Rapid nongenomic actions of thyroid hormoneYukio Hiroi*, Hyung-Hwan Kim*, Hao Ying†, Fumihiko Furuya†, Zhihong Huang‡, Tommaso Simoncini§,Kensuke Noma*, Kojiro Ueki¶, Ngoc-Ha Nguyen�, Thomas S. Scanlan�, 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 GeneralHospital 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 �Departments of PharmaceuticalChemistry 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 tran-scriptional effects of TR mediated by the thyroid response element(TRE) cannot explain many actions of thyroid hormone. We pos-tulate that TR can initiate rapid, non-TRE-mediated effects in thecardiovascular system through cross-coupling to the phosphatidyl-inositol 3-kinase (PI3-kinase)�protein kinase Akt pathway. In vas-cular endothelial cells, the predominant TR isoform is TR�1. Treat-ment of endothelial cells with L-3,5,3�-triiodothyronine (T3)increased the association of TR�1 with the p85� subunit of PI3-kinase, leading to the phosphorylation and activation of Akt andendothelial nitric oxide synthase (eNOS). The activation of Akt andeNOS by T3 was abolished by the PI3-kinase inhibitors, LY294002and wortmannin, but not by the transcriptional inhibitor, actino-mycin D. To determine the physiological relevance of this PI3-kinase�Akt pathway, we administered T3 to mice undergoingtransient focal cerebral ischemia. Compared with vehicle, a singlebolus infusion of T3 rapidly increased Akt activity in the brain,decreased mean blood pressure, reduced cerebral infarct volume,and improved neurological deficit score. These neuroprotectiveeffects of T3 were greatly attenuated or absent in eNOS�/� andTR�1

�/���/� mice and were completely abolished in WT micepretreated with LY294002 or a T3 antagonist, NH-3. These findingsindicate that the activation of PI3-kinase�Akt pathways can medi-ate some of the rapid, non-TRE effects of TR and suggest that theactivation of Akt and eNOS contributes to some of the acutevasodilatory and neuroprotective effects of thyroid hormone.

nitric oxide � phosphatidylinositol 3-kinase � protein kinase Akt � stroke

Thyroid hormone exerts many physiological effects. It in-creases 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 theseeffects lead to a dramatic increase in cardiac output, which is aprominent feature of hyperthyroidism. In contrast, elevatedSVR is observed in thyroid hormone deficiency or hypothyroid-ism and is rapidly reversed with thyroid hormone replacement.However, the precise mechanism by which thyroid hormoneregulates vascular tone and SVR is not known.

The actions of thyroid hormone occur through its binding tothe thyroid hormone receptor (TR) (2). TR is a nuclear hormonereceptor, which heterodimerizes with retinoid X receptor, or insome cases, with itself. The dimers bind to the thyroid responseelements (TREs) in the absence of ligand and act as transcrip-tional repressors. An active form of thyroid hormone, L-3,5,3�-triiodothyronine (T3), binds to TR with much greater affinitythan the more abundant L-3,5,3�5�-tetraiodothyronine (T4).Binding of T3 to TR derepresses TRE-dependent genes andinduces the expression of target genes such as �-myosin heavychain, sarcoplasmic reticulum Ca2�-ATPase, �1-adrenergic re-ceptors, guanine-nucleotide-regulatory proteins, Na��K�-ATPase, and voltage-gated potassium channels (Kv1.5, Kv4.2,

and Kv4.3) in heart (1). Through TRE, T3 can also down-regulate the expression of �-myosin heavy chain, phospholam-ban, adenylyl cyclase types V and VI, Na��Ca2� exchanger, andthe TR isoform TR�1 (1). In addition to these genomic orTRE-mediated effects of T3, non-nuclear or TRE-independentactions of T3 have recently been described. For example, T3

rapidly modulates membrane potential, cellular depolarization,and contractile activity by regulating ion flux across plasmamembrane ion channels (3–5). Furthermore, in mice possessinga mutant form of TR� that cannot bind to TRE, thyroidhormone, which is known to regulate outer hair cell developmentin the ear via TR�, is still able to induce the development of thesehair cells (6). These findings suggest that TR may have actionsbeyond TRE-mediated gene transcription and that non-TRE-dependent effects of TR may contribute to important physio-logical effects of thyroid hormone.

