Epidermal-growth-factor-induced proliferation of astrocytes requires Egr transcription factors

In the nervous system, a key feature of astrocyte reactivity is their proliferative response. Many types of insults to the central nervous system induce reactive gliosis, an astrocyte response that includes an increase in cell number, an extension of cellular processes, and an upregulation of glial fibrillary acidic protein (GFAP). Tissue culture studies showed that mitogens such as epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) stimulate DNA synthesis and proliferation in primary astrocytes and glioma cells (Kaufmann and Thiel, 2001; Riboni et al., 2001). Mitogenic stimulation of astrocytes with EGF and other mitogens leads to an activation of extracellular signal-regulated protein kinase (ERK, also known as MAPK) (Tournier et al., 1994; Kaufmann et al., 2001; Wang et al., 2002). The phosphorylated form of ERK translocates to the nucleus and induces a change in the genetic expression pattern of the cells, which is required for the proliferative response. One of the transcription factors that is synthesized as a result of ERK activation is the zinc finger protein Egr-1 (Thiel and Cibelli, 2002; Rössler et al., 2006). Originally, the gene expressing Egr-1 was discovered in a search for genes induced by growth factors (Lim et al., 1987; Christy et al., 1988; Lemaire et al., 1988; Sukhatme et al., 1988). Egr-1 preferentially binds to the GC-rich sequence 5′-GCGG/TGGGCG-3′ (Christy and Nathans, 1989; Cao et al., 1993; Nakagama et al., 1995), with its DNA-binding domain comprised of three zinc finger motifs. The N-terminus of Egr-1 functions as an extended transcriptional activation domain (Thiel et al., 2000). In addition, an inhibitory domain between the activation domain and the DNA-binding domain has been identified that functions as a binding site for two transcriptional co-repressors, NAB1 and NAB2. Both NAB1 and NAB2 block the biological activity of Egr-1 (Russo et al., 1993; Svaren et al., 1996; Thiel et al., 2000). The fact that Egr-1 and NAB2 expression is induced by the same stimuli, although on different time scales, suggests that expression of NAB2 functions as a negative feedback loop to reduce Egr-1 activity (Svaren et al., 1996).

A previous study using established human glioma cell lines revealed that EGF stimulation rapidly activated ERK, induced the biosynthesis of Egr-1, and triggered cell growth (Kaufmann and Thiel, 2001). In this study, it was concluded that Egr-1 may be an important `late' component of the EGF-initiated signaling cascades and may function as a `third messenger' connecting growth-factor stimulation with changes in gene transcription. Thus, induction of Egr-1 expression may be an integral part of the mitogenic pathway and continues the mitogenic signaling cascade via stimulation of the synthesis of growth factors or growth factor receptors. However, the proposed role for Egr-1 in controlling cell growth is largely based on the correlation between mitogenic responses and the induction of Egr-1 biosynthesis by mitogens. In addition, there are reports to the contrary, attributing to Egr-1 a growth inhibitory function in tumor cells, including gliomas (Huang et al., 1995; Calogero et al., 2001; Calogero et al., 2004).

The objective of this study was to show whether a causality exists between proliferation of astrocytes and the presence of Egr-1 transcriptional activity. The results show that Egr proteins are required to connect mitogenic stimulation with enhanced proliferation. In Egr-1-deficient mice, other Egr proteins compensate for the loss of Egr-1. Moreover, we show that nuclear ERK and ternary complex factors are essential for stimulating proliferation in EGF-treated astrocytes.

Stimulation of the EGF receptor (a receptor tyrosine kinase) by its cognate ligand EGF activates the ERK signaling cascade. Using a phospho-specific antibody we showed that ERK2 is phosphorylated in astrocytes as a result of EGF stimulation (Fig. 1A). The activation of ERK by EGF was transient. Within 15 minutes of stimulation, phospho-ERK could be detected. ERK was mostly dephosphorylated and inactivated 45 minutes after stimulation.

To evaluate astrocyte proliferation, we measured the incorporation of 5-bromo-2′-deoxyuridine (BrdU) into DNA as an indicator of DNA synthesis. Astrocytes were serum-starved for 24 hours and then treated with EGF for 24 hours. Fig. 1B shows that EGF stimulation of astrocytes induced a significant increase in BrdU incorporation (as estimated by an immunoassay) compared with unstimulated control cells, indicating that EGF functions as a mitogen for these cells. To elucidate the role of ERK in EGF-induced cell proliferation, we pre-incubated primary astrocytes with PD98059, a compound that inhibits the phosphorylation of the mitogen-activated protein kinase kinase (MAPK kinase, also known as MEK), before stimulation of the cells with EGF. Fig. 1B shows that PD98059 efficiently blocked the incorporation of BrdU into the DNA of EGF-treated astrocytes. Likewise, expression of DA-Raf1, a splicing form of A-Raf that functions as an antagonist of the Ras-Raf-ERK signaling pathway, abrogated EGF-induced proliferation of astrocytes following stimulation with EGF (Fig. 1C-E). The phosphorylated and activated ERK translocates into the nucleus and changes the transcriptional program by phosphorylating transcriptional regulatory proteins. In the nucleus, MKP-1 is part of a negative feedback loop leading to the inactivation of ERK via dephosphorylation. Having shown that ERK activation is a key step in the mitogenic signaling cascade leading to proliferation of astrocytes, we tested whether overexpression of MKP-1 counteracts the EGF-induced proliferation of astrocytes. Fig. 1F reveals that expression of MKP-1 completely blocked the proliferative response in EGF-treated astrocytes. Moreover, the incorporation of BrdU into the DNA was even reduced in comparison to unstimulated cells, suggesting that MKP-1 dephosphorylated residual phosphorylated ERK in the nucleus. As a negative control, we expressed the protein phosphatase 2C (PP2C) in astrocytes. Fig. 1G shows that PP2C did not play any role in the mitogenic signaling cascade induced by EGF.

Activation of the ERK signaling cascade is the major stimulus for induction of the biosynthesis of Egr-1 in many cellular systems (Thiel and Cibelli, 2002; Rössler et al., 2006). Primary astrocytes were serum-starved for 24 hours and then stimulated with EGF. The cells were harvested, nuclear extracts prepared and Egr-1 expression analyzed by immunoblotting. Egr-1 immunoreactivity was almost undetectable in unstimulated cells. EGF stimulation significantly increased its levels (Fig. 2A) with peak expression of Egr-1 occurring after 2 hours of stimulation. Preincubation of the cells with PD98059 blocked the biosynthesis of Egr-1 in EGF-stimulated astrocytes (Fig. 2B), indicating that ERK activation is a key event in the signaling cascades that connect EGF receptor stimulation with Egr-1 gene transcription. Thus, a correlation between EGF-induced proliferation and the biosynthesis of Egr-1 is observed in astrocytes.

The 5′-flanking region of the gene expressing Egr-1 contains five serum response elements (SREs) encompassing the consensus sequence CC[A/T]₆GG, also known as the CArG box. These motifs are responsible for the induction of Egr-1 gene transcription by various extracellular signaling molecules (Thiel and Cibelli, 2002; Rössler et al., 2006). The SREs occur in two clusters in the Egr-1 promoter, a distal 5′ cluster of three SREs and a proximal 3′ cluster of two SREs. In addition, multiple binding sites for ternary complex factors (TCFs) are adjacent to the CArG boxes having the TCF consensus core sequence GGAA/T (Bauer et al., 2005).

To test the involvement of the proximal and distal SRE clusters of the Egr-1 promoter in mediating the responsiveness of the Egr-1 gene to EGF, we inserted two Egr-1 promoter/luciferase reporter genes into the chromatin of astrocytes using lentiviral gene transfer. The transfer vectors pFWEgr-1.1luc and pFWEgr-1.2luc, used to generate recombinant lentiviruses, encode Egr-1 promoter/luciferase reporter genes that contained 239 or 490 nucleotides of the human EGR-1 gene 5′-upstream region, respectively, together with 235 nucleotides of the 5′-nontranslated region. The transfer vector pFWEgr-1SREluc encodes the luciferase reporter gene under the control of the proximal SREs #1 and #2 of the Egr-1 promoter. Fig. 2C shows a schematic representation the integrated proviruses encoding Egr-1 promoter/luciferase reporter genes. Astrocytes were infected with recombinant lentiviruses and stimulated with vehicle or EGF. The addition of EGF significantly induced reporter gene transcription controlled by either 490 or 239 nucleotides of the human EGR-1 promoter (Fig. 2D). The fact that the reporter genes #1 and #2 directed almost similar luciferase expression as a result of EGF stimulation indicates that the proximal, and not the distal SRE cluster, was important for the upregulation of Egr-1 transcription in EGF-stimulated cells.

