Diversity in the Complementarity-Determining Region 3 (CDR3) of Antibodies from Mice with Evolving Anti-Thyroid-Stimulating Hormone Receptor Antibody Responses

Diversity in the Complementarity-Determining Region 3(CDR3) of Antibodies from Mice with EvolvingAnti-Thyroid-Stimulating Hormone ReceptorAntibody Responses

Osvaldo Martinez, Eryn Gangi, David Mordi, Sonal Gupta, Samuel Dorevitch, Marie-Paule Lefranc, andBellur S. Prabhakar

Department of Microbiology and Immunology (E.G., D.M., S.G., B.S.P.), School of Public Health, Epidemiology, andBiostatistics (S.D.), University of Illinois at Chicago, Chicago, Illinois 60612; Laboratoire d’ImmunoGenetique Moleculaire(M.-P.L.), Universite Montpellier II, Unite Propre de Recherche, Centre National de la Recherche Scientifique 1142, Institutde Genetique Humaine, 34396 Montpellier, France; and Department of Microbiology (O.M.), Mount Sinai School ofMedicine, New York, New York 10029

In a mouse model of autoimmune Graves’ disease, stimulatoryanti-TSH receptor (TSHR) antibodies (TSAbs) slowly evolveupon repeated immunization with TSHR and lead to hyper-thyroidism. Although all immunized mice developed high lev-els of TSH-binding inhibitory Ig (TBII), only a subset of thesemice become hyperthyroid, suggesting that the generation ofpathogenic antibodies (Abs) may require affinity maturation.We analyzed the complementarity-determining region 3(CDR3) of IGHV1 and IGHV5 heavy chains from mice at dif-ferent stages of disease development. Subcloned CDR3 PCRproducts were amplified from RNA isolated from enrichedsplenic B/plasma cells of a control mouse, and mice with lowTBII and normal T4 levels (LTNT4), high TBII and normal T4levels (HTNT4), and high TBII and high T4 levels (HTHT4).

Using statistical analyses, we correlated usage of D and Jgenes and the amino acid composition and length of and mu-tations within the CDR3 with different outcomes after TSHRimmunization. CDR3 sequences from TSHR-immunized micecontained a higher frequency of D gene SP2.9 relative to con-trol, whereas sequences from HTHT4 contained a higher fre-quency of D gene Q52 compared with sequences from LTNT4.Furthermore, HTHT4 sequences also contained higher CDR3replacement mutations, relative to LTNT4 and HTNT4 mice,that are indicative of somatic hypermutation. Collectively,our results suggest that higher somatic mutations within theCDR3 may correlate with pathogenic antibodies against theTSHR. (Endocrinology 148: 752–761, 2007)

THE EUTHYROID STATE is maintained through a hy-pothalamo-pituitary-thyroid feedback loop (1) that is

regulated by the levels of TSH and thyroid hormones. Bind-ing of TSH or thyroid-stimulating antibodies (TSAbs) to theTSH receptor (TSHR) results in thyroid hormone secretion(2). Several monoclonal TSAbs have been described (3–6),and most of them bind to TSHR with high affinity (Kd � 1 �10�9 to 7.1 � 10�11) and block TSH binding, suggesting thatTSH and TSAbs may have a common binding specificity (7).

In patients with Graves’ disease (GD), the TSAbs are re-stricted to IgG isotypes (8, 9) and show evidence of affinitymaturation (10, 11). High-affinity antibodies (Abs) are pro-duced through a regulated process in which somatic muta-

tions are randomly introduced into the Ig genes. After mu-tating, B cells that lose their affinity for the cognate antigenare eliminated, whereas B cells that competitively bind theremaining antigen survive and may increase their affinity forthe antigen by undergoing additional somatic mutations.The complementarity-determining regions (CDR) and theframework regions (FR) of the Ab are primarily important forantigen binding and maintaining the Ab structure, respec-tively (12, 13). The Ig gene heavy chain CDR3 that resultsfrom a recombination of variable-diversity-joining (V-D-J)genes is translated into a protein loop (14) that comes indirect contact with the antigen and thus plays a critical rolein determining the specificity and affinity of Abs. Therefore,antigen-driven selection that results in affinity maturation isa consequence of higher frequencies of mutations in the CDRrelative to those in FR of Ig genes.

It is likely that pathogenic Abs to TSHR (15) appear slowlyover a period of time in patients with GD. Whether this slowappearance of pathogenic Abs is dependent upon repeatedexposure to the target antigen (i.e. TSHR) and emergence ofhigh-affinity Abs through somatic hypermutation, or reper-toire spreading, is not known. We have developed a mousemodel of experimental autoimmune GD (EAGD) in whichthe disease is induced through repeated immunization ofBALB/c mice with isogenic cells expressing high levels of

First Published Online October 26, 2006Abbreviations: Ab, Antibody; CDR, complementarity-determining

region; EAGD, experimental autoimmune GD; ECD, extracellular do-main; FR, framework region; GD, Graves’ disease; HTHT4, high TSH-binding inhibitory Ig and high T4 levels; HTNT4, high TSH-bindinginhibitory Ig and normal T4 levels; IGH, Ig heavy chain; LTNT4, lowTSH-binding inhibitory Ig and normal T4 levels; mAb, monoclonal Ab;OVA, ovalbumin; R, replacement; RT, room temperature; S, silent; TBII,TSH-binding inhibitory Ig; TsAb, thyroid-stimulating Ab; TSHR, TSHreceptor; V-D-J, variable-diversity-joining.Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving theendocrine community.

