0013-7227/02/$15.00/0 Printed in U.S.A.

The Journal of Clinical Endocrinology & Metabolism 87(2):906 –914 Copyright © 2002 by The Endocrine Society

TR Expression and Function in Human Bone Marrow Stromal and Osteoblast-Like Cells AYESHA SIDDIQI, MARIE P. PARSONS, JOHN L. LEWIS, JOHN P. MONSON, GRAHAM R. WILLIAMS, AND JACKY M. BURRIN Department of Endocrinology, St. Bartholomew’s and the Royal London School of Medicine and Dentistry (A.S., M.P.P., J.P.M., J.M.B.), West Smithfield, London, United Kingdom EC1A 7BE; and Department of Hematology, LRF Center for Adult Leukemia (J.L.L.), and ICSM Molecular Endocrinology Group, Division of Medicine and Medical Research Council Clinical Sciences Center (G.R.W.), Imperial College School of Medicine, Hammersmith Hospital, London, United Kingdom W12 0NN were also apparent, with TR␣1> TR␤1 > TR␤2 in all cell types. Incubation with [125I]T3 confirmed reversible T3 binding to cell nuclei. Specific binding was greatest in MG63 > hBMS > SaOs-2 > hOb. Finally, endogenous TR activity was determined in transfections using a thyroid hormone response element derived from the rat GH gene linked to the luciferase reporter gene. In MG63 and hBMS cells T3 treatment increased luciferase activity 5.5 ⴞ 0.7-fold (P < 0.05), confirming the presence of endogenous receptors. In SaOs-2 and hOb cells, T3 treatment had no effect on thyroid hormone response element-thymidine kinase-luciferase expression, suggesting that in these cells TR expression was too low to be detected. These results indicate that three main TR isoforms are expressed in cells of the human osteoblast lineage, but that expression and endogenous TR activity are predominantly present in hBMS cells. Whether there are distinct mechanisms of thyroid hormone action mediated by TR␣1, TR␤1, and TR␤2 in hOb and hBMS cells remains to be shown. (J Clin Endocrinol Metab 87: 906 –914, 2002)

Thyroid hormones influence both bone formation and bone resorption. In vitro studies demonstrate direct effects of thyroid hormones on cells of the osteoblast lineage. Transcriptional regulation by thyroid hormones is mediated by liganddependent transcription factors called TRs. The three main T3-binding TR isoforms are TR␣1, TR␤1, and TR␤2. TRs have been identified in cells of the osteoblast lineage, but it is still not known whether TR isoform expression differs in primary cultures of human osteoblasts. We used immunocytochemistry, Western blotting, nuclear binding assays, and transient transfection studies to examine the expression of functional TR isoforms in primary cultures of osteoblasts (hOb) derived from explants of trabecular bone, in human bone marrow stromal cells (hBMS), which are believed to be the source of osteoblast progenitor cells, and for comparison in the transformed human osteosarcoma cell lines MG63 and SaOs-2. TR␣1, TR␤1, and TR␤2 proteins were expressed in all cells, although expression was greatest in MG63 > hBMS > SaOs-2 > hOb. Differences between isoforms

T

HYROID HORMONES PLAY an important role in the growth, development, and turnover of bone by influencing the rate of bone resorption and formation. Children with juvenile hypothyroidism have growth arrest, delayed bone maturation, and epiphyseal dysgenesis. In contrast, thyrotoxic children have accelerated growth velocity and advanced bone maturation. Thyrotoxic adults are at risk of osteoporosis (1, 2) and increased mortality from fractures of the femur (3). Increased bone formation and resorption in thyrotoxicosis suggest that both osteoblasts and osteoclasts are potential targets for T3 action. Direct responsiveness of osteoclasts to T3 has been disputed, and the majority of the actions of T3 on bone are thought to be mediated via osteoblasts. In response to T3, osteoblasts are able to increase DNA synthesis and alkaline phosphatase (ALP) (4) and osteocalcin (5) production and are permissive for the action of osteoclasts on bone resorption (6, 7). TRs are central to conferring T3 responsiveness to cells. They may bind to DNA as homodimers (8) or as heterodimers with 9-cis-RXR, which is also a common het-

erodimer partner for all-trans-RAR (9). RXR and RAR are coexpressed with TRs in osteoblasts and modify the regulation of endogenous gene (ALP, osteocalcin, osteopontin) expression by T3 in these cells (10). There are two classes of TR in man, designated TR␣ and TR␤. Four principal TR isoforms have been described, namely ␣1, ␣2, ␤1, and ␤2, which result from alternative splicing. TR␣ isoforms differ in the carboxyl-terminal region of the ligand-binding domain, and ␤ isoforms differ in the amino-terminal. Thus, TR␣1, TR␤1, and TR␤2 bind T3, but TR␣2, lacking the terminal region of the ligand-binding domain, does not. TR␣1 and ␤1 mRNA and protein are expressed in a variety of tissues, although relative ratios of the isoforms may vary between tissues and during development. The ␤2 variant is also expressed more widely than initially thought, and protein expression has now been documented outside the brain, namely in heart, liver, and kidney. The majority of studies on TR expression in osteoblasts have used rodent cell lines. Binding of radiolabeled [125I]T3 has been documented in ROS 17/2.8, UMR 106, and MC3T3-E1 cell lines. Similarly, expression of isoform-specific TR mRNA by Northern blotting and of TR protein by Western analyses has been documented in both rodent cell lines (10) and primary cultures of rat osteoblasts (11). No differences in TR isoform expression have been reported between

Abbreviations: ALP, Alkaline phosphatase; 1,25-(OH)2D3, 1,25-dihydroxyvitamin D3; hBMS, human bone marrow stromal cells; hOb, human osteoblasts; PBS-T, PBS plus 0.1% Tween 20; TKLUC, thymidine kinase gene promoter linked to a luciferase reporter gene; TRE, thyroid hormone response element.

906

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

Siddiqi et al. • TR Expression in Bone

bone derived from rat femoral and vertebral sites (12), although subtle differences may be present in T3 responsiveness with respect to ALP activity and gene expression of osteocalcin and IGF-I (13). Less is known about the expression and function of TRs in human osteoblasts. To date, TR protein has been identified by means of immunocytochemistry in the osteosarcoma-derived MG63 cell line (14) and in heterotopic bone (15). We have demonstrated elevated levels of IL-6 and IL-8 both in patients with thyrotoxicosis (16) and in cells of the osteoblast lineage stimulated with T3 (17). We documented differential responses to T3 in osteoblasts derived from different sources. Human bone marrow stromal cells (hBMS), the source of osteoblast progenitor cells, are able to release the bone-resorbing and osteoclast-activating cytokines IL-6 and IL-8 in response to T3; in contrast, primary human osteoblasts (hOb), cultured from explants of human trabecular bone, fail to produce IL-6 and IL-8 in response to T3. From these results a possible mechanism involving hBMS cells may be postulated for T3-induced bone resorption. Studies of TR expression and function are pertinent to the understanding of bone physiology in normal and diseased states. The aim of our study therefore was to extend TR studies to human species to address the question of whether human osteoblasts possess TRs and whether the difference in T3-stimulated cytokine release between hOb and hBMS cells can be explained by differences in isoform-specific expression or function of TRs in these cells. We therefore studied the expression and function of the main T3-binding TR isoforms in primary cultures of human osteoblasts derived from trabecular bone and human bone marrow. For comparison, we also studied two human osteoblast-like cell lines, SaOs-2 and MG63. Materials and Methods Cell culture The hOb-like cell lines MG63 and SaOs-2 were maintained in ␣MEM (Life Technologies, Inc., Paisley, UK) containing 10% FCS (Life Technologies, Inc.), penicillin, streptomycin, and fungizone (hereafter referred to as medium) at 37 C in a humidified atmosphere containing 5% CO2 in air. hBMS. The procedures for establishing primary osteoblast cultures from human bone marrow cells have been described previously (17, 18). Bone marrow cells were obtained by iliac aspiration from normal healthy donors (age range, 18 –55 yr) who volunteered for the allogenic bone marrow transplantation program at the Hammersmith Hospital. Ethical approval was obtained from the Hammersmith Hospital research ethics committee. Marrow was diluted 1:1 with medium and centrifuged at 500 ⫻ g for 10 min. Pelleted cells were resuspended in equal volumes of medium and Histopaque (specific gravity, 1077) and centrifuged for an additional 30 min at 1800 rpm. Mononuclear cells present in the interface were then harvested. Cells (5 ⫻ 106) were plated in a 25-cm2 tissue culture flask in medium containing 1% 200 mm l-glutamine (Life Technologies, Inc.) and 2 ␮m methylprednisolone (Solumedrone, Upjohn Pharmaceuticals, Milton Keynes, UK). Cells were allowed to attach without disturbance for 7 d, after which medium was changed at 2- to 3-d intervals. At confluence, cells were trypsinized, counted, and plated into multiwell culture plates. First passage cells were used in all experiments. hOb. The procedures for culturing hOb have been described previously (17) and are based on the method of Robey and Termine (19). Femoral tissue was obtained from bone discarded during hip replacement surgery for osteoarthritis from 3 men, aged 70 – 85 yr, and 4 women, aged

J Clin Endocrinol Metab, February 2002, 87(2):906 –914 907

50 –75 yr. Given potential site-dependent differences in T3 responsiveness (13), we used bone derived from the femoral site only. Trabecular bone was removed by scraping the surface with a bone curette and was minced into 2- to 4-mm pieces with dissecting scissors. The bone chips were then washed in PBS and incubated in serum-free medium containing 1 mg/ml collagenase (type IV, Sigma, St. Louis, MO). After 30 min, the supernatant fraction containing collagenase-released cells was discarded, and the bone chips were placed in tissue culture flasks in medium. Cells growing out from the bone chips were passaged once, and all assays were performed on cells subcultured at the end of the first passage. Exclusion criteria for all hOb and hBMS donors included the absence of medication that might affect the bone response to T3, specifically estrogens, glucocorticoids, and bisphosphonates.

Characterization of cell phenotype Our culture conditions were based on the methodology of Haynesworth (20) and Rickard (21), both of whom have extensively characterized their stromal cell population and found them to be predominantly of the osteogenic lineage following appropriate culture conditions. We also confirmed this (17). Experiments were performed on first passage cells plated at a density of 10,000 cells/cm2 and incubated overnight in medium. The osteoblastic phenotype of confluent cells from all four sources was confirmed by measuring ALP positivity and 1,25-dihydroxyvitamin D3 [1,25-(OH)2D3] induction of osteocalcin. ALP staining. Nitro blue tetrazolium chloride (Sigma, Poole, UK) in 0.1 m Tris buffer, pH 9.5, was added to the cells fixed in formalin. The ALP substrate 5-bromo-4-chloro 3-indolyl phosphate gives a blue oxidation product after the addition of nitro blue tetrazolium chloride. Blue staining, confirming ALP positivity, was seen in all cell types. Osteocalcin. Osteocalcin was measured in the media of cells stimulated for 24 h with 10 nm 1,25-(OH)2D3. The OSTK-PK (CIS Biointernational, Gif-sur-Yvette, France) RIA was used. hBMS, SaOs-2, and MG63 cells had only 1,25-(OH)2D3-dependent osteocalcin secretion, in contrast to hOb cells, which had detectable constitutive and 1,25-(OH)2D3-stimulated osteocalcin secretion (Table 1). For chondrocyte and adipocyte analyses, cells were fixed in formalin and stained for 30 min with Alcian blue (10% in 0.1 m HCl) for chondrocytes or overnight with Oil red O (0.5% in propylene glycol) for adipocytes (22). No staining for mature chondrocytes or adipocytes was observed in any of the cell types.

Plasmids All plasmids were obtained from Prof. V. K. K. Chatterjee (University of Cambridge, Cambridge, UK): TKLUC, containing the thymidine kinase (TK) gene promoter linked to a luciferase (LUC) reporter gene (23), was used as a control. Thyroid hormone response element (TRE)TKLUC, which incorporates two copies of a TRE (24), was used to detect the presence of functioning endogenous T3 receptors. RSVerbA␤, an expression vector containing the human thyroid hormone ␤1 receptor cDNA (25), was used to determine the effect of excess T3 receptors on TRE-TKLUC responsiveness. The expression vector Bos␤Gal contains the promoter of the human elongation factor 1␣ gene driving the expression of ␤-galactosidase (26) and was cotransfected as an internal control plasmid to normalize transfection efficiencies. TABLE 1. Basal and 10⫺8 M 1,25-(OH)2D3-stimulated osteocalcin production in human osteoblasts and osteoblast-like cells Osteocalcin (ng/mg protein)

MG63 SaOs-2 hBMS hOb

Basal

⫹10 nM 1,25-(OH)2D3

0.9 ⫾ 0.05 0.1 ⫾ 0.05 0.3 ⫾ 0.09 3.2 ⫾ 1.8a

45.5 ⫾ 8.9 32.1 ⫾ 8.0 38.3 ⫾ 9.3 50.9 ⫾ 10.8

a P ⬍ 0.05 for basal osteocalcin production in hOb compared to hBMS.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

908

J Clin Endocrinol Metab, February 2002, 87(2):906 –914

Siddiqi et al. • TR Expression in Bone

Transient expression assays All cells were plated at a density of 0.5 ⫻ 10 cells/60-mm plate and incubated overnight. Thereafter, cells were transfected using the lipofection reagent dioleoyloxypropyl trimethyl ammonium methylsulfate (Roche Molecular Biochemicals, Lewes, UK). Dioleoyloxypropyl trimethyl ammonium methylsulfate (30 ␮g) and purified plasmid DNA were each diluted in 20 mm HEPES and 150 mm NaCl, pH 7.4, before addition to each well. Cells were transfected with TKLUC (10 ␮g), TRE-TKLUC (10 ␮g), or a combination of RSVerbA␤ (1 ␮g) and TRE-TKLUC (10 ␮g) together with 3 ␮g Bos␤Gal. Twenty-four hours later, cells were incubated in medium containing 10% charcoal-stripped FCS (GlobePharm, Surrey, UK) in the presence or absence of a physiological concentration (1 nm) of T3 (Sigma). After overnight incubation, medium was removed, and cellular extracts were prepared and assayed for luciferase activity as described previously (27). Luciferase data were normalized for ␤galactosidase activity, which was assayed in the same cell extracts using a spectrophotometric microtiter plate assay with d-nitrophenyl-␤O-galactopyranoside as substrate. 6

Nuclear protein extraction and Western blotting

for 1 h, washed in 0.1% PBS, and left overnight in primary antibody diluted 1:400 in 0.1 m PBS-T. Controls were processed in the absence of primary antibody. After three 15-min washes in 0.1 m PBS-T, cells were incubated for 1 h in secondary antibody (horseradish peroxidase, Amersham Pharmacia Biotech) diluted 1:800 in 0.1 m PBS-T. Peroxidase activity was visualized with diaminobenzamine tetrahydrochloride plus 0.1% H2O2. After three 15-min washes in 0.1 m PBS, coverslips were dehydrated in graded alcohol and mounted with Depex (BDH Laboratories). To standardize staining, the protocol was adhered to strictly, and cells were stained in batches that included samples from all cell types. TR expression was compared semiquantitatively within each batch. The total cell count and the number of stained cells were assessed in at least three coverslips (three to five fields per coverslip at ⫻100 magnification) using an Olympus Corp. CK-2 microscope (New Hyde Park, NY). For colocalization of ALP-positive and TR-positive cells, a duplicate set of cells in each experiment was stained for alkaline phosphatase immediately after fixation and before immunocytochemistry for TR as described above.

[125I]T3 binding studies with intact cells

Cells were plated into multiwell plates at a density of 1.5 ⫻ 10 cells/60-mm plate in medium. After overnight incubation, cells were placed in fresh medium containing charcoal-stripped serum. Twentyfour hours later, they were treated with either 1 or 10 nm T3. Wells plated with osteoblastic cells were incubated with no additions and used as negative controls, and wells plated with GH3 somatolactotroph cells were used as positive controls. GH3 cells have been shown previously to express TRs (28). After a further 24 h, medium was aspirated, and the cells were harvested for extraction of nuclear proteins as described previously (10). The protein content of the nuclear extracts was determined using the Bradford protein assay (Bio-Rad Laboratories, Inc., Hemel Hempstead, UK). Prestained SDS protein standards (Bio-Rad Laboratories, Inc.) and 20 ␮g protein extract were diluted 1:1 in loading buffer [4% SDS, 20% glycerol, 0.1% dithiothreitol, 125 mm Tris-HCl (pH 6.8), and 0.005% bromophenol blue dye], and loaded onto 12.5% polyacrylamide gel (Bio-Rad Laboratories, Inc.). The samples were electrophoresed at 120 mV for 40 min, electroblotted at 100 mV for 1 h onto nitrocellulose paper, and then blocked for an additional 1 h in PBS plus 0.1% Tween 20 (PBS-T; Sigma) containing 20% nonfat milk (Marvel, Premier Brands, Stafford, UK). Proteins were then incubated overnight with primary antibody diluted 1:400 in PBS-T and then with secondary antibody diluted 1:20,000 in PBS-T for 1 h. TR protein was detected using enhanced chemiluminescence (ECL Western Blotting Reagents, Amersham Pharmacia Biotech, Little Chalfont, UK) after exposure of the nitrocellulose to x-ray film (ECL Hyperfilm, Amersham Pharmacia Biotech) for 10 sec to 15 min. Using Whole-Band Analysis software (BioImage Computer Systems, Crewe, UK), each negative was scanned, the image was digitized, and the OD of each band was measured. The coefficients of variations for repeated measurements were less than 20%. The linear quantitative detection of each receptor isoform was optimized to enable TR isoform expression to be compared among cell types. 6

Immunocytochemistry Specific rabbit polyclonal anti-TR-␣1 and -␤1 antibodies were obtained from Affinity BioReagents, Inc. (Golden, CO). We have shown that these antibodies react specifically with the TR isoforms in osteoblastic cell lines (10) and have used them previously in immunocytochemistry studies (29). The TR␤2 antibody was obtained from Prof. V. K. K. Chatterjee (University of Cambridge). This was also a rabbit polyclonal directed to the N-terminal domain of the human ␤2 fused to glutathione-S-transferase. Cells were plated at a density of 3 ⫻ 105 to 5 ⫻ 105 on coverslips (BDH Laboratories, Poole, UK) and allowed to settle overnight, followed by incubation for 24 h in ␣MEM with 10% CS-FCS. Cells were then treated with either no stimulant (negative control) or T3 (1 or 10 nm) for an additional 24 h before being fixed in 10% formalin for 10 min. Endogenous peroxidase activity was removed with 1.5% H2O2. Coverslips were then incubated in blocking buffer (0.1 m PBS-T and 1% goat serum)

T3 binding studies were performed to confirm the presence of T3binding receptors in human stromal and osteoblast cell cultures. A modification of the method of Samuels and Tsai (30) was used. MG63, SaOs-2, hOb, or hBMS cells were plated into multiwell dishes at a density of 104 cells/cm2 in medium. After overnight incubation, cells were washed in PBS and left for 24 h in medium containing 10% charcoalstripped serum. After three washes with PBS, cells were treated with 1 ml medium containing 10% CS-FCS and 5 pmol [125I]T3 (SA, 800 ␮Ci/␮g; 8.2 ⫻ 105 cpm; NETRIA, London, UK) in the presence or absence of 500 nmol unlabeled T3 for 30, 60, or 120 min at 37 C. Nonspecific binding was calculated as binding of [125I]T3 in the presence of an excess of unlabeled T3. Specific binding was defined as the difference between binding of [125I]T3 in the presence or absence of an excess of unlabeled T3. The reversibility of binding was determined by adding an excess (500 nmol/ ml) of unlabeled T3 for 60 min to cells previously incubated for 60 min in the presence of [125I]T3 alone. At the end of each specified incubation period, nuclear extracts were prepared as described above. The radioactivity associated with nuclear extracts was analyzed using an NE 1600 ␥-counter (Nuclear Enterprises, Dorset, UK). The protein content of the nuclear extracts was determined using the Bradford protein assay. Each incubation was performed in duplicate, and experiments on each cell type were performed on two separate occasions.

Statistical analysis Results are expressed as the mean ⫾ sem for each experimental group and were compared by one-way ANOVA. When the F statistic was significant (P ⬍ 0.05), the analysis was continued using Fischer’s multiple comparisons test.

Results Immunocytochemistry

The immunostaining characteristics of hOb, hBMS, MG63, and SaOs-2 cells were determined using specific antibodies to TR␣1, TR␤1, and TR␤2. Figure 1 (A–C) shows that all cell types demonstrated expression of TR␣1, TR␤1, and TR␤2. Staining was compared among the four cell types within each batch. Staining with all three isoform-specific antibodies was similar in MG63, SaOs-2, and hBMS cells, but was less in hOb. All TR antibodies gave rise to some cytoplasmic staining, as noted previously (8, 14). Dual staining for ALP and TR (␤1) confirmed that the positive nuclear stain for thyroid receptors was within ALP-positive cells (Fig. 1D). However, not all ALP-positive cells were positive for TR staining. There was no staining visible on control slides processed in the absence of primary antibody.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

Siddiqi et al. • TR Expression in Bone

J Clin Endocrinol Metab, February 2002, 87(2):906 –914 909

FIG. 1. A, Nuclear immunostaining for TR␣1 in MG63 (a), SaOs-2 (b), hBMS (c), and hOb (d). Magnification, ⫻100. B, Nuclear immunostaining for TR␤1 in MG63 (a), SaOs-2 (b), hBMS (c), and hOb (d). Magnification, ⫻100. C, Nuclear immunostaining for TR␤2 in MG63 (a), SaOs-2 (b), hBMS (c), and hOb (d). NC, negative control. Magnification, ⫻100. D, Dual staining for ALP and TR␤1 in hBMS (a) and hOb (b) cells. Nuclear staining (brown) for TR␤1 can be seen against background staining (blue) ALP.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

910

J Clin Endocrinol Metab, February 2002, 87(2):906 –914

Western analyses

Protein expression was determined by Western blotting using the same antibodies as those used for immunocytochemistry. TR␣1 (48 kDa), TR␤1 (55 kDa), and TR␤2 (52 kDa) proteins of appropriate sizes (31, 32) were detected in each cell type (Fig. 2) and comigrated with 5–10 ␮g of either in vitro translated cDNA for TR␤1 and TR␤2 (Prof. V. K. K. Chatterjee, University of Cambridge) or nuclear protein extracts from the well characterized, TR-positive, pituitary GH3 cells. Levels of TR proteins were quantified using scanning densitometry. TR␣1 expression was higher in MG63 (5.0 ⫾ 0.9fold) and hBMS (4.3 ⫾ 0.6-fold) cells than in SaOs-2 (2.2 ⫾ 0.3-fold) or hOb (1.0 ⫾ 0.09-fold). TR␤1 expression predominated in SaOs-2 (2.5 ⫾ 0.4-fold) compared with MG63 (2.0 ⫾ 0.2-fold), hBMS (1.0 ⫾ 0.08-fold), and hOb (1.0 ⫾ 0.07-fold). TR␤2 expression predominated in SaOs-2 (4.5 ⫾ 0.2-fold) and hOb (5.2 ⫾ 0.4-fold) compared with MG63 (4.2 ⫾ 0.2-fold) and hBMS (1.0 ⫾ 0.6-fold). The effects of treatment with T3 were also assessed by Western analysis. Twenty-four-hour treatments with T3 (1 and 10 nm) did not alter levels of TR␣1, TR␤1, and TR␤2 in any cell type. Nuclear binding

Figure 3 demonstrates [125I]T3 bound relative to the amount of protein present. 125I-labeled T3 binding appeared

Siddiqi et al. • TR Expression in Bone

to peak between 30 and 60 min, in keeping with previous studies (4, 33, 34), and was greater in hBMS and MG63 than in SaOs-2 and hOb. Nonspecific binding was evaluated in the presence of 500 nm T3, and at 60 min represented 25%, 21.7%, 19.4%, and 44% of the specific binding in MG63, hBMS, SaOs-2, and hOb cells, respectively. Reversibility of binding was confirmed when specifically bound [125I]T3 was displaced within 60 min of addition of 500 nm unlabeled T3. Transient expression assays

To determine whether the endogenous receptors expressed in the cells were functional, all cell types were transfected with a plasmid containing TREs linked to a luciferase reporter gene (Fig. 4). Basal expression of the TRE-containing reporter gene was approximately 1.8-fold greater than the control TKLUC in the untreated cells. In MG63 and hBMS cells, T3 treatment stimulated further expression of this construct, and luciferase activity increased 4.7 ⫾ 0.9- and 5.5 ⫾ 0.7-fold, respectively, compared with controls (P ⬍ 0.05), confirming the presence of endogenous receptors. A further, highly significant (P ⬍ 0.001), response to T3 stimulation was demonstrated in MG63 (13.1 ⫾ 2.3-fold) and hBMS (11.2 ⫾ 2.0-fold) cotransfected with the T3 receptor expression vector RSVerbA␤. This rise was absent in unstimulated cultures. In SaOs-2 and hOb cells, T3 treatment had no further effect

FIG. 2. Western blot analysis for TR␣1 (48 kDa; A), TR␤1 (55 kDa; B), and TR␤2 (52 kDa; C) in MG63, SaOs-2, hBMS, and hOb cells. A, Each blot includes GH3 nuclear extracts as a positive control (left lane), followed by control (C) and 1 and 10 nM T3-treated nuclear extracts. B, Each lane includes in vitro translated ␤1 protein as a positive control, followed by control (C) and 1 and 10 nM T3-treated nuclear extracts. C, Each lane includes in vitro translated ␤2 protein as a positive control, followed by control (C) and 10 nM T3-treated nuclear extracts.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

Siddiqi et al. • TR Expression in Bone

J Clin Endocrinol Metab, February 2002, 87(2):906 –914 911

FIG. 3. Time course of nuclear binding of [125I]T3 in MG63 (A), SaOs-2 (B), hBMS (C), and hOb (D) cells. Binding was measured in the absence (f) or presence (Œ) of 500 nM T3. Reversibility of binding was determined by adding an excess of unlabeled T3 (500 nM) to cells previously incubated for 60 min in the presence of [125I]T3 alone (dotted line). Each point represents the mean ⫾ SEM of two experiments performed in duplicate.

on TRE-TKLUC expression (2.0 ⫾ 0.4- and 1.5 ⫾ 0.2-fold, respectively; P ⬎ 0.05). However, in cells transfected with RSVerbA␤, a significant response was seen, albeit lower than in MG63 and hBMS cells (SaOs-2, 5.2 ⫾ 0.6-fold; hOb, 4.6 ⫾ 0.4-fold; P ⬍ 0.01). The luciferase activity of the control TKLUC remained unaffected by the presence or absence of T3. In none of the above studies were there any apparent difference observed in T3R expression or response to T3 between hBMS or hOb cells from different donors. Discussion

We have described for the first time the expression of TR isoforms in primary cultures of human osteoblasts and bone marrow stromal cells and shown that the TR␣1, TR␤1, and TR␤2 isoforms are present in primary cultures as well as osteoblast cell lines. Their relative expression varies with cell type, but receptor expression does not necessarily predict function. In particular, this work has highlighted the importance of human bone marrow in thyroid and bone physiology. Immunocytochemistry demonstrated similar staining intensities for all three isoforms in MG63, SaOs-2, and hBMS cells, but weaker staining in hOb. However, it is important to remember that in analyses based on the use of specific antibodies, levels of the isoforms are not strictly comparable within any one cell because of the use of separate antibodies with unquantified affinities for TR antigen. Some diffuse

cytoplasmic staining was observed, although the nuclear staining was more intense and localized. This has been observed previously in MG63 cells and also in other tissues (8, 14). TRs are constitutively localized to the nucleus, and thus cytoplasmic immunostaining may be nonspecific, representing leakage of antigen from the nucleus or weak cross-reactivity of these antibodies with unknown proteins. Western analyses demonstrated protein for TR␣1, TR␤1, and TR␤2 in all cell types studied. Overall, TR expression was low in hOb compared with MG63 and hBMS cells. Other studies have also demonstrated differential expression of TR. In rat cell lines, Williams et al. (10) demonstrated variation in TR protein with osteoblast phenotype. In the preosteoblast UMR106 and the mature ROS 17/2.8 cells, the expression of TR␤1 was 2- and 5-fold greater, respectively, than that of TR␣1. In fibroblast-like immature ROS 25/1, this relationship was reversed. An analogous situation was seen in the developing rat and chick brain, where TR␣1 and TR␣2 appear very early before thyroid hormone is available (35), whereas TR␤1 expression correlates with the onset of endogenous T3 production (36). The relatively low level of TR protein expression in hOb cells is consistent with cytokine secretion data. These data show that in contrast to MG63 and hBMS cells, hOb are poorly responsive to hormones such as T3 (17), PTH, or 1,25-(OH)2D3 (37). Moreover, IL-6, normally a powerful stimulator of bone resorption, has no effect in hOb (38). These results argue against a major role for these cells in osteotropic hormone-induced bone resorption. In contrast,

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

912

J Clin Endocrinol Metab, February 2002, 87(2):906 –914

Siddiqi et al. • TR Expression in Bone

FIG. 4. Effect of T3 1 nM on luciferase activity in MG63 (A), SaOs-2 (B), hBMS (C) and hOb (D) cells. Cells were either untransfected (䡺) or transfected with the control TKLUC either alone (vertical lines) or in the presence of TRE (diagonal lines) or TRE and RSVerbA␤ (f). After 24 h culture in charcoal-stripped medium, cells were transfected in the presence or absence of T3. All results are relative to luciferase activity of TKLUC (6101.7 ⫾ 3536.6 light units). Each bar represents the mean ⫾ SEM of four experiments performed in duplicate. *, P ⬍ 0.05; **, P ⬍ 0.001 (vs. TKLUC).

hBMS cells express TR and are T3 responsive, and our data would suggest that they, rather than hOb, represent the site of T3 action. A relative predominance of TR␣1 was also apparent in MG63 and hBMS, consistent with skeletal defects seen in TR␣⫺/⫺ knockout mice. These mice show growth arrest and delayed bone maturation with disrupted endochondral bone formation associated with biochemical hypothyroidism (39). In contrast, TR␤⫺/⫺ mice have little disruption of skeletal development (40), and double knockout of both TR␣ and TR␤ genes (TR␣⫺/⫺TR␤⫺/⫺) fails to worsen the TR␣⫺/⫺ phenotype (41). Thus, TR␣ seems essential for bone development and function, whereas TR␤ appears not to be as critical. However, the skeletal abnormalities of TR␣⫺/⫺ knockout mice are reversible with T3 replacement, demonstrating firstly that the growth disruption results from biochemical hypothyroidism and secondly that TR␤ can be functional and compensate for TR␣ if required (39). In the light of this, the predominance of TR␣ in the highly responsive MG63 and hBMS cells may be particularly relevant. We also demonstrated for the first time evidence for TR␤2 protein expression in human osteoblasts. TR␤2 expression in rodent bone was recently investigated (12), but no TR␤2 mRNA was demonstrated. Their negative and our positive

results may be related to discrepancies in the mRNA/protein ratios of TRs in bone, which are well described (42, 43), particularly for ␤2. For example, in rodent liver, kidney, and heart, despite only trace detectable mRNA, TR␤2 contribution to total nuclear binding is up to 20% (44). ␤2 mRNA may be difficult to detect because it is either expressed at very low levels (and efficiently translated) or is unstable and has a high turnover, making the RNA hard to detect. The contribution of the ␤2 isoform to bone physiology remains a subject for future investigations. We used transient transfections assays to determine whether cell-specific differences in TR expression correlated with T3 responsiveness. In MG63 and hBMS cells, which expressed relatively high levels of all isoforms, there was a significant T3-dependent induction of LUC activity, indicating that the endogenous TRs were functional. In hOb there was no such rise seen in LUC activity. We would suppose that the level of TR expression in these cells is too low for endogenous receptor function to be detected by the transfection assays employed. Interestingly, in SaOs-2 cells, which expressed TR protein in equivalent amounts to MG63 and hBMS cells, TRE activation was also absent. The difference in responses between the different cell types may also reflect variable expression of RXR and RAR, both of which are

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

Siddiqi et al. • TR Expression in Bone

heterodimerization partners for T3. Given that there are two TR genes (␣ and ␤) and three RXR and RAR (␣, ␤, and ␥) genes, coexpression of multiple receptor isoforms in a single cell may provide a basis for cell-specific responses. When exogenous RSVerbA␤ was transfected into cells, T3-dependent expression of the LUC gene was significantly enhanced in MG63 and hBMS cells, respectively. This suggests that the abundance of endogenous TRs may modify T3-dependent regulation of genes. In these experiments cells contain multiple copies of the TRE. Several TREs may act cooperatively, as in the case of the estrogen response element (45), and in the presence of excess TREs the availability of TRs may become important. We found significant reversible binding of [125I]T3 in MG63 and hBMS cells. SaOs-2 and hOb had significantly lower specific binding compared with these T3-responsive cells. In contrast, Sato et al. (4) found no difference in the number of T3-binding sites between the T3-responsive ROS 17/2.8-2 and the T3-unresponsive ROS 17/2.8-3 cells, although their conclusions were based on rodent cell lines. We have thus demonstrated cell-specific differences between hOb and hBMS cells with respect to receptor expression and function. From our work, which demonstrates lower cytokine secretion (17) and paucity of TR immunostaining, T3-binding sites, and functioning receptors, it would seem that hOb cells play a lesser role and hBMS cells a greater role than previously envisaged in T3 regulation of bone remodeling. Physiological interpretation and extrapolation to in vivo models is premature, and there are no doubt many other cell-specific differences aside from differential TR isoform expression that may influence T3 responsiveness. Regulatory mechanisms for thyroid hormones are coupled with retinoids. Possible differential expression of these nuclear receptors and/or differences in heterodimerization patterns in hOb vs. hBMS may correlate with the observed differences. Our study does establish a basis for T3 action along the osteoblast lineage and emphasizes the importance of human bone marrow cells in future studies of T3 action on bone.

Acknowledgments We are grateful to Prof. V. K. K. Chatterjee (University of Cambridge, Cambridge, UK) for providing the plasmids TKLUC, TRE-TKLUC, and RSVerbA␤ as well as the antibody TR␤2 and the in vitro translated human TR␤1 and TR␤2; Prof. Myrtle Gordon (Imperial College School of Medicine, London, UK) for her support in establishing the bone marrow cultures; Dr. R. Edwards (Netria, London, UK) for providing [125I]T3; and C. Nott (St. Bartholomew’s and The Royal London School of Medicine) for help with immunocytochemistry. Received July 3, 2001. Accepted November 1, 2001. Address all correspondence and requests for reprints to: Dr. Ayesha Siddiqi, Molecular Endocrinology Laboratory 1.4, First Floor Dominion House, 59-60 Bartholomew Close, London, United Kingdom EC1A 7BE. E-mail: [email protected]. This work was supported by a Clinical Endocrinology Trust Research Fellowship (to A.S.), the Leukemia Research Fund (to J.L.L.), and Medical Research Council Career Establishment Grant G9803002 and Wellcome Trust Project Grant 050570 (to G.R.W.).

J Clin Endocrinol Metab, February 2002, 87(2):906 –914 913

References 1. Franklyn J, Betteridge J, Holder R, Daykin J, Lilley J, Sheppard M 1994 Bone mineral density in thyroxine treated females with or without a previous history of thyrotoxicosis. Clin Endocrinol (Oxf) 41:425– 432 2. Siddiqi A, Burrin JM, Noonan K, James I, Wood DF, Price CP, Monson JP 1997 A longitudinal study of markers of bone turnover in Graves’ disease and their value in predicting bone mineral density. J Clin Endocrinol Metab 82: 753–759 3. Franklyn JA, Maisonnneuve P, Sheppard MC, Betteridge J, Boyle P 2000 Mortality after the treatment of hyperthyroidism with radioactive iodine. N Engl J Med 338:712–718 4. Sato K, Han DC, Fujii Y, Tsushima T, Shizume K 1987 Thyroid hormone stimulates alkaline phosphatase activity in cultured rat osteoblastic cells (ROS 17/2.8) through 3,5,3⬘-triiodo-l-thyronine nuclear receptors. Endocrinology 120:1873–1881 5. Ohishi K, Ishida H, Nagata T, Yamauchi N, Tsurumi C, Nishikawa S, Wakano Y 1994 Thyroid hormone suppresses the differentiation of osteoprogenitor cells to osteoblasts, but enhances functional activities of mature osteoblasts in cultured rat calvaria cells. J Cell Physiol 161:544 –552 6. Allain TJ, Chambers TJ, Flanagan AM, McGregor AM 1992 Tri-iodothyronine stimulates rat osteoclastic bone resorption by an indirect effect. J Endocrinol 133:327–331 7. Britto JM, Fenton AJ, Holloway WR, Nicholson GC 1994 Osteoblasts mediate thyroid hormone stimulation of osteoclastic bone resorption. Endocrinology 134:169 –176 8. Yen PM, Sunday ME, Darling DS, Chin WW 1992 Isoform specific thyroid hormone receptor antibodies detect multiple thyroid hormone receptors in rat and human pituitaries. Endocrinology 130:1539 –1546 9. Kliewer SA, Umesono K, Mangelsdorf DJ, Evans RM 1992 Retinoid X receptor interacts with nuclear receptors in retinoic acid, thyroid hormone and vitamin D3 signalling. Nature 355:446 – 449 10. Williams GR, Bland R, Sheppard MC 1994 Characterization of thyroid hormone (T3) receptors in three osteosarcoma cell lines of distinct osteoblast phenotype: interactions among T3, Vitamin D3 and retinoid signaling. Endocrinology 135:2375–2385 11. Bland R, Sammons RL, Sheppard MC, Williams GR 1997 Thyroid hormone, vitamin D and retinoid receptor expression and signaling in primary cultures of rat osteoblastic and immortalized osteosarcoma cells. J Endocrinol 154:63–74 12. Milne M, Kang M, Cardona G, Quail J, Braverman L, Chin W, Baran D 1999 Expression of multiple thyroid hormone receptor isoforms in rat femoral and vertebral bone and in bone marrow osteogenic cultures. J Cell Biochem 74: 684 – 693 13. Milne M, Kang M, Quail J, Baran D 1998 Thyroid hormone excess increases insulin-like growth factor-1 transcripts in bone marrow cell cultures: divergent effects on vertebral and femoral cell cultures. Endocrinology 139:2527–2534 14. Allain TJ, Yen PM, Flanagan AM, McGregor AM 1996 The isoform specific expression of the T3 receptor in osteoblasts and osteoclasts. Eur J Clin Invest 26:418 – 425 15. Abu EO, Bord S, Horner A, Chatterjee VKK, Compston JE 1997 The expression of thyroid hormone receptors in human bone. Bone 21:137–142 16. Siddiqi A, Monson JP, Wood DF, Besser GM, Burrin JM 1999 Serum cytokines in thyrotoxicosis. J Clin Endocrinol Metab 84:435– 439 17. Siddiqi A, Burrin JM, Wood DF, Monson JP 1998 T3 regulates the production of IL-6 and IL-8 in human bone marrow stromal and osteoblast-like cells. J Endocrinol 157:453– 461 18. Gordon MY, Lewis JL, Marley SB, Grand FH, Goldman JM 1997 Stromal cells negatively regulate primitive haemopoietic progenitor cell activation via a phosphatidylinositol-anchored cell adhesion/signalling mechanism. Br J Haematol 96: 647– 653 19. Robey PG, Termine JD 1985 Human bone cells in vitro. Calcif Tissue Int 37:453– 460 20. Haynesworth SE, Goshima J, Goldberg VM, Caplan AI 1992 Characterisation of cells with osteogenic potential from human marrow. Bone 13:81– 88 21. Rickard DJ, Kassem M, Hefferan TE, Sarkar G, Spelsberg TC, Riggs BL 1996 Isolation and characterisation of osteoblast precursor cells from human bone marrow. J Bone Miner Res 11:312–324 22. Bancroft JD, Stevens A 1996 Theory and practice of histological techniques. London: Churchill Livingstone 23. Maxwell IH, Harrison GS, Wood WM, Maxwell F 1989 A DNA cassette containing a trimerized SV40 polyadenylation signal which effectively blocks spurious plasmid initiated transcription. BioTechniques 7:276 –280 24. Brent GA, Harvey JW, Chen Y, Warne RL, Moore DD, Larsen PR 1989 Mutations of the rat growth hormone promoter which increase and decrease response to thyroid hormone define a consensus thyroid response element. Mol Endocrinol 3:1996 –2004 25. Weinberger C, Thompson CC, Ong ES, Lebo R, Gruol DJ, Evans RM 1986 The c-erbA gene encodes a thyroid hormone receptor. Nature 324:641– 646 26. Mizushima S, Nagata S 1990 pEF-BOS, a powerful mammalian expression vector. Nucleic Acids Res 18:5322–5326 27. Burrin JM, Aylwin SJB, Holdstock JG, Sahye U 1988 Mechanism of action of pituitary adenylate cyclase-activating polypeptide on human glucoprotein

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.

914

28. 29.

30. 31. 32.

33. 34. 35. 36. 37.

J Clin Endocrinol Metab, February 2002, 87(2):906 –914 hormone ␣-subunit transcription in ␣T3-1 gonadotrophs. Endocrinology 139: 1731–1737 Jones KE, Chin WW 1991 Differential regulation of thyroid hormone receptor mRNA levels by thyrotropin releasing hormone. Endocrinology 128:1763–1768 Stevens DA, Hasserjian RP, Robson H, Siebler T, Shalet SM, Williams GR 2000 Thyroid hormones regulate hypertrophic chondrocyte differentiation and expression of parathyroid hormone-related peptide and its receptor during endochondral bone formation. J Bone Miner Res 15:2431–2442 Samuels HH, Tsai JS 1973 Thyroid hormone action in cell culture: demonstration of nuclear receptors in intact cells and isolated nuclei. Proc Natl Acad Sci USA 70:3488 –3492 Falcone M, Miyamoto T, Fierro-Renoy F, Macchia E, DeGroot LJ 1992 Antipeptide polyclonal antibodies specifically recognise each thyroid hormone receptor isoform. Endocrinology 131:2419 –2429 Macchia E, Falcone M, Giorgilli G, Bogazzi F, Antonangeli L, Baccarini S, Fontanini G, Torresani J, DeGroot LJ, Pinchera A 1990 Characterisation of site-specific polyclonal antibodies to c-erb A peptides recognising thyroid hormone receptors ␣1, ␣2 and ␤ and study of tissue distribution of the antigen. Endocrinology 126:3232–3239 Rizzoli R, Poser J, Burgi U 1986 Nuclear thyroid hormone receptors in cultured bone cells. Metabolism 35:71–74 Kreiger NS, Stappenbeck TS, Stern PH 1988 Characterisation of specific TR in bone. J Bone Miner Res 3:473– 478 Mellstrom B, Naranjo JR, Santos A 1991 Independent expression of the ␣ and ␤ c-erb A genes in developing rat brain. Mol Endocrinol 5:1339 –1350 Chatterjee VK, Tata JT 1992 Thyroid hormone receptors and their role in development. Cancer Surv 14:147–167 Littlewood AJ, Russell J, Harvey GR, Hughes DE, Russell GG, Gowen M

Siddiqi et al. • TR Expression in Bone

38. 39.

40.

41.

42. 43. 44. 45.

1991a The modulation of the expression of IL6 and its receptor in human osteoblasts in vitro. Endocrinology 129:1513–1520 Littlewood AJ, Aarden LA, Evans DB, Russell GG, Gowen M 1991b Human osteoblast-like cells do not respond to IL-6. J Bone Miner Res 6:141–148 Fraichard A, Chassande O, Plateroti M, Roux JP, Trouillas J, Dehay C, Legrand C, Gauthier K, Kedinger M, Malaval L, Rousset B, Samarut J 1997 The TR␣ gene encoding a thyroid hormone receptor is essential for post-natal development and thyroid hormone production. EMBO J 16:4412– 4420 Forrest D, Hanebuth E, Smeyne RJ, Everds N, Stewart CL, Wehner JM, Curran T 1996 Recessive resistance to thyroid hormone in mice lacking thyroid hormone receptor ␤: evidence for tissue specific modulation of receptor function. EMBO J 15:3006 –3015 Gauthier K, Chassande O, Plateroti M, Roux JP, Legrand C, Pain B, Rousset B, Weiss R, Trouillas J, Samarut J 1999 Different functions for the thyroid hormone receptors TR␣ and TR␤ in the control of thyroid hormone production and post-natal development. EMBO J 18:623– 631 Chamba A, Neuberger J, Strain A, Hopkins J, Sheppard MC, Franklyn JA 1996 Expression and function of thyroid hormone receptor in normal and chronically diseased human liver. J Clin Endocrinol Metab 81:360 –367 Tagami T, Hirotoshi N, Shigekazu S, Miyoshi Y, Imura 1993 Estimation of the protein content of thyroid hormone receptor ␣1 and ␤1 in rat tissues by Western blotting. Endocrinology 132:275–279 Schwartz HL, Lazar MA, Oppenheimer JH 1994 Widespread distribution of immunoreactive TR␤2 in the nuclei of extrapituitary rat tissues. J Biol Chem 269:24777–24782 Ernst M, Parker MG, Rodan GA 1991 Functional estrogen receptors in osteoblastic cells demonstrated by transfection with a reporter gene containing an estrogen response element. Mol Endocrinol 5:1597–1606

Summer Institute on Aging Research 2002 The National Institute on Aging (NIA) announces the annual Summer Institute on Aging Research, a weeklong workshop for investigators new to aging research, focused on current issues, research methodologies, and funding opportunities. The program will also include consultations on the development of research interests. The 2002 Summer Institute will be held July 27–August 2 in Airlie, Virginia. Support is available for travel and living expenses. Applications are due March 8th. To increase the diversity of participants, minority investigators are strongly encouraged to apply. For additional information and an application form, contact: Office of the Director, Office of Special Populations, National Institute on Aging, National Institutes of Health, Building 31, Room 5C-35, 31 Center Drive MSC-2292, Bethesda, Maryland 20892-2292. Telephone: (301) 496-0756; Fax: (301) 496-2525; E-mail: [email protected]; or see the What’s New section of the NIA Web page at http://www.nih.gov/nia.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 January 2017. at 05:40 For personal use only. No other uses without permission. . All rights reserved.