The phosphatidylinositol 3-kinase (PI3-kinase)�protein ki-nase Akt pathway is an important regulator of cellular growth,metabolism, and survival (7, 8). For example, Akt is known toblock apoptosis via the serine-threonine phosphorylation ofmultiple targets, including phosphorylation and inhibition ofglycogen synthase kinase (GSK)-3, inactivation of the BCL-2family member BAD, and inhibition of cell death pathwayenzyme caspase-9 (8–10). Another important downstream tar-get of Akt is endothelial nitric oxide synthase (eNOS), which isphosphorylated and activated by Akt (11, 12). Mice with targeteddeletion of eNOS have enlarged cerebral and myocardial infarctsize after transient ischemia (13, 14). Therefore, it is likely thatthe regulation of eNOS activity by Akt in endothelial cells is animportant mediator of vascular function.

Recently, members of the steroid hormone receptor super-family, such as the estrogen, vitamin D, and glucocorticoidreceptors, have been shown to cross-couple to the PI3-kinase�Akt pathway (15–18). In vascular endothelial cells, the activationof the PI3-kinase�Akt pathway by steroid hormones leads toeNOS activation and cardiovascular protection (16, 18). Indeed,thyroid hormone has recently been shown to modulate theinteraction 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�analytic tools; Y.H., H.-H.K., T.S., M.A.M., and J.K.L. analyzeddata; 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, endothelialnitric oxide synthase; T3, L-3,5,3�-triiodothyronine; CBF, cerebral blood flow; NDS, neuro-logical deficit score; GSK, glycogen synthase kinase; SH2, Src homology 2; BP, bloodpressure; MCA, middle cerebral artery; MCAO, MCA occlusion.

**To whom correspondence should be addressed. E-mail: jliao@rics.bwh.harvard.edu.

© 2006 by The National Academy of Sciences of the USA

14104–14109 � PNAS � September 19, 2006 � vol. 103 � no. 38 www.pnas.org�cgi�doi�10.1073�pnas.0601600103

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0

effects of TR in the cardiovascular system remains to bedetermined.

The purpose of this study, therefore, is to show that PI3-kinase�Akt and eNOS can mediate some of the rapid, non-nuclear, cardiovascular effects of TR. The ability of TR toactivate PI3-kinase�Akt and eNOS may be physiologically im-portant and perhaps therapeutically beneficial because steroidhormone therapy may exert side effects, which may limit theiroverall clinical use.

ResultsTR Expression in Endothelial Cells. There are two major isoforms ofTR, TR�1 and TR�1. TR�1 is expressed in heart, brain, skeletalmuscle, and adipose tissue, whereas TR�1 is expressed at higherlevels in liver and kidney (22, 23). Because vascular tone is, inpart, regulated by endothelium-derived NO, we examined theexpression of TR isoforms in endothelial cells. By Northern blotanalysis, TR�1 and TR�2 mRNA were detected in endothelialcells from bovine aorta and human umbilical vein (Fig. 1A). Incontrast, little, if any, TR�1 mRNA was detected in endothelialcells. The robust expression of TR�1 mRNA in NIH 3T3fibroblasts, mouse embryonic fibroblasts (MEFs), and HeLacells served as positive controls. These findings indicate thatTR�1 is the predominant TR isoform in vascular endothelialcells. Similar findings were observed by Western blot analysiswhere 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 whetherT3 can activate Akt in endothelial cells, we serum-starvedendothelial cells for �8 h before T3 stimulation. Treatment withT3, at concentrations as low as 1 nM, increased Ser-473 phos-phorylation of Akt within 20 min (Fig. 2 A and B). Thephosphorylation of Akt by T3 was blocked by pretreatment withthe PI3-kinase inhibitors, LY294002 and wortmannin (Fig. 6A,which is published as supporting information on the PNAS website), but not by the transcriptional inhibitor, actinomycin D (Fig.2C), suggesting a nontranscriptional effect involving the PI3-kinase pathway. Interestingly, LY294002, but not wortmannin,decreased Akt phosphorylation below basal levels (Fig. 6). Theincrease in Akt phosphorylation by T3 corresponded to anincrease in Akt kinase activity as determined by the Akt down-stream phosphorylation target, GSK-3�. T3-induced GSK-3�phosphorylation was blocked by LY294002 (Fig. 2D). Becausehigher levels (i.e., nM) of T3 were required to activate Akt(compared with TRE-dependent responses, i.e., pM), we inves-

tigated whether TR mediated the effects of T3 on Akt kinaseactivity in brain tissues derived from TR�1

�/���/� mice, whichlack all known T3 binding receptors. Compared with WT mice,T3 did not stimulate Akt kinase activity in brain tissues fromTR�1

�/���/� mice (Fig. 2E). These findings indicate that T3

rapidly activates the PI3-kinase�Akt pathway through TR.

Interaction of TR�1 with PI3-Kinase. To determine the mechanismof PI3-kinase�Akt activation by TR, we investigated whether TRcan interact with the regulatory subunit of PI3-kinase, p85�, byusing GST-p85� pull-down assays with in vitro-translated, radio-labeled [35S]TR�1 and [35S]TR�1. In a ligand-dependent man-ner, treatment with T3 increased the association of TR�1, but notTR�1, with p85� (Fig. 3A). To determine whether endogenousp85� can interact with TR�1 in intact endothelial cells, weperformed coimmunoprecipitation studies using p85� and TR�1

antibodies. Lysates from endothelial cells treated with T3 wereimmunoprecipitated with p85� antibody followed by immuno-blotting for TR�1 or TR�1. Using this coimmmunoprecipitationassay, we found that T3 increased the association of TR�1, butnot TR�1 (data not shown), with p85� (Fig. 3B). Similarly,lysates treated with T3, which were then coimmunnoprecipitatedwith a TR antibody (sc-739), showed greater amounts of ligand-

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

Fig. 2. Phosphorylation and activation of Akt by T3. (A) Time-dependenteffects of T3 (10 nM) on Ser-473 phosphorylation of Akt (p-Akt). Results werestandardized to total Akt (t-Akt) and expressed as fold induction comparedwith baseline. *, P � 0.05; †, P � 0.01. (B) Concentration-dependent effects ofT3 (at 30 min) on Ser-473 phosphorylation of Akt. Results were standardizedto 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), oractinomycin 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. Stimu-lation with insulin growth factor (IGF, 50 ng�ml) served as a positive control.Results were standardized to total Akt (t-Akt) and expressed as fold inductioncompared with baseline. †, P � 0.01. (D) Effect of LY294002 (LY) or actinomycinD (ActD) on T3-induced Akt activity. The level of GSK-3� phosphorylation (foldinduction) was used to assess Akt activity. *, P � 0.05; †, P � 0.01. (E) Inductionof Akt activity by T3 (fold induction of GSK-3� phosphorylation compared withbaseline) in the brain tissues from WT and TR�1

�/���/� mice. Mice were givenT3 (500 ng, i.v. bolus) and brains were harvested at 30 min after administration.

*, P � 0.05.

Hiroi et al. PNAS � September 19, 2006 � vol. 103 � no. 38 � 14105

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0

dependent p85� (data not shown). These findings indicate thatTR�1 and p85� can associate in a ligand-dependent manner.

To determine whether the Src homology 2 (SH2) domains ofp85� 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 ofp85� in a GST pull-down assay with radiolabeled TR�1 andTR�1. Both SH2 domains of p85� were important for ligand-dependent interaction of p85� with TR�1 (Fig. 3C). The TRantagonist, NH-3 (24), blocked the ligand-dependent interactionof TR�1 with p85� (Fig. 3D).

Phosphorylation and Activation of eNOS by T3. Because eNOS is animportant downstream phosphorylation target of Akt in endo-thelial cells, we investigated whether T3 can lead to eNOSphosphorylation and activation. Treatment of endothelial cellswith T3, at a concentration as low as 0.1 nM, increased eNOSphosphorylation within 10–20 min (Fig. 4A). The phosphoryla-tion of eNOS was correlated with an increase in eNOS activityas measured by [3H]L-arginine to [3H]L-citrulline conversion andnitrite accumulation. In a time-dependent manner, T3 (10 nM)increased eNOS activity with maximal activity occurring 30–60min after stimulation (Fig. 4B). The time course and dosedependency of eNOS activation are slightly earlier and lower,respectively, compared with Akt activation. For example, al-though a minimum of 1 nM of T3 was required to observe anincrease in Akt activation, increase eNOS activation was ob-served at T3 concentrations as low as 0.1 nM (Fig. 4C). Thisresult may be caused by a greater sensitivity of and differencesin the assays used to detect eNOS versus Akt activation (i.e.,enzymatic assay versus antibody detection). Nevertheless, thetime and dose dependency of Akt and eNOS activation werewithin 10–20 min and 0.1–1 nM of each other. The activation ofboth Akt and eNOS by T3 was blocked by LY294002 andwortmannin, indicating that PI3-kinase mediates their activationby T3 (Fig. 4D).

Effects of T3 on BP. To determine the physiological relevance ofthis pathway in the cardiovascular system, we investigated theeffects of T3 on BP and cerebral blood flow (CBF) in mice.

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 mmHgat 30 min (P � 0.01) (Fig. 5A). The changes in mean BP weremore substantial when propylthiouracil-treated hypothyroid WTmice were used (88.0 to 78.9 mmHg at 30 min after T3 admin-istration, P � 0.05) (Fig. 5B). The acute decreases in mean BPwith T3 administration were greatly attenuated or abolished ineuthyroid or hypothyroid eNOS�/� mice. These findings indicatethat eNOS mediates most, if not all, of the rapid effects of T3 onvascular tone and BP.

Effects of T3 on CBF and Infarct Size. To determine the hemody-namic consequences of T3 on vascular tone, we measuredabsolute CBF by an indicator fractionation technique usingradiolabeled [14C]iodoamphetamine (25–27). Despite acute de-creases in mean BP, T3 rapidly increased absolute CBF in WTmice (134 � 13 to 190 � 27 ml�100 g per min, n � 9 and 8, P �0.05) (see Supporting Text, which is published as supportinginformation on the PNAS web site). In contrast, in eNOS�/�

mice, there was a small, but nonsignificant, increase in absoluteCBF with T3 (135 � 12 to 162 � 14 ml�100 g per min, n � 4 and5, P � 0.05). These findings suggest that eNOS is the primarymediator of a T3-induced increase in CBF, although anothermechanism could not be completely excluded.

To determine whether the increase in CBF by T3 correspondsto neuroprotection after focal ischemia, we subjected mice tointrafilament transient middle cerebral artery (MCA) occlusion(MCAO) (i.e., 2-h occlusion followed by 22-h reperfusion) asdescribed (27, 28). Cerebral infarct volumes were determined bysumming up the infarcted areas as determined by 2,3,5-triphenyltetrazolium chloride staining in 2-mm-thick coronal

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 associationof TR�1 with p85� by T3 (10 nM) in coimmunoprecipitation study of intactendothelial cells. Cell lysates were immunoprecipitated with p85� antibody orIgG, and then the immunoprecipitate was immunoblotted for TR�1 and p85�.There were little or no detectable levels of TR�1 in the p85� immunoprecipi-tate. (C) Effect of T3 (10 nM) on the interaction of N-SH2 and C-SH2 domainsof p85� with 35S-labeled TR in GST-p85� pull-down assay. The molecularmasses of TR�1 and TR�1 are indicated. (D) Concentration-dependent inhibi-tory effects of TR antagonist, NH-3, on T3-induced interaction of 35S-labeledTR�1 with p85� in GST-p85� pull-down assay.

Fig. 4. Phosphorylation and activation of eNOS by T3. (A) Concentration-dependent effects of T3 (at 30 min) on Ser-1179 phosphorylation of eNOS.Results were standardized to total eNOS (t-eNOS) and expressed as foldinduction compared with baseline. *, P � 0.05; †, P � 0.01 compared withbaseline control. (B) Time-dependent effects of T3 (10 nM) on eNOS activity asmeasured by [3H]arginine to [3H]citrulline conversion (�, pmol�mg) or nitriteaccumulation (■ , nmol per million cells). *, P � 0.05; †, P � 0.01 compared with0 time point. (C) Concentration-dependent effects of T3 (at 30 min) on eNOSactivity (�, [3H]citrulline, pmol�mg; ■ , nitrite accumulation, nmol per millioncells). *, P � 0.05; †, P � 0.01 compared with baseline control. (D) Effect ofLY294002 (10 �M) on T3-induced eNOS activity (�, [3H]citrulline, pmol�mg; ■ ,nitrite accumulation, nmol per million cells). †, P � 0.01 compared withbaseline control.

14106 � www.pnas.org�cgi�doi�10.1073�pnas.0601600103 Hiroi et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0

sections of the brain. Administration of T3 (30 min beforeocclusion) decreased cerebral infarct volume by 25% comparedwith 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 assupporting information on the PNAS web site).

To determine whether reductions in cerebral infarct sizecorrelated with improvement in neurological motor function, weassessed the neurological deficit score (NDS) in each mouseafter MCAO. NDS was scored by two observers blinded totreatment protocol as follows: 0, no motor deficits (normal); 1,f lexion of the contralateral torso and forelimb on lifting theanimal by the tail (mild); 2, circling to the contralateral side butnormal posture at rest (moderate); 3, leaning to the contralateralside at rest (severe); and 4, no spontaneous movement (critical).

The decrease in cerebral infarct volume by T3 correlated withqualitative improvement in NDS (P � 0.05 by Mann–Whitneyanalysis of noncontinuous variables) (Fig. 5E).

Treatment with T3, however, had little or no effect on cerebralinfarct volume and NDS in eNOS�/� mice (P � 0.05 for bothcompared with vehicle-treated mice) (Fig. 5 D and F), suggestingthat the neurological benefits were caused primarily by eNOS.Furthermore, the neuroprotective effects of T3 were not ob-served 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 TRantagonist, NH-3 (33.5 � 4.4% vs. 32.1 � 4.6% of infractedhemisphere, P � 0.05 compared with vehicle-treated mice).Indeed, administration of T3 had no beneficial effects on cere-bral infarct volume in TR�1

�/���/� mice (31.3 � 2.4% vs. 28.6 �4.1% of infracted hemisphere, P � 0.05 compared with vehicle-treated mice) (Fig. 5D). These findings indicate the critical rolesof TR, PI3-kinase�Akt, and eNOS in mediating the acuteneurovascular protective effects of T3.

DiscussionWe have shown that thyroid hormone can nontranscriptionallyactivate the PI3-kinase�Akt pathway. In a ligand-dependentmanner, TR�1 was shown to interact with PI3-kinase in intactendothelial cells by coimmunoprecipitation assay and in vitro byGST pull-down assay. The rapid activation of the PI3-kinase�Aktpathway by T3 led to an increase in eNOS activity, decrease inmean BP, augmentation of CBF, and reduction in cerebralinfarct size. These results indicate that some of the hemodynamiceffects of thyroid hormone are attributable to eNOS activationand suggest that the activation of Akt and eNOS by T3 may betherapeutically beneficial in cardiovascular disease. Indeed, theactivation of PI3-kinase�Akt by TR is similar to other nuclearhormone receptors such as estrogen receptor and glucocorticoidreceptor, which also activate the PI3-kinase�Akt pathway andmediate cardiovascular protection (15, 16, 18). However, admin-istration of steroid hormones is often associated with significantuntoward side effects such as increased risks of breast anduterine cancers and the development of hypertension and os-teoporosis. These side effects of steroid hormones have pre-cluded their use in cardiovascular diseases. Thus, the acuteadministration of thyroid hormone, compared with steroidhormones, may perhaps be a safer alternative for cardiovascularprotection.

The obligatory role of TR in T3-induced eNOS activation andstroke protection is demonstrated by studies showing lack ofstroke protection in the presence of TR antagonist, NH-3, andin TR�1

�/���/� or eNOS�/� mice. Although two differentgenes, THRA and THRB, encode the TR isoforms, TR� andTR�, respectively (2), we have shown that the predominant TRisoform in vascular endothelial cells is TR�1. TR�1 is a func-tional receptor for T3 and is highly expressed in heart, brain,skeletal muscle, and brown fat (22). In contrast to TR�1, TR�2does not bind thyroid hormone and act as a weak antagonist invitro. TR�1 is highly expressed in heart, brain, liver, and kidney(29), but we found that its expression is barely detectable invascular endothelial cells. The alternatively spliced THRB geneencoding TR�2 is almost exclusively expressed in the anteriorpituitary and hypothalamus. Thus, in vascular endothelial cells,TR�1 is the most likely TR isoform that mediates PI3-kinase�Akt�eNOS activation. Indeed, TR�1, but not TR�1, interactswith PI3-kinase in a ligand-dependent manner.

