Thyroxine Promotes Association of Mitogen-activated Protein Kinase and Nuclear Thyroid Hormone Receptor (TR) and Causes Serine Phosphorylation of TR*

Activated nongenomically by L -thyroxine (T 4 ), mito- gen-activated protein kinase (MAPK) complexed in 10–20 min with endogenous nuclear thyroid hormone receptor (TR b 1 or TR) in nuclear fractions of 293T cells, resulting in serine phosphorylation of TR. Treatment of cells with the MAPK kinase inhibitor, PD 98059, prevented both T 4 -induced nuclear MAPK-TR co-immuno- precipitation and serine phosphorylation of TR. T 4 treatment caused dissociation of TR and SMRT (silenc-ing mediator of retinoid and thyroid hormone receptor), an effect also inhibited by PD 98059 and presumptively a result of association of nuclear MAPK with TR. Transfection into CV-1 cells of TR gene constructs in which one or both zinc fingers in the TR DNA-binding domain were replaced with those from the glucocorticoid receptor localized the site of TR phosphorylation by T 4 -acti- vated MAPK to a serine in the second zinc finger of the TR DNA-binding domain. In an in vitro cell- and hor-mone-free system, purified activated MAPK phosphorylated recombinant human TR b 1 (102–461). Thus, T 4 acti- vates MAPK and causes MAPK-mediated serine phosphorylation of TR b 1 and dissociation an

We have demonstrated in cultured cells that L-thyroxine (T 4 ) 1 can nongenomically activate signal transduction proteins such as mitogen-activated protein kinase (MAPK) (1) and, through serine phosphorylation by MAPK, can enhance the activity of several nuclear transactivator proteins. Among the latter are the signal transducer and activator of transcription (STAT) proteins that mediate growth factor (2) and cytokine (1,3) signals. This T 4 effect is initiated by a G protein-coupled receptor in the plasma membrane (1) and has been observed in cells that contain endogenous nuclear thyroid hormone receptor (TR), such as 293T cells and human skin fibroblasts (BG-9), and in cells which are devoid of functional TR (CV-1 and HeLa (4)).
In contrast, genomic actions of thyroid hormone require binding of the hormone, predominantly 3,5,3Ј-triiodo-L-thyronine (T 3 ), by specific receptors in the cell nucleus. T 3 -liganded TR may bind as a monomer or a homo-or heterodimer with retinoid X receptor to thyroid hormone response elements in the regulatory upstream regions of specific hormone-responsive genes (5)(6)(7). In the absence of T 3 , TR exists in the transcriptionally inactive (repressed) state. This state is imposed by the binding to unliganded TR of the co-repressor proteins, SMRT (silencing mediator of retinoid and thyroid hormone receptors) and NCoR (nuclear co-repressor) (8). SMRT binding to TR has been localized to the hinge region of the receptor (amino acids 211-240) (9). Binding of T 3 to TR results in dissociation of co-repressor proteins from TR and the recruitment of activator proteins that facilitate enhanced transcriptional activity of the receptor (10).
Serine phosphorylation of TR isoforms has been described by several laboratories (11)(12)(13)(14)(15). In such studies phosphorylation has been inferred from stimulation of the activity of cellular cAMP-dependent protein kinase (PKA) (11,12), a serine/threonine kinase, or from serine kinase inhibition or phosphatase inhibition in treated cells (14,16). There have been several results of such phosphorylation in these model systems. For example, serine phosphorylation of TR␣1 has been shown to decrease TR monomer binding to DNA (11). TR␤1 is selectively stabilized against protease degradation by serine phosphorylation and transcriptional activity of the receptor is significantly increased (14,15). Leitman et al. (12) have also shown increased transcriptional activity of TR␤1 in response to phosphorylation. Comparing phosphorylatable and nonphosphorylatable forms of TR␣2, Katz et al. (17) concluded that serine phosphorylation of the receptor isoform decreased its ability to heterodimerize with retinoid X receptor at a thyroid hormone response element site. Using a serine/threonine kinase inhibitor, H7, Jones et al. (16) showed a reduction in T 3 -induced transcriptional activity of both TR␣1 and TR␤1. In some of these studies (12), the results were highly cell line-specific.
