Abbreviations used: BMP, bone morphogenetic protein; CNTF, ciliary neurotrophic factor; E, embryonic day; GFAP, glial fibrillary acidic protein; IL, interleukin; LIF, leukemia inhibitory factor; LIFR, LIF receptor; MAP2, microtubule-associated protein 2; MAPK, mitogen-activated protein kinase; MAPKK, MAPK kinase; OSM, oncostatin M; OSMR, OSM receptor; sIL-6R, soluble IL-6 receptor; STAT3, signal transducer and activator of transcription 3.
Address correspondence and reprint requests to Prof. T. Taga at Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, 2-3-10 Kanda-Surugadai, Chiyoda-ku Tokyo 101-0062, Japan. E-mail: email@example.com
Abstract: The interleukin (IL)-6 family cytokines utilize membrane glycoprotein gp130 in common as a critical signal-transducing receptor component. IL-11, a cytokine initially identified as a plasmacytoma growth factor, belongs to this family. We show here that IL-11 and its cognate receptor components are expressed in fetal mouse neuroepithelial cells. We also show that after 4 days of culture with IL-11, cells with typical astrocytic morphologies expressing glial fibrillary acidic protein (GFAP; a marker for astrocytes) come out. This differentiation process is totally dependent on the gp130-mediated signal-transduction pathway involving activation of a latent cytoplasmic transcription factor, STAT3 (for signal transducer and activator of transcription 3), because (a) IL-11-induced astrocyte differentiation is not observed when neuroepithelial cells prepared from gp130-deficient mice were used, (b) stimulation of neuroepithelial cells by IL-11 rapidly induces tyrosine-phosphorylation of STAT3, and (c) transfection of neuroepithelial cells with a dominant-negative form of STAT3 inhibits IL-11-induced activation of the GFAP gene promoter. We have further identified, in the GFAP promoter region, a STAT3 site at which nucleotide substitutions almost completely abolished the IL-11-induced GFAP promoter activation. Taken together, it is suggested that IL-11 contributes to astrocytogenesis in fetal brain via activation of gp130 and STAT3.
It has been proposed that neuronal and glial lineages arise from common precursors under the regulation of various mediators present in their environment within the developing brain (McKay, 1997). Mouse embryonic neuroepithelia are known to contain cells that have a potential to differentiate into both neurons and glial cells like astrocytes (Reynolds et al., 1992); Johe et al., 1996; Rajan and McKay, 1998). Ciliary neurotrophic factor (CNTF) and leukemia inhibitory factor (LIF) are known to induce differentiation of astrocytes in cultures of the embryonic neuroepithelial cells (Bonni et al., 1997; Nakashima et al., 1999a,b). These two cytokines belong to the interleukin (IL)-6-type cytokine family, which also includes IL-6, IL-11, oncostatin M (OSM), and cardiotrophin-1 (Taga, 1996). All six IL-6-type cytokines share a signal-transducing receptor component, gp130, in their respective receptor complexes, which clearly explains functional overlaps observed with these cytokines (Taga and Kishimoto, 1997). In addition to CNTF and LIF, IL-6 was also reported to induce astrocytes from cultured fetal brain cells, which is enhanced by the addition of soluble IL-6 receptor (sIL-6R) (Bonni et al., 1997).
In response to the IL-6-type cytokines, gp130 becomes dimerized either with itself or with another dimer partner like LIF receptor (LIFR) or OSM receptor (OSMR). IL-6 binds to IL-6 receptor, and the resulting complex is then associated with gp130, leading to the formation of its homodimers (Taga et al., 1989; Murakami et al., 1993). CNTF, LIF, OSM, and cardiotrophin-1 are known to form heterodimers of gp130 and LIFR, either via direct binding to them (e.g., LIF to LIFR and OSM to gp130) or in association with the cytokine-specific third component (e.g., CNTF with CNTF receptor) (Davis et al., 1993; for review, see Taga and Kishimoto, 1997). In addition to gp130/LIFR heterodimers, OSM has been demonstrated to signal through heterodimers composed of gp130 and OSMR (Tanaka et al., 1999). As for IL-11, it is still controversial whether it induces gp130 homodimers or heterodimers of gp130 and a so far unidentified membrane protein, although it is clear that gp130 is critical for signal transduction (Yin et al., 1993; Neddermann et al., 1996). Each IL-6-type cytokine leads to activation of a transcription factor, STAT3 (for signal transducer and activator of transcription 3), regardless of the type of gp130-containing receptor dimers that it induces (Kishimoto et al., 1994).
