Unique Attributes of Orbital Fibroblasts and Global Alterations in IGF-1 Receptor Signaling Could Explain Thyroid-Associated Ophthalmopathy

Abstract

Tissue remodeling associated with thyroid-associated ophthalmopathy (TAO) involves the complex interplay between resident cells (endothelium, vascular smooth muscle, extraocular muscle, and fibroblasts) and those recruited to the orbit, including members of the “professional” immune system. Inflammation early in the disease can later culminate in fibrosis and diminished extraocular muscle motility. TAO remains a poorly understood process, in large part because access to tissues early in the disease is limited and because no robust and complete animal models of Graves' disease have yet been devised. Remaining uncertainty as to the identity of a pathogenic autoantigen(s) that underlies lymphocyte trafficking to the orbit complicates matters. These limitations in our understanding of extrathyroidal Graves' disease have resulted in poorly served patients with severe TAO. Therapies have targeted symptoms rather than the underlying disease processes. Our laboratory group has focused over the last several years on defining the peculiarities of the human orbital fibroblasts as a strategy for shedding more light on the pathologies occurring in TAO. We have reasoned that unique properties of these cells might ultimately prove the basis for why the manifestations of Graves' disease occur in an anatomically selective manner. In this brief review we attempt to survey our findings. We believe that they might provide a “roadmap” for further discovery into the pathogenesis of TAO. Clearly, more questions remain than those thus far answered.

Introduction

Thyroid-associated ophthalmopathy (TAO) represents a clinically vexing and poorly understood component of Graves' disease (1). The hallmarks of TAO include expansion of the orbital connective tissue/fat pad, infiltration of that tissue and extraocular muscle with mononuclear cells, and tissue remodeling that can culminate in fibrosis and diminished eye motility. How TAO relates to the intrathyroidal processes and the cutaneous manifestation, localized dermopathy, remains uncertain. At its heart, TAO involves targeting and activating orbital fibroblasts, cells exhibiting distinct characteristics (2). In fact, a host of evidence now supports the concept that these fibroblasts represent a critical agent for the pathological changes occurring in TAO (3).

Can the Orbital Pathology in TAO Be Attributed to Unique Phenotypic Attributes of Orbital Fibroblasts?

Tissue changes observed in TAO can range from minimal to dramatic. Hyaluronidase-sensitive material staining positively with Alcian blue accumulates in fatty connective tissue and, early in the disease, intercalates between intact muscle fibers (4). The extraocular muscles can hypertrophy (4,5). Moreover, volumetric expansion of muscles and connective tissue results from an accumulated hyaluronan and the substantially enhanced water binding capacity it conveys to infiltrated tissues (4). This increased tissue volume can in turn compromise the mechanical function of muscle and result in diminished eye motility. Increased orbital volume can also result from enlargement of the fat depot, a process that may involve the de novo differentiation of adipocytes (6). The potential for subsets of orbital fibroblasts to undergo adipogenic differentiation in cell culture suggests that enhanced adipogenesis might underlie the increased fat mass. A number of mechanisms have thus far been proposed for explaining how this might occur (7). Mononuclear cells often dominate the histological picture in TAO, especially early in the disease process. Conspicuous among the infiltrating cells are lymphocytes and mast cells (5,8,9). In muscle, the aggregate of lymphocytes may be confined to the endomysium. Later in the disease, fibrosis and fatty infiltration of the muscle correspond to defects in eye motility (1). Analysis of the infiltrating cells has revealed evidence for lymphocyte activation and for a broad spectrum of cytokine generation (10–14). Cytokines and other small molecules elaborated by T and B lymphocytes and mast cells appear to drive the remodeling seen in TAO. A schematic of our current model for how cytokines and fibroblasts might interact to yield the tissue remodeling associated with TAO is presented in Figure 1. Evidence supporting cytokines playing roles in disease pathogenesis consists of their detection in affected tissues (10–12) and their robust actions on orbital fibroblasts (15–19). Thus, anti-cytokine therapy remains an attractive direction in which efforts toward therapy development might be aimed. In allied processes such as rheumatoid arthritis, this approach has found wide application and has literally changed disease management (20).

FIG. 1.

Cartoon of our current model for the interaction between orbital fibroblasts and members of the professional immune system and the small molecules they produce. Chemoattractant molecules such as IL-16 and regulated on activation, normal T cell expressed (RANTES) are generated in response to Graves' disease–IgG (GD-IgG) acting on the fibroblast. This in turn leads to the recruitment of T cells and other mononuclear cell members of the immune system. When activated, these cells produce a number of proinflammatory cytokines such as IL-1α, IL-1β, CD154 (CD40 ligand), and IL-6. Cytokines in turn activate proinflammatory genes such as those encoding prostaglandin endoperoxide H synthase-2 (PGHS-2), IL-6, IL-8, hyaluronan synthase (HAS), and UDP glucose dehydrogenase (UGDH). The major factor thus far identified as explaining the exaggerated responses to cytokines concerns the low levels of IL-1 receptor antagonist (IL-1RA) expressed by orbital fibroblasts. In addition, IL-4 and IL-13 induce 15-lipoxygenase exclusively in orbital fibroblasts from patients with GD, perhaps accounting for the different patterns of inflammation found in TAO.

Fibroblasts constitute a population of cells directly insinuated in host defense and wound repair. Those inhabiting the human orbit exhibit a set of phenotypic characteristics that set them apart from fibroblasts in other tissues. Notable among these features are exaggerated responses to cytokines such as interleukin-1 β (IL-1β), leukoregulin, and CD154 (15–17). When treated with these cytokines, several proinflammatory genes are dramatically upregulated. These include prostaglandin endoperoxide H synthase-2 (PGHS-2) (17,21,22), IL-1α (18), IL-1β (18), IL-6 (23), RANTES (24), and IL-16 (24). The magnitude of response and the particular array of genes and their products induced suggest that orbital fibroblasts represent highly reactive sentinel cells involved as the “first alert” for coping with tissue injury. But what biological basis underlies the particularly responsive nature of these cells? Differences in the expression and induction of IL-1 receptor antagonist (IL-1RA) might help explain the marked divergence between orbital and nonorbital fibroblast responses to cytokines (18). Specifically, the levels of IL-1RA released by orbital fibroblasts into the culture medium in response to IL-1β and related cytokines are substantially lower than those in other fibroblasts. These diminished IL-1RA levels result in inadequately opposed actions of IL-1 and cytokines acting through the intermediate generation of IL-1 agonists (IL-1α and IL-1β). Therefore, excessive responses in orbital fibroblasts compared with those in other cell populations might result from the divergence in cytokine signaling. These unusual patterns of IL-1RA expression undoubtedly serve some normal function in orbital tissue, but under abnormal circumstances could result in disease.

Orbital Tissue Expansion Involves Increased Fat Volume and Accumulation of Glycosaminoglycans Such As Hyaluronan

Anterior displacement of the eye results from expansion of either the extraocular muscles or the fat pad investing the posterior orbit. In some patients, both compartments are dramatically enlarged. Forward propulsion of the globe can compromise the neurovascular tether of the eye, including the optic nerve, and potentially lead to diminished visual acuity and blindness. A hallmark feature of TAO relates to the disordered accumulation of the glycosaminoglycan, hyaluronan (4). This nonsulfated complex sugar is extremely hydrophilic, and because of its immense molecular mass and rheological properties, it occupies an enormous volume when hydrated. Unlike the other abundant glycosaminoglycans, hyaluronan is synthesized at the cell membrane (4). In human fibroblasts, its production is activated by several cytokines (15). Specifically in fibroblasts from patients with Graves' disease, immunoglobulins directed at the insulin-like growth factor receptor (IGF-1R) can upregulate hyaluronan synthesis (25). Three mammalian hyaluronan synthases (HAS) have been identified and cloned (26). Each isoform is associated with a characteristic pattern of tissue distribution and product chain length. All three enzymes are expressed by orbital fibroblasts from patients with TAO in response to IL-1β (19). Their induction results in the upregulation of hyaluronan synthesis in these cells. Thus, the accumulation of hyaluronan in TAO may be attributed to the induction of one or more of these enzymes, either by inflammatory cytokines or through the actions of disease-specific immunoglobulins. What provokes their activation in vivo in the disease remains uncertain. In addition to the HAS isoforms, uridine diphosphate-glucose dehydrogenase (UGDH) is expressed at higher levels in orbital fibroblasts than those found in cultures from other anatomic regions (Tsui, Chen, and Smith, unpublished observations). It can be induced in these cells by IL-1β (27). UGDH lies immediately upstream from HAS and catalyzes the conversion of UDP-glucose to UDP-glucuronate. Thus, it appears that multiple components of the molecular machinery for glycosaminoglycan synthesis are poised for generating particularly high levels of these complex carbohydrates in orbital fibroblasts.

Orbital Fibroblasts Produce Extraordinarily High Levels of Prostaglandin E₂ When Activated by IL-1β and Related Cytokines

Another notable property exhibited by orbital fibroblasts is the robust induction by IL-1β of the prostanoid biosynthetic pathway. In particular, the upregulation of prostaglandin E₂ (PGE₂) synthesis and the expression of the inflammatory cyclooxygenase, PGHS-2 (aka COX-2), are dramatically greater in these cells than in fibroblasts from skin, thyroid, and other tissues (15,17,18,22,28,29). PGE₂ may play an important role in conditioning the immune responses seen in TAO. It can bias the differentiation of naïve T cells to the T helper 2 (Th2) subtype at the expense of Th1 responses (30,31). Moreover, the prostanoid alters the behavior of mast cells (32) and T and B lymphocytes (33). PGHS-2 induction in the orbit involves a modest and transient activation of the gene promoter and a substantial cytokine-dependent stabilization of mature PGHS-2 mRNA (22). The signaling pathways utilized by IL-1β have been identified and at least two components of the mitogen-activated protein kinase pathway, namely, Erk1/2 and p38, appear to be involved (22). Moreover, the inducible microsomal form of the PGE₂ synthases (mPGES) is coordinately upregulated by the cytokine (22). The functional coupling between PGHS-2 and mPGES yields an efficient generation of PGE₂ in response to IL-1β in orbital fibroblasts. Recent evidence confirms that PGHS-2 is among the inflammatory genes upregulated in orbital fat in situ (34).

IL-4 Can Induce 15-Lipoxygenase-1 Expression and 15-HETE Generation in Orbital Fibroblasts—Identification of a Putative Profibrotic, Antiinflammatory Pathway That Might Self-Limit TAO?

Other bioactive lipids beside PGE₂ are generated preferentially by orbital fibroblasts. 15-Hydroxyeicosatetranoic acid (15-HETE) represents a major metabolite resulting from the oxidation by 15-lipoxygenase (15-LOX-1) of plasma membrane. 15-LOX-1 can be induced by Th2 cytokines, including IL-4 and IL-13, in orbital fibroblasts from patients with TAO (35). In contrast, cultures from donors without Graves' disease fail to express the enzyme. The mechanism underlying the induction concerns a stabilization of 15-LOX-1 mRNA and a modest increase in gene promoter activity. Similar levels of IL-4 receptor display on the surface of fibroblasts from normal tissue and those affected by Graves' disease suggest that important postreceptor events in fibroblasts from the two sources might play a role in the divergent capacity for TAO-derived fibroblasts to respond. These findings carry potential significance in that the 15-LOX-1/15-HETE pathway has been associated with the dampening of proinflammatory signaling and therefore could self-limited the course of TAO. Thus, it remains conceivable that the transition from Th1 to Th2 dominance seen as TAO progresses from early to late disease might drive 15-LOX-1 expression and therefore dampen inflammation and promote end-stage fibrosis.

Orbital Fibroblasts Comprise Subsets with Divergent Capacities for Terminal Differentiation

Another component of tissue expansion in TAO relates to the apparent enlargement of the orbital fat depot (1). In fact, some patients with TAO manifest increased fat volume and vanishingly little muscle enlargement. Expanded fat could reflect reduced apoptosis or enhanced adipogenesis. Sorisky et al. (6) first reported that orbital fibroblasts could differentiate de novo into fat-accumulating cells when subjected to medium containing cAMP-enhancing agents and prostacyclin. Smith et al. reported subsequently that the differentiation into fat cells induced by these agents might be restricted to a sub-population of fibroblasts (36). Moreover, agents that ligate and activate peroxisome proliferator activated receptor γ (PPARγ), such as the thiazolidinediones, promote the adipogenesis of orbital fibroblasts; findings have been substantiated by those of other groups (37–39). Orbital fibroblasts are a heterogeneous population comprising cell subsets with distinct phenotypic attributes (40). When segregated on the basis of the cell surface marker, Thy-1, their divergent capacities for terminal differentiation become apparent. Thy-1⁺ fibroblasts can differentiate into myofibroblasts when treated with TGF-β. These cells play important roles in wound healing and tissue contraction. They also help generate scar tissue, and their continued activation can lead to fibrosis. TGF-β signaling involves downstream pathways where several kinases called Smads become activated. The hallmark of the myofibroblast phenotype is overexpression of smooth muscle-specific actin. On the other hand, Thy-1⁻ fibroblasts, when treated with cAMP-enhancing agents, PPARγ ligands such as PGJ₂ or the thiazolidinediones such as Rosiglitazone, and prostacylin differentiate into mature adipocytes that accumulate triglycerides in their cytoplasm (36). Besides divergent capacities for terminal differentiation, these cell subsets exhibit distinctly different biosynthetic repertoires.

The diversity displayed by orbital fibroblasts in culture mirrors the differences that set them apart from cells derived elsewhere. By virtue of their embryological pedigree consisting of both neural crest ectoderm and mesodermal elements (41,42), orbital fibroblasts appear to respond particularly vigorously to a wide array of inflammatory signals. In light of the complex story unfolding concerning kidney fibrosis (43–46), consideration of the potential for trans-differentiation of epithelial, smooth muscle, and endothelial elements found in orbital connective tissue may ultimately account for the cellular diversity encountered in the human orbit. Specifically, we now know that under certain conditions, cells exhibiting a particular well-differentiated phenotype might be provoked to differentiate into another, as has been the case in kidney disease (43,44). For instance, Oncostatin M has been shown to induce epithelial cell–myofibroblast trans-differentiation, an action mediated through the Jak/STAT pathway (46). Moreover, circulating fibroblasts such as fibrocytes might become trafficked to the orbit in TAO from the bone marrow. Fibrocyte generation has been linked to the pathogenesis of kidney and lung diseases (47).

Orbital Fibroblasts Express Several Important Cytokines Playing Roles in Lymphocyte Recruitment and Inflammation

A remarkable feature of orbital inflammation in TAO concerns trafficking of professional immune cells to the disease. When activated by IL-1β, TNF-α, TGF-β, CD154, or leukoregulin, fibroblasts from this tissue express cytokines such as IL-1α, IL-1β, and IL-6 (23) and several powerful mononuclear cell chemoattractants, including IL-8, RANTES, and IL-16 (16,24). IL-8 and RANTES are CXC and CC chemokines, respectively. IL-8 targets several cell types, including neutrophils and T cells, while RANTES, working through CCR4, CCR5, and other GTP protein–coupled receptors, targets monocytes and lymphocytes. IL-16 lacks the requisite cysteine signatures to be classified as either chemokine type but plays an important role in CD4⁺ T cell trafficking to sites of inflammation (48). In those cultures from patients with TAO, specific IgGs derived from the same donors provoke IL-16 and RANTES production (49). These IgGs act through the IGF-1R (50) and target specifically the fibroblasts derived from patients with the disease (49). It would appear that the anatomic region from which the responsive fibroblasts derive is immaterial as those from the abdominal wall, neck, and pretibium exhibit an equivalent induction of IL-16 and RANTES. With regard to IL-16 induction, the Akt/FRAP/mTOR/p70^s6k pathway mediates the action of these IgGs, effects which can be blocked with rapamycin (49). One key difference between control fibroblasts and those from patients with GD relates to IGF-1R density. Surface display of the receptor as assessed by flow cytometry using a specific Ab against IGF-1Rα as well as Western blotting with anti-IGF-1Rβ suggest that fibroblasts from patients with Graves' disease overexpress the receptor (50). Moreover, the fraction of IGF-1R⁺ fibroblasts in cultures from patients with TAO is considerably higher than that in fibroblasts from control donors.

A major question remains concerning IGF-1R–dependent signaling and the trafficking of T cells to the orbit. Since fibroblasts from anatomic regions not ordinarily manifesting the disease in patients with Graves' disease also overexpress the receptor and engage in signaling that culminates in chemoattractant synthesis, how can we reconcile the localized manifestations found in the orbit? We believe that the answer lies in the unique set of environmental factors found in the orbit (and shin). Thus, while T cell recruitment in this disease might prove global, the very different consequences of their presence in orbital tissues could result from the differential susceptibilities of orbital fibroblasts to cytokines. Factors such as the differential production of IL-1RA alluded to previously in connective tissues outside and within the orbit could account, at least in part, for the inflammation occurring selectively in that tissue.

Evidence for Direct Crosstalk between Professional Immune Cells and Orbital Fibroblasts

Current models of orbital Graves' disease place great importance on the release of soluble cytokines from activated T cells and their impact on target fibroblasts. While this picture seems plausible, it also appears incomplete. Orbital fibroblasts express and display particularly high levels of CD40 and respond to CD154 (16,17). Autologous T cells can drive fibroblast proliferation in mixed culture, an action dependent on cell–cell contact and on MHC class II and CD40/CD154 signaling (51). Activated T cells from patients with TAO can drive differentiation of orbital fibroblasts to adipocytes, through the action of lymphocyte-derived PGD₂, presumable mediated through a PPARγ-dependent mechanism (39). When incubated with HMC-1 mast cells, orbital fibroblasts generate increased levels of PGE₂, the consequence of pretranslational induction of PGHS-2 (52). This induction can be blocked with anti-IL-4 receptor Abs. Moreover, hyaluronan synthesis is also increased in fibroblasts cocultured with HMC-1 cells. Thus, a number of nontraditional signaling pathways including CD40, prostanoids, and IL-4 may serve as important molecular conduits between activated T cells, mast cells, and residential fibroblasts inhabiting the orbit in TAO.

IGF-1R Overexpression Might Help Explain the Expansion of T Cell Populations in Graves' Disease

Not only does IGF-1R overexpression pertain to a divergent pattern of signaling found in fibroblasts from patients with Graves' disease, but T cell activation might also involve this receptor. Expansion of memory T cells has been detected in several human autoimmune diseases, including Graves' disease (53). Given the findings concerning IGF-1R overexpression in fibroblasts, Douglas et al. (54) found a disproportionately large number of IGF-1R⁺ T cells in the circulation of patients with Graves' disease and TAO. Moreover, T cells harvested from the affected orbital connective tissue mirrored this skewed T cell population. Importantly, dramatically greater numbers of IGF-1R⁺ cells were found among CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ memory T cells (54). Increased IGF-1R display on these lymphocytes appears to carry substantial functional importance. Addition of IGF-1 or IgG from these patients enhanced T cell proliferation and promoted resistance to apoptosis. Thus, IGF-1R expression and signaling could underlie the peculiar profile of T cells driving the pathogenesis of TAO. Moreover, the potential interplay between this receptor and the activating antibodies directed against it could define a proximate link between B cells and T cells in this disease.

Conclusions

Our current understanding of the pathogenesis of TAO remains incomplete. Clearer definition of the important interactions between the molecular mediators of the disease and their targeted cells has led us to conclude that susceptibility to TAO involves multiple factors. Notable among them are the unique phenotypes exhibited by orbital fibroblasts. These appear to underlie the orbital manifestations of the disease. The proximate signaling pathways involved in orbital fibroblast activation have emerged as attractive targets for interrupting the disease and lessening its morbidity. Further investigation into the role of IGF-1R in disease pathogenesis and defining the dimensions of the phenotypic divergence displayed by orbital fibroblasts remain formidable objectives. They should occupy our focus in future studies.

Acknowledgments

The authors thank Ms. Debbie Hanaya for her expert help in preparing this manuscript. This work was supported in part by National Institutes of Health grants EY008976, EY11708, DK063121, EY016339, RR017304, and RR00425, and unrestricted grant from Research to Prevent Blindness, Research to Prevent Blindness Career Development Award, and through the generosity of the Bell Charitable Foundation.