Rapid nongenomic actions of thyroid hormone Yukio Hiroi*, Hyung-Hwan Kim*, Hao Ying†, Fumihiko Furuya†, Zhihong Huang‡, Tommaso Simoncini§, Kensuke Noma*, Kojiro Ueki¶, Ngoc-Ha Nguyen储, Thomas S. Scanlan储, Michael A. Moskowitz‡, Sheue-Yann Cheng†, and James K. Liao*,** *Vascular Medicine Research, Brigham and Women’s Hospital and Harvard Medical School, Cambridge, MA 02139; †Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892; ‡Laboratory of Stroke and Neurovascular Regulation, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114; §Department of Reproductive Medicine and Child Development, University of Pisa, 56126 Pisa, Italy; ¶Department of Metabolic Diseases, Graduate School of Medicine, University of Tokyo, Tokyo 113-0033, Japan; and 储Departments of Pharmaceutical Chemistry and Cellular and Molecular Pharmacology, University of California, San Francisco, CA 94143 Edited by John D. Baxter, University of California, San Francisco, CA, and approved July 28, 2006 (received for review February 26, 2006)

The binding of thyroid hormone to the thyroid hormone receptor (TR) mediates important physiological effects. However, the transcriptional effects of TR mediated by the thyroid response element (TRE) cannot explain many actions of thyroid hormone. We postulate that TR can initiate rapid, non-TRE-mediated effects in the cardiovascular system through cross-coupling to the phosphatidylinositol 3-kinase (PI3-kinase)兾protein kinase Akt pathway. In vascular endothelial cells, the predominant TR isoform is TR␣1. Treatment of endothelial cells with L-3,5,3ⴕ-triiodothyronine (T3) increased the association of TR␣1 with the p85␣ subunit of PI3kinase, leading to the phosphorylation and activation of Akt and endothelial nitric oxide synthase (eNOS). The activation of Akt and eNOS by T3 was abolished by the PI3-kinase inhibitors, LY294002 and wortmannin, but not by the transcriptional inhibitor, actinomycin D. To determine the physiological relevance of this PI3kinase兾Akt pathway, we administered T3 to mice undergoing transient focal cerebral ischemia. Compared with vehicle, a single bolus infusion of T3 rapidly increased Akt activity in the brain, decreased mean blood pressure, reduced cerebral infarct volume, and improved neurological deficit score. These neuroprotective effects of T3 were greatly attenuated or absent in eNOSⴚ/ⴚ and TR␣1ⴚ/ⴚ␤ⴚ/ⴚ mice and were completely abolished in WT mice pretreated with LY294002 or a T3 antagonist, NH-3. These findings indicate that the activation of PI3-kinase兾Akt pathways can mediate some of the rapid, non-TRE effects of TR and suggest that the activation of Akt and eNOS contributes to some of the acute vasodilatory and neuroprotective effects of thyroid hormone. nitric oxide 兩 phosphatidylinositol 3-kinase 兩 protein kinase Akt 兩 stroke

T

hyroid hormone exerts many physiological effects. It increases tissue thermogenesis and metabolism, decreases systemic vascular resistance (SVR) and arterial blood pressure (BP), enhances renal sodium reabsorption and blood volume, and augments cardiac inotropy and chronotropy (1). All of these effects lead to a dramatic increase in cardiac output, which is a prominent feature of hyperthyroidism. In contrast, elevated SVR is observed in thyroid hormone deficiency or hypothyroidism and is rapidly reversed with thyroid hormone replacement. However, the precise mechanism by which thyroid hormone regulates vascular tone and SVR is not known. The actions of thyroid hormone occur through its binding to the thyroid hormone receptor (TR) (2). TR is a nuclear hormone receptor, which heterodimerizes with retinoid X receptor, or in some cases, with itself. The dimers bind to the thyroid response elements (TREs) in the absence of ligand and act as transcriptional repressors. An active form of thyroid hormone, L-3,5,3⬘triiodothyronine (T3), binds to TR with much greater affinity than the more abundant L-3,5,3⬘5⬘-tetraiodothyronine (T4). Binding of T3 to TR derepresses TRE-dependent genes and induces the expression of target genes such as ␣-myosin heavy chain, sarcoplasmic reticulum Ca2⫹-ATPase, ␤1-adrenergic receptors, guanine-nucleotide-regulatory proteins, Na⫹兾K⫹ATPase, and voltage-gated potassium channels (Kv1.5, Kv4.2, 14104 –14109 兩 PNAS 兩 September 19, 2006 兩 vol. 103 兩 no. 38