IL-11 was isolated initially for its biological activity to induce proliferation of plasmacytoma cells (Yang, 1993). This molecule was later found to have a wide variety of biological activities in, for instance, hematopoietic precursor cell regulation, bone metabolism, and hepatic acute-phase protein synthesis (Taga, 1996). Regarding its activity in the nervous system, IL-11 was reported to induce neuronal differentiation of a hippocampal cell line (Mehler et al., 1993), but to our knowledge this cytokine has not been reported to induce astrocyte differentiation. In the present study, we show that IL-11 and its receptor components are expressed in fetal mouse neuroepithelial cells, and that IL-11 induces astrocyte differentiation from these cells. We further demonstrate that the IL-11-induced astrocyte differentiation is entirely dependent on the signal-transducing receptor component, gp130, and its downstream transcription factor, STAT3.
MATERIALS AND METHODS
Timed-pregnant ICR mice were used to prepare fetal neuroepithelial cells. In this study, fetuses from intercrossings of heterozygous gp130 mutant mice (gp130 +/-) on the genetic background of ICR were also used (Yoshida et al., 1996; Kawasaki et al., 1997). Mice were treated according to the guidelines of the Tokyo Medical and Dental University Animal Committee.
Telencephalons from embryonic day 14.5 (E14.5) mice were triturated in Hanks' balanced salt solution by mild and frequent pipetting with 1-ml pipet tips (Gilson) as previously described (Nakashima et al., 1999a,b). Dissociated cells were cultured for 4 days in N2-supplemented Dulbecco's modified Eagle's medium/F-12 medium containing 10 ng/ml of basic fibroblast growth factor (R&D) (N2/DMEM/F-12/bFGF) on a 10-cm dish that had been precoated with poly-L-ornithine (Sigma) and fibronectin (GibcoBRL). Cells were then detached in Hanks' balanced salt solution and replated on chamber slides (Nunc) precoated as above, and cultured in the N2/DMEM/F-12/bFGF medium supplemented with various cytokines: IL-11 (kindly provided by Yamanouchi Pharmaceutical), LIF (GIBCO), CNTF (Genzyme), and IL-6/sIL-6R (kindly provided by Dr. K. Yasukawa, Tosoh Corp.).
Total RNAs were isolated from 4-day-cultured E14.5 neuroepithelial cells from normal and gp130-deficient littermates. cDNAs were synthesized with 5 μg of total RNAs as templates in 20 μl of reaction mixture using Superscript II reverse transcriptase (GibcoBRL). After reverse transcription, each reaction mixture was diluted fivefold with H2O, and 1 μl of each mixture was subjected to PCR using AmpliTaq Gold DNA polymerase (Perkin Elmer) for hot-start PCR with the following settings: 95°C for 9 min; 40 cycles of 94°C for 20 s, 60°C for 20 s, and 72°C for 30 s; and 72°C for 5 min. Used primer sets were as follows: 5′-TGCTGACAAGGCTTCGAGTAG-3′, 5′-CAGTCGAGTCTTTAACAACAGC-3′ (for IL-11); 5′-GCAATACCGACCAGCACAGC-3′, 5′-GGGATCATGGGTGCCAAGAG-3′ (for IL-11 receptor); 5′-TGTCAGCACCAAGGATTTGGC-3′, 5′-GTAGCTGACCATACATGAAGTG-3′ (for gp130).
Immunofluorescent staining was performed as previously described (Nakashima et al., 1999a,b). In brief, cells cultured on chamber slides were fixed in 4% paraformaldehyde in phosphate-buffered saline and stained with anti-glial fibrillary acidic protein (anti-GFAP) antibody (Dako)/rhodamine-conjugated second antibody (Chemicon) for detection of astrocytes, and anti-microtubule-associated protein 2 (anti-MAP2) antibody (Sigma)/fluorescein isothiocyanate-conjugated second antibody (Jackson ImmunoResearch) for detection of neurons. Bisbenzimide H33258 fluorochrome trihydrochloride (Nakarai) was used to stain nuclei.
Cells stimulated with IL-11 (100 ng/ml) were solubilized with NP-40 lysis buffer [0.5% NP-40, 10 mM Tris-Cl, pH 7.4, 150 mM NaCl, 3 mMp-amidinophenylmethanesulfonyl fluoride (Wako), 5 mg/ml aprotinin (Sigma), 2 mM sodium orthovanadate (Wako), 5 mM EDTA]. Lysates were immunoprecipitated with anti-STAT3 antibody (kindly provided by Dr. S. Akira, Osaka University). Precipitates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting with antibodies to phosphotyrosine (4G10, UBI) or STAT3 (Transduction Laboratories). Detection was done with the enhanced chemiluminescence system (Amersham).
