The Isoflavone Equol Mediates Rapid Vascular Relaxation

We recently reported that soy isoflavones increase gene expression of endothelial nitric-oxide synthase (eNOS) and antioxidant defense enzymes, resulting in improved endothelial function and lower blood pressure in vivo. In this study, we establish that equol (1-100 nm) causes acute endothelium- and nitric oxide (NO)-dependent relaxation of aortic rings and rapidly (2 min) activates eNOS in human aortic and umbilical vein endothelial cells. Intracellular Ca2+ and cyclic AMP levels were unaffected by treatment (100 nm, 2 min) with equol, daidzein, or genistein. Rapid phosphorylation of ERK1/2, protein kinase B/Akt, and eNOS serine 1177 by equol was paralleled by association of eNOS with heat shock protein 90 (Hsp90) and NO synthesis in human umbilical vein endothelial cells, expressing estrogen receptors (ER)α and ERβ. Inhibition of phosphatidylinositol 3-kinase and ERK1/2 inhibited eNOS activity, whereas pertussis toxin and the ER antagonists ICI 182,750 and tamoxifen had negligible effects. Our findings provide the first evidence that nutritionally relevant plasma concentrations of equol (and other soy protein isoflavones) rapidly stimulate phosphorylation of ERK1/2 and phosphatidylinositol 3-kinase/Akt, leading to the activation of NOS and increased NO production at resting cytosolic Ca2+ levels. Identification of the nongenomic mechanisms by which equol mediates vascular relaxation provides a basis for evaluating potential benefits of equol in the treatment of postmenopausal women and patients at risk of cardiovascular disease.

We recently reported that soy isoflavones increase gene expression of endothelial nitric-oxide synthase (eNOS) and antioxidant defense enzymes, resulting in improved endothelial function and lower blood pressure in vivo. In this study, we establish that equol (1-100 nM) causes acute endothelium-and nitric oxide (NO)-dependent relaxation of aortic rings and rapidly (2 min) activates eNOS in human aortic and umbilical vein endothelial cells. Intracellular Ca 2؉ and cyclic AMP levels were unaffected by treatment (100 nM, 2 min) with equol, daidzein, or genistein. Rapid phosphorylation of ERK1/2, protein kinase B/Akt, and eNOS serine 1177 by equol was paralleled by association of eNOS with heat shock protein 90 (Hsp90) and NO synthesis in human umbilical vein endothelial cells, expressing estrogen receptors (ER)␣ and ER␤. Inhibition of phosphatidylinositol 3-kinase and ERK1/2 inhibited eNOS activity, whereas pertussis toxin and the ER antagonists ICI 182,750 and tamoxifen had negligible effects. Our findings provide the first evidence that nutritionally relevant plasma concentrations of equol (and other soy protein isoflavones) rapidly stimulate phosphorylation of ERK1/2 and phosphatidylinositol 3-kinase/Akt, leading to the activation of NOS and increased NO production at resting cytosolic Ca 2؉ levels. Identification of the nongenomic mechanisms by which equol mediates vascular relaxation provides a basis for evaluating potential benefits of equol in the treatment of postmenopausal women and patients at risk of cardiovascular disease.
The interaction of estrogen with estrogen receptors (ER) 7 leads to the transcriptional activation of estrogen-responsive genes (1)(2)(3), including eNOS (4) and key antioxidant defense genes (5). In addition to its genomic actions, estrogen rapidly stimulates eNOS activity in cultured endothelial cells. ER␣, localized in plasmalemmal caveolae, has been implicated in the rapid activation of eNOS via pathways involving ERK1/2 and PI 3-kinase/Akt (6 -12). In pulmonary artery endothelial cells, coupling of ER␣ to G␣ i activates downstream signaling pathways leading to NO production (13). Inhibition of the functional association of the chaperone protein Hsp90 with eNOS abolishes 17␤-estradiol-stimulated NO release (14), and studies in HUVEC have shown that fluid shear stress stimulates phosphorylation of eNOS via PI 3-kinase/Akt, leading to increased NO production independent of cytosolic Ca 2ϩ mobilization (15). A similar Ca 2ϩ insensitivity has been reported for eNOS activation by 17␤-estradiol (E 2 ) in HUVEC (16), although in bovine and human aortic endothelial cells E 2 appears to elevate intracellular Ca 2ϩ (17,18).
