Probing structural variability at the N terminus of the TSH receptor with a murine monoclonal antibody that distinguishes between two receptor conformational forms.

Despite elucidation of the crystal structure of M22, a human thyroid-stimulating autoantibody (TSAb) bound to the TSH receptor (TSHR) leucine-rich repeat domain (LRD), the mechanism by which TSAs activate the TSHR and cause Graves' disease remains unknown. A nonstimulatory murine monoclonal antibody, 3BD10, and TSAb interact with the LRD N-terminal cysteine cluster and reciprocally distinguish between two different LRD conformational forms. To study this remarkable phenomenon, we investigated properties of 3BD10, which has a linear epitopic component. By synthetic peptide ELISA, we identified 3BD10 binding to TSHR amino acids E34, E35, and D36 within TSHR cysteine-bonded loop 2 (C31-C41), which includes R38, the most N-terminal contact residue of TSAb M22. On flow cytometry, despite not contributing to the 3BD10 and M22 epitopes, chimeric substitution (but not deletion) of TSHR cysteine-bonded loop 1 (C24-C29) eliminated 3BD10 binding to the TSHR ectodomain (ECD) expressed on the cell surface, as found previously for TSAb including M22. Furthermore, 3BD10 did not recognize all cell surface TSHR ECDs, consistent with recognition of only one conformational receptor form. Reversion to wild-type of small components of the loop 1 chimeric substitution partially restored 3BD10 binding to the TSHR-ECD but not to synthetic peptides tested by ELISA. Molecular modeling supports the concept that modification of TSHR C-bonded loop 1 influences loop 2 conformation as well as LRD residues further downstream. In conclusion, the present study with mouse monoclonal antibody 3BD10 confirms TSHR conformational heterogeneity and suggests that the N-terminal cysteine cluster may contribute to this structural variability.

H yperthyroidism in Graves' disease is caused by autoantibodies that mimic the action of TSH by engaging and activating the TSH receptor (TSHR) (reviewed in Ref. 1). The TSHR is unique among the glycoprotein hormone receptors in undergoing intramolecular proteolytic cleavage into two subunits, A and B, which remain linked by disulfide bonds. Some of the A subunits on the cell surface are shed, and there is evidence that this component of the TSHR is the primary immunogen leading to the generation of thyroid-stimulating autoantibodies (TSAbs) (2). Al-though a human monoclonal stimulating antibody, M22, has been crystallized in complex with the TSHR A subunit and the three-dimensional structure determined (3), the mechanism by which TSAbs such as M22 activate the TSH holoreceptor remains unknown. 3BD10 is a nonstimulatory mouse monoclonal antibody (mAb) raised to the TSHR A subunit with some striking similarities and differences to both polyclonal human TSAb and human TSAb M22. As summarized in Table 1, properties of 3BD10 that are similar to those of M22 include the fol-lowing: (i) a conformational epitope (4), (ii) steric hindrance in binding to the TSH holoreceptor relative to the glycosylphosphatidyl inositol (GPI)-anchored ectodomain (ECD) or free A subunit (5), and (iii) an epitopic component within TSHR amino acid residues 22-51 (4) at the N terminus of the TSHR (residues 1-21 being the signal peptide). This N-terminal region contains a cluster of four cysteines, now known to form two disulfide-linked loops (6), hereafter termed N-terminus loop 1 (C24-C29) and loop 2 (C31-C41) (3). M22 is known to contact TSHR residue R38 within cysteine loop 2, schematically depicted in Fig. 1. Whether the 3BD10 epitopic component lies within or downstream of the cysteine loops is not known.
The importance of the mouse mAb 3BD10 is that, despite the foregoing similarities with TSAb (including M22), these antibodies reciprocally distinguish between two conformational forms of recombinant TSHR A subunits secreted by eukaryotic cells (4,(7)(8)(9). "Active" TSHR A subunits are recognized by TSAb but not 3BD10, whereas 3BD10 binds to "inactive" but not active A subunits. Indeed, these properties permit the separate purification from conditioned culture medium of both types of A subunits secreted in approximately equal proportions (7). Denaturation (reduction and alkylation) of TSHR A subunits eliminates 3BD10 binding (4), indicating the requirement for conformationally intact antigenic protein.
The foregoing similarities and differences between TSAbs and the mouse mAb 3BD10 suggest that further characterization of 3BD10 binding to the TSHR could provide insight into the interaction of TSAb with the TSHR. Therefore, in the present study, we examined the effect on 3BD10 binding of deletion or chimeric substitution of TSHR N-terminal loop 1, the latter involving replacement of loop 1 with the corresponding LH receptor region, a modification known to influence polyclonal and monoclonal (M22) TSAb interaction with the TSHR (9 -12).