This conclusion was confirmed in an experiment in which we directly measured the impact of the proximal SREs of the Egr-1 promoter on EGF-induced reporter gene transcription. Astrocytes were infected with a lentivirus encoding the luciferase gene under the control of the two proximal SREs of the Egr-1 promoter upstream of a minimal promoter (Fig. 2D). The transcriptional activity measured was no different to that of a reporter gene controlled by Egr-1 promoter sequences from –239 to +235. Thus, the most proximal SREs of the Egr-1 promoter are sufficient to transduce EGF signaling to the Egr-1 gene.

Given the importance of SREs within the Egr-1 promoter, we next assessed the impact of TCF activation on the regulation of proliferation of astrocytes. To overcome the problem associated with redundancy of functions between the TCFs, we expressed a dominant-negative mutant of the TCF Elk-1, termed REST/Elk-1ΔC (Fig. 3A). This mutant retains the DNA-binding and SRF interaction domains, but lacks the C-terminal activation domain of Elk-1. REST/Elk-1ΔC additionally contains the N-terminal repression domain of the transcriptional repressor REST (Thiel et al., 1998), a FLAG epitope for immunological detection and a nuclear localization signal (NLS). Nuclear proteins of mock-infected astrocytes or astrocytes infected with a REST/Elk-1ΔC encoding lentivirus were fractionated by SDS-PAGE. The fusion protein was identified by western blot analysis using antibodies targeting the FLAG epitope (Fig. 3B). Fig. 3C shows that expression of REST/Elk-1ΔC prevented the proliferation of astrocytes that had been stimulated with EGF. We conclude that TCF activation is essential for conversion of mitogenic stimulation by EGF into a proliferative response.

In addition to five SREs, the Egr-1 promoter contains a cyclic AMP response element (CRE) encompassing the sequence 5′-TCACGTCA-3′. It had been suggested that the CRE may control Egr-1 gene transcription via activating transcription factor 2 (ATF2) activation by p38 stress-activated protein kinases (Rolli et al., 1999). To assess the role of ATF2 in the regulation of proliferation in astrocytes, we expressed a dominant-negative mutant of ATF2 termed ATF2ΔN that lacked the N-terminal regulatory region and the transcriptional activation domain. The modular structure of this mutant is depicted in Fig. 3D. The biological activity of this mutant has been demonstrated (Steinmüller and Thiel, 2003; Mayer et al., 2008). Fig. 3E shows that the dominant-negative mutant ATF2ΔN was synthesized in astrocytes following infection with recombinant lentiviruses. Fig. 3F reveals that expression of ATF2ΔN did not impair the proliferative response of astrocytes after stimulation with EGF.

So far, we have shown that Raf, ERK and TCF activation are essential for the mitogenic responses in astrocytes. ERK and TCF activation are additionally required in many cellular systems for induction of the biosynthesis of Egr-1 (Thiel and Cibelli, 2002; Bauer et al., 2005). Thus, we can correlate the mitogenic responses and the induction of Egr-1 biosynthesis by mitogens. We prepared astrocytes from wild-type and Egr-1-deficient mice in order to investigate whether a causality exists between proliferation of astrocytes and the presence of Egr-1 transcriptional activity. Fig. 4A reveals that there were no differences between the EGF-induced growth rates of astrocytes derived from either wild-type or Egr-1-deficient mice.

bFGF is a mitogen for astrocytes, and elevated bFGF levels have been detected in human glioma cells (Morrison, 1991; Holland and Varmus, 1998; Riboni et al., 2001). Interestingly, the human bFGF gene is transactivated by Egr-1 (Wang et al., 1997; Biesiada et al., 1998). In the murine bFGF gene, an Egr-1 binding motif was detected in the 5′-untranslated region (supplementary material Fig. S1). Thus, Egr-1 may regulate growth of astrocytes via transactivation of the gene encoding bFGF and the subsequent synthesis of bFGF. We therefore assessed the bFGF levels in EGF-stimulated astrocytes derived from wild-type and Egr-1 knockout mice. Fig. 4B,C shows that EGF treatment induced the expression of bFGF, regardless of the presence or absence of Egr-1 in the cells.