and Kv4.3) in heart (1). Through TRE, T3 can also downregulate the expression of ␤-myosin heavy chain, phospholamban, adenylyl cyclase types V and VI, Na⫹兾Ca2⫹ exchanger, and the TR isoform TR␣1 (1). In addition to these genomic or TRE-mediated effects of T3, non-nuclear or TRE-independent actions of T3 have recently been described. For example, T3 rapidly modulates membrane potential, cellular depolarization, and contractile activity by regulating ion flux across plasma membrane ion channels (3–5). Furthermore, in mice possessing a mutant form of TR␤ that cannot bind to TRE, thyroid hormone, which is known to regulate outer hair cell development in the ear via TR␤, is still able to induce the development of these hair cells (6). These findings suggest that TR may have actions beyond TRE-mediated gene transcription and that non-TREdependent effects of TR may contribute to important physiological effects of thyroid hormone. The phosphatidylinositol 3-kinase (PI3-kinase)兾protein kinase Akt pathway is an important regulator of cellular growth, metabolism, and survival (7, 8). For example, Akt is known to block apoptosis via the serine-threonine phosphorylation of multiple targets, including phosphorylation and inhibition of glycogen synthase kinase (GSK)-3, inactivation of the BCL-2 family member BAD, and inhibition of cell death pathway enzyme caspase-9 (8–10). Another important downstream target of Akt is endothelial nitric oxide synthase (eNOS), which is phosphorylated and activated by Akt (11, 12). Mice with targeted deletion of eNOS have enlarged cerebral and myocardial infarct size after transient ischemia (13, 14). Therefore, it is likely that the regulation of eNOS activity by Akt in endothelial cells is an important mediator of vascular function. Recently, members of the steroid hormone receptor superfamily, such as the estrogen, vitamin D, and glucocorticoid receptors, have been shown to cross-couple to the PI3-kinase兾 Akt pathway (15–18). In vascular endothelial cells, the activation of the PI3-kinase兾Akt pathway by steroid hormones leads to eNOS activation and cardiovascular protection (16, 18). Indeed, thyroid hormone has recently been shown to modulate the interaction of TR␤ or a mutant form of TR␤ with PI3-kinase (19–21). Whether this accounts for some of the rapid, non-TRE Author contributions: Y.H., H.-H.K., Z.H., T.S., and J.K.L. designed research; Y.H., H.-H.K., Z.H., T.S., K.N., and K.U. performed research; H.Y., F.F., N.-H.N., T.S.S., M.A.M., and S.-Y.C. contributed new reagents兾analytic tools; Y.H., H.-H.K., T.S., M.A.M., and J.K.L. analyzed data; and Y.H., S.-Y.C., and J.K.L. wrote the paper. The authors declare no conflict of interest. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: MEF, mouse embryonic fibroblasts; TR, thyroid hormone receptor; TRE, thyroid response element; PI3-kinase, phosphatidylinositol 3-kinase; eNOS, endothelial nitric oxide synthase; T3, L-3,5,3⬘-triiodothyronine; CBF, cerebral blood flow; NDS, neurological deficit score; GSK, glycogen synthase kinase; SH2, Src homology 2; BP, blood pressure; MCA, middle cerebral artery; MCAO, MCA occlusion. **To whom correspondence should be addressed. E-mail: [email protected]. © 2006 by The National Academy of Sciences of the USA

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601600103

effects of TR in the cardiovascular system remains to be determined. The purpose of this study, therefore, is to show that PI3kinase兾Akt and eNOS can mediate some of the rapid, nonnuclear, cardiovascular effects of TR. The ability of TR to activate PI3-kinase兾Akt and eNOS may be physiologically important and perhaps therapeutically beneficial because steroid hormone therapy may exert side effects, which may limit their overall clinical use. Results TR Expression in Endothelial Cells. There are two major isoforms of TR, TR␣1 and TR␤1. TR␣1 is expressed in heart, brain, skeletal muscle, and adipose tissue, whereas TR␤1 is expressed at higher levels in liver and kidney (22, 23). Because vascular tone is, in part, regulated by endothelium-derived NO, we examined the expression of TR isoforms in endothelial cells. By Northern blot analysis, TR␣1 and TR␣2 mRNA were detected in endothelial cells from bovine aorta and human umbilical vein (Fig. 1A). In contrast, little, if any, TR␤1 mRNA was detected in endothelial cells. The robust expression of TR␤1 mRNA in NIH 3T3 fibroblasts, mouse embryonic fibroblasts (MEFs), and HeLa cells served as positive controls. These findings indicate that TR␣1 is the predominant TR isoform in vascular endothelial cells. Similar findings were observed by Western blot analysis where TR␣1 protein, and to a much lesser extent, TR␤1 protein, is detected in vascular endothelial cells (Fig. 1B). Phosphorylation and Activation of Akt by T3. To determine whether T3 can activate Akt in endothelial cells, we serum-starved endothelial cells for ⬎8 h before T3 stimulation. Treatment with T3, at concentrations as low as 1 nM, increased Ser-473 phosphorylation of Akt within 20 min (Fig. 2 A and B). The phosphorylation of Akt by T3 was blocked by pretreatment with the PI3-kinase inhibitors, LY294002 and wortmannin (Fig. 6A, which is published as supporting information on the PNAS web site), but not by the transcriptional inhibitor, actinomycin D (Fig. 2C), suggesting a nontranscriptional effect involving the PI3kinase pathway. Interestingly, LY294002, but not wortmannin, decreased Akt phosphorylation below basal levels (Fig. 6). The increase in Akt phosphorylation by T3 corresponded to an increase in Akt kinase activity as determined by the Akt downstream phosphorylation target, GSK-3␤. T3-induced GSK-3␤ phosphorylation was blocked by LY294002 (Fig. 2D). Because higher levels (i.e., nM) of T3 were required to activate Akt (compared with TRE-dependent responses, i.e., pM), we invesHiroi et al.