Flow cytometry
TSHR-expressed CHO cells were harvested from 10-cm dishes using 1 mM EDTA and 1 mM EGTA in PBS. After washing twice with PBS containing 10 mM HEPES (pH 7.4) and 2% fetal bovine serum, the cells were incubated for 1 h on ice in 100 l of the same buffer with or without monoclonal TSHR antibodies at the concentrations indicated in individual experiments. After rinsing, the cells were incubated for 1 h on ice with 100 l fluorescein isothiocyanate-conjugated goat antimouse IgG (1:100; Caltag, Burlingame, CA). After rinsing, fluorescence was measured using a Beckman Coulter FACScan flow cytofluorimeter (Brea, CA). Cells stained with propidium iodide (6 g/ml final concentration) were excluded from analysis.

Enzyme-linked immunosorbent assay
TSHR and TSHR variant peptides were synthesized by Peptide 2.0 (Chantilly, VA) and confirmed for size and purity (Ͼ90%) by mass spectroscopy and HPLC, respectively. ELISA wells were coated with individual peptides (10 g/ml) or with purified, recombinant TSHR A subunits (10 g/ml) (5) by incubation in 0.035 M NaHCO 3 , 0.015 M Na 2 CO 3 (pH 9.3). Wells were incubated with mAbs at the concentrations indicated in the text for individual experiments. Antibody binding was detected with horseradish peroxidase-conjugated mouse anti-IgG (Sigma Aldrich, St. Louis, MO). The signal was developed with o-phenylenediamine and H 2 O 2 and the OD read at 490 nm.

TSHR molecular modeling
The three-dimensional model of the TSHR with the TSH-LHR-A1 substitution was generated using the crystal structure of TSHR ligand binding domain (pdb code: 2XWT) as a template. The loop structure was modeled using Biopolymers module in Sybylx2.0 (Tripos, Inc., St. Louis, MO). Subsequently, to obtain a stereochemically and energetically favorable model, the chimeric model was subjected to molecular simulation. The molecular dynamics was performed using Desmond (Desmond Molecular Dynamics System, version 2.4, 2010; D. E. Shaw Research, New York, NY). The model was simulated for 12ns using the default parameters incorporated in Desmond and structure analyses was performed using Pymol (The PyMOL Molecular Graphics System, version 1.5.0.4; Schrödinger, LLC, Cambridge, MA). The quality of the model was assessed by a Ramachandran plot (19). Comparison of the wild-type TSHR leucine-rich repeat domain (LRD) and the TSHR LRD with the A1 chimeric substitution were done using Superpose (20,21).

Statistical analysis
Significance of differences between receptor variants was assessed by a Student's t test (SigmaPlot; Systat Software, San Jose, CA). The binding kinetics of mAbs 3BD10 and 2C11 were determined using a nonlinear regression, one-site model (Graph-Pad Prism, La Jolla, CA).

Amino acid residues in the linear component of the 3BD10 epitope
Although mAb 3BD10 has a conformational epitope (binding is eliminated by reduction and alkylation of the antigen), recognition of 50 -200 amino acid polypeptides generated by a bacteriophage TSHR cDNA fragment library revealed that the 3BD10 epitope also has a linear component contained within TSHR residues 22-51 (4).

FIG. 1. Monoclonal antibody recognition of TSHR N terminal synthetic peptides. Upper panel,
TSHR 20-mer peptides encompassed cysteine loops 1 and 2 (peptide A; residues 22-41), neither loop (peptide B; residues 37-56), and an overlapping peptide containing only loop 2 (peptide AB; residues 30 -49). Cysteine residues known to form disulfide bonds in the TSHR (6) are shown as black disks. Residue R38 in loop 2, the most N-terminal contact residue for human TSAb M22 (3) is depicted as a gray disk. Lower panel, ELISA with the peptides indicated in panel A. Binding of the indicated mAb with each peptide (mean Ϯ SEM of triplicate assays, each measured in duplicate). Antibody concentrations were 5 g/ml 3BD10, 5 g/ml 2C11, and 1:4 dilution of 3E5 containing culture medium.

564
Hamidi et al. This information was insufficient to establish whether the 3BD10 epitope included TSHR N-terminal cysteine loop 1 (residues 24 -29), cysteine loop 2 (residues 31-41), both loops, or neither loop. We therefore tested by ELISA 3BD10 recognition of TSHR 20-mer peptides containing amino acid residues within cysteine loops 1 and 2 (peptide A; residues 22-41), neither loop (peptide B; residues 37-56), and an overlapping peptide containing only loop 2 (Peptide AB; residues 30 -49) ( Fig.  1). It should be noted that multiple cysteine residues in synthetic peptides are likely to form disulfide bonds unless prevented by the presence of a reducing agent. The mAb 3BD10 recognized TSHR peptides A and AB but not peptide B (Fig. 1). As a control, TSHR mAb 2C11 with an epitope far downstream gave no signal. Of interest, the TSHR mAb 3E5 which, like 3BD10 and human TSAbs, displays steric hindrance to the TSH holoreceptor but not the GPI-tethered ectodomain (Rapoport B., S. M. McLachlan, and G. D. Chazenbalk, unpublished data), behaved similarly to 3BD10 in terms of TSHR peptide recognition (Fig. 1).
Having localized the linear component of the 3BD10 epitope to individual amino acids lying within TSHR cysteine-bonded loop 2, we performed ELISAs using alanine scanning substitutions of residues 32-40 within this segment in peptide AB ( Fig. 2A). Alanine substitutions of TSHR residues E34, E35, and D36 severely diminished 3BD10 binding, with a more limited decrease in binding with F37A and V39A (Fig. 2B). Binding of 3E5 overlapped with, but was not identical with, that of 3BD10, being shifted approximately one residue further downstream (Fig. 2B).

Influence of TSHR N-terminal cysteine-bonded loop 1 on 3BD10 binding
The preceding peptide ELISA data localizing a linear component of the 3BD10 epitope to amino acid residues contained within TSHR N-terminal cysteine-bonded loop 2 suggest that cysteine-bonded loop 1 does not play a role in 3BD10 binding. Human monoclonal TSAb M22 also does not interact with TSHR cysteine-bonded loop 1 (3), yet, surprisingly, deletion or chimeric substitution of loop 1 reduces TSAb and M22 binding (summarized in Table 1). We therefore wanted to examine the effect of these same TSHR N-terminal modifications on 3BD10 binding to the TSHR. Because of severe steric hindrance for 3BD10 binding to the TSH holoreceptor, we performed flow cytometry on CHO cells stably expressing the TSHR ECD tethered to the plasma membrane by a GPI anchor.
To serve as internal, quantitative controls for mAb 3BD10, the levels of TSHR-ECD-GPI expression were determined using mAb 2C11 and CS-17 whose epitopes are far removed from the 3BD10 epitope. Monoclonal antibodies 2C11 and CS-17 provided similar fluorescence values and were normalized to 100%. Relative to these quantitative controls, mAb 3BD10 recognition of the wild-type TSHR-ECD-GPI was reduced to approximately 40% of controls (Fig. 3A). Moreover, deletion of TSHR N terminal loop 1 amino acids 22-30 (residues 1-21 being the signal peptide) had no further effect on 3BD10 binding (Fig. 3A). In contrast to loop 1 deletion, chimeric substitution of TSHR residues 25-30 with the corresponding segment of the LH receptor (TSH-LHR-A1) (10) greatly reduced 3BD10 binding to less than 10% of the control mAb 2C11 (Fig. 3B). Monoclonal antibody 3E5 was similarly affected by the A1 chimeric substitution (Fig. 3B).

Monoclonal antibody 3BD10 does not see all TSHRs expressed on the cell surface
On flow cytometry, mAb 3BD10 fluorescence with CHO cells expressing the wild-type TSHR-ECD was approximately 40% that of the 2C11 signal (see above). Even though 3BD10 and 2C11 were of the same isotype (IgG), we wanted to exclude the possibility that this quantitative discrepancy was an artifact. We therefore performed a dose-response analysis using these two mAbs. At the highest practical concentrations (10 g/ ml), the signal with 3BD10 attained a plateau (Fig. 3C), indicating that the second antibody concentration was not limiting in these experiments. Nonlinear regression analysis revealed projected maximum fluorescence values of 36.9 Ϯ 4.2 and 73.2 Ϯ 9.4 (mean Ϯ SEM; five experiments) for 3BD10 and 2C11, respectively. Of note, despite not attaining a plateau with 2C11 nonlinear regression analysis was able to provide binding affinities for 3BD10 and 2C11 (1.2 Ϯ 0.1 vs. 3.5 Ϯ 0.6 nM, respectively). However, because of the absence of a plateau for 2C11, its binding affinity must be regarded as an approximate value.

Flow cytometric analysis of TSHR loop 1 amino acids influencing 3BD10 binding
Given the importance on 3BD10 binding of substituting TSHR loop 1 amino acids 25-30 with the corresponding LH receptor residues (chimeric receptor A1-ECD-GPI) (Fig. 3B), we generated stably transfected CHO cells expressing the TSHR-ECD-GPI with smaller modifications within this region (Fig. 4). We focused on the striking feature of potential structural importance in TSHR loop 1, namely the pair of adjacent proline residues that contribute to tight turns in the polypeptide. In the A series of mutations introduced into the wild-type TSHR-ECD, we tested whether deletion or rearrangement of these prolines would reproduce the TSH-LHR-A1 phenotype. Thus, we separated the two prolines (wt-1), deleted or substituted one proline (wt-2 and wt-3, respectively), or shortened loop 1 by removing a serine residue (wt-4). In wt-5 we substituted the TSHR residues with the corresponding residues of the FSH receptor. Flow cytometric analysis of 3BD10 binding confirmed (see Fig. 3) that 3BD10 bound fewer wild-type TSHR-ECD-GPI expressed on the cell surface relative to the control mAb 2C11 (fluorescence values with the latter normalized to 100%) (Fig. 5A). However, with the exception of the small signal decreases observed for wt-3 and wt-4, the majority of the series A limited re- . Fluorescence values obtained using mAb 3BD10 were expressed relative to the control mAb normalized to 100%. Two experiments used both 2C11 and CS-17 as controls, and a third experiment used CS-17. All mAbs were at 10 g/ml. Each bar represents the mean Ϯ SEM of values obtained in three experiments. B, Chimeric substitution of loop 1 (TSHR amino acids 25-30 replaced with the corresponding LHR residues) A1-ECD-GPI (10). These experiments tested mAb 3BD10 (10 g/ml) and also 3E5 (undiluted culture medium) relative to the control mAb 2C11 (10 g/ml), fluorescence with the latter normalized to 100%. Each bar represents values (mean Ϯ SEM) obtained in three experiments. C, The mAb 3BD10 binding kinetics. Flow cytometry was performed with increasing concentrations of mAb 3BD10 and mAb 2C11 as a control. Both antibodies are of the same IgG1 isotype. Each point represents the mean Ϯ SEM of fluorescence values obtained in five experiments. Data were analyzed by nonlinear regression.

566
Hamidi et al. arrangements or substitutions had no significant effects on 3BD10 binding relative to the wild-type TSHR-ECD-GPI. For mAb 3E5, only wt-5 recognition was slightly, but significantly, reduced (Fig. 5A).
In the B series of TSHR loop 1 modifications (Fig. 4), the goal was the converse to the series A mutations. Thus, using the TSH-LHR-A1 chimeric substitution as the template, we attempted to restore the wild-type phenotype by limited reversion back to the wild-type TSHR. The LHR, like the TSHR, has two prolines between the first two cysteine residues. However, in the LHR these prolines are not adjacent, and the intercysteine segment has one fewer residue than the TSHR. Therefore, in TSH-LHR-A1-revert1, the two LHR prolines were rearranged to be adjacent to each other. In TSH-LHR-A1-revert2, a small flexible glycine residue was added to the segment in addition to the adjacent prolines. In TSH-LHR-A1-revert3, the LHR segment was left intact but was extended by an additional glycine. On flow cytometry with mAb 3BD10, using mAb 2C11 as a control, all three series B modifications partially restored the wild-type TSHR phenotype (Fig. 5B). A similar partial reversion to the TSHR wild-type was obtained with mAb 3E5 (Fig. 5B).