Because most of T3 is bound to carrier proteins such asthyroxine-binding globulin, albumin, and thyroid-binding preal-bumin in vivo, only 0.3% of T3 is unbound and free to interactwith TR (29). Although the concentrations of T3 that lead toTRE-dependent responses normally occur in the picomolarrange, we found that the minimum concentrations of T3, which

Fig. 5. Acute hemodynamic and neuroprotective effects of T3. (A) Time-dependent effects of T3 (500 ng, i.v. bolus) on mean BP (mmHg) in euthyroidWT and eNOS�/� mice. BP changes were examined by the paired Student’s ttest. *, P � 0.05; **, P � 0.01 compared with WT mice. (B) Time-dependenteffects of T3 (500 ng, i.v. bolus) on mean BP (mmHg) in propylthiouracil-treated (hypothyroid) WT and eNOS�/� mice. BP changes were examined bythe paired Student’s t test. *, P � 0.05 compared with WT mice. (C) Effect ofT3 (500 ng, i.v. bolus) on cerebral infarct volume (mm3) after MCAO. *, P � 0.05compared with vehicle-treated mice. (D) Effects of T3 (500 ng, i.v. bolus) oncerebral infarct volume (percentage of infarcted hemisphere) in WT, eNOS�/�,and TR�1

�/���/� mice. The percentage of infarcted hemisphere was calculatedby the formula: (contralateral hemisphere-ipsilateral nonischemic hemi-sphere)�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 pre-sented 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 aspercentage of total mice in each category.

Hiroi et al. PNAS � September 19, 2006 � vol. 103 � no. 38 � 14107

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0

activate Akt and eNOS, are somewhat higher, within the disso-ciation constant for TR (i.e., 0.1–1 nM) (22). Indeed, in euthy-roid human volunteers, injection of 100 �g of T3 increases freeT3 in the serum to levels �62 pM and is required to rapidlydecrease BP and systemic vascular resistance (30). These find-ings indicate that there are pharmacological effects of thyroidhormone on the cardiovascular system, which occur at concen-trations well above what is required to initiate TRE-dependentresponses. It remains to be determined, however, why higherconcentrations of T3 are required for Akt and eNOS activationcompared with that of TRE-dependent responses.

An interesting, and perhaps clinically important, finding ofthis study is that eNOS may contribute to some of the rapid,vasodilatory effects of thyroid hormone (i.e., decrease in BP andincrease in CBF). For example, T3 has been shown to inducerapid relaxation of preconstricted resistance arterioles fromisolated skeletal muscle (31). Treatment with T3 decreased BPin euthyroid WT mice, and to a greater extent, in hypothyroidWT mice. The decrease in BP by T3 was greatly attenuated orabsent in euthyroid and hypothyroid eNOS�/� mice, indicatingthe important contribution of eNOS in mediating the acutehemodyamic response to thyroid hormone. Interestingly, afterMCAO, BP was substantially higher in eNOS�/� mice comparedwith 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. Takentogether, these findings suggest that NO-mediated vasodilationis an important contributor to thyroid hormone’s physiologicaleffects on vascular tone and BP.

Although a reduction in systemic BP usually leads to greaterseverity in stroke, we found that T3 paradoxically decreasescerebral infarct volume and improved NDS. This result, in part,is probably caused by the ability of T3 to rapidly dilate cerebralblood vessels, leading to increases in CBF. Higher CBF has beenshown to closely correlate with neuronal protection after focalcerebral ischemia (16). We found that CBF was considerablyhigher in T3-treated WT mice compared with that in T3-treatedeNOS�/� mice or vehicle-treated WT mice. Thus, it is likely thatthe increase in CBF and stroke protection by T3 is predominantlymediated by eNOS, because these changes were relativelysmaller or absent in eNOS�/� mice. However, in our study, wecannot exclude the possibility that T3 could have additionalneuroprotective effects via mechanisms beyond eNOS. Forexample, the activation of Akt in neuronal and inflammatorycells by T3 may also contribute to the overall neuroprotectiveeffects of thyroid hormone.

In summary, we have shown that T3 rapidly activates eNOS viathe TR�PI3-kinase�Akt pathway in vitro and in vivo. Treatmentwith T3 increases the association of TR�1 with PI3-kinase,leading to decreased stroke size and improved NDS after focalcerebral ischemia. These findings suggest an important non-TRE-dependent effect of thyroid hormone. Further clinicalstudies, however, are required to determine whether acutethyroid hormone, either alone or as adjunctive therapy, could bebeneficial in patients with ischemic strokes.

Materials and MethodsMaterials. T3, LY294002, and actinomycin D were purchasedfrom EMD Biosciences (San Diego, CA). NH-3 was synthesizedand dissolved in DMSO as described (32). Anti-TR antibody(sc-739 and sc-737) was purchased from Santa Cruz Biotech-nology (Santa Cruz, CA), and anti-p85� antibody was fromUpstate Biotechnology (Lake Placid, NY). Unless specified, allother antibodies were obtained from Cell Signaling Technology(Danvers, MA).

Cell Culture. Human umbilical vein endothelial cells and bovineaortic endothelial cells were isolated and cultured as described

(16, 18). COS7, NIH 3T3, and HeLa cells were purchased fromAmerican Type Culture Collection (Manassas, VA). MEF werea gift from M. Kasuga (Kobe University, Kobe, Japan). Cellswere stimulated under serum-starved conditions consisting ofphenol-red-free Medium 199 (Gibco�BRL, St. Louis, MO) orDMEM (Gibco�BRL) with 0.4% charcoal-stripped FCS (Hy-Clone, Logan, UT).

Northern Blotting Analysis. Total RNA was extracted by usingRNAzolB (Tel-Test, Friendswood, TX). Twenty micrograms oftotal RNA was separated by electrophoresis on 1% agarose geland transferred onto Hybond N membrane (Amersham Phar-macia Biotech, Piscataway, NJ). The membrane was hybridizedwith 32P-labeled PstI fragment of mouse TR�1 cDNA in Per-fectHyb 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-coldPBS and incubated with 500 �l of lysis buffer (1% Triton�20mM Tris, pH 7.4�150 mM NaCl�1 mM EDTA�1 mM EGTA�2.5 mM sodium pyrophosphate�1 mM �-glycerolphosphate�1mM PMSF�1 mM Na3VO4). The cell lysates were centrifuged,and the supernatant were recovered. Forty microgram ofproteins was separated by SDS�PAGE, blotted onto nitrocel-lulose membranes (Osmonics, Trevose, PA), and probed withthe indicated antibody. Detection of protein bands was per-formed by using ECL (Pierce, Rockford, IL). Band intensitieswere analyzed by using National Institutes of Health Image.

Immunoprecipitation. Immunoprecipitation was performed byusing 800 �g of cell lysates and 1 �g of anti-p85� or anti-TRantibody at 4°C overnight. Protein G Sepharose (GE Healthcare,Buckinghamshire, U.K.) was added, and the mixture was incu-bated for 2 h and washed three times with lysis buffer.

Akt Kinase Assay. Cells or tissues were washed twice with ice-coldPBS and incubated with lysis buffer. Approximately 400–600 �gof protein was used for the Akt kinase activity. The assay kitdetected a downstream phosphorylation target of Akt, GSK-3�(Cell Signaling Technology).

eNOS Activity Assay. eNOS activity was determined by measuringnitrite accumulation or the conversion [3H]L-arginine to [3H]L-citrulline in the presence or absence of the competitive NOSinhibitor, L-NAME (1 mM), as described (Calbiochem-Novabiochem). Cells were homogenized in ice-cold PBS con-taining 1 mM EDTA. The homogenates were centrifuged, and5 �g of protein extracts from the supernatant was used for theeNOS assay as described. Unlabeled L-arginine was added to[3H]L-arginine (specific activity, 60 Ci�mmol) at a ratio of 3:1.

GST Pull-Down Assay. Mouse TR�1 and rat TR�1 cDNAs weresubcloned into pSPUTK vector (Stratagene, La Jolla, CA).35S-methionine-labeled TR�1 and TR�1 proteins were synthe-sized by using the TNT SP6 Quick Coupled Transcription�Translation System (Promega). GST-p85�, GST-N-SH2, andGST-C-SH2 proteins were purified with Glutathione Sepharose4B beads (GE Healthcare) and incubated with 35S-labeled TR�1or TR�1 protein in PBS with 0.2% Tween 20 at 4°C for 2 h. Beadswere washed three times, and proteins were separated by 10%SDS�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.

Transient MCAO Model. All experiments were conducted in ac-cordance with institutional guidelines on animal experimenta-tion from the National Institutes of Health, Brigham and Wom-

14108 � www.pnas.org�cgi�doi�10.1073�pnas.0601600103 Hiroi et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0

en’s Hospital, and Massachusetts General Hospital. WT andeNOS�/� mice (both on C57BL�6 backgrounds) were purchasedfrom Jackson Laboratory (Bar Harbor, ME). Eight-week-oldmice were made hypothyroid by daily i.p. injection of propyl-thiouracil (250 �g) for 3 weeks (33). TR�1

�/� mice (34) andTR��/� mice (35) were used to generate TR�1

�/���/� mice.Transient intraluminal occlusion of the MCA in mice wasperformed as described (see Supporting Text). One hundredmicroliters of PBS with or without 500 ng of T3 was administeredin an i.v. bolus 30 min before MCAO. LY294002 was adminis-tered 30 min before T3 injection. Two hundred molar excess ofNH-3 against T3 was given for 7 days and injected 30 min beforeT3 injection.

Absolute CBF Measurement. Absolute CBF was quantified with anindicator fractionation technique as described (see SupportingText) (25–27). CBF was calculated according to the method of

Van Uitert and Levy (36) and Betz and Iannotti (25). CBF(ml�100 g per min) � [brain count (cpm) � 0.3 (ml�min)�bloodcount (cpm) � brain weight (g)] � 100.

Statistical Analysis. All values are expressed as mean � SE. BPchanges were examined by the paired Student’s t test. Differ-ences of cerebral infarction volumes between groups weredetermined by the one-way ANOVA test. The difference inNDS, a noncontinuous variable, was determined by Mann–Whitney analysis. Values of P � 0.05 were considered statisti-cally significant.

We thank Richard T. Lee and John Gannon (Brigham and Women’sHospital) for technical support. This study was supported by NationalInstitutes of Health Grants HL070274, HL080187, NS010828, andDK062729. Y.H. was a recipient of a Research Fellowship from theNortheast Affiliate of the American Heart Association and an UeharaMemorial Fellowship.

1. Klein I, Ojamaa K (2001) N Engl J Med 344:501–509.2. Harvey CB, Williams GR (2002) Thyroid 12:441–446.3. Sakaguchi Y, Cui G, Sen L (1996) Endocrinology 137:4744–4751.4. Sun ZQ, Ojamaa K, Coetzee WA, Artman M, Klein I (2000) Am J Physiol

278:E302–E307.5. Sen L, Sakaguchi Y, Cui G (2002) Am J Physiol 283:H2119–H2129.6. Shibusawa N, Hashimoto K, Nikrodhanond AA, Liberman MC, Applebury

ML, Liao XH, Robbins JT, Refetoff S, Cohen RN, Wondisford FE (2003) J ClinInvest 112:588–597.

7. Coffer PJ, Jin J, Woodgett JR (1998) Biochem J 335:1–13.8. Franke TF, Kaplan DR, Cantley LC (1997) Cell 88:435–437.9. Burgering BM, Coffer PJ (1995) Nature 376:599–602.

10. Cardone MH, Roy N, Stennicke HR, Salvesen GS, Franke TF, Stanbridge E,Frisch S, Reed JC (1998) Science 282:1318–1321.

11. Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R, Zeiher AM (1999)Nature 399:601–605.

12. Fulton D, Gratton JP, McCabe TJ, Fontana J, Fujio Y, Walsh K, Franke TF,Papapetropoulos A, Sessa WC (1999) Nature 399:597–601.

13. Huang Z, Huang PL, Ma J, Meng W, Ayata C, Fishman MC, Moskowitz MA(1996) J Cereb Blood Flow Metab 16:981–987.

14. Jones SP, Girod WG, Palazzo AJ, Granger DN, Grisham MB, Jourd’Heuil D,Huang PL, Lefer DJ (1999) Am J Physiol 276:H1567–H1573.

15. Hafezi-Moghadam A, Simoncini T, Yang E, Limbourg FP, Plumier JC,Rebsamen MC, Hsieh CM, Chui DS, Thomas KL, Prorock AJ, et al. (2002) NatMed 8:473–479.

16. 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.

17. Rebsamen MC, Sun J, Norman AW, Liao JK (2002) Circ Res 91:17–24.18. 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 Endocrinol19:102–112.

20. Storey NM, Gentile S, Ullah H, Russo A, Muessel M, Erxleben C, ArmstrongDL (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.

Hiroi et al. PNAS � September 19, 2006 � vol. 103 � no. 38 � 14109

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 10

, 202

0