The mechanism by which serine phosphorylation of TR␤1 is achieved is unclear, aside from the possible involvement of PKA (11,12). In the study reported here, we describe a signal transduction mechanism by which T 4 , acting at the cell surface, promotes MAPK-dependent serine phosphorylation of TR␤1 and resultant dissociation of TR and SMRT. We also identify a region of TR␤1 that is required for MAPK binding to the receptor.

EXPERIMENTAL PROCEDURES
Materials-L-T 4 , L-T 3 , D-T 4 , D-T 3 , 3,3Ј,5Ј-triiodothyronine (rT 3 ), tetraiodothyroacetic acid (tetrac), 3,5,3Ј-triiodothyroacetic acid (triac), 8-bromo-cyclic-AMP (8-Br-cAMP), myelin basic protein (MBP), 6-n-propyl-2-thiouracil (PTU), T 4 -agarose, and protein A-agarose were obtained from Sigma. Stock solutions of thyroid hormone and analogues were prepared in 0.04 N KOH, 4% propylene glycol, and dilutions were made to final analogue concentrations as indicated. In all experiments in which T 4 was added to cultured cells, the total and free T 4 concentrations were 10 Ϫ7 and 10 Ϫ10 M, respectively, and total and free T 3 concentrations were below the limits of detection. PD 98059 was obtained from Calbiochem (La Jolla, CA), and geldanamycin was from the Drug Synthesis and Chemistry Branch of the National Cancer Institute (Bethesda, MD). Stock solutions of these inhibitors were prepared in 100% Me 2 SO, so that a final concentration of 0.1% Me 2 SO was achieved in cell cultures and had no effect on experimental results. KT5720 was obtained from Kamiya Biomedical Co. (Thousand Oaks, CA). Lipo-fectAMINE Plus was obtained from Life Technologies, Inc.
TR nucleotides, expressed in pcDNA1/Amp and including full-length hTR␤1, rTR-⌬N (containing DBDs and LBDs), and rTR-LBD (LBD only), were generously provided by Dr. Paul M. Yen (NIDDK, NIH, Bethesda, MD), as were a TGT hybrid construct of TR␤1 with the DBD of the glucocorticoid receptor (GR) substituted for that of TR (provided with permission of Dr. Ronald Evans, Salk Institute, La Jolla, CA) and two TR hybrid mutants in which one or the other zinc finger of the TR DBD is replaced with the corresponding zinc finger of GR (T-TG-T or T-GT-T; obtained with permission of Dr. J. Larry Jameson, Northwestern University, Chicago, IL). Recombinant TR␤1 (residues 102-461) was generously provided by Dr. Brian L. West (Metabolic Research Unit, University of California-San Francisco, CA).
Cell Cultures, Treatment, and Preparation of Nuclear Fractions-TRreplete 293T cells were obtained from Dr. K. Pumiglia (Albany Medical College, Albany, NY); CV-1 cells, which lack TR, were on hand in the laboratory (1). All cells were maintained and grown in Dulbecco's modified essential medium supplemented with 10% fetal bovine serum. Almost confluent cells were placed for 2 days in medium containing 0.25% serum that had been previously depleted of thyroid hormone by the method of Samuels et al. (18), as modified by Weinstein et al. (19), and then in serum-free medium for 2 h. Cells were then treated with T 4 or other analogues for the times indicated. In selected experiments the inhibitors PD 98059 and geldanamycin were added to cells for 5 h, and T 4 was added for the last 30 min; KT5720 was added for 70 min, and 8-Br-cAMP and T 4 were added for the last 30 min of KT5720 incubation.
After cell treatment, cells were harvested, and nuclear extracts were prepared as follows: the cells were washed twice with ice-cold phosphate-buffered saline and lysed in hypotonic buffer as described previously (1-3). After sample centrifugation at 4°C and 13,000 rpm for 1 min, supernatants were collected as cytoplasmic extracts. Nuclear extracts were prepared according to the method of Wen et al. (20) by resuspension of the crude nuclei in high salt buffer (hypotonic buffer with 20% glycerol and 420 mM NaCl) at 4°C with rocking for 30 min. The supernatants were collected after centrifugation at 4°C and 14,000 ϫ g for 10 min.
Immunoprecipitation and Immunoblotting-Following normalization of sample protein content, immunoprecipitation was performed using polyclonal anti-phosphotyrosine (Transduction Laboratories, Lexington, KY), monoclonal antibody to the amino-terminal or carboxyl-terminal halves of the amino-terminal (AB) domain (amino acids 1-93 of TR␤1 (Santa Cruz, Santa Cruz, CA), polyclonal antibody to amino acids 73-93 of rat TR␤1 (generously provided by Dr. William W. Chin, Indianapolis, IN), monoclonal antibody to amino acids 235-414 of human TR␤1 (C3 antibody, Affinity Bioreagents, Inc., Golden, CO), monoclonal antibody to MAPK (Transduction), polyclonal anti-phosphoserine (Research Diagnostics, Flanders, NJ), or polyclonal antibody to tyrosine/threonine-phosphorylated MAPK (New England Biolabs, Beverly, MA). After overnight incubation of samples at 4°C with rocking, protein A-agarose was added, and samples were rocked for 1 h at 4°C. After two washes with hypotonic buffer containing 0.2% Nonidet P-40, immunoprecipitates were eluted with 2ϫ sample buffer and proteins (10 g of protein/sample) were resolved on discontinuous SDS-PAGE gel (7.5-9.0%). Proteins were transferred to Immobilon membranes (Millipore, Bedford, MA) by electroblotting. After blocking with 5% milk in Tris-buffered saline containing 0.1% Tween, membranes were immunoblotted with one of several antibodies including monoclonal anti-MAPK or -TR␤1, polyclonal anti-tyrosine/threonine-phosphorylated MAPK, anti-phosphoserine, anti-phosphothreonine (Research Diagnostics), anti-TR␤1, or anti-SMRT (Santa Cruz). The secondary antibodies were rabbit anti-mouse, goat anti-rabbit, or donkey anti-goat IgG (1: 1000, DAKO Corporation, Carpenteria, CA). Immunoblots were visualized by chemiluminescence (ECL; Amersham Pharmacia Biotech) and quantitated by digital imaging (BioImage, Millipore). The possibility of nonspecific immunoprecipitation was ruled out by use of anti-HLA-DR␣ for immunoprecipitation, followed by MAPK immunoblot of the immunoprecipitated proteins; no MAPK was detectable. Supernatants from TR␤1 immunoprecipitation contained no TR␤1, indicating that the immunoprecipitation was complete. Immunoblots shown in the figures are representative of three or more experiments.
Transfection of TR␤1 and Mutants of TR␤1-The DNA samples for TR␤1, two truncated mutants (TR-⌬N and TR-LBD), and three hybrid constructs (TGT, T-GT-T, and T-TG-T), in the pcDNA expression vector were transformed into competent DH5␣ cells and plated on ampicillintreated agar plates. Single colonies were chosen and grown overnight in Circle Grow culture broth (Bio101, Carlsbad, CA) containing ampicillin (10 mg/ml). Plasmid DNA was isolated and purified by a modified alkaline lysis protocol (RPM4G kit, Bio101). CV-1 cells were transfected with TR␤1 or mutants in pcDNA or with pcDNA alone using Lipo-fectAMINE Plus according to supplier's instructions. Cells grown to 60% confluence in serum-free, antibiotic-free medium were replenished with 5 ml of fresh medium, and LipofectAMINE/DNA mixtures previously prepared were then added, followed by incubation for 6 h at 37°C. An equal volume of medium containing 20% T 4 -depleted serum was then added to the DNA-LipofectAMINE Plus mixture, and the cells were incubated for an additional 6 -8 h. The medium was replaced with Dulbecco's modified essential medium containing 10 Ϫ3 M PTU and 0.25% T 4 -depleted serum, and the cells were incubated for 16 h and then treated with T 4 (10 Ϫ7 M) or diluent for 30 min and harvested for preparation of nuclei as described above. Nuclear samples were immunoblotted with a TR␤1 antibody that detects TR␤1 and all mutants or were immunoprecipitated with the same antibody, and precipitated proteins were then separated by PAGE and immunoblotted with antibodies to MAPK, phosphoserine, or phosphothreonine.
In Vitro Serine Phosphorylation of Recombinant TR␤1 by Activated MAPK-5 g of recombinant TR␤1 (amino acids 102-461) were immunoprecipitated with C3 antibody, and the immunoprecipitated protein eluted from protein A-agarose. The immunoprecipitate was incubated for 30 min at 30°C with 5 units of activated MAPK (New England Biolabs) in 50 mM Tris-HCl (pH 8.0), 0.5 mM EDTA, 25 mM MgCl 2 , 1 mM dithiothreitol, 10% glycerol, and 20 M ATP containing 0.05 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences), following the method of Kato et al. (21). Immunoprecipitated TR␤1 from 293T cells was also exposed to activated MAPK. Control samples contained 1 g of MBP instead of TR␤1. After incubation proteins were solubilized, separated by PAGE, and radioautographed.

T 4 Induces Activation of MAPK and Nuclear Translocation of
Co-immunoprecipitated MAPK and TR␤1-Human embryonic kidney (293T) cells contain endogenous TR␤1. To study coimmunoprecipitation, or complexing, of TR␤1 and MAPK, nuclear fractions were prepared from 293T cell lysates of control samples and of cells treated with T 4 (10 Ϫ7 M) for 10 -90 min and were immunoprecipitated with antibody to the carboxyl-terminal half of the AB domain of TR␤1. The immunoprecipitated proteins were eluted, separated by SDS-polyacrylamide gel electrophoresis, and transferred to membranes for immunoblotting with antibody to MAPK. Shown in Fig. 1A is a representative immunoblot. There is an increase in nuclear MAPK complexed with TR␤1 immunoprecipitates that reaches a peak in 40 min (22-fold increase over control in the figure, decreasing to 12-fold in 90 min). In nine experiments, the fold increases in nuclear MAPK associated with TR␤1 in 30 and 40 min were 6.7 Ϯ 1.5 (mean Ϯ S.E.) and 8.6 Ϯ 2.5, respectively, as shown in Fig. 1B. These changes parallel the timing of T 4 -induced increases in tyrosine phosphorylation and nuclear translocation of MAPK that we have previously reported (1).
In studies using a selective antibody to tyrosine/threoninephosphorylated, or activated, MAPK for immunoblotting of TR␤1 immunoprecipitates, we observed 21-and 24-fold in-creases in activated MAPK in 30 and 40 min, respectively (Fig.  1C). In additional experiments, nuclear fractions were immunoprecipitated with antibody to phosphorylated MAPK, and the resulting proteins were immunoblotted with anti-TR␤1, thus reversing the antibody order. Again, complexing of TR␤1 and MAPK was found, with 1.4-and 7.8-fold increases in TR␤1 bands in 30 and 40 min of T 4 treatment, respectively (not shown). We have therefore demonstrated that T 4 in a physiologic concentration can transiently 1) activate MAPK causing its translocation to the nucleus and 2) promote the association of activated MAPK with TR␤1 in an immunoprecipitable complex. This effect of T 4 occurs in as early as 10 min and persists for up to 90 min. Studies utilizing antibody to the aminoterminal half of the AB domain of TR␤1 demonstrated similar results, as did studies with a polyclonal antibody to amino acids 73-93 of the rat TR␤1 AB domain (not shown).
In time course experiments similar to the study shown in Fig. 1A, nuclear extracts of 293T cells treated with T 4 were immunoblotted with TR␤1 antibody without prior immunoprecipitation; Western blots demonstrated 2-5-fold increases in nuclear TR␤1 in 10 -60 min, respectively, as shown in Fig. 2A. Nuclear samples were also immunoprecipitated with monoclonal TR␤1 antibody, and the precipitates were subjected to PAGE and immunoblotted with polyclonal anti-TR␤1 (Fig. 2B). Again, progressive nuclear accumulation of the receptor during T 4 treatment is seen.
We determined in 293T cells whether T 4 treatment for 30 min altered the abundance of TR␤1 per unit of nuclear fraction protein (10 g of protein/lane). A representative blot is shown in Fig. 2C. It is assumed that the increase in nuclear TR␤1 originated from the pool of cytosolic TR␤1 also shown in this figure. We have also found cytosolic TR␤1 in another cell line (NIH3T3 cells; not shown), and others have reported the presence of TR in cytoplasm (22). Extranuclear TR␤1 concentration in our studies was 4.5 Ϯ 2.5% of nuclear receptor concentration, where relative concentration was expressed as band integrated optical density per 10 g of lane protein. The pool of cytosolic TR␤1 may be larger than Fig. 2C implies, however, given the 2-fold greater concentration of total protein in cytoplasm than in nuclear fractions and the relative volumes of nucleus and cytosol in intact cells.