ELISA with TSHR loop 1 amino acids modifications
TSHR synthetic peptide A contains a linear component of the 3BD10 epitope localized to amino acid res-idues in cysteine-bonded loop 2 (Fig.  2). Because 3BD10 recognizes this peptide on ELISA and because the peptide also includes upstream cysteine-bonded loop 1, we performed an ELISA on 20-mer synthetic peptides incorporating the same loop 1 modifications (Fig. 4) introduced into the TSHR-ECD-GPI expressed on the surface of CHO cells. Monoclonal antibodies tested (3BD10 and 3E5) were titrated to provide OD values of approximately 1.0. As positive and negative antigen controls, we used the CHO cell-generated TSHR A subunit (amino acid residues 22-289) and peptide X (TSHR amino acids 367-386, adjacent to the insertion of the TSHR ECD into the plasma membrane and far downstream of the 3BD10 epitope), respectively.
3BD10 and 3E5 recognition of peptide A was similar to that of the recombinant TSHR A subunit (Fig. 6). However, no conclusions can be drawn from this comparison because on an ELISA with a small peptide, it is not possible to use a reference antibody with a far downstream epitope, as used for TSHR-ECD-GPI flow cytometry (mAb 2C11; Fig. 5). Instead, peptide A containing the amino acid sequence of the wild-type TSHR can be used to provide a reference ELISA value. Results with the series A peptides were largely parallel to the flow cytometry data (Fig. 5A). With the exception of a small decrease in 3BD10 recognition of peptide wt-2, none of the altered peptides significantly affected 3BD10 or 3E5 recognition (Fig. 6A). However, a remarkable difference with the flow cytometry data (see Fig. 5B) was that the A1 substitution diminished only 3BD10, but not 3E5, binding (Fig. 6B). Moreover, unlike with the flow cytometry data (Fig. 5B), the partial reversions of peptide A1 to wild-type had no effect on 3BD10 binding (Fig. 6B). In this respect, 3E5 serves as an important internal control.

How does the TSHR loop 1 A1-LHR substitution alter 3BD10 binding?
This conundrum arises because our present evidence indicates that 3BD10 makes contact with TSHR loop 2 but not with loop 1. Molecular modeling of the TSHR LRD suggests that the A1 substitution in loop 1 introduces radical shifts in the orientation of the side chains of loop 2 residues E34, E35, and D36 (Fig. 7D), residues shown to contribute to the linear component of the 3BD10 epitope (Fig. 2B). Monoclonal TSAb M22 presented a greater enigma. As mentioned above and reported previously (9), M22 binding is greatly reduced by the TSHR-A1 loop 1 substitution, even though the only contact residue in the N terminal cysteine-bonded loops is R38 (3) in loop 2, and the A1 substitution in loop 1 had little effect on the position of R38 or the orientation of its side chain (Fig. 7D). A clue to this puzzle was provided by the discrepancy in the ability to partially reverse the inhibitory effect of the A1 substitution on 3BD10 binding using flow cytometry with the TSHR-ECD-GPI (Fig. 5B) but not with a synthetic peptide containing the identical substitution (Fig. 6B). Both the TSHR-ECD-GPI and peptide TSH-LHR-A1 contain loop 1 and loop 2; however, only the former contains downstream elements. These data suggested that the chimeric LHR loop 1 substitution influences the TSHR ECD conformation downstream of loop 2. The M22 epitope is predominantly downstream of TSHR loop 2, extending to residue R255 (3). Indeed, molecular modeling suggests that the A1 substitution generates alterations in the phenolic side chains of hydrophobic residues F130 and F153 (Fig. 7C) that are important contributors to the M22 epitope (3).

Discussion
Crystallization of the TSHR LRD (amino acids  in complex with a human monoclonal TSAb (M22) (3), as well as with a human monoclonal TSH blocking antibody (K1-70) (6), represents a major landmark in defining the binding sites of autoantibodies that cause human disease. Overlap between the binding sites of nonstimulatory antibody K1-70 and TSH clearly indicates the mechanism of action of the former in causing  Fig. 4. Receptor expression levels were determined using the mAb 2C11 (10 g/ml) whose epitope is far downstream in the ECD. The 2C11 fluorescence value for each receptor was normalized to 100%. Fluorescence values for 3BD10 (10 g/ml) and 3E5 (undiluted culture medium) were expressed as a percentage of the 2C11 value. For the series A substitutions, significant reductions in 3BD10 and 3E5 binding relative to the wild-type TSHR-ECD-GPI (wt) are indicated by asterisks. In series B, suppressed 3BD10 recognition of TSHR-LHR-A1-GPI (see Fig. 3B) was partially restored toward wild-type fluorescence values. Each bar indicates the mean Ϯ SEM of four experiments. *, P Ͻ 0.05, **, P ϭ 0.01, Student's t test.  To understand the mechanism of TSAb action, it will be necessary to elucidate the structure of the TSHR hinge region and the orientation to one another of the LRD, hinge, and transmembrane domain. The very recent elucidation of the crystal structure of FSH in complex with the entire extracellular domain of FSH recep-tor including the hinge region now supports homology modeling for the TSHR (22). One clue to the mechanism of TSHR activation is provided by evidence that TSAb access to the TSH holoreceptor is sterically hindered (5,9). Completion of high-affinity TSAb binding after initial limited access to the full epitope can cause receptor torsion with subsequent transmission of a signal via the transmembrane domain. Given the limited information available from the crystal structure, in the present study we focused on another unexplained phenomenon that could shed light on antibody-mediated TSHR conformational changes. As mentioned above (introductory text), the importance of murine mAb 3BD10 is that it and human TSAb reciprocally distinguish between two conformational forms of soluble TSHR A subunits, the latter comprising primarily the LRD. These two forms, termed active (TSAb specific) and inactive (3BD10 specific) are secreted by CHO cells in approximately equal proportions. Because a major linear component of the 3BD10 epitope was previously localized to the TSHR N terminus (amino acid residues 22-51) (4), we suggested that the TSHR Nterminal region contributed to this conformational difference.
Our present study exploring additional 3BD10 properties provides the following new information: (i) Flow cytometric data reveal that 3BD10 sees only approximately half of the TSHR ECD (LRD plus hinge) expressed on the cell surface. These data indicate that the phenomenon observed with isolated TSHR A subunits does not only occur for soluble protein in vitro but can now be extrapolated to the TSHR ECD tethered to the cell surface. (ii) The linear component of the 3BD10 epitope is now defined more precisely than TSHR residues 22-51 (the latter encompassing both N-terminal C-bonded loops and 10 residues further downstream). Thus, TSHR amino acids E34, E35, and D36 are the primary contributors to this epitopic component. The The orientation of the side chain of pivotal residue E30 between loops 1 and 2 is shown in this panel (dotted oval) but, for simplicity, not in subsequent panels because of no effect on its position by the A1 substitution. Although the protein crystallized contained TSHR residues 22-260, residues 258 -260 were not depicted because of molecular disorder in the crystal. B, Molecular model (see Materials and Methods) of the same TSHR LRD with the chimeric TSH-LHR-A1 substitution (TSHR residues 25-SSPPCE-30 replaced with LHR residues PEPCD) (orange) (10). C, Superimposition (overlap) of the two structures shown in panels A and B. The key residues in the M22 epitope (F130 and F153) that show significant conformational disposition changes are indicated by dotted ovals. D, TSHR C-bonded loop 2 ( Fig. 1) is magnified to illustrate the conformational dispositions of residues E34, E35, E36, and R38 in the wild-type TSHR LRD (blue) and the LRD with the A1 substitution (orange). Changes in the regions shown in panels C and D would be anticipated to influence antibody binding to the LRD through electrostatic and solvent mediated interactions (26).
Endocrinology, January 2013, 154(1):562-571 endo.endojournals.org importance of this observation is that, based on the crystal structure of the TSHR LRD (3,6), 3BD10 interacts with residues within TSHR C-bonded loop 2 (C31-C41), the latter also containing residue R38, the most N-terminal contact residue in the M22 epitope. It is of interest that mAb 3E5, with some properties similar to 3BD10, also contains a linear epitopic component within TSHR loop 2 but shifted slightly downstream of 3BD10. (iii) Although the epitopic components of 3BD10 and M22 localize to the TSHR N-terminal C-bonded loop 2 (C31-C41), manipulation of loop 1 (C24-C29) influences the binding of both 3BD10 and M22 but in different manners. Thus, the A1 LH receptor chimeric substitution of loop 1, but not deletion of loop 1, largely eliminates 3BD10 binding. The M22 binding is similarly abrogated by the A1 substitution in loop 1 (9). However, The deletion of loop 1 blinds M22 to some, but not all, forms of cell surface receptors (12). (iv) Reversion to wild-type of small components of the loop 1 A1 chimeric substitution partially restores 3BD10 binding when studied with the TSHR-ECD cell surface protein but not when the same substitutions are tested with synthetic peptides.
The foregoing lines of evidence indicate that the TSHR C-bonded loop 1 influences receptor conformation in a manner that is important for the binding of both 3BD10 and TSAb (including monoclonal M22), even though these antibodies interact with C-bonded loop 2 (not loop 1). Previously, based on observations with M22, we hypothesized that flexibility around a fulcrum of E30, the intervening residue between loop 1 (C24 -29) and loop 2 (C31-41) (see Fig. 1), could explain the enigmatic role of loop 1 (12). We subsequently considered the possibility that the A1 LH receptor chimeric substitution in the TSHR could convert the two-loop structure in the TSHR to a sushi domain, as occurs in the FSH receptor (23). The crystal structure of the LH receptor has not been solved. However, contrary to our hypothesis, molecular modeling suggests that the E30 link between loop 1 and loop 2 in the TSHR is unlikely to be flexible (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org) and that the LH receptor N terminus is more likely to resemble the TSHR two-loop structure (Fig. 7 B) rather than the FSHR sushi domain structure.
Molecular modeling of the TSHR loop 1 A1 LHR substitution provided plausible, but different, explanations for the effect of this modification on 3BD10 and M22 binding. In the case of 3BD10, the A1 substitution in loop 1 introduces major shifts in the orientation of the side chains of loop 2 residues E34, E35, and D36 (Fig. 7D), which contribute to the linear component of the 3BD10 epitope (Fig. 2B). It should be appreciated that even though 3BD10 has a linear epitopic component within TSHR loop 2, many antibodies to large proteins have a footprint in the range of 20 residues (for example, Refs. 24 and 25), and it is therefore likely that there are discontinuous, conformational components to the 3BD10 epitope further downstream in the TSHR ECD that could be affected by the A1 substitution.
The mechanism by which the A1 substitution greatly reduces binding to the TSHR of polyclonal TSAb in patients' sera (10) as well as of human mAb M22 was more difficult to explain because this chimeric substitution did not significantly alter the position of TSHR residue R38 or the orientation of its side chain (Fig. 7D). Rather, molecular modeling suggested that the A1 substitution induces conformational changes far downstream in the TSHR LRD, namely the phenolic side chains of hydrophobic residues F130 and F153 (Fig. 7C) that are important contributors to the M22 epitope (3). This concept is consistent with the discrepancy in the ability to partially reverse the inhibitory effect of the A1 substitution on 3BD10 binding using flow cytometry with the TSHR-ECD-GPI (Fig. 5B) but not with a synthetic peptide containing the identical substitution (Fig. 6B). Only the TSHR-ECD-GPI contains the entire LRD including residues F130 and F153.
In conclusion, the present study with the mouse mAb 3BD10 illustrates the complexity of the TSHR structure and indicates that the crystal structure of the isolated TSHR LRD bound by TSAb M22 does not provide all the answers regarding the mechanism by which TSAb activate the receptor. Our data support the concept that there is likely to be more than a single conformational form of the TSHR and suggest further that the N-terminal cysteine cluster may contribute to this structural variability.