Striatin assembles a membrane signaling complex necessary for rapid, nongenomic activation of endothelial NO synthase by estrogen receptor α

Abstract

Steroid hormone receptors (SHRs) are ligand-activated transcription factors that regulate gene expression. SHRs also mediate rapid, nongenomic cellular activation by steroids. In vascular endothelial cells, the SHR for estrogen, estrogen receptor (ER) α, is targeted by unknown mechanisms to a functional signaling module in membrane caveolae that enables estrogen to rapidly activate the mitogen-activated protein kinase and phosphatidylinositol 3–Akt kinase pathways, and endothelial NO synthase (eNOS). Here we identify the 110-kDa caveolin-binding protein striatin as the molecular anchor that localizes ERα to the membrane and organizes the ERα–eNOS membrane signaling complex. Striatin directly binds to amino acids 183–253 of ERα, targets ERα to the cell membrane, and serves as a scaffold for the formation of an ERα–Gαi complex. Disruption of complex formation between ERα and striatin blocks estrogen-induced rapid activation mitogen-activated protein kinase, Akt kinase, and eNOS, but has no effect on ER-dependent regulation of an estrogen response element-driven reporter plasmid. These findings identify striatin as a molecular scaffold required for rapid, nongenomic estrogen-mediated activation of downstream signaling pathways. Furthermore, by demonstrating independent regulation of nongenomic vs. genomic ER-dependent signaling, these findings provide conceptual support for the potential development of “pathway-specific” selective ER modulators.

Estrogen, acting via its two receptors, ERα and ERβ, exerts diverse effects on a variety of tissues (reviewed in ref. 1). Estrogen receptors (ERs) regulate cellular function via two signaling pathways referred to as “genomic” and “nongenomic.” Genomic effects of estrogen are mediated by nuclear ERs that act as ligand-activated transcription factors. Nongenomic effects of estrogen are also mediated by ERs, but they occur relatively rapidly and do not involve alterations in gene expression. We and others (2–9) have shown that vascular cells express ERs capable of mediating both genomic and nongenomic effects. Nongenomic effects of estrogen in the vasculature have been best studied in vascular endothelial cells. In these cells, estrogen rapidly and sequentially activates the mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase–Akt kinase pathways, which increases the activity of endothelial NO synthase (eNOS) (4, 10–13). This pathway contributes to the rapid, estrogen-induced arterial vasodilation observed in clinical studies (14, 15). ERα is required for rapid activation of eNOS by estrogen (4, 10–12), and recent reports demonstrate that these nongenomic effects are transduced by a subpopulation of ERα localized to caveolae (3), cell membrane microdomains important for many signaling pathways. However, the molecular mechanisms by which ERα localizes to the caveolae and the details by which ERα activates membrane-associated eNOS remain obscure.

We now report the identification of striatin as an ERα-binding protein. Striatin is a 780-aa protein first isolated from rat brain synaptosomes (16, 17). Striatin is a calmodulin-binding member of the WD-repeat family of proteins and contains at least four protein–protein interaction domains, including caveolin-binding (amino acids 55–63), coiled-coil (amino acids 70–116), Ca²⁺-calmodulin-binding (amino acids 149–166), and a series of eight WD repeat (amino acids 419–780) motifs (16) (Fig. 1 A). WD repeats are found in a wide variety of proteins, including the signal-transducing G protein β-subunit, as well as other proteins that organize and regulate signal transduction, cytoskeletal organization, and vesicular fusion (18). Striatin also forms a complex with protein phosphatase 2A (19), which we have shown recently also binds directly to and regulates the function of ERα (20). We now show further that striatin targets ERα to the cell membrane and serves as a scaffold for the assembly of proteins required for rapid, estrogen-induced activation of eNOS.

Fig. 1.