Fig. 2. Phosphorylation and activation of Akt by T3. (A) Time-dependent effects of T3 (10 nM) on Ser-473 phosphorylation of Akt (p-Akt). Results were standardized to total Akt (t-Akt) and expressed as fold induction compared with baseline. *, P ⬍ 0.05; †, P ⬍ 0.01. (B) Concentration-dependent effects of T3 (at 30 min) on Ser-473 phosphorylation of Akt. Results were standardized to total Akt (t-Akt) and expressed as fold induction compared with baseline. *, P ⬍ 0.05. (C) Effects of PI3-kinase inhibitor, LY294002 (LY, 10 ␮M), or actinomycin D (ActD, 5 ␮M) on T3-induced Ser-473 phosphorylation of Akt. Cells were treated with LY or ActD for 30 min before T3 stimulation. Stimulation with insulin growth factor (IGF, 50 ng兾ml) served as a positive control. Results were standardized to total Akt (t-Akt) and expressed as fold induction compared with baseline. †, P ⬍ 0.01. (D) Effect of LY294002 (LY) or actinomycin D (ActD) on T3-induced Akt activity. The level of GSK-3␤ phosphorylation (fold induction) was used to assess Akt activity. *, P ⬍ 0.05; †, P ⬍ 0.01. (E) Induction of Akt activity by T3 (fold induction of GSK-3␤ phosphorylation compared with baseline) in the brain tissues from WT and TR␣1⫺/⫺␤⫺/⫺ mice. Mice were given T3 (500 ng, i.v. bolus) and brains were harvested at 30 min after administration. *, P ⬍ 0.05.

tigated whether TR mediated the effects of T3 on Akt kinase activity in brain tissues derived from TR␣1⫺/⫺␤⫺/⫺ mice, which lack all known T3 binding receptors. Compared with WT mice, T3 did not stimulate Akt kinase activity in brain tissues from TR␣1⫺/⫺␤⫺/⫺ mice (Fig. 2E). These findings indicate that T3 rapidly activates the PI3-kinase兾Akt pathway through TR. Interaction of TR␣1 with PI3-Kinase. To determine the mechanism

of PI3-kinase兾Akt activation by TR, we investigated whether TR can interact with the regulatory subunit of PI3-kinase, p85␣, by using GST-p85␣ pull-down assays with in vitro-translated, radiolabeled [35S]TR␣1 and [35S]TR␤1. In a ligand-dependent manner, treatment with T3 increased the association of TR␣1, but not TR␤1, with p85␣ (Fig. 3A). To determine whether endogenous p85␣ can interact with TR␣1 in intact endothelial cells, we performed coimmunoprecipitation studies using p85␣ and TR␣1 antibodies. Lysates from endothelial cells treated with T3 were immunoprecipitated with p85␣ antibody followed by immunoblotting for TR␣1 or TR␤1. Using this coimmmunoprecipitation assay, we found that T3 increased the association of TR␣1, but not TR␤1 (data not shown), with p85␣ (Fig. 3B). Similarly, lysates treated with T3, which were then coimmunnoprecipitated with a TR antibody (sc-739), showed greater amounts of ligandPNAS 兩 September 19, 2006 兩 vol. 103 兩 no. 38 兩 14105

MEDICAL SCIENCES

Fig. 1. Expression of TR in vascular endothelial cells. (A) Northern blotting analysis showing the expression of TR mRNA and protein in bovine aortic endothelial cells (BAEC), human umbilical vein endothelial cells (HUVEC), COS7, NIH 3T3 fibroblasts, MEF, and HeLa cells. (B) Western blotting analysis using an antibody that recognizes both TR␣1 and TR␤1 and with a TR␤1-specific antibody.

Fig. 3. Ligand-dependent interaction of TR with PI3-kinase. (A) Effect of T3 (10 nM) on 35S-labeled TR and p85␣ association in GST-p85␣ pull-down assay. The molecular masses of TR␣1 and TR␤1 are indicated. (B) Increased association of TR␣1 with p85␣ by T3 (10 nM) in coimmunoprecipitation study of intact endothelial cells. Cell lysates were immunoprecipitated with p85␣ antibody or IgG, and then the immunoprecipitate was immunoblotted for TR␣1 and p85␣. There were little or no detectable levels of TR␤1 in the p85␣ immunoprecipitate. (C) Effect of T3 (10 nM) on the interaction of N-SH2 and C-SH2 domains of p85␣ with 35S-labeled TR in GST-p85␣ pull-down assay. The molecular masses of TR␣1 and TR␤1 are indicated. (D) Concentration-dependent inhibitory effects of TR antagonist, NH-3, on T3-induced interaction of 35S-labeled TR␣1 with p85␣ in GST-p85␣ pull-down assay.

dependent p85␣ (data not shown). These findings indicate that TR␣1 and p85␣ can associate in a ligand-dependent manner. To determine whether the Src homology 2 (SH2) domains of p85␣ could be important for the association of p85␣ with TR␣1, we used deletional N-terminal (N-SH2, amino acids 332–428) and C-terminal (C-SH2, amino acids 624–718) SH2 constructs of p85␣ in a GST pull-down assay with radiolabeled TR␣1 and TR␤1. Both SH2 domains of p85␣ were important for liganddependent interaction of p85␣ with TR␣1 (Fig. 3C). The TR antagonist, NH-3 (24), blocked the ligand-dependent interaction of TR␣1 with p85␣ (Fig. 3D). Phosphorylation and Activation of eNOS by T3. Because eNOS is an

important downstream phosphorylation target of Akt in endothelial cells, we investigated whether T3 can lead to eNOS phosphorylation and activation. Treatment of endothelial cells with T3, at a concentration as low as 0.1 nM, increased eNOS phosphorylation within 10–20 min (Fig. 4A). The phosphorylation of eNOS was correlated with an increase in eNOS activity as measured by [3H]L-arginine to [3H]L-citrulline conversion and nitrite accumulation. In a time-dependent manner, T3 (10 nM) increased eNOS activity with maximal activity occurring 30–60 min after stimulation (Fig. 4B). The time course and dose dependency of eNOS activation are slightly earlier and lower, respectively, compared with Akt activation. For example, although a minimum of 1 nM of T3 was required to observe an increase in Akt activation, increase eNOS activation was observed at T3 concentrations as low as 0.1 nM (Fig. 4C). This result may be caused by a greater sensitivity of and differences in the assays used to detect eNOS versus Akt activation (i.e., enzymatic assay versus antibody detection). Nevertheless, the time and dose dependency of Akt and eNOS activation were within 10–20 min and 0.1–1 nM of each other. The activation of both Akt and eNOS by T3 was blocked by LY294002 and wortmannin, indicating that PI3-kinase mediates their activation by T3 (Fig. 4D). Effects of T3 on BP. To determine the physiological relevance of

this pathway in the cardiovascular system, we investigated the effects of T3 on BP and cerebral blood flow (CBF) in mice. 14106 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601600103

Fig. 4. Phosphorylation and activation of eNOS by T3. (A) Concentrationdependent effects of T3 (at 30 min) on Ser-1179 phosphorylation of eNOS. Results were standardized to total eNOS (t-eNOS) and expressed as fold induction compared with baseline. *, P ⬍ 0.05; †, P ⬍ 0.01 compared with baseline control. (B) Time-dependent effects of T3 (10 nM) on eNOS activity as measured by [3H]arginine to [3H]citrulline conversion (䊐, pmol兾mg) or nitrite accumulation (■, nmol per million cells). *, P ⬍ 0.05; †, P ⬍ 0.01 compared with 0 time point. (C) Concentration-dependent effects of T3 (at 30 min) on eNOS activity (䊐, [3H]citrulline, pmol兾mg; ■, nitrite accumulation, nmol per million cells). *, P ⬍ 0.05; †, P ⬍ 0.01 compared with baseline control. (D) Effect of LY294002 (10 ␮M) on T3-induced eNOS activity (䊐, [3H]citrulline, pmol兾mg; ■, nitrite accumulation, nmol per million cells). †, P ⬍ 0.01 compared with baseline control.

Treatment of euthyroid WT mice with T3 (500 ng, i.v. bolus) rapidly decreased mean BP within 5 min from 84.2 to 79.9 mmHg at 30 min (P ⬍ 0.01) (Fig. 5A). The changes in mean BP were more substantial when propylthiouracil-treated hypothyroid WT mice were used (88.0 to 78.9 mmHg at 30 min after T3 administration, P ⬍ 0.05) (Fig. 5B). The acute decreases in mean BP with T3 administration were greatly attenuated or abolished in euthyroid or hypothyroid eNOS⫺/⫺ mice. These findings indicate that eNOS mediates most, if not all, of the rapid effects of T3 on vascular tone and BP. Effects of T3 on CBF and Infarct Size. To determine the hemodynamic consequences of T3 on vascular tone, we measured absolute CBF by an indicator fractionation technique using radiolabeled [14C]iodoamphetamine (25–27). Despite acute decreases in mean BP, T3 rapidly increased absolute CBF in WT mice (134 ⫾ 13 to 190 ⫾ 27 ml兾100 g per min, n ⫽ 9 and 8, P ⬍ 0.05) (see Supporting Text, which is published as supporting information on the PNAS web site). In contrast, in eNOS⫺/⫺ mice, there was a small, but nonsignificant, increase in absolute CBF with T3 (135 ⫾ 12 to 162 ⫾ 14 ml兾100 g per min, n ⫽ 4 and 5, P ⬎ 0.05). These findings suggest that eNOS is the primary mediator of a T3-induced increase in CBF, although another mechanism could not be completely excluded. To determine whether the increase in CBF by T3 corresponds to neuroprotection after focal ischemia, we subjected mice to intrafilament transient middle cerebral artery (MCA) occlusion (MCAO) (i.e., 2-h occlusion followed by 22-h reperfusion) as described (27, 28). Cerebral infarct volumes were determined by summing up the infarcted areas as determined by 2,3,5triphenyltetrazolium chloride staining in 2-mm-thick coronal Hiroi et al.

Fig. 5. Acute hemodynamic and neuroprotective effects of T3. (A) Timedependent effects of T3 (500 ng, i.v. bolus) on mean BP (mmHg) in euthyroid WT and eNOS⫺/⫺ mice. BP changes were examined by the paired Student’s t test. *, P ⬍ 0.05; **, P ⬍ 0.01 compared with WT mice. (B) Time-dependent effects of T3 (500 ng, i.v. bolus) on mean BP (mmHg) in propylthiouraciltreated (hypothyroid) WT and eNOS⫺/⫺ mice. BP changes were examined by the paired Student’s t test. *, P ⬍ 0.05 compared with WT mice. (C) Effect of T3 (500 ng, i.v. bolus) on cerebral infarct volume (mm3) after MCAO. *, P ⬍ 0.05 compared with vehicle-treated mice. (D) Effects of T3 (500 ng, i.v. bolus) on cerebral infarct volume (percentage of infarcted hemisphere) in WT, eNOS⫺/⫺, and TR␣1⫺/⫺␤⫺/⫺ mice. The percentage of infarcted hemisphere was calculated by the formula: (contralateral hemisphere-ipsilateral nonischemic hemisphere)兾contralateral hemiphere ⫻ 100%. *, P ⬍ 0.05 compared with vehicle. (E) Effect of vehicle (Veh) or T3 (500 ng, i.v. bolus) on NDS in WT mice. Noncontinuous data were examined by Mann–Whitney analysis and presented as percentage of total mice in each category. (F) Effect of vehicle (Veh) or T3 (500 ng, i.v. bolus) on NDS in eNOS⫺/⫺ mice. Data are presented as percentage of total mice in each category.

sections of the brain. Administration of T3 (30 min before occlusion) decreased cerebral infarct volume by 25% compared with vehicle-treated WT mice (85 ⫾ 4 mm3 vs. 114 ⫾ 7 mm3, n ⫽ 12 and 13, P ⬍ 0.05) (Fig. 5C and Fig. 7, which is published as supporting information on the PNAS web site). To determine whether reductions in cerebral infarct size correlated with improvement in neurological motor function, we assessed the neurological deficit score (NDS) in each mouse after MCAO. NDS was scored by two observers blinded to treatment protocol as follows: 0, no motor deficits (normal); 1, flexion of the contralateral torso and forelimb on lifting the animal by the tail (mild); 2, circling to the contralateral side but normal posture at rest (moderate); 3, leaning to the contralateral side at rest (severe); and 4, no spontaneous movement (critical). Hiroi et al.

Discussion We have shown that thyroid hormone can nontranscriptionally activate the PI3-kinase兾Akt pathway. In a ligand-dependent manner, TR␣1 was shown to interact with PI3-kinase in intact endothelial cells by coimmunoprecipitation assay and in vitro by GST pull-down assay. The rapid activation of the PI3-kinase兾Akt pathway by T3 led to an increase in eNOS activity, decrease in mean BP, augmentation of CBF, and reduction in cerebral infarct size. These results indicate that some of the hemodynamic effects of thyroid hormone are attributable to eNOS activation and suggest that the activation of Akt and eNOS by T3 may be therapeutically beneficial in cardiovascular disease. Indeed, the activation of PI3-kinase兾Akt by TR is similar to other nuclear hormone receptors such as estrogen receptor and glucocorticoid receptor, which also activate the PI3-kinase兾Akt pathway and mediate cardiovascular protection (15, 16, 18). However, administration of steroid hormones is often associated with significant untoward side effects such as increased risks of breast and uterine cancers and the development of hypertension and osteoporosis. These side effects of steroid hormones have precluded their use in cardiovascular diseases. Thus, the acute administration of thyroid hormone, compared with steroid hormones, may perhaps be a safer alternative for cardiovascular protection. The obligatory role of TR in T3-induced eNOS activation and stroke protection is demonstrated by studies showing lack of stroke protection in the presence of TR antagonist, NH-3, and in TR␣1⫺/⫺␤⫺/⫺ or eNOS⫺/⫺ mice. Although two different genes, THRA and THRB, encode the TR isoforms, TR␣ and TR␤, respectively (2), we have shown that the predominant TR isoform in vascular endothelial cells is TR␣1. TR␣1 is a functional receptor for T3 and is highly expressed in heart, brain, skeletal muscle, and brown fat (22). In contrast to TR␣1, TR␣2 does not bind thyroid hormone and act as a weak antagonist in vitro. TR␤1 is highly expressed in heart, brain, liver, and kidney (29), but we found that its expression is barely detectable in vascular endothelial cells. The alternatively spliced THRB gene encoding TR␤2 is almost exclusively expressed in the anterior pituitary and hypothalamus. Thus, in vascular endothelial cells, TR␣1 is the most likely TR isoform that mediates PI3-kinase兾 Akt兾eNOS activation. Indeed, TR␣1, but not TR␤1, interacts with PI3-kinase in a ligand-dependent manner. Because most of T3 is bound to carrier proteins such as thyroxine-binding globulin, albumin, and thyroid-binding prealbumin in vivo, only 0.3% of T3 is unbound and free to interact with TR (29). Although the concentrations of T3 that lead to TRE-dependent responses normally occur in the picomolar range, we found that the minimum concentrations of T3, which PNAS 兩 September 19, 2006 兩 vol. 103 兩 no. 38 兩 14107

MEDICAL SCIENCES

The decrease in cerebral infarct volume by T3 correlated with qualitative improvement in NDS (P ⬍ 0.05 by Mann–Whitney analysis of noncontinuous variables) (Fig. 5E). Treatment with T3, however, had little or no effect on cerebral infarct volume and NDS in eNOS⫺/⫺ mice (P ⬎ 0.05 for both compared with vehicle-treated mice) (Fig. 5 D and F), suggesting that the neurological benefits were caused primarily by eNOS. Furthermore, the neuroprotective effects of T3 were not observed in WT mice pretreated with the PI3-kinase inhibitor, LY294002 (32.1 ⫾ 4.6% vs. 33.5 ⫾ 4.4% of infracted hemisphere, P ⬎ 0.05 compared with vehicle-treated mice) or with the TR antagonist, NH-3 (33.5 ⫾ 4.4% vs. 32.1 ⫾ 4.6% of infracted hemisphere, P ⬎ 0.05 compared with vehicle-treated mice). Indeed, administration of T3 had no beneficial effects on cerebral infarct volume in TR␣1⫺/⫺␤⫺/⫺ mice (31.3 ⫾ 2.4% vs. 28.6 ⫾ 4.1% of infracted hemisphere, P ⬎ 0.05 compared with vehicletreated mice) (Fig. 5D). These findings indicate the critical roles of TR, PI3-kinase兾Akt, and eNOS in mediating the acute neurovascular protective effects of T3.

activate Akt and eNOS, are somewhat higher, within the dissociation constant for TR (i.e., 0.1–1 nM) (22). Indeed, in euthyroid human volunteers, injection of 100 ␮g of T3 increases free T3 in the serum to levels ⬎62 pM and is required to rapidly decrease BP and systemic vascular resistance (30). These findings indicate that there are pharmacological effects of thyroid hormone on the cardiovascular system, which occur at concentrations well above what is required to initiate TRE-dependent responses. It remains to be determined, however, why higher concentrations of T3 are required for Akt and eNOS activation compared with that of TRE-dependent responses. An interesting, and perhaps clinically important, finding of this study is that eNOS may contribute to some of the rapid, vasodilatory effects of thyroid hormone (i.e., decrease in BP and increase in CBF). For example, T3 has been shown to induce rapid relaxation of preconstricted resistance arterioles from isolated skeletal muscle (31). Treatment with T3 decreased BP in euthyroid WT mice, and to a greater extent, in hypothyroid WT mice. The decrease in BP by T3 was greatly attenuated or absent in euthyroid and hypothyroid eNOS⫺/⫺ mice, indicating the important contribution of eNOS in mediating the acute hemodyamic response to thyroid hormone. Interestingly, after MCAO, BP was substantially higher in eNOS⫺/⫺ mice compared with that in WT mice (126 ⫾ 7 mmHg vs. 86 ⫾ 10 mmHg, P ⬍ 0.01), suggesting that the postvasodilatory response to ischemia– reperfusion injury also appears to be mediated by eNOS. Taken together, these findings suggest that NO-mediated vasodilation is an important contributor to thyroid hormone’s physiological effects on vascular tone and BP. Although a reduction in systemic BP usually leads to greater severity in stroke, we found that T3 paradoxically decreases cerebral infarct volume and improved NDS. This result, in part, is probably caused by the ability of T3 to rapidly dilate cerebral blood vessels, leading to increases in CBF. Higher CBF has been shown to closely correlate with neuronal protection after focal cerebral ischemia (16). We found that CBF was considerably higher in T3-treated WT mice compared with that in T3-treated eNOS⫺/⫺ mice or vehicle-treated WT mice. Thus, it is likely that the increase in CBF and stroke protection by T3 is predominantly mediated by eNOS, because these changes were relatively smaller or absent in eNOS⫺/⫺ mice. However, in our study, we cannot exclude the possibility that T3 could have additional neuroprotective effects via mechanisms beyond eNOS. For example, the activation of Akt in neuronal and inflammatory cells by T3 may also contribute to the overall neuroprotective effects of thyroid hormone. In summary, we have shown that T3 rapidly activates eNOS via the TR兾PI3-kinase兾Akt pathway in vitro and in vivo. Treatment with T3 increases the association of TR␣1 with PI3-kinase, leading to decreased stroke size and improved NDS after focal cerebral ischemia. These findings suggest an important nonTRE-dependent effect of thyroid hormone. Further clinical studies, however, are required to determine whether acute thyroid hormone, either alone or as adjunctive therapy, could be beneficial in patients with ischemic strokes. Materials and Methods Materials. T3, LY294002, and actinomycin D were purchased

from EMD Biosciences (San Diego, CA). NH-3 was synthesized and dissolved in DMSO as described (32). Anti-TR antibody (sc-739 and sc-737) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and anti-p85␣ antibody was from Upstate Biotechnology (Lake Placid, NY). Unless specified, all other antibodies were obtained from Cell Signaling Technology (Danvers, MA).

