The Androgen Metabolite, 5α-androstane-3β,17β-diol, Decreases Cytokine-Induced Cyclooxygenase-2, Vascular Cell Adhesion Molecule-1 Expression, and P-Glycoprotein Expression in Male Human Brain Microvascular Endothelial Cells

Abstract

P-glycoprotein (Pgp), a multiple drug resistance transporter expressed by vascular endothelial cells, is a key component of the blood-brain barrier and has been shown to increase after inflammation. The nonaromatizable androgen, dihydrotestosterone (DHT), decreases inflammatory markers in vascular smooth muscle cells, independent of androgen receptor (AR) stimulation. The principal metabolite of DHT, 5α-androstane-3β,17β-diol (3β-diol), activates estrogen receptor (ER)β and similarly decreases inflammatory markers in vascular cells. Therefore, we tested the hypothesis that either DHT or 3β-diol decrease cytokine-induced proinflammatory mediators, vascular cell adhesion molecule-1 (VCAM-1) and cyclooxygenase-2 (COX-2), to regulate Pgp expression in male primary human brain microvascular endothelial cells (HBMECs). Using RT-qPCR, the mRNAs for AR, ERα, and ERβ and steroid metabolizing enzymes necessary for DHT conversion to 3β-diol were detected in male HBMECs demonstrating that the enzymes and receptors for production of and responsiveness to 3β-diol are present. Western analysis showed that 3β-diol reduced COX-2 and Pgp expression; the effect on Pgp was inhibited by the ER antagonist, ICI-182,780. IL-1β-caused an increase in COX-2 and VCAM-1 that was reduced by either DHT or 3β-diol. 3β-diol also decreased cytokine-induced Pgp expression. ICI-182,780 blocked the effect of 3β-diol on COX-2 and VCAM-1, but not Pgp expression. Therefore, in cytokine-stimulated male HBMECs, the effect of 3β-diol on proinflammatory mediator expression is ER dependent, whereas its effect on Pgp expression is ER independent. These studies suggest a novel role of 3β-diol in regulating blood-brain barrier function and support the concept that 3β-diol can be protective against proinflammatory mediator stimulation.

A key group of drug efflux transporters at the blood brain barrier (BBB) are the superfamily of ATP-binding cassette transporters (ABC-transporters) (1), which are classified into seven families (2, 3). These drug efflux pumps are located in the luminal membrane of endothelial cells that form the BBB and play an important role in pharmacotherapeutic efficacy by acting as a selective barrier to drugs or other small molecules by transporting them out of the brain tissue. The most highly expressed ABC transporter of the BBB is P-glycoprotein (Pgp; ABCB1), which actively prevents the uptake of some anticancer drugs, antiepileptic drugs, antidepressants, antipsychotics, and corticosteroids (4–7). In rodent studies, inhibition of Pgp increases the movement of chemotherapy drugs across the BBB and thereby increases the therapeutic efficacy of these drugs (8–10).

The BBB is defined as the monolayer of vascular endothelial cells connected by high-resistance tight cell-cell junctions devoid of fenestrations and serves as a selective barrier for the movement of molecules from the bloodstream into the brain parenchyma (11). The integrity of the BBB can be compromised by inflammation (12), which is of interest because inflammation occurs in a wide range of brain disorders (7, 13–15). Further, BBB breakdown is correlated with advancing disease states such as multiple sclerosis, epilepsy, and Alzheimer's disease (11). Inflammatory responses elicited by neural trauma, such as after ischemic stroke and traumatic brain injury, have also been shown to compromise BBB integrity (11, 16, 17).

Aside from endothelial cell and tight junction damage, BBB integrity may also be altered by changes in the expression of transporter proteins such as Pgp. Studies have shown that factors that promote inflammation increase Pgp expression/activity in both human and rodent cells/tissue (6, 18–23). For example, the cytokine, TNFα, has been shown to increase Pgp protein and mRNA expression in human and rat brain endothelial cell lines (6, 22). Both TNFα and the cytokine, IL-1β, also increased Pgp protein expression in primary cultures of rat astrocytes (23). In rat brain endothelial cells, hypoxia followed by reoxygenation increased Pgp expression and nitric oxide, a product of the proinflammatory enzyme, inducible nitric oxide synthase (19). In some cases, increased expression and activity of Pgp could be blocked by TNFα receptor inhibition or did not occur in TNFα receptor null mice (20).

The proinflammatory mediator cyclooxygenase-2 (COX-2) has been shown to play a role in the increases in Pgp protein expression and activity observed after seizures (15, 24). This may be the result of exposure of capillaries to the excitatory neurotransmitter glutamate, which is in excess during seizure activity. Exposure to glutamate, both in vivo and in vitro, increased Pgp expression and transport activity within several hours. This effect could be blocked with a selective COX-2 inhibitor and was not observed in COX-2 null mice. Similar studies in rats also found that seizures induced Pgp expression and this effect was blocked by COX-2 inhibition (15, 24). Of importance, we have recently reported that in both brain and coronary human vascular smooth muscle cells, dihydrotestosterone (DHT), a potent nonaromatizable androgen, decreases COX-2 protein expression that was induced with cytokines or by hypoxia with glucose deprivation (25, 26).

Both androgens and estrogens have been shown to have antiinflammatory effects (25–29); however, it is important to note that androgens also have proinflammatory effects as well (30–32). In regards to its antiinflammatory properties, DHT can suppress levels of COX-2 after hypoxia with glucose deprivation or cytokine stimulation in human vascular smooth muscle cells, independent of androgen receptor (AR) stimulation (25, 26, 33). Interestingly, and perhaps not commonly appreciated, DHT can have properties beyond that of an AR agonist because it can be further metabolized to 5α-androstane-3β,17β-diol (3β-diol), an estrogen receptor β (ERβ) agonist (34–36). 3β-diol has recently been shown to reduce expression of inflammatory markers in human brain vascular smooth muscle cells (33) and human umbilical vein endothelial cells (37) and in a similar fashion, the dehydroepiandrosterone (DHEA) metabolite/ERβ agonist 5α-androstene-3β,17β diol (Adiol), has also been shown to have antiinflammatory effects (38). Therefore, in these studies we tested the hypothesis that DHT and 3β-diol, act through estrogen receptors to decrease cytokine-induced expression of the inflammatory mediator COX-2, leading to decreased inflammation-induced expression of Pgp in male human brain microvascular endothelial cells (HBMECs). In addition to COX-2, we examined expression of vascular cell adhesion molecule-1 (VCAM-1), another common maker for vascular inflammation. Like COX-2, VCAM-1 is up-regulated by cytokine stimulation in endothelial cells. Both DHT and 3β-diol have been shown to decrease cytokine-induced VCAM-1 expression in human umbilical vein endothelial cells (27, 37); however, their effects in human brain endothelial cells are unknown. The results of these studies indicate that 3β-diol can inhibit Pgp, COX-2, and VCAM-1 expression but that its actions on Pgp are not coupled with its antiinflammatory properties.

Materials and Methods

Cell culture and hormone/drug treatment

Primary HBMECs isolated from a 30-yr-old male donor (Lot no. ACBRI 419) were purchased from Cell Systems Corp. (Kirkland, WA). HBMECs were grown in 5% CO₂, room air atmosphere at 37 C, in CSC media supplemented with culture boost (Cell Systems Corp.) and 10% fetal bovine serum.

Hormone/drug treatments were performed on cells at 80 to 90% confluency and at passage 7 in which HBMECs expressed the endothelial-specific protein von Willebrand factor (data not shown). Cell treatments were carried out in hormone-free media supplemented with charcoal-stripped fetal bovine serum (Invitrogen Life Technologies, Gaithersburg, MD). A 6-h time point for COX-2 induction, 10 nm DHT dose, and a 5 ng/ml IL-1β dose were selected for these studies based on our previous studies (25, 26). A 6-h time point for Pgp induction was chosen based on a published report (21). Cells were treated with DHT (10 nm), 3β-diol (1 nm or 10 nm), 17β-estradiol (E2; 10 nm), DHEA (10 nm), Adiol (10 nm), or vehicle (VEH; 0.001% ethanol) for 18 h followed by IL-1β (5 ng/ml) for an additional 6 h. In a separate set of experiments, cells were pre-treated for 1 h with the nonselective estrogen receptor (ER) antagonist ICI 182,780 (ICI, 1 μm; Tocris Bioscience; dissolved in ethanol) followed by 18 h of cotreatment with either VEH (0.001% ethanol) or 3β-diol (10 nm), and then 6 h exposure to IL-1β (5 ng/ml). ICI doses of 1 μm were chosen based on results from previous studies (25, 26). In some studies, cells were treated with VEH or hormone alone (Adiol or 3β-diol) for 6 h to determine their effects in the absence of inflammation.

Quantitative real-time RT-PCR (RT-qPCR)

RT-qPCR was used to measure mRNA expression of gonadal steroid hormone receptors and steroid-metabolizing enzymes in adult male cells treated with hormone-free media for 6 h as previously described (33) Negative controls, where water was used in place of template, were used in all experiments. Levels of mRNA expression for each sample were determined by comparison to the standard curve and reported as the absolute concentration of cDNA. Specificity was confirmed via thermal melting curve analysis which showed a single peak at the predicted melting temperature for each primer set. The list of primers used is shown in Supplemental Table 1 published on The endocrine Society's Journals Online web site at http://endo.endojournals.org.

Western blot

Levels of COX-2, VCAM-1, or Pgp protein were examined using standard immunoblotting methods, as previously described (26). Briefly, cells were rinsed with ice-cold PBS containing 100 μm sodium orthovanadate, scraped from flasks over ice, and centrifuged at 200 × g for 10 min. The pellet was resuspended in ice-cold lysis buffer (50 mm Tris, 150 mm NaCl, 0.1% sodium dodecyl sulfate, 1% Nonidet P-40, 1 mm EGTA, 1 mm EDTA, 1 mm dithiothreitol, 20 μm Pepstatin, 20 μm Leupeptin, 0.1 U/ml Aprotinin, 0.1 mm phenylmethylsulfonyl fluoride). Cell/lysis buffer mixture was divided in half to collect the whole-cell lysate for COX-2 and VCAM-1 analysis and membrane fraction lysate for Pgp analysis separately. Whole-cell lysate was prepared as previously described (25, 33). For membrane fraction isolation, the cell/lysis buffer mixture was centrifuged at 17,000 × g for 30 min at 4 C. Pellet (crude membrane fraction) and supernatant (crude cytosolic fraction) were collected. The crude membrane fraction was used for Pgp protein analysis. After total protein content determination, samples were loaded into 7.5% Smart gels (LI-CORE), and proteins were separated via SDS-PAGE and transferred to nitrocellulose membranes. Nonspecific binding was blocked by incubation at room temperature for 30 min in PBS containing 1% Tween (TPBS) and 3% dried milk. Membranes were incubated with COX-2 (1:1000) mouse monoclonal antibody (Cayman Chemical, Ann Arbor, MI), VCAM-1 (H-276; 1:1000) rabbit antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), Pgp (1:500) mouse antibody (Calbiochem, Gibbstown, NJ), and/or β-actin (1:5000) mouse monoclonal antibody (Sigma Aldrich Corp., St. Louis, MO) overnight at 4 C in TPBS. After TPBS washes, the membranes were incubated in goat antimouse IR 800 Dye (1:15,000), goat antirabbit IR 680 Dye (1:15,000), secondary antibody (LI-COR), or goat antimouse IgG-horseradish peroxidase (1:5000) secondary antibody (Bio-Rad Laboratories, Hercules, CA) for 1 h at room temperature. COX-2 antibody specificity was verified with lipopolysaccharide (LPS) stimulated Raw-264.7 (mouse macrophage) cell lysate (Santa Cruz Biotechnology), which is a positive control for COX-2 protein. Endothelial cell phenotype was confirmed (data not shown) in adult male HBMECs at passage 7 with Von Willebrand factor rabbit antibody (1:500, Santa Cruz Biotechnology). After additional TPBS washes, proteins were visualized and quantitated using an Odyssey Infrared Imager, and data were analyzed using Odyssey V3.0 software (LI-COR) or using a Fujifilm Imager, and data were analyzed using the Multi gauge V3.0 Software (Fujifilm Medical Systems USA, Inc., Stamford, CT).

Reagents

All reagents were purchased from Sigma Aldrich Corp. unless otherwise noted.

Statistical analysis

Samples from each treatment group were run on the same Western blot to allow for direct comparison. For each treatment, individual measures were repeated to achieve an appropriate number for statistical analysis (n = 3-17). Data from Western blots are expressed as an optical density ratio relative to VEH and normalized to the optical density values for β-actin bands. Multiple vehicles and treatment groups were included on the same gel and normalized to the first vehicle on the gel to account for variance. Normalizing to both a vehicle and a loading control limits variance between blots and allows pooling of multiple blots from each experiment. For VCAM-1 Western blot data, the combined optical density of the doublet was used for quantification. Data from RT-qPCR studies are expressed as femtograms of cDNA, and all measures were repeated a sufficient number of times for statistical analysis (n = 3-4). All values are reported as means ± sem. Unless otherwise noted, data were compared using one-way ANOVA across treatment groups using Prism Software (Irvine, CA), and when indicated, differences were compared post hoc using Student Newman-Keuls test. A level of P < 0.05 was considered significant.

Results

Gonadal steroid receptor and steroid-metabolizing enzyme mRNAs are expressed in adult male HBMECs

To confirm the expression of necessary gonadal steroid receptors and steroid-metabolizing enzymes for DHT and 3β-diol metabolism/receptor activation, adult male HBMECs were grown in hormone-free media, and mRNA levels were measured by RT-qPCR. The size of the amplified cDNA was confirmed by 2% agarose gel electrophoresis for each primer set (data not shown). Expression of mRNA for AR, ERα, ERβ, 3α-hydroxysteroid dehydrogenase (HSD), 3β-HSD, 17β-HSD, and CYP7B1 was detected in HBMECs (Fig. 1). AR mRNA expression was greatest compared with ERα or ERβ expression (Fig. 1A). 3α-HSD mRNA expression was greatest compared with other steroid-metabolizing enzyme genes tested (Fig. 1B).

Fig. 1.

Gonadal steroid receptors and steroid-metabolizing enzyme mRNAs are expressed in male HBMECs. A, Relative levels of AR, ERα, and ERβ mRNA expression assessed by RT-qPCR in adult male HBMECs treated for 6 h with hormone-free media (n = 3-4 per group). B, Relative levels of steroid metabolism enzymes 3α-HSD, 3β-HSD, 17β-HSD, and CYP7B1 mRNA expression assessed by RT-qPCR in adult male HBMECs treated for 6 h with hormone-free media (n = 3-4 per group).

3β-diol decreased COX-2 and Pgp protein expression

To determine the effect of 3β-diol on COX-2 and Pgp expression, male HBMECs were treated for 6 h with VEH or 3β-diol (1 nm and 10 nm). Additionally, to test whether 3β-diol's effects were ER mediated, some cells were pretreated with the non-subtype-selective ER antagonist ICI (1 μm) followed by 6 h of 10 nm 3β-diol. One-way ANOVA showed that a 10 nm dose of 3β-diol significantly decreased COX-2 and Pgp protein expression (P < 0.05 vs. VEH; Fig. 2). 3β-diol's effects on Pgp, but not COX-2, were blocked by ICI.

Fig. 2.

3β-diol decreased COX-2 and Pgp expression. COX-2 and Pgp protein expression were assessed using Western analysis in adult male HBMECs treated with VEH, 1 nm 3β-diol, 10 nm 3β-diol, or 10 nm 3β-diol + ICI (1 μm; 1 h pretreatment) for 6 h. A, Representative Western blot for COX-2 and β-actin. Molecular mass is indicated in parentheses. A molecular weight standard (STD) is shown in the far left lane. B, Relative levels of COX-2 after treatment with VEH, 1 nm 3β-diol, 10 nm 3β-diol, or 10 nm 3β-diol + 1 μm ICI. C, Representative Western blot for Pgp and β-actin. Molecular weight is indicated in parentheses. D, Relative levels of Pgp after treatment with VEH, 1 nm 3β-diol, 10 nm 3β-diol, or 10 nm 3β-diol + 1 μm ICI. For panels B and D, COX-2 and Pgp expression were each normalized to β-actin and calculated as the ratio compared with VEH. For panels B and D, each bar represents the mean ± sem of three to six determinations. *, P < 0.05 vs. VEH.

DHT and 3β-diol decreased IL-1β-induced COX-2 and VCAM-1 expression

To determine whether 3β-diol could inhibit inflammatory markers induced by IL-1β, male HBMECs were treated for 18 h with VEH, DHT (10 nm), or 3β-diol (10 nm) followed by VEH or IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. One-way ANOVA showed that IL-1β treatment significantly increased levels of COX-2 (P < 0.05, Fig. 3) and VCAM-1 (P < 0.05; Fig. 4). Furthermore, both DHT and 3β-diol reduced the IL-1β-induced increases in COX-2 (P < 0.05 vs. VEH + IL-1β; Fig. 3) and VCAM-1 protein (P < 0.05 vs. VEH+IL1β; Fig. 4).

Fig. 3.

DHT and 3β-diol inhibited cytokine-induced increases in COX-2. COX-2 protein expression was assessed in adult male HBMECs treated with VEH, DHT (10 nm), or 3β-diol (10 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. Relative changes in COX-2 protein expression are shown after treatment with VEH, 10 nm DHT, or 10 nm 3β-diol in combination with IL-1β. COX-2 expression was normalized to β actin and reported as an intensity ratio compared with VEH-treated cells. Each bar represents the mean ± sem of six to 17 determinations. *, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

Fig. 4.

DHT and 3β-diol inhibited cytokine-induced increases in VCAM-1. VCAM-1 protein expression was assessed in adult male HBMECs treated with VEH, DHT (10 nm), or 3β-diol (10 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. A, Representative Western blot for VCAM-1 and β actin. Numbers in parentheses indicate the calculated molecular mass of the protein shown. A molecular weight standard (STD) is shown in the far left lane. B, Relative changes in VCAM-1 protein expression after treatment with VEH, 10 nm DHT, or 10 nm 3β-diol in combination with IL-1β. VCAM-1 expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. For panel B each bar represents the mean ± sem of six to 17 determinations. *, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

3β-diol, but not DHT or E2, inhibited cytokine-induced increases in Pgp

We sought to determine whether sex steroids would affect cytokine-induced Pgp expression in male HBMECs in a fashion similar to the inflammatory markers COX-2 and VCAM-1. HBMECs were treated for 18 h with VEH, DHT (10 nm), E2 (10 nm), or 3β-diol (10 nm), followed by VEH or IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. One-way ANOVA revealed that IL-1β significantly increased Pgp protein expression (P < 0.05 vs. VEH; Fig. 5). 3β-diol, but not DHT or E2, attenuated the IL-1β-induced increases in Pgp protein (P < 0.05 vs. VEH + IL-1β; Fig. 5).

Fig. 5.

3β-diol inhibited cytokine-induced increases in Pgp. Pgp protein expression was assessed in HBMECs treated with VEH, DHT (10 nm), E2 (10 nm), or 3β-diol (10 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. Relative levels of Pgp are shown after treatment with VEH, DHT, E2, or 3β-diol in combination with IL-1β. Pgp expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. Each bar represents the mean ± sem of four to 11 determinations. *, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

3β-diol inhibits cytokine-induced increases in COX-2 and VCAM-1 via ER activation

To test ER dependence of 3β-diol's effects, male HBMECs were pretreated for 1 h with VEH or the nonselective ER antagonist ICI (1 μm), and then treated (18 h) with VEH or 3β-diol (10 nm) followed by VEH or IL-1β (5 ng/ml; 6 h) in the continued presence of hormone/antagonist. One-way ANOVA revealed that 3β-diol significantly reduced the IL-1β-induced increases in COX-2 protein (P < 0.05 vs. IL-1β/VEH), and this effect was prevented by the ER antagonist, ICI (Fig. 6A). 3β-diol also reduced the IL-1β-induced increases in VCAM-1 protein expression (P < 0.05 vs. IL-1β/VEH), an effect that was also prevented by the ER antagonist, ICI (Fig. 6B). ICI alone did not have an effect on COX-2 or VCAM-1 levels.

Fig. 6.

3β-diol inhibited cytokine-induced increases in COX-2 and VCAM-1 via ER. HBMECs were pretreated for 1 h with VEH or the nonselective ER antagonist ICI (1 μm), followed by VEH or treated with VEH or 3β-diol (10 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone/antagonist. A, Relative levels of COX-2 after treatment with VEH, 3β-diol, or 3β-diol + ICI in combination with IL-1β. COX-2 expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. B, Relative levels of VCAM-1 after treatment with VEH, 3β-diol, or 3β-diol + ICI in combination with IL-1β. VCAM-1 expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. Each bar represents the mean ± sem of five to 17 determinations (panel A) or two to 12 determinations (panel B).*, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

3β-diol inhibited cytokine-induced increases in Pgp independent of ER activation

To test ER dependence of 3β-diol's effects on Pgp, male HBMECs were pretreated for 1 h with VEH or the nonselective ER antagonist ICI (1 μm), and then treated (18 h) with VEH or 3β-diol (10 nm) followed by VEH or IL-1β (5 ng/ml; 6 h) in the continued presence of hormone/antagonist. 3β-diol prevented the IL-1β-induced increases in Pgp protein expression (P < 0.05 vs. IL-1β/VEH); however, this effect was not inhibited by the ER antagonist ICI (Fig. 7).

Fig. 7.

3β-diol inhibited cytokine-induced increases in Pgp independent of ER activation. Pgp protein expression was assessed in HBMECs pretreated for 1 h with VEH or the nonselective ER antagonist ICI (1 μm), followed by VEH or 3β-diol (10 nm or 100 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone/antagonist. Relative changes in Pgp expression are shown after treatment with VEH, 10 nm 3β-diol, 100 nm 3β-diol, or 10 nm 3β-diol + ICI in combination with IL-1β. Pgp expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. Each bar represents the mean ± sem of four to 11 determinations. *, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

Effects of DHEA and its metabolite/ERβ agonist Adiol on cytokine-induced COX-2, VCAM-1, and Pgp

Similar to our recent findings with 3β-diol (33), it was recently reported that the DHEA metabolite, Adiol, has antiinflammatory properties in microglia and activates ERβ (38). To determine whether Adiol and DHEA have similar effects as 3β-diol on brain endothelial cells, male HBMEC were treated for 18 h with VEH, DHEA (10 nm), or Adiol (10 nm) followed by VEH or IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. Neither DHEA nor Adiol altered IL-1β-induced COX-2 (Fig. 8A) or Pgp expression (Supplemental Fig. 1). Although DHEA itself did not alter IL-1β-induced VCAM-1 expression, its metabolite, Adiol, modestly decreased IL-1β-induced VCAM-1 expression (P < 0.05 vs. VEH + IL-1β; Fig. 8B).

Fig. 8.

Effects of DHEA and the DHEA metabolite/ERβ agonist Adiol on cytokine-induced increases in COX-2 and VCAM-1. Adult male HBMECs were treated with VEH, DHEA (10 nm), or Adiol (10 nm) for 18 h, and then exposed to IL-1β (5 ng/ml; 6 h) in the continued presence of hormone. A, Relative changes in COX-2 expression after treatment with VEH, DHEA, or Adiol in combination with IL-1β. COX-2 expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. B, Relative changes in VCAM-1 expression after treatment with VEH, DHEA, or Adiol in combination with IL-1β. VCAM-1 expression was normalized to β-actin and reported as an intensity ratio compared with VEH-treated cells. Each bar represents the mean ± sem of three to seven determinations.*, P < 0.05 vs. VEH; #, P < 0.05 vs. IL-1β VEH.

Discussion

This study addressed the effects of the androgen DHT and its metabolite 3β-diol on vascular inflammatory markers and Pgp expression in male HBMECs and determined whether they were mechanistically linked. Using COX-2 and VCAM-1 protein expression as markers for inflammation, we examined the effects of DHT and DHEA and their metabolites, 3β-diol and Adiol, on cytokine-induced inflammation and Pgp expression. Results show that DHT decreases cytokine-induced COX-2 and VCAM-1 and that this may possibly occur via metabolism to 3β-diol and subsequent activation of ER. Interestingly, 3β-diol's effect to reduce cytokine-induced Pgp expression appears to be ER independent and not associated with the up-regulation of inflammatory markers.

In the current studies, we demonstrate that the potent androgen, DHT, reduces cytokine-induced expression of the proinflammatory mediators, COX-2 and VCAM-1, in male HBMECs. These data are supported by numerous studies showing antiinflammatory effects of androgens in a variety of cells/tissues including human monocytes (39), human macrophages (40), mouse macrophages (41–43), human prostate cancer cells (44), human endothelial cells (27, 45–47), human coronary artery vascular smooth muscle cells (26), and human brain vascular smooth muscle cells (25, 33). Androgen treatment in vivo has also been shown to decrease inflammatory markers (42, 48, 49).

Our results point to ERs as a potential alternate receptor pathway used by DHT during cytokine-induced inflammation in endothelial cells, because the effects of the DHT metabolite 3β-diol on COX-2 and VCAM-1 were blocked by ER antagonism. Because DHT can be converted to 3β-diol by the enzymes 3α-HSD, 3β-HSD, and 17β-HSD (35, 51–54), we asked whether these enzymes are found in HBMECs. Our data show that primary HBMECs express the steroid receptors AR, ERα, ERβ, as well as enzymes involved in 3β-diol metabolism, 3α-HSD, 3β-HSD, 17β-HSD, and CYP7B1. This is consistent with previous studies showing that blood vessels contain AR, ERα, ERβ, and 3β-HSD (55–58). Although we did not measure expression of these enzymes or receptors after hormone treatment or cytokine treatment, we have previously measured mRNA expression after IL-1β treatment in human brain vascular smooth muscle cells and found that ERβ expression is unchanged (33). In many cases treatment with hormones causes an up-regulation of their receptor. For example, with DHT treatment we have found that AR protein expression increases (26). 3β-diol has also been shown to increase ERβ expression (59). As for effects that inflammation may have on hormone metabolism, it has been shown that cytokine stimulation (IL-1α, IL-4) increases 3β-HSD expression and activity (60). However, it should be pointed out that changes in any one of these enzymes may not necessarily alter levels of 3β-diol because metabolism of DHT can be through the actions of several different enzymes (35, 51–54).

These results also show that both DHT and 3β-diol decrease cytokine-induced COX-2 and VCAM-1 expression in HBMECs and that 3β-diol's effects are mediated via ER because a non-subtype-selective ER antagonist can block the actions of 3β-diol. These data are consistent with the recent study by Norata et al. (37) showing that 3β-diol has antiinflammatory actions in human umbilical vein endothelial cells and mouse aorta, and that these effects are mediated by ERβ. Furthermore, we have recently found that DHT reduces cytokine-induced COX-2 expression in human brain vascular smooth muscle cells via ERβ (33). Interestingly, in the absence of cytokine stimulation, 3β-diol decreases COX-2 expression and this effect is not ER mediated, whereas, during cytokine stimulation the inhibition of COX-2 expression was ER mediated. These results suggest that targeting the 3β-diol/ER pathway may be beneficial for reducing vascular inflammation. This could be accomplished therapeutically with 3β-diol treatment, or perhaps treatment with another ERβ-selective agonist or inhibition of CYP7B1, the enzyme that converts 3β-diol to inactive metabolites (35, 61).

HBMECs also express 17β-HSD. This enzyme not only converts DHT to 3β-diol, but also is the enzyme primarily responsible for the conversion of the adrenal androgen, DHEA to Adiol (62), another ERβ agonist (38). Consequently, the possibility exists that other products of androgen metabolism can have similar effects as those in the DHT/3β-diol pathway. However, our results show that at a similar dose, DHEA itself did not alter COX-2 or VCAM-1 expression in HBMECs, whereas its metabolite, Adiol, only modestly decreased cytokine-induced VCAM-1 expression via ER activation. Such data indicate that DHEA conversion to Adiol is likely not an active pathway in HBMEC cells although HBMEC cells can respond to Adiol through its ER activity.

Recent studies show that ERβ activation has a number of antiinflammatory effects in a variety of cells/tissues such as rodent microglial (63, 64), mouse brain (65), a rodent motor neuron cell line (66), astrocytes (67), rat vascular smooth muscle cells (29), and human endothelial cells (37). Saijo et al. (38) identified Adiol as an ERβ agonist that can act on microglia to reduce inflammatory markers. Further, they found that Adiol suppresses inflammatory responses by binding to ERβ and recruiting C-terminal binding protein (CtBP), a corepressor that binds to activator protein-1-dependent promoters, thereby repressing expression of proinflammatory genes. Interesting, E2 does not recruit CtBP and was unable to elicit this antiinflammatory response. CtBP recruitment is likely also required for ERβ-mediated repression of IL-1β induced proinflammatory mediators because the endotoxin LPS, their inflammatory stimulant of choice, and IL-1β act via a converging transcriptional pathway. Furthermore, they found that IL-1β treatment, like LPS treatment, activated protein kinase A, which they showed was one of the key steps in ERβ-mediated repression of proinflammatory mediators (38). Although it is possible that 3β-diol may, at a molecular level, work similarly to Adiol and recruit CtBP to inhibit inflammation, in the current studies 3β-diol was much more effective than Adiol, suggesting that its binding to ER may induce a more effective conformation change to attract CtBP or an alternate set of coregulatory proteins to the transcriptional machinery in HBMECs. Nonetheless, these studies show that ERβ activation has similar protective effects as do androgens and support the theory that some of the beneficial effects of androgens, particularly those associated with antiinflammatory responses, may occur through activation of ERβ. Whether the ER-dependent effects of 3β-diol we observed in the current studies are mediated by ERβ is yet to be determined. Based upon previous studies in other human vascular cell types, including brain vascular smooth muscle (33) and umbilical vein endothelial cells (37), suggesting ERβ receptor subtype involvement leads us to suspect that ERβ is involved. However, future studies using ERβ-selective antagonists or ERβ knockdown would need to be performed to be certain that 3β-diol is working through ERβ.

Our current data show that the cytokine IL-1β increases protein expression of Pgp in a fashion similar to that reported by others, in epithelial cells (18), rat astrocytes (23), a human brain endothelial cell line (6), rat brain endothelial cells (19), and isolated rat brain capillaries (20, 21). This is consistent with the effects elicited by other cytokines such as TNFα, which has been shown to increase Pgp protein and mRNA expression in a human brain endothelial cell line (6) and to increase Pgp protein expression and activity in rat brain capillaries (21).

To our knowledge, this is the first report showing effects of 3β-diol on any multiple drug-resistance transporter in any cell type. We have identified the DHT metabolite 3β-diol, but not DHT itself, as an inhibitor of Pgp expression in HBMECs. Thus, although all enzymes necessary for the conversion of DHT to 3β-diol are present in HBMECs, there may be insufficient levels to convert DHT at a rate sufficiently high to affect Pgp expression. Moreover, the ability of 3β-diol to inhibit Pgp expression occurred in HBMECs even in the absence of an inflammatory signal such as IL-1β. Thus, although it appears that 3β-diol is capable of preventing the inflammatory mediated increase in Pgp, it is more likely that 3β-diol's effect on Pgp expression occurs independent of its effect on COX2 inhibition. Support for this comes from our results after treatment with 3β-diol alone, showing that the effect on Pgp expression is blocked by ICI 182,780, but the effect on COX-2 is not. Furthermore, treatments with 3β-diol inhibited the IL-1β-induced up-regulation of Pgp, and cotreatment with ICI 182,780 decreased the effects of 3β-diol on IL-1β-induced COX-2 expression but did not alter IL-1β induced Pgp expression. In both cases, there was a disassociation of inflammatory marker regulation and of Pgp indicating that their expressions are similar but independently controlled. Interestingly, although 3β-diol was able to inhibit Pgp expression, E2 was not. Another recent paper showed that protective effects of 3β-diol to reduce cell proliferation, cell adhesion, and tumor invasion, which were mediated via ERβ, also could not be mimicked by E2 (68). In addition, we recently showed that recruitment of steroid receptor coactivator 1 and cAMP response element-binding protein-binding protein to the composite response element in the oxytocin promoter occurs after 3β-diol, but not after E2 treatment (69). Further, when examining the regulation of the vasopressin promoter by 3β-diol, occupancy of ERβ by 3β-diol is more effective in inducing transcriptional activity than E2 (70).

In previous studies, we found that DHT decreases COX-2 expression via an AR-independent mechanism in human brain vascular smooth muscle cells in the presence of an inflammatory stimulus (25, 26, 33). Further, we have recently demonstrated that the effect of DHT on COX-2 expression in human vascular smooth muscle cells is mediated by ERβ (33). Our current data support and extend those data in that we show that DHT's antiinflammatory effects in vascular endothelial cells are likely also occurring via metabolism to 3β-diol and activation of ERs. Because COX-2 has been shown to mediate the increases in Pgp expression that are seen after epileptic seizures (15, 24), it would be congruent to suggest that 3β-diol's effect to decrease Pgp in HBMECs could also be COX-2 mediated. However, several observations suggest that this is not the case. After IL-1β administration, 3β-diol inhibited Pgp expression to a much greater degree than it had on COX-2 and VCAM-1 expression. Further, ICI did not block the actions of 3β-diol on Pgp expression, whereas it prevented its actions on COX2 and VCAM-1. Because 3β-diol's effect on Pgp was not blocked by ICI, it is possible that in this case 3β-diol may be acting via a nongenomic mechanism that is not blocked by ER antagonists. This might involve GPR30 or other membrane-bound estrogen receptors, even potentially membrane-bound ERβ (71–73). Metabolism to other steroids is also possible (74). Taken together, these data suggest that 3β-diol's ability to block Pgp expression is independent of its actions as an antiinflammatory agent.

Previous studies have shown that Pgp expression can be regulated by steroid hormones. Kim and Benet (75), showed that ethinyl estradiol can increase Pgp and MDR1 expression levels by up to 3-fold (75). Progesterone and E2 have also been shown to up-regulate Pgp expression via transcriptional regulation (76). E2, working via ERβ, has been shown to down-regulate Pgp protein expression via a posttranslational mechanism in human breast cancer cells (77). However, another study showed that E2 treatment can increase Pgp expression in breast cancer cells via ERα (78). Further, hormones such as glucocorticoids have been described as targets for Pgp-mediated transport (79). Thus the possibility exists that changes in gonadal steroid hormones or their metabolites may alter sensitivity to glucocorticoids in the brain through changes in Pgp expression.

In the present studies, the inability of ICI to block 3β-diol's effect on Pgp expression during IL-1β stimulation suggests that, in the presence of an inflammatory stimulus such as IL-1β, 3β-diol is acting through an ER-independent mechanism. A similar mechanism has been reported for E2's effect on another gene encoding a multiple drug resistance transporter, breast cancer-related protein. It was found that E2 rapidly decreases breast cancer-related protein activity via nongenomic signaling in mouse brain capillaries, and that this effect could not be blocked with the nonselective ER antagonist ICI (80). However, Hartz et al., (80) examined only short-term (90 min or less) E2 effects, so alternative genomic effects after longer treatment time may have been missed (80). Similarly, our studies did not examine Pgp regulation after acute exposure to hormones. Because the current studies show that E2 did not alter cytokine-induced Pgp expression, it is likely that the effect of E2 on Pgp expression varies based on experimental conditions/cell types and would not be expected to correspond to changes seen after 3β-diol.

The functional significance for the up-regulation of Pgp expression by cytokines, and the down-regulation after 3β-diol treatment is not currently known. Up-regulation of Pgp would indicate a decrease in BBB permeability, although this has not been tested under the conditions used in the current study. However, because Pgp does not work alone in controlling efflux across the BBB, it is possible that it represents a compensatory change in response to alterations in other ABC cassette transporters. Up-regulation of Pgp may also be working to prevent permeability of substances that can be neurotoxic or neuroendangering. An example of these are the glucocorticoid hormones. Synthetic glucocorticoids such as prednisone (81), dexamethasone (82), and endogenous hormones such as progesterone and cortisol have been suggested to be targets for Pgp (79, 83) and may alter the negative feedback regulation of the hypothalamo-pituitary-adrenal axis (50), particularly because Pgp restricts accumulation of radiolabeled cortisol in hypothalamus, but not other brain areas (82). However, recent studies suggest that this is not always the case (85, 86) and ultimately, transport mechanisms may be tissue specific.

This study used primary cultures of HBMECs to investigate the effects of sex steroids on vascular inflammation and Pgp expression. This model may be more translationally relevant compared with studies using rodent models; however, there are certain limitations with the use of human primary cell cultures. Although these cells originated from a single donor, our previous studies in human coronary artery vascular smooth muscle cells, originating from a different donor, also showed that DHT attenuated IL-1β-induced COX-2 via an AR-independent mechanism (26). Further, we have found that DHT decreases COX-2 expression in response to hypoxia with glucose deprivation in human brain vascular smooth muscle cells, rat pial arteries, and in the human brain vascular smooth muscle cells. The antiinflammatory effect of DHT was also AR independent (25). Our findings correspond with a previous study showing that 3β-diol decreased cytokine-induced vascular inflammation via an ER-dependent mechanism in human umbilical vein endothelial cells (37). Together, these studies minimize the concern of using cells isolated from a single donor. A second limitation of using cultured primary cells that should be noted is that their properties can change over time. Therefore, to minimize the risk of culture-induced changes, all experiments were performed after a small number of passages (7) when the HBMEC still expressed the endothelial cell marker, von Willebrand factor (data not shown). The third limitation of this study is that responses were examined in male cells. There have been reports of sex differences in inflammatory responses (87, 88) and in future studies we hope to determine whether the DHT/3β-diol/ER pathway also plays a role in the female response to vascular inflammation.

In summary, we have shown that DHT, 3β-diol, and Adiol decrease cytokine-induced vascular inflammatory markers in male HBMECs and that 3β-diol decreases Pgp expression in the presence and absence of IL1-β. Although 3β-diol's antiinflammatory effects during cytokine stimulation occur via an ER-dependent mechanism, its ability to decrease Pgp expression appears to be ER independent. Thus, the results of these studies have identified an ER-dependent antiinflammatory effect for DHT in vascular endothelial cells by the DHT metabolite 3β-diol, and the DHEA metabolite Adiol. Further, we have identified 3β-diol, a naturally occurring metabolite of DHT, as a steroid that can down-regulate Pgp expression and potentially influence the function of the BBB. Because Pgp functions to limit the uptake of many therapeutic drugs (4–6) and hormones (79, 83), the ability to down-regulate its expression could have a major impact on the efficacy of therapeutic approaches to manage brain diseases or traumatic insults as well as inhibit any accompanying inflammatory responses.