The experiments performed with astrocytes prepared from Egr-1-deficient mice suggests that either Egr-1 plays no role in the control of astrocyte proliferation, or that other members of the Egr family compensate for the loss of Egr-1. Before addressing this problem, we confirmed the importance of Raf, nuclear ERK and TCF activation in this system. We prepared astrocytes from Egr-1-deficient mice and infected the cells with lentiviruses encoding either DA-Raf1 (Fig. 5A), MKP-1 (Fig. 5B), or REST/Elk-1ΔC (Fig. 5D). The results show that proliferation of astrocytes was completely blocked after stimulation with EGF. Likewise, expression of PP2C or ATF2ΔN did not change the proliferative response of EGF-treated astrocytes (Fig. 5C,E). These data indicate that activation of ERK, translocation of ERK into the nucleus, and subsequent activation of TCF are key events in the regulation of astrocyte proliferation.

To directly address the problem of compensation between the members of the Egr family, we expressed a dominant-negative mutant of Egr-1, Egr-1/Zn, that encodes the zinc finger DNA-binding domain of Egr-1 (Fig. 6A). In comparison with Egr-1, the mutant lacks the N-terminal transcriptional activation domain and the binding site for the transcriptional corepressors NAB1 and NAB2. Egr-1/Zn additionally contains a FLAG epitope for immunological detection and an NLS. Nuclear proteins of mock-infected astrocytes or astrocytes infected with an Egr–1/Zn-encoding lentivirus were fractionated by SDS-PAGE. The protein was identified by western blot analysis using antibodies targeting the FLAG epitope. Fig. 6B shows that the mutant protein was synthesized as expected. Fig. 6C,D reveal that expression of Egr-1/Zn blocked the proliferation of EGF-stimulated astrocytes, derived either from wild-type (Fig. 6C) or Egr-1 knockout mice (Fig. 6D). These data indicate that the Egr proteins control proliferation of astrocytes following stimulation of the cells with EGF.

The dominant-negative Egr-1/Zn mutant inhibits DNA binding of wild-type Egr proteins by blocking the cognate GC-rich binding sites for DNA binding. Thus, an impairment of DNA binding should reduce or abolish the biological activity of the mutant. Therefore, we mutated crucial cysteine residues at positions 368 and 396 of zinc fingers 2 and 3, respectively, to serine residues (Fig. 7A). Similar mutations, introduced at the zinc finger domain of Sp1, abolished DNA binding (Lee et al., 2005). Electrophoretic mobility shift assays showed that the mutated Egr-1/Zn protein did not bind anymore to cognate Egr-1 binding motifs, derived from either the Egr-1 or the synapsin I promoter (Fig. 7B). Egr-1/Zn and the mutated version of Egr-1/Zn were synthesized in astrocytes following infection with recombinant lentiviruses (Fig. 7C). An analysis of EGF-induced proliferation revealed that these mutations of the DNA-binding domain reduced or abolished the activity of Egr-1/Zn to suppress the mitogenic signaling cascade in astrocytes (Fig. 7D,E). These data clearly show that transcriptional activation of Egr target genes is essential to couple EGF stimulation with enhanced proliferation of the cells.

To confirm the previous results, we expressed a FLAG-tagged NAB2 protein in astrocytes via lentiviral gene transfer. NAB2 is a transcriptional co-repressor that does not bind to DNA, but interacts with Egr-1, Egr-2 and Egr-3, leading to impairment of the transcriptional activity of the Egr transcription factors. We analyzed nuclear proteins derived from mock-infected astrocytes or astrocytes infected with a FLAG–NAB2-encoding lentivirus by SDS-PAGE. The protein was identified by western blot analysis using antibodies targeting the FLAG epitope. Fig. 8A shows that FLAG-NAB2 was synthesized as expected. Fig. 8B,C show that expression of NAB2 blocked the proliferation of EGF-stimulated astrocytes, derived either from wild-type (Fig. 8B) or Egr-1 knockout mice (Fig. 8C). We conclude that Egr proteins are essential for transforming the mitogenic signal into a proliferative response. Moreover, Egr target genes need to be activated for continuation of the mitogenic signaling cascade.