Iodothyronine Analogues and Formation of TR␤1-MAPK Complexes-We have previously reported that physiologic concentrations of T 4 are more effective in nongenomic models of hormone action than physiologic concentrations of T 3 (1-3) and that triac and tetrac may block the action of T 4 at the plasma membrane (1, 3). We therefore studied whether D-T 4 , D-T 3 , rT 3 , tetrac, and triac caused complexing of TR␤1 and MAPK in  1 and 2). PD 98059 inhibited the T 4 effect by 88% in the study shown (lane 2 versus lane 3). Additional studies examined the effect of PD 98059 on serine phosphorylation of TR␤1. Samples were immunoprecipitated with anti-TR␤1, and precipitates were immunoblotted with anti-phosphoserine. In the experiment shown in Fig. 4B, there is 2.9-fold enhancement of TR␤1 serine phosphorylation by T 4 (lanes 1 and 2) and 97% inhibition of serine phosphorylation by PD 98059 (lanes 2 and 3).
T 4 Effect on TR␤1/MAPK Complex Formation Is PKA-mediated-Because a role for PKA in the serine phosphorylation of TR␤1 has been suggested (11,12), we examined the effect of KT5720 and 8-Br-cAMP, respectively, an inhibitor and an agonist of PKA (24). In the study shown in Fig. 5A, T 4 enhanced 293T cell nuclear uptake of TR␤1 complexed with MAPK 7.7fold (lane 2). KT5720 inhibited T 4 -induced complexing of TR␤1 and MAPK (lanes 3 and 4, 33 and 61% inhibition, respectively). 8-Br-cAMP added to the effect of T 4 (lane 8) and was itself agonistic (lane 5). The effect of 8-Br-cAMP was also partially blocked by KT5720 (lanes 6 and 7).
The PKA-mediated T 4 3 and 5, respectively). These studies suggest that PKA, whether activated by T 4 or by 8-Br-cAMP, requires MEK action on MAPK to bring about serine phosphorylation of TR␤1.

Inhibition of both T 4 and 8-Br-cAMP effects by PD 98059 was evident (lanes
T 4 -induced Nuclear Complexing of MAPK with Transfected TR␤1 and Mutants of TR␤1 in CV-1 Cells-Having established that T 4 induces TR␤1/MAPK complexing and serine phosphorylation of the endogenous receptor in 293T cells, we studied CV-1 cells transfected with TR␤1 and found that T 4 produced the same effects. Preliminary studies with two truncated mutants of TR␤1, rTR-⌬N (amino acids 94 -461, containing both DBD and LBD) and rTR-LBD (amino acids 174 -461, without DBD) showed that the DBD must be present for T 4 -induced complexing of MAPK and receptor to occur (results not shown). Based on this finding, we developed a strategy to localize the site in the TR DBD of MAPK-receptor interaction. CV-1 cells were transfected with three hybrid constructs, substituting human GR sequences for corresponding segments of human TR␤1: T-TG-T, containing the second zinc finger of GR instead of that of TR; T-GT-T, with the first zinc finger of GR replacing that of TR; and TGT, which contains the entire DBD of GR in lieu of the TR DBD (Fig. 6A). Results in Fig. 6B confirm transfection of the probes. These cells were treated with PTU and T 4 as described above under "Experimental Procedures," and examined for TR␤1/MAPK complexing. T 4 -stimulated coimmunoprecipitation of TR␤1 and MAPK is seen only in cells with intact TR␤1 (Fig. 6C, upper panel, lane 2) or with the T-GT-T construct, in which the second zinc finger of TR is intact (lane 6). In contrast, in cells with the second zinc finger of TR replaced with that of GR or with the entire DBD replaced with that of GR, no T 4 -induced complex formation of TR and MAPK is seen (lanes 4 and 8, respectively). The TGT construct substitutes GR (446 -488), containing neither serine nor threonine, for TR (132-176) which has two serines (142, 144) and three threonines. However, only one of these amino acids in TR (serine 142) is located next to a proline, a basic requirement for MAPK substrates (25). Anti-phosphothreonine immunoblots of TR immunoprecipitates revealed no receptor threonine phosphorylation (not shown). These results support an essential role for the second zinc finger of TR␤1 in T 4 -induced receptor complexing with activated MAPK and suggest that serine 142 is the phosphorylation site.
Effect of Activated MAPK on Serine Phosphorylation of Recombinant TR␤1- Kato et al. (21) have demonstrated that activated MAPK can cause serine phosphorylation of the estrogen receptor in vitro. To determine whether MAPK can directly serine phosphorylate TR␤1, in vitro studies of thyroid hormone receptor phosphorylation were performed with activated MAPK and TR␤1. Recombinant TR␤1 (amino acids 102-461, 5 g), was further purified by immunoprecipitation with the C3 antibody and then incubated for 30 min at 30°C with activated MAPK (5 units/l), and 20 M ATP including 0.05 Ci of [␥-32 P]ATP in the buffer medium described by Kato et al. (21). A control sample contained 1.0 g of MBP and activated MAPK. Phosphorylation of proteins was evaluated by PAGE and radioautography, and results are shown in Fig. 7. Phosphorylation of MBP by activated MAPK is seen in lane 1. In the lane 2 sample MAPK was absent, and no phosphorylation of recombinant TR␤1 is seen. The lane 3 sample contained activated MAPK and immunoprecipitated recombinant TR␤1, and a 43-kDa labeled band is seen, consistent with the size of the TR␤1 fragment. A sample of immunoprecipitated endogenous TR␤1 from 293T cells was also exposed to activated MAPK and radiolabeled ATP, and in lane 4 a band is seen at a molecular mass of 56 kDa, consistent with that of intact TR␤1.
Effect of T 4 -induced Nuclear Complexing of Activated MAPK and TR␤1 on Binding of SMRT to TR␤1-293T cells were treated with T 4 and/or tetrac (10 Ϫ7 M) for 30 min, and selected samples were pretreated for 1 h with 30 M PD 98059 or 10 M geldanamycin. The latter causes depletion of cellular Raf-1, a serine/threonine kinase responsible for activation of MEK (1). Shown in Fig. 8A is the presence of SMRT in the TR␤1 immunoprecipitate of control cells (lane 1) and a marked reduction in SMRT binding to TR␤1 when cells were treated with T 4 (lane 2). Geldanamycin and PD 98059 both inhibited the T 4 effect on SMRT binding to TR␤1 (lanes 3 and 4, respectively). Tetrac alone had no effect on SMRT binding to TR␤1 (lane 5) but completely blocked T 4 -induced displacement of SMRT from the receptor (lane 6). From prior studies we know that tetrac inhibits T 4 binding to plasma membranes (26) and blocks T 4induced activation of MAPK (1), STAT1␣ (3), and STAT3. 2 In additional studies, T 4 -agarose (10 Ϫ7 M T 4 ) inhibited SMRT binding to TR␤1 by 50% (not shown). These findings support a role for T 4 in the release of SMRT from TR␤1 in a manner that is dependent on 1) an intact MAPK pathway and 2) tetracinhibitable binding of T 4 to a plasma membrane receptor.
The effect of T 3 on SMRT binding to TR␤1 was compared with that of T 4 . In four experiments, levels of SMRT complexed with TR␤1 were reduced by 38 Ϯ 9 and 80 Ϯ 6% of control levels in cells treated with 10 Ϫ8 and 10 Ϫ7 M T 4 , respectively, and the SMRT levels were reduced by 49 Ϯ 18 and 70 Ϯ 9% of control in cells treated with T 3 (10 Ϫ10 and 10 Ϫ7 M, respectively). A representative blot is shown in Fig. 8B. Similar findings were obtained in studies with NCoR antibody (not shown). Thus, a physiologic concentration of T 4 (10 Ϫ7 M) is more effective than a physiologic concentration of T 3 (10 Ϫ10 M) in displacing SMRT from endogenous TR␤1 in 293T cells. DISCUSSION A number of hormones have recently been reported to influence activity of kinase cascades, including the MAPK pathway. Activation of steps in this pathway by gonadotropin-releasing hormone (27), norepinephrine (28), insulin (29), estradiol (21,30), and T 4 (1) are several examples of such hormones. MAPK has been shown to serine phosphorylate the nuclear estrogen receptor, and overexpression of MEK or Ras in Cos-1 cells potentiates estrogen-induced transcriptional activity of the estrogen receptor (21). Modulation of glucocorticoid receptor activity by MAPK has also been described (31). Chick oviduct nuclear progesterone receptor is another phosphoprotein whose hormone-induced activation is serine protein kinase-dependent (32). A variety of changes in the activity and stabilization of TR have been ascribed to serine phosphorylation (11,13,15,16,33,34), and PKA activity has been implicated in the process (11,12), but no further mechanism of control of serine phosphorylation of TR has been suggested.
We have found that T 4 nongenomically promotes the phosphorylation and nuclear uptake of MAPK and the co-immunoprecipitation of nuclear MAPK with STAT1␣ and STAT3 (1,2). We have shown that this effect results in phosphorylation of serine 727 of STAT1␣ (1). We therefore examined the possibility that iodothyronines, particularly T 4 , might promote serine phosphorylation of TR␤1 by activation of MAPK. Were this to be the case, thyroid hormone in the form of T 4 could modulate the activity of the nuclear receptor by a mechanism distinct from the binding of T 3 directly to TR␤1. That is, without competing for the T 3 -binding site on TR␤1, T 4 might affect the transcriptional activity of the liganded nuclear T 3 receptor by promoting serine phosphorylation of the protein.
In the present studies we demonstrated in 293T cells that treatment with physiologic concentrations of T 4 caused endogenous nuclear TR␤1 to increase and promoted MAPK and TR␤1 co-immunoprecipitation in nuclear fractions. The time course of these effects was relatively rapid, being apparent in 10 min and maximal by 30 -40 min. The association of TR␤1 and MAPK was evident whether we immunoprecipitated TR␤1 first and then reprobed with anti-MAPK antibody or immunopre- In three studies, 100 and 75% inhibition of the T 4 effect, respectively, were found with geldanamycin and PD 98059. Tetrac was ineffective alone (lane 5) but completely blocked the action of T 4 in this study (lane 6; 44% reduction in three studies). B, 293T cells were treated with T 4 or T 3 in the concentrations shown for 30 min and then studied as described in A, above. T 4 (10 Ϫ8 and 10 Ϫ7 M) reduced SMRT binding to TR␤1 by 34 and 87%, respectively, in the study shown, and by 38 Ϯ 9 and 80 Ϯ 6% in four experiments. T 3 (10 Ϫ10 and 10 Ϫ7 M) reduced SMRT binding to TR␤1 by 49 Ϯ 18 and 70 Ϯ 9% in four experiments. cipitated MAPK and then identified TR␤1 in that precipitate.
The structure-activity relationships of the iodothyronine analogues studied in this MAPK-TR model were identical to those we have previously reported in the nongenomic activation by thyroid hormone of MAPK and its association with STAT proteins (1,2). L-T 4 was more active at physiologic concentrations than L-T 3 , and T 4 -agarose was as active as T 4 . The fact that deaminated analogues (tetrac and triac) were not thyroid hormone agonists but were capable of blocking the action of T 4 is consistent with the existence of a cell surface receptor for thyroid hormone that we have previously described (26,35,36), which is pertussis toxin-and GTP␥S-sensitive (37). We have previously demonstrated that tetrac and triac block T 4 potentiation of the antiviral (38) and immunomodulatory (3) actions of interferon-␥, even though these analogues themselves have no effect on interferon-␥ action.
We have detected TR␤1 in cytosol of 293T cells (Fig. 2B). TR␤1 is generally regarded to be restricted to the nucleus but has been reported recently by Zhu et al. (22) to be present in cytosol. In our studies of 293T cells, we found TR␤1 in cytosol in much lower concentrations than those in corresponding nuclear fractions. The role of cytosolic TR␤1 is not clear; it may be nascent in an inactivated state or may be undergoing degradation. Although nuclear estrogen receptor and plasma membrane estrogen receptor apparently represent transcripts of the same gene (39), the TR␤1 detected in cytosol in our studies is not currently thought by us to represent trafficking of TR␤1 to the cell surface. We base this conclusion on the structureactivity relationships of iodothyronine analogues in their actions at the cell surface (1,3,26), compared with those at the nuclear TR (40,41).
In the present experiments, we showed that MEK activity was required to cause nuclear translocation of phosphorylated MAPK and association of MAPK with TR␤1 in cell nuclei. PD 98059, an inhibitor of MEK, prevented activation of MAPK and the nuclear association of phosphorylated MAPK and TR␤1 and inhibited T 4 -induced serine phosphorylation of TR␤1. Thus, the MEK-MAPK cascade that has been implicated in serine phosphorylation of estrogen receptor (21,30) is also operative in T 4 -treated cells in which TR␤1 is serine-phosphorylated. In studies not presented here, we have found that T 4 can also promote the nuclear association of estrogen receptor and MAPK. 2 The fact that TR␤1 and MAPK were found to be associated in the nucleus did not prove that the receptor was a substrate for MAPK. However, using a phosphoserine antibody we documented that endogenous nuclear TR␤1 complexed with MAPK was serine-phosphorylated in T 4 -treated cells and that this serine phosphorylation was inhibited by PD 98059. In contrast, we found no evidence for threonine phosphorylation of TR.
Studies with two truncated mutants of TR␤1 suggested that to demonstrate T 4 -induced co-immunoprecipitation of MAPK and the receptor, the DBD must be present. With the use of hybrid constructs of TR␤1 and GR in which the zinc fingers of the DBDs were exchanged, we identified the second zinc finger of the TR DBD as a necessary participant in MAPK action. In cells with either this portion of TR or the entire TR DBD replaced with that of GR, T 4 treatment did not result in coimmunoprecipitation of TR and MAPK or serine phosphorylation of the TR/GR hybrid. The second zinc finger of the DBD has also been found to be important for homodimerization of TR and transcriptional activation (42).
We also demonstrated that constitutively activated MAPK phosphorylates purified recombinant human TR␤1 in vitro. The receptor fragment we used (residues 102-461) contains 2 serines in the DBD and 14 in the LBD. From our studies described above, we concluded that MAPK binds to the DBD, and serine phosphorylates the DBD. None of the serines in TR are in environments that qualify as optimum consensus phosphorylation sites, namely, PX n (S/T)P, where X is a neutral or basic amino acid and n ϭ 1 or 2, or as the minimum sequence site ((S/T)P; Ref. 21). However, MAPK substrates may lack consensus sequences, and the enzyme may phosphorylate a PS sequence (25) similar to that which occurs in the TR DBD at residues 141-142. Another PS sequence occurs at amino acids 98 -99 of TR, but this segment was not present in the recombinant TR␤1 used in our studies.
It is the hinge region of TR (amino acids 211-240), located in the amino-terminal portion of the LBD, that binds the corepressor proteins SMRT and NCoR (9), which contribute to basal repression of transcriptional activation. Our studies demonstrate that T 4 alone, in the absence of T 3 , is effective in displacing SMRT and NCoR from the receptor. This action of T 4 has characteristics similar to T 4 -induced activation of MAPK and TR␤1/MAPK complexing, including dependence on MEK activity, effectiveness of T 4 -agarose (1), and inhibition by tetrac and triac (1). We therefore postulate that T 4 treatment causes serine phosphorylation in the DBD of TR␤1, leading to allosteric changes in the proximal region of the LBD resulting in dissociation of SMRT from TR␤1. Such a change in the state of the receptor would result, even in the absence of T 3 and binding of T 3 by TR␤1, in derepression of the receptor without ligandinduced transcriptional activation. That this may be the case has recently been shown in a preliminary study, where T 4 in the absence of T 3 caused return of TR␤1 from a state of transcriptional repression to the basal state but did not cause transcriptional activation of the receptor. 2 We have previously shown that thyroid hormone can potentiate the action of interferon-␥ by a signal transduction pathway that involves both protein kinase C and PKA (24). We therefore studied the possibility that T 4 might also promote the serine phosphorylation of TR␤1 by a PKA-dependent mechanism and found that the MEK-MAPK pathway by which T 4 acts to phosphorylate TR␤1 is indeed subject to activation by 8-Br-cAMP and inhibition by KT5720. Our observations suggest that the involvement of PKA in TR␤1 phosphorylation reported by others depends upon MAPK activation and that T 4 -stimulated PKA contributes to activation and nuclear translocation of MAPK.