GFAP promoter assay
Neuroepithelial cells cultured for 4 days on a 10-cm dish as described above were replated on 12-well plates (Nunc) and transfected on the next day with a plasmid containing firefly luciferase gene under the regulation of 2.5-kb GFAP promoter (GF1L-pGL3) or a modified construct (GF1L-SBSPM-pGL3; see Fig. 4b) (Nakashima et al., 1999b). Control transfection was made with sea pansy luciferase gene conjugated with human elongation factor 1α promoter pEF-RLuc (kindly provided by Drs. S. Nagata and K. Shimozaki, Osaka University). Transfection was done by using Trans-It LT1 (Pan Vera) according to the manufacturer's procedures. On the following day, cells were stimulated with cytokines for 8 h and solubilized, and luciferase activity was measured according to the recommended procedures for the Pikkagene Dual Luciferase Assay System (Toyo Ink Inc.) with some modification: trypsin inhibitor (1 mg/ml, type III-0 from chicken egg white; Sigma) was supplemented in cell lysis buffer. Luminous CT-9000D (Dia-Iatron) luminometer was used for detection.
Expression of IL-11 and its receptor components in fetal mouse neuroepithelial cells
We and others have shown that CNTF, LIF, and a combination of IL-6 and sIL-6R induce astrocyte differentiation from fetal neuroepithelial cells when cultured for, e.g., 4 days, via stimulation of a common signal-transducing receptor component, gp130 (Bonni et al., 1997; Nakashima et al., 1999a, b, c). In addition, we have observed previously the expression of LIF, IL-6, and their receptor components in fetal mouse neuroepithelial cells, suggesting that these gp130-stimulating cytokines contribute to astrocytogenesis in the developing brain (Nakashima et al., 1999c). In the present study, we examined the expression of IL-11, another gp130-stimulating cytokine initially isolated as a plasmacytoma growth factor, and its receptor components in fetal mouse neuroepithelial cells, as a first step to test possible involvement of IL-11 in astrocytogenesis in vivo. As shown in Fig. 1, transcripts for IL-11 and its receptor components, i.e., IL-11 receptor and gp130, were clearly detectable in fetal mouse neuroepithelial cells. The deficit of gp130 molecule had no effect on IL-11 and IL-11 receptor expression.
gp130-dependent astrocyte differentiation by IL-11
Based on the above results, we intended to know whether IL-11 has a potential to induce astrocytes. Neuroepithelial cells from E14.5 mouse telencephalon were cultured for 4 days with IL-11, LIF, CNTF, or a combination of IL-6 and sIL-6R. Differentiation of astrocytes was monitored by immunofluorescent detection of their marker protein, GFAP. As shown in Fig. 2 (a, c, e, g, and i), IL-11 did actually induce GFAP-positive astrocytes in cultured neuroepithelial cells to the extent comparable to that observed with LIF, CNTF, and a combination of IL-6 and sIL-6R. When neuroepithelial cells from gp130-deficient littermates were used for the same assay, none of these gp130-stimulating cytokines, including IL-11, induced astrocyte differentiation (Fig. 2b, d, f, h, and j). Astrocytic and neuronal phenotypes of normal neuroepithelial cells cultured with gp130-stimulating cytokines were examined by immunofluorescent staining and are summarized in Table 1. gp130-stimulating cytokines did not appear to affect neuronal differentiation as assessed by MAP2 expression.
Table 1. Astrocytic and neuronal phenotype of neuroepithelial cells cultured with IL-11 and other gp130-stimulating cytokines
No. of cells counted
Neuroepithelial cells were prepared and cultured as in Fig. 2 and stained for astrocytic and neuronal markers, GFAP and MAP2, respectively. The total cell numbers were assessed by H33258 staining.
Activation of the gp130-signaling pathway in neuroepithelial cells by IL-11
We then attempted to confirm that IL-11 actually activates the gp130-mediated signal-transduction pathway in neuroepithelial cells. It has been shown that tyrosine-phosphorylation of STAT3 rapidly occurs upon stimulation of cells with gp130-stimulating cytokines (Kishimoto et al., 1994). Thus, we examined whether neuroepithelial cells stimulated by IL-11 undergo this tyrosine-phosphorylation event. Lysates from unstimulated and IL-11-stimulated cells were subjected to immunoprecipitation with anti-STAT3 antibody and subsequent immunoblotting with anti-phosphotyrosine antibody. As shown in Fig. 2k, STAT3 protein was clearly phosphorylated on tyrosine residue in response to IL-11. In contrast, no tyrosine-phosphorylation of STAT3 was observed in cells prepared from gp130-deficient mice.
