Interactions between the mannose receptor and thyroid autoantigens

Abstract

Thyroid autoantigens require internalization and processing by antigen-presenting cells to induce immune responses. Besides pinocytosis, antigen uptake can be receptor-mediated. The mannose receptor (ManR) has a cysteine rich domain (CR) and eight carbohydrate recognition domains (CRD) that bind glycosylated proteins. The TSH receptor (TSHR), thyroid peroxidase (TPO) and thyroglobulin (Tg) are glycoproteins. To investigate a role for the ManR in thyroid autoimmunity, we tested the interaction between these autoantigens and chimeric ManRs. Plasmids encoding the CR-domain linked to IgG-Fc (CR-Fc) and CDR domains 4–7 linked to IgG-Fc (CDR4-7-Fc) were expressed and purified with Protein A. Enzyme-linked immunosorbent assay (ELISA) plates were coated with human thyroid autoantigens and CR-Fc or CRD4-7-Fc binding detected with peroxidase-conjugated anti-IgG-Fc. CRD4-7-Fc binding was highest for the TSHR, followed by Tg and was minimal for TPO. CR-Fc bound to Tg but not to TSHR or TPO. The interaction between the TSHR and CRD-Fc was calcium-dependent; it was inhibited by mannose (not galactose), and required a glycosylated TSHR A-subunit. Moreover, precomplexing the TSHR A-subunit with CRD-Fc (but not CR-Fc), or adding mannose (but not galactose), decreased in vitro responses of splenocytes from TSHR-immunized mice. Our data indicate that the ManR may participate in autoimmune responses to Tg and the TSHR but not to TPO. Most important, ManR binding of heavily glycosylated TSHR A-subunits suggests a mechanism by which the minute amounts of A-subunit protein shed from the thyroid may be captured by antigen-presenting cells located in the gland or in draining lymph nodes.

Introduction

Autoantigens, like other antigens, require uptake, internalization and processing by antigen-presenting cells (APC) for presentation to T cells. Soluble antigen uptake by macrophages can occur by fluid phase internalization (pinocytosis). In addition, glycoproteins can be internalized following binding to cell surface mannose receptors (ManR) [1]. Compared with pinocytosis, ManR-mediated uptake enhances markedly the efficacy of T cell responses (for example [2]. The ManR receptor is a calcium-dependent lectin comprising an amino-terminal cysteine rich (CR) domain, eight carbohydrate recognition domains (CRD) and transmembrane and intracellular domains [3](Fig. 1a). The CR domain binds to sulphated carbohydrate side-chains [4], while the CRDs interact with sugars such as mannose, fucose and N-acetylglucosamine but not galactose [5].

Fig. 1.

Structure of the mannose receptor (a) and the chimeric proteins CR-Fc (b) and CDR4-7-Fc (c). CRDs, carbohydrate recognition domains; CR, cysteine-rich domain; FNII, fibronectin type II domain; TM, transmembrane region; CT, cytoplasmic tail. Adapted with permission from Linehan et al. [12] and Martinez-Pomares et al. [13], with kind permission from the publishers: WILEY-VCH Verlag GmbH & Co. KGaA [12]; and reproduced from the Journal of Experimental Medicine 1996, 184: 1927–37, copyright permission of The Rockefeller University Press.

Thyroid autoimmunity involves immune responses to three autoantigens, the thyrotrophin receptor (TSHR), thyroid peroxidase (TPO) and thyroglobulin (Tg). TSHR autoantibodies that stimulate the thyroid gland are the direct cause of Graves’ hyperthyroidism (reviewed in [6]). Besides being markers of thyroid autoimmunity, autoantibodies to TPO and Tg are associated with lymphocytic infiltration of the thyroid and thyrocyte destruction in Hashimoto's thyroiditis (reviewed in [6]). Tg is a soluble protein while TPO and the TSHR are membrane-associated molecules present at the apical and basal surfaces of thyroid epithelial cells, respectively. On reaching the thyrocyte surface, many TSHR undergo intramolecular cleavage to form an A-subunit tethered by disulphide bonds to the membrane-spanning B-subunit (reviewed in [6]). Subsequent to cleavage, some A-subunits may be shed from the cell surface. Although the intact TSHR is required for thyroid cell stimulation by TSH or TSHR autoantibodies, recent evidence suggests that the free A-subunit is the major autoantigenic component responsible for initiating or amplifying the autoimmune response to the TSHR in Graves’ hyperthyroidism [7,8].

The TSHR, TPO and Tg are all glycoproteins (for example [9–11]). For these reasons, we considered the possibility that the ManR on APC could be involved in the autoimmune response to thyroid antigens. Indeed, there is prior evidence that porcine Tg binds to the ManR [12]. However, there are no available data on TSHR or TPO binding to the ManR. To study interactions between all three human thyroid autoantigens and the ManR, we used soluble, recombinant forms of the two extracellular components of the ManR, namely the CR [13] and the CRD containing repeats 4–7 (CRD4-7) [12]. To facilitate secretion, solubility and purification, each ManR component was expressed as a fusion protein linked to the Fc region of human IgG, namely CR-Fc (Fig. 1b) and CRD4-7-Fc (Fig. 1c). Our data revealed that TPO did not bind to the ManR and we obtained new information regarding the Tg–ManR interaction. In particular, we observed an interaction between the TSHR A-subunit and the carbohydrate recognition domain of the ManR.

Methods and materials

Mannose receptor Fc-fusion proteins

The plasmids encoding CRD4-7-Fc [12] and CR-Fc [13] (Drs Louisa Martinez-Pomares and Siamon Gordon, Oxford, UK) were expressed in 293T cells (American Type Culture Collection, with the agreement of Dr Michele Calos, Stanford, CA, USA). Fusion proteins were affinity-purified using protein A as described previously [12]. Some purified CR-Fc was also made available to us by Dr Martinez-Pomares.

Thyroid autoantigens and other proteins

Recombinant TSHR A-subunit is a variant of the receptor expressed in Chinese hamster ovary (CHO) cells (TSHR-289) [14]. The protein was isolated from culture medium by affinity chromatography using mouse MoAb 3BD10 [14,15], dialysed against 50 mm NaCl, 10 mm Tris, pH 7·4 and its purity and concentration determined by sodium dodecyl sulphate– polyacrylamide gel electrophoresis (SDS-PAGE). In some experiments, we used enzyme-linked immunosorbent assay (ELISA) plates coated with the porcine TSHR (holoreceptor) that are used to detect TSHR autoantibodies by the ability of the latter to inhibit radio-labelled TSH binding to the receptor (Kronus, Boise, ID, USA).

Recombinant human TPO, secreted by CHO cells, was affinity purified from culture medium as reported [16] and buffer exchanged to phosphate-buffered saline using Centricon 50 units (Millipore, Billerica, MA, USA). The concentration of TPO was determined by spectrophotometry at OD 280 nm (extinction coefficient 17·9) and its purity was assessed by SDS-PAGE. Human Tg was purchased (Calbiochem, La Jolla, CA, USA) and was 96% pure. Bovine serum albumin (BSA), mannose and galactose were obtained from Sigma Chemical Co. (St Louis, MO, USA).

Deglycosylation of human thyroid autoantigens

Purified TSHR A-subunit, TPO and Tg were deglycosylated with endoglycosidase F according to the manufacturer's protocol (New England Laboratories, Beverly, MA, USA). TSHR A-subunit (0·5 mg), TPO and Tg (∼ 9 µg each) were denatured (10 min, 95°C) in 0·05% SDS and 1%β-mercaptoethanol. After cooling on ice, the proteins were digested overnight with endoglycosidase F (100 units per µg) in the presence of 50 mm Nonidet P-40 (Sigma). A second aliquot of TSHR A-subunit (0·5 mg) was denatured and treated with 50 mm Nonidet P-40 without enzyme to provide control denatured protein. Denatured and deglycosylated TSHR A-subunit proteins were dialysed extensively against phosphate buffered saline to remove β-mercaptoethanol and permit their use in inhibition studies (see below). Aliquots of TSHR A-subunit, TPO and Tg before and after digestion were re-suspended in Laemmli buffer and subjected to SDS-PAGE (12% for TSHR A-subunit, 10% for Tg and 7·5% for TPO) under reducing conditions followed by Coomassie Blue staining.

