Bidirectional TSH and IGF-1 Receptor Cross Talk Mediates Stimulation of Hyaluronan Secretion by Graves' Disease Immunoglobins

Abstract

Context:

There is no pathogenetically linked medical therapy for Graves' ophthalmopathy (GO). Lack of animal models and conflicting in vitro studies have hindered the development of such therapy. Recent reports propose that Graves' Igs bind to and activate thyrotropin receptors (TSHRs) and IGF-1 receptors (IGF-1Rs) on cells in orbital fat, stimulating hyaluronan (HA) secretion, a component of GO.

Objective:

The objective of the study was to investigate potential cross talk between TSHRs and IGF-1Rs in the pathogenesis of GO using a sensitive HA assay.

Design/Setting/Participants:

Orbital fibroblasts from GO patients were collected in an academic clinical practice and cultured in a research laboratory. Cells were treated with TSH, IGF-1, and a monoclonal Graves' Ig M22.

Main Outcome Measures:

HA was measured by a modified ELISA.

Results:

Simultaneous activation by TSH and IGF-1 synergistically increased HA secretion from 320 ± 52 for TSH and 430 ± 65 μg/mL for IGF-1 alone, to 1300 ± 95 μg/mL. IGF-1 shifted the TSH EC₅₀ 19-fold to higher potency. The dose response to M22 was biphasic. An IGF-1R antagonist inhibited the higher potency phase but had no effect on the lower potency phase. M22 did not cause IGF-1R autophosphorylation. A TSHR antagonist abolished both phases of M22-stimulated HA secretion.

Conclusions:

M22 stimulation of HA secretion by GO fibroblasts/preadipocytes involves cross talk between TSHR and IGF-1R. This cross talk relies on TSHR activation rather than direct activation of IGF-1R and leads to synergistic stimulation of HA secretion. These data propose a model for GO pathogenesis that explains previous contradictory results and argues for TSHR as the primary therapeutic target for GO.

Graves' disease (GD) is an autoimmune disease comprised of two major components: hyperthyroidism and ophthalmopathy [or Graves' orbitopathy (GO)] (1). It is clear that Graves' hyperthyroidism is caused by the activation by circulating Igs (GD-IgGs or thyroid stimulating antibodies) of TSH receptors (TSHR) on thyroid cells leading to stimulated synthesis and secretion of thyroid hormones. The pathogenesis of GO, however, is less clear. Although it appears that GD-IgG activation of TSHR on fibroblasts/preadipocytes and adipocytes in the soft tissue of the eye plays a role in GO pathogenesis, it has been proposed that GD-IgG may also directly activate IGF-1 receptors (IGF-1Rs) on these cells to contribute to disease development (2, 3). A functional relationship between TSHR and IGF-1R signaling has been previously established in thyroid cells wherein simultaneous activation of the two receptors leads to the synergistic up-regulation of DNA synthesis and cell proliferation (4–6). In support of this idea in the pathogenesis of GO, it has been suggested that patients with GO may have circulating antibodies which bind TSHR and IGF-1R, but whether IGF-1R is a secondary GO target has not been established (7–9). Because GD-IgGs are polyclonal, it is possible that different antibodies within a patient's GD-IgG may bind to and activate TSHR and IGF-1R. Recently, however, it was reported that a human monoclonal antibody M22, in addition to stimulating cAMP (10), also activates phosphatidylinositol 3-kinase-Akt signaling (11), which is downstream of both TSHR and IGF-1R pathways.

A major component of GO is the excessive deposition of hyaluronan [hyaluronic acid (HA)] in the extracellular matrix of orbital soft tissue. Because attempts at generating an animal model for GO (12) have yet to be reproduced, most research in this field has been performed in tissue culture using GO fibroblasts/preadipocytes (GOFs) and adipocytes obtained from GO patients at orbital decompression surgery (13). GOFs express TSHR and IGF-1R, and selective activation of both receptors by their cognate ligands TSH and IGF-1, respectively, has been shown to stimulate HA secretion by these cells (14, 15). It is therefore likely that cross talk between TSHR and IGF-1R occurs in GOFs (2) as has been shown for G protein-coupled receptors (GPCRs) including TSHR and receptor tyrosine kinases (RTKs) including IGF-1R (16, 17).

Herein we demonstrate that TSHR and IGF-1R on GOFs are functionally dependent. We show that simultaneous treatment with TSH and IGF-1 synergistically increased HA secretion by GOFs, wherein increasing IGF-1 concentration augmented potency and efficacy of TSH on TSHR, and that dose-dependent stimulation of HA secretion by M22 was biphasic, with the higher potency phase mediated in part by IGF-1R. These data provide evidence of M22-induced cross talk between TSHRs and IGF-1Rs to synergistically increase HA secretion. We suggest this GD-IgG-induced bidirectional cross talk plays a pivotal role in the pathogenesis of GO.

Materials and Methods

Materials

Thyrotropin from bovine pituitary (TSH), human IL1β, and (R)-(+)-trans-4-(1-aminoethyl)-N-(4-pyridyl)-cyclohexanecarboxamide dihydrochloride (Y-27632) were purchased from Sigma-Aldrich. Recombinant human IGF-1, human platelet-derived growth factor-AB, human fibroblast growth factor-2, and human TGFβ1 were purchased from PeproTech. Thyroid-stimulating human monoclonal autoantibody (M22) was purchased from Kronus. Thyroid-stimulating hamster monoclonal antibody was kindly provided by Dr Terry Davies (Mount Sinai Hospital, New York, New York). The TSHR antagonist NCGC00229600 (C1) was synthesized by the National Center for Advancing Translational Science, National Institutes of Health, as previously reported (18). IGF-1R receptor kinase inhibitor linsitinib was purchased from Selleckchem. HA ELISA kits were purchased from Corgenix. One MDa HA was purchased from Lifecore Biomedical.

Cell culture

Retrooribital adipose tissue from two female patients and one male patient with GD was generously supplied by Drs Neil Miller, Prem Subramanian, and Shannath Merbs (Johns Hopkins School of Medicine, Baltimore, Maryland). Institutional review board approval was obtained from Johns Hopkins Hospital. Adherent cells were isolated from tissue as previously described (14). Cells were maintained in F-media containing DMEM with FBS [10% (vol/vol)], penicillin (100 U/mL), streptomycin (100 μg/mL), L-glutamine (2 mM), Ham's F-12 nutrient mixture [25% (vol/vol)], hydrocortisone (25 ng/mL), epithelial growth factor (0.125 ng/mL), insulin (5 μg/mL), cholera toxin (11.7 nM), gentamicin (10 μg/mL), Fungizone (250 ng/mL), and Y-27632 (5 μM) in a humidified 7% CO₂ incubator at 37°C.

Measurement of HA secretion in orbital fibroblasts

To measure HA secretion, cells were grown to confluence and then pretreated with hyaluronidase [1 U/mL in Hanks' balanced salt solution (HBSS)] for 1 hour at 37°C. After the digestion of preexisting HA, GOFs were switched to DMEM with fetal bovine serum (FBS) [10% (vol/vol)], penicillin (100 U/mL), and streptomycin (100 μg/mL) with individual or combination treatments of TSH, IGF-1, M22, platelet-derived growth factor, fibroblast growth factor-2, TGFβ1, or IL-1β and incubated for 5 days in 7% CO₂ at 37°C. For experiments inhibiting TSHR or IGF-1R, cells were pretreated with antagonist in low-serum DMEM (1% FBS) at 37°C for 1 hour before the addition of FBS (final concentration 10%) and TSH, IGF-1, or M22. Conditioned media were collected and stored at −20°C. Conditioned media were assayed using a modified Corgenix HA ELISA kit as previously described (14).

Measurement of IGF-1R phosphorylation

GOFs were grown to confluence in six-well plates and then serum starved in DMEM with 2% BSA for 24 hours. Cells were pretreated with antagonist in HBSS at 37°C for 1 hour and then incubated at 37°C for 30 minutes with maximally effective doses of IGF-1 or M22 in HBSS with or without antagonist. Lysates were prepared using the Bio-Plex Pro cell signaling kit (Bio-Rad Laboratories; catalog number 171-304006M) according to the manufacturer's directions and adjusted so that all samples had equal protein concentrations. Phosphorylated IGF-1R levels were measured with a Bio-Plex MAGPIX multiplex reader (Bio-Rad Laboratories; catalog number 171-015001) using the Phospho-IGF-1R (Tyr1131) set (Bio-Rad Laboratories; catalog number 171-V50009M) according to the manufacturer's directions.

Statistical analysis

Statistical analysis was performed by GraphPad Prism version 6.04 for Windows (GraphPad Software). Significance was determined using a Student's t test. Model discrimination for dose-response curves was conducted using the Extra sum-of-squares F-test.

Results

TSH and IGF-1 synergistically stimulate HA secretion

To assess possible TSHR and IGF-1R cross talk, we compared the effects of individual receptor activation by their cognate ligands with simultaneous activation of the two receptors with concurrent treatment by TSH and IGF-1 (Figure 1). Both TSH and IGF-1 stimulated monophasic dose-dependent increases in HA secretion when GOFs were treated with them alone [the EC₅₀s were 150 ± 21 nM (mean ± SD) and 0.45 ± 0.15 nM for TSH and IGF-1, respectively], and the maximal increases over basal in HA secretion were 320 ± 52 and 430 ± 65 μg/mL for TSH and IGF-1, respectively. Combined TSH and IGF-1 treatments synergistically up-regulated HA secretion. Compared with TSH alone, the presence of increasing doses of IGF-1 increased the potency of TSH to maximal 19-fold (EC₅₀ was 7.8 ± 1.9 nM) in the presence of the highest doses of IGF-1. There was no effect of TSH on the potency of IGF-1. Moreover, there was a synergistic increase in the maximal stimulation of HA secretion when the highest doses of TSH and IGF-1 were combined to 1300 ± 95 μg/mL. Experiments were done in two strains of GOFs for these experiments, and we found no differences in the responses to TSH and maximally effective IGF-1 when we studied three strains of GOFs individually (Supplemental Figure 1).

Figure 1.

TSH and IGF-1 synergistically stimulate HA secretion. Strains 1 and 2 GOFs were stimulated to secrete HA as described in Materials and Methods. The cells were incubated in medium containing various concentrations of TSH and IGF-1. After 5 days, the mediums were collected, and HA was assayed by our modified ELISA. The data points represent the mean ± SE of four independent experiments. Open and closed symbols indicate different IGF-1 concentrations for the TSH dose-response curves. Solid and dotted lines show curve fits to a monophasic sigmoidal dose response. The EC₅₀s for TSH decreased from 150 ± 21 nM in control cultures to 7.8 ± 1.9 nM in cultures exposed to the highest concentrations of IGF-1 (P < .001). The EC₅₀s for IGF-1 did not change (0.45 ± 0.15 nM). Total number of samples, n = 192.

To determine whether the receptors stimulated HA secretion using parallel pathways or whether TSHR or IGF-1R was a downstream signaling partner of the other, GOFs were pretreated with either TSHR or IGF-1R small molecule antagonists before exposure to maximally effective doses of TSH and/or IGF-1 (Figure 2). The TSHR antagonist C1 (NCGC00229600) had no effect on basal secretion, whereas the IGF-1R antagonist linsitinib had a small inhibitory effect. C1 fully inhibited HA secretion stimulated by TSH (100% ± 1.5%) and had a partial inhibitory effect on IGF-1 (32% ± 1.5%). Similarly, linsitinib completely inhibited IGF-1 stimulation (94% ± 1.2%) and only partially inhibited TSH induction (44% ± 4.0%). C1 (63% ± 1.0%) and linsitinib (67% ± 1.0%) partially inhibited combined TSH and IGF-1 treatment. Neither C1 nor linsitinib affected secretion of HA stimulated by other cytokines that also signal via receptor tyrosine kinases (Supplemental Figure 2), showing that the TSHR and IGF-1R antagonists we used were specific inhibitors of these receptors (although linsitinib is known to inhibit the insulin receptor tyrosine kinase also). These data are consistent with the idea that both TSHRs activated by TSH and IGF-1R activated by IGF-1 exhibit bidirectional cross talk but that both receptors when activated by their cognate ligands stimulate HA secretion by independent pathways as well.

Figure 2.

Inhibition of TSH- and IGF-1-stimulated HA secretion by TSHR antagonist C1 and IGF-1R antagonist linsitinib. Strains 1, 2, and 3 GOFs were cultured as described in the legend to Figure 1 in medium containing TSH (1.8 μM), IGF-1 (13 nM), C1 (10 μM), and/or linsitinib (Lins; 10 μM). Bars represent the mean ± SEM for six independent experiments. *, P < .03, **, P < .001 compared with control. Total number of samples, n = 146.

M22 stimulation of HA secretion is mediated by both TSHR and IGF-1R

Although it was possible for TSHR and IGF-1R to work together to regulate HA, whether this is the case for GD-IgG activation in the pathogenesis of GO is not known. To this end, we used the stimulatory monoclonal antibody M22 to model the effects of GD-IgG. In a typical experiment, HA induction by maximally effective doses of M22 (540 ± 64 μg/mL) approximated the additive effects of TSH alone (330 ± 90 μg/mL) and IGF-1 alone (350 ± 60 μg/mL) and was similar to that of combined TSH plus IGF-1 (650 ± 58 μg/mL). This could be interpreted as simultaneous, parallel activation of TSHR and IGF-1R pathways. However, unlike combined TSH and IGF-1, M22 stimulated HA secretion by GOFs in a biphasic dose-dependent manner with EC₅₀s of 0.010 nM and 0.78 nM (Figure 3). The high potency phase accounted for approximately 30% of the maximal response and was eliminated with linsitinib pretreatment. The resulting monophasic dose response had an EC₅₀ of 0.93 nM. Simultaneous treatment with IGF-1 also resulted in a monophasic dose response to M22 with an EC₅₀ of 0.16 nM and baseline 3-fold higher than control. Maximal stimulation, however, was not different from M22 alone. Hence, the M22 effect appears to be mediated by TSHR and IGF-1R.

Figure 3.

Effects of IGF-1 and the IGF-1R antagonist linsitinib on M22-stimulated HA secretion. Strains 1, 2, and 3 GOFs were cultured as described in the legend to Figure 1 in medium containing various concentrations of M22; M22 and IGF-1 (13 nM); or M22, IGF-1, and linsitinib (Lins; 10 μM). The best fit curve for M22 was biphasic (F-test, P < .0001, EC₅₀s 0.010 nM and 0.78 nM). The best fit curves for M22+IGF-1 and M22+IGF-1+Lins were monophasic (EC₅₀s 0.16 nM and 0.93 nM, respectively). Data points represent mean ± SEM. For M22+IGF-1 and M22+IGF-1+Lins, data were from six independent experiments. For M22-control, data were from 12 independent experiments. Total number of samples, n = 328.

Because previous reports assert that M22 binds to and activates IGF-1R, we tested the ability of M22 to stimulate IGF-1R autophosphorylation, a primary effect of IGF-1R activation. As expected, IGF-1 increased IGF-1R phosphorylation 6.4- ± 1-fold and linsitinib abolished this effect. Unlike with HA secretion, C1 did not significantly change the IGF-1 effect. In contrast, M22 treatment had no effect on IGF-1R phosphorylation (Figure 4). Therefore, the linsitinib-sensitive phase of the M22 dose response was not the result of direct activation of IGF-1R by M22.

Figure 4.

Effects of IGF-1 and M22 on IGF-1R stimulation. Strains 1 and 2 were treated with IGF-1 (13 nM) or M22 (2 nM) for 30 minutes with or without C1 or linsitinib (Lins). M22 did not affect IGF-1R phosphorylation in contrast to IGF-1, which increased IGF-1R phosphorylation approximately 6.4-fold. IGF-1 stimulation was not affected by C1 but was completely inhibited by Lins. Bars represent mean ± SEM from six independent experiments, except for Lins, which was from two experiments. Total number of samples, n = 42.

We next examined the effects of TSHR antagonist C1 on M22 stimulation of HA secretion by GOFs (Figure 5). Compared with linsitinib, which inhibited M22 stimulation by only 27% ± 9.7%, C1 abolished M22 stimulation (inhibition was 100% ± 4%). These data are inconsistent with a process in which TSHRs and IGF-1Rs are simultaneously activated. Rather, M22 activation of TSHR most likely initiates two signaling pathways: a major one that relies solely on TSHR and a minor one based on TSHR-dependent activation of IGF-1R. HA secretion by a second monoclonal TSHR-stimulating antibody MS1 (19) was also partially inhibited by linsitinib and totally inhibited by C1 (Supplemental Figure 3), indicating that indirect IGF-1R activation by M22 is not unique.

Figure 5.

Inhibition of M22-stimulated HA secretion by a TSHR antagonist C1 and an IGF-1R antagonist linsitinib. Strains 1, 2, and 3 GOFs were cultured as described in the legend to Figure 1 in medium containing M22 (2 nM) (control), C1 (10 μM), and/or linsitinib (10 μM). Bars represent mean ± SEM from at least four independent experiments. *, P < .01, **, P < .001 compared with control. Total number of samples, n = 60.

Discussion

Our goal in this study was to determine the roles of TSHRs and IGF-1Rs in the stimulated secretion of HA by GOFs to gain insight into the mechanism by which GD-IgGs cause GO. We used HA secretion by GOFs as an end point because it is a major component in the pathogenesis of GO (1). We treated GOFs with TSH or IGF-1 to selectively activate TSHR or IGF-1R, respectively, and with a combination of both to determine whether cross talk between these receptors occurred. Treatment of GOFs with TSH or IGF-1 stimulated HA secretion and simultaneous treatment with TSH and IGF-1 acted synergistically, wherein increasing IGF-1 concentrations augmented the potency and efficacy of TSH. To our knowledge, synergism in which the presence of a growth factor increases the effective potency of a ligand to its cognate receptor has not been previously described. Of note, full HA induction occurred only in metabolically active cells, requiring the use of media containing 10% FBS. These media likely contain nanomolar concentrations of IGF-1 (20); therefore, it is possible that basal IGF-1R activity is necessary for TSHRs to induce HA. Given the ubiquitous expression of IGF-1R, cooperative signaling may occur in every tissue in which these two receptors are present with tissue-specific biological outcomes. In thyroid tissue, this outcome is increased proliferation, for orbital tissue, HA secretion.

The monoclonal TSHR stimulatory antibody M22 was used as a surrogate for polyclonal GD-IgGs to determine whether a single antibody could activate both receptors. This was not the case because M22 did not change the phosphorylation state of IGF-1R. Prior studies indirectly demonstrated IGF-1R activation by M22 by studying IGF-1R downstream signaling events rather than the receptor itself (7, 11). However, phosphatidylinositol 3-kinase-Akt and MAPK signaling pathways activated by IGF-1R are also downstream of TSHRs, and M22 effects could have been mediated through that receptor. Because earlier experiments were conducted in systems in which both receptors were present and M22 undeniably binds to TSHRs, whether its effects are in part mediated by IGF-1R could not have been determined. In our investigation, the biphasic dose dependence of M22 stimulation of HA secretion was shown to be in part dependent on IGF-1R activation. We found that an IGF-1R-selective kinase antagonist linsitinib, which abolished IGF-1 stimulation of HA secretion, inhibited the higher potency component of the biphasic dose response to M22 but did not inhibit the lower potency HA secretion. By contrast, a selective TSHR antagonist C1 abolished M22 stimulation. These data support a process wherein M22 activates TSHRs, leading to the stimulation of IGF-1R pathways and synergistic increase HA secretion but does not address whether M22 directly binds to and activates IGF-1R.

To address the question as to whether M22 binds to and thereby activates IGF-1R, we measured IGF-1R phosphorylation, which is the initial step in activation of IGF-1R by IGF-1 (21, 22). As expected, IGF-1 stimulated IGF-1R phosphorylation in GOFs and linsitinib abolished this effect, but C1 did not inhibit IGF-1-stimulated IFG-1R phosphorylation. By contrast, M22 did not stimulate IGF-1R phosphorylation. We can conclude therefore that M22 does not directly activate IGF-1R. Because IGF-1-stimulated IGF-1R phosphorylation was not inhibited by C1 but IGF-1 stimulation of HA secretion was partially inhibited by the TSHR-selective antagonist, it appears that cross talk between TSHR and IGF-1R by M22 in GOFs does not require direct activation of IGF-1R.

Cross talk between GPCRs and RTKs is well documented, with multiple mechanisms occurring between different receptors in different cells (16, 17) including one-way models in which GPCRs influence RTKs or vice versa and more rarely bidirectional systems in which both receptors affect each other. Our results are consistent with a bidirectional model between TSHR and IGF1-R based on the findings that the stimulation of HA secretion by TSH or M22 was partially inhibited by linsitinib and that IGF-1 stimulation was similarly suppressed by the TSHR antagonist C1. Our data also support a model by which IGF-1R and TSHR can independently stimulate HA production in GOFs because both linsitinib and C1 completely abolished HA secretion when only their cognate receptors were activated. Previous studies have reported the stimulation of HA secretion by IGF-1 in cells that do not express TSHR (23), and it is likely that IGF-1R in GOFs may use a similar pathway. In addition, the similar results seen with another, albeit less effective, stimulatory antibody MS1 demonstrate that this pathway may be common to stimulatory GD-IgG. It has already been shown by coimmunoprecipitation that TSHRs and IGF-1Rs are present in close proximity in the cell surface membrane (24). Because it is unlikely that both M22 and MS1 bind to TSHRs the same way, the plausible explanation is that TSHRs and IGF-1Rs are in the same signaling complex.

Previous investigators have interpreted the finding that an antibody 1H7, which inhibits IGF-1 binding to the α-subunit of IGF1-R, inhibited activation of IGF1-R by GD-IgG (7, 11) as evidence of GD-IgG directly binding to and activating IGF1-R. These findings, however, do not allow for this conclusion. 1H7 is an inverse agonist that inhibits IGF-1-independent (or basal) activity as well as IGF-1-dependent activation (25), and therefore the effect of 1H7 to apparently inhibit GD-IgG activation of IGF-1R may have been caused by 1H7 inhibition of the agonist-independent activity of IGF1-R or activation mediated by GD-IgG-activated TSHR. Also, it is possible that binding of 1H7 to IGF-1R might sterically interfere with M22 binding to TSHRs. A second monoclonal antibody that inhibits binding of IGF-1 to the IGF-1R (26) has been shown to inhibit TSH stimulation also (27), but whether it is an inverse agonist has not been reported. The IGF-1R antagonist that we used is a small molecule that binds to the active tyrosine kinase site of the intracellular portion of IGF-1R and therefore would not interfere with GD-IgG or M22 binding. It is important to note that our findings do not exclude the possibility that there are antibodies within GD-IgGs that bind to and activate IGF-1Rs. However, it is clear that binding to IGF-1R is not necessary for its involvement in the pathogenesis of GO.

In conclusion, we have shown that the activation of both TSHR and IGF-1R leads to synergistic stimulation of HA secretion by GOFs and that activation by a monoclonal stimulatory antibody M22 generated from a patient with GD uses this dual-signaling cascade. We suggest that GD-IgG, like M22 and MS1, activate TSHR and IGF-1R cross talk and that this cross talk plays a major role in the pathogenesis of GO. The sequence of molecular events leading to this cross talk is beyond the scope of this work and warrants further investigation. Nevertheless, this work strongly argues that the inhibition of TSHR would be an effective strategy for GO; however, a combination therapy with antagonists to both receptors may be more effective.