Introduction

Top Summary Introduction Materials & Methods Results Discussion References

An increasing body of evidence indicates that oxidative damage plays a key role in the pathological events occurring in AD (Good et al., 1996; Yan et al., 1996), with use of antioxidants [i.e., any substance that under physiological conditions significantly delays or inhibits oxidation when present in low concentrations compared with those of an oxidizable substrate (Halliwell and Gutteridge, 1989)] emerging as a possibly useful therapy. Substances that fit this description and that are currently used in clinical trials include vitamin E and estrogens. A role for estrogens in the prevention of AD has been implicated by a retrospective epidemiological study, showing a dose- and duration-dependent relationship of ERT with a reduction in the incidence of AD (Paganini-Hill and Henderson, 1994). Recent studies revealed that ERT delays the onset of AD symptoms regardless of the ethnic origin of the subjects (Tang et al., 1996b). As such, there is much to be gained in elucidating the mechanism by which ERT alters the course of AD.

In some cases of familial AD, genetic mutations result in the increased production of AP (Selkoe, 1997), a 39-43 amino acid peptide which upon aggregation (Pike et al., 1991) and contact with the plasma membrane (Mattson et al., 1993) is toxic to neurons in culture. The toxic portion of this peptide seems to be an 11-amino acid sequence (AP 25-35) (Yankner et al., 1989), and increased amounts of H₂O₂ (Behl et al., 1994) and lipid peroxides (Behl et al., 1992; Goodman et al., 1996; Gridley et al., 1997; Keller et al., 1997) result from a AP 25-35 challenge. We have observed (Green et al., 1996) that this fragment rapidly aggregates, causing cell death with a dose-dependence and time-course similar to that reported for in primary neurons and B12 cells (Behl et al., 1994). This provides an in vitro model system for studying the mechanism by which estrogens ameliorate APinduced toxicity on a variety of cell types.

Estrogens have long been recognized as antioxidants in a variety of in vitro and in vivo models, and evidence is emerging to suggest that their antioxidant activity is involved in their neuroprotective capacity. Structure-activity relationship studies revealed that a phenolic A ring (Behl et al., 1997; Green et al., 1997b) and at least three rings of the steroid structure (Green et al., 1997b) are required for the molecule to demonstrate neuroprotection. Additionally, estrogens reduce the cellular toxicity of AP and other oxidative insults (Behl et al., 1995; Goodman et al., 1996; Green et al., 1996) and we have recently demonstrated that this correlates to a decrease in the amount of generated lipid peroxidation (Gridley et al., 1997). The mechanism by which estrogens act as an antioxidant at higher concentrations may directly relate to their ability to scavenge free radicals (Mooradian, 1993). We hypothesized that estrogens at physiologically relevant concentrations may be participating in a cycle whereby estrogens are regenerated by other endogenous antioxidants. Because reparation of lipid membranes relies on the glutathione peroxidase enzyme system (Meister and Anderson, 1983), and low concentrations of estrogen can reduce lipid peroxidation (Gridley et al., 1997), we targeted glutathione as a likely candidate for this interaction. The present studies were undertaken to investigate the possibility of an interaction between E2 and glutathione in protecting cells against the presence of AP 25-35.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Materials. Lyophilized AP 25-35 (1 mg; Bachem, Torrance, CA) was initially dissolved in 200 µl of double deionized H₂O and with the addition of 800 µl of PBS, rapid aggregation was observed. E2 (Steraloids, Wilton, NH) was initially dissolved at 10 mg/ml in absolute ethanol (Fisher Scientific, Orlando, FL) and diluted in cell culture media to obtain the necessary concentrations. ICI (Zeneca, Chesire, England) was dissolved in absolute ethanol and spiked into individual cell culture wells to obtain the 200 nM concentrations. -Tocopherol acetate was initially dissolved in 200 µl of absolute ethanol and diluted in cell culture media to the appropriate concentrations. Lipoic acid (thiotic acid), taurine (2-aminoethanoic acid), and ascorbic acid were initially dissolved in cell culture media and used at the concentrations indicated. Unless otherwise noted, materials were obtained from Sigma Chemical Corp (St. Louis, MO).

SK-N-SH neuroblastoma cell culture. SK-N-SH neuroblastoma cells were obtained from American Type Culture Collection (Rockville, MD). Cell cultures were grown to confluency in RPMI-1640 media (Fisher Scientific, Pittsburgh, PA) supplemented with 10% charcoal/dextran-treated FBS, (Hyclone, Logan, UT), 100 units/ml of penicillin G and 100 mg/ml of streptomycin (Sigma) in monolayers in Corning 150-cm² flasks (Fisher Scientific) at 37° under 5% CO₂, 95% air. Media was changed three times weekly. Cells were observed with a phase contrast microscope (Nikon Diaphot-300; Nikon, Tokyo, Japan).

Primary rat cortical cultures. Primary neuronal cultures were prepared according to methods described elsewhere (Chandler et al., 1993). Briefly, Female Sprague-Dawley rats (Charles River Farms, Wilmington, MA) were housed and bred in our animal facility. Primary cortical neurons were prepared from 1-day-old rat pups as follows: brain tissue was removed from rat pups and placed in isotonic salt solution containing 100 units of penicillin G, 100 µg of streptomycin and 0.25 µg of amphotericin B (Fungizone; Life Technologies) per ml (pH 7.4). After removal of blood vessels and pia mater, the tissue was sectioned into approximately 2-mm chunks, suspended in 25 ml of 0.25% trypsin (weight/volume) in isotonic salt solution (pH 7.4), and placed in a shaking water bath for 10 min at 37° to dissociate the cells. The dissociated cell suspension was then removed and combined with 10 ml of DMEM containing 10% PDHS (Central Biomedia, Irwin, MO) and the undissociated chunks were mixed with 160 µg of DNase 1 and triturated until the cells dissociated. The cell suspensions were then combined, centrifuged at 1000 × g for 10 min, and the resulting cell pellet washed with 50 ml of DMEM with 10% PDHS and plated on precoated poly-L-lysine 35-mm culture dishes at a density of 4 × 10⁶ cells per dish and incubated in a humidified incubator containing 95% air and 5% CO₂ at 37°. On day 3, cells were treated with -cytosine arabinoside (10 µM) for 48 hr and media was then aspirated and replaced with DMEM containing 10% PDHS and incubated for an additional 5 days before being used in experiments. At this time, cultures contain approximately 90% neurons and 10% astroglia. These appeared as many phase-contrast bright cells with characteristic neuronal morphology overlaying a number of flat phase dark cells that had typical astroglial morphology.

Statistics. The significant treatment effects on cell viability were determined using ANOVA followed by Scheffé's post hoc test, with significance determined at p < 0.05. For dose-response evaluations, EC₅₀ values were calculated by randomly assigning cell counts at the indicated doses to generate 3-5 lines per treatment and determining the average value for those lines. Mann-Whitney rank sum nonparametric analyses were used on EC₅₀ values because the variances for the standard deviation were not equivalent. Comparisons between dose response relationships were calculated using two-way ANOVA to determine the significance of GSH or E2 presence or absence.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

The neuroprotective capability of physiologically relevant concentrations of E2 relies on the presence of glutathione in the cell culture milieu (Fig. 1). Using DMEM media, which lacks GSH in the cell culture recipe, and RPMI-1640 manufactured specifically without GSH, we demonstrated that the addition of AP 25-35 reduced the number of viable cells by 41% and 47%, respectively, in these culture media. Concomitant treatment with E2 (2 nM) had no effect, which contrasted with our earlier work that showed these same concentrations of E2 to be neuroprotective in RPMI-1640 media (Green et al., 1996). Based on our hypothesis that GSH may play a role in this system, we supplemented GSH (3.25 µM) to the cell culture milieu. The addition of GSH was not neuroprotective alone; when added with low concentrations of E2 (2 nM), however, it increased the number of viable cells by 88% and 47% in DMEM and RPMI-1640 lacking GSH, respectively.

View larger version (54K): [in this window] [in a new window]   Fig. 1.   Effects of 17- estradiol (E2; 2 nM) and GSH (3.25 µM) on the toxicity induced by AP 25-35 (20 µM) in different cell culture media that lack GSH in the cell culture recipe. Cells were plated at 106 cells/ml and were exposed to vehicle, AP 25-35, E2, GSH, or a combination as indicated for 72 hr. Controls, cell numbers for the vehicle, GSH, E2, and GSH+E2 control groups (no AP 25-35) pooled together after it was determined they were not statistically different from each other. Depicted are the mean values ± standard error for four or five wells/group. *, p < 0.05 versus controls determined by ANOVA followed by Scheffé's post hoc test.

To further evaluate the interaction noted between E2 and GSH, we assessed the neuroprotective capacity of GSH (Fig. 2). We noted that in the absence of estrogen, high concentrations of GSH were necessary to achieve neuroprotection, with a significant reduction (50%) in AP 25-35-induced toxicity at the 325 µM dose. By contrast, in the presence of E2 (2 nM), the neuroprotective dose of GSH was reduced to 0.325 µM. Two-way ANOVA revealed a highly significant effect of E2 on the GSH induced neuroprotection (F, 41.4; p < 0.001). The neuroprotective EC₅₀ value generated for GSH without E2 in the media was 82.6 ± 60 µM, contrasted by 0.04 ± 0.02 µM when E2 was present, an approximate 2000-fold increase in GSH potency in the presence of E2. Using this information, we analyzed the ability of E2 to increase the number of viable cells exposed to AP 25-35 (Fig. 3). In the absence of GSH, high concentrations of E2 (200 nM) were necessary for significant neuroprotection (Fig. 3). However, in the presence of a nonprotective dose of GSH (3.25 µM), the neuroprotective dose of E2 shifted from 200 nM to 0.2 nM (Fig. 3). Again, two-way ANOVA revealed a highly significant effect of the presence of GSH on the neuroprotective effect of E2 (F, 44.33; p < 0.001). The calculated EC₅₀ values likewise shift to the left, from 126 ± 87 nM in the absence of GSH to 0.033 ± 0.031 nM in the presence of GSH, a ~4000-fold increase in the potency of E2 in the presence of GSH.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Effects of GSH concentration on live SK-N-SH cell number subjected to a AP 25-35 (20 µM) challenge in the presence and absence of a nonprotective dose of E2 (2 nM). Cells were plated at 106 cells/ml and exposed to treatments at the doses indicated. Controls represents cell numbers for the vehicle, GSH, E2, and GSH+E2 control groups (no AP 25-35), pooled together after it was determined they were not statistically different from each other. Depicted are the mean values ± standard error for six wells per group. When error bars are absent, they are smaller than the symbol used to depict mean values. The effect of E2 on the response to GSH was highly significant (F, 41.48; p < 0.001). A comparison of the EC50 values for the different dose response curves using the Mann-Whitney rank sum test showed p = 0.036.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Effects of E2 concentrations on live SK-N-SH cell number subjected to a AP 25-35 (20 µM) challenge in the presence and absence of a nonprotective dose of GSH. Cells were plated at 106 cells/ml and exposed to treatments at the doses indicated. Controls, cell numbers for the vehicle, GSH, E2, and GSH+E2 control groups (no AP 25-35), pooled together after it was determined they were not statistically different from each other. Depicted are the mean values ± SEM for four wells per group. When error bars are absent, they are smaller than the symbol used to depict mean values. The effect of GSH on the response to E2 was highly significant (F, 44.33; p < 0.001). A comparison of the EC50 values for the different dose response curves using the Mann-Whitney rank sum test showed a p < 0.057.

To ensure that this synergy was not caused simply by cell origin or tumorigenicity, we performed similar experiments in rat primary cortical neurons. Again, the ability of E2 to protect neurons was evaluated in the presence and absence of GSH (3.25 µM) (Fig. 4). The addition of AP 25-35 (10 µM) to primary cortical neurons resulted in a 39% to 40% reduction in the average number of viable cells per field in the absence and presence of GSH, respectively (Fig. 4). When increasing concentrations of E2 were evaluated against AP 25-35, 200 nM E2 was the lowest concentration found to be protective (Fig. 4), which is in full agreement with our SK-N-SH cell line studies (Fig. 3). With the addition of GSH (3.25 µM), all concentrations of E2 of 2 nM or higher were neuroprotective (Fig. 4). Evaluation of EC₅₀ values demonstrated similar changes in potency, from 68.1 ± 79 nM in the absence of GSH, to 4.3 ± 5.9 nM in the presence of GSH. Likewise, evaluation of the effect of GSH on the neuroprotective effect of E2 in rat primary cultures using two-way ANOVA demonstrated a significant effect (F, 8.53; p < 0.005).

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Effects of concentrations of E2 on the average number of live rat primary cortical neurons per photographic field when subjected to a AP 25-35 (10 µM) challenge in the presence and absence of a nonprotective dose of GSH. Cells were plated and exposed to treatments as previously indicated at the doses noted. Controls, cell numbers for the vehicle, GSH, E2, and GSH+E2 control groups (no AP 25-35), pooled together after it was determined they were not statistically different from each other. Depicted are the mean values ± standard error for four to seven plates per group. The effect of GSH on the response to E2 was highly significant (F, 8.53; p < 0.005). A comparison of the EC50 values for the different dose response curves using the Mann-Whitney rank sum test showed p < 0.057.

We examined the specificity of this estrogen-antioxidant interaction for the GSH system by evaluating protection of SK-N-SH cells from AP 25-35-induced toxicity. This was achieved by addition of a concentration of E2 (20 nM) that was at or near the protective threshold in the presence or absence of nonprotective concentrations of other well known antioxidants (Table 1). In four separate experiments, the reduction in viable cells ranged from 35% to 68% when subjected to a AP 25-35 (20 µM) challenge. The 20 nM dose of E2 alone was either nonprotective or slightly protective (Table 1). No significant effect was observed on AP 25-35-induced toxicity with the addition ascorbic acid (100 µM), -tocopherol acetate (50 µM), taurine (5 mM), or lipoic acid (10 µM) alone or in combination with E2 (Table 1). However, the addition of GSH (3.25 µM) or the oxidized form of glutathione (1.5 µM) was enough to significantly reduce the cytotoxicity of AP 25-35 when 20 nM E2 was present (Table 1).

View this table: [in this window] [in a new window]   TABLE 1 Effects of E2, a variety of antioxidant treatments, and their combination on the AP 25-35-induced toxicity in SK-N-SH live cell number after 72 hr of treatment.

The importance of an estrogen receptor in this interaction was determined by using ICI as an antiestrogen (Fig. 5). Again, AP 25-35 (20 µM) was added to SK-N-SH cells in the presence and absence of E2 (2 nM), GSH (3.25 µM), and/or ICI (200 nM). AP 25-35 reduced viable cells after 72 hr of exposure by 54% when compared with vehicle controls. Using concentrations of E2 with GSH that were not protective when administered alone but were neuroprotective when added together, the addition of ICI in 100-fold excess of the E2 concentration did not significantly alter the protective effects of E2 and GSH in combination. Indeed, ICI addition alone exerted neuroprotective activity.

View larger version (69K): [in this window] [in a new window]   Fig. 5.   Effects of the antiestrogen ICI (200 nM) on 20 µM AP 25-35-induced toxicity in SK-N-SH cells using doses of E2 (2 nM) and GSH (3.25 µM), which alone are not protective but will exhibit protection when used in combination. Cells were plated at 106 cells/ml and were exposed to vehicle, AP 25-35, E2, GSH, or a combination as indicated for 72 hr in the presence and absence of ICI. Depicted are the mean values ± standard error for three wells/group. *, p < 0.05 versus controls determined by ANOVA followed by Scheffé's post hoc test.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

It is the purpose of this article to report a novel synergistic interaction between E2 and glutathione for neuroprotection. This interaction is independent of species origin and the tumorigenicity of the cells, as we have demonstrated protection in a human neuroblastoma cell line, in rat primary cortical neurons (present report), and the HT-22 transformed mouse hippocampal cell line (Green et al., 1998). Furthermore, this effect seems to be independent of the type of cytotoxic insult used, as we have observed the same synergy with both serum deprivation and zinc toxicity (Gridley KE, Pariokar KS, Simpkins JW unpublished observations) as herein reported with AP toxicity in SK-N-SH cells. This effect does not depend on the form of AP used, as we have seen synergistic interactions against both AP 1-40 and AP 25-35 in HT-22 cells (Green et al., 1998). Although differences exist in the glutathione-induced shift in the neuroprotective potency of E2 in these cell types, these may directly relate to the different cell culture techniques used to assess viability. Furthermore, intracellular concentrations of glutathione may play a role as we have determined that primary rat cortical neurons have higher intracellular glutathione concentrations (172 ± 12 µM) than SK-N-SH cells (15 ± 2 µM). In either case, we can resolve discrepancies with regard to differences reported between our work (Green et al., 1996) and others (Behl et al., 1995; Goodman et al., 1996) for the neuroprotective concentrations of estrogens. In our previous work, we used a cell culture milieu containing GSH, whereas others used media that lacked GSH.

Mounting evidence supports the hypothesis that part of the neuroprotective activity of estrogens resides in their antioxidant capacity. Decreases in oxidative by-products are correlated with decreases in cellular toxicity in our previous studies (Gridley et al., 1997), as well as those done by others (Behl et al., 1997; Goodman et al., 1996). Because the actions of most antioxidants are multifaceted, we expect that estrogens may be working through several mechanisms to provide this neuroprotection. Estrogen can participate in the nonenzymatic reduction of free radicals (Mukai et al., 1990), further demonstrated in a cell-free system where peroxy-nitrite radicals were found to be reduced by estrogens (Mooradian, 1993) at the same concentration at which we see neuroprotection in the absence of glutathione. In addition, estrogens can participate in iron reduction (Ruiz-Larrea et al., 1995), which may be paramount to decreasing the production of free radicals. Multiple studies have demonstrated that estrogens decrease lipid peroxidation in a variety of model systems (Sugioka et al., 1987; Behl et al., 1995; Lacort et al., 1995; Goodman et al., 1996; Tang et al., 1996a; Behl et al., 1997; Gridley et al., 1997; Keller et al., 1997). Estradiol has also been shown to reduce oxidative impairment of membrane transporters for ions and glucose resulting from AP 25-35 exposure (Keller et al., 1997). However, estrogens did not prevent impairment of membrane transport systems from toxic lipid peroxidation by-products (Keller et al., 1997) providing additional support for the idea that their antioxidant nature is the basis for their activity in neuroprotection.

Likewise, GSH exerts its antioxidant activity through several mechanisms (Meister and Anderson, 1983). GSH can scavenge free radicals via a nonenzymatic mechanism (Meister and Anderson, 1983). In our system, high concentrations of GSH (325 µM) were necessary to protect cells from the AP insult (20 µM). This neuroprotective concentration is a much higher concentration of GSH than is present in extracellular fluids (Smith et al., 1996). Perhaps a more practical explanation involves the ability of GSH to act on intracellular peroxides via GSH peroxidases and GSH S-transferases (Meister and Anderson, 1983). This system functions in the defense against free radicals through the reduction of hydrogen peroxide. This action of GSH may be relevant to the AP insult, because generation of increased amounts of H₂O₂ has been demonstrated in both AP 25-35- and AP 1-40-induced toxicity (Behl et al., 1994) and increases in the activity of glutathione peroxidase are correlated with increases in cell survivability for both peptides (Sagara et al., 1996).

The observed synergistic interaction between E2 and GSH seems to be mediated through an estrogen-receptor independent mechanism. Both the SK-N-SH human neuroblastoma and the HT-22 mouse hippocampal cell lines lack a functional estrogen receptor as determined by nuclear exchange assay (Green et al., 1998). Further, a variety of estratrienes that act only transiently at the estrogen receptor exhibit neuroprotection equivalent to that of E2 (Green et al., 1996; Green et al., 1997a, 1997b; Green et al., 1998). Additionally, we (Green et al., 1997b) and Behl (Behl et al., 1997) have demonstrated that an intact phenolic group is necessary for neuroprotection. The use of ICI as an antiestrogen is further support for this idea. The amount of ICI used, which in 100-fold excess of the E2 concentration satisfies the criteria required for demonstration of competitive inhibition, is at concentrations at which other phenolic A ring containing compounds demonstrate protection in the absence of glutathione (Green et al., 1998).

Given the phenolic nature of estrogens, they could activate the antioxidant response element/electrophilic response element (Montano and Katzenellenbogen, 1997). The antioxidant response element/electrophilic response element has been shown to be activated by phenolic antioxidants to increase phase II enzyme production, which includes GSH S-transferase (Jaiswal, 1994). Yet another possibility involves the conjugation of glutathione to estrogens (Jellnick et al., 1967). This action proceeds via glutathione S-transferases, where conjugation on the 4 position of the A ring of the steroid molecule provides bulky substituents (Elce and Harris, 1971), which has been shown in vivo to increase greatly the antioxidant potential (Miller et al., 1996). Given that 20-30% of the cellular mitochondrial pool resides in the mitochondria in some cell types (Smith et al., 1996), and estrogens stabilize mitochondrial function (Mattson et al., 1997), the estrogen-glutathione synergy could function to protect mitochondria against oxidative damage.

Lipophilic estrogens that partition to the plasma membrane should associate their phenolic A rings with the charged hydrophilic head groups of the membrane phospholipids. Conjugation of estradiol with bovine serum albumin at the 17- (Green et al., 1997b) or 6- carbon positions (Green et al., 1998), which prevents the appropriate orientation of the molecule into the plasma membrane, blocks the neuroprotective action of estradiol. Based upon the observation that AP aggregates extracellularly and causes membrane lipid peroxidation (Behl et al., 1995; Goodman et al., 1996; Gridley et al., 1997), we predict the hydroxyl hydrogen of estradiol is donated to prevent the cascade of membrane lipid peroxidation. Additionally, the enhanced potency of estrogens may result from its ability to donate hydrogen ions from several positions on the A ring (Jellnick and Bradlow, 1990). A relatively stable oxidized form of estradiol could result from this hydrogen ion donation and glutathione peroxidase could regenerate the reduced form of estrogen. This would operate by using GSH as a substrate for donation of the hydrogen group back to estrogen, and thus explain the synergy between the two molecules.

The specificity of estrogen's interaction for this glutathione system is supported by two lines of evidence. First, there are no apparent interactions noted between estrogen and the other thiols tested, lipoic acid or taurine, or any other antioxidants, including ascorbic acid or -tocopherol. It is interesting to note that although -tocopherol is a powerful antioxidant in its own right, estrogen has been argued to be even more powerful. Sugioka et al. (Sugioka et al., 1987) postulate that this may be because of the ability of the tocopheroxyl radical to regenerate estrogen. We have not observed any such apparent interactions in our system. Second, the ability of oxidized glutathione to work in this system supports the argument that estrogens may be interacting with the glutathione peroxidase/reductase process. In vivo evidence supports this idea, in that oral contraceptive use has been correlated with an increase in glutathione peroxidase activity (Capel et al., 1981; Massafra et al., 1993).

Finally, the identification of this synergistic interaction between estrogens and glutathione in neuroprotection has implications for the area of drug development for neurodegenerative diseases. Our observation that the potency of E2 was markedly affected by physiological concentrations of GSH in the cell culture milieu indicates that careful consideration for antioxidant defenses must accompany experimental design when assessing antioxidant drugs in vitro. Additionally, our data indicate that estrogen therapy can be improved by agents that enhance intracellular or extracellular glutathione concentrations.