Binding of CRD4-7-Fc and CR-Fc to thyroid autoantigens in ELISA

ELISA wells were coated by overnight incubation (4°C) with Tg (2 and 8 µg/ml), TSHR A-subunit (1 and 4 µg/ml) or TPO (1 and 4 µg/ml) diluted in 50 mm NaCl, 10 mm CaCl₂, 10 mm Tris, pH 7·5 (Tris-Na-Ca buffer). Wells were rinsed, blocked with BSA (5% in Tris-Na-Ca buffer) and exposed to CRD4-7-Fc or CR-Fc (4·0, 0·4 or 0·04 µg/ml) in Tris-Na-Ca buffer (45 min, 37°C). After washing (Tris-Na-Ca buffer + 0·05% Tween 20, Sigma), horseradish peroxidase conjugated antihuman IgG-Fc (Caltag, Burlingame, CA, USA) was added (45 min, 37°C). Colour was developed with o-phenylene diamine, the reaction terminated with H₂SO₄ (diluted 20% in water) and colour read at 490 nm. A similar approach was used for ELISA wells coated with porcine TSHR (see above).

Parallel studies were performed with CaCl₂ omitted from the coating buffer, washing buffer and diluent for the ManR-chimeric proteins. In other studies, EDTA (final concentration 50 mm) was included in the Tris-Na-Ca buffer for CRD4-7-Fc or CR-Fc binding to antigen. Specificity was determined by comparing the effect of soluble mannose or galactose (0·4–10 mm) on binding by CRD4-7-Fc to TSHR A-subunit coated wells. A similar approach was used to determine the effect of soluble denatured or deglycosylated TSHR A-subunit to the binding of CRD4-7-Fc to TSHR A-subunit coated wells. For these studies, CRD4-7-Fc was preincubated for 30 min at room temperature with 1–10 µg denatured or deglycosylated TSHR A-subunits prior to addition of the mixture to the ELISA wells.

ManR involvement in TSHR antigen-induced responses in immunized mice

Female BALB/c mice (∼6 weeks old; Jackson Laboratories, Bar Harbor, ME, USA) were injected intramuscularly (10¹¹ particles) on three occasions at 3-weekly intervals with adenovirus expressing the TSHR or control adenovirus expressing β-galactosidase, as described previously [8]. Mice were euthanized 8 weeks after the third injection and spleens were removed to prepare splenocyte suspensions. Animal studies were approved by the local Institutional Animal Care and Use Committee and performed in accordance with the highest standards of care in a pathogen-free facility.

Recall responses were studied in vitro as described previously for adenovirus immunized mice [17], with the following modifications to block A-subunit binding to the ManR: (i) using mannose (or galactose as control) together with TSHR A-subunit; and (ii) precomplexing A-subunit with CRD4-7-Fc (or CR-Fc) before addition to splenocyte cultures. For the first approach, splenocytes (duplicate 200 µl aliquots; ∼5 × 10⁵ cells) were incubated with A-subunit protein (4 µg/ml) alone or together with 10 mm mannose or galactose (final concentration). In the second approach, TSHR A-subunit protein (8 µg) was preincubated (30 min at 20°C) in culture medium (50 µl; see below) alone or together with CRD4-7-Fc (16 µg) or an equimolar amount of CR-Fc (8 µg). In addition, splenocytes were preincubated with purified rat antimouse CD16/CD32 (2·4G2; 2 µg/10⁶ cells, BD Pharmingen, San Diego, CA, USA), an antibody shown to block Fc receptors on murine lymphocytes and macrophages [18]. Subsequently, precomplexed TSHR A-subunit was added to these Fc receptor blocked splenocytes (∼5 × 10⁵ cells, as above) to give a final concentration of 4 µg/ml A-subunit protein. Culture medium was RPMI-1640, 10% heat-inactivated fetal bovine serum, 2 mm glutamine, 1 mm sodium pyruvate, 50 µg/ml gentamycin, 50 µmβ-mercaptoethanol and 100 units/ml penicillin. Splenocyte cultures were incubated for 5–6 days (37°C, 5% CO₂), centrifuged to remove cell debris and supernatants assayed for interferon (IFN)-γ by ELISA (100 µl aliquots) using capture and biotinylated detection antibodies from BD Pharmingen. IFN-γ production is reported as pg/ml calculated from recombinant IFN-γ standards (BD Pharmingen).

Results

Glycan content of the TSHR A-subunit, TPO and Tg

Before studying their interactions with the ManR chimeric proteins, we compared the carbohydrate contents of the three thyroid autoantigen preparations. Because of their size differences, each autoantigen was examined on a different gel (TSHR A-subunit 12%, Tg 10% and TPO 7·5%). Removal of N-linked carbohydrate side chains by endoglycosidase F reduced the sizes of TPO, Tg and the TSHR A-subunit to values consistent with the molecular masses of their protein backbones. In particular, enzymatic deglycosylation converted the TSHR A-subunit from a broad ∼60 KDa band to a sharp 34 kDa band (Fig. 2, left panel), as observed previously [9,15]. The molecular weight differences between untreated and endoglycosidase F digested samples were less marked for Tg and TPO (Fig. 2, middle and right panels). From these data, carbohydrate constitutes ∼43% of the TSHR A-subunit versus∼12% and ∼10% for Tg and TPO, respectively. These values are consistent with previous findings of ∼30% or more carbohydrate in the full-length TSHR [19], and for the lesser proportion of carbohydrate in Tg (8–10%) [20] and recombinant or human TPO (12% and 14%, respectively) [16,21].

Fig. 2.

Enzymatic deglycosylation of the three major thyroid autoantigens. TSHR A-subunit and TPO ectodomain (purified from transfected CHO cell cultures) and Tg (Calbiochem) were digested with endoglycosidase F (see Methods). Aliquots of each protein (∼ 5 µg) before and after digestion were subjected to SDS-PAGE (12% for TSHR A-subunit, 10% for Tg and 7·5% for TPO) under reducing conditions followed by staining with Coomassie blue. The solid arrows indicate the size of each protein before and after deglycosylation. In the TSHR A-subunit panel, the dotted arrow indicates endoglycosidase F.

Binding of thyroid autoantigens to mannose receptor CR and CRD components

To examine these interactions, CR-Fc and CRD4-7-Fc were applied to ELISA wells coated with human Tg, recombinant human TSHR A-subunit or recombinant human TPO. BSA-coated wells provided negative controls. Purified CR-Fc did not bind to the TSHR A-subunit, TPO or BSA. In contrast, similar CR-Fc concentrations (4 and 0·4 µg/ml) provided strong signals on Tg-coated wells (Fig. 3a). At 0·04 µg/ml, CR-Fc binding to Tg approached background values.

Fig. 3.

Interaction between CRD4-7-Fc or CR-Fc and the TSHR A-subunit, Tg and TPO. For both panels, ELISA wells were coated with two concentrations of each autoantigen and binding was tested using three concentrations of CRD4-7-Fc or CR-Fc. Data are reported as the OD 490 nm (mean + s.e.m., n = 3). (a) CR-Fc binds to ELISA wells coated with Tg but not to wells coated with the TSHR A-subunit, TPO or (as a control) BSA. (b) CRD4-7-Fc binds to ELISA wells coated with the TSHR A-subunit and Tg but not to TPO or BSA.

A different profile of thyroid autoantigen binding by the ManR carbohydrate recognition domain was observed. Consistent with findings from a previous study using a different methodology [12], CRD4-7-Fc bound to Tg (Fig. 3b). However, in the same assay, we observed even higher binding between CRD4-7-Fc and the TSHR A-subunit. Thus, at an intermediate concentration (0·4 µg/ml), CRD4-7-Fc binding was clearly evident to the TSHR A-subunit but not to Tg. Again, as for the ManR cysteine rich domain, CRD4-7-Fc bound to neither TPO nor BSA (Fig. 3b). It should be emphasized that in these assays, CRD4-7-Fc binding was studied in the presence of CaCl₂ (final concentration 10 mm).

TSHR binding to the mannose receptor CRD requires calcium

The ManR is a calcium-dependent lectin (see Introduction). To determine whether the TSHR A-subunit interaction with the ManR CRD involved the lectin properties of the latter, we repeated the binding studies with and without calcium. As observed previously using a calcium buffer (Fig. 3), CRD4-7-Fc bound in a dose-dependent manner to recombinant TSHR A-subunit (Fig. 4a). The addition of EDTA abrogated binding to background levels (Fig. 4b). Similar results were obtained when calcium was omitted from the binding buffer (data not shown). Previously, the TSHR A-subunit was not found to bind to CR-Fc in buffer containing calcium (Fig. 3a). The addition of EDTA did not alter this lack of binding (Fig. 4a,b).

Fig. 4.

Binding of CRD4-7-Fc to the TSHR requires calcium. ELISA wells were coated with recombinant TSHR A-subunit (a, b) or the porcine TSH holoreceptor (c, d). Binding of CRD4-7-Fc or CR-Fc was studied in the presence of 10 mm CaCl₂ (a, c) or the presence of 50 mm EDTA (b, d). Data are reported as OD 490 nm (mean + s.d., n = 2).

In addition to the recombinant human TSHR A-subunit, we studied the interaction between CRD4-7-Fc and TSH holoreceptors extracted from porcine thyroids. In the presence of calcium, CRD4-7-Fc bound strongly to ELISA wells coated with porcine TSHR and this binding was abrogated by EDTA (Fig. 4c,d). As for TSHR A-subunits, CR-Fc binding to porcine TSH holoreceptors was minimal, regardless of the presence of calcium or EDTA.

Specificity of TSHR A-subunit binding to the mannose receptor CRD

We investigated the specificity of TSHR A-subunit binding to the ManR carbohydrate recognition domain in two ways. First, we tested the ability of deglycosylated A-subunit to inhibit this interaction. Effective enzymatic deglycosylation requires protein denaturation. Increasing concentrations of denatured A-subunits inhibited CRD4-7-Fc binding to plate-bound TSHR A-subunits (Fig. 5a). In contrast, denatured, deglycosylated A-subunits were devoid of inhibitory activity. Secondly, we examined the sugar specificity of the interaction between TSHR A-subunits and the ManR CRD. Increasing concentrations of mannose progressively reduced the interaction between the A-subunit and CRD4-7-Fc and at 10 mm mannose, binding was totally inhibited (Fig. 5b). Addition of similar concentrations of galactose to the buffer had no effect on the TSHR A-subunit interaction with the ManR CRD. These findings for the TSHR A-subunit are similar to earlier observations for Tg, namely complete inhibition by mannose (but not galactose) of Tg binding to the soluble mannose receptor [12].

Fig. 5.

Specificity of TSHR A-subunit binding to the mannose receptor CRD. ELISA wells were coated with native, glycosylated TSHR A-subunit protein (4 µg/ml). For both panels, data are shown as OD 490 nm (duplicates + s.d.); *P < 0·05; buffer versus denatured TSHR A-subunit (10 µg/ml) or mannose (10 mm), Kruskal–Wallis one-way analysis of variance. (a) CRD4-7-Fc (0·4 µg/ml) was preincubated (30 min at room temperature) with buffer alone (‘0’) or with the indicated concentrations of either denatured TSHR A-subunits or denatured, deglycosylated TSHR A-subunits. The mixtures were then added to the ELISA wells and CRD4-7-Fc binding detected with conjugated anti-IgG-Fc. (b) CRD4-7-Fc (0·4 µg/ml) was mixed with buffer alone (‘0’) or with the indicated concentrations of either mannose or galactose. The mixtures were then added to the ELISA wells and CRD4-7-Fc binding detected with conjugated anti-IgG-Fc.

ManR and splenocyte responses to TSHR A-subunit

Further investigations of a role for the ManR were performed using splenocytes from BALB/c mice immunized with TSHR-adenovirus that, as described previously [17], produce IFN-γ following in vitro challenge with TSHR A-subunit protein. We used two different approaches. First, adding mannose (10 mm) to splenocytes from a TSHR-immunized mouse reduced by approximately 75% the IFN-γ response to TSHR A-subunit, whereas the same concentration of galactose had a minimal effect (Fig. 6a). Secondly, we tested the outcome of precomplexing TSHR A-subunit protein with CRD4-7-Fc, CR-Fc or culture medium alone before addition to splenocytes. To block Fc-receptor binding by ManR chimeras, all splenocytes were treated with an appropriate antibody (see Methods) that was present throughout the culture period. Compared with splenocytes challenged with TSHR A-subunit alone, preincubation with CRD4-7-Fc (but not CR-Fc) decreased the amount of IFN-γ secreted by approximately 75% (Fig. 6b). The data shown are representative of two separate experiments. Splenocytes from control-adenovirus immunized mice did not respond to TSHR A-subunit (data not shown).

Fig. 6.

Interactions between the TSHR A-subunit and mannose receptors tested in memory responses of splenocytes from TSHR-immunized mice. Data for both (a) and (b) are shown as the mean (+ range) of duplicate cultures with or without TSHR A-subunit (4 µg/ml). (a) Inclusion of mannose (10 mm) decreases IFN-γ secretion following challenge with A-subunit protein. In contrast, inclusion of galactose (also 10 mm) had a minimal inhibitory effect. (b) Precomplexing TSHR A-subunit protein with CRD4-7-Fc (but not CF-Fc) decreases IFN-γ production by splenocytes from TSHR-adenovirus (TSHR-Ad) immunized BALB/c mice.

Discussion

Antigen capture, processing and presentation to T cells is a vital component of the normal immune response, as well as in the pathogenesis of autoimmune disease. Macrophages and dendritic cells, professional antigen-presenting cells (APC), can effectively engulf particulate fragments of dead cells and therefore do not necessarily require the ManR on their surface for this purpose. Soluble proteins present at high concentration in the extracellular fluid can be captured by fluid phase pinocytosis. However, the ManR is important for the capture of glycosylated particulate antigens (microorganisms) as well as soluble antigens [3], particularly if they are present at low concentrations. The extracellular region of the ManR comprises two components: (i) the cysteine rich (CR) component interacts primarily with proteins containing sulphated carbohydrate moieties; and (ii) the carbohydrate recognition domain (CRD) contains multiple repeats and functions as a lectin (requiring divalent cations for glycan binding). In the present study, we used recombinant forms of each ManR component to study their interaction with the major thyroid autoantigens, Tg, TPO and the TSHR. All three autoantigens are glycoproteins [9–11].

Tg is the most abundant thyroid component; it is a soluble protein and it is readily detectable in serum (for example [22]. Therefore, ManR-mediated enhanced Tg uptake may not be necessary for immune responses to this autoantigen. Nevertheless, as described previously for porcine Tg [12], our present data indicate that human Tg bound to the CRD of the ManR. A novel observation is that, consistent with sulphation of its glycan moieties [23,24], human Tg also binds to the CR-ManR component. Therefore, Tg can be added to the endocrine-associated proteins (including lutropin and thyroid stimulating hormone) that interact with both the CR and CRD domains of a mannose receptor family member [25]. Lutropin binding to this receptor on hepatic endothelial cells is involved in regulation of its half-life in the circulation [4,26]. Of interest, the ability of Tg to interact with both the CRD- and CR- domains of the ManR resembles observations for the organism Klebsiella pneumoniae [27]. Regardless of the relative importance of the ManR versus pinocytosis in autoimmunity to Tg, macrophages and/or dendritic cells play an important role in presenting Tg in vivo in rats [28] and in vitro in humans [29]. However, a role for macrophages/dendritic cells in autoimmunity to Tg does not preclude a role for other APC such as B cells (reviewed in [30]). Indeed, Tg antibodies can enhance or suppress processing of a pathogenic epitope to T cells [31].

TPO is less abundant in the thyroid than Tg. It is a membrane-bound protein and there is controversial evidence regarding its presence in the circulation [32,33]. Evidence from some T cell clones indicates that TPO is one of the thyroid autoantigens presented by thyroid cells that aberrantly express MHC class II (for example [34]). On the other hand, in vitro studies of patients’ lymphocytes and splenocytes from immunized mice suggest that antibodies may be involved in TPO presentation to T cells [35,36]. Irrespective of whether thyroid cells or autoantibodies (or both) play a role, our present data indicate that the mannose receptor is not involved in TPO-associated autoimmunity.

Because of the very small number of TSH receptors on thyroid cells (reviewed in [37]), a mechanism for capture and presentation of the this autoantigen is likely to be required. It has long been recognized that antigen-specific B cells capture and present minute amounts of antigen to T cells (reviewed in [30]). However, T cell responses studied in mice vaccinated with TSHR-DNA in a plasmid or an adenovirus vector do not support a role for TSHR-specific B cells [17,38,39]. Consequently, an interaction between the TSHR and the ManR is potentially important. Moreover, recent evidence indicates that, although thyroid stimulation requires the membrane-bound TSHR, initiation or amplification of the autoimmune response to the TSHR in Graves’ disease appears to involve recognition of a soluble autoantigen, the shed A-subunit. First, stimulatory TSHR autoantibodies preferentially recognize the shed A-subunit [7,9]. Secondly, Graves’ disease is induced preferentially by injecting mice with an adenovirus expressing the free TSHR A-subunit than with adenovirus encoding a receptor mutated to prevent cleavage and A-subunit shedding [8].

We observed that the ManR CRD component bound to recombinant human TSHR A-subunit as well as to the porcine TSH holoreceptor. The interaction with the membrane-bound TSHR is consistent with previous observations of CRD4-7-Fc binding to an unidentified protein present in the basal membrane of mouse thyroid tissue [12]. ManR-CRD binding to the TSHR required calcium (as expected for a C-type lectin), depended on A-subunit glycosylation and was inhibited by mannose but not galactose. Glycosylation by CHO cells is not identical to that in humans. However, recombinant TSHR A-subunits and TPO are produced in the same CHO cell line; both are glycosylated and yet only the former is recognized by the ManR. Overall, these findings, together with our observation regarding the non-recombinant porcine receptor, suggest that ManR-CRD recognition of the TSHR is not related to the cells in which the antigen is generated. In contrast, neither the TSHR A-subunit nor the TSH holoreceptor interacted with the CR domain. Complementing the information from our ELISA studies, memory responses of splenocytes from TSHR-immunized mice were inhibited strongly by precomplexing the TSHR A-subunit with ManR CRD (but not ManR CR) or by coincubating A-subunit protein with mannose (but not with galactose).

Can the TSHR A-subunit be detected in the circulation? A description of unpublished data suggesting detection of TSHR A-subunits in serum [40] has not been confirmed and seems unlikely for several reasons: (a) unlike abundant Tg, the TSHR is expressed at low levels in the thyroid and its turnover is slow; (b) the A-subunit (∼60 kDa) is likely to be filtered by the glomerulus because it is smaller than albumin; and (c) the high carbohydrate content (∼45%) of the A-subunit contributes to a high non-specific background on immunodetection. We suggest that lymphatic drainage from the thyroid to regional lymph nodes provides the route by which small amounts of soluble A-subunit can be captured by ManR-bearing APC. Alternatively, dendritic cells in human thyroid tissue [41] are positioned optimally to capture thyroid autoantigens.

In conclusion, we provide evidence that the ManR interacts with Tg and the TSHR A-subunit but not with TPO. Because of its abundance, no specialized mechanism may be required for Tg uptake by APC. Of greater importance, our data suggest a mechanism whereby minute amounts of TSHR A-subunits shed from the thyroid could be captured by ManR-bearing dendritic cells in the gland or in draining lymph nodes. Mature dendritic cells stimulate T cells but immature dendritic cells induce tolerance in naive T cells [42]. Whether the outcome of TSHR A-subunit (or Tg) capture by dendritic cells promotes tolerance or stimulation, our data suggest a role for the mannose receptor in the autoimmune response to two major thyroid autoantigens.