(16, 18). COS7, NIH 3T3, and HeLa cells were purchased from American Type Culture Collection (Manassas, VA). MEF were a gift from M. Kasuga (Kobe University, Kobe, Japan). Cells were stimulated under serum-starved conditions consisting of phenol-red-free Medium 199 (Gibco兾BRL, St. Louis, MO) or DMEM (Gibco兾BRL) with 0.4% charcoal-stripped FCS (HyClone, Logan, UT). Northern Blotting Analysis. Total RNA was extracted by using

RNAzolB (Tel-Test, Friendswood, TX). Twenty micrograms of total RNA was separated by electrophoresis on 1% agarose gel and transferred onto Hybond N membrane (Amersham Pharmacia Biotech, Piscataway, NJ). The membrane was hybridized with 32P-labeled PstI fragment of mouse TR␣1 cDNA in PerfectHyb Plus Hybridization Buffer (Sigma, St. Louis, MO) solution at 68°C. The membrane was washed with 2⫻ SSC, 0.1% SDS twice and 1⫻ SSC, 0.1% SDS twice at 42°C. Western Blotting Analysis. Cells were washed twice with ice-cold

PBS and incubated with 500 ␮l of lysis buffer (1% Triton兾20 mM Tris, pH 7.4兾150 mM NaCl兾1 mM EDTA兾1 mM EGTA兾 2.5 mM sodium pyrophosphate兾1 mM ␤-glycerolphosphate兾1 mM PMSF兾1 mM Na3VO4). The cell lysates were centrifuged, and the supernatant were recovered. Forty microgram of proteins was separated by SDS兾PAGE, blotted onto nitrocellulose membranes (Osmonics, Trevose, PA), and probed with the indicated antibody. Detection of protein bands was performed by using ECL (Pierce, Rockford, IL). Band intensities were analyzed by using National Institutes of Health Image. Immunoprecipitation. Immunoprecipitation was performed by

using 800 ␮g of cell lysates and 1 ␮g of anti-p85␣ or anti-TR antibody at 4°C overnight. Protein G Sepharose (GE Healthcare, Buckinghamshire, U.K.) was added, and the mixture was incubated for 2 h and washed three times with lysis buffer.

Akt Kinase Assay. Cells or tissues were washed twice with ice-cold PBS and incubated with lysis buffer. Approximately 400–600 ␮g of protein was used for the Akt kinase activity. The assay kit detected a downstream phosphorylation target of Akt, GSK-3␤ (Cell Signaling Technology). eNOS Activity Assay. eNOS activity was determined by measuring nitrite accumulation or the conversion [3H]L-arginine to [3H]Lcitrulline in the presence or absence of the competitive NOS inhibitor, L -NAME (1 mM), as described (CalbiochemNovabiochem). Cells were homogenized in ice-cold PBS containing 1 mM EDTA. The homogenates were centrifuged, and 5 ␮g of protein extracts from the supernatant was used for the eNOS assay as described. Unlabeled L-arginine was added to [3H]L-arginine (specific activity, 60 Ci兾mmol) at a ratio of 3:1. GST Pull-Down Assay. Mouse TR␣1 and rat TR␤1 cDNAs were

subcloned into pSPUTK vector (Stratagene, La Jolla, CA). TR␣1 and TR␤1 proteins were synthesized by using the TNT SP6 Quick Coupled Transcription兾 Translation System (Promega). GST-p85␣, GST-N-SH2, and GST-C-SH2 proteins were purified with Glutathione Sepharose 4B beads (GE Healthcare) and incubated with 35S-labeled TR␣1 or TR␤1 protein in PBS with 0.2% Tween 20 at 4°C for 2 h. Beads were washed three times, and proteins were separated by 10% SDS兾PAGE. The gels were fixed with acetic acid and methanol. Signals were enhanced by using Enlightning (PerkinElmer, Wellesley, MA), and the gels were subjected to autoradiography. 35S-methionine-labeled

Transient MCAO Model. All experiments were conducted in acCell Culture. Human umbilical vein endothelial cells and bovine

aortic endothelial cells were isolated and cultured as described 14108 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0601600103

cordance with institutional guidelines on animal experimentation from the National Institutes of Health, Brigham and WomHiroi et al.

Van Uitert and Levy (36) and Betz and Iannotti (25). CBF (ml兾100 g per min) ⫽ [brain count (cpm) ⫻ 0.3 (ml兾min)兾blood count (cpm) ⫻ brain weight (g)] ⫻ 100. Statistical Analysis. All values are expressed as mean ⫾ SE. BP

changes were examined by the paired Student’s t test. Differences of cerebral infarction volumes between groups were determined by the one-way ANOVA test. The difference in NDS, a noncontinuous variable, was determined by Mann– Whitney analysis. Values of P ⬍ 0.05 were considered statistically significant.

Absolute CBF Measurement. Absolute CBF was quantified with an indicator fractionation technique as described (see Supporting Text) (25–27). CBF was calculated according to the method of

We thank Richard T. Lee and John Gannon (Brigham and Women’s Hospital) for technical support. This study was supported by National Institutes of Health Grants HL070274, HL080187, NS010828, and DK062729. Y.H. was a recipient of a Research Fellowship from the Northeast Affiliate of the American Heart Association and an Uehara Memorial Fellowship.

Klein I, Ojamaa K (2001) N Engl J Med 344:501–509. Harvey CB, Williams GR (2002) Thyroid 12:441–446. Sakaguchi Y, Cui G, Sen L (1996) Endocrinology 137:4744–4751. Sun ZQ, Ojamaa K, Coetzee WA, Artman M, Klein I (2000) Am J Physiol 278:E302–E307. Sen L, Sakaguchi Y, Cui G (2002) Am J Physiol 283:H2119–H2129. Shibusawa N, Hashimoto K, Nikrodhanond AA, Liberman MC, Applebury ML, Liao XH, Robbins JT, Refetoff S, Cohen RN, Wondisford FE (2003) J Clin Invest 112:588–597. Coffer PJ, Jin J, Woodgett JR (1998) Biochem J 335:1–13. Franke TF, Kaplan DR, Cantley LC (1997) Cell 88:435–437. Burgering BM, Coffer PJ (1995) Nature 376:599–602. Cardone MH, Roy N, Stennicke HR, Salvesen GS, Franke TF, Stanbridge E, Frisch S, Reed JC (1998) Science 282:1318–1321. Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R, Zeiher AM (1999) Nature 399:601–605. Fulton D, Gratton JP, McCabe TJ, Fontana J, Fujio Y, Walsh K, Franke TF, Papapetropoulos A, Sessa WC (1999) Nature 399:597–601. Huang Z, Huang PL, Ma J, Meng W, Ayata C, Fishman MC, Moskowitz MA (1996) J Cereb Blood Flow Metab 16:981–987. Jones SP, Girod WG, Palazzo AJ, Granger DN, Grisham MB, Jourd’Heuil D, Huang PL, Lefer DJ (1999) Am J Physiol 276:H1567–H1573. Hafezi-Moghadam A, Simoncini T, Yang E, Limbourg FP, Plumier JC, Rebsamen MC, Hsieh CM, Chui DS, Thomas KL, Prorock AJ, et al. (2002) Nat Med 8:473–479. Limbourg FP, Huang Z, Plumier JC, Simoncini T, Fujioka M, Tuckermann J, Schutz G, Moskowitz MA, Liao JK (2002) J Clin Invest 110:1729–1738. Rebsamen MC, Sun J, Norman AW, Liao JK (2002) Circ Res 91:17–24. Simoncini T, Hafezi-Moghadam A, Brazil DP, Ley K, Chin WW, Liao JK (2000) Nature 407:538–541.

19. Cao X, Kambe F, Moeller LC, Refetoff S, Seo H (2005) Mol Endocrinol 19:102–112. 20. Storey NM, Gentile S, Ullah H, Russo A, Muessel M, Erxleben C, Armstrong DL (2006) Proc Natl Acad Sci USA 103:5197–5201. 21. Furuya F, Hanover JA, Cheng SY (2006) Proc Natl Acad Sci USA 103:1780–1785. 22. Yen PM (2001) Physiol Rev 81:1097–1142. 23. Cheng SY (2005) Trends Endocrinol Metab 16:176–182. 24. Lim W, Nguyen NH, Yang HY, Scanlan TS, Furlow JD (2002) J Biol Chem 277:35664–35670. 25. Betz AL, Iannotti F (1983) J Cereb Blood Flow Metab 3:193–199. 26. Fujii M, Hara H, Meng W, Vonsattel JP, Huang Z, Moskowitz MA (1997) Stroke 28:1805–1810, discussion 1811. 27. Aoki Y, Huang Z, Thomas SS, Bhide PG, Huang I, Moskowitz MA, Reeves SA (2000) FASEB J 14:1965–1973. 28. Huang Z, Huang PL, Panahian N, Dalkara T, Fishman MC, Moskowitz MA (1994) Science 265:1883–1885. 29. Williams GR (2000) Mol Cell Biol 20:8329–8342. 30. Schmidt BM, Martin N, Georgens AC, Tillmann HC, Feuring M, Christ M, Wehling M (2002) J Clin Endocrinol Metab 87:1681–1686. 31. Park KW, Dai HB, Ojamaa K, Lowenstein E, Klein I, Sellke FW (1997) Anesth Analg 85:734–738. 32. Nguyen NH, Apriletti JW, Cunha Lima ST, Webb P, Baxter JD, Scanlan TS (2002) J Med Chem 45:3310–3320. 33. Moreno M, Lombardi A, Beneduce L, Silvestri E, Pinna G, Goglia F, Lanni A (2002) Endocrinology 143:504–510. 34. Wikstrom L, Johansson C, Salto C, Barlow C, Campos Barros A, Baas F, Forrest D, Thoren P, Vennstrom B (1998) EMBO J 17:455–461. 35. Forrest D, Hanebuth E, Smeyne RJ, Everds N, Stewart CL, Wehner JM, Curran T (1996) EMBO J 15:3006–3015. 36. Van Uitert RL, Levy DE (1978) Stroke 9:67–72.

1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11. 12. 13. 14. 15.

16. 17. 18.

Hiroi et al.

PNAS 兩 September 19, 2006 兩 vol. 103 兩 no. 38 兩 14109

MEDICAL SCIENCES

en’s Hospital, and Massachusetts General Hospital. WT and eNOS⫺/⫺ mice (both on C57BL兾6 backgrounds) were purchased from Jackson Laboratory (Bar Harbor, ME). Eight-week-old mice were made hypothyroid by daily i.p. injection of propylthiouracil (250 ␮g) for 3 weeks (33). TR␣1⫺/⫺ mice (34) and TR␤⫺/⫺ mice (35) were used to generate TR␣1⫺/⫺␤⫺/⫺ mice. Transient intraluminal occlusion of the MCA in mice was performed as described (see Supporting Text). One hundred microliters of PBS with or without 500 ng of T3 was administered in an i.v. bolus 30 min before MCAO. LY294002 was administered 30 min before T3 injection. Two hundred molar excess of NH-3 against T3 was given for 7 days and injected 30 min before T3 injection.