ERα binds directly to striatin via amino acids 183–253 of the ER. (A) Schematic representation of relevant domains of striatin. (B) Striatin coimmunoprecipitates with ERα in a variety of cells, including cells that endogenously express ERα [EAhy926 cells (human endothelial cell line), MCF7 cells (human breast cancer cell line), and GH3B6 cells (rat pituitary cell line)] and ERα-null cells transfected with an ERα expression plasmid [Rad91 cells (human radial artery-derived vascular smooth muscle cell line)]. IP, immunoprecipitation of ERα; NI, nonimmune immunoprecipitation; L, fraction of total cell lysates. (C) Short-term E2 treatment enhances complex formation between ERα and striatin. ERα was immunoprecipitated from Rad91 cells transfected with ERα, with or without exposure to 10^–8 M E2 for 20 min. (D) Striatin binds ERα directly. GST fusion proteins were incubated with lysates of Rad91 cells transfected with ERα (Left) or with recombinant ERα (Right). (E) Striatin binds to amino acids 183–253 of ERα. (Upper) Schematic representation of ERα fragments used in GST pulldown experiments. (Lower) GST fusion proteins containing full-length ERα or ERα fragments were incubated with lysates of Cos1 cells.

Methods

Cell Culture, Transfection, and Luciferase Assay. Rad91 cells are a spontaneously immortalized vascular smooth muscle cell line derived from a human radial artery. EAhy926 cells, a human aortic endothelial cell hybridoma, were the kind gift of C.J. Edgell (University of North Carolina, Chapel Hill). GH3B6 cells are a rat pituitary tumor cell line; Cos1 cells and MCF7 human breast cancer cells were obtained from American Type Culture Collection. Where indicated, cells were treated with 10^–8 M 17β-estradiol (E2) or an identical amount of vehicle for 20 min. In coincubation experiments, cells were preincubated with either the ER antagonist ICI 182,780 (10^–7 M) for 30 min or pertussis toxin (PTX) (100 ng/ml) for 120 min before the E2.

For the protein expression studies, cells were transfected by electroporation as described (6, 21). For luciferase assays, cells were transfected with PolyFect (Qiagen, Valencia, CA). Cos1 cells were plated in 12-well plates and cultured in DMEM without phenol red with 10% estrogen-deficient FBS overnight. After transfection (20 h), the cells were washed and treated with either 10^–8 M E2 or vehicle for 9 h. Cells were lysed in reporter lysis buffer (Promega), and luciferase assays (luciferase assay system, Promega) and β-galactosidase assays (Tropix, Bedford, MA) were performed according to the manufacturer's guidelines.

Plasmids. The bait plasmid PGBT9-ERα was constructed by inserting the 1.8-kb full-length cDNA of human ERα into the EcoRI site of the Gal4 DNA-binding domain vector pGBT9 (Clontech). The pGAD10-ERα and pGAD10-ERβ plasmids were cloned by inserting full-length ERα and ERβ into pGAD10 vector (Clontech). The expression plasmids pCMV3-ERα, pCMV3 ER-271, and the reporter plasmid ERE-Luc have been described (20, 21). pCMV3–striatin was constructed by cloning full-length rat striatin cDNA into pCDNA 3.1 (Invitrogen). The pEGFP-H1 and pEGFP-ER176–253 constructs were made by cloning N-terminal striatin 1–203 and PCR-derived ERα 176–253 into pEGFP-C2 vector (Clontech) into the EcoRI site and EcoRI/SmaI sites, respectively. The Gαi2 expression plasmid was obtained from the Guthrie Research Institute (Sayre, PA). The plasmids for the GST pull-down experiments were constructed by cloning the full-length Gαi2, H1 (striatin_1–203), full-length ERα, truncated mutants ERα 1–271 and ERα 176–595, and PCR-derived ERα fragments including ERα 176–253, 254–370, 176–282, and 183–253 into pGEX-4T-1 (Amersham Pharmacia). GST–striatin was a kind gift from A. Monneron (Université de la Méditerranée, Marseille, France). The pTAT-HA vector with 6-His-tag was a kind gift from S. F. Dowdy (Washington University School of Medicine, St. Louis). A PCR-generated KpnI-EcoRI fragment containing ERα_176–253 was cloned into the corresponding sites of plasmid pTAT-HA vector.

Yeast Two-Hybrid Screen. The yeast two-hybrid screen was performed by using a human heart cDNA library (Clontech). The human full-length ERα was cloned as “bait” into the Gal4 DNA-binding domain vector pGBT9. PGBT9 ERα and the human heart library were transformed into yeast strain Y190 (lacZ and HIS3 reporter genes) by using the lithium acetate method. The transformants were selected for growth on synthetic dropout selective medium plus 3-amino 1,2,4-triazole. Clones were screened for the ability to transactivate both histidine and β-galactosidase reporters. The interaction between ERα and H1 was confirmed by reintroduction of the individual plasmids into the host yeast Y190.

