News and Views: At Long Last, an Animal Model of Graves' Orbitopathy

Patients with Graves' disease (GD) are classically hyperthyroid due to excess production of thyroid hormones. They often have a goiter and may experience weight loss, sensitivity to heat, tremor, and nervousness. Some 30% to 60% of these patients also have signs and symptoms of Graves' orbitopathy (GO; also termed Graves' ophthalmopathy or thyroid eye disease). Although the majority experience only mild eye irritation and redness, some 3% to 5% of GO patients suffer from more severe disease that presents variably as painful soft tissue swelling and redness of the eyes and lids, forward protrusion of the globes (proptosis), and/or debilitating double vision. Some patients risk sight loss due to compressive optic neuropathy or breakdown of the cornea. Orbital imaging generally shows extraocular muscle enlargement and, in some patients, increased orbital fat volume. The former is caused by deposition of hydrophilic glycosaminoglycans (GAGs) between the muscle fibers, and the latter develops as a result of de novo adipogenesis within the orbit. Also evident on histologic examination is an infiltration of mononuclear cells including predominately CD4⁺ T cells with occasional populations of CD8⁺ cells, B cells, and macrophages (1).

Although it is well known that the hyperthyroidism of GD is caused by stimulation of the TSH receptor (TSHR) on thyroid follicular cells by autoantibodies directed against the receptor (TSHR antibodies [TRAbs]), the pathogenesis of GO is less clear. The clinical observation that GO is frequently diagnosed concurrent with or shortly after the diagnosis of Graves' hyperthyroidism led early on to the concept that these conditions comprise a syndrome and share a common etiology. It wasn't, however, until after the cloning of TSHR that circulating TRAb could be reliably measured and the possible expression of TSHR within the orbit explored. Studies followed demonstrating that prognosis can be predicted from the TRAb level (2) and activity and severity assessed using a sensitive bioassay that detects IgG stimulation of TSHR in a reporter cell line (3). Although some patients with GO have never been hyperthyroid or may even be hypothyroid, very sensitive TRAb assays can measure circulating antibody in essentially all patients diagnosed with GO (4). The demonstration by several laboratories of TSHR expression in the orbit and specifically on orbital fibroblasts further connected the ocular and thyroidal manifestations of GD. In addition, in response to GD patients' IgGs or a monoclonal stimulatory TRAb, orbital fibroblasts differentiate into mature adipocytes (5) and produce excessive GAGs (6). Furthermore, these processes can be blocked using a small-molecule antagonist of TSHR activation (7). However, although these and other lines of circumstantial evidence supported a central role for TSHR in the pathogenesis of GO, confirmatory in vivo data were not available owing primarily to the lack of a spontaneous or inducible animal model of the disease.

Several animal models of the hyperthyroidism of GD have been developed over the past 2 decades using various vectors to deliver TSHR (8, 9). Although no convincing evidence of GO was reported in any of these models, in most, the orbits were not carefully examined. The earliest animal models involved conventional immunization with human TSHR protein expressed in bacteria or insect cells. Although antibodies reacting with the receptor were induced, the animals did not produce thyroid stimulating antibodies (TSAbs) or develop hyperthyroidism. In contrast, transient TSAb activity was measured in severe combined immunodeficient (SCID) mice engrafted with human Graves' thyroid tissue, with some also receiving patients' lymphocytes or IgGs. However, this technique was cumbersome, owing in part to the need for major histocompatibility complex (MHC) matching of T cell lines and grafts. Other studies introduced the receptor using fibroblasts, B cells, or human embryonic kidney (HEK293) cells stably expressing TSHR or a portion of the receptor. More recently, several groups used genetic immunization requiring in vivo expression of TSHR. In these models, BALB/c or C57BL6 mice are injected multiple times with an adenovirus or plasmid encoding the holoreceptor or TSHR A-subunit. These mice almost invariably produce TSAb with many developing thyroid hypertrophy and hyperthyroidism. The finding that TSHR-knockout mice are particularly susceptible to immunization with the mouse TSHR suggested another approach in which splenocytes from TSHR-knockout mice were transferred into athymic nude mice expressing the endogenous receptor (10). Some of these recipient mice showed transient increases in TSAbs and later developed hypothyroidism due to the production of antibodies capable of blocking TSH stimulation of the thyroid (TSBAbs).

The first animal model with ocular changes suggestive of GO involved splenocyte transfer from BALB/c and NOD mice previously immunized with TSHR DNA or with a TSHR fusion protein (11). Cells were cultured with the fusion protein and then injected into nonimmunized recipients. The animals developed thyroid inflammation and, in the BALB/c mice only, orbital changes were described. These included immune cell infiltration, adipose tissue expansion, and muscle fibers with widely separated ground substance suggestive of GAGs. Unfortunately, the model could not be reproduced in another facility despite attempts by some of the same investigators using the same protocol. This group later determined that most of the orbital changes likely represented artifacts related to the histologic preparation technique (12). More recently, a mouse model using genetic immunization of an activating mutant of TSHR was generated. Although these animals did not develop orbital inflammation or tissue changes, the investigators did describe a novel technique for orbital examination designed to avoid artifacts based on in toto removal of the orbit with perfusion fixing before preparation (13).

