Immunoglobulin Activation of T Cell Chemoattractant Expression in Fibroblasts from Patients with Graves’ Disease Is Mediated Through the Insulin-Like Growth Factor I Receptor Pathway¹

Graves’ disease (GD)³ is an autoimmune process in which the thyroid gland becomes infiltrated by lymphocytes and activated by IgGs directed against the thyrotropin (TSH) receptor (TSHR) (1). IgG/TSHR interactions result in excess thyroid hormone synthesis and thyroid gland enlargement (2). Another component of GD involves the infiltration of orbital and pretibial connective tissue depots, processes referred to as thyroid-associated ophthalmopathy (TAO) and dermopathy, respectively (3, 4). Orbital tissues in TAO become inflamed (5), and it is currently believed that infiltrating immunocompetent cells synthesize inflammatory mediators, such as cytokines, which activate fibroblasts. CD4⁺ and CD8⁺ T lymphocytes, mast cells, and B cells have been identified in affected orbital tissue, but the predominance of any particular cell type in early disease has been a topic of considerable debate (6, 7). The role postulated for fibroblasts in the pathogenesis of these GD manifestations rests largely on circumstantial evidence. Fibroblasts possess a newly appreciated potential for orchestrating inflammatory responses and initiating critical elements of tissue remodeling, such as those occurring in GD (8).

While the central role for anti-TSHR IgGs in provoking disordered thyroid function is firmly established (9), little direct evidence has been advanced to implicate them in the peripheral components of GD. Temporal discordance between the onset of hyperthyroidism and the development of TAO in some patients suggests that causative factors provoking extraglandular and thyroidal GD might differ. Moreover, the majority of patients with GD fail to manifest clinical TAO or dermopathy, while others with TAO never become hyperthyroid. Little progress has been made in establishing a mechanistic basis for TAO or its link to the thyroid gland. As a result, therapy for this orbital disease has remained inadequate.

We reported recently that IgGs from patients with GD (GD-IgG) can activate fibroblasts derived from those individuals to express IL-16 and RANTES (10). IL-16 is a soluble ligand that cross-links its receptor, CD4⁺, by binding to the D4 region (11). IL-16 ligation of CD4⁺ cells has substantial biochemical and functional consequences, such as migration and regulation of cell activation (12, 13, 14). While maintaining responsiveness to IL-2 and IL-15, IL-16 stimulation of CD4⁺ T cells results in the loss of responses to Ags. Elevations of IL-16 levels have been detected in sera from patients with asthma (15), systemic lupus erythematosus (16), rheumatoid arthritis (17), and inflammatory bowel disease (18). A positive correlation has been established between IL-16 expression and CD4⁺ T cell infiltration (15).

RANTES is a C-C chemokine that activates resting and activated T lymphocytes and monocytes, effects mediated through CCR5 and other GTP protein-coupled receptors (19, 20). In T cells RANTES activates Janus kinase kinases and p38 mitogen-activated protein kinase (21). Like IL-16, RANTES has been implicated in human autoimmune diseases and has been detected in thyroid tissue from patients with GD (22). IL-16 and RANTES are expressed by cytokine-activated human fibroblasts from several tissues, including orbit, thyroid, skin, and synovial membrane (23). In fibroblasts from patients with GD, IL-16 and RANTES are up-regulated by GD-IgG (10). Induction of IL-16 involves activation of the Akt/FRAP/mammalian target of rapamycin/p70^s6k pathway and is susceptible to inhibition with rapamycin (10). In contrast, RANTES is induced through a rapamycin-insensitive pathway.

We hypothesize that the up-regulation of IL-16 and RANTES by GD-IgG underlies the infiltration of connective tissues by immunocompetent cells in GD. While some of the biological consequences of GD fibroblast activation by GD-IgG have been identified, the nature of the Ag has not been previously elucidated. Some features of the connective tissue remodeling observed in TAO suggest the possible involvement of growth factors. Thus, we investigated whether any of the growth factor receptors displayed by fibroblasts might prove relevant to our earlier findings. Surprisingly, we can now implicate the insulin-like growth factor I (IGF-I) receptor (IGF-IR) pathway in the mediation of IL-16 and RANTES induction in GD fibroblasts provoked by GD-IgG. IGF-IR is a widely expressed tyrosine kinase receptor that resembles the insulin receptor (24). Interfering with IGF-IR function can block the chemoattractant expression provoked by GD-IgG. GD-IgGs fail to induce IL-16 or RANTES in control fibroblasts, indicating an intrinsic difference between disease-derived and normal fibroblasts. Moreover, GD-IgG can attenuate Ab detection of IGF-IR using flow cytometric analysis and compete for ¹²⁵I-labeled IGF-I ([¹²⁵I]IGF-I) binding on the surface of intact fibroblasts. Our findings strongly implicate IGF-IR in the stimulation by GD-IgG of IL-16 and RANTES expression in GD fibroblasts. These observations suggest that IGF-IR is a previously unrecognized self-Ag involved in the pathogenesis of GD.

Human rIL-1β and a RANTES ELISA were purchased from BioSource International (Camarillo, CA). Polyclonal rabbit anti-rIL-16 Ab was prepared from rIL-16-immunized rabbit sera. IGF-IR-blocking Ab 1H7 was obtained from BD PharMingen (San Diego, CA). Neutralizing anti-RANTES Abs and human rIGF-I were purchased from R&D Systems (Minneapolis, MN). Des(1–3) IGF-I and [Leu 24]IGF-I were obtained from GroPep (Adelaide, Australia). [¹²⁵I]iodotyrosyl-IGF-I (2000 Ci/mmol) was purchased from Amersham Pharmacia Biotech (Piscataway, NJ). Rapamycin was obtained from Calbiochem (La Jolla, CA). Human recombinant TSH and dexamethasone were purchased from Sigma-Aldrich (St. Louis, MO).

Human fibroblasts were obtained from individuals with GD and from donors without known thyroid disease. Orbital fibroblast strains were initiated from surgical waste. Dermal fibroblasts were derived from biopsies of normal-appearing skin or were purchased from American Type Culture Collection (Manassas, VA). Sera were collected from patients with GD, either without or with clinical TAO, and from individuals without thyroid disease (controls). The diagnosis of GD was confirmed as described previously (10). IgG was prepared by protein A affinity chromatography (25). Fibroblasts were characterized as described previously (26). The institutional review board of Harbor-University of California-Los Angeles Medical Center has approved these protocols.

Fibroblasts were treated with nothing (control), IL-1β (10 ng/ml), IGF-I (10 nM), human serum (final concentration, 1%), or protein A-purified human IgG added to the medium for the times indicated. Culture medium was collected, and chemotaxis was assessed in a microchemotaxis chamber using human NWNA-T lymphocytes and 8-μm Micropore nitrocellulose filters (NeuroProbe, Cabin John, MD) as previously described (27). To assess the chemoattractant activity attributable to IL-16 or RANTES, neutralizing experiments were conducted by incubating culture supernatants for 15 min with either affinity-purified anti-IL-16 mAb (clone 14.1 (10 μg/ml), which neutralizes 50 ng/ml rIL-16) or anti-RANTES mAb (5 μg/ml; neutralizing dose 50 of 200 ng/ml for rRANTES) before the migration assay.

Quantification of IL-16 protein released from the fibroblast monolayers was accomplished by subjecting aliquots of conditioned medium to commercial IL-16 and RANTES ELISAs (BD PharMingen and BioSource International, respectively).

Plasmid containing the soluble DN mutant IGF-IR, 486/STOP (provided by Dr. R. Baserga, Jefferson Medical School, Philadelphia, PA) (28), was transiently transfected into cells at 80% confluence using the Lipofectamine Plus system (Invitrogen, San Diego, CA). The plasmid (0.4 μg) was mixed with Plus reagent for 15 min before being combined with Lipofectamine. Some cultures received empty vector DNA or were sham-transfected as controls. The DNA-lipid mixture was added to the culture medium for 3 h at 37°C. DMEM containing 10% FBS replaced the transfection mixture overnight. Transfected cultures were then serum-starved, and some received GD-IgG or normal IgG for 16 h. Medium samples were collected and frozen until they were assayed.

[¹²⁵I]IGF-I binding to the surface of intact fibroblasts was determined by incubating confluent monolayers in 100 μl of medium containing 0.5 μCi of 3[¹²⁵I]iodotyrosyl-IGF-I without or with the additives specified. Cultures were incubated for 2 h at 4°C, and then cells were scraped off the substratum in ice-cold buffer containing EDTA. Bound radioactivity was determined by rapid filtration of duplicate cell suspensions through GF/C microfiber filters (Whatman, Maidstone, U.K.) using a vacuum manifold. Data are presented as the fractional displacement of the IGF-I binding occurring with radiolabeled IGF-I alone. To visualize the binding association, dissociated GD or control fibroblasts were bound by FITC-labeled anti-IGF-IR Ab for 30 min at 4°C and analyzed on a FACScan Flow Cytometer (BD BioSciences, San Diego, CA). To determine specific binding, cells were pretreated (10 min, 4°C) with either control IgG or GD-IgG (100 μg/ml) before addition of the FITC-labeleddetection Ab and FACScan assessment.