GST Pull-Down and in Vitro Binding Assay. GST fusion pulldown experiments were performed as described (20). Recombinant human ERα (50 ng; Calbiochem) was used instead of cell lysates in a subset of these experiments.

Immunoprecipitation. Immunoprecipitation and coimmunoprecipitation experiments were performed as described (20).

In Vitro Binding Assay. GST or GST fusion proteins were incubated with rERα in the presence or absence of partially purified full-length striatin or N-terminal striatin_1–203 (H1). Partially purified striatin was prepared as follows. Striatin was immunoprecipitated from Cos1 cell lysates as described above, and the immunopellet was resuspended in 0.2 ml of resolubilization buffer (50 mM Tris, pH 7.5/5 mM DTT/0.5% SDS). This solution was boiled for 5 min, and after addition of 1 ml of Tris lysis buffer (TLB) it was centrifuged for 5 min at 14,000 × g. A second immunoprecipitation was carried out on the supernatant, and these immunopellets were resolubilized in 0.2 ml of resolubilization buffer, diluted with 1 ml of TLB, and used for in vitro binding assays. The remainder of the GST pulldown protocol described above was then carried out.

Cell Fractionation. EAhy926 cells were cotransfected with ERα and striatin plasmids or ERα and control vector by electroporation. The cells were plated in 100-mm dishes and grown in phenol red-free DMEM in 10% estrogen-deficient FBS for 24 h after transfection. Lysates were prepared in 1 ml of homogenization buffer (20 mM Tris·HCl, pH 7.5/100 mM NaCl/1 mM EDTA/1 mM PMSF/protease inhibitor mixture). Crude lysates were centrifuged at 3,000 × g for 10 min at 4°C. The pellet containing nuclear proteins was recovered. The postnuclear supernatant fraction was next centrifuged at 45,000 × g for 1 h at 4°C, and the cell membrane pellet was recovered. The fractionated pellets were resuspended in Tris lysis buffer. The purity of isolation of the plasma membrane and nuclear fractions was examined by immunoblotting for membrane proteins [epidermal growth factor receptor (EGFR) and insulin-like growth factor receptor (IGFR)] and the nuclear protein histone.

Immunoblotting. Protein samples were resolved by SDS/PAGE, transferred to nitrocellulose membranes, and immunoblotted with the appropriate primary Ab. Abs used include anti-ERα (Ab7; 1:200, NeoMarker), anti-striatin (1:1,000, Transduction Laboratories), anti-EGFR, anti-IGFR, and anti-histone (1:500; Santa Cruz Biotechnology), anti-Gαi (1:1000, Calbiochem), anti-eNOS, phospho-eNOS, MAPK, phospho-MAPK (1:1,000, Cell Signaling Technology), and AKT and phospho-AKT (1:1,000, Pharmingen). The immunoblots were then visualized by ECL (Amersham Pharmacia).

Immunostaining. EAhy926 cells cotransfected with ERα and striatin plasmids or ERα and control vector, were plated in 12-well tissue culture plates (Becton Dickinson Labware) on coverslips and grown in phenol red-free DMEM in 10% estrogen-deficient FBS for 24 h. The cells were fixed in 3.7% paraformaldehyde for 10 min and permeabilized with 0.3% Triton X-100 for 15 min. After blocking with 10% donkey serum for 1 h, double immunostaining was performed by incubating the cells with rabbit anti-ERα HC 20 (1:500, from Santa Cruz Biotechnology) and mouse anti-striatin Ab (1:1,000) for 1 h, washing three times with PBS followed by incubation with FITC-conjugated donkey anti-rabbit secondary Ab (1:200) and Cy3-labeled donkey anti-mouse secondary Ab (1:1,000) (secondary Abs were purchased from Jackson ImmunoResearch) and then examined by fluorescence microscopy.

