated secondary antibody and avidinhorseradish peroxidase. Few, if any, cells were positive for TGF-fi ( A ) whereas nearly all were positive for G F A P (B). Magnification x 125. Photomicrograph of a representative area from neonatal cortex shows no staining with antibody against TGF-/3 (C). A coronal section of forebrain anterior to hippocampus was immunostained for TGF-/3 as described in Materials and methods and counterstained with fast red to reveal nuclei. Insert to C shows TGF-/3-1ike immunoreactivity detected in the meninges in this tissue section which served as an internal positive control for staining with anti-TGF-/3 antibody. The results are representative of two independent experiments. Magnification x 400.
tensity (Fig. 2). This indicates that TGF-/3 immunoreactivity was below detection or absent from astrocytes in vitro. We assessed immunocytochemically the presence of TGF-/3 protein in coronal sections of neonatal and adult rat cortex. No TGF-/3 immunoreactivity was detected in either neonatal (Fig. 2) or adult rat cortex (data not shown). Identical results were obtained when coronal sections were immunostained with and without hyaluronidase digestion. These results are consistent with the negative staining for TGF-/3 in at least adult mouse brain (Thompson et al.,
163 1989). In the neonatal rat brain arachnoid membrane and trabeculi of meninges were the only structures stained with anti-TGF-/3 antibody serving as an internal positive control (Fig. 2C, insert). These structures were also positive for TGF-/3 in 11- to 15-day-old mouse embryos (Heine et al., 1987). In adult rat brain even meninges were negative for TGF-/3 (data not shown), consistent with the results published on adult murine brain (Thompson et al., 1989). These findings suggest that TGF-/3 immunoreactivity in meninges persists after birth but decreases during postnatal development. The absence or non-detectability of TGF-/3 protein in cortex was consistent with the results in cultured astrocytes. A lack of correlation between TGF-/31 m R N A and protein has been observed in organs other than brain: gut, lung, kidney, and heart (Heine et al., 1987; Lehnert and Akhurst, 1988; Wilcox and Derynck, 1988a, b). In this respect astrocytes closely resemble monocytes which also express TGF-/31 m R N A but not detectable protein (Assoian et al., 1987). The results suggest that TGF-/31 mRNA is not translated in astrocytes and brain parenchyma. The results also indicate that TGF-/31 expression in astrocytes in vitro was similar to TGF-/3 expression in vivo suggesting that astrocytes in culture may have retained their original phenotype with respect to TGF-/31 expression.
TGF-[31 gene expression is regulated at the mRNA le~,el in developing astrocytes During murine development TGF-/3 protein expression is regulated in a cell- and tissuespecific manner (Heine et al., 1987; Thompson et al., 1989). For example, in brain tissue TGF-/3 was detected only in the meninges and pia mater of the 11- to 15-day-old embryo (Heine et al., 1987) but not in the adult (Thompson et al., 1989). Data presented here indicate that TGF-/3 immunoreactivity was present in the meninges of neonatal but not adult rat brain. Thus, TGF-/~ protein appears to be transiently expressed during rodent brain development only in structures derived from neural crest mesenchyme (Heine et al., 1987). In contrast to TGF-/3 protein, TGF-/3 m R N A appears to be constitutively expressed as shown above in both neonatal and adult rat cortex and in cultured astrocytes derived from
4 < Z
24 h) may induce TGF-/3 mRNA (Wesselingh et al., 1990) and protein.
Fig. 5. IL-1 induces TGF-/3 immunoreactivity in primary astrocytes. Photomicrograph of IL-l-stimulated astrocytes (250 units/ml; 6 h) stained with rabbit anti-TGF-/3 antibody followed by biotinylated sheep anti-rabbit IgG (A). A parallel culture was not exposed to TGF-/3 antibody but stained with biotinylated sheep anti-rabbit IgG (control; B). TGF-/3 immunoreactivity was detected as described in Materials and methods. Note positive staining of cells in A but not in B. Magnification X 250.
We measured TGF-/31 mRNA levels in astrocytes exposed to increasing concentrations of IL-1 for 6 and 24 h by slot blot. The astrocytes assayed were derived from the cultures found to contain elevated concentrations of extraceIlular TGF-/31 (Fig. 4). In contrast to the increase in TGF-/31 protein (Fig. 4), no change in the steady-state levels of TGF-/31 mRNA was detected with increasing doses of IL-1 for 6 (Fig. 6) and 24 h (data not shown). Thus, the expression of TGF-fll