Orbital and skin fibroblasts, including those from patients with GD, express substantial T cell chemotaxis-promoting activity in response to IL-1β, of which the vast majority can be attributed to IL-16 and RANTES (Fig. 1) (23). Moreover, 25 of 26 of the GD-IgG preparations isolated from patients, without or with overt TAO, activated GD fibroblasts to express high levels of both chemoattractants (Fig. 1) (10). Ten of 11 fibroblast strains from patients with GD were found to respond to GD-IgG with a substantial induction of IL-16 and RANTES (10). In contrast, five of five strains derived from donors without GD failed to respond to GD-IgG, but became activated following treatment with IL-1β (10, 23). Because sera from most patients with GD contain anti-TSHR Abs (1, 2), and human fibroblasts display TSHR (29, 30), we determined the potential effects of TSH on chemoattractant expression in fibroblasts. GD fibroblasts were treated with TSH (10 mU/ml) for 24 h and then assayed for chemoattractant expression. TSH failed to induce either IL-16 or RANTES or influence the overall generation of T cell migratory activity (Fig. 2), strongly suggesting that TSHR activation does not mediate the actions of GD-IgG on fibroblasts.

To identify the receptor involved in GD-IgG stimulation, molecules known to bind cell surface receptors expressed by human fibroblasts were examined for their ability to induce IL-16 and RANTES. Among these, IGF-I (10 nM) significantly increased IL-16- and RANTES-dependent T cell migration after a 24-h treatment (Figs. 1 and 2). The magnitude of the induction was similar to that following treatment with IL-1β. To determine whether IGF-IR was mediating the effect, the IGF-IR-specific ligand, Des(1–3) IGF-I, was tested and also found to induce chemoattractant expression. Des(1–3) IGF-I binds to the IGF-IR with high affinity and leads to activation of the receptor (31). In contrast, it does not bind to the IGF-I-binding proteins (IGFBPs). On the other hand, [Leu 24]IGF-I, a molecule that interacts with IGFBPs but has a markedly decreased affinity for IGF-IR (32), failed to induce IL-16 or RANTES in these fibroblasts (Fig. 2).

Induction of IL-16 and RANTES by GD-IgG is limited to fibroblasts from patients with GD (10). To test whether IGF-I effects are similarly restricted, four different fibroblast strains, each obtained from a separate control donor, were challenged with the growth factor (Fig. 3). None of these fibroblast strains responded to IGF-I treatment under conditions identical to those used in studies with GD fibroblasts. However, all control strains responded to IL-1β with substantial increases in IL-16- and RANTES-dependent T cell chemotaxis and protein expression. Thus, it appears that intrinsic differences in GD and normal fibroblasts maintained in culture underlie the divergent responses to GD-IgG and exogenous IGF-I. These findings imply that the effects of IGF-I and GD-IgG are related.

To directly determine whether interfering with IGF-IR function could block GD-IgG-dependent increases in IL-16 and RANTES expression, fibroblast cultures were challenged with IgGs in the absence or the presence of a highly specific receptor-blocking Ab, clone 1H7 (Fig. 1). This Ig binds to the α subunit of the receptor, but fails to initiate signaling (33). As the figure demonstrates, clone 1H7 completely attenuates the effects of GD-IgG on cell migration and chemoattractant expression. An isotype-matched control Ab failed to alter the stimulation by GD-IgG (Fig. 1). The ability of IGF-IR neutralization to block both IL-16 and RANTES induction by GD-IgG was subsequently demonstrated using sera from five patients with GD (Fig. 4). In each case 1H7 blocked the expression of T cell chemotaxis in GD fibroblasts.

To further demonstrate the important role of IGF-IR in mediating GD-IgG stimulation of IL-16 and RANTES expression in GD fibroblasts, a DN mutant receptor, 486/STOP, was introduced by transient transfection into these fibroblasts. 486/STOP encodes a 486-amino acid protein conforming to IGF-IRα that is secreted into the culture medium (28). The secreted mutant thus functions as a nonsignaling sink for IGF-I and is effective even if the transfection efficiency is <100%. Cultures were transiently transfected with IGF-IR DN before GD-IgG stimulation. The IGF-IR DN completely attenuated the induction of both IL-16 and RANTES by GD-IgG (Fig. 5). In contrast, sham transfections and those involving an empty vector control failed to alter the effects of GD-IgG. This result provides important confirmation with regard to the role of IGF-IR in mediating the actions of GD-IgG, considering the complexities of the IGF-I system. For instance, IGFBPs have been shown to mediate the biological effects of IGF-I (34). Taken together with the results obtained with 1H7, the IGF-IR-blocking Ab (Fig. 4), and the IGF-IR-specific agonist, Des(1–3) (Fig. 2), IGF-IR appears to mediate the actions of GD-IgG in fibroblasts.

To determine whether the difference in GD-IgG stimulation of GD and control fibroblasts was attributable to IGF-IR display, receptor expression on these cells was analyzed by flow cytometry. As shown in Fig. 6,A, control fibroblasts expressed lower levels of IGF-IR (2- to 4-fold) than fibroblasts from patients with GD. Since the detection Ab specifically recognizes IGF-IRα, the expression of IGF-Iβ was determined by Western blot analysis of cell lysates. Fig. 6,B demonstrates that the β subunit of the receptor was detected in all seven fibroblast strains (three from controls and four from GD donors), but the levels were higher in the fibroblasts from patients with GD (GD, 0.845 ± 0.35 arbitrary units; control, 0.181 ± 0.1 arbitrary units (mean ± SD). To determine a direct association between GD-IgG and IGF-IR, competitive binding studies were conducted. Scatchard analysis revealed that [¹²⁵I]IGF-I binds to these cells with a K_d of 0.3 nM. As shown in Fig. 6,C, increasing concentrations of unlabeled IGF-I displace nearly 100% of total [¹²⁵I]IGF-I binding to these cells. Addition of Des(1–3) IGF-I also inhibits this binding (65% at the highest concentration tested), indicating substantial displaceable binding of [¹²⁵I]IGF-I to IGF-IR. GD-IgG displaced [¹²⁵I]IGF-I binding up to 80%, while control IgG attenuated binding by a maximum of only 18% at the highest concentration tested (500 μg/ml). Flow cytometric analysis confirms the specific interaction between GD-IgG and IGF-IR displayed on GD fibroblasts. Disease-specific Ig blocks FITC-conjugated anti-IGF-IRα Ab binding, so that the number of IGF-IR-positive cells is reduced from 74 to 32% (Fig. 6,D). Importantly, GD-IgG fails to influence the detection of IGF-IRα on control cells (Fig. 6 D). Thus, GD-IgG specifically recognizes IGF-IR on the GD fibroblast surface.

GD-IgG induces the expression of active IL-16 and RANTES in cultured fibroblasts from patients with GD (10). These effects of disease-specific Igs can now be attributed to interactions with IGF-IR, which is widely expressed on human fibroblasts (35). Activating TSHR with TSH fails to elicit these responses. GD-IgG, like IGF-I, can provoke the release of T cell chemotactic activity from fibroblasts derived from several anatomic regions of patients with GD, including those ordinarily not manifesting the disease, but not in control fibroblasts from donors without the disease. These findings indicate that GD fibroblasts from multiple tissues differ intrinsically from their counterparts derived from control donors. If fibroblasts from apparently uninvolved tissues are also activated, what is the basis for the apparent anatomic site selectivity of GD? We hypothesize that the peculiar disease distribution involvesadditional susceptibility factors existing in orbital and pretibial fibroblasts. For instance, compared with cells from tissues irrelevant to GD, orbital fibroblasts exhibit exaggerated responses to proinflammatory cytokines (36, 37). The profile of proteins they express following treatment with T cell-derived cytokines differs (38), and they display particularly high levels of CD40 (39). Ligation of that receptor on orbital fibroblasts with CD154 results in the robust up-regulation of key proinflammatory molecules, such as IL-6, IL-8, and PG endoperoxide H synthase-2, and the generation of hyaluronan (39, 40). Thus, the consequences of infiltrating T lymphocytes in disease-affected connective tissue depots may differ from regions not manifesting GD. Since the majority patients with GD fail to manifest clinically obvious TAO or dermopathy, we suspect that orbital and pretibial fibroblasts in individuals without connective tissue involvement may lack exaggerated responses to proinflammatory cytokines.

IGF-I exerts multiple actions on growth and metabolism (24, 34). IGF-IR resembles the insulin receptor and can mediate the growth-promoting activities of insulin as well as those of IGF-I and IGF-2. The receptor comprises two subunits, is displayed on the cell surface as disulfide-linked dimers, and exhibits both high and low affinity binding states and negative cooperativity (41, 42). Cell signaling involves domain rearrangements instead of oligomerization. IGF-IR has been implicated in the pathogenesis of growth and neoplastic disorders. Activated IGF-IR functions as an inhibitor of apoptosis, and its overexpression tends to offer protection against cell death (43). IGF-IR had not been implicated previously in the realm of human autoimmunity.

A number of questions emerge from our current findings. For instance, if IGF-IR activation by GD-IgG is widespread and chronic, do patients with GD manifest growth abnormalities or exhibit alterations in glucose homeostasis as a consequence? Children with GD often experience accelerated linear growth associated with precocious maturation of the epiphysis (44). The impact of active GD on glucose homeostasis is widely recognized (45). Heretofore, both derangements have been attributed solely to the thyrotoxic component of GD. Our current assessment of the ability of GD-IgG to initiate downstream signaling through the IGF-IR to date has been limited to studies involving cells of the fibroblast lineage. Therefore, the impact of GD-IgG directed against IGF-IR on other cell types, including those relevant to growth and glucose metabolism, remains uncertain. Clearly, defining the consequences of IGF-IR activation by GD-IgG on several metabolic pathways relevant to this receptor will require further investigation in patients with GD.

Activation by GD-IgG of IGF-IR may represent the basis for T cell infiltration of multiple tissues, including the orbit and thyroid. Identification of a second autoantigen in GD in addition to TSHR could help explain the temporal discordance observed with regard to the onset of glandular and extrathyroidal manifestations. Our findings are consonant with GD-IgG interacting with IGF-IR displayed on human fibroblasts. GD-IgG displaces specific [¹²⁵I]IGF-I binding to IGF-IR in GD fibroblasts (Fig. 6 C). Earlier observations had demonstrated that [¹²⁵I]IGF-I could bind to orbital fibroblasts obtained from normal tissue (46). Twelve of 23 GD-IgG preparations tested in that earlier study significantly displaced binding regardless of whether the donor manifested clinical TAO. The K_d of IGF-I binding was 0.5 nM (46), consistent with that found in dermal fibroblasts (35). These earlier studies failed to identify the IGF-I binding site in fibroblasts. IGF-I can bind with high affinity to several cell surface proteins, including multiple IGFBPs, as well as IGF-IR (34).

Our results demonstrate a coordinate induction of IL-16 and RANTES by GD-IgG. The expression of these molecules in tandem has very specific implications with regard to the nature of the immune response. Both chemoattractants can target resting and activated CD4⁺ T cells. Coupled secretion of IL-16 and RANTES may create a selective process for T cell recruitment, because IL-16/CD4 stimulation desensitizes RANTES/CCR5-induced signaling and results in the partial inhibition of T cell migration (47). This attenuation is incomplete, since RANTES binds to several other members of the chemokine receptor family. However, a decrease in recruited CCR5⁺ cells is noted. It is conceivable that this functional relationship between IL-16 and RANTES contributes substantially to a selective process of T cell recruitment. Further investigation is required to establish its relevance to GD.

The fundamental nature of human autoimmunity remains uncertain and might be intimately related to infectious processes (48). This is certainly true for GD, where a number of disease-initiating triggers have been considered (49). Whatever the earliest events provoking development of the disease, the novel model of GD we now propose involves the productive interaction between at least two self-Ags, TSHR and IGF-IR, and disease-specific Igs, leading to downstream signaling. Whether the epitopes present on these Ags are related remains to be determined. It is widely accepted that anti-TSHR Igs promote thyroid growth and disordered activity. We postulate that interactions between IGF-IR and GD-IgG initiate T cell migration and lead to lymphocyte infiltration of the thyroid gland, orbital connective tissue, skin, and potentially other anatomic regions of the body. Because many cell types express functional IGF-IR, including lymphocytes, it is possible that GD-IgGs directed against this receptor might provoke responses in multiple tissues. It is possible that components of the IGF-IR pathway may prove attractive targets for therapeutic intervention of the disease.

We thank Drs. R. Baserga (Jefferson University) and P. Cohen (University of California-Los Angeles) for helpful discussions and providing useful reagents.