Purification of Permeable Tat–ERα Fragment Fusion Peptides. The pTat-HA ERα_176–253 plasmid was then transformed into Escherichia coli XL10-gold strain. Protein expression was induced by using 0.2 mM isopropyl β-d-thiogalactoside for 5 h at 37°C. Bacterial pellets were resuspended in 10 ml of buffer Z (8 M urea/100 mM NaCl/20 mM Hepes, pH 8.0), sonicated, and centrifuged at 12,000 × g for 20 min, and the supernatant was equilibrated in 10 mM imidazole. The fusion proteins were purified by using Ni-NTA agarose (Invitrogen). The Tat fusion peptides were then eluted in 5 ml of buffer Z containing 100, 250, and 500 mM imidazole, followed by desalting on a PD-10 column (Amersham Biosciences).

Statistical Analysis. Unless indicated otherwise, all experimental results presented represent a minimum of three independent experiments. In Fig. 4, mean values were compared by using Student's t test, and P ≤ 0.05 was considered significant.

Fig. 4.

Disruption of striatin-ERα binding prevents E2-induced nongenomic, but not genomic, signaling. (A) Overexpression of a peptide consisting of the striatin-binding domain within ERα, ER_176–253, prevents complex formation between ERα and striatin. (B) Overexpression of the blocking peptide ER_176–253 does not interfere with transcriptional transactivation of ERα by E2. Cos1 cells were transiently transfected with an ERα expression plasmid and an estrogen response element-driven luciferase reporter plasmid, with a plasmid encoding ER_176–253 or an empty vector. Cells were treated with vehicle alone (open bars) or 10^–8 M E2 (filled bars). E2-induced activation of ERα was somewhat greater in the presence of ER_176–253. Bars represent mean ± SE from four independent experiments. *, P < 0.05 vs. empty vector. (C) Overexpression of ER_176–253 blocks E2-induced increases in phosphorylation of MAPK. Lysates of Cos1 cells transfected with ERα and a control vector, or ERα and ER_176–253, were immunoblotted for phospho-MAPK (pMAPK) or total MAPK, after treatment with E2 for various durations. Bar graphs show the mean ± SE for four independent experiments. *, P < 0.01; **, P < 0.05 vs. time 0. (D) Overexpression of the blocking peptide ER_176–271 prevents E2-induced phosphorylation of Akt kinase. EAhy926 cells expressing endogenous ERα were incubated with Tat-GFP (○) or Tat-ER_176–253 (▪) for 6 h before E2 treatment. Total protein lysates were then immunoblotted for phospho-AKT (pAKT) normalized for the amount of total Akt. These ratios were normalized to one for the vehicle-treated cells. Results represent mean ± SE derived from four independent experiments. *, P < 0.05 vs. time 0. (E) Overexpression of the blocking peptide ER_176–253 prevents E2-induced phosphorylation of eNOS. EAhy926 were incubated with Tat-GFP (○) or Tat-ER_176–253 (▪) for 6 h before E2 treatment. Total protein lysates were then immunoblotted for phospho-eNOS (peNOS) normalized for the amount of total eNOS. These ratios were normalized to one for the vehicle-treated cells. Results represent mean ± SE derived from four independent experiments. *, P < 0.05 vs. time 0.

Results

A yeast two-hybrid system screen by using full-length human ERα as bait led to isolation of clone H1 (data not shown), which proved identical to the first 203 aa of striatin.

Immunoprecipitation studies demonstrated that striatin complexes with ERα in mammalian cells. Striatin and ERα were consistently coimmunoprecipitated from a variety of cell types, including EAhy926 endothelial cells, which contain the functional ERα–eNOS membrane signaling complex (22) (Fig. 1B). Notably, in some experiments, striatin appears as a doublet which might be due to different phospho-forms of the protein. The ERα–striatin interaction was enhanced substantially by short-term exposure of cells to E2 (Fig. 1C). In vitro studies with GST–striatin fusion proteins also demonstrated complex formation between striatin and ERα in cell lysates and direct binding of striatin to recombinant ERα (Fig. 1D). Studies with ERα deletion mutants identified the region of ERα that mediates striatin binding as amino acids 183–253 in the N-terminal A/B domain of ERα (Fig. 1E), and this was confirmed by reciprocal experiments by using a GST–striatin fusion peptide (amino acids 1–203) with lysates of cells expressing either full-length ERα or ERα deletion mutants (data not shown).