Activation of the GFAP gene promoter by IL-11 and related cytokines
Previous studies have shown that stimulation of gp130 molecules on neuroepithelial cells by LIF, CNTF, or the IL-6/sIL-6R complex induces activation of the GFAP gene promoter (Bonni et al., 1997; Nakashima et al., 1999b). We thus examined whether IL-11 could induce GFAP promoter activation by using a reporter construct, GF1L-pGL3, that is composed of a 2.5-kb GFAP promoter and firefly luciferase gene. As shown in Fig. 3, IL-11 activated the GFAP promoter in a dose-dependent manner. LIF and CNTF induced the GFAP promoter activation at much lower concentrations, and their effect was already saturable at 8 ng/ml. Although the extent of the IL-6-induced GFAP promoter activation was small, it was enhanced by the addition of sIL-6R. The level of maximum promoter activation by each cytokine was roughly similar among the cytokines examined.
STAT3-mediated activation of the GFAP gene promoter by IL-11
The phosphorylation on Tyr705 in response to gp130-stimulating cytokines has been reported to be important for the formation of STAT3 dimers and critical for gp130-mediated signal transduction. Thus, Tyr705 to phenylalanine substitution in the STAT3 molecule completely abolishes its signaling capability (Minami et al., 1996). This mutated form of STAT3 (STAT3Y705F), when overexpressed in cells that are normally responsive to gp130-stimulating cytokines, makes the cells unresponsive to them, behaving as a dominant-negative molecule against endogenous STAT3. Stimulation of gp130 by IL-11 is known to activate a latent cytoplasmic transcription factor STAT3 (Yang, 1993). Involvement of STAT3 in IL-11-induced transcriptional activation of the GFAP gene was examined by using the above-mentioned promoter construct (GF1L-pGL3) in combination with cDNA for the dominant-negative form of STAT3. Cotransfection of the STAT3Y705F cDNA with the GF1L-pGL3 reporter into neuroepithelial cells dramatically impeded the IL-11-induced activation of the GFAP promoter (Fig. 4a).
Identification of a STAT3 site in the GFAP promoter critical for IL-11-induced activation
The above results clearly indicated that STAT3 is a prerequisite for IL-11-induced activation of the GFAP promoter. We next examined which part of the GFAP promoter region is important for the transcriptional activation of the GFAP gene by IL-11 in mouse neuroepithelial cells. In the 2.5-kb mouse GFAP promoter region (see GF1L-pGL3 in Fig. 4b), there exist some potential STAT3 binding sites. Among them, the TTCCGAGAA sequence (indicated in the figure) is 100% conserved between mouse and rat GFAP promoters at a comparable position (Nakashima et al., 1999b). This sequence in the rat GFAP promoter was reported previously to be important for its activation induced by LIF, CNTF, and the IL-6/sIL-6R complex (Bonni et al., 1997). As shown in Fig. 4b, when nucleotide substitutions were introduced in it (GF1L-SBSPM-pGL3), IL-11-induced GFAP-promoter activation disappeared almost completely. Taken together, the TTCCGAGAA sequence located ∼1.5 kb upstream of the transcriptional start site is critical for IL-11-induced activation of GFAP gene transcription.
The IL-6-type cytokines that share a signal-transducing receptor component gp130 are known to signal through either homodimers of gp130 or heterodimers composed of gp130 and a dimer partner such as LIFR (Taga and Kishimoto, 1997). For instance, IL-6 triggers gp130/gp130 homodimers, and LIF and CNTF induce gp130/LIFR heterodimers. Although it is still controversial whether IL-11 signals through gp130 homodimers or heterodimers composed of gp130 and a so far unidentified component (Yin et al., 1993; Neddermann et al., 1996), the present study has clearly demonstrated that IL-11 has a potential to induce astrocyte differentiation via gp130 and a downstream transcription factor STAT3, as has been suggested for LIF, CNTF, and the IL-6/sIL-6R complex (Bonni et al., 1997; Nakashima et al., 1999b,c). Taken together with our recent results that OSM also induces astrocyte differentiation (Yanagisawa et al., 1999), it is generally the case that stimulation of gp130 triggers cytoplasmic signaling that leads to astrocyte differentiation, regardless of the formation of gp130/gp130 homodimers, gp130/LIFR heterodimers, or gp130/OSMR heterodimers.