Genistein and daidzein are polyphenolic isoflavones contained in soy protein (19), and intestinal bacteria metabolize daidzein to equol. Isoflavones are structurally similar to estrogen and generally bind with higher affinity to estrogen receptor ␤ (ER␤) compared with ER␣ (20). In humans consuming a soydeficient diet, plasma concentrations of genistein are Ͻ40 nM but can reach 4 M in Japanese consuming a soy-rich diet (19,21). Dietary phytoestrogens have been associated with a favorable cardiovascular risk profile in postmenopausal women (22). However, the limited number of clinical trials only describe marginal benefits of isoflavones in healthy postmenopausal women, highlighting the need for further studies with postmenopausal women at risk of diabetes, cardiovascular disease, breast cancer, and osteoporosis (23).
Long term treatment with genistein improves endothelial function in humans and animal models with mild to moderate hypertension (24 -27), whereas acute treatment enhances flowmediated dilation and endothelium-dependent relaxation (28 -32). Isoflavones may thus affect vascular tone by regulating either eNOS expression and/or bioavailability of NO (33). We recently reported that a soy protein diet rich in isoflavones increases the expression of eNOS and key antioxidant defense genes in aged male rats, resulting in improved NO-dependent relaxation and reduced arterial blood pressure in vivo (34).
Our present findings provide the first evidence that equol (0.1-100 nM) causes rapid activation of ERK1/2 and NO release in HUVEC and HAEC at basal cytosolic Ca 2ϩ levels and NOdependent relaxation. Equol-stimulated phosphorylation of Akt and eNOS serine 1177 was rapid (2 min) and accompanied by a dissociation of eNOS from caveolin-1 and association with Hsp90. Rapid activation of eNOS by equol was abrogated following inhibition of ERK1/2 and PI 3-kinase, and was not substantially affected by inhibitors of G protein-coupled receptors, Src family kinases, or the estrogen antagonists ICI 182,780 and tamoxifen.
Detection of Estrogen Receptors ER␣ and ER␤-HUVEC were seeded into chamber slides and cultured overnight in M199. For light microscopy, cells were fixed with 4% paraformaldehyde for 10 min. Fixative was replaced with 0.01 M Kreb's phosphate-buffered saline (KPBS); one group of cells was permeabilized with KPBS containing 0.1% Triton X-100 for 10 min to facilitate penetration of reagents and another group remained nonpermeabilized. After blocking with 5% bovine serum albumin (BSA), cells were incubated with primary antibodies raised against the carboxyl terminus of human ER␣ (amino acids 582-595; Calbiochem, 0.5 g ml Ϫ1 ) or human ER␤ (amino acids 458 -477; Zymed Laboratories Inc.; D7N, 0.5 g/ml) for 48 h (37). Biotinylated horse anti-mouse IgG (1:500) or goat antirabbit IgG (1:500; Vector Laboratories), respectively, were applied as secondary antibodies for 3 h, followed by avidin-biotinylated enzyme complex for 2 h. Nickel-enhanced diaminobenzidine was used to detect the immunoreactive loci in cells. After removing the chambers, HUVEC were air-dried and cover-slipped with DEPEX.
For double labeling, cells were fixed with 4% paraformaldehyde (10 min) and then placed in 0.01 M PBS followed by application of ER␣ and ER␤ antibodies together for 48 h. Alexa 488conjugated goat anti-rabbit IgG and Alexa 568-conjugated goat anti-mouse IgG (1:500) were mixed and applied together for 16 h, and cells were air-dried and cover-slipped with anti-fading agent (SlowFade, Molecular Probes).