In EAhy926 cells, ERα is localized primarily in the nucleus with only a small proportion in the region of the plasma membrane (Fig. 2A), whereas striatin is detectable only weakly (Fig. 2B). Overexpression of striatin markedly changes the distribution of ERα by substantially increasing the proportion of ERα that is distributed along the plasma membrane (Fig. 2D). Membrane-associated ERα appears in a highly concentrated, punctate pattern, as does a subpopulation of striatin (Fig. 2E) that colocalizes with the plasma membrane ERα (Fig. 2F). In unstimulated cells, striatin also colocalizes along the plasma membrane with eNOS (Fig. 2 G–I). Membrane-associated ERα has previously been shown to activate growth factor receptor tyrosine kinases such as IGFR (23) and EGFR (24), which reside in caveolae (25). Overexpression of striatin substantially increases the amount of ERα in the membrane-enriched fraction containing EGFR (Fig. 2 J) and IGFR (data not shown) and slightly decreases the amount of ERα in the nuclear fraction (Fig. 2 J), although it has no effect on the overall abundance of ERα in the cell (data not shown).

Fig. 2.

Striatin targets ERα to the plasma membrane. (A–C) EAhy926 cells without overexpression of striatin were immunostained for ERα (A) or striatin (B), and the images were merged (C). Under these conditions only minimal amounts of membrane-associated ERα is detected. (D–I) EAhy926 cells transfected with a striatin expression plasmid were immunostained for ERα (A and D), striatin (B, E, and H), or eNOS (G), and the images were merged (C, F, and I). Under these conditions substantial membrane-associated ERα is detected, which colocalizes with striatin, as does eNOS. (J) Overexpression of striatin increases the proportion of ERα detectable in membrane fractions. Lysates of EAhy926 cells transfected with a control vector (–) or a striatin expression plasmid (+) were fractionated by differential centrifugation. The purity of plasma membrane and nuclear fractions was identified by immunoblotting for predominantly membrane proteins [EGFR and IGFR (data not shown)] or the nuclear protein histone.

Previous reports have shown that E2-induced activation of both the MAPK and Akt kinase cascades (4) is preceded by formation of a complex between ERα and Gαi in endothelial cells and in Cos cells transfected with eNOS and ERα (11, 12, 26). E2 activation of eNOS was reported to be blocked by PTX (26) and the ER antagonist ICI 182,780, supporting that eNOS stimulation by E2 requires activation of both ERα and Gαi (26–28). As shown in Fig. 3A, in endothelial cells, E2 stimulates formation of a complex containing ERα, Gαi, and striatin, and formation of this complex is inhibited by either PTX or ICI 182,780. PTX also abolishes E2-induced phosphorylation of MAPK (Fig. 3B) and Akt kinase (data not shown). We tested whether Gαi binds directly to ERα or striatin by using recombinant proteins. Although an interaction was observed between striatin and ERα, no interaction was detectable between Gαi and ERα under identical experimental conditions (Fig. 3C). In contrast, full-length, but not N-terminal striatin, bound Gαi (Fig. 3D), suggesting that striatin mediates, and is necessary for complex formation between Gαi and ERα. To more directly test this possibility, recombinant ERα was incubated with Gαiinthe absence or presence of full-length striatin or an N-terminal striatin peptide. Minimal complex formation was observed between Gαi and ERα in the absence of striatin (Fig. 3E). In the presence of full-length striatin, more abundant Gαi–ERα complex formed. This complex was not observed in the presence of a truncated, N-terminal striatin peptide (Fig. 3E). These data support that striatin mediates the formation of a complex between ERα and Gαi, and that the G protein binds to a domain(s) of striatin outside of the N-terminal 203 aa containing the ERα-binding site (compare Fig. 1D). These data are consistent with a model in which the C-terminal WD domain(s) of striatin mediate Gαi binding, similar to the role of the WD domains in the interaction between subunits of the heterotrimeric G proteins (29). GST fusion pulldown experiments and coimmunoprecipitation approaches both showed that striatin also forms a complex that contains eNOS (Fig. 3F).

Fig. 3.

