Transforming growth factor alpha acts as a gliatrophin for ... - Nature

Oncogene (2006) 25, 4076–4085

& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Transforming growth factor alpha acts as a gliatrophin for mouse and human astrocytes A Sharif1,2,3, V Pre´vot4, F Renault-Mihara1,2,3, C Allet4, J-M Studler1,2,3, B Canton1,2,3, H Chneiweiss1,2,3 and M-P Junier1,2,3 1 4

Inserm U752, Paris F-75013, France; 2University of Paris 5, Paris F-75013, France; 3Inserm U114, Paris F-75005, France and Inserm U816, Lille F-59000, France

Astrocyte death has been implicated in several neuropathological diseases, but the identification of molecules susceptible of promoting astrocyte survival has been elusive. We investigated whether transforming growth factor alpha (TGFa), an erbB1/EGFR ligand, which promotes glioma progression and affects astrocyte metabolism at embryonic and adult stages, regulates astrocyte survival. Primary serum-free astrocyte cultures from postnatal mouse and fetal human cortices were used. Transforming growth factor alpha protected both species of astrocytes from staurosporine-induced apoptosis. In serum-free medium, mouse astrocytes did not survive beyond 2 months while TGFa-treated astrocytes survived up to 12 months. Transforming growth factor alpha also promoted long-term survival of human astrocytes. We additionally extended TGFa proliferative effects to human astrocytes. After 3 days of permanent application, TGFa induced a major downregulation of both erbB1 and erbB2. This downregulation did not impair the functional activation of the receptors, as ascertained by their tyrosine phosphorylation and the continuous stimulation of both ERK/MAPK and PI3K/Akt pathways up to 7 days, the longest time examined. The full cellular effects of TGFa required activation of both transduction pathways. Enhanced proliferation and survival thus define TGFa as a gliatrophin for mammalian astrocytes. These results demonstrate that in normal, non-transformed astrocytes, sustained and functional erbBs activation is achieved without bypassing ligand-induced receptors downregulation. Oncogene (2006) 25, 4076–4085. doi:10.1038/sj.onc.1209443; published online 13 March 2006 Keywords: erbB1; apoptosis; survival; trophic factor; oncogenesis; EGF

Correspondence: Dr M-P Junier, Inserm U752, Paris F-75013, France. E-mail: [email protected] Received 25 July 2005; revised 20 December 2005; accepted 10 January 2006; published online 13 March 2006

Introduction Astrocytes, as defined by their expression of the specific intermediate filament GFAP, are the most abundant cell type of the CNS. They form intimate contacts with all other neural cells and play an essential role in the proper development and functioning of the CNS (Hertz and Zielke, 2004; Takuma et al., 2004). They directly affect the formation and activity of synapses, and coordinate cerebral blood flow with neuronal activity. They are also involved in neurogenesis, angiogenesis, and the development and maintenance of the blood–brain barrier. In the injured CNS, they participate in the immune response and modulate neural tissue repair. Accordingly, damage to these cells ultimately affects neuronal function. Patients suffering from Alexander disease bear GFAP mutations, which directly affect astrocytes, and result in profound fatal cerebral alterations (Rodriguez et al., 2001). In addition, astrocytes death participates in several neuropathological diseases ranging from HIVassociated dementia to ischemic and traumatic brain injury (Thomas et al., 1995; Thompson et al., 2001; Takuma et al., 2004; Giffard and Swanson, 2005). The identification of molecules capable of promoting the survival of astrocytes is therefore essential. Transforming growth factor alpha (TGFa), a member of the EGF family, with which it shares the same tyrosine kinase receptor, erbB1 or EGFR (Lee et al., 1995), has been implicated in the control of astrocytes genesis and mature functions. Mutant mice defective in TGFa or erbB1 expression possess reduced numbers of astrocytes in adulthood (Weickert and Blum, 1995; Kornblum et al., 1998; Sibilia et al., 1998). This may reflect defective astrocyte genesis and/or altered survival of these macroglial cells. In the adult CNS, TGFa is synthesized by subsets of astrocytes and neurons. In rodents, TGFa exerts mitogenic effects on mature astrocytes, stimulates their synthesis of various trophic factors and cytoskeletal proteins, and its overexpression results, in vivo, in the induction of astrogliosis, a phenomenon that accompanies all types of CNS injuries (reviewed in Xian and Zhou, 1999; Junier, 2000). Although TGFa is a survival factor for some neurons in rodents, and glial tumoral cells in humans (Morrison et al., 1987; Tang et al., 1997; Hanson et al., 1998; Boillee et al., 2001; Wechsler-Reya and Scott, 2001; Kerr

TGFa is a gliatrophin for astrocytes A Sharif et al

4077

et al., 2003), its pro-survival properties on astrocytes remain unexplored. In an attempt to identify molecules that promote both the growth and survival of astrocytes, and could therefore be considered as gliatrophins, we evaluated TGFa’s survival effects using primary cultures of human and mouse astrocytes. The lack of reports on the mitogenicity of erbB1 ligands on normal human astrocytes led us to additionally compare the proliferative effects of the factor in both species. We report that TGFa enhances human and rodent astrocytes survival and inhibits their apoptosis. Identification of the mechanisms underlying the effects of TGFa revealed for the first time that a sustained activation over several days of the erbB1 receptor and its associated transduction pathways can be achieved by prolonged ligand exposure, despite a robust downregulation of receptor protein levels.

Results Ligand binding to erbB1 induces receptor dimerization, activation of the kinase domain and receptor autophosphorylation on tyrosine residues (Schlessinger, 2004). Heterodimerization of ligand-bound erbB1 with the other members of the erbB family (erbB2, erbB3 and erbB4) may alter the cellular responses to the ligand (Riese and Stern, 1998; Moghal and Sternberg, 1999). Therefore, prior to evaluating the biological effects of TGFa on astrocytes, we performed a thorough characterization of the erbB receptors set expressed and/or activated by TGFa under our cell culture conditions. The lack of available data on erbBs activation beyond a few hours in TGFa-treated astrocytes prompted us to perform this characterization after both short (minutes) and long-term (days) TGFa treatments. erbB1 and erbB2 are expressed in mouse cortical astrocytes and activated by TGFa Only erbB1 and erbB2 were observed in untreated cortical astrocytes (Figure 1a–d), whereas the four members of the erbB receptor family were detected in mouse post-natal days 1–2 whole cortices. A possible induced expression of erbB3 and erbB4 was sought in astrocytes treated for a short (5 min) or prolonged (3 days) periods with either TGFa or Heregulin b1 (HRGb1, a product of the neuregulin 1 gene is a specific erB3 and erbB4 ligand; Yarden and Sliwkowski, 2001). Detection of the activated, tyrosine-phosphorylated forms of erbB3 and erbB4 receptors was achieved using immunoprecipitation with anti-erbB3 or erbB4 antibodies followed by immunoblotting with anti-phosphotyrosine antibodies, in 1–2 post-natal day-old whole cortices. However, they remained undetectable in cultured astrocytes, regardless of the length of the TGFa or HRGb1 treatment (Figure 1e and f). ErbB receptors, and especially erbB1, are subject to a tight regulation of their activation through ligandinduced receptor internalization and subsequent lyzoso-

Figure 1 ErbB receptors expression in mouse whole cortices and cortical astrocytes cultures. (a–d) Western blot analysis of erbB1, 2, 3 and 4, respectively. Ast: cultured cortical astrocytes. Ctx: whole cortex. IB: immunoblotting. Note that the four erbBs are present in whole cortices, while only erbB1 and erbB2 are detected in cultured cortical astrocytes. 50 mg protein/lane. (e, f) Lack of erbB3 and erbB4 expression and activation in mouse astrocytes treated with TGFa or Heregulin b1. Phosphorylated erbB3 (e) and erbB4 (f) were analysed under control conditions and following 5 min and 3 days of TGFa or Heregulin b1 (HRGb1) treatment. Note that phosphorylated erbB3 and erbB4 are absent in cultured astrocytes, while they are detected in mouse cortices. Ctx: whole cortex. IPP: immunoprecipitation. IB: immunoblotting. d: day.

mal targeting within a few minutes (Wiley, 2003). We therefore performed direct assays of whole-cell protein lysates with erbB1 and erbB2 antibodies to verify whether they could still be detected in astrocytes after 3 days of TGFa exposure, using 5 min stimulation as a positive control. Both receptors were observed at either time point, but a marked downregulation of their protein levels was noted after 3 days of TGFa treatment (Figure 2). This observation led us to analyse their activation status. Phosphorylations of both erbB1 and erbB2 were still detected after a long-term treatment of 3 days in the continuous presence of TGFa (Figure 3), indicating that TGFa-induced erbBs activation remains effective despite downregulation in the cellular contents of the receptors. In addition, formation of erbB1/erbB2 heterodimers was verified by immunoprecipitation followed by immunoblotting (Figure 3b). Thus, TGFa continuously activates both erbB1 and erbB2 in mouse post-natal astrocytes over several days in culture. Functionality of the long-term erbB1 activation induced by TGFa To further verify the functionality of TGFa-activated erbB1 and erbB2 receptors, we examined the mobilization of two erbB1-linked transduction pathways, the ERK/MAPK and PI3K/Akt pathways in astrocytes treated from 24 h to 7 days with TGFa. Presence of the phosphorylated, activated forms of the proteins was evaluated in control and treated astrocytes. In the absence of TGFa, low levels of phosphorylated ERK1/2 and Akt were observed (Figure 4a and b). Application Oncogene


4078

Figure 2 Long-term TGFa treatment results in the downregulation of erbB1 and erbB2 expression in mouse astrocytes. ErbB1 and erbB2 levels were evaluated using Western blot analysis in mouse astrocytes cultures under control conditions and following TGFa treatment for 5 min and 3 days. IB: immunoblotting. d: day. 50 mg protein/lane. Figure 4 Transforming growth factor alpha induces activation of the MAPK/ERK and PI3K/Akt transduction pathways in mouse astrocytes. ERK1/2 (a) and Akt (b) phosphorylation in mouse astrocytes cultures were analysed using Western blots under control conditions and following TGFa treatment for 1, 2, 3 and 7 days. Note that TGFa-induced ERK1/2 and Akt phosphorylation is observed at all times tested. IB: immunoblotting. d: day. 50 mg protein/lane.

Figure 3 ErbB1 and erbB2 are activated in astrocytes upon longterm TGFa treatment. Phosphorylated erbB1 (a) and erbB2 (b) were analysed by Western blot in mouse astrocytes cultures under control conditions and following 5 min and 3 days of TGFa treatment. Note that TGFa activates erbB1 and erbB2 at both times tested, and that erbB1 co-precipitates with erbB2 (b). IPP: immunoprecipitation. IB: immunoblotting. 4G10: monoclonal anti-phosphotyrosine antibody. d: day.

of the growth factor strongly enhanced phosphorylations of ERK1/2 and Akt at all times tested from 24 h to 7 days (Figure 4a and b). An increase in the intensity of the total Akt signal was consistently noted after 7 days of TGFa treatment (Akt/actin relative densitometric units ¼ 144714.4% of controls, mean7s.e.m., n ¼ 3), indicating that TGFa also increased Akt expression at this time point. To further analyse cellular responses upon long-term TGFa application, we determined cellular proliferation using BrdU incorporation. In the mouse, the mitogenic effects of erbB1 activation have only been studied using EGF as a ligand (Leutz and Schachner, 1981), and in the human they have been studied on cells derived from low-grade astrocytomas, but not on normal astrocytes Oncogene

(Barna et al., 2001). In the absence of TGFa, astrocytes presented a low level of proliferation (Figure 5a, percentage of BrdU-immunoreactive nuclei over total nuclei numbers labelled with hematoxylin: 1 DIV, 1.570.1; 1.5 DIV, 2.770.4; 3 DIV, 9.471.9; 7 DIV, 11.271.0, mean7s.e.m., n ¼ 3). In presence of TGFa, a twofold increase in the number of cycling cells was visible at 24 h post-treatment (Figure 5b). Maximal proliferation was observed at 3 and 7 days of TGFa treatment, the numbers of BrdU-immunoreactive cells increasing sevenfold in comparison to control conditions. A large majority of astrocytes were sensitive to TGFa mitogenic effects in our cultures with 80% of cells having entered the cell cycle at 3 days (Figure 5b). Finally, AG1478, a specific inhibitor of erbB1 tyrosine kinase activity (Yaish et al., 1988) abolished TGFainduced proliferation as observed at 7 DIV (percentage of BrdU-immunoreactive nuclei over total nuclei numbers labelled with hematoxylin: untreated, 18.371.8; AG1478-treated, 1171.2; TGFa-treated, 70.671; TGFa plus AG1478-treated, 13.470.8, mean7s.e.m., n ¼ 4). To evaluate the transduction pathways underlying this mitogenic effect, astrocytes were treated with TGFa in the additional presence of U0126, a specific inhibitor of both MEK1 and MEK2 kinase activities, which prevents phosphorylation, and thus activation of ERK1/2 (Favata et al., 1998). LY294002, a specific inhibitor of PI3K activity, was used to suppress Akt phosphorylation and activation (Duronio et al., 1998). The efficacy and specificity of these inhibitors were verified by immunoblotting assays in our experimental conditions. Each inhibitor specifically prevented the


4079

Figure 5 Transforming growth factor alpha enhances mouse and human astrocytes proliferation. Proliferating cells in astrocytes cultures were quantified using BrdU incorporation. (a) Examples of BrdU-immunoreactive nuclei (black staining, brown in HTML version) in control, and TGFa-treated mouse astrocytes cultures. Hematoxylin counterstaining appears as gray (blue in HTML version). Scale bar ¼ 100 mm. (b) Quantification of BrdU-immunoreactive nuclei in control and TGFa-treated mouse astrocytes cultures from 1 to 7 days. Results are expressed as percentage of BrdU-immunoreactive cells following TGFa treatment as compared to time-matched untreated cultures. Mean7s.e.m., n ¼ 6. (c) Suppression of TGFa mitogenic effects in mouse astrocytes cultures treated with U0126 or LY294002, specific inhibitors of the MAPK/ERK and PI3K/Akt transduction pathways, respectively. Results are expressed as in (b). Mean7s.e.m., n ¼ 6. ***Po0.0001, as compared to untreated astrocytes. (d) Quantification of BrdUimmunoreactive nuclei in untreated and TGFa-treated cultures of human astrocytes at 3 days post-treatment. Results are expressed as in (b). Mean7s.e.m., n ¼ 9–10. **Po0.01.

activation of its respective target in both control and TGFa-stimulated conditions without affecting the other (Figure 6a and b). U0126 and LY294002 treatments were limited to 36 h since U0126 proved to be toxic to astrocytes when applied for longer periods (data not shown). At that time, TGFa alone induced close to a sixfold increase in BrdU-immunoreactive cells (Figure 5c). The application of either U0126 or LY294002 was sufficient to abrogate TGFa mitogenic effect (Figure 5c). These results show that both the ERK/MAPK and the PI3K/Akt transduction pathways are necessary for the mitogenic effect of TGFa. This mitogenic effect of TGFa was also observed on human astrocytes (Figure 5d). TGFa enhances survival of astrocytes through increased lifespan and protection from induced apoptosis To evaluate TGFa’s effect on astrocyte survival, we undertook different approaches. We determined whether TGFa could alter the in vitro lifespan of

astrocytes, which are naturally submitted to senescence. We also determined whether TGFa could protect astrocytes from induced death. Astrocytes were grown in defined serum-free medium in the presence or absence of TGFa, and visualized using actin labelling with Alexa 488-coupled phalloidin and DAPI counterstaining. Transforming growth factor alpha caused a change from a polygonal morphology to a process-bearing morphology, as previously reported in rat astrocytes (Zelenaia et al., 2000). In the absence of TGFa, astrocytes began to die after 30 DIV, and were all dead within 2 months (Figure 7a). In contrast, astrocytes cultured in the continuous presence of TGFa survived (Figure 7a). Similar results were observed with human astrocytes. Only scattered cells were observed in cultures maintained for 3 months in absence of TGFa, while numerous cells populated the cultures maintained in TGFa-supplemented medium (Figure 7b). During the course of this study, we noted that replacement of TGFa-supplemented medium with serum-free medium triggered astrocyte death. This effect Oncogene


4080

Figure 6 Densitometric analyses of phosphorylated ERK1/2 (a) and Akt (b) levels over total ERK1/2 and Akt levels in mouse astrocytes cultures in the absence and presence of TGFa, U0126 and LY294002 for 36 h. Note that U0126 inhibits ERK1/2 phosphorylation and is without effect on Akt phosphorylation. Conversely, LY294002 suppresses Akt phosphorylation without affecting ERK1/2 phosphorylation.

was quantified using astrocyte cultures treated for 3 days with TGFa, and then further cultured for 3 days in either serum-free medium, TGFa-supplemented medium, EGF-supplemented medium or bFGF-supplemented medium. Growth factor deprivation induced apoptosis of 40% of the cell population, as revealed by the presence of apoptotic bodies after DAPI nuclear staining. In contrast, no apoptotic bodies were observed when TGFa-treated astrocytes were transferred in TGFa-supplemented medium. Replacement of TGFa with EGF, the functional analog of TGFa, also rescued astrocytes from growth factor deprivation-induced death (Figure 7c), but bFGF, another mitogenic factor for astrocytes (see below), was without effect. The capability of TGFa to protect astrocytes from induced death was further evaluated using staurosporine as a death inducer. Astrocytes were treated with staurosporine for 3 h and death was evaluated 24 h after application (Di Iorio et al., 2004). DAPI nuclear staining revealed the presence of apoptotic bodies, which are indicative of apoptotic death (Figure 8a). In the absence of TGFa, 50% of the mouse astrocytes survived to staurosporine-induced death, as compared with control conditions, while 75–80% of cells survived Oncogene

Figure 7 Long-term survival of mouse (a) and human (b) astrocytes in serum-free medium supplemented with TGFa. The astrocytes’ actin cytoskeleton was visualized using Alexa 488coupled phalloidin. (a) Examples of mouse astrocytes cultures maintained for 7 days and 9 months. At 9 months, only cellular debris were observed in control cultures, while TGFa-treated cells populated the whole culture dishes at 7 days. Cell nuclei labelled with DAPI. Scale bars ¼ 50 mm. (b) Example of human astrocytes cultured for 3 months. At that time, only scattered cells were observed in the absence of TGFa, as opposed to cultures maintained in TGFa-supplemented medium. Scale bar ¼ 50 mm. (c) TGFa protects mouse astrocytes from growth factor deprivation-induced apoptosis. DAPI-stained healthy nuclei were quantified in cultures in which TGFa was replaced by defined serum-free medium containing or not TGFa, EGF or bFGF. Results are expressed as percentage of surviving cells as compared to cultures supplemented with TGFa. Mean7s.e.m., n ¼ 4–9. ***Po0.0001.

when TGFa was applied simultaneously with staurosporine, and maintained in the medium until cell fixation (Figure 8b). Staurosporine being a large spectrum kinase inhibitor (Ruegg and Burgess, 1989), we next verified its effects on TGFa-induced ERK and Akt activations. Staurosporine did not prevent TGFa-induced ERK and Akt activations, but enhanced by itself ERK phosphorylation (Figure 8d). Nevertheless, staurosporine-induced


4081

Figure 8 Transforming growth factor alpha protects mouse and human astrocytes from staurosporine-induced apoptosis. (a) Example of DAPI labelling of mouse astrocytes nuclei. Upper panel: in control conditions, astrocytes nuclei appeared as rhomboidshaped structures uniformly labelled by DAPI. Lower panel: following staurosporine application, numerous apoptotic bodies strongly stained with DAPI (arrow) were observed. Scale bars ¼ 20 mm. (b) Quantification of DAPI-stained healthy nuclei in control and staurosporine-treated mouse cultures in the absence or presence of TGFa. In TGFa-treated cultures, the growth factor was added at the time of staurosporine application. Results are expressed as percentage of surviving cells as compared to their respective timematched cultures that have not been exposed to staurosporine. Mean7s.e.m., n ¼ 12. **Po0.01. (c) Quantification of DAPI-stained healthy nuclei in control and staurosporine-treated human astrocytes cultures in the absence or presence of TGFa. Results are expressed as in (b). Mean7s.e.m., n ¼ 3. *Po0.05 as compared to control astrocytes that have not been exposed to TGFa. (d) Effects of staurosporine on ERK and Akt activations in control and TGFa-treated astrocytes. Note that staurosporine induces ERK phosphorylation. (e) Modulation of TGFa protective effects on staurosporine-induced mouse astrocytes apoptosis by U0126 and LY294002. Results are expressed as in (b). Mean7s.e.m., n ¼ 4–5. *Po0.05, ***Po0.0001 as compared to TGFa-treated astrocytes. Note that the combined inhibition of the MAPK/ERK and PI3K/Akt pathways is required to suppress TGFa protective effects.

apoptosis was maintained in the presence of U0126 and LY294002 (4772.7% of surviving astrocytes in staurosporine-treated cultures, versus 57.678.4% in cultures treated with staurosporine þ U0126 þ LY294002, mean7s.e.m., n ¼ 3), thus showing that staurosporineinduced apoptosis does not depend on ERK activation. In contrast, inhibition of the activation of the ERK/ MAPK using U0126 induced a nearly twofold decrease in TGFa protective effect (Figure 8e). Inhibition of the activation of the PI3K/Akt pathway using LY294002 did not significantly alter TGFa’s protective effect (Figure 8e). A total abrogation of TGFa protective effects was observed with the simultaneous inhibition of both transduction pathways (Figure 8e). These results show that the ERK/MAPK and the PI3K/Akt pathways act synergistically to mediate TGFa protective effect. TGFa also protected human astrocytes from staurosporine-induced death, since only 40% of human astrocytes survived to staurosporine treatment, whereas TGFa induced the survival of nearly 75% of the cells (Figure 8c). Transforming growth factor alpha is thus endowed with pro-survival properties, which can be observed both in murin and human astrocytes. Furthermore, these effects were specific to TGFa since the

growth factor bFGF, which is able to enhance astrocyte proliferation (450.977.4% of controls after 3 DIV) and is known to activate ERK and Akt (Schlessinger, 2004), did not protect mouse astrocytes from staurosporineinduced death (percentage of surviving cells in presence of staurosporine as compared to their time-matched cultures unexposed to staurosporine: control, 57.272.15; bFGF-treated, 46.971.43, mean7s.e.m., n ¼ 3).

Discussion Heterodimerization between the erbBs results in signal diversification and amplification (Prenzel et al., 2001). For example, erbB2 prolongs the TGFa-induced intracellular signalling when co-expressed with erbB1 (Klapper et al., 1999). Astrocytes derived from 1- to 2day-old mouse cortices expressed only two of the four members of the erbB family, erbB1 and 2. Expression of erbB3 and 4 in similar cultures of mouse astrocytes was not investigated. ErbB4 receptors have either not been detected (Prevot et al., 2003) or Oncogene


4082

were observed at low levels (Elenius et al., 1997; Francis et al., 1999) in astrocytes cultures derived from newborn rat cortices. ErbB3 expression has been found in one study (Francis et al., 1999) and not in another (Pinkas-Kramarski et al., 1997). Lack or weak expression of erbB3 and 4 in cultures of cortical astrocytes appears coherent with the in vivo situation since in the adult rat cortex erbB3 and erbB4 immunostainings are primarily observed in GFAPimmunonegative cells (Gerecke et al., 2001). In the adult mouse cortex, erbB4 expression is detected at low levels in some astrocytes (Erlich et al., 2000). Astrocyte expression of erbB3 receptors is only reported during astrogliosis following injury of the rat adult cortex (Tokita et al., 2001). ErbB1 and erbB2 activation upon ligand binding to erbB1 has been studied in various cellular models over a maximal time frame of 24 h. Their activation was shown to occur within minutes and to be transient, with the activation subsiding within a few hours up to 1 day (see for example; Waterman et al., 1998; Kornblum et al., 1999; Prevot et al., 2005). Our results show that at least in astrocytes, erbB1 and erbB2 may be activated over several days when continuously exposed to TGFa. Most surprisingly, this activation is functional, translating into a sustained mobilization of the MAPK/ERK and Akt pathways all along the treatment, despite a major downregulation in the expression of the receptors. Receptor internalization and subsequent degradation are indeed at the core of the mechanisms responsible for erbB1 signal termination (Sweeney and Carraway, 2004). Yet, our results show that the marked and sustained downregulation of both erbB1 and erbB2 whole cellular levels induced by TGFa does not prevent activation of the receptors in astrocytes, mobilization of the underlying transduction pathways, and the manifestation of the biological effects. Escape from ligandinduced downregulation participates in the deregulation of tyrosine kinase receptors and their subsequent oncogenic effects (Peschard and Park, 2003). Accordingly, in gliomas, and most tumors exhibiting a TGFa/ erbB1 trophic loop, erbB1 protein levels are enhanced (Wechsler-Reya and Scott, 2001). In addition, pancreatic tumors induced in mice by the combined transgenic overexpression of TGFa and the inactivation of one of the alleles of the tumor suppressor gene p53 exhibit amplification of the erbB1 gene (Schreiner et al., 2003a, b). Likewise, mammary tumors induced by a TGFa transgene display enhanced erbB1 mRNA levels (Matsui et al., 1990). These observations, associated with the occurrence of erbB1 gene amplification in highgrade gliomas, suggest that maintenance of elevated erbB1 levels is required for erbB1 and its ligand to exert their full oncogenic effects. As consequences of enhanced expression of erbB1, tumoral epithelial cells and glioma cells exhibit increased survival, resistance to induced-cell death and deregulated proliferation (Wu et al., 2000; Yarden and Sliwkowski, 2001; Chakravarti et al., 2002; Ciardiello and Tortora, 2003; Kari et al., 2003). Our results show that TGFa may achieve similar effects on normal astrocytes, without the help of Oncogene

enhanced receptors levels. The capability of normal astrocytes to ensure an appropriate regulation of erbB1 and 2 proteins levels despite their continuous exposure to TGFa is likely to be instrumental in limiting the wellknown oncogenic properties of this erbB1 ligand (Lee et al., 1995). Transforming growth factor alpha cellular effects on astrocytes were not restricted to mitosis but extended to survival, through enhanced lifespan and protection from induced death. Our results show that TGFa possesses the striking capability of maintaining viable astrocytes in serum-free medium for over a year. The density of the human astrocytes cultures was stable over time, while it increased in mouse cultures. This indicates that for this last species, cell proliferation contributes in part to the long-term maintenance of the astrocytes, although aged astrocytes have been shown to lose their mitogenic responsiveness to EGF, the structural and functional analog of TGFa (Levison et al., 2000). However, we observed within 4 days the massive death of 12-monthold cultured mouse astrocytes when TGFa was removed from the medium, whereas a progressive cell depletion should have been observed if proliferation was at the sole origin of these long-term cultures. Transforming growth factor alpha’s survival effect appears therefore distinct from its proliferative effect in both mouse and human astrocyte. Short-term TGFa exposure was moreover capable of protecting both mouse and human astrocytes from staurosporine-induced death. Interestingly, bFGF, a well-known astrocyte mitogen acting through tyrosine kinase receptors, was ineffective. In mouse astrocytes, TGFa’s protective effects on staurosporine-induced death were unaltered by blockade of the PI3K/Akt pathway. This observation was unexpected given the central role attributed to the Akt kinase in cell survival (Franke et al., 2003). In contrast, inhibition of the MAPK/ERK pathway reduced by half TGFa’s protective effects. Nevertheless, total blockade was only obtained when both pathways were inhibited, and suggests a synergistic cooperation between Akt and ERK through yet unidentified mechanisms. Exploration of the signalling pathways underlying TGFa’s mitogenic effects showed that blockade of either the MAPK/ERK or the PI3K/Akt pathway is sufficient to suppress TGFa’s proliferative signal in astrocytes. Although the Ras/MAPK/ERK pathway has been proposed as the major mitogenic signalling pathway initiated by erbB1 activation (Malumbres and Pellicer, 1998), the fact that TGFa-induced activation of ERK was unaltered following blockade of the PI3K/Akt pathway and vice versa rules out eventual reciprocal modulations of these pathways. Our results raise the possibility that each pathway intervenes at different steps of the cell cycle, a possibility supported by studies of TGFa mitogenic effects in epithelial cell lines, where Akt stimulation is required for G1/S transition (Busse et al., 2000; Justman and Clinton, 2002). Altogether, our results show that the mobilization of both MAPK/ERK and PI3K/Akt transduction pathways are required for TGFa to exert its full mitogenic and pro-survival effects.


4083

To this day, TGFa is considered as a factor that promotes proliferation and metabolic activity in mammalian astrocytes. Its effects also extend to the induction of astroglial reactivity (Rabchevsky et al., 1998). Although its neurotrophic properties were known (Morrison et al., 1987; Hanson et al., 1998; Boillee et al., 2001; Kerr et al., 2003), a survival effect on glial cells remained to be demonstrated. In mammals, astrocyte death contributes to various pathological events (Takuma et al., 2004; Giffard and Swanson, 2005), and regulation of astrocytes numbers through developmental death has been documented (Soriano et al., 1993; Krueger et al., 1995). Nevertheless, only scarce studies have been devoted to the identification of growth factors capable of promoting the survival of these glial cells (Yokoyama et al., 1993; Bakhiet et al., 2001; Saas et al., 2002). In contrast, several studies point to products of neuregulin1 gene, a member of the EGF family, as survival factors during the development of both Schwann cells (Lemke, 2001) and oligodendrocytes (Fernandez et al., 2000). Through the demonstration that TGFa promotes long-term survival and enhances resistance of astrocytes to induced death, we provide evidence that the EGF family also includes a survival factor for astrocytes. These results, associated with the previous demonstration that Spitz and Vein, the respective TGFa and neuregulin orthologs in drosophila, protect the fly’s glial cells from developmental cell death (Hidalgo et al., 2001; Beck and Fainzilber, 2002; Bergmann et al., 2002) suggest that the trophic role of the members of the EGF family on glial cells are strongly conserved through evolution. Promotion of neuronal growth and survival are the criteria that have been the most frequently used to define molecules as neurotrophic factors. The capability of TGFa to promote both astrocyte growth, through enhanced proliferation, and survival, through increased lifespan and protection from induced-death, thus allows to designate TGFa as a gliatrophin for human and mouse astrocytes. It remains to be determined whether TGFa gliatrophic properties extend to in vivo situations as shown in drosophila, a possibility supported by the reduced astrocyte numbers in the forebrains of adult mutant mice defective in TGFa or erbB1 expression (Weickert and Blum, 1995; Kornblum et al., 1998; Sibilia et al., 1998), and the protection provided by TGFa in rodent models of ischemia, a pathological situation known to be accompanied with astrocyte death (Justicia and Planas, 1999).

Materials and methods All procedures were approved by the Institutional Animal Care and Use Committee of the Institut de Biologie du Colle`ge de France, using C57Bl6/J mice (Janvier, France). All experiments carried on cells derived from human fetuses were performed in compliance with the French laws on bioethics. Recombinant human TGFa was from AbCys (France), bFGF from RnDSystems (France) and EGF from Invitrogen (France). Recombinant human Heregulin b1 was from

Neomarkers (France). Tyrphostin AG1478, LY294002, U0126, and staurosporine were from Calbiochem (La Jolla, CA, USA), and dissolved in dimethyl-sulfoxide. Statistical analyses Statistical analyses were carried out using ANOVA followed by a Fisher test, the level of significance was set at Po0.05, and results are shown as mean7s.e.m. Cell culture preparations Mouse astrocytes were prepared from cortices of 1–2-day-old mice following previously described procedures (Prevot et al., 2005), using dishes coated with 1.5 mg/ml poly-L-ornithine (MW 40 000, Sigma, France). The cultures were transferred in serum-free medium for 3 days, prior to being subjected to specific treatments. Immunocytochemical detection of the astrocyte-specific marker glial fibrillary acidic protein (monoclonal mouse anti-GFAP antibody, 1:400, ICN, France) and of the microglia CD11b receptor (monoclonal rat Mac1 antibody, 1:200, PharMingen, France) were used to determine the astrocytes enrichment of the cultures. For staurosporineinduced apoptosis assays, cortical cells suspensions were directly seeded in 24-well plates and grown for 6 days in 10% fetal calf serum. Contaminating cells were removed through washes with ice-cold PBS at 1, 4 and 6 DIV. The astrocytes were then submitted to the appropriate treatments. Human astrocytes cultures were prepared from fetal cortices (9–12 gestational weeks), and grown on poly-D-lysine-coated plastic dishes and glass slides (100 ng/ml, Sigma, France). Human fetal cortices were crushed on an 80 mm nylon (Buisin, France), and filtered through a 20 mm Nitex cloth (Buisin, France) to obtain single-cell suspension. After a growth period of 1 month in DMEM-F12 supplemented with 10% (v/v) fetal calf serum (Invitrogen, France), astrocytes were used for experiments. In all mouse and human cultures, GFAPimmunoreactive astrocytes accounted for 91–96% of the cells. In mouse cultures, the 4–9% remaining cells were immunoreactive for the CD11b receptor microglia-specific antigen. Immunoprecipitation and Western blot analyses After treatment, the cells were rinsed once with ice-cold PBS and snap-frozen on dry ice. Preparation of cell lysates, erbBs immunoprecipitations and Western blot analysis followed the procedures described in Prevot et al. (2005). Cell lysates were normalized according to protein concentration with the BSA kit (Bio-Rad, Hercules, CA, USA). Immunoprecipitations were carried with equal amounts of protein (0.8–1 mg), using 2 mg of polyclonal antibodies directed against epitopes contained within the intracellular domain of the receptors (anti-erbB1 sc-03-G, anti-erbB2 sc-284, anti-erbB3 sc-285, Santa Cruz Biotechnologies, Santa Cruz, CA, USA), and of a monoclonal anti-erbB4 antibody (Ab-1, Neomarkers) that recognizes an epitope present in the extracellular domain of erbB4. Membranes immunoblotting was achieved with the monoclonal anti-phosphotyrosine antibody 4G10 (1.5 mg/ml; UpstateBiotechnologies, Lake Placid, NY, USA), or with polyclonal antibodies against erbB1 (1.5 mg/ml; UpstateBiotechnologies, Lake Placid, NY, USA), erbB-2 (1:500, sc-284), erbB4 (1:200, sc-283; Santa Cruz Biotechnologies, Santa Cruz, CA, USA) or with the monoclonal anti-erbB3 antibody (clone 2F12, 1.5 mg/ml; UpstateBiotechnologies, Lake Placid, NY, USA). Western blot analysis of the MAPK/ERK and PI3K/ Akt transduction pathways were performed using the following antibodies: rabbit polyclonal anti-Akt, 1:1000 (Cell Signaling, France), rabbit polyclonal anti-phosphorylated Akt, 1:1000 (Cell Signaling, France), rabbit polyclonal Oncogene


4084 anti-ERK1/2 1:1000 (extracellular signal-regulated kinases 1 and 2, Santa Cruz, France), monoclonal mouse anti-phosphorylated ERK1/2, 1:1000 (Cell Signaling, France). The densities of the immunoreactive bands were quantified using the NIH Image software 1.60/ppc. Densitometric results are presented for each condition as the ratio of phosphorylated ERK1/2 over total ERK1/2 and the ratio of phosphorylated Akt over total Akt levels. Experiments were repeated 2–3 times in an independent manner. Proliferation assay Astrocytes were cultivated for 1–7 days in defined medium supplemented or not with TGFa (50 ng/ml) in the continuous presence of 10 mM 5-bromo-2-deoxyuridine (BrdU, Sigma, France), and for 36 h when U0126 (10 mM) or LY294002 (30 mM) were used to avoid the toxicity of U0126 beyond this period. Cells that had entered the cell cycle were identified by immunocytochemical detection of BrdU in their nuclei following previously described procedures (Rabchevsky et al., 1998). Cells were counterstained with Harris’ hematoxylin (Reáctifs RAL, Bordeaux, France). The nuclei were counted over four distinct areas of 3 mm2, and averaged for each culture slide. Results are expressed as percentage of BrdU-immunoreactive nuclei over total nuclei numbers. Experiments were repeated 2–4 times in an independent manner. Survival assays Long-term survival of astrocytes was examined in astrocytes cultures maintained in defined medium containing or not TGFa (50 ng/ml), renewed twice a week. One-week and 9 month-old control and TGFa-treated cultures were fixed in 4% paraformaldehyde and processed for actin labelling with phalloidin coupled to Alexa 488 (1/1500 in PBS, 30 min at room temperature, Molecular Probes, France). Nuclei were counterstained with DAPI (3 mM in PBS, 10 min at room

temperature, Sigma, France). Staurosporine-induced astrocytes death was performed as described (Di Iorio et al., 2004), exposing astrocytes for 3 h to 100 nM staurosporine (Calbiochem, La Jolla, CA, USA). Transforming growth factor alpha (50 ng/ml), U0126 (10 mM), LY294002 (30 mM) or the corresponding vehicles were added simultaneously with staurosporine. After 3 h, the cells were rinsed, and defined medium containing or not TGFa (50 ng/ml), U0126 and LY294002 was further added for 21 h. At that time, astrocytes were fixed in 4% paraformaldehyde, and the cells nuclei labelled with DAPI. Astrocytes death was evaluated by counting the total number of intact, uniformly labelled nuclei over two distinct areas of 0.16 mm2, using a Nikon fluorescent microscope and a 20 magnification lens, and averaged for each culture slide. Results are expressed as percentage of healthy DAPI-labelled nuclei in staurosporine-treated cultures as compared to their respective time-matched cultures that have not been exposed to staurosporine. The numbers of astrocytes surviving growth factor deprivation were determined in an identical manner. Experiments were repeated 3–6 times in an independent manner.

Acknowledgements We are grateful to Professor Jacques Glowinski for his constant support. We thank Amelia Dias-Morais and Yvette Thorens for their expert technical help, and Annette Koulakoff for generously sharing her knowledge of cell cultures. This research was supported by the Association pour la Recherche contre le Cancer (ARC, grant# 3500 to HC, and study fellowship to AS), the Fondation pour la Recherche Medicale (FRM, grant to VP), the Acade´mie Nationale de Me´decine (study fellowhip to FR-M), the Centre Hospitalier Universitaire de Lille and the Re´gion Nord Pas de Calais (study fellowships to CA).

References Bakhiet M, Tjernlund A, Mousa A, Gad A, Stromblad S, Kuziel WA et al. (2001). Nat Cell Biol 3: 150–157. Barna BP, Mattera R, Jacobs BS, Drazba J, Estes ME, Prayson RA et al. (2001). Cell Immunol 208: 18–24. Beck G, Fainzilber M. (2002). Neuron 33: 673–675. Bergmann A, Tugentman M, Shilo BZ, Steller H. (2002). Dev Cell 2: 159–170. Boillee S, Cadusseau J, Coulpier M, Grannec G, Junier MP. (2001). J Neurosci 21: 7079–7088. Busse D, Doughty RS, Ramsey TT, Russell WE, Price JO, Flanagan WM et al. (2000). J Biol Chem 275: 6987–6995. Chakravarti A, Chakladar A, Delaney MA, Latham DE, Loeffler JS. (2002). Cancer Res 62: 4307–4315. Ciardiello F, Tortora G. (2003). Eur J Cancer 39: 1348–1354. Di Iorio P, Ballerini P, Traversa U, Nicoletti F, D’Alimonte I, Kleywegt S et al. (2004). Glia 46: 356–368. Duronio V, Scheid MP, Ettinger S. (1998). Cell Signal 10: 233–239. Elenius K, Paul S, Allison G, Sun J, Klagsbrun M. (1997). EMBO J 16: 1268–1278. Erlich S, Shohami E, Pinkas-Kramarski R. (2000). Mol Cell Neurosci 16: 597–608. Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley DA, Feeser WS et al. (1998). J Biol Chem 273: 18623–18632. Fernandez PA, Tang DG, Cheng L, Prochiantz A, Mudge AW, Raff MC. (2000). Neuron 28: 81–90. Oncogene

Francis A, Raabe TD, Wen D, DeVries GH. (1999). J Neurosci Res 57: 487–494. Franke TF, Hornik CP, Segev L, Shostak GA, Sugimoto C. (2003). Oncogene 22: 8983–8998. Gerecke KM, Wyss JM, Karavanova I, Buonanno A, Carroll SL. (2001). J Comp Neurol 433: 86–100. Giffard RG, Swanson RA. (2005). Glia 50: 299–306. Hanson Jr MG, Shen S, Wiemelt AP, McMorris FA, Barres BA. (1998). J Neurosci 18: 7361–7371. Hertz L, Zielke HR. (2004). Trends Neurosci 27: 735–743. Hidalgo A, Kinrade EF, Georgiou M. (2001). Dev Cell 1: 679–690. Junier MP. (2000). Prog Neurobiol 62: 443–473. Justicia C, Planas AM. (1999). J Cereb Blood Flow Metab 19: 128–132. Justman QA, Clinton GM. (2002). J Biol Chem 277: 20618–20624. Kari C, Chan TO, Rocha de Quadros M, Rodeck U. (2003). Cancer Res 63: 1–5. Kerr DA, Llado J, Shamblott MJ, Maragakis NJ, Irani DN, Crawford TO et al. (2003). J Neurosci 23: 5131–5140. Klapper LN, Glathe S, Vaisman N, Hynes NE, Andrews GC, Sela M et al. (1999). Proc Natl Acad Sci USA 96: 4995–5000. Kornblum HI, Hussain R, Wiesen J, Miettinen P, Zurcher SD, Chow K et al. (1998). J Neurosci Res 53: 697–717.


4085 Kornblum HI, Zurcher SD, Werb Z, Derynck R, Seroogy KB. (1999). Eur J Neurosci 11: 3236–3246. Krueger BK, Burne JF, Raff MC. (1995). J Neurosci 15: 3366–3374. Lee DC, Fenton SE, Berkowitz EA, Hissong MA. (1995). Pharmacol Rev 47: 51–85. Lemke G. (2001). Annu Rev Neurosci 24: 87–105. Leutz A, Schachner M. (1981). Cell Tissue Res 220: 393–404. Levison SW, Jiang FJ, Stoltzfus OK, Ducceschi MH. (2000). Glia 32: 328–337. Malumbres M, Pellicer A. (1998). Front Biosci 3: d887–d912. Matsui Y, Halter SA, Holt JT, Hogan BL, Coffey RJ. (1990). Cell 61: 1147–1155. Moghal N, Sternberg PW. (1999). Curr Opin Cell Biol 11: 190–196. Morrison RS, Kornblum HI, Leslie FM, Bradshaw RA. (1987). Science 238: 72–75. Peschard P, Park M. (2003). Cancer Cell 3: 519–523. Pinkas-Kramarski R, Eilam R, Alroy I, Levkowitz G, Lonai P, Yarden Y. (1997). Oncogene 15: 2803–2815. Prenzel N, Fischer OM, Streit S, Hart S, Ullrich A. (2001). Endocr Relat Cancer 8: 11–31. Prevot V, Lomniczi A, Corfas G, Ojeda SR. (2005). Endocrinology 146: 1465–1472. Prevot V, Rio C, Cho GJ, Lomniczi A, Heger S, Neville CM et al. (2003). J Neurosci 23: 230–239. Rabchevsky AG, Weinitz JM, Coulpier M, Fages C, Tinel M, Junier MP. (1998). J Neurosci 18: 10541–10552. Riese II DJ, Stern DF. (1998). BioEssays 20: 41–48. Rodriguez D, Gauthier F, Bertini E, Bugiani M, Brenner M, N’guyen S et al. (2001). Am J Hum Genet 69: 1134–1140. Ruegg UT, Burgess GM. (1989). Trends Pharmacol Sci 10: 218–220. Saas P, Walker PR, Quiquerez AL, Chalmers DE, Arrighi JF, Lienard A et al. (2002). NeuroReport 13: 1921–1924. Schlessinger J. (2004). Science 306: 1506–1507.

Schreiner B, Baur DM, Fingerle AA, Zechner U, Greten FR, Adler G et al. (2003a). Genes Chromosomes Cancer 38: 240–248. Schreiner B, Greten FR, Baur DM, Fingerle AA, Zechner U, Bohm C et al. (2003b). Oncogene 22: 6802–6809. Sibilia M, Steinbach JP, Stingl L, Aguzzi A, Wagner EF. (1998). EMBO J 17: 719–731. Soriano E, Del Rio JA, Auladell C. (1993). J Histochem Cytochem 41: 819–827. Sweeney C, Carraway III KL. (2004). Br J Cancer 90: 289–293. Takuma K, Baba A, Matsuda T. (2004). Prog Neurobiol 72: 111–127. Tang P, Steck PA, Yung WK. (1997). J Neurooncol 35: 303–314. Thomas LB, Gates DJ, Richfield EK, O’Brien TF, Schweitzer JB, Steindler DA. (1995). Exp Neurol 133: 265–272. Thompson KA, McArthur JC, Wesselingh SL. (2001). Ann Neurol 49: 745–752. Tokita Y, Keino H, Matsui F, Aono S, Ishiguro H, Higashiyama S et al. (2001). J Neurosci 21: 1257–1264. Waterman H, Sabanai I, Geiger B, Yarden Y. (1998). J Biol Chem 273: 13819–13827. Wechsler-Reya R, Scott MP. (2001). Annu Rev Neurosci 24: 385–428. Weickert CS, Blum M. (1995). Dev Brain Res 86: 203–216. Wiley HS. (2003). Exp Cell Res 284: 78–88. Wu CJ, Chen Z, Ullrich A, Greene MI, O’Rourke DM. (2000). Oncogene 19: 3999–4010. Xian CJ, Zhou XF. (1999). Mol Neurobiol 20: 157–183. Yaish P, Gazit A, Gilon C, Levitzki A. (1988). Science 242: 933–935. Yarden Y, Sliwkowski MX. (2001). Nat Rev Mol Cell Biol 2: 127–137. Yokoyama M, Black IB, Dreyfus CF. (1993). Exp Neurol 124: 377–380. Zelenaia O, Schlag BD, Gochenauer GE, Ganel R, Song W, Beesley JS et al. (2000). Mol Pharmacol 57: 667–678.

Oncogene