For electron microscopy, HUVEC were treated with sucrose solutions in 0.01 M KPBS (10 -30% for 10 min). After three cycles of rapid freezing and thawing, cells were transferred to 10% sucrose and washed with KPBS. After blocking with 5% BSA, cells were incubated with rabbit anti-human ER␣ (amino acids 458 -477; Zymed Laboratories Inc.; D7N, 0.5 g/ml) for 48 h. Cells were thoroughly rinsed with KPBS and blocked with a mixture of 0.1% cold water fish gelatin (Electron Microscopy Sciences, Hatfield, PA) and 1% BSA in KPBS. Nanogold goat anti-rabbit IgG, FabЈ fragment (Molecular Probes) diluted in the same blocking solution 1:100, was used for 3 h followed by glutaraldehyde fixation (1.25%) in KPBS. Cells were washed in 0.2 M sodium citrate, pH 7.5, and gold particles were silver intensified with IntenSE kit (Amersham Biosciences). Cells were treated with 1% osmium tetroxide in 0.1 M phosphate buffer for 30 min, dehydrated in ascending series of ethanol, embedded in Durcupan ACM epoxy resin (Fluka), and polymerized at 56°C for 2 days. Ultrathin 50 -60 nm thin sections were cut using a Leica ultracut UCT ultramicrotome; ribbons were collected onto Formvar-coated single slot grids and examined with a JEOL electron microscope.
Immunocytochemical controls revealed no punctate staining after processing cells without primary antibodies (Fig. 1, B and D) or when ER antibodies were pre-absorbed with peptide antigens or recombinant proteins.
Single Cell [Ca 2ϩ ] i Measurements-HUVEC were loaded with 1 M fura 2-AM for 60 min in HEPES-buffered Dulbecco's modified Eagle's medium containing 20% FCS and then maintained in HEPES-buffered balanced salt solution (36,38). Fura 2-AM-loaded cells were mounted on a thermostated stage (37°C) of a Zeiss Axiovert 135 inverted microscope with a ϫ40 oil immersion objective and excited alternately at 350 and 380 nm via a LEP dual filter wheel system (Ludl, Hawthorne, NY). Fluoresced light, Ͼ420 nm, was captured with a 14-bit cooled CCD camera (Hamamatsu C4880-80) using Openlab software (Improvision, Coventry, UK). An image pair was captured every 1-3 s, and the ratio of fluorescence at 350 and 380 nm excitation was used as a measure of intracellular calcium [Ca 2ϩ ] i .
Intracellular cGMP Accumulation-Basal and stimulated cGMP levels were determined by radioimmunoassay, with inhibition of cGMP accumulation by L-NAME (100 M, 15 min) serving as an index of NO production (36). Confluent monolayers were washed twice with warmed Krebs-Henseleit buffer and then preincubated for 15 min with buffer containing L-arginine (100 M) and the phosphodiesterase inhibitor 3-isobutyl-1methylxanthine (IBMX, 500 M). Cells were then stimulated (30 s to 15 min) with 1-100 nM equol, daidzein, or genistein in the presence of IBMX and L-arginine (100 M), and responses were compared with E 2 or E 2 conjugated to bovine albumin (E 2 -BSA).
Immunoblotting Phosphorylated ERK1/2, Akt, and eNOS Ser 1177 -HUVEC were equilibrated in M199 containing 1% FCS for 4 h and then preincubated with Krebs-Henseleit buffer containing L-arginine (100 M) and IBMX (0.5 mM). In other experiments, HUVEC were pretreated with vehicle or U0126 (1 M, 30 min) or LY294,002 (10 M, 30 min) and then stimulated for 30 s to 5 min with equol, using thrombin (1 unit/ml, 2 min) as an internal control. We reported previously (36) that the structurally distinct MEK1/2 inhibitors U0126 and PD98059 do not have inhibitory effects on p38 MAPK or c-Jun NH 2terminal kinase (JNK) pathways in HUVEC. Reactions were stopped with ice-cold PBS containing 200 M sodium orthovanadate, and cell lysates were separated by SDS-PAGE, and proteins were electrotransferred onto polyvinyldifluoride membranes and then probed with a polyclonal antibody against dually phosphorylated (threonine 183/tyrosine 185) ERK1/2 or phosphospecific antibodies against Akt or eNOS serine 1177. Protein bands were detected by ECL, and densitometric analyses were performed using Scion Image software (Scion Corp.).
Immunoprecipitation of eNOS, Association with Hsp90, and Dissociation from Caveolin-1-HUVEC in T75 culture flasks were equilibrated in M199 containing 1% FCS for 4 h prior to treatment for 2 min with equol (100 nM) or thrombin (1 unit ml Ϫ1 ) in Krebs-Henseleit buffer containing FIGURE 1. Localization of ER␣ and ER␤ immunoreactivity in HUVEC. Cells were incubated with antibodies to ER␣ and ER␤. Weak fixation together with Triton X-100 treatment facilitated penetration of immunoreagents into cell nuclei, and nuclei immunopositive for ER␣ (A) or ER␤ (C) are shown with dotted immunoreactivity scattered in the chromatin. Using the antibody diluent (2% BSA) instead of the primary antibodies during immunocytochemical procedures (B-D) or preabsorption of primary antibodies with the corresponding peptide antigens (not shown) abolished staining of cells. Stronger fixation conditions together with omission of Triton X-100 from preincubation steps preserved IR in extranuclear sites but reduced nuclear signals for estrogen receptors; thus IR for ER␣ (E) or ER␤ (F) appears as faint dots throughout the cell. Under these conditions, double label immunofluorescence for the two receptors (G) shows limited overlapping of immunoreactive loci (red dots represent ER␣-IR and green dots ER␤-IR); the white rectangle (G) is shown at higher magnification (H). At the EM level, ER␤-IR was revealed by silver-intensified gold particles in the cell membrane (white arrows, I). n, nucleus. Black rectangle in I is shown at higher magnification in J. K, cytosolic Ca 2ϩ ([Ca 2ϩ ] i ) levels in HUVEC are unaffected by acute exposure to 10 and 100 nM equol but elevated biphasically in response to histamine. Images represent simultaneous measurements taken from a range of 6 -61 cells at any one time in two different cell cultures.
L-arginine (100 M) and IBMX (0.5 mM). Cells were washed with ice-cold PBS, and cell lysates were collected in RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) containing a protease inhibitor mixture (Sigma). Lysates were centrifuged at 13,000 ϫ g for 10 min, and supernatants were incubated overnight at 4°C with ExactaCruz immunoprecipitation matrix (Santa Cruz Biotechnology) precoated with an eNOS antibody. Immobilized immune complexes were washed twice with RIPA buffer, eluted from the matrix by boiling for 5 min in SDS lysis buffer, and analyzed by SDS-PAGE using antibodies against eNOS, caveolin-1, and Hsp90.
Statistical Analysis-Data are expressed as means Ϯ S.E. of measurements in 3-7 different endothelial cell cultures. Statistical analysis was performed by use of analysis of variance followed by a Student's t test, and a level of p Ͻ 0.05 was considered significant.

Intracellular Localization of ER␣ and ER␤ in HUVEC-Mild
fixation and permeabilization with Triton X-100 were used to detect ER␣ or ER␤ immunoreactivity within the nucleus of HUVEC (Fig. 1, A and C), with staining absent when the primary antibody was omitted (Fig. 1, B and D). Stronger fixation of cells not treated with Triton X-100 showed lower signal intensity within the nucleus, but immunoreactive loci were preserved throughout the cytoplasm (Fig. 1, E and F). Doubly labeled immunofluorescence showed ER␣ and ER␤ reactivity in different loci of the cell (Fig. 1, G and H), and electron microscopy revealed ER␤ reactivity in the nucleus and cytoplasm of cells, with a few immunoreactive loci also detected at the plasma membrane (Fig. 1, I and J (36,38). Closer examination of single cells revealed no localized [Ca 2ϩ ] i responses to equol. Similar responses were observed following acute (2 min) application of 17␤-estradiol (data not shown), confirming previous findings in HUVEC (16).

Equol Causes Rapid Endothelium-and NO-dependent
Relaxation-Endothelium intact aortic rings were preconstricted with PE (1 M) and exposed to cumulative doses of equol. As shown in representative original records (Fig. 2, A and  B), equol caused a dose-dependent relaxation that was inhibited by pretreatment with L-NAME (100 M). Fig. 2C summarizes the results from three different animals.
Isoflavones Do Not Acutely Increase cAMP or PGI 2 Release-Treatment of HAEC (data not shown) or HUVEC for 2 min with 100 nM equol, daidzein, or genistein had no significant effect on intracellular cAMP levels or PGI 2 production, whereas forskolin and histamine increased cAMP levels and PGI 2 production, respectively (Table 1).
Rapid Activation of ERK1/2 and eNOS by Equol, Daidzein, and Genistein-Because ERK1/2 has been implicated in the activation of eNOS by estrogen (6, 10, 12), we compared acute effects of E 2 , equol, daidzein, and genistein on EKR1/2 phosphorylation in HUVEC and HAEC. Treatment with isoflavones (100 nM) led to a rapid (2 min) phosphorylation of ERK1/2 (Fig.  3, A and C) in both cell types, and densitometric analyses of different cell cultures are shown in Fig. 3, B and D.
We reported previously that activation of A 2a -purinoceptors results in rapid phosphorylation of ERK1/2 and NO production in HUVEC independent of detectable increases in cytosolic Ca 2ϩ (36). To examine whether equol, daidzein, or genistein also acutely stimulate NO synthesis in endothelial cells, we treated HUVEC and HAEC with these isoflavones (100 nM, 2 min) and measured changes in intracellular cGMP accumulation in the absence or presence of L-NAME (100 M). Equol, daidzein, and genistein (and E 2 ) evoked rapid increases in cGMP accumulation, which were inhibited by L-NAME (Fig. 3E).
We then further examined concentration-and time-dependent actions of equol on ERK1/2 phosphorylation and cGMP accumulation in HUVEC. ERK1/2 was phosphorylated by equol concentrations as low as 1 nM (data not shown), with phosphorylation at higher concentrations (100 nM) detected within 30 s and maintained for up to 5 min (Fig. 4, A and B). Activation of ERK1/2 was paralleled by an increase in intracellular cGMP, with maximal NO release achieved at 2 min (Fig. 4C). All subsequent experiments were performed using 100 nM isoflavones or E 2 .
Equol Acutely Modulates Interactions of eNOS with Hsp90 and Caveolin-1-As eNOS activity is regulated by post-translational modifications, involving phosphorylation by Akt and interactions with regulatory proteins such as Hsp90 (14,39,40), we hypothesized that equol would acutely modulate eNOS interactions with caveolin-1 and Hsp90. Treatment of HUVEC for 2 min with equol (100 nM) or thrombin (1 unit ml Ϫ1 ) resulted in a rapid dissociation of eNOS from caveolin-1 and association with Hsp90 (Fig. 6H).
Effect of Pertussis Toxin on Isoflavone-stimulated ERK1/2 Phosphorylation and cGMP Accumulation-Previous studies have shown that activation of eNOS by E 2 requires plasma membrane ER␣ coupling to G␣ i , leading to activation of downstream signaling events (13). To evaluate the potential role of G proteins in equol-stimulated eNOS activity, HUVEC were pretreated for 120 min with vehicle or pertussis toxin (PTX) and exposed to equol (100 nM) for 2 min. Pertussis toxin treatment inhibited ERK1/2 phosphorylation in response to equol, daidzein, E 2 , and E 2 -BSA, whereas stimulated cGMP accumulation was decreased only marginally (Fig. 7B).
Effect of Src Inhibition on ERK1/2 and eNOS Activation-Src kinase has been implicated in the activation of Akt and eNOS in  HUVEC in response to E 2 (41). To examine the potential involvement of Src kinase in the actions of equol and other isoflavones, HUVEC were pretreated for 30 min with an Src kinase inhibitor PP2 (10 M). PP2 abrogated acute phosphorylation of ERK1/2 in response to equol, daidzein, and genistein (100 nM, 2 min) but had negligible effects on stimulated NO release (data not shown).
Estrogen Receptor Antagonists Do Not Inhibit Acute Activation of ERK1/2 and eNOS by Equol and Other Isoflavones-In view of previous reports for E 2 (11), we hypothesized that estrogen receptor antagonists would inhibit isoflavone-stimulated activation of ERK1/2 and eNOS. Pretreatment of HUVEC with ICI 182,780 (10 M) or tamoxifen (10 M) had no effect on acute (2 min) phosphorylation of ERK1/2 or cGMP accumulation in response to 100 nM equol, daidzein, genistein, or E 2 (Fig. 8, A  and B).
To further evaluate whether estrogen receptor antagonists affected equol and E 2 -stimulated cGMP accumulation over a longer time interval, we pretreated HUVEC with ICI 182,780 (10 M) and then measured intracellular cGMP after 15 min of treatment with either equol or E 2 . As shown in Fig. 8C, ICI 182,780 inhibited E 2 but not equol-stimulated cGMP production, confirming earlier reports of inhibition of E 2 -stimulated eNOS activity (after 15-30 min) in human endothelial cells by ER antagonists (9,11,14).

DISCUSSION
We have demonstrated that nutritionally relevant plasma concentrations of equol cause rapid activation of ERK1/2 and Akt in human endothelial cells, leading to phosphorylation of eNOS, NO release, and endothelium-dependent relaxation of aortic rings. Although HUVEC express both ER␣ and ER␤, acute stimulation of NO synthesis was insensitive to inhibition by the ER antagonists ICI 182,780 and tamoxifen. Rapid activation of eNOS in HUVEC by equol at basal cytosolic Ca 2ϩ levels is consistent with previous reports of shear stress and E 2 -stimulated NO production at low intracellular Ca 2ϩ (15,16). Moreover, as reported for E 2 (14,39,40), equol causes a rapid dissociation of eNOS from caveolin-1 and association with the chaperone protein Hsp90. To our knowledge, our findings provide the first evidence that equol, a metabolite of the isoflavone daidzein, enhances association of Hsp90 with eNOS in human endothelium.
We have shown, for the first time, a similar distribution of ER␣ and ER␤ immunoreactivity in HUVEC, confirming earlier reports of immunocytochemical staining (42) and detection of mRNA and protein for ER␣ in HUVEC (14). After mild fixation, immunoreactivity for both receptors appears in the nuclei of HUVEC, implicating a genomic function. Fixation of cells without permeabilization revealed punctate immunoreactivity over both extranuclear and nuclear sites. Such strategy has been used to detect estrogen receptors at extranuclear sites in other cell types (43). Because ER␣ and ER␤ can form heterodimers (44 -46), we further examined whether receptors were co-localized at immunoreactive sites. Negligible co-localization at nuclear or extranuclear sites suggests that formation of ER complexes may not required for acute activation of eNOS in HUVEC in response to equol or E 2 . A study of en face arterial endothelium has also revealed negligible co-localization of ER␣ and ER␤ at extranuclear or nuclear sites (45), and dimerization of ER␣ is not required for rapid ER␣ coupling to eNOS in ovine endothelial cells (46).
The distribution of ER␤ immunoreactivity in HUVEC was independent of the markers used (nickel-enhanced diaminobenzidine, fluorochromes, or ultrasmall gold particles intensified with silver), and the latter marker allowed an accurate determination of the subcellular location of ER␤. Silver-intensified gold grains representing ER␤ were detected in the nucleus and cytoplasm of cells, with some labeling at the cell membrane. Previous studies have identified and implicated membrane ERs in nongenomic activation of signaling pathways (46 -50). Further studies are required to establish whether ER␤ or ER␣ associate with caveolar protein complexes in HUVEC, particularly in view of reports that subpopulations of ER␤, ER␣, and alternatively spliced and truncated ER␣ isoforms appear to be coupled to eNOS in ovine and EAhy.926 endothelial cells (8, 46 -48).
We have shown previously in HUVEC that vasoactive agonists such as histamine activate eNOS via an increase in cytosolic Ca 2ϩ (38), whereas A 2a -purinoceptor stimulation elicits rapid Ca 2ϩ -independent and ERK1/2-dependent NO release (36). We report here that acute treatment of HUVEC with equol does not alter cytosolic Ca 2ϩ (Fig. 1K), reminiscent of E 2 -mediated activation of eNOS independent of Ca 2ϩ mobilization (16). However, acute regulation of NO synthesis varies in endothelial cells from different vascular beds or species, because E 2 causes a transient rise in cytosolic Ca 2ϩ in bovine aortic and human arterial endothelial cells (17,18). Ca 2ϩ -independent activation of eNOS in HUVEC in response to equol (Fig. 1K), E 2 (14,16), adenosine (36), or shear stress (15) may reflect the ability of phosphorylation to activate eNOS at resting cytosolic Ca 2ϩ levels by enhancing electron flux through the reductase domain of the enzyme and reducing calmodulin dissociation (51).
We have also demonstrated that nanomolar concentrations of equol, daidzein, and genistein acutely (30 s to 2 min) stimulate phosphorylation of ERK1/2 and NO synthesis in HUVEC and HAEC. Increases in NO synthesis were inhibited by the NOS inhibitor L-NAME. As eNOS inhibition had no effect on equolstimulated ERK1/2 phosphorylation (data not shown), this suggests that activation of ERK1/2 lies upstream of eNOS. Inhibition of ERK1/2 with U0126 significantly attenuated equol-stimulated eNOS serine 1177 phosphorylation and NO production. Previous studies established that E 2 causes a rapid-  phosphorylation of ERK1/2 and eNOS (6 -12), but to our knowledge this is the first report that the isoflavones equol, daidzein, and genistein stimulate ERK1/2 and eNOS phosphorylation in human endothelium within 2 min.
Association of Hsp90 with eNOS is critically important for eNOS-mediated NO production (14,39), and post-translational regulation of eNOS by tyrosine kinases and Hsp90 inhibits uncoupling of the enzyme and eNOS-dependent O 2 Ϫ generation (54). We have shown that rapid (2 min) association of Hsp90 with eNOS induced by equol is accompanied by NO release. Studies in porcine aortic endothelial cells have implicated AMP-activated protein kinase (AMPK) in eNOS activation by 100 nM E 2 and suggest that catechol conversion of E 2 leads to AMPK activation and association of Hsp90 with eNOS (40). Interestingly, E 2 -mediated AMPK activation is unaffected by inhibition of Src family kinases, PI 3-kinase, ERK1/2, protein kinase C, and the ER antagonist ICI 182,780 (40). Thus, in porcine aortic endothelial cells co-activation of AMPK and PI 3-kinase by E 2 appears to mediate eNOS activation.
Isoflavones may act as inhibitors of cAMP phosphodiesterase (55). However, as cAMP levels in HUVEC were unaffected by acute treatment with equol and other isoflavones, effects of cAMPdependent protein kinase inhibitors on eNOS activation by isoflavones were not examined. Previous studies with bovine aortic endothelial cells (BAEC) reported that longer term treatment with genistein (100 nM, 10 -120 min) stimulates activation of eNOS independent of ERK1/2, PI 3-kinase/Akt, or the involvement of estrogen receptors (56). The lack of inhibition of genistein-stimulated eNOS activity by ICI 182,780 is consistent with our present results. The discrepancies between our findings and those of Liu et al. (56), based primarily on experiments with BAEC, may reflect differences in cell type, species, and/or the time interval of genistein treatment (30 s versus 10 min). Moreover, Liu et al. (56) only reported increases in cAMP levels in BAEC in response to 10 nM to 100 M genistein. In our study, treatment of HUVEC with isoflavone concentrations of Ͼ100 nM actually inhibited ERK1/2 phosphorylation and did not lead to enhanced NO synthesis (data not shown). The inhibition of tyrosine kinase activity by genistein (10 -50 M) is well known (57), and our present and previous results in HUVEC (36) highlight the concentration-dependent actions of genistein and other isoflavones on ERK1/2 activity. Natural polyphenolic compounds in black and green tea and red wine also activate eNOS (58 -60). Epigallocatechin-3-gallate (40 -100 M) stimulates NO release in bovine endothelial cells only after 15 min, with moderate activation of Akt and eNOS serine 1177 (58). Epigallocatechin-3-gallate-stimulated phosphorylation of eNOS was abolished by Akt and cAMP-dependent protein kinase inhibitors but notably unaffected by ERK1/2 inhibition. In porcine endothelial cells, black tea polyphenols stimulate Akt and eNOS phosphorylation within 5 min, involving a Ca 2ϩ -and p38 MAPK -dependent activation of PI 3-kinase/Akt (59). Red wine polyphenols rapidly activate ERK1/2 and NO production in several endothelial cell types (60) and stimulate PI 3-kinase/Akt-dependent phosphorylation of eNOS in porcine coronary arteries (61). Although actions of trans-resveratrol on ERK1/2 and NO production in cultured endothelial cells are inhibited by ICI 182,780 and tamoxifen (60), studies with tea polyphenols have not assessed the potential inhibitory actions of ER antagonists.
Although pertussis toxin inhibited isoflavone-stimulated ERK1/2 phosphorylation, cGMP levels were only reduced marginally, suggesting activation of eNOS via PI 3-kinase/Akt. The lack of significant inhibition of equol-stimulated NO production by pertussis toxin contrasts with previous studies with E 2 in ovine endothelial cells, and suggests that rapid stimulation of eNOS by isoflavones in HUVEC is not mediated via G proteincoupled receptors.
We found that isoflavone-stimulated eNOS activation was independent of ERs, because the classical ER antagonists ICI 182,780 and tamoxifen failed to inhibit equol-, daidzein-, and genistein-stimulated cGMP production. The concentration and preincubation times used in our study are similar to those in previous studies examining the actions of E 2 . The only difference is the much shorter treatment (2 min versus 10 -30 min) used in our experiments. When we compared the inhibitory action of ICI 182,780 on E 2 -and equol-stimulated cGMP accumulation over 15 min, the ER antagonist selectively inhibited the actions of E 2 , confirming that equol-stimulated NO production was independent of ER activation. As longer term responses to 17␤-estradiol were inhibited by ICI 182,750, this implicates interactions of 17␤-estradiol with the classical ERs detected in HUVEC in this study (Fig. 1).
In summary, this study in human umbilical vein endothelial cells establishes that rapid activation of ERK1/2 and PI 3-kinase/ Akt signaling pathways by equol, daidzein, and genistein is paralleled by phosphorylation of eNOS serine 1177, Hsp90 association with eNOS, and NO release (Fig. 9). Furthermore, rapid phosphorylation of eNOS serine 1177 in response to equol is abrogated by inhibitors of ERK1/2 and PI 3-kinase, and we have demonstrated cross-talk between these two signaling pathways (Fig. 6, E and F). Based on our present findings, we cannot exclude the possibility that equol also modulates NO synthesis via other NOS isoforms expressed at much lower levels in subcultured HUVEC (62,63). Activation of tyrosine kinases and Hsp90 may act in concert to regulate the balance of NO ⅐ and O 2 . generation by eNOS (54). In this context, prolonged treatment of J774 cells with equol increases the bioavailability of NO by inhibiting O 2 . production (64). We have also documented that equol induces endotheliumand NO-dependent relaxation of isolated aortic rings. Similar studies have reported that dilatory responses to dehydroequol in isolated rat aortic rings are insensitive to inhibition of eNOS (65), whereas dilation of human forearm resistance arteries is largely NO-dependent (66). Equol is derived from the soy isoflavone daidzein, and humans have acquired the ability to exclusively synthesize (S)equol (67). Equol and other isoflavones may activate similar intracellular signaling pathways as a result of their phenolic ring structure. In a recent study assessing bioavailability and metabolism of soy isoflavones, it was noted that only 30% of the subjects were equol producers with negligible differences in equol production observed with age or gender (68). Thus, use of dietary sources of isoflavones as alternative treatment for postmenopausal women and patients at risk of cardiovascular disease will require evaluation of the ability of subjects to metabolize daidzein to equol. Our findings provide insight into the signaling pathways regulating eNOS activity in human endothelium and identify equol as a potent activator of acute NO production.