Striatin assembles a complex containing ERα and Gαi. (A) E2 induces Gαi to complex with striatin and ERα, and this is blocked by coincubation with PTX or the ER antagonist ICI 182,780 (ICI). Equivalent recovery of Gαi was shown by immunoblotting (data not shown). SFM, serum-free medium. (B) PTX abolishes E2-induced phosphorylation of MAPK. *, P < 0.05 vs. SFM. n = 3 independent experiments. (C)Gαi does not bind directly to ERα. (D)Gαi binds to the C terminus of striatin. GST fusion proteins containing either full-length striatin or the N terminus of striatin (striatin_1–203) were incubated with lysates of Cos1 cells transfected with Gαi. (E) Striatin is required for complex formation between ERα and Gαi. GST or GST–Gαi fusion proteins were incubated with recombinant (rER), in the absence or presence of purified full-length striatin or striatin_1–203. rERα, recombinant ERα. (F) Striatin forms a complex with eNOS. eNOS was identified by immunoblotting in GST fusion pulldowns and by coimmunoprecipitation of lysates from EAhy926 cells.

To test whether striatin is required for activation of the ERα–phosphatidylinositol 3-kinase–Akt–eNOS signaling pathway, blocking peptides were first developed and tested for their ability to disrupt the ERα–striatin interaction. Mammalian expression vectors and Tat fusion proteins (30) were created that express amino acids 176–253 of ERα, the region that mediates its interaction with striatin, an approach that we have used previously to disrupt protein–protein interactions required for signal transduction (20, 31, 32). The amino acids 176–253 blocking peptide disrupts binding between ERα and striatin both in overexpression studies (Fig. 4A) and when introduced into cells as a Tat fusion peptide (30) (data not shown). Disruption of the ERα–striatin interaction moderately enhances E2-mediated transcriptional activation of an estrogen response element-driven reporter plasmid by the ER (Fig. 4B), supporting that disruption of the ERα–striatin interaction does not abolish transcriptional activation of the ER. These data do not, however, rule out the possibility that striatin binding of ERα participates in the regulation of ERα-mediated transcription of endogenous genes. E2, via ERα, also rapidly activates MAPK in endothelial cells and in COS cells transfected with ERα and eNOS (4, 33). Overexpression of the ERα–striatin blocking peptide prevented E2-induced MAPK activation (Fig. 4C). In all cases, similar results were obtained when blocking peptide was delivered to cells by using the Tat fusion peptide approach (data not shown). Endothelial cells expressing endogenous ERα and eNOS were studied to test the importance of the ERα–striatin interaction in the rapid activation of eNOS by E2 in intact cells. E2 activated Akt and eNOS, as assessed by an increased level of phosphorylation of each protein, and this was completely abolished when the ERα–striatin interaction was disrupted by the blocking peptide (Fig. 4 D and E). These data show that the interaction between ERα and striatin is required for E2 to activate Akt and eNOS in endothelial cells and that disruption of the ERα– striatin interaction does not abolish ligand-activated transcription by ERα.

Discussion

These studies identify striatin as a scaffold that promotes localization of ERα to the plasma membrane and assembly of the signaling complex of ERα and Gαi that is required for ERα-dependent activation of MAPK, phosphatidylinositol 3–Akt kinase, and eNOS, a critical regulator of many physiologic and pathophysiologic processes. Estrogen regulates fundamental physiological processes in both reproductive and nonreproductive target tissues, including bone, brain, and the cardiovascular system. The diversity of estrogen action results in part from the ability of ERs to act as both transcription factors that regulate gene expression (i.e., genomic effects) and as signaling proteins that rapidly recruit and activate kinase-dependent signaling pathways (nongenomic effects). The genomic effects of estrogen are themselves quite diverse, in part as a result tissue-specific differences in expression levels of the ERs and ER-interacting proteins (e.g., coactivators and corepressors). The existence of these tissue-specific differences in estrogen action has given rise to the concept of “tissue-specific” selective ER modulators. Selective ER modulators are pharmacologic agents that act as ER agonists in some tissues but ER antagonists in others.

Although nongenomic effects of estrogen were first described almost three decades ago (34), and related effects have been described for all major steroid hormones (35), the molecular mechanisms that mediate rapid effects of steroid hormones are not yet well understood. Identification of striatin as a molecule specific to and required for nongenomic activation of eNOS by estrogen provides a new opportunity to dissect nongenomic steroid hormone receptor signaling pathways from ER-signaling pathways that regulate transcription. These data also potentially provide an opportunity to design pathway-specific selective ER modulators capable of differentially regulating nongenomic vs. genomic effects that may prove useful ultimately as specific therapies in breast cancer, osteoporosis, and cardiovascular disease.