FIBROBLASTS HAVE IMPORTANT functions in the orchestration

Transcription

1 /02/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 87(1): Printed in U.S.A. Copyright 2002 by The Endocrine Society Orbital Fibroblast Heterogeneity May Determine the Clinical Presentation of Thyroid- Associated Ophthalmopathy TERRY J. SMITH, LAURA KOUMAS, ANNEMARIE GAGNON, ANDREA BELL, GREGORY D. SEMPOWSKI, RICHARD P. PHIPPS, AND ALEXANDER SORISKY Division of Molecular Medicine (T.J.S.), Harbor-UCLA Medical Center, Torrance, California 90502, and School of Medicine, University of California, Los Angeles, Los Angeles, California 90095; Cancer Center and Department of Microbiology and Immunology (L.K., G.D.S., R.P.P.), University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; and Ottawa Health Research Institute and Departments of Medicine, Biochemistry, Microbiology, and Immunology (A.G., A.B., A.S.), University of Ottawa, Ottawa, Canada K1Y 4E9 Thyroid-associated ophthalmopathy, a process in which the orbital tissues become inflamed and are remodeled, occurs with a variable presentation. In some patients, eye muscle enlargement predominates. In others, the connective/adipose tissue enlargement appears the more significant problem. Orbital fibroblasts exhibit heterogeneous phenotypes in culture. Here we report that fibroblasts derived from the connective/ adipose tissue depot are distinct from those investing the extraocular muscles. Connective tissue fibroblasts represent a bimodal population of cells with regard to the surface display of the glycoprotein, Thy-1. Perimysial fibroblasts in contrast express Thy-1 uniformly. In that regard, they resemble those from the skin. When subjected to a newly defined set of culture conditions, adipocyte differentiation occurs in up to 43% of FIBROBLASTS HAVE IMPORTANT functions in the orchestration of tissue remodeling and participate in the recruitment of lymphocytes and other bone marrow-derived cells to sites of inflammation (1, 2). Fibroblasts derived from the connective/adipose tissues of the human orbit differ from those of other anatomic regions with regard to the display of surface receptors (3), gangliosides (4), and CD40 (5) as well as the synthesis and regulation of glycosaminoglycans (6, 7) and PGE 2 (8 10). They exhibit particularly robust responses to proinflammatory cytokines (9), endothelin (3), CD154 (5, 11), and prostanoids (12 14). The distinct phenotype exhibited by orbital fibroblasts we believe may underlie the susceptibility of the orbit to manifestations of Graves disease, known as thyroid-associated ophthalmopathy (TAO). TAO involves the infiltration of the orbital tissues with T lymphocytes and the disordered accumulation of hyaluronan, a nonsulfated glycosaminoglycan, often in the setting of intense inflammation (15 17). Another important aspect of the phenotype exhibited by orbital fibroblasts is their newly recognized potential to differentiate in culture into mature adipocytes (18). The fraction of fibroblasts undergoing adipogenesis has been small, in the range of 5 10%. Orbital fibroblasts from the connective/adipose depots are heterogeneous with regard to the expression and surface Abbreviations: cpgi 2, Carbaprostacyclin; HRP, horseradish peroxidase; IBMX, isobutylmethylxanthone; RT, room temperature; TAO, thyroid-associated ophthalmopathy; TSHr, TSH receptor. the cells. All adipocytes examined failed to display Thy-1. Fibroblasts derived from perimysium and dermis uniformly do not differentiate into adipocytes when incubated under identical culture conditions. Both Thy-1 and Thy-1 connective tissue fibroblasts express the adipogenic trigger, peroxisome proliferator activator, suggesting that differences in the potential for differentiation may reside with phenotypic attributes downstream from this receptor/adipogenic transcription factor. These observations enhance our understanding of orbital adipogenesis and define previously unrecognized differences between fibroblasts from the extraocular muscle and connective tissue. (J Clin Endocrinol Metab 87: , 2002) display of Thy-1 (19), a glycoprotein found on a variety of cell types (20). Although the function of Thy-1 as a receptor is unknown, a thoughtful investigation into the functional consequences of segregating fibroblasts on the basis of Thy-1 display has been undertaken. In murine lung fibroblasts, Phipps et al. (21) reported that Thy-1 cells express IL-1, but Thy-1 failed to express either IL-1 or IL-1. Moreover, the expression of immunologically relevant molecules such as HLA-DR class II appears to differ in the two subpopulations. The full scope of cellular differences imposed by the bimodal distribution of Thy-1 on orbital fibroblasts is not yet understood but may relate to discrete functions of the two populations, as has been proposed previously (22). Some patients with TAO present with muscle enlargement as the dominant feature of their orbital disease. The muscles of other patients appear normal in volume, but some of these individuals present with evidence of enlargement of the posterior orbital connective/adipose tissue depot and develop optic neuropathy (23). Although the connective tissuederived fibroblasts of the orbit have been studied extensively by several groups of investigators, those from the extraocular muscles have received little attention. Kiljanski et al. (24) have long focused their efforts on defining the primary importance of the muscle fibers in the pathogenesis of TAO. But most of the evidence generated to date would support the notion that the muscle fibers are spared until very late in the disease process (16) when they may become the casualty of 385

2 386 J Clin Endocrinol Metab, January 2002, 87(1): Smith et al. Thyroid-Associated Ophthalmopathy fibrosis. Thus, most authorities now attribute the muscle dysfunction associated with TAO to the misdeeds of perimysial fibroblasts. To our knowledge, few studies have appeared comparing these fibroblasts with those from the connective/adipose tissue of the orbit. Potential differences in the two populations of fibroblasts, such as their adipogenic potential, could underlie the differences among patients with regard to the anatomic distribution of orbital disease manifestations. In this article, we report that perimysial fibroblasts derived from the extraocular muscles represent a population of cells that are distinct from those originating in the adipose/connective tissue depot of the orbit. These cells are homogeneous with regard to the surface display of Thy-1 and, unlike their counterparts from connective tissue, do not appear to possess in vitro the capacity for differentiation into adipocytes. Adipogenic differentiation in the connective tissue derived fibroblasts occurs in nearly half of the cells when the cultures are incubated in a medium containing the PPAR agonist, rosiglitazone. These current findings have direct implications with regard to the distribution of disease manifestations within the orbital contents associated with TAO. They more precisely define the highly specialized phenotypes of fibroblasts investing orbital tissues. Materials Materials and Methods Anti-Thy-1 antibody (F ) was kindly provided by Dr. J. W. Fabre, University of London (UK). Dexamethasone was from Steraloids (Wilton, NH). Carbaprostacyclin (cpgi 2 ) was from Cayman Chemical Co. (Ann Arbor, MI). Rosiglitazone was from Zen-Bio (Research Triangle Park, NC). Other reagents were of the highest purity commercially available. Anti-PPAR antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) or Calbiochem (San Diego, CA). Cell culture Human orbital and dermal fibroblasts were cultivated as reported previously (25). Tissue explants were obtained from individuals undergoing surgical decompression for severe TAO or orbital surgery for some other, noninflammatory problem. Dermal tissues were obtained from punch biopsies of normal-appearing skin, following informed consent. These activities have been approved by the institutional review boards of Albany Medical College and Harbor-UCLA Medical Center. Some of the fibroblast strains were generously provided by Dr. R. S. Bahn (Mayo Clinic, Rochester, MN). The array of culture strains examined in these studies included fibroblasts from patients with inactive orbital disease. A total of 10 different strains were examined in these studies. Donors were euthyroid at the time of culture harvest. Explants were allowed to attach to the bottom of culture plates and were covered with Eagle s medium containing 10% FBS, antibiotics, and glutamine. They were incubated in a 37 C, 5% CO 2, humidified environment. The resulting fibroblast monolayers were passaged serially by gentle treatment with trypsin/edta. Strains were stored in liquid N 2 until needed and were used between the 2nd and 12th passage. They were determined to not express Factor VIII, cytokeratin, or smooth muscle-specific actin. Immunofluorescence and flow cytometric analysis of Thy-1 display by fibroblasts Indirect immunofluorescence was used to determine the levels of Thy-1 expression in individual cells and to screen for its display in various strains of human fibroblasts. Cell-free supernatant prepared from a mouse hybridoma (F ) was used as the primary antihuman Thy-1 antibody. For flow cytometric studies, cultures were disrupted with gentle treatment with trypsin (0.5%, Worthington Biochemical Corp., Freehold, NJ) for 30 sec and suspended in PBS containing BSA (1%) and sodium azide (0.01%). This solution was used as a staining medium. We have documented that trypsin treatment under these circumstances does not cleave Thy-1 from the surface of fibroblasts. Individual samples were incubated with F or an isotype control antibody for 45 min at 4 C. Samples were then washed three times with staining medium and incubated with goat antimouse immunoglobulin conjugated with fluorescein (Organon Teknika Corp., West Chester, AL) for 45 min at 4 C. After three washes, cell pellets were resuspended in 0.5 ml staining medium and subjected to flow cytometric analysis using a Coulter Profile analyzer (Coulter Electronics, Hialeah, FL). Forward and side-scatter gates were set to exclude nonviable cells and 5000 cellular events were collected for each sample analyzed. For staining of Thy-1 in situ, fibroblasts were cultured on chamber slides (Nunc, Naperville, IL). They were stained under identical conditions as those used in the flow cytometric studies. Slides were removed from the chambers, coverslips applied, and cell monolayers visualized with a BX 40 microscope (Olympus Corp. America, Lake Success, NY) fitted with a UPlan-FL (Ph2) X 40 objective. Induction of fibroblast differentiation into adipocytes Orbital fibroblasts between passage 3 and 12 were allowed to proliferate to near-confluence in DMEM containing 10% FBS and antibiotics, usually on 6-well plastic culture arrays (Costar, Cambridge, MA). Some cultures were then shifted to fresh medium and underwent one of two differentiation protocols. Protocol A was conducted essentially as described previously (18). It consisted of shifting cultures to serum-free medium containing DMEM:F12 (1:1, vol/vol) supplemented with 33 mol/liter biotin, 17 mol/liter pantothenate, 10 g/ml transferrin, 0.2 nmol/liter triiodothyronine, 1 mol/liter insulin, 0.2 mol/liter cpgi 2, 10 nmol/liter TSH, and for the first 4 d only, 1 mol/liter dexamethasone and 0.1 mmol/liter isobutylmethylxanthone (IBMX). This protocol was conducted for 2 wk. Protocol B involved shifting cultures to a medium consisting of DMEM:F-10 (1:1) (vol/vol) supplemented with 3% FBS, 100 nmol/liter insulin, 1 mol/liter dexamethasone, and for the first 3 d only, 1 mol/liter rosiglitazone and 0.2 mmol/liter IBMX. Cultures were maintained in this medium for 18 d. Control cultures were prepared in parallel, as described previously (18). The extent of differentiation was assessed morphologically by phase contrast microscopy using a 1-70 microscope (Olympus Corp.) linked to a dual-color charge-coupled device camera (Sony, Tokyo, Japan). The fraction of cells undergoing differentiation was quantified by the methods of Crandall et al. (26) and Sottile and Seuwen (27) with the following modifications. Following fixation in 0.5% paraformaldehyde, cells were incubated for 10 min at room temperature (RT) in a solution containing 0.1 g/ml Nile Red to stain cytoplasmic triacylglycerol. An excitation light of 488 nm generated a forward linear scatter signal depicted along the ordinates, and fluorescence emission was detected at 565 nm shown along the abscissas of the relevant figures. Western blot and immunohistochemistry analysis of PPAR expression For immunoblots, cultures of orbital fibroblast subsets were allowed to proliferate to confluence in standard culture conditions. They were then subjected to differentiation protocol B and the monolayers were solubilized in a buffer containing 15 mmol/liter Tris-HCl (ph 8.3), 15 mmol/liter CHAPS, 1 mmol/liter EDTA, 10 g/ml soybean trypsin inhibitor, and 10 mol/liter phenylmethylsulfonyl fluoride. An aliquot of each sample was used for protein determination using the bicinchoninic acid method (Pierce Chemical Co., Rockford, IL). Equal amounts of protein were loaded onto a 4.5% stacking gel and run on a 10% polyacrylamide gel at 150 V for 80 min. The gel was transferred to Hybond-C nitrocellulose membrane (Amersham Pharmacia Biotech, Piscataway, NJ), blocked with 10% nonfat dry milk and incubated with anti-ppar polyclonal antibody (1:2000, Calbiochem) for 1hatRT. Following washes, the membrane was incubated with goat antirabbit IgG horseradish peroxidase (HRP) secondary antibody (1:2000, Caltag Laboratories, Inc., Burlingame, CA) for 1hatRT.Blots were developed using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech) and exposed to x-ray film (Kodak, Rochester, NY).

3 Smith et al. Thyroid-Associated Ophthalmopathy J Clin Endocrinol Metab, January 2002, 87(1): For immunohistochemical analysis of PPAR expression, monolayers of fibroblasts were stained with anti-ppar (Santa Cruz Biotechnology) or isotype control mouse IgG1 (Caltag Laboratories, Inc.) at 4 g/ml overnight at 4 C. Biotinylated horse antimouse IgG (heavy and light chain, 1:200, Vector Laboratories, Inc., Burlingame, CA) was added as the secondary antibody for 2 h at 4 C.Cells were washed three times and then streptavidin-hrp (1:1000, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) was added as the substrate for1hatrt.samples were visualized by adding an aminoethyl-carbachol chromogen (Zymed Laboratories, Inc., South San Francisco, CA) and cover-slips added using Immu-mount (Shandon, Pittsburgh, PA). Results Perimysial fibroblasts have a similar morphology to that of orbital connective/adipose tissue fibroblasts Orbital connective/adipose tissue-derived fibroblasts proliferate as a monolayer and have a morphology that is very similar to that of perimysial fibroblasts. Both exhibit fusiform and angular shapes with multiple dendritic processes. In general, both types of orbital fibroblast exhibit a rather granular-appearing cytoplasm. Dermal fibroblasts, in contrast, are predominantly angular appearing and are more homogeneous and less granular than those from either orbital source. No differences in cellular appearance or proliferative pattern were observed with regard to whether strains were harvested from diseased or normal tissues. Perimysial fibroblasts uniformly express Thy-1 We have found that orbital fibroblasts from the connective/adipose tissue depots exhibit a bimodal distribution with regard to the expression and surface display of Thy-1. Approximately 50% of these cells, when cultivated in vitro, stain positively for the antigen. Importantly, this proportion remains stable through many population doublings and passage to daughter cultures. The flow cytometric analysis of Thy-1 surface expression shown in Fig. 1D exhibits this bimodal distribution of cells in parent strains from the orbital connective tissue. In contrast, those from the extraocular muscles appear to uniformly display Thy-1. This is the case whether the patient from whom the cultures derived had been diagnosed with TAO (Fig. 1B) or some other condition not affecting the orbit (Fig. 1A). In this regard, perimysial fibroblasts resemble dermal fibroblasts. Pretibial fibroblasts also express Thy-1 uniformly (Fig. 1C). A number of fibroblast strains from the connective tissue and perimysium were assessed by flow cytometry for Thy-1 expression and the results are summarized in Table 1. Those derived from muscle are uniformly homogeneous. Moreover, treatment with glucocorticoids and a wide array of proinflammatory cytokines failed to influence Thy-1 expression in any of the fibroblasts thus far examined (data not shown). Connective/adipose tissue fibroblasts from the orbit can undergo adipocyte differentiation in vitro. Perimysial fibroblasts do not exhibit this capacity. A subpopulation of connective/adipose tissue-derived orbital fibroblasts can be induced to differentiate in vitro into adipocytes. The fraction of cells undergoing such conversion is small under conditions we first reported (about 5 10% after 18 d) (18). Using protocol A, monolayers were incubated FIG. 1. Flow cytometric analysis of surface Thy-1 expression in human fibroblasts. Cultures were derived from normal perimysial (A), TAO perimysial (B), normal pretibial (C), and TAO orbital connective tissue (D). Cultures were analyzed as described in Materials and Methods. The frequency of Thy-1 positivity is indicated in each panel. Single-cell suspensions were stained with anti-thy-1 mab (F ) followed by fluorescein isothiocyanate-labeled goat antimouse Ig secondary antibody (solid peaks). Background staining was assessed with an isotype-matched control primary antibody, followed by the same secondary Ab (unshaded peak). TABLE 1. Flow cytometric analysis of several human fibroblast strains for the surface expression of Thy-1 Fibroblast source Percent Thy-1 Percent Thy-1 Graves connective/adipose orbital tissue Graves connective/adipose orbital tissue Dermal Normal perimysium Normal perimysium Graves perimysium Graves perimysium in medium containing insulin, cpgi 2, dexamethasone, triiodothyronine, and IBMX. As the photomicrographs contained in Fig. 2 indicate, cells undergoing adipogenic differentiation under these conditions uniformly failed to exhibit Thy-1 expression as determined by immunofluorescence. With regard to their morphology, they take on large cytoplasmic inclusions that stain positively with Oil Red O and represent accumulations of triglycerides. It would appear that a subset of orbital connective/adipose tissue fibroblasts not displaying Thy-1 in culture represent preadipocytes. Because of the relatively poor yield in the fractional differentiation of orbital fibroblasts from the connective tissue using protocol A, we modified the conditions under which the studies were conducted. Thus, we examined a number of additional compounds for their ability to enhance adipogenesis when added to the differentiation medium. When protocol B was employed and the PPAR ligand, rosiglitazone, was added to the differentiation mixture, the yield of cells undergoing adipogenesis was increased dramatically (Fig. 3). These cells accumulate droplets of triglyceride that stain with Oil Red O. Unlike fibroblasts from orbital connective

4 388 J Clin Endocrinol Metab, January 2002, 87(1): Smith et al. Thyroid-Associated Ophthalmopathy FIG. 2. Orbital adipocytes do not express Thy-1 in culture. Orbital fibroblasts, in this case from an individual with severe TAO, were allowed to proliferate and then were incubated under standard culture conditions (A and B) or in a differentiation medium using protocol A (C and D), as described in Materials and Methods. They were fixed and stained with anti-thy-1 mab (F ), followed by incubation with a fluorescein isothiocyanate-labeled goat antimouse secondary Ig. The same field was photographed in phase contrast (A and C) and under fluorescence (B and D). Final magnification, 132. tissue, those from the perimysium fail to accumulate triglyceride droplets and continue to exhibit their native morphologies. In this regard, too, they resemble those from the skin. Dermal fibroblasts, derived from the pretibium and abdominal wall, are also refractory to the effects imposed by the differentiation medium. It would appear from our studies to date that the level of disease activity and the duration of TAO when the tissues were obtained play little or no role in determining the adipogenic potential of the fibroblast strains. Moreover, cultures initiated from normal and TAO tissues behaved in a very similar manner with regard to their response to the differentiation protocols. Because the differentiation response under the modified conditions using protocol B was considerably more robust than that we had observed previously, we next applied a newly described technique based on flow cytometry to quantify the extent of adipogenesis. Orbital connective tissue fibroblasts were subjected to these culture conditions and the fraction of cells differentiating was determined. As the flow diagrams displayed in Fig. 4 indicate, in a typical study, nearly 50% of the fibroblasts from the connective/adipose tissue depot differentiate after 18 d. In contrast, there was no evidence of adipocyte differentiation in dermal fibroblast cultures subjected to the same protocol (Fig. 5). Orbital fibroblasts express high levels of PPAR PPAR is a key molecular trigger for the differentiation of preadipocytes into mature adipocytes (28). We thus determined whether orbital fibroblasts express this receptor/adipogenic transcription factor. As the immunofluorescence photomicrographs in Fig. 6 demonstrate, the protein is easily detectable in both Thy-1 and Thy-1 connective tissue fibroblasts under culture conditions not favoring the differentiation of the cells. The cellular distribution of PPAR staining is similar in both Thy-1 and Thy-1 subsets and contains a prominent nuclear component. These observations were then quantified by subjecting cell lysates to Western blot analysis for PPAR. The receptor protein is abundantly expressed in cells from both subsets and the levels are very similar (Fig. 7). This result suggests that the inability of the Thy-1 subset to undergo adipogenic differentiation following exposure to a PPAR ligand is a consequence of phenotypic differences downstream from receptor occu-

5 Smith et al. Thyroid-Associated Ophthalmopathy J Clin Endocrinol Metab, January 2002, 87(1): FIG. 3. Adipose/connective tissue derived fibroblasts exhibit a robust differentiation into mature adipocytes when subjected to culture conditions that include rosiglitazone. Fibroblasts, in this case from normal orbital connective tissue, were allowed to proliferate to confluence in medium supplemented with 10% FBS. Controls were continued in this growth medium (A) and others were shifted for 2 wk to the differentiation medium described in protocol B (B). The medium contained 3% FBS, 100 nmol/liter insulin, 1 mol/liter dexamethasone, and for the first 3 d only 0.2 mmol/liter IBMX and 1 mol/liter rosiglitazone. Cultures were photographed as described in Materials and Methods. FIG. 4. Flow-cytometric quantitation of orbital connective/adipose tissuederived fibroblast differentiation to adipocytes in vitro. Orbital fibroblasts, in this case from an individual with normal orbital tissue, were allowed to proliferate in growth medium containing 10% FBS, and then some were shifted to the differentiation medium (protocol B) described in Materials and Methods. Monolayers were disrupted and subjected to flow cytometric analysis. pancy. When parent strains of the fibroblasts are subjected to the differentiation protocol B, the levels of PPAR were slightly enhanced (50%, data not shown), consistent with the findings in other differentiating cell types (29). Discussion Orbital fibroblasts exhibit a remarkable phenotype in culture and retain those attributes despite many population doublings (30). We hypothesize that it is the excessive responses observed in these fibroblasts that underlie the siterestricted involvement of orbital tissues in TAO. With regard to fibroblasts derived from the extraocular muscles, their behavior may ultimately explain the limitations in eye movement observed late in the course of TAO. It is at that stage that fibrosis becomes an important concern. Thus, the fibroblast, by virtue of its vast biosynthetic repertoire, remains a prime suspect in the mediation of end-stage tissue remodeling including the deposition of collagen in the muscle. It is noteworthy that Smith and Parikh (31) have demonstrated an activation of connective tissue-derived orbital fibroblasts when cocultured with HMC-1 mast cells involving IL-4. Mast cells have been demonstrated in the cellular infiltrate associated with TAO (16) and have a putative role in the development of fibrosis, as does IL-4. Disease manifestations within the orbit appear to be distributed differently in old and young patients. In those over 60 yr, severe eye muscle involvement in the absence of adipose tissue expansion is often found. In contrast, patients

6 390 J Clin Endocrinol Metab, January 2002, 87(1): Smith et al. Thyroid-Associated Ophthalmopathy FIG. 5. Flow-cytometric analysis of adipocyte differentiation in dermal fibroblast strains. The cultures, in this case from normal skin, were subjected to normal growth medium or to the differentiation medium using protocol B as described in Materials and Methods. FIG. 6. Immunohistochemical analysis of PPAR expression in Thy-1 and Thy-1 orbital fibroblasts discloses that both express the receptor protein. Cells were seeded in chamber slides and allowed to attach over 24 h. Antihuman PPAR antibody (4 g/ml) was used as the primary and biotinylated horse antimouse IgG as the secondary antibody, as described in Materials and Methods. Streptavidin-HRP was added as the substrate and cells were visualized using an AEC substrate kit. Left panels show results with an isotype control primary antibody and right panels show results with anti-ppar. younger than 40 yr often present with relatively little inflammation or involvement of the muscle but substantial adipose/connective tissue expansion (23). These differences in the distribution of orbital tissue involvement could be related to the distinct phenotypes of muscle and connective tissue-derived fibroblasts. It is possible that the adipogenic potential of orbital fibroblasts diminishes as a consequence of aging, as has been demonstrated in abdominal preadipocytes (32). Clearly, greater insight into the cellular divergence existing in the two fibroblast populations will be necessary before this hypothesis can be tested adequately. The current findings raise important questions concerning the role of fibroblasts as participants in both physiological and pathological states. Previously it was thought that these cells were relatively inert and played limited reactive roles. Instead, they were viewed in the narrow context of producing extracellular matrix molecules such as collagen. We now know that fibroblasts are sentinel cells, capable of reacting rapidly to a wide array of molecular cues and orchestrating complex interactions with immunocompetent cells (1). Fibroblasts are the source of many proinflammatory molecules such as cytokines and growth factors and release chemoattractant molecules involved in the recruitment and activation of lymphocytes and other bone marrow-derived cells (1, 2). The extent of phenotypic divergence exhibited by fibroblasts from the adipose and muscle depots of the orbit had not been appreciated previously. It will be necessary to assess the capacity of these different cell populations to elaborate collagen and other profibrotic molecules and to define cell type-specific signaling pathways that might be interrupted.

7 Smith et al. Thyroid-Associated Ophthalmopathy J Clin Endocrinol Metab, January 2002, 87(1): FIG. 7. Western blot analysis of PPAR expression in orbital fibroblasts. Thy-1 and Thy-1 orbital fibroblasts, in this case from a patient with TAO, were lysed and 10 g cellular protein subjected to SDS-PAGE followed by Western blot analysis for PPAR as described in Materials and Methods. A lung fibroblast line, L828, was used as a positive control. These could represent important therapeutic targets for modifying the course of TAO. Of substantial relevance to this discussion are observations made concerning the extraocular muscles of patients with TAO. Some muscles exhibit, with magnetic resonance and computer tomography imaging, areas that are consistent with fatty infiltration (33). These findings have led to speculation that under certain conditions, endomysial and perimysial fibroblasts might differentiate in situ into adipocytes. It is possible that in these individuals, some contribution to the muscle-associated fibroblast population might derive from the Thy-1 subset. Orbital fibroblasts, when activated with cytokines, synthesize substantial levels of hyaluronan, a nonsulfated glycosaminoglycan (7, 34). This enhanced synthesis can be attributed to a coordinated induction of the uridine 5 diphosphate-glucose dehydrogenase (35) and hyaluronan synthase genes (36). Hyaluronan accumulation is an important component in the volume expansion seen in TAO. This is thought to result from the rheologic properties of these complex sugars, which occupy extraordinary volumes when hydrated. From the findings we report here, it would appear that the orbital expansion could result from a greater adipose tissue mass as well. Our current findings differ somewhat from those reported recently by Jeney et al. (37). Those investigators found widespread, spontaneous adipogenic changes in fibroblasts from the skin and orbit that were allowed to propagate on glass. Their observations are interesting and represent potentially important findings. They may underscore the critical physical/chemical nature of the substratum in defining, at least in part, the behavior of fibroblast lineage cells. On the other hand, that study failed to provide important negative controls demonstrating examples of cells not undergoing adipogenic differentiation. Just how their findings can be reconciled with our results concerning the specialized differentiation of orbital connective tissue-derived fibroblasts will require further investigation. We now report the important contribution of the PPAR agonist, rosiglitazone, to the adipogenesis of orbital fibroblasts. PPAR functions as a transcriptional factor that has been implicated in adipogenic differentiation (38). This receptor recognizes a number of ligands that elicit such cellular changes through its ability to activate a constellation of key downstream genes involved in adipogenesis and insulin sensitization (39). PPAR possesses biological properties that are distinct from those of PPAR and PPAR, the latter of which does not exhibit adipogenic activity (40). A possible endogenous ligand for PPAR is 15-deoxy- 12, 14-PGJ 2 (38), although considerable debate remains concerning the relationship existing between that prostanoid and PPAR. Studies involving the interruption of the PPAR gene strongly suggest the central importance of this receptor in the development of fat as well as heart and placenta (41). This contention was strengthened by the demonstration that PPAR overexpression drives adipogenic differentiation in cells not ordinarily considered preadipocytes (42). With regard to our current findings, a concern arises regarding the potential worsening of fat expansion in TAO, should patients ingest PPAR agonists as therapy for concomitant diabetes mellitus. Conversely, the PPAR antagonists currently under development may prove of therapeutic value in modifying the adipogenic potential of orbital tissue in TAO. Such a strategy might limit expansion of the fat compartment in TAO. Recent reports have appeared suggesting that adipogenesis of orbital fibroblasts results in increased expression of the TSH receptor (TSHr) (43, 44). The authors of these studies have concluded that differentiation also results in a substantial enhancement of camp generation provoked by the ligation of the TSHr with TSH. The second of these reports documented that a majority of cells undergoing differentiation display the receptor (44). The enhanced TSHr expression they observed in orbital fibroblasts is consistent with earlier studies in 3T3-L1 cells (45) and rat adipocytes (46, 47). These are potentially important observations. They may have identified an important consequence of adipocyte differentiation, namely that the abundance of a putative orbital selfantigen is increased. It is also important to note that TSHr expression has been documented in a wide array of fibroblast lineage cells (48, 49), including those from anatomic regions ordinarily not manifesting Graves disease. Whether the levels of TSHr might be enhanced in those depots following adipogenesis as well remains to be determined. The observations we report here extend substantially our earlier work in which we introduced the concept of heterogeneity exhibited by fibroblasts from the orbital connective tissue (19). The current findings imply important functional consequences from the cellular diversity and provide evidence suggesting divergent biological roles for Thy-1 and Thy-1 fibroblast subsets. It would appear that Thy-1 fibroblasts represent preadipocytes that, under certain conditions, can be induced to differentiate. When prompted by the constellation of growth factors and cytokines that are expressed as a consequence of TAO, these cells may undergo differentiation into adipocytes and thus contribute to the increased tissue volume associated with the disease. The Thy-1 fibroblast must now be singled out as a potential target for disease-modifying therapeutics designed to retard adipogenesis in TAO. Acknowledgments Received June 14, Accepted October 1, Address all correspondence and requests for reprints to: Terry J. Smith, M.D., Division of Molecular Medicine, Building C-2, Harbor- UCLA Medical Center, 1124 West Carson Street, Torrance, California tjsmith@ucla.edu.

8 392 J Clin Endocrinol Metab, January 2002, 87(1): Smith et al. Thyroid-Associated Ophthalmopathy Present address for G.D.S.: Department of Medicine, Duke University School of Medicine, Durham, North Carolina This work was supported in part by NIH Grants EY08976, EY11708, DE11390, H150002, and ES01247; a Merit Review award from Research Service of the Department of Veterans Affairs; Heart and Stroke Foundation of Canada Grant T A.G. was supported by a CDA postdoctoral research award. A.B. was supported by a CIHR/HSFC doctoral research award. A.S. is a Career Investigator of the Heart and Stroke Foundation of Ontario. References 1. Smith RS, Smith TJ, Blieden TM, Phipps RP 1997 Fibroblasts as sentinel cells: synthesis of chemokines and regulation of inflammation. Am J Pathol 151: Sciaky D, Brazer W, Center DM, Cruikshank WW, Smith TJ 2000 Cultured human fibroblasts express constitutive IL-16 mrna: cytokine induction of active IL-16 protein synthesis through a caspase-3-dependent mechanism. J Immunol 164: Smith TJ, Kottke RJ, Lum H, Andersen TT 1993 Human orbital fibroblasts in culture bind and respond to endothelin. Am J Physiol 265:C138 C Berenson CS, Smith TJ 1995 Human orbital fibroblasts in culture express ganglioside profiles distinct from those in dermal fibroblasts. J Clin Endocrinol Metab 80: Sempowski GD, Rozenblit J, Smith TJ, Phipps RP 1998 Human orbital fibroblasts are activated through CD40 to induce proinflammatory cytokine production. Am J Physiol 274:C707 C Smith TJ, Bahn RS, Gorman CS, Cheavens M 1991 Stimulation of glycosaminoglycan accumulation by interferon gamma in cultured human retroocular fibroblasts. J Clin Endocrinol Metab 72: Smith TJ, Wang H-S, Evans CH 1995 Leukoregulin is a potent inducer of hyaluronan synthesis in cultured human orbital fibroblasts. Am J Physiol 268:C382 C Wang H-S, Cao HJ, Winn VD, Rezanka LJ, Frobert Y, Evans CH, Sciaky D, Young DA, Smith TJ 1996 Leukoregulin induction of prostaglandin-endoperoxide H synthase-2 in human orbital fibroblasts: an in vitro model for connective tissue inflammation. J Biol Chem 271: Young DA, Evans CH, Smith TJ 1998 Leukoregulin induction of protein expression in human orbital fibroblasts: evidence for anatomical site-restricted cytokine-target cell interactions. Proc Natl Acad Sci USA 95: Cao HJ, Smith TJ 1999 Leukoregulin upregulation of prostaglandin endoperoxide H synthase-2 expression in human orbital fibroblasts. Am J Physiol 277:C1075 C Cao HJ, Wang H-S, Zhang Y, Lin H-Y, Phipps RP, Smith TJ 1998 Activation of human orbital fibroblasts through CD40 engagement results in a dramatic induction of hyaluronan synthesis and prostaglandin endoperoxide H synthase-2 expression. J Biol Chem 273: Smith TJ, Wang HS, Hogg MG, Henrikson RC, Keese CR, Giaever I 1994 Prostaglandin E 2 elicits a morphological change in cultured orbital fibroblasts from patients with Graves ophthalmopathy. Proc Natl Acad Sci USA 91: Wang HS, Keese CR, Giaever I, Smith TJ 1995 Prostaglandin E 2 alters human orbital fibroblast shape through a mechanism involving the generation of cyclic adenosine monophosphate. J Clin Endocrinol Metab 80: Reddy L, Wang H-S, Keese CR, Giaever I, Smith TJ 1998 Assessment of rapid morphological changes associated with elevated camp levels in human orbital fibroblasts. Exp Cell Res 245: Smith TJ, Bahn RS, Gorman CA 1989 Connective tissue, glycosaminoglycans, and diseases of the thyroid. Endocr Rev 10: Hufnagel TJ, Hickey WF, Cobbs WH, Jakobiec FA, Iwamoto T, Eagle RC 1984 Immunohistochemical and ultrastructural studies on the exenterated orbital tissues of a patient with Graves disease. Ophthalmology 91: Weetman AP, Cohen S, Gatter KC, Fells P, Shine B 1989 Immunohistochemical analysis of the retrobulbar tissues in Graves ophthalmopathy. Clin Exp Immunol 75: Sorisky A, Pardasani D, Gagnon A, Smith TJ 1996 Evidence of adipocyte differentiation in human orbital fibroblasts in primary culture. J Clin Endocrinol Metab 81: Smith TJ, Sempowski GD, Wang H-S, Del Vecchio PJ, Lippe SD, Phipps RP 1995 Evidence for cellular heterogeneity in primary cultures of human orbital fibroblasts. J Clin Endocrinol Metab 80: Morris RJ, Ritter MA 1980 Association of Thy-1 cell surface differentiation antigen with certain connective tissues in vivo. Cell Tissue Res 206: Phipps RP, Baecher C, Frelinger JG, Penney DP, Keng P, Brown D 1990 Differential expression of interleukin 1 by Thy-1 and Thy-1 lung fibroblast subpopulations: enhancement of interleukin 1 production by tumor necrosis factor-. Eur J Immunol 20: Fries KM, Blieden T, Looney RJ, Sempowski GD, Silvera MR, Willis RA, Phipps RP 1994 Evidence of fibroblast heterogeneity and the role of fibroblast subpopulations in fibrosis. Clin Immunol Immunopathol 72: Anderson RL, Tweeten JP, Patrinely JR, Garland PE, Thiese SM 1989 Dysthyroid optic neuropathy without extraocular muscle involvement. Ophthalmic Surg 20: Kiljanski JI, Nebes V, Wall JR 1995 The ocular muscle cell is a target of the immune system in endocrine ophthalmopathy. Int Arch Allergy Immunol 106: Smith TJ 1987 n-butyrate inhibition of hyaluronate synthesis in cultured human fibroblasts. J Clin Invest 79: Crandall DL, Armellino DC, Busler DE, McHendry-Rinde B, Kral JG 1999 Angiotensin II receptors in human preadipocytes: role in cell cycle regulation. Endocrinology 140: Sottile V, Seuwen K 2000 Bone morphogenetic protein-2 stimulates adipogenic differentiation of mesenchymal precursor cells in synergy with BRL (rosiglitazone). FEBS Lett 475: Tontonoz P, Nagy L, Alvarez JG, Thomazy VA, Evans RM 1998 PPAR gamma promotes monocyte/macrophage differentiation and uptake of oxidized LDL. Cell 93: Tontonoz P, Hu E, Spiegelman BM 1994 Stimulation of adipogenesis in fibroblasts by PPAR 2, a lipid-activated transcription factor. Cell 79: Smith TJ, Bahn RS, Gorman CA 1989 Hormonal regulation of hyaluronate synthesis in cultured human fibroblasts: evidence for differences between retroocular and dermal fibroblasts. J Clin Endocrinol Metab 69: Smith TJ, Parikh SJ 1999 HMC-1 mast cells activate human orbital fibroblasts in coculture: evidence for up-regulation of prostaglandin E 2 and hyaluronan synthesis. Endocrinology 140: Hauner H, Entenmann G, Wabitsch M, Gaillard D, Ailhaud G, Negrel R, Pfeiffer EF 1989 Promoting effect of glucocorticoids on the differentiation of human adipocyte precursor cells cultured in a chemically defined medium. J Clin Invest 84: Kemp EG, Rootman J 1989 Lipid deposition within the extra-ocular muscles of a patient with dysthyroid ophthalmopathy. Orbit 8: Smith TJ, Bahn RS, Gorman CA, Cheavens M 1991 Stimulation of glycosaminoglycan accumulation by interferon gamma in cultured human retroocular fibroblasts. J Clin Endocrinol Metab 72: Spicer AP, Kaback LA, Smith TJ, Seldin MF 1998 Molecular cloning and characterization of the human and mouse UDP-glucose dehydrogenase genes. J Biol Chem 273: Kaback LA, Smith TJ 1999 Expression of hyaluronan synthase messenger ribonucleic acids and their induction by interleukin-1 in human orbital fibroblasts: potential insight into the molecular pathogenesis of thyroidassociated ophthalmopathy. J Clin Endocrinol Metab 84: Jeney F, Bazso-Dombi E, Oravecz K, Szabo J, Nagy IZ 2000 Cytochemical studies on the fibroblast-preadipocyte relationships in cultured fibroblast cell lines. Acta Histochemica 102: Forman BM, Tontonoz P, Chen J, Brun RP, Spiegelman BM, Evans RM deoxy-delta 12, 14-prostaglandin J2 is a ligand for the adipocyte determination factor PPAR-gamma. Cell 83: Miles PD, Barak Y, He W, Evans RM, Olefsky JM 2000 Improved insulin sensitivity in mice heterozygous for PPAR-gamma deficiency. J Clin Invest 105: Brun RP, Tontonoz P, Forman BM, Ellis R, Chen J, Evans RM, Spiegelman BM 1996 Differential activation of adipogenesis by multiple PPAR isoforms. Genes Dev 10: Barak Y, Nelson MC, Ong ES, Jones YZ, Ruiz-Lozano P, Chien KR, Koder A, Evans RM 1999 PPAR gamma is required for placental, cardiac, and adipose tissue development. Mol Cell 4: Rosen ED, Spiegelman BM 2000 Molecular regulation of adipogenesis. Annu Rev Cell Dev Biol 16: Valyasevi RW, Erickson DZ, Harteneck DA, Dutton CM, Heufelder AE, Jyonouchi SC, Bahn RS 1999 Differentiation of human orbital preadipocyte fibroblasts induces expression of functional thyrotropin receptor. J Clin Endocrinol Metab 84: Crisp M, Starkey KJ, Lane C, Ham J, Ludgate M 2000 Adipogenesis in thyroid eye disease. Invest Ophthalmol Vis Sci 41: Haraguchi K, Shimura H, Lin L, Saito T, Endo T, Onaya T 1996 Functional expression of thyrotropin receptor in differentiated 3T3 L1 cells: a possible model cell line of extrathyroidal expression of thyrotropin receptor. Biochem Biophys Res Commun 223: Haraguchi K, Shimura H, Lin L, Endo T, Onaya T1996 Differentiation of rat preadipocytes is accompanied by expression of thyrotropin receptors. Endocrinology 137: Haraguchi K, Shimura H, Kawaguchi A, Ikeda M, Endo T, Onaya T 1999 Effects of thyrotropin on the proliferation and differentiation of cultured rat preadipocytes. Thyroid 9: Bell A, Gagnon A, Grunder L, Parikh SJ, Smith TJ, Sorisky A 2000 Functional TSH receptor in human abdominal preadipocytes and orbital fibroblasts. Am J Physiol 279:C335 C Janson A, Rawet H, Perbeck L, Marcus C 1998 Presence of thyrotropin receptor in infant adipocytes. Pediatr Res 43: