Involvement of Insulin/Phosphoinositide 3-Kinase/Akt Signal Pathway in 17β-Estradiol-mediated Neuroprotection*

In the present study, we tested the hypothesis that 17β-estradiol (βE2) is a neuroprotectant in the retina, using two experimental approaches: 1) hydrogen peroxide (H2O2)-induced retinal neuron degeneration in vitro, and 2) light-induced photoreceptor degeneration in vivo. We demonstrated that both βE2 and 17α-estradiol (αE2) significantly protected against H2O2-induced retinal neuron degeneration; however, progesterone had no effect. βE2 transiently increased the phosphoinositide 3-kinase (PI3K) activity, when phosphoinositide 4,5-bisphosphate and [32γATP] were used as substrate. Phospho-Akt levels were also transiently increased by βE2 treatment. Addition of the estrogen receptor antagonist tamoxifen did not reverse the protective effect of βE2, whereas the PI3K inhibitor LY294002 inhibited the protective effect of βE2, suggesting that βE2 mediates its effect through some PI3K-dependent pathway, independent of the estrogen receptor. Pull-down experiments with glutathione S-transferase fused to the N-Src homology 2 domain of p85, the regulatory subunit of PI3K, indicated that βE2 and αE2, but not progesterone, identified phosphorylated insulin receptor β-subunit (IRβ) as a binding partner. Pretreatment with insulin receptor inhibitor, HNMPA, inhibited IRβ activation of PI3K. Systemic administration of βE2 significantly protected the structure and function of rat retinas against light-induced photoreceptor cell degeneration and inhibited photoreceptor apoptosis. In addition, systemic administration of βE2 activated retinal IRβ, but not the insulin-like growth factor receptor-1, and produced a transient increase in PI3K activity and phosphorylation of Akt in rat retinas. The results show that estrogen has retinal neuroprotective properties in vivo and in vitro and suggest that the insulin receptor/PI3K/Akt signaling pathway is involved in estrogen-mediated retinal neuroprotection.

Oxidative stress-induced neuronal cell death has been implicated in different neurological disorders and neurodegenerative diseases (1,2). There is substantial evidence that light injures photoreceptors by increasing the formation of reactive oxygen species (3) and that antioxidants can rescue light-damaged photoreceptors (4 -6). The role of oxidant stress as a mediator of apoptosis has been examined (7,8), and hydrogen peroxide (H 2 O 2 ), a by-product of oxidative stress, has been implicated in triggering apoptosis in various cell types including cultured retinal neurons (9 -11). Apoptosis has been described in a wide variety of hereditary retinal degenerations (12,13), in light-damaged retinas (14,15), and following retinal detachment (16). Other types of retinal degeneration, such as retinal ischemia (17) and glaucoma (18), have also been associated with apoptosis. During the past decade, several therapeutic approaches, including retinal transplantation (19), gene therapy (20), growth factors (21,22), and antioxidants (4 -6), have been used to treat retinal degeneration. However, no effective medical therapy is currently available in humans, although vitamin A supplementation has shown some beneficial effect by slowing down the progression of retinitis pigmentosa (23).
The female sex hormone estrogen has a variety of metabolic activities including numerous effects on neurons (24). Therefore, the notion that estrogen is only important for sex differentiation and maturation has changed, to include its function as a neuromodulator and neuroprotectant. It appears that estrogen specifically maintains verbal memory in women and may prevent the deterioration in short and long term memory which is associated with normal aging (25). In addition, estrogen is not restricted to females because the male sex hormone testosterone (and other steroids with a 19-carbon atom structure, so-called C-19 steroids) can be converted chemically to estradiol in various tissues, including the brain, by an aromatase P450 enzyme (26).
17␤-Estradiol (␤E2) 1 is a steroid hormone synthesized enzymatically mainly in the ovaries from acetate, cholesterol, progesterone, and testosterone, but also by the placenta during pregnancy and, to a lesser extent, in the adrenal cortices, testes, and peripheral tissues (26). Several lines of evidence suggest that estrogen has neurotrophic and neuroprotective properties (27,28). For many years, it has been known that ␤E2 promotes viability and survival of neurons in primary neuronal cultures. Addition of ␤E2 to defined culture media increased the viability, survival, and differentiation of primary neuronal cultures from different brain areas including amygdala (29), hypothalamus (30), and neocortex (31). However, the role of estrogen in the protection of retinal neurons is not well understood. We have previously used both H 2 O 2 -induced cell death of cultured retinal neurons and light-induced photoreceptor degeneration as model systems to study the role of neurotrophic factors, pigment epithelium-derived factor, or basic fibroblast survival factor on retinal neuronal cell protection (11,22). In the present study, we have used in vitro and in vivo approaches to examine the role of estrogen as a neuroprotectant in the retina.

EXPERIMENTAL PROCEDURES
Primary Culture of Retinal Neurons-Timed pregnant Sprague-Dawley rats were ordered each week and the retinas of 10 -15 pups, 0 -2 days old, were removed with the aid of a dissecting microscope under sterile conditions in a tissue culture hood. The retinas were suspended in 25 ml of Dulbecco's modified Eagle's medium with F-12 medium plus 10% fetal calf serum in a plastic bag and mechanically dissociated for 2 min using a Stomacher set on low power. The suspension was first filtered through a 230-m sieve, which was then rinsed once with medium, and the combined filtrates were passed through a 140-m sieve followed by a rinse with undiluted fetal calf serum. The filtered suspension was centrifuged at 800 rpm in a clinical centrifuge for 5 min, the supernatant decanted, and the cell pellets resuspended in 25 ml of media using a sterile 5-ml pipette. The concentration of cells was determined with a cell counter or hemocytometer and the suspension diluted with medium to 1 ϫ 10 5 cells/ml. The cells (1 ml) were plated in 24-well tissue culture plates on 12-mm coverslips that had been pretreated overnight with 10 g/ml poly-D-lysine. The cells were maintained in either Dulbecco's modified Eagle's medium with F-12 medium and 2% fetal calf serum or in synthetic serum-free media. The cultures were used in experiments 7-10 days after plating.
Immunocytochemistry-Cells grown in culture on poly-D-lysinecoated coverslips were fixed for 30 min in 4% paraformaldehyde in 0.1 M Tris-buffered saline (TBS, pH 7.5) and then rinsed three times with 1.0 ml of 0.1 M Tris-HCl, pH 7.5, and maintained at 4°C in that buffer until processed for immunocytochemistry as described in a previous report (33). Briefly, nonspecific binding sites were blocked by 2% normal goat serum for 30 min. The cells were incubated overnight with antirecoverin antibody (1:5,000) and anti-estrogen receptor-␣ antibody (1: 100) in TBS with 1% bovine serum albumin, 1% goat serum, and 0.01% Triton X-100. Recoverin was detected using goat anti-rabbit IgG conjugated to Texas Red, whereas estrogen receptor-␣ was visualized using biotinylated goat anti-mouse IgG followed by incubation with streptavidin conjugated to fluorescein isothiocyanate. The control experiments were performed using normal IgGs at the same concentration as the primary antibodies as well as omitting primary antibodies as control for nonspecific labeling by secondary antibody. Cells treated without primary antibody were unlabeled. The cells on coverslips were then mounted with anti-fade mounting medium and viewed and photographed with an Eclipse 800 Nikon microscope equipped with fluorescence and Nomarski optics and a digital camera. The images were transferred and stored in a computer.
MTT Assay-As described previously (33), MTT (3,(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Sigma) was dissolved at a concentration of 5 mg/ml in phosphate buffered saline. Lysing buffer was prepared as follows. 20% w/v of SDS (Sigma) was dissolved at 37°C in a solution of 50% of each N,N-dimethylformamide (Sigma) and deionized water. 25 l of the 5 mg/ml stock solution of MTT was added to each well, and after 2 h of incubation at 37°C, 100 l of the lysing buffer was added. After an overnight incubation at 37°C, absorbance of the samples was read at 562 nm using a microtiter plate enzyme-linked immunosorbent assay reader.
TUNEL Assay-Detection of apoptosis using the TUNEL (TdT-mediated digoxigenin-dUTP nick-end labeling) method was carried out with a commercially available in situ apoptosis detection kit as described previously (11). Staining for the TUNEL assay was performed according to the manufacturer's protocol. TUNEL-positive cells were identified with a Nikon Eclipse 800 microscope, and images were captured by a digital camera and stored in a computer. The percentage of apoptotic cells was calculated by dividing TUNEL-positive cells by the total number of cells visualized by Nomarski optics in the same field. Three digitized images of similar total cell numbers were selected from each coverslip for counting and averaging and were considered as one independent experiment. Three independent experiments were then averaged.
DNA Fragmentation-DNA laddering was carried out essentially as described previously (34). The cells were homogenized in 1 ml of extraction buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 0.5% SDS, and 0.5 mg/ml freshly prepared proteinase K) using a Tissue Tearor TM (Biospec Products, Inc., Bartlesville, OK). Each sample was placed on ice for 20 min and then centrifuged at 15,000 ϫ g for 10 min. After centrifugation the supernatant from each sample was extracted with phenol/chloroform until the white precipitate was no longer visible in the aqueous fraction. This usually took three to six extractions. The genomic DNA was then precipitated overnight at Ϫ20°C with 0.1 volume of 3 M sodium acetate, pH 5.2, and 2.5 volumes of 100% ethanol. The samples were then centrifuged at 5,000 ϫ g for 20 min, and the resulting pellets were resuspended in 100 l of TE buffer (Tris-Cl and EDTA, pH 8.0). RNase A was then added to a final concentration of 20 g/ml, and the samples were incubated at 37°C for 2 h. Finally 3-5 l of each sample was run on a 2% agarose gel at 40 volts for 2 h.
PI3K Assay-Enzyme assays were carried out essentially as described previously (35,36). Briefly, assays were performed directly on total cells in 50 l of the reaction mixture containing 0.2 mg/ml phosphoinositide 4,5-bisphosphate, 50 M ATP, 0.2 Ci of [␥-32 P]ATP, 5 mM MgCl 2 , and 10 mM HEPES buffer, pH 7.5. The reactions were performed for 15 min at room temperature and stopped by the addition of 100 l of 1 N HCl followed by 200 l of chloroform/methanol (1:1, v/v). Lipids were extracted and resolved on oxalate-coated TLC plates (silica gel 60) with a solvent system of 2-propanol and 2 M acetic acid (65:35, v/v). The TLC plates were prepared by placing in 1% (w/v) potassium oxalate in 50% methanol (v/v) and baked in an oven at 100°C for 1 h before use. TLC plates were exposed to x-ray film overnight at Ϫ70°C, and radioactive lipids were scraped and quantified by liquid scintillation counting.
GST-p85 Fusion Proteins and Pull-down Experiments-Glutathione S-transferase (GST-p85-N-SH2) (314 -446 amino acids) fusion proteins were generated by PCR amplification of the p85␣ cDNA and cloned into a GST vector (37). The sequence of each clone was verified by DNA sequencing. All inductions yielded proteins of the expected size, as judged by Coomassie Blue staining. Pull-down experiments were performed as described previously (38), using 5 g of GST fusion proteins that had been absorbed onto GST-Sepharose 4B matrix. Cultured retinal neurons were incubated with GST/GST-p85 fusion proteins with continuous mixing at 4°C for 1.5 h. The Sepharose beads were washed three times in 500 l of HNTG buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and 10% glycerol)and centrifuged at 5,000 rpm for 1 min at 4°C. Proteins bound to GST-p85 were eluted by boiling in 2ϫ SDS sample buffer and subjected to SDS-PAGE. The gels were transferred to nitrocellulose membranes followed by Western blot analysis with anti-IR␤ antibody.
Light-induced Photoreceptor Degeneration-Female Sprague-Dawley albino rats were raised and maintained in a cyclic light environment (12 h on, 12 h off at an in-cage illumination of less than 10 lux). Ovariectomy was performed 2 weeks before exposure to constant light for 24 h at age 3-4 months. Constant light at an illumination level of 1,700 lux was provided by two 40-W white fluorescent light bulbs that were suspended 50 cm above the floor of the cage. During light exposure, rats were maintained in transparent polycarbonate cages with stainlesssteel wire bar covers. A water bottle was kept in the appropriate depression in the cage cover, but food was placed in the bottom of the cage on the bedding. Animals received intraperitoneal injection of ␤E2 (500 g/kg of body weight) or vehicle 1 h before exposure to constant light for 24 h.
Functional Rescue of Photoreceptor Cells Evaluated by Electroretinogram-Electroretinogram (ERG) recordings were performed as described previously (22). Briefly, animals were kept in total darkness overnight before the ERG recording. Pupils were dilated with 1% atropine and 2.5% phenylephrine HCl. Animals were anesthetized intra-muscularly with a ketamine/xylazine mixture. ERG responses were recorded with a silver chloride needle electrode placed in the cornea with 1% tetracaine topical anesthesia. A reference electrode was positioned at the nasal fornix and a ground electrode on the foot. The duration of light stimulation was 10 s, and the band pass was set at 0.3-500 Hz. 14 responses were measured with flash intervals of 20 s. For quantitative analysis, the B-wave amplitude was measured at saturating light intensity.
Morphological Protection of Photoreceptors Evaluated by Quantitative Histology-After electroretinographic testing, animals were killed by an overdose of carbon dioxide. Eyes were enucleated, fixed, and embedded in paraffin, and 5-m-thick sections were taken along the vertical meridian to allow comparison of all regions of the eye. Outer nuclear layer (ONL) thickness was measured at nine defined points along the vertical meridian in the superior and inferior hemispheres, using the optic nerve head as a point of reference. The distance between each point was 450 m. In addition to the mean ONL thickness for the entire retinal section, ONL thickness of the region of retina most sensitive to the damaging effects of light was compared among different groups of rats. In each of the experiments where ONL thickness was quantified, a single section from the retinas of at least 6 (usually 10 or more) eyes was measured.
Serum Levels of ␤E2-Serum (50 l) was removed by tail vein 1 h after intraperitoneal injection of ␤E2, and ␤E2 was measured in triplicate by a competitive enzyme immunoassay kit (Estradiol EIA kit, Cayman Chemical Company, Ann Arbor, MI).
Statistical Analysis-Data were analyzed by means of analysis of variance and assessed further by Dunnett tests. Statistical significance was set at p Ͻ 0.05.

RESULTS
␤E2 Attenuates H 2 O 2 -induced Cytotoxicity-We tested the cytoprotective effects of ␤E2 in our well characterized retinal neuronal culture system, using H 2 O 2 to generate an oxidant stress. We first examined the cytotoxicity of ␤E2 on neurons by treating the cultures with different concentrations of ␤E2 ranging from 0.0001 to 100 M. A significant reduction in cell viability was observed only at the relatively high 100 M concentration (Fig. 1B). To evaluate the role of ␤E2 in protecting retinal neurons from H 2 O 2 -induced cytotoxicity, we pretreated retinal neuron cultures with different concentrations of ␤E2 (ranging from 0.001 to 10 M) 30 min prior to H 2 O 2 (100 M) treatment; significant increases in cell viability were observed between 0.1 and 10 M of ␤E2 (Fig. 1C). We used absolute ethanol to dissolve ␤E2. When culture medium contained 10 M ␤E2, the concentration of ethanol was 0.01%. There was no significant cytotoxicity of 0.01% ethanol treatment for up to 24 h as measured by the MTT assay (Fig. 1A).
Effect of the Estrogen Receptor Inhibitor Tamoxifen-To investigate whether the cytoprotective effects of ␤E2 are mediated through the estrogen receptor, we repeated the experiments described above in the presence of tamoxifen, an estrogen receptor blocker. Fig. 2A shows that additions of tamoxifen at 5 M for 24 h did not significantly affect cell viability, although decreases in cell viability were observed at 10 and 100 M concentrations. Pretreatment with 5 M tamoxifen for 30 min prior to the addition of ␤E2 did not significantly block the neuroprotective effect of ␤E2 in this study (Fig. 2B). Immunocytochemical analysis of cultured primary retinal neurons showed that a very small population of cells (less than 1%) was estrogen receptor-positive. These cells were recoverin-negative, indicating that they were not photoreceptor cells, and were usually double or triple the size of recoverin-positive cells (Fig. 3).   (Fig. 4B,  lane 9).
Effects of 17␣-Estradiol (␣E2) and Progesterone-The specificity of the ␤E2 effect was determined by testing two other steroids, ␣E2, an isomer of ␤E2 that does not activate the estrogen receptor, and progesterone. Neither was cytotoxic to the cells at the concentration of 10 M when added in the absence of H 2 O 2 . A few positive staining cells were noted in control cultures (Fig. 5A), whereas large numbers of cells undergoing apoptotic cell death (Fig. 5B) were present in cultures  (Fig. 5D). However, pretreatment with 10 M progesterone for 30 min prior to H 2 O 2 (100 M) exposure did not appear to inhibit apoptosis induced by H 2 O 2 significantly (Fig.  5F). Quantification of the number of apoptotic cells showed that cultures treated with either ␣E2 or progesterone had no more apoptotic cells than untreated cultures (4 -7%). However, pretreatment with ␣E2 decreased the percentage of apoptotic cells in H 2 O 2 -treated cultures from 49 to 22%, whereas progesterone-pretreated cultures remained 42% TUNEL-positive (Fig. 5G). Thus, ␣E2 significantly reduced the number of apoptotic cells, although to a lesser extent than ␤E2 (Fig. 4A). Another indication of the effectiveness of ␣E2 was that it significantly reduced DNA fragmentation, whereas progesterone did not (Fig. 4B, lanes 5 and 7).
Morphological and Functional Evaluations of Photoreceptor Cell Rescue in Vivo-Neuroprotection by systemic administration of ␤E2 was evaluated by quantitative histology, ERG, and TUNEL assay. 24 h of exposure to fluorescent light (1,700 lux) reduced the thickness of the ONL of photoreceptor cell nuclei from the normal 10 -13 rows in control animals (Fig. 6, A and B) to 3 rows in the most degenerated region of the retinas in vehicle-injected animals (Fig. 6C). However, in ␤E2-treated animals, there was significant rescue of photoreceptors with the ONL having ϳ7-8 rows of nuclei (Fig. 6D). Quantitative analysis of ONL thickness as a function of the retinal location showed a significant protection of photoreceptors by systemic injection of ␤E2 across the entire retina (Fig. 7A). ␤E2 also protected retinal function, as demonstrated in the ERG tracings in Fig. 6. Both A-and B-wave responses were greater in the light-stressed rats given ␤E2 than in controls given a placebo. Measurement of B-wave amplitudes at different flash intensities clearly demonstrates the functional protection provided by ␤E2 (Fig. 7B).
The TUNEL assay was performed to determine whether ␤E2-mediated photoreceptor protection is through inhibition of apoptosis. Fig. 8A showed intensive TUNEL-positive cells in the ONL in the superior region (the most degenerated area) of the retinas in vehicle-treated animals after exposure to constant light for 24 h at an illumination of 1,700 lux. Systemic injection of ␤E2 significantly reduced light-induced photoreceptor cell apoptosis in this region (Fig. 8B).
Involvement of Insulin/PI3K/Akt Signal Pathway-The PI3K cascade has been shown to provide neuroprotection to stressed neuronal cells (39). PI3K activity was increased by a 30-min  treatment of cultured retinal neurons with ␤E2 and reached a plateau at 1 and 3 h (Fig. 9A), then activity decreased after 6 h and returned to the control level after 12 h. A similar pattern of PI3K activation was observed in rat retina after systemic injection of ␤E2 (Fig. 9B, upper panel). Synthesis of phosphoinositide trisphosphate increased up to 12 h and returned to base line by 24 h. Over the same time course, Western blots showed that the expression of the p85␣ regulatory subunit of PI3K did not respond to ␤E2 treatment (Fig. 9B, lower panel), indicating that the increased enzymatic activity of PI3K was not caused by increased p85 expression in the retina.
Pretreatment with LY294002, a PI3K inhibitor, for 30 min prior to the addition of ␤E2 greatly inhibited the ␤E2-induced protective effect (Fig. 10C). Neither the carrier (dimethyl sulfoxide) nor 100 M LY294002 drug was cytotoxic after 24 h (Fig.  10, A and B). These results suggest that the ␤E2 cytoprotective effect may be mediated through some downstream effector generated by activation of the PI3K pathway. The cytoprotective effect of the two estrogen isomers and the absence of a tamoxifen effect as well as the absence of estrogen receptor expression in the photoreceptor cells (Fig. 3) suggested that the estrogen receptor is not involved in the estrogen-mediated photoreceptor protection.
Recently it has been shown that IR␤ is involved in the regulation of PI3K activity in the retina (37,38). We examined whether IR␤ activation may be involved in the estrogen effect. We presented preliminary data on this at the Retinal Degeneration meeting held at Burginstock, Switzerland (40) that 1) IR␤ is expressed in our cultured rat retinal neurons; 2) 5 M ␤E2 and 5 M ␣E2 activated IR␤; and 3) 5 M progesterone was ineffective. In the present study, using GST (GST-p85-N-SH2) pull-down experiments to study IR␤ phosphorylation (activation), we further demonstrated that both insulin and ␤E2 (Fig.  11A) activated IR␤. The data from in vivo experiments showed that both IR␤ and insulin-like growth factor-I receptor (IGF-IR) were present in the retina (Fig. 11B). Systemic administration of ␤E2 activated IR␤ but not IGF-IR (Fig. 11C) using GST (GST-p85-N-SH2) pull-down experiments. We also demonstrated by immunohistochemistry that insulin receptor is present in the inner and outer segments of photoreceptor cells (Fig.  12). In addition, we found that the insulin receptor blocker, HNMPA (200 M), did not completely inhibit ␤E2-activated PI3K (Fig. 13A), and incubation of ␤E2 (10 M) directly with a homogenate of cultured retinal neurons did not activate PI3K (Fig. 13B).
It is well known that Akt is the downstream target of PI3K in receptor-mediated signal cascades and is activated by phosphorylation (41). To determine whether Akt can be activated by ␤E2, retinal neuron cultures were treated with 10 M ␤E2 for different times (0.5, 1, 3, 6, and 12 h), and extracts were subjected to Western blot analysis with anti-phospho-Akt (pS473) antibody (1:2,000) (preliminary data presented at the Retinal Degeneration meeting held at Burginstock, Switzerland (40)). As shown in Fig. 14A, addition of 10 M ␤E2 to the culture for 30 min significantly increased phosphorylated Akt expression more than 2-fold, and this increase was maintained for at least 3 h before declining to 1.5-fold by 6 h. Systemic injection of ␤E2 at the concentration of 500 g/kg also transiently increased pAkt level in vivo (Fig. 14B, upper panel). Over the same time course, Western blots showed that the expression of the pan-Akt protein did not respond to ␤E2 treatment (Fig. 14B, lower panel), indicating that the increased pAkt was not caused by increased Akt expression in the retina. DISCUSSION In the present study, we demonstrated that estrogen has a neuroprotective effect in the retina which is mediated via the insulin/PI3K/Akt signal transduction pathway. Although it has been known that ␤E2 promotes viability and survival of other primary neuronal culture systems, such as cortex (42), hippocampus (43), hypothalamus (30), this is the first observation demonstrating that ␤E2 attenuated H 2 O 2 -induced cytotoxicity, inhibits light-induced photoreceptor apoptosis, and activates the IR␤/PI3K/Akt signaling pathway.
In some tissues, neuroprotection of ␤E2 may be mediated by the estrogen receptor and protein synthesis (27,44) because ␤E2 can bind intracellular specific estrogen receptors, and the complex binds to specific sites on genomic DNA and control its transcription. Indeed, ␤E2 was reported to provide neuroprotection mediated by estrogen receptors in cultured cortical neurons (45) and in hippocampus-derived cell line (46). We observed that a competitive estrogen receptor antagonist, tamoxifen, did not significantly attenuate the protection provided by ␤E2. Although the absence of any effects by tamoxifen did not completely exclude the possibility that the neuroprotection was mediated by estrogen receptors, ␣E2, a biologically inactive stereoisomer that has little effect as a female sex steroid hormone, also provides neuroprotection of retinal neurons against H 2 O 2 -induced apoptotic cell death. Furthermore, we could not detect estrogen receptors in photoreceptor cells in vitro. These results suggest that neuroprotective effects of ␤E2 in our culture system do not involve activation of estrogen receptors.
In addition to the activation of genome transcription mediated by estrogen receptors, estrogens have been reported to activate IGF-IR kinase, resulting in enhanced binding of p85, the regulatory subunit of PI3K, to insulin receptor substrate-1 and -2 in the mouse uterus (47). However, we found that ␤E2 does not activate IGF-IR in the retina. Insulin receptors are present in neuronal retina (48), and the interaction of the IR␤ with PI3K has been demonstrated (38,49). As shown in Fig  11A, both ␤E2 and insulin activate the IR␤. Whether the estrogen effect on IR␤ is direct or indirect is not known at present. However, insulin implants significantly reduced the number of apoptotic cells in retinas of streptozotocin diabetic rats (50), and insulin has been shown to prevent apoptotic cell death of proliferating neuroepithelial cells in the embryonic retina (51) and to rescue retinal neurons by inhibiting caspase-3 (50).
Studies examining the role of PI3K, mainly using inhibitors, suggest involvement of PI3K in numerous biological responses that encompass the regulation of cell growth (52). It is clear in the present study that ␤E2 significantly increased PI3K activity in cultured retinal neurons and in the retina in vivo. This neuroprotective effect provided by ␤E2 can be blocked by a PI3K inhibitor, suggesting a direct role of PI3K in estrogenmediated neuroprotection. Akt, also known as protein kinase B, is a serine/threonine, mitogen-regulated protein kinase involved in the protection of cells from apoptosis (53). Anti-Akt pS473 is a polyclonal antibody developed against the singly phosphorylated Akt. We used this antibody to show clearly that addition of ␤E2 to retinal neuronal cultures significantly activates Akt, suggesting the possible involvement of Akt in ␤E2mediated retinal neuronal protection. It is generally thought that insulin acts as a neurotrophic factor via the PI3K/Akt pathway, which in turn inhibits many proapoptotic targets (49,54). These observations led to the hypothesis that estrogenmediated insulin receptor/PI3K/Akt signaling could be one of the mechanisms of ␤E2-mediated retinal neuron protection. Because the insulin receptor blocker did not completely inhibit ␤E2-activated PI3K, this suggests that additional mechanisms might be involved in this protection. It has been reported that the extracellular signal-regulated kinase/mitogen-activated protein kinase signaling pathway is involved in estrogen-mediated cell survival (55,56). We have found that systemic injection of ␤E2 increased Ras activity in a time-dependent manner in mouse retina. 2 The effects of ␤E2 and ␣E2 on the insulin receptor are likely not through a nonspecific membrane effect, based on following evidence: 1) ethanol and acetic acid did not activate insulin receptor; 2) progesterone at the same concentration as ␤E2 did not activate the insulin receptor; 3) in vivo injection of ␤E2 activates insulin receptor but not the IGF-IR. The mechanism that triggers the phosphorylation of IR␤ in response to ␤E2 and ␣E2 is not known. Recently light-induced tyrosine phosphorylation of IR␤ independent of insulin in the retina has been reported (37). We speculate that there could be at least two possible mechanisms that lead to the phosphorylation of IR␤. The first could involve ligand(s) other than insulin, which are induced or released in response to estrogens. The second mechanism involves the activation of a nonreceptor tyrosine kinase(s) in response to estradiols. Nonreceptor tyrosine kinase Src phosphorylates insulin receptors and IGF-IRs on autophosphorylation sites, and Src kinase has been shown to substitute for the ligand-dependent receptor activation (59,60). Consistent with this mechanism, we have also reported previously the in vitro phosphorylation of IR␤ by c-Src in reactive oxygen species (38). It has been shown that ␣E2-induced vascular endothelial growth factor-A gene expression in rat pituitary tumor cells is mediated through an estrogen receptor-indepen-2 W. Cao, unpublished observation. dent but PI3K-Akt-dependent signaling pathway (61). Akt activation has also been shown by estrogen in estrogen receptornegative breast cancer cells (32). In those studies, insulin receptor phosphorylation was not studied.
It is worth noting that in our in vitro primary rat retinal neuronal system, the concentration of ␤E2 needed to provide a neuroprotective effect is much higher than the normal estrogen level in vivo in rat serum (0.6 nM). Cultured retinal neurons survive in a chemically defined serum-free environment that is completely different from in vivo conditions. The advantage of using an in vitro system is that the experiments can be well controlled in the chemically defined environment to study complex and specific mechanisms like hormonal or drug responses. However, to determine physiological relevance, we further tested the effect of ␤E2 in vivo. The serum level of ␤E2 in ovariectomized rats with systemic administration of ␤E2 at the dose of 500 g/kg of body weight was 2.1 nM. This dose was sufficient to protect the retina against light-induced photoreceptor degeneration through inhibition of apoptosis and transiently increase PI3K activity and the pAkt level. These in vivo findings further support the hypothesis that insulin/PI3K/Akt is involved in estrogen-mediated neuroprotection and is correlated with our in vitro studies. Experiments are under way in our laboratory to help an understanding the molecular mechanisms underlying the neuroprotective effect of estrogen in the retina.