The previous results suggest that other Egr proteins compensate for the loss of Egr-1 in Egr-1-deficient mice. To test this assumption we analyzed the stimulus-induced binding of Egr proteins to the gene expressing bFGF. Fig. 9A shows a schematic representation of the 5′-upstream region of the murine bFGF gene, including the Egr-1 binding site in the 5′-untranslated region and the location of the primers used for PCR. The bFGF gene is in an open configuration in astrocytes, as shown by the fact that it is embedded into a nucleosomal context with histone H3 molecules carrying trimethylated lysine 4 (Fig. 9B), which is a marker of actively transcribed genes. Next, cross-linked and sheared chromatin, prepared from unstimulated astrocytes and astrocytes stimulated with EGF, was immunoprecipitated with antibodies directed against either Egr-1, Egr-2 or Egr-3. Fig. 9C (upper panel) shows that Egr-1, Egr-2 and Egr-3 bound to the regulatory region of the bFGF gene, but only when the cells had been stimulated with EGF. No binding of Egr proteins was detected to a distal region of the bFGF gene (Fig. 9C, lower panel). Naturally, no Egr-1 binding to the bFGF gene could be detected in astrocytes prepared from Egr-1-deficient mice. However, binding of Egr-2 and Egr-3 to the bFGF gene was similar to that in astrocytes prepared from wild-type mice (Fig. 9D).

The results obtained in expression experiments involving Egr-1/Zn and Egr-1/ZnC368SC396S suggested that Egr-1/Zn binds to Egr target genes and competes with wild-type Egr proteins for DNA binding. Chromatin immunoprecipitation (ChIP) experiments confirmed that Egr-1/Zn bound to the bFGF gene under physiological conditions (Fig. 9E, upper panel). In addition, ChIP experiments were performed with astrocytes that had been infected with a lentivirus encoding Egr-1/Zn and stimulated with EGF. Fig. 9E (lower panel) shows that expression of Egr-1/Zn prevented binding of Egr-1, Egr-2 and Egr-3 to the bFGF gene. Thus, the Egr-1 mutant binds to the 5′-untranslated region of the bFGF gene and inhibits DNA binding of wild-type Egr proteins.

Since the discovery of the Egr-1 gene as an `early growth response gene' (Sukhatme et al., 1988), research has been directed towards a function of Egr-1 in growth and proliferation. In fact, induction of Egr-1 gene transcription was monitored in many cell types in response to mitogens, and a direct role of Egr-1 in controlling proliferation has been proposed for astrocytes, glioma cells, glomerular mesangial cells, keratinocytes and T-cells (Perez-Castillo et al., 1993; Biesiada et al., 1996; Hofer et al., 1996; Kaufmann and Thiel, 2002; Rössler and Thiel, 2004). Egr-1 biosynthesis is strongly stimulated by activation of ERK (Kaufmann et al., 2001), the protein kinase that is activated by mitogens. The fact that genes encoding growth factors such as insulin-like growth factor II, platelet-derived growth factors A and B, and transforming growth factor β1 have been identified as target genes of Egr-1 (Khachigian et al., 1995; Svaren et al., 2000) suggests that Egr-1 continues the mitogenic signaling cascade via stimulation of growth factor synthesis. However, it is necessary to emphasize that the proposed role for Egr-1 in controlling cell growth is largely based on the correlation between mitogenic responses and the induction of Egr-1 biosynthesis by mitogens. The use of cells derived from Egr-1-deficient mice has helped to demonstrate a causal relationship between Egr-1 expression and cellular growth in a few cases. Delayed hepatocellular mitotic progression and impaired liver regeneration, as well as impairment of tumor progression of the prostate, has been observed (Abdulkadir et al., 2001; Liao et al., 2004). The objective of this study was to show whether a causality exists between proliferation of astrocytes and the presence of Egr-1 transcriptional activity.