It has been shown that stimulation of cells with the IL-6 family of cytokines induces rapid tyrosine-phosphorylation of gp130. Activation of a latent cytoplasmic transcription factor, STAT3, and activation of a mitogen-activated protein kinase (MAPK) are the two major downstream pathways triggered by stimulation of gp130. Of two pathways, the STAT3 activation has been suggested to play an important role in astrocyte differentiation induced by LIF, CNTF, IL-6/sIL-6R, and OSM (Rajan et al., 1996; Bonni et al., 1997; Nakashima et al., 1999b; Yanagisawa et al., 1999). In these studies, differentiation of astrocytes was monitored by expression of their typical marker, GFAP. Our current study has provided evidence that the GFAP gene expression is mediated directly by STAT3 following IL-11 stimulation (Fig. 4), by showing complete inhibition of IL-11-induced activation of the GFAP promoter by a dominant-negative form of STAT3. This STAT3-mediated promoter activation depends primarily on the presence of one particular STAT3 binding sequence, which is evidenced by the result that nucleotide substitutions in this sequence completely abolish IL-11-mediated activation of the GFAP promoter. As for a role of the MAPK pathway in gp130-mediated astrocyte differentiation, two conflicting results have been reported. Rajan and McKay (1998) reported down- and up-regulation of CNTF-induced astrocyte differentiation by treatment with MAPK kinase (MAPKK) inhibitor, PD98059, and overexpression of constitutively active form of MAPKK, respectively, suggesting a positive contribution of MAPK. In contrast, Bonni et al. (1997) reported enhanced GFAP promoter activation by expression of a dominant-negative form of MAPKK in CNTF-stimulated cells, implying a negative regulatory role of MAPK. Although we have not examined the role of MAPK in our culture system, this issue remains to be clarified.
Previous studies by other groups and our present work suggest that loss of signals from gp130-stimulating cytokines may lead to a defect in astrocyte development in vivo for several reasons: (a) At least five of six gp130-stimulating cytokines are now known to have a potential to induce astrocytes in cultured fetal neuroepithelial cells. (b) When the brain in the gp130-deficient mice was histologically examined on E18.5, the number of GFAP-positive cells and the intensity of GFAP expression were reduced dramatically (Nakashima et al., 1999a). (c) Perinatal mice deficient for LIFR also exhibit significant reduction of astrocytes as histologically examined for GFAP expression (Ware et al., 1995; Bartlett et al., 1998). We have demonstrated recently that the gp130 signaling synergizes with the bone morphogenetic protein (BMP) 2 signaling in the induction of astrocytes from fetal neuroepithelial cells via the formation of STAT3-Smad1 complex bridged by a transcriptional coactivator p300 (Nakashima et al., 1999b). In that report, LIF and BMP2 act in synergy to induce GFAP-positive astrocytes in the 2-day-cultured neuroepithelial cells. In the present study using 4-day cultures, BMP2 or equivalent cytokine might be expressed and accumulated, which may cooperatively induce astrocytes with exogenously added IL-11. In fact, BMP2, BMP4, and their receptor components are expressed in fetal neuroepithelial cells either in vivo or in cultures (Nakashima et al., 1999c).
It has been suggested that there exist common neural precursor cells that have a potential to differentiate into both neurons and astrocytes (Gage et al., 1995; McKay, 1997). In our culture system, stimulation of fetal neuro-epithelial cells by gp130-stimulating cytokines does not appear to affect neuronal differentiation, whereas it does induce astrocytes. An interesting notion is that mice deficient for gp130 show a loss of motor neurons in the facial nucleus, nucleus ambiguus, and spinal motor column and sensory neurons in dorsal root ganglia at the late stage of development (E18.5), whereas they show normal neuron numbers and shapes at the earlier stage (E14.5) (Nakashima et al., 1999a). This is also true for mice deficient for LIFR (Li et al., 1995) or JAK1 (Rodig et al., 1998), a kinase that is considered critical for gp130-mediated signaling. It is thus suggested that signals from gp130 may act on already differentiated neurons to support their survival. This is in contrast to a role of gp130 in gliogenesis proposed by the present work and a previous report (Bonni et al., 1997) in which the gp130 signals are suggested to promote astrocyte differentiation at the neural precursor level. As astrocytes have been suggested to play important roles in the survival of neurons and maintenance of neuronal functions, as well as in the regeneration of injured neurons, IL-11 and related cytokines may play important roles in the regulation of the brain by regulating astrocyte development, neuronal survival, and nerve regeneration.
We are very much grateful to Ms. Yuko Nakamura for her excellent secretarial assistance. We also thank Ms. Kyoko Saito and Mr. Ryotaro Watabe for technical help. This work was supported by a Grant-in-Aid from the Ministry of Education, Science, Sports and Culture, Human Frontier Science Program, and Cell Fate Modulation Research Unit of Medical Research Institute, Tokyo Medical and Dental University.