Moshkelgosha and colleagues (14) report in this issue of Endocrinology the development of the first true animal model of GO. Female BALB/c mice were immunized with human TSHR A-subunit, the portion of the receptor that initiates or augments immune responses (15). The direct gene transfer into muscle was accomplished by injecting plasmid into both thighs, followed by low-voltage electroporation. Orbital changes, very similar to and as varied as those seen in humans with GO, were apparent in each of the immunized animals. Although orbital congestion with eyelid swelling and redness was seen in some animals, unilateral or bilateral proptosis was apparent in others, either visually or using magnetic resonance imaging. Most mice showed extraocular muscle enlargement, and some demonstrated expansion of the orbital adipose tissues. GAG deposition within the muscles and infiltration of the orbit by CD3⁺ T cells, macrophages, and mast cells was evident. All immunized mice produced measurable TRAbs. However, rather than producing TSAbs, the majority generated persistent TSBAbs and showed thyroid histology compatible with hypothyroidism. Transient TSAbs were produced in 5 animals, and histology indicative of hyperthyroidism without thyroiditis was seen in a few. Of note is that similar ocular changes developed regardless of the TRAb type generated. The study involves the immunization of 22 mice in total, representing 3 repetitions of the experiment with animals killed in each at a different time point after immunization.

A previous study from Banga and colleagues (16) used an experimental protocol very similar to the one used in the current report. In that study, however, the orbital changes did not recapitulate human GO nearly to the same extent, and the TRAb profile was also quite dissimilar to that currently described. The orbits of some mice showed moderate connective tissue fibrosis and deposition of collagen-rich material within the extraocular muscles. Almost half developed TSAbs and hyperplastic thyroid histology with evidence of lymphocytic microinfiltration. The authors indicate that the only apparent difference between the two protocols was the deeper injection of the plasmid given over a larger area of the thigh muscle in the current study. It was postulated that this enhanced technique may have produced greater transfection efficiency resulting in a modified antigenic stimulus. Studies unraveling the differences between the immune responses to TSHR elicited in these two studies will add much to our understanding of the earliest events in disease development, its evolution, and factors impacting the type and degree of orbital remodeling in GO.

The study by Moshkelgosha and colleagues (14) represents the strongest evidence to date that TSHR, and likely its A-subunit, is the primary autoantigen in GO. Chen and colleagues (17) have described TSHR as “the culprit as well as the victim” in Graves' hyperthyroidism. Similarly, the current study reveals TSHR to be the culprit in GO and its role as a victim is supported by numerous cell-based studies. The concept that TSHR might interact in GO with the IGF-1 receptor (IGF-1R) as a partner in crime has been suggested in studies from Smith and colleagues (18). That group has reported elevated levels of IGF-1R in GO orbital tissue and on B and T cells from patients with GD and demonstrated the displacement of IGF-1 from IGF-1R by activating IgGs from patients with GD. GD IgGs also induce the expression of lymphocyte-homing molecules on orbital fibroblasts via the IGF-1R pathway (19). To address the potential role of IGF-1R as an autoantigen in GO, Moshkelgosha and colleagues (14) immunized some animals with IGF-1Rα plasmid instead of TSHR cDNA. They found that although these mice generated high levels of IGF-1Rα antibody, they developed no apparent pathology. Of particular interest, however, is that some animals immunized with TSHR A-subunit alone developed low-titer IGF-1Rα antibodies shortly after immunization. How these secondary antibodies are produced and whether they might impact TSHR signaling in orbital fibroblasts are topics of considerable interest.

No single animal model ever recreates exactly the diverse elements of a human disease, and each different model carries advantages and limitations and highlights particular aspects of the disease. Together they allow for the study of genetic and environmental factors, effector mechanisms, and novel therapies in a manner not possible using humans or in vitro systems. The robust mouse model developed by Moshkelgosha and colleagues (14) and future animal models evolving from it will allow investigators to invent new approaches to the study of GO. These studies will also enrich our understanding of autoimmune diseases in general. Insights to be gained include the delineation of immunologic processes and molecular events at play early in GO development and later as orbital tissue remodeling ensues. In addition, analyses of the evolving immune milieu within the orbit using a systematic technique for orbital examination will be possible. The natural history of the disease can be better understood. Finally, novel approaches to disease prevention and new therapies for established disease can now be studied in vivo. These studies will no doubt facilitate the initiation of randomized clinical trials of novel agents in patients with GO.