The analysis of astrocytes derived from wild-type and Egr-1-deficient mice revealed that there were no differences in the proliferation of the cells following EGF stimulation. We hypothesized that other members of the Egr family, which all bind to a similar GC-rich target sequence, compensated for the loss of Egr-1. To overcome the problem associated with redundancy of functions between the Egr proteins, we chose a dominant-negative approach. Expression of a dominant-negative mutant of Egr-1 that blocked the cognate binding sites of all Egr proteins abrogated proliferation of astrocytes following EGF stimulation, indicating that Egr activity is required to transform a mitogenic signal into a proliferative response. Furthermore, we showed that DNA binding of the Egr-1/Zn mutant is essential for interference with the mitogenic signaling in astrocytes, indicating that transcriptional upregulation of Egr target genes is essential for connecting EGF stimulation with proliferation of the cells. Likewise, proliferation of astrocytes was prevented in cells overexpressing NAB2. This cofactor blocks transcription mediated by Egr-1, Egr-2 and Egr-3, all of which contain binding sites for NAB2 (O'Donovan et al., 1999). Together, these data establish a mechanistic link between Egr activity and proliferation of astrocytes and indicate that loss of Egr-1 in astrocytes prepared from Egr-1 knockout mice is compensated by other Egr proteins. This view was corroborated by ChIP experiments. In Egr-1-deficient astrocytes, Egr-2 and Egr-3 still bound to the bFGF gene when the cells had been stimulated with EGF, thus explaining the increased levels of bFGF mRNA following EGF stimulation in the absence of Egr-1 expression. We conclude that the proliferation of astrocytes following stimulation with EGF is regulated by Egr transcription factors. The bFGF gene was analyzed as a known Egr-1 target gene and as a mitogen for astrocytes (supplementary material Fig. S2). Pharmacological inhibition of the bFGF receptor revealed that bFGF is not an essential intermediate for EGF-induced cell proliferation (supplementary material Fig. S2). Thus, the identification of Egr target genes encoding growth regulatory proteins will be the aim of future studies.

The analysis of Egr-1 promoter/luciferase reporter genes revealed that the most proximal SREs of the Egr-1 promoter are sufficient for transduction of EGF signaling to the Egr-1 gene. Transcriptional activation of Egr-1 is often preceded by an activation of Elk-1, a TCF. Phosphorylation of Elk-1 connects the ERK signaling cascade with SRE-mediated transcription. Elk-1 and other TCFs contact DNA and the serum response factor (SRF) to exhibit biological activity. Genetic inactivation of Elk-1 or other TCFs in transgenic mice revealed minimal changes of the phenotype (Ayadi et al., 2001; Cesari et al., 2004; Costello et al., 2004), suggesting that functional redundancy may exist. Therefore, we have assessed the necessity of TCF activation for EGF-induced proliferation of astrocytes by using a dominant-negative version of Elk-1 in loss-of-function experiments. Due to the binding to DNA and SRF, the Elk-1 mutant REST/Elk-1ΔC most probably also inhibits the activity of two other TCFs, SAP-1 and SAP-2. Expression of this Elk-1 mutant revealed that TCF activation is essential for the connection of mitogen stimulation and proliferation of astrocytes. This observation is in accordance with a previous study showing that TCF activation is required for proliferation of fibroblasts (Vickers et al., 2004).

Finally, we have shown that ERK activation is a key step in the mitogenic signaling cascade leading to proliferation of astrocytes. Expression of DA-Raf1, an antagonist of the Ras-Raf-ERK signaling pathway, or incubation of the cells with the compound PD98059 interfered with the mitogenic signaling cascade. Similary, overexpression of MAPK phosphatase 1 (MKP-1; an enzyme that catalyzes the dephosphorylation and inactivation of ERK in the nucleus) blocked cell growth in EGF-stimulated astrocytes, indicating that ERK needs to translocate to the nucleus in order to stimulate cell proliferation. Thus, MKP-1 functions as a nuclear shut-off device that interrupts the mitogenic signaling cascade in the nucleus.

Transgenic mice containing an inactivated Egr-1 gene have been described (Topilko et al., 1997). These mice have an insertion of a LacZ-neo cassette between the promoter and the coding region that inactivates the Egr-1 gene. Moreover, an additional frame-shift mutation has been introduced at the beginning of the DNA-binding region. Homozygous Egr-1^–/– mice were produced by mating of heterozygous Egr-1^+/– mice.

Primary astrocyte cultures were prepared from the forebrain of 1- to 2-day-old neonatal mice. Excised brains were placed in buffer containing 137 mM NaCl, 5.4 mM KCl, 197 μM Na₂HPO₄, 35 μM KH₂PO₄, 5 mM glucose, and 58 mM saccharose (pH 6.5) on ice. Olfactory lobes and cerebellum were removed. The cerebral hemispheres were separated, and meninges and blood vessels removed. The cortices were cut into pieces and incubated in buffer containing 0.5% trypsin at 37°C for 10 minutes. The cell suspension was transferred to basal Eagle's medium (BME) (Sigma #B9638) containing 10% fetal calf serum (FCS). Further dissection of the tissue was obtained by aspiration of the suspension through a narrowed Pasteur pipette. Astrocytes were used for the experiments after 7-10 days in culture.

Before stimulation, cells were seeded in 96-well plates at a cell density of 1×10⁴ cells per well and incubated for 24 hours. Cells were incubated for 24 hours in medium without serum before stimulation. Stimulation with EGF (Biochrom, Berlin, Germany, #W1325.950.500; used at a concentration of 10 ng/ml) and bFGF (Biochrom, #W1370950050; used at a concentration of 10 ng/ml) was performed as indicated. The MAPK kinase inhibitor PD98059 was purchased from Axxora (Lörrach, Germany, #385-023), dissolved in DMSO and used at a concentration of 50 μM. The bFGF receptor inhibitor PD173074 (Skaper et al., 2000) was purchased from Calbiochem (Bad Soden, Germany, #341607), dissolved in DMSO and used at a concentration of 10 nM. Induction of DNA synthesis was measured by the incorporation of the pyrimidine analog BrdU instead of thymidine into the DNA of proliferating cells using the Cell Proliferation ELISA kit from Roche Diagnostics (Mannheim, Germany, #1647229). The assay was performed according to the instruction manual with minor modifications. The labeling time with BrdU was 4 hours and incubation with the anti-BrdU peroxidase-conjugated antibody was for 90 minutes. Peroxidase activity was determined spectrophotometrically as described in the instruction manual. Each experiment was performed at least four times and the means ± s.e.m. (Student's t-test) are depicted.

The lentiviral transfer vectors pFUW-REST/Elk-1ΔC, pFUWATF2ΔN, pFUW-MKP-1, pFUW-mycPP2C, and pFUWEgr-1/Zn have been described elsewhere (Stefano et al., 2006; Bauer et al., 2007; Mayer et al., 2008). Site-directed mutagenesis was performed with the Quick-change kit from Stratagene (La Jolla, CA, #210518), using plasmid pFUWEgr-1/Zn as template and the oligonucleotides 5′-GCCAGAAGCCCTTCCAGTCTCGAATCTGCATG-3′, 5′-CATGCAGATTCGAGACTGGAAGGGCTTCTGGC-3′, 5′-GCGAGAAGCCTTTTGCCTCTGACATTTGTGGGAG-3′, and 5′-CTCCCACAAATGTCAGAGGCAAAAGGCTTCTCGC-3′ as primers. To construct a FLAG-tagged NAB2 expression vector, we cloned an Ecl136II/XmnI-fragment, derived from plasmid pCMV-SPORT6.1-mNAB2 (accession #BC045138, obtained from the RZPD, Deutsches Ressourcenzentrum für Genomforschung) into the filled-in EcoRI site of plasmid p3XFLAG-CMV-7.1 (Sigma), generating plasmid pCMV-FLAGmNAB2. This plasmid was cut with Ecl136II and EcoRV. The fragment was inserted into the HpaI site of plasmid pFUW (Lois et al., 2002), generating the lentiviral transfer vector pFUWmNAB2. The lentiviral transfer vector pFUW-mycDA-Raf1 was generated via cloning of a filled-in XbaI-fragment, derived from plasmid pEF-BOS/Myc-DA-Raf1 (Yokoyama et al., 2007), into plasmid pFUW. The lentiviral transfer vectors pFWEgr-1.1luc and pFWEgr-1.2luc, encoding the luciferase reporter gene under the control of 239 or 490 nucleotides of the human EGR-1 5′-flanking region, have been described elsewhere (Rössler et al., 2008). The transfer vector pFWEgr-1SREluc encodes the luciferase reporter gene under the control of the proximal SREs #1 and #2 of the Egr-1 promoter. A minimal promoter was inserted downstream of the SREs consisting of a TATA box derived from the HIV long terminal repeat and an initiator element from the adenovirus major late promoter. The regulatory region of this transcription unit was derived from plasmid pEgr-1SREluc (Bauer et al., 2005). The viral particles were produced as previously described (Stefano et al., 2006) by triple transfection of 293T/17 cells with the gag-pol-rev packaging plasmid, the env plasmid encoding VSV-glycoprotein and the transfer vector.

Binding was measured using the electrophoretic mobility shift assay (EMSA). Assays were carried out for 15 minutes at room temperature in a 20-μl reaction containing 42 mM HEPES (pH 7.9), 40 mM KCl, 2.7 mM MgCl₂, 0.13 mM ZnCl₂, 0.09 mM Na₂EDTA, 1.16 mM dithiothreitol, 5.3 mM spermidine, 44 mg/ml double-stranded poly(dI-dC), 130 mg/ml bovine serum albumin, 8.7% glycerol and double-stranded DNA probe. To generate the probe, we isolated fragments from plasmids pEBS1⁴CAT and pEBS2⁴CAT, containing four copies of an Egr-1-binding site derived either from the Egr-1 or the synapsin I gene, using SalI and XhoI restriction enzymes. The isolated fragments were incubated with the Klenow fragment of Escherichia coli DNA polymerase I in the presence of [αP³²]dCTP. Nuclear extracts prepared from 293T cells that had been mock-transfected or transfected with plasmids pFUWEgr-1/Zn or pFUWEgr-1/ZnC368SC396S were added to the reaction mix, and the incubation was continued for 15 minutes. Free DNA and DNA-protein complexes were resolved by electrophoresis at 4°C on a 4% polyacrylamide gel (29:1 acrylamide:bisacrylamide) in 0.25× TBE (1× TBE is 50 mM Tris, 50 mM boric acid, 1 mM Na₂EDTA). The gel was dried and exposed to X-ray film.

Nuclear extracts were prepared as described (Kaufmann and Thiel, 2002). Nuclear proteins (20 μg) were separated by SDS-PAGE and the blots were incubated with antibodies directed against Egr-1 (Santa Cruz Biotechnology, Heidelberg, Germany, #sc-189). To detect phospho-ERK, 20 μg of proteins derived from whole-cell extract preparations were separated on SDS-PAGE and transferred to nitrocellulose membranes. Blots were probed with an antibody directed against the phosphorylated form of ERK2 (Promega, Mannheim, Germany, #V8031). Immunoreactive bands were detected using the ECL Plus system (Amersham, Braunschweig, Germany).

Reverse transcription polymerase chain reaction (RT-PCR) to amplify bFGF mRNA was performed as previously described (Bauer et al., 2007) using the primers 5′-GGAAACAGAGGCAGGATGAA-3′ and 5′-GAATAAGGGTTGCCCAGACA-3′.

ChIP experiments were performed as described elsewhere (Rössler et al., 2008). To amplify the proximal region of the murine bFGF promoter, the primers 5′-GCCTAGCGGGACAGATTCTT-3′ and 5′-GAGGGAGCCCCTTGAGTGTA-3′ were used. To amplify a distal region of the murine bFGF 5′-upstream region, the primers 5′-CACTAAGCCACCACACATGG-3′ and 5′-TTGGTCCCCCTCATAATCTG-3′ were used. The antibodies used for chromatin immunoprecipitation were anti-dimethyl H3K9 (Abcam, Cambridge, UK, #ab7312), anti-trimethyl H3K4 (Abcam, #ab8580), anti-Egr-1 (Santa Cruz Biotechnology, #sc-189), anti-Egr-2 (Santa Cruz Biotechnology, #sc-20690), and anti-Egr-3 (Santa Cruz Biotechnology, #sc-191). To detect binding of the Egr-1 mutant Egr-1/Zn to DNA, M2-agarose (Sigma, #A2220) was used, which interacts with the FLAG epitope of the mutant as described (Hohl and Thiel, 2005).