0013-7227/07/$15.00/0 Endocrinology 148(2):752–761Printed in U.S.A. Copyright © 2007 by The Endocrine Society

doi: 10.1210/en.2006-1096

752

TSHR (mM12 cells) (16, 17). Anti-TSHR Abs that can bereadily detected in an ELISA are induced shortly after thesecond immunization. However, the TSH-binding inhibitoryIg (TBII) activity emerges after three to four immunizations,but the animals remain euthyroid. The TBII activity steadilyincreases with each immunization, and TSAbs appear in asubset of these mice leading to hyperthyroidism. Therefore,the mere presence of high levels of TBII activity does notcorrelate with hyperthyroidism (18–21). Little is knownabout the evolution of somatic hypermutation and its rela-tion to pathogenic anti-TSHR Abs in both humans and mice(15). We hypothesized that the pathogenic anti-TSHR Abproduction is related, at least in part, to affinity maturationas indicated by increased somatic mutations in the CDR3. Tobegin to understand the molecular evolution of TSAbs, weanalyzed the CDR3 sequences from mice with differentstages of disease for Ig heavy chain (IGH) V-D-J gene usageand somatic mutations.

Materials and MethodsCells and antibodies

The M12 cells (BALB/c isogenic B cell line) or M12 cells stably trans-fected with a TSHR cDNA (mM12 cells) that allows cell surface expres-sion of the protein (17) were used for immunization. The HEK293 cellsexpressing chimeric protein consisting of the extracellular (or ectodo-main) domain of the TSHR (ECD-TSHR), intervening thrombin site, anda C-terminal CD8 transmembrane domain (17, 20) (i.e. TTCD8 cells) wereused as a source of soluble TSHR ectodomain, sufficient for anti-TSHRantibody induction (22). The antimouse CD4, antimouse CD8, anti-mouse B220, antimouse IgG, antimouse CD19, antimouse IgM, andantimouse CD11b antibodies were purchased from Becton Dickinson(PharMingen, San Diego, CA).

Induction of GD

Six- to eight-week-old BALB/c mice were purchased from JacksonLaboratories (Bar Harbor, ME). Mice were immunized seven times withmitomycin C-treated M12 or mM12 cells 2 wk apart. Each mouse wasimmunized with approximately 107 cells. Harvested cells were washedand resuspended in 50 �g/ml mitomycin C and incubated for 30 minin the dark at 37 C followed by three washings in PBS. Each mouse wasimmunized ip with mitomycin C-treated cells and 5 �g cholera toxin Bas described before (17). All animal studies were carried out accordingto the procedure approved by the Animal Care and Use Committee ofthe University of Illinois at Chicago.

TBII and T4 assays

Sera were obtained from blood samples that were collected period-ically from mouse tail vein. Fifty microliters sera were tested for TBIIactivity using a commercial radioreceptor (Kronus, Dana Point, CA)assay following the manufacturer’s protocol. Mouse serum (25 �l) wasused to determine the blood levels of total T4, using a commercial kit,Coat-a-Count (Diagnostic Products, Los Angeles, CA), following themanufacturer’s protocol.

Purification of the ECD-TSHR protein

TTCD8 cells were grown to confluence in T-150 flasks and washedonce gently, and serum-free medium containing 100 U/ml �-thrombinwas added. Flasks were incubated overnight at 37 C in a humidifiedincubator containing 5% CO2. Supernatants were collected, and histi-dine-tagged TSHR was purified from the spent medium using a nickelcolumn (Pharmacia Amersham, Piscataway, NJ), following the manu-facturer’s protocol. Purity of the protein was confirmed using SDS-PAGE followed by Western blotting for TSHR.

Expansion of anti-TSHR Ab-producing cells

Mice previously immunized with mM12 cells were rechallenged ivwith either 100 �g purified ECD-TSHR or 100 �g ovalbumin (OVA).Three days later, mice were killed and splenocytes harvested. Red bloodcells were removed by resuspending splenocytes in 0.83% ammoniumchloride for 5 min at room temperature followed by washing with PBS.To enrich for plasma cells, IgM�, CD4�, CD8�, and CD11b� cells weredepleted by staining splenocytes with anti-CD4, anti-CD8, anti-IgM, andanti-CDllb phycoerythrin- or fluorescein-isothiocyanate-conjugated an-tibodies at 5 �g/ml for 30 min in PBS plus 2% fetal calf serum at 4 C.Stained cells were washed with PBS and then incubated with anti-phycoerythrin and anti-fluorescein-isothiocyanate MACS beads(Milteny Biotech, Auburn, CA) for 10 min. The beads were washed, andbound cells were removed with a MACS sorter using the deplete pro-gram. The depleted cells from different mice were tested for viability(�75%) and B220� staining (55–60%) and were found to be comparable(not shown). Depleted and nondepleted cells were washed and resus-pended in 5% RPMI 1640 medium and plated into 96-well plates at 1 �105 cells per well. Six days later, culture supernatants were tested for thepresence of spontaneously secreted anti-BSA, anti-OVA, and anti-TSHRIgG antibodies by ELISA.

ELISA

For ELISA, 96-well flat-bottomed Maxi-Sorp (Nunc, Rochester, NY)plates were coated with 200 ng/well of OVA or ECD-TSHR protein inbuffer overnight at 4 C. Plates were then washed three times with PBScontaining 0.05% Tween 20 and blocked with PBS plus 10% FCS at roomtemperature (RT) for 1 h. Plates were washed three times with PBS/0.05% Tween 20, and culture supernatants were added to appropriatewells and incubated for 2 h at RT followed by five washes. AntimouseIgG conjugated to horseradish peroxidase (Caltag, Burlingame, CA) wasadded at a 1/2000 dilution to all wells and incubated at RT for 1 hfollowed by seven washes. 3,3�,5,5�-Tetramethylbenzidine substrate wasadded to each well and incubated for 15 min in the dark until thehorseradish peroxidase reaction was stopped with 1/4 vol of 1 N HCl.Plates were read using an ELISA plate reader (Bio-Rad, Hercules, CA)at 450 nm wavelength.

RNA purification and cDNA synthesis

Total RNA was harvested from splenocytes using TRIzol reagentfollowing the manufacturer’s protocol (Invitrogen, Carlsbad, CA). TheRNA was resuspended in water containing 1 �l RNAGuard (Pharmacia-Amersham, Piscataway, NJ) at 5 � 106 cellular equivalents/50 �l. Thefirst-strand cDNA was synthesized using Superscript II (Invitrogen)following the manufacturer’s protocol with a minor modification. TotalRNA (11 �l) was added to 1 �l oligo-dT (Fermentas, Hanover, MD) andheated to 60 C for 10 min. Two microliters of of 10� Superscript bufferwas added along with 1 �l of 10 mm dNTP mix and 1 �l of 0.1 mdithiothreitol. The rest of the protocol was identical to that suggested bythe manufacturer.

Subcloning of CDR3

PCR samples were prepared in UV-irradiated sterile hoods with allcomponents on ice before placing them into an iCycler PCR machine(Bio-Rad, Hercules, CA). CDR3 was amplified using 0.4 �m of IGHVsubgroup-specific primers and 0.2 �m of each of the four IGHJ primers(see Fig. 2) in 1� PCR buffer (Roche Molecular Biochemicals, India-napolis, IN) in a final volume of 50 �l. The reaction mix was supple-mented with 0.2 mm dNTP mix, 3 U high-fidelity polymerase mix (RocheMolecular Biochemicals, Mannheim, Germany), and 1 �l sample cDNA.The PCR was performed with a 2-min hot start at 94 C and 30 cycles at91 C, 72 C, and 60 C, each for a period of 30 sec ending with a 5-minextension at 72 C. The PCR product was then separated on a 1.5%agarose gel to confirm size and purity of the product. Each PCR productwas concentrated and purified using the PCR purification and concen-tration kit (QIAGEN, Valencia, CA) and eluted in a 10-�l vol of water,and 1.5 �l of the products was then used in a ligation reaction consistingof 1� ligation buffer, 0.5 �l T4 ligase, and 0.5 �l pGEM vector (Fisher,Pittsburgh, PA) and incubated overnight at 16 C. Then, 2 �l was used

Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies Endocrinology, February 2007, 148(2):752–761 753

to transform MAX efficiency DH5�-competent cells (Invitrogen).Ampicillin-resistant colonies were harvested and grown overnight in LBbroth, and their DNA was prepared using QIAGEN mini-prep kit. Thepresence of insert was confirmed using SacI and SacII digests, andpositive clones were sequenced using universal T7 short primers.

Real-time PCR

One microliter of cDNA was used for each real-time PCR using theiCycler PCR system (Bio-Rad). In the final PCR cocktail, a single IGHVprimer (0.4 �m) and a combination of four IGHJ primers (0.2 �m each)that can prime all four IGHJ genes in mouse were used to produce anapproximately 165-bp PCR product. PCR was carried out for 50 cyclesat 91 C, 72 C, and 60 C, each for a period of 30 sec. The cocktail of primersand cDNA was diluted with an equal volume of 2� Sybr Green qPCRmix (Invitrogen). For each mouse, two to three PCR using all the sub-group-specific primers were performed. To estimate the copy numbers,a standard curve, using cloned VH1 cDNA was used in each reaction.This plasmid was diluted 10-fold over a previously determined range of10–35 cycle number thresholds. Cycle number thresholds for the IGHVPCR were typically in the 20s except for IGHV1 (18–20). Copy numberscalculated from the standard curve using regression equations wereadded together to represent 100% of all IGHV expressed per mouse. Thecopy number per mouse per individual IGHV reaction was then dividedby the total per mouse and multiplied by 100 to determine the percentusage of that particular IGHV subgroup.

CDR3-IMGT analysis

Sequences were aligned and analyzed using the Seqlab program fromthe University of Illinois’ online GCG Wisconsin package software. Thepileup program was used to determine the nucleotide and amino acidsimilarities of all sequences and their translated products. Sequenceswere then exported in FASTA format. The IMGT/V-QUEST tool (23)from IMGT, the international ImMunoGeneTics information systemhttp://imgt.cines.fr (initiator and coordinator Marie-Paule Lefranc,Montpellier, France) (24) was used to determine IGHV gene and IGHJgene usage for each sequence, which was then used in the IMGT/JunctionAnalysis tool (25) to determine IGHD gene usage, V-D-J JUNC-TION diversity, and JUNCTION mutation analysis. IMGT/JunctionAn-alysis explores the JUNCTION that is delimited in 5� by the V-REGION2nd-CYS 104 and in 3� by the J-REGION J-TRP 118 (for the IGHJ). TheJUNCTION includes 2nd-CYS 104 and J-TRP 118, whereas the CDR3-IMGT does not include these positions and extends from position 105–117. Possible permutations of exact positions of mutations within theJUNCTION were determined using a newly improved IMGT/Junctio-nAnalysis program. For all sequences, the permutation with the highestlevels of mutations was chosen. Finding two replacement mutationswithin one codon was counted as a single-replacement mutation.Sequences used for the analysis have been deposited into GenBank andare identified with accession numbers AY877924–AY877941 andAY877943–AY878052.

Statistical analysis

Associations were analyzed between D gene usage, J gene usage, andCDR3 amino acid usage on the one hand and TSHR-immunization statusof mice (immunized vs. control) and the thyroid status [hyperthyroidwith high TBII and high T4 levels (HTHT4) vs. euthyroid with normalTBII and normal T4 levels and high TBII and normal T4 levels (HTNT4)]on the other. The �2 or odds ratio with confidence intervals (95%) werecalculated using SAS, version 8.12 (SAS Institute, Cary, NC). Becausefrequencies of mutations among gene sequences were not normallydistributed among groups, Wilcoxon rank-sum test was performed todetermine whether statistically significant differences existed in thenumber of CDR3 R and S mutations, FWR3 R and S mutations and CDR3lengths for sequences from TSHR-immunized mice vs. control mouseand high-T4 mouse vs. other TSHR-immunized mice with normal T4mice existed. A P value � 0.05 was deemed to be significant.

Results

We monitored the production of anti-TSHR antibodies bytesting serum TBII activity in mice immunized with mM12

cells. Mice had developed either high (�40%) (HTNT4) orlow (�20%) TBII activity, and some mice with high TBIIactivity had also developed higher than normal T4 levels(Table 1). We chose to study the IGH CDR3 derived frommice highlighted in Table 1 (in bold) that featured either lowTBII activity (20%) with normal T4 (4 �g/dl) (LTNT4), highTBII activity (50%) with little T4 perturbation (5 �g/dl)(HTNT4), or high TBII values (49%) with high (8.5 �g/dl) T4

levels (HTNT4), compared with control (12% TBII, 3.9 �g/dlT4).

Before isolating IGH CDR3 sequences, we tested a plasmacell enrichment protocol (26) modified to deplete IgM� cells.IgM� cells were depleted because the TSAb activity is foundin serum IgG in GD patients (8, 9, 27) and hyperthyroidmouse (not shown). Table 2 demonstrates an example of theresults of plasma cell enrichment. In this experiment, micepreviously immunized with mM12 cells were immunizedwith either 100 �g OVA or 100 �g purified recombinantECD-TSHR protein while the other served as a control. Threedays after immunization, plasma cells were enriched fromharvested splenocytes and tested for viability (�75%) andB220� staining (55–60%) and found to be similar in all de-pleted populations (not shown). MACS negative depletionprevents nonspecific activation of target cells and allowed usto test for spontaneous antibody production. To test for thepresence of Ab-producing cells, spent supernatants of theplasma-cell-enriched cultures were tested for the presence ofanti-BSA (irrelevant antigen), anti-OVA, and anti-TSHR IgGAbs by ELISA. Table 2 shows that supernatants from bothcontrol and OVA-immunized mice were negative for anti-TSHR, anti-BSA, or anti-OVA IgG Abs. Splenocytes frommM12 and OVA-immunized mice failed to produce anti-OVA Abs, likely because these mice were primed once withOVA 3 d before being killed and the assay detected onlyIgGs. In contrast, splenocytes from mM12 and ECD-TSHR-immunized mice readily produced anti-TSHR IgGs. Thenumber of anti-TSHR IgG Abs secreted per splenocyte per6 d was estimated to be approximately 158,000 for the de-pleted population (Table 2). There was approximately a

TABLE 1. Serum TBII and T4 hormone levels in TSHR-immunized mice

Mouse TBII (%) T4 �g/dl

1 14 52 31 53 30 4.84 64 6.55 51 4.86 20 47 56 4.38 22 59 49 8.510 50 5Control mice 12.5 2.1 3.9 0.52

Three days after the seventh immunization with mM12 or M12cells along with 5 �g cholera toxin adjuvant, mice were bled from thetail vein. Harvested serum was then tested for TBII activity and T4hormone levels, and the results are shown. Shown in the bottom roware the average T4 and TBII levels of four M12-immunized mice. Allmice were subjected to an identical immunization protocol, and miceshown in bold were selected for this study.

754 Endocrinology, February 2007, 148(2):752–761 Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies

3-fold increase in the levels of anti-TSHR Abs in depleted,relative to the nondepleted, populations (Table 2).

Mice primed with TSHR (Table 1), but not the control,were immunized iv with 100 �g purified ECD-TSHR proteinand killed 3 d later when plasma cell populations peak (d3–5) in secondary lymphoid organs (28–31). After enrich-ment, depleted cells were tested for B220 expression levelsand were found to be similar in all mice (55–65% B220� cells,not shown). To take advantage of the increase in steady-stateIg RNA levels in plasma cells (32, 33) compared with restingB cells (�10- to 60-fold), total RNA was extracted from en-riched B/plasma cells and cDNA synthesized using oligo-dTprimers. To amplify the heavy chain CDR3, previously pub-lished IGHV subgroup-specific and IGHJ primers were used(34, 35) (Fig. 1). These primers have been used to distinguishCDR3 sequences corresponding to the antihapten Ab re-sponse from those derived from total spleen by CDR3 lengthspectratyping (i.e. searching for overrepresented CDR3 mo-lecular characteristic). Using these IGHV subgroup-specific

primers (36), we obtained approximately 165 nucleotidesproducts consisting of approximately 129 nucleotides of FR3and approximately 36 nucleotides of the CDR3. PCR condi-tions consistently produced expected size products withoutspurious bands as demonstrated by single-peak melt curvesand single-sized products when separated in 1.5% agarosegels (not shown).

Before B cells enter the periphery during their develop-ment, Ab diversity is generated through V-D-J recombina-tion (13). V-D-J recombination is a random process that re-sults in the formation of the CDR3 (Fig. 1). This involves the3� end of a V gene recombining with a D gene that has alreadyrecombined with the 5� end of a J gene. By using real-timePCR and subgroup-specific primers, we determined the fre-quency of usage of IGHV subgroups (Fig. 2A). For compar-ison, we have shown the IGHV subgroup usage as a pro-portion of their genomic distribution (Fig. 2B). The repertoirein each mouse was similar but showed distinct differencesbetween each other. For example, the HTHT4 mouse showedincreased usage of IGHV5 in the IGHV-D-J repertoire. Tran-scripts in all mice used mostly IGHV1 and IGHV5 subgroupgenes.

Therefore, IGHV1 and IGHV5 CDR3 were PCR amplifiedand cloned into the pGEM cloning vector, and a represen-tative number (�35–50 per mouse) of PCR products weresequenced. Of the more than 200 IGHV1 and IGHV5 rear-ranged sequences analyzed in this study, most translatedinto proteins without stop codons (�70%), none was a pseu-dogene, and subgroup-specific IGHV primers amplifiedgenes from only their corresponding subgroup (not shown).There was an average 4-fold increase in IGHV1 transcriptsamong immunized mice relative to the naive control, butthere was no statistical difference between immunized mice(not shown).

CDR3 from all mice was analyzed both at the nucleotideand amino acid level. No identical or similar (defined as�98% identical at nucleotide level) sequences were identifiedwhen comparing sequences from one mouse to any othermouse. However, similar CDR3 sequences were found whensequences were compared within each mouse, except for thecontrol mouse (Table 3). For example, one sequence from theLTNT4 mouse was repeated twice, whereas another one wasrepeated five times (see Table 3). Repeated sequences fromthe HTNT4 and HTHT4 mice were not always 100% identical,and interestingly, some sequences displayed a few nucleo-tide substitutions, which could represent clonal sets.

TABLE 2. Spontaneous secretion of anti-TSHR Ab by splenocytes from TSHR-immunized mice

Mouse Primedwith

Immunizedwith

IgG molecules per splenocyte per 6 d

BSA-specific OVA-specificTSHR-specific

Depleted Not depleted

Mouse 1 (control) ND ND ND NDMouse 2 mM12 OVA ND ND ND NDMouse 3 mM12 TSHR ND ND 158,000 (�/�40,000) 47,000 (�/�2,000)

Mice previously immunized with mM12 cells were injected iv with 100 �g of either OVA or purified recombinant TSHR protein. Three dayslater, splenocytes from these mice and a control mouse were harvested and depleted of CD4�, CD8�, IgM�, and CD11b� cells. Splenocytes,depleted or not depleted, were cultured in 96-well dishes (1 � 105 cells per well), and spent supernatant was harvested 6 d later and testedin an ELISA for spontaneous production of anti-BSA, anti-OVA, and anti-TSHR IgG antibodies. ELISA plates were coated with 200 ng BSA,OVA, and TSHR protein. Mouse anti-IgG antibody conjugated to horseradish peroxidase was used as a secondary. Known amounts of murineIgG were used to develop a standard curve to determine amounts of Ab secreted. ND, Not detected.

IGHJ1 GACGGTGACCGTGGTCCCTGIGHJ2 GACTGTGAGAGTGGTGCCTTGIGHJ3 GACAGTGACCAGAGTCCCTTGIGHJ4 GACGGTGACTGAGGTTCCTTG

IGHV1 CAAGGCCACATTGACTGTAGACIGHV2 CAGACTGAGCATCACCAAGGACIGHV3 TCGAATCTCCATCACTCGAGACIGHV4 CAAGGCCACATTGACTGTAGACIGHV5 CCGATTCACCATCTCAAGAGACIGHV6 GAGGTTCACCATATCAAGAGATIGHV7 TCGGTTCATCGTCTCCAGAGACIGHV8 CCAGCTCACAATCTCCAAGGATIGHV9 ACGGTTTGCCTTCTCTTTGGAAIGHV10 CAGGTTCACCATCTCCAGAGATIGHV11 TCGATTCACTATCTTCAGAGACIGHV12 CCCCATCTCCATTACTAGAGAAIGHV13 CAGATTCGCCATTTCAAGAGATIGHV14 CAAGGCCACTATGATCACAGAC

~160 bp

VHDJH

Sense primers Anti-sense primers

FIG. 1. Primers used to amplify CDR3 of murine IGHV-D-J se-quences. The end result of heavy-chain V-D-J recombination is theCDR3 (boxed). IGHV subgroup-specific primers prime within the FR3and were used with a cocktail of four IGHJ antisense primers, rep-resenting all four IGHJ genes in mouse. In the final PCR cocktail, asingle IGHV primer (0.4 �M) and a combination of all four IGHJprimers (0.2 �M each) that are able to prime all four IGHJ genes inmouse were used to produce an approximately 165-bp PCR product.On average, the length of the CDR3 was 36 nucleotides long, and theupstream FR3 sequence amplified was 129 nucleotides long.

Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies Endocrinology, February 2007, 148(2):752–761 755

V-D-J recombination results in the formation of the CDR3(Fig. 1). We determined the IGHD and IGHJ genes used bythe IGHV-D-J transcripts analyzed. IGHD gene usage in allCDR3 (Fig. 3A) and in repeated sequences (Fig. 3B) relativeto the control is shown. The IGHD genes used by the controldemonstrated a bias toward the IGHD-FL16.1*01 gene,whereas other IGHD genes were used at a lower frequency.The IGHD genes used by the repeated sequences from theLTNT4 mouse were IGHD-FL16.1 (�30%) and IGHD-SP2.9(�70%). The HTNT4 mouse preferentially used IGHD-

FL16.1*01 (�55%) and several others to a much lower extentincluding IGHD-SP2.9 (11%). In contrast, the HTHT4 mouseused mostly IGHD-Q52*01 (33%), IGHD-FL16.1*01 (22%),IGHD-SP2.9 (22%), and IGHD-FL16.2 (22%) (Fig. 3B). TheIGHD-SP2.9 gene was used exclusively by transcripts fromall immunized mice tested and not the control.

The IGHJ gene usage in all CDR3 (Fig. 4A) and in repeatedsequences (Fig. 4B) showed that the control sequences mostlyused IGHJ4 followed by IGHJ2 and IGHJ3 (Fig. 4, A and B).Repeated sequences from the LTNT4 mouse used exclusively

0

10

20

30

40

50

60

70

80

90

100

1 2 3 4 5 6 7 8 9 01 11 21 31 41

ControlLTNT4HTNT4HTHT4

0

10

2 0

3 0

4 0

5 0

6 0

7 0

8 0

9 0

10 0

A

B

IGHV subgroup

noit

atn

es

erp

er p

uor

gb

us

en

eg

VH

GI %

0

10

20

30

40

50

60

70

80

90

100

1 2 3 4 5 6 7 8 9 01 11 21 31 41

ControlLTNT4HTNT4HTHT4

0

10

2 0

3 0

4 0

5 0

6 0

7 0

8 0

9 0

10 0

oitat

ne

ser

per

uor

gb

us

en

eg

VH

GI %

FIG. 2. Frequency of IGHV subgroup usage in representative mice. Total RNA from depleted splenocytes was used to generate oligo-dT-primedcDNAs, which were used to amplify each IGHV separately using an IGHV subgroup-specific primer and a combination of all four IGHJ primers.A standard curve was used to determine IGHV subgroup usage as described in Materials and Methods. The SD are shown for the real-time PCRperformed on three separate occasions. A, Each IGHV gene subgroup usage shown as a percentage of all IGHV subgroups amplified; B, eachIGHV subgroup shown as a proportion of their genomic distribution.

TABLE 3. Repeated sequences from control and immunized mice

Mouse Conditions Totalsequences

Sequence repeats(X times)

Total number ofrepeated sequences

Naive Normal T4, control 28 0/28 0LTNT4 Normal T4, low TBII 30 1/30, 2X 7

1/30, 5XHTNT4 Normal T4, high TBII 45 6/45, 2X 18

2/45, 3XHTHT4 High T4, high TBII 29 2/29, 2X 8

1/29, 4X

The total sequences column shows the number of sequences from each mouse. Sequences were derived from cloning of at least two separatePCR for each mouse. The sequences are expected to be nonquantitative. The sequence repeats column shows the number of different sequencesthat were repeated. For example, for sequences from the LTNT4 mouse, there were two different sets in which one set had five and anotherset had two repeated sequences. Repeated sequences were grouped if they contained a minimum of 98% similarity at the nucleotide level. Thelast column shows the total number of repeated sequences.

756 Endocrinology, February 2007, 148(2):752–761 Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies

IGHJ3, whereas repeated sequences from the HTNT4 mouseused all four IGHJ genes. In contrast, repeated sequencesfrom the HTHT4 mouse used IGHJ1 and IGHJ4 (Fig. 4B).

V-D-J recombination also generates diversity in the CDR3region through nuclease-mediated shortening of the recom-bining V, D, and J genes and in the random N and P additionsbetween the recombining V-D-J genes (13). All sequenceswere analyzed by using IMGT/JunctionAnalysis (http://imgt.cines.fr) (20) to delineate the V, D, and J genes in therearrangement and to determine the junctional diversity ofeach sequence. In Fig. 5, the junctional diversity of the re-peated CDR3 nucleotide sequences from TSHR-immunizedmice (Table 3) is shown. This includes the N and P nucleotideadditions and nuclease-mediated shortening of the IGHV,IGHD, and IGHJ genes represented by dots (Fig. 5). Boxedgroups in Fig. 5 represent sequences that contain less than 2%difference at the nucleotide level.

Mature B cells may further diversify their antibodiesthrough random somatic mutations in the variable regions

(including FR3 and CDR3) of their Igs. Therefore, we deter-mined the number of mutations within the 3� part of the FR3(codons at positions 82–104) by comparing with the germlineIGHV genes identified by IMGT/V-QUEST. Similarly, thenumber of mutations in the CDR3 (codons at positions 105–117) was determined by comparing with the nucleotidesencoded by the germline 3�V-REGION, D-REGION, and 5�-J-REGION and identified by IMGT/JunctionAnalysis. Theanalysis was performed for all sequences and repeated se-quences compared with control sequences (Table 4). The dataare presented as the average number of mutations per se-quence. Table 4 shows that there were 0.46, 0.07, 0.4, and 0.93replacement (R) mutations per sequence in the CDR3 and0.96, 0.57, 0.96, and 1 R mutation per sequence in the 3� partof the FR3 for the control, LTNT4, HTNT4, and HTHT4 mice,respectively.

The R and silent (S) mutations denote nucleotide changesthat lead to and do not lead to amino acid changes, respec-tively. Therefore, a higher R/S ratio would indicate accu-mulation of mutations that are more relevant for affinitymaturation. Importantly, the R/S values in these groupswere 6.6, 1, 2.5, and 9.3 for the CDR3 and 1.9, 0.6, 1.1, and 1.8for the 3� part of the FR3, respectively. The trends for themutations from all sequences compared with repeated se-quences (Table 4) did not change significantly. Within CDR3,the LTNT4 sequences contained fewer R mutations (approx-imately 10-fold less than HTHT4), whereas control andHTNT4 sequences contained a similar number of mutations.However, HTHT4 sequences showed twice as many R mu-tations as those of either control or HTNT4 sequences. More-over, the CDR3 R/S values were higher in the HTHT4 se-quences than in any of the other mouse sequences. Incontrast, the R mutations and R/S values in the 3� part of theFR3 for HTHT4 sequences were not high. Although the R/Svalue in the CDR3 of the HTHT4 sequences was a minimumof 9, R/S values in the 3� part of the FR3 of all or repeatedsequences were never higher than 1.9.

To determine whether associations between molecularcharacteristics of sequences and TSHR immunization statusor T4 hormone levels could be explained by chance alone,statistical analyses were performed (Table 5). Specifically, wetested to see whether there were any associations betweenthe V, D, or J gene usage (Figs. 2–4), CDR3 amino acid usage(i.e. basic, acid, aliphatic, aromatic, S-containing, proline, al-cohol, or amides) (not shown), mean CDR3 length (notshown) and mean R and S mutations in CDR3 and FR3 (Table

A

B

05

101520253035404550

10*1

.61

LF

20*1

. 61

LF

2.61

LF

10 *

4T

S

10*1

.2P

D

10*2

.2P

S

1 0*3

.2P

S

1 0*x

.2P

S

1 0

*7.2

PS

10*8

. 2P

S

10*9

. 2P

S

4.2

PS

10*2

1 .2

PS

10*3

1 .2

PS

10*2

5Q

%

control

LTNT4

HTNT4

HTHT4

01020304050607080

10*1

.61

LF

20*1

.61

LF

2.61

LF

10*

4T

S

10* 1

. 2P

D

10* 2

.2P

S

10*3

.2P

S

1 0*x

.2P

S

1 0

*7.2

PS

10*8

. 2P

S

10* 9

. 2P

S

4.2

PS

1 0* 2

1 .2

PS

1 0*3

1.2

PS

10* 2

5Q

%

control

LTNT4

HTNT4

HTHT4

FIG. 3. CDR3 IGHD gene usage. All CDR3 (A) or repeated CDR3 (B)from immunized mice and control CDR3 were analyzed for the rel-ative frequency of IGHD gene usage. Note that only those IGHD genesthat were used at least once are shown in the histogram.

0

10

20

30

40

50

60

70

1JH

GI

2JH

GI

3 JH

G I

4JH

GI

%

control

LTNT4

HTNT4

HTHT4

0

20

40

60

80

100

120

1JH

GI

2JH

GI

3 JH

GI

4JH

GI

%

control

LTNT4

HTNT4

HTHT4

FIG. 4. CDR3 IGHJ gene usage. All CDR3 (left) or repeated CDR3 (right) from immunized mice and all control CDR3 were analyzed to determinethe relative frequency of IGHJ gene usage.

Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies Endocrinology, February 2007, 148(2):752–761 757

4), with TSHR immunization status, and TBII and T4 levels.In Table 5, we show that the IGHD-SP2.9 gene was found ata higher frequency in sequences from TSHR-immunizedmice relative to control (P � 0.04), but the sequences fromHTHT4 used IGHD-Q52 at a higher frequency than se-quences from the other TSHR-immunized mice (P � 0.05).IGHJ1 was found at a higher frequency in sequences fromTSHR-immunized mice compared with control (P � 0.005),and the reverse was found with IGHJ4 (P � 0.0001). IGHJ3was found in a higher frequency (P � 0.004) of TSHR-im-munized mice sequences compared with the HTHT4sequences.

There were fewer sulfur-containing amino acids in se-quences from TSHR-immunized mice, particularly in CDR3of HTHT4, compared with control (P � 0.0002). There wereno statistically significant associations with CDR3 length,which averaged 12 amino acids (Table 5). There were sig-nificantly fewer FR3 S mutations in the sequences fromTSHR-immunized mice (P � 0.048). There were no signifi-

cant differences in the R mutations or S mutations of CDR3between TSHR-immunized mice compared with control.However, there was a significant (P � 0.002) difference in themean CDR3 R mutations when CDR3 from the HTHT4 wascompared with the other TSHR-immunized mice with nor-mal T4 levels.

Discussion

In our EAGD model, animals are repeatedly immunizedwith TSHR, and anti-TSHR antibodies slowly evolve intoTSAbs resulting in hyperthyroidism. TSHR-immunized micecontained an average 4-fold increase in IGHV1 CDR3 tran-script levels compared with naive control; however, therewas no significant difference between different immunizedmice (not shown). The percentage of B220� cells after de-pletion was similar for all mice (not shown) and indicatedthat the increase in Ig transcripts was probably not due toincreased cell numbers and was a consequence of TSHR

FIG. 5. CDR3 diversity of similar se-quences. CDR3 of repeated sequences isshown. IMGT/V-QUEST (23) andIMGT/JunctionAnalysis (25) tools fromIMGT (http://imgt.cines.fr) (24) wereused to determine the V-D-J diversity ofeach sequence with respect to nucleo-tide (N) additions (TdT enzyme medi-ated), palindrome (P) additions and V,D, and J gene end-shortening (shown asdots). The number of repeated se-quences that each CDR3 represents isshown at the end of CDR3. CDR3 thatare boxed form part of a set of repeatedsimilar sequences.

TABLE 4. Mutations in the CDR3 and FR3

CDR3-IMGTR/S

FR3-IMGTR/S R(CDR3/R(FR3)

� R S � R S

All sequencesControl (n � 28) 0.54 0.46 0.07 6.6 1.5 0.96 0.5 1.9 0.48LTNT4 (n � 30) 0.13 0.07 0.07 1 1.6 0.57 1 0.6 0.12HTNT4 (n � 45) 0.58 0.4 0.16 2.5 1.8 0.96 0.89 1.1 0.42HTHT4 (n � 29) 1 0.93 0.1 9.3 1.6 1 0.55 1.8 0.93

Repeated sequencesControl (n � 28) 0.54 0.46 0.07 6.6 1.5 0.96 0.5 1.9 0.48LTNT4 (n � 7) 0 0 0 0 1.4 0 1.4 0 Ua

HTNT4 (n � 18) 0.78 0.56 0.22 2.5 1.72 0.72 1 0.7 0.78HTHT4 (n � 8) 1.13 1.13 0 Ub 1.63 0.88 0.75 1.2 1.28

The CDR3 of all sequences and repeated sequences were analyzed for R or S mutations using the IMGT/Junction Analysis tool (http://imgt.cines.fr) (25). FR3 were analyzed from positions 82–104, nucleotides 244–312, according to the IMGT unique numbering (36). Numbersrepresent the average mutation per CDR3. The R/S value was calculated as the average R mutation divided by the average S mutation in theCDR3.

a R(CDR3)/R(FR3) is unknown (U) because R (CDR3) � O.b R/S value is unknown (U) because CDR3 S mutations were not found.

758 Endocrinology, February 2007, 148(2):752–761 Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies

immunization. This is consistent with the robust spontane-ous secretion of anti-TSHR IgG Abs from splenocytes iso-lated from mice that were immunized with TSHR (Table 2).

The IGHV subgroup usage in control mice mirrored thepercentages of genes reported for each subgroup (Fig. 2). Thissuggested that the IgG repertoire is chosen, at least in part(37, 38), stochastically (39). A comparison of data from allmice showed that the IGHV response was similar (Fig. 2A)and the observed moderate increase in IGHV5 frequency inthe HTHT4 mouse might indicate a more focused response.It is interesting to note that the percentage of IGHV1 genesubgroup usage in our control mouse is approximately 15%higher than that reported in other studies (38, 40). The reasonfor this discrepancy is unclear, although it could be due todifferent assays used to determine IGHV usage (hybridiza-tions as opposed to PCR) and/or due to the depletion ofIgM� B cells.

We sequenced a representative number of the two mostabundant (34) IGHV subgroups (i.e. IGHV1 and IGHV5) thataccount for 55–80% of the mouse Ab repertoire (Fig. 2),including the 3� part of the FR3 and CDR3 of the rearrangedtranscripts (Fig. 1). The translated CDR3 showed no identitywith each other when sequences from different mice werecompared, indicating the diversity of responses in thesemice. There were, however, CDR3 amino acid sequences thatwere repeated (Table 3) within the same immunized mousebut not in the naive mouse. Because immunized mice hadhigher levels of CDR3 transcripts, it is likely that the repeatedproducts in these mice are not a consequence of limited initialtranscripts but dominant CDR3 usage in what is most likelyan oligoclonal B cell response. It is interesting to note that thegroups of similar sequences or clonal groups (Fig. 3) in thehigh-TBII mice (HTNT4 and HTHT4) contained small nucle-otide differences that translate into different amino acids (Rmutations) in their CDR3 that are indicative of an evolvingAb response.

Somatic hypermutation of Igs allows further diversifica-tion and refinement of Ab-antigen binding including in-creased affinity. The highest R mutation value in the CDR3(�0.93) was found in the diseased mouse (HTHT4) (Table 4).The frequency of R mutations in HTHT4 was 0.93, which ishigher than the frequencies of 0.46, 0.07, and 0.4 from control,LTNT4, and HTNT4 mice, respectively. The R mutation valuein the CDR3 for repeated sequences (Table 4) from HTHT4(1.13) was also higher than the frequency of R mutations in

LTNT4 (0), HTNT4 (0.56), and control (0.46) sequences. Onaverage, CDR3 sequences from the HTHT4 mouse had ac-cumulated more amino acid changes than any other mouse.Statistical analysis of HTHT4 sequences demonstrated nosignificant difference in the mean CDR3 S and FR3 R and Smutations relative to sequences from other TSHR-immu-nized mice. However, a statistically higher mean value ofCDR3 R mutations was noted in HTHT4 sequences relativeto sequences from other TSHR-immunized mice. These re-sults indicate that Igs from HTHT4 mice contained a higherfrequency of somatic mutations that alter the proteinsequence.

Several sets of murine anti-TSHR monoclonal Abs (mAbs)with TSAb activity have been generated. Analyses of two ofthese murine mAbs showed that one set of 2 Abs used IGHJ2and IGHD-ST4 or -Q52 (41) whereas the other set of 3 mAbsused IGHJ4 and IGHD-Q52 (5). Although repeated se-quences from the hyperthyroid mouse demonstrate a pref-erential use of IGHJ1 and IGHJ4 and IGHD-Q52 (Fig. 5),additional studies are required to draw a firm conclusionregarding an association with hyperthyroidism. More inter-estingly, a set of murine monoclonal TSAbs used IGHD-Q52and IGHJ4 (5) and contained average R/S values of 1.2, 0.63,and 1.0 in the FR and 6, 6, and 4 in the CDR of the Abs. TheR/S values of the HTNT4 sequences (Table 4) were (�2.5) lessthan the R/S values (4–6) of the monoclonal TSAbs. In con-trast, the sequences from the HTHT4 mouse showed R/Svalues of more than 9. Based on these results, we speculatethat in an evolving anti-TSHR Ab response, there is an in-creased probability that a pathogenