GLIA 50:270–275 (2005)
Estrogen Receptor-a Is Associated With the Plasma Membrane of Astrocytes and Coupled to the MAP/Src-Kinase Pathway JUSTYNA PAWLAK,1 MAGDALENA KAROLCZAK ,2 ANDRE KRUST,3 P. CHAMBON,3 AND CORDIAN BEYER 1* 1 Anatomisches Institut, Universita¨t Tu¨bingen, Tu¨bingen, Germany 2 Anatomie II, J.W. Goethe Universita¨t, Frankfurt, Germany 3 Institut de Genetique et de Biologie Mole´culaire et Cellulaire, CNRS/INSERM, Colle`ge de France, Illkirch Cedex, France
KEY WORDS
estrogen; astrocyte; signaling; mouse; ER-a; ko
ABSTRACT Estrogens influence CNS development and a broad spectrum of neural functions. Several lines of evidence also suggest a neuroprotective role for estrogen. Different modes of estrogen action have been described at the cellular level involving classical nuclear estrogen receptor (ER)-dependent and nonclassical membrane ER-mediated rapid signaling. We have previously shown that nonclassical estrogen signaling is implicated in the control of dopamine cell function and protection. Since nonclassical interactions between estrogens and glia may contribute to these effects, our aim was to demonstrate the presence of membrane-associated ERs and their putative coupling to intracellular signaling pathways in astrocytes. Confocal image analysis and fluorescence-activated cell sorting (FACS) studies indicated the attachment of ER-a but not ER-b to the plasma membrane of astrocytes. ERs were located in the cell soma region and glial processes. FACS analysis revealed that only a subpopulation of midbrain astrocytes possesses membrane ER-a. In FACS studies on ER-a knockout astrocytes, only a few membrane ER-positive cells were detected. The activation of membrane ERs appears to be coupled to the MAP-kinase/Src signaling pathway as shown by Western blotting. In conclusion, our data provide good evidence that nonclassical estrogen action in astrocytes is mediated by membrane ER-a. The physiological consequence of this phenomenon is not yet understood, but it might have a pivotal role in estrogen-mediated protective effects on midbrain dopamine neurons. V 2005 Wiley-Liss, Inc. C
The steroid hormone estrogen directs the cellular differentiation and controls the function of distinct neural networks (McEwen, 1992; Beyer, 1999). In addition, in vitro and in vivo approaches have proved the neuroprotective efficacy of estrogens by rescuing neurons from oxidative stress, chemically induced apoptosis, excitotoxic glutamate overflow, and ischemia-induced cell death (reviewed by Garcia-Segura et al., 2001; Behl, 2002; Kajta and Beyer, 2003). The general importance of estrogens for CNS development and function is underlined by the widespread expression of both nuclear estrogen receptors (ER-a/b) in different neural cells, including neurons as well as C 2005 V
Wiley-Liss, Inc.
astrocytes, oligodendrocytes, and microglia (Santagati et al., 1994; Azcoitia et al., 1999; McEwen et al., 2001). Concerning the intracellular action, there is now general acceptance that estrogens, just as other steroid hormones, can act in a classical way as transcription factors that involves binding to nuclear *Correspondence to: Cordian Beyer, Anatomisches Institut, Universita¨t Tu¨bingen, o¨sterbergstr. 3, D-72074 Tu¨bingen, Germany. E-mail:
[email protected] Received 18 August 2004; Accepted 29 October 2004 DOI 10.1002/glia.20162 Published online 11 February 2005 in Wiley InterScience (www.interscience. wiley.com).
MEMBRANE ESTROGEN RECEPTORS IN ASTROCYTES
receptors, its translocation to the nucleus, and interaction with the promoter of target genes. In contrast, estrogens can interact with membrane-associated and/or cytoplasmic receptors, thereby rapidly affecting intracellular signaling processes (Falkenstein et al., 2000; Beyer et al., 2003a). Such effects, termed rapid nonclassical or nongenomic estrogen effects, include the activation of different signal transduction systems in the brain, such as cAMP/protein kinase A (PKA), protein kinase C (PKC), phosphatidylinositol 3 kinase (PI3K), and Ca2þ-dependent calmodulin kinase (CaMK) (Beyer and Raab, 1998; Wong et al., 1996; Ku¨ppers et al., 2001). In addition, specific nonclassical estrogen action that always incorporates ER activation must be distinguished from nonspecific rapid estrogen effects comprising interactions with ion channels and ionotropic neurotransmitter receptors as well as the potency of estrogens to act as radical scavenger (Wong et al., 1996; Behl, 2002). It becomes increasingly apparent that astrocytes besides neurons represent important target cells for 17-b-estradiol. Astroglial cells express all types of estrogen receptors during development and in the adult brain, as mentioned above. With regard to the clinical aspects, a high number of human astrocytomas appear to be estrogen-sensitive and contain estrogen receptors (Gonzalez-Aguero et al., 2001). Developmental and neuroprotective estrogen effects in the CNS also seem to be mediated in part by estrogen–astroglia interactions (Garcia-Segura et al., 2001; Ku¨ppers et al., 2001). Thus, estrogens can regulate the expression of growth factors and interfere with numerous physiological properties of astrocytes (Garcia-Segura et al., 1999; Dhandapani and Brann, 2002; Beyer et al., 2003b; Sato et al., 2003). Despite the cumulative relevance of estrogen interactions with astroglia, information on the intracellular action of estrogen in this cell type is still scanty. Isolated and preliminary observations already suggest that nonclassical estrogen action may be found in astrocytes and implicated in the physiological regulation of glial cells (Smitherman and Sontheimer, 2001; Zhang et al., 2002; Beyer et al., 2003b; Sato et al., 2003). The aim of this study was therefore to demonstrate nonclassical estrogen action in and the presence of estrogen receptors at the plasma membrane of astroglial cells. Highly enriched astroglial cultures were established from the midbrain of 1-day-old BALB/c mice (Ivanova et al., 2001) or from ER-a knockout newborns (Dupont et al., 2000). Cultures consisted of 95% astrocytes and few oligodendrocytes (precursors) and were virtually free of neurons and microglial cells. In brief, tissues were dissociated using 0.1% trypsin and 0.02% EDTA in Dulbecco’s phosphate-buffered saline (PBS), then transferred to Hank’s-balanced salt solution (HBSS) containing 10% fetal calf serum (FCS), filtered through a 50-mm nylon mesh, centrifuged at 400g for 5 min, resuspended in minimal essential medium (MEM Biochrom, Germany), and finally seeded at a density of 1–2 106 cells/cm2. Upon reaching confluency, cells
271
were trypsinized and replated. After the second passage, cells were maintained with serum-free neurobasal medium (NBM) and used for experiments. Cultures from knockout mice were prepared in the same way, except that the midbrains of fetuses were not pooled but were cultured individually yielding single pub-derived cultures that were afterward analyzed by reversed transcription (RT) and polymerase chain reaction (PCR) for wild-type, heterozygote, or homozygote status, according to a previously published protocol (Dupont et al., 2000). Treatment with estrogen (108 M) was performed for 30 min. For immunoblotting, proteins were isolated in lysis buffer (65.2 mM Tris-HCl, 2% sodium dodecyl sulfate [SDS], 10% saccharose, 0.5 mM phenylmethylsulfonylfluorid [PMSF], 2 mg/ml aprotinin, and 0.5 mg/ml leupeptin). Samples were sonicated and denaturated at 958C for 5 min, and protein content was determined using the BCA protein estimation kit (Perbi Science, Germany). To separate proteins into a nuclear fraction and a fraction containing cytoplasmic and plasma membranes, harvested cells were centrifuged at 800g (10 min, 48C) to remove the nuclear fraction, and then at 100,000g (48C, 45 min) resembling the cytoplasmic/membrane fraction. The purity of both fractions was analyzed by immunoblotting for the nuclear marker Ki-67 and the adhesion molecule pancadherin (antibodies from Sigma), which were found exclusively in the respective fraction (data not shown). In this study, 20 mg protein of each sample was loaded onto SDS-polyacrylamide gels (10%). After electrophoretic separation, proteins were transferred to nitrocellulose membranes, blocked for 1 h with 5% nonfat dried milk in Tris-buffered saline (TBS) containing 0.5% Tween 20, and incubated with polyclonal antibodies directed against phosphorylated or nonphosphorylated extracellular signal-regulated kinase (ERK1/2, 1:1,000) and against phosphorylated or nonphosphorylated Src (1:1,000, both Cell Signaling, Germany) overnight at 48C. Filters were washed and incubated with secondary peroxidase-conjugated antibodies (anti-guinea pig IgG, 1:2,000; Jackson ImmunoResearch, Germany for pMAPK/MAPK/pSrc/Src). Blots were washed, and immunoreactive proteins were visualized with an enzyme-linked chemoluminescence kit (ECL, Amersham). Quantitative analysis of Western blots was accomplished densitometrically with a fluorescence scanner (ImageMaster, Pharmacia, Germany) using the manufacturer’s software (ImageMaster, Germany, USD version 2,0). Absolute optical densities (ODs) of pERK1/ 2 and pSrc were normalized to the ODs of the corresponding nonphosphorylated proteins and expressed in arbitrary units. For ER immunoblots, we used the antibodies mentioned below). For immunocytochemistry and double labeling, cells were fixed in 4% paraformaldehyde and incubated with polyclonal antisera against ER-a and -b (1:500, Santa Cruz Biochemical, Santa Cruz, CA) for 12 h at 48C and GFAP (1:500; Santa Cruz). As secondary antibodies, Cy3 and Oregon greenconjugated anti-rabbit IgGs were applied (1:500; Jackson ImmunoResearch). Immunocytochemistry was
272
PAWLAK ET AL.
Fig. 1. A: RT-PCR analysis demonstrates the presence of estrogen receptor (ER)-a and ER-b mRNAs in cultured midbrain astrocytes. B: Western blotting of mouse ovarian tissue (a), nuclear (b), and membrane/cytoplasmic fraction (c) of midbrain astrocytes probed with an antiserum against ER-a and ER-b. Note the presence of different ER-a isoforms in the ovary. In the nuclear fraction of astrocytes, ER-a-and ER-b-positive bands were observed with molecular weights of 65 and 60 kDa, respectively. In the cytoplasmic/membrane fraction, only a clear-cut ER-a IR-band is seen, whereas very
faint immunoreaction for ER-b is visible, most likely due to impurity during the experimental purification. C–E: Immunostaining and confocal microscopy of ER-a(C), ER-b/GFAP- (D), and ER-a/GFAP double-labeled astrocytes (E). Immunocytochemistry was performed with nonpermeabilized cells. ER-a immunoreactivity is found at the plasma membrane (red labeling in C and green arrowhead in E). Punctuated ER-a IR-clusters are located along astrocyte processes (E, white arrowheads). Note the absence of membrane labeling with an ER-b antiserum (D). Scale bars ¼ 5 mm.
performed with nonpermeabilized cells. Stained cells were viewed with a Leica TCS-SP confocal laser-scanning microscope. Reverse transcription-polymerase chain reaction (RT-PCR) analysis for ER-a/b mRNA was performed as previously described (Ivanova and Beyer, 2000). RNA isolation and RT was achieved using standard protocols. PCR was carried out with 3 ml samples of RT reaction, 0.2 mM sense/antisense primers, 0.2 mM dNTP, 2.5 mM MgCl2, and 1.2 U Taq polymerase (all from Gibco, Germany). The following primers were used: ER-a sense nt 1682–1702 and antisense nt 1961– 1981 (Genbank M38651); ER-b sense nt 10–31 and antisense nt 262–282 (Genbank AJ000220). The amplification protocol was: 36 cycles with denaturation for 1 min at 958C, annealing for 1 min at 628C, and elongation for 2 min at 728C. Specificity of reaction was tested by Southern analysis (not shown). For fluorescence-activated cell sorting (FACS) studies, cells were dispersed with a nonenzymatic ready-to-use cell dissociation solution (Sigma) for 5 min. After washing with PBS supplemented with 0.2% bovine serum albumin (BSA) and 0.1% sodium azide, cells were incubated with antibodies directed against ER-a and ER-b (1:200) for 30 min or with the membrane-impermeable FITC-coupled E-BSA construct (1 mg/ml Sigma) for 10 min. Anti-ER-labeled cells were then incubated with FITC-conjugated donkey anti-rabbit IgGs (1:500; Jackson ImmunoResearch) for
30 min. Cells were washed and transferred to the FACScan flow cytometer (Becton Dickinson, Germany) equipped with a 488-nm argon laser; 10,000 cells per experiment were scanned. Data analysis and graphic presentation were performed using Cellquest (Becton Dickinson) and the WinMDI software. Untreated cells were used to set the autofluorescent background. The polygonal gate R0 was defined for this cell population. The polygonal gate R2 was set for FITC-labeled cells. Membrane ER-positive cells were expressed as the number of cells shifted from R0 to gate R2. Dead cells were marked with propidium iodide (5 min, 2 mg/ml). We have investigated the expression and localization of ER-a/b at the plasma membrane of astrocytes using different experimental approaches. RT-PCR revealed the presence of ER-a and ER-b mRNAs in cultured astroglial cells (Fig. 1A). Western blot analysis after subcellular fractionation showed a single IR-band at 65-kDa resembling the full-length ER-a in the membrane/cytoplasmic fraction as well as in the nuclear fraction (Fig. 1B). A single ER-b protein was also found in the nuclear fraction but was almost undetectable in the membrane/cytoplasmic compartment (Fig. 1B). Membrane-associated binding sites for estrogen were visualized with an antisera directed against ER-a (Fig. 1C,E), but not with an ER-b antiserum (Fig. 1D). Specific surface staining was observed in the cell soma
MEMBRANE ESTROGEN RECEPTORS IN ASTROCYTES
273
Fig. 2. Fluorescence-activated cell sorting (FACS) analysis of midbrain astrocyte(A-C) and mixed astroglia-neuron cultures (D). A: Labeling with an antiserum directed against ER-b. Note that only autofluorescence of unmarked astrocytes (field R0, defined by the triangle between R2 and R1) and no shifting into the polygonal field R2 is observed. B: Labeling of a subpopulation of astrocytes with an antiserum against ER-a (shifting into R2). C: A massive reduction in the number of ER-a-labeled astrocytes is detected in ER-a knockouts. D: Simultaneous cell sorting with E-BSA-FITC (R1) and propidium iodide (R2) to demonstrate that estrogen membrane binding is different from dead cells. These data are from mixed neuron-glia midbrain cell cultures (panel D was previously published as fig 2D, J. Neurochem. 87: 545–550, 2003).
and along astrocyte processes, where a punctuated fluorescence signal was obtained (Fig. 1E). To rule out the possibility that the ER-b antiserum is insufficient for immunocytochemistry, we have performed immunostaining on hypothalamic slices from adult mice. This experiment revealed high numbers and a strong nuclear staining in the paraventricular and supraoptic nuclei (not shown). This is in agreement with an earlier study that has demonstrated high numbers of ERb-positive cells in these brain areas (Li et al., 1997). Membrane-associated ERs were also depicted on living astrocytes by applying FACScan (Fig. 2). A subpopulation of astrocytes (31 9% of all scanned cells) was assigned with an ER-a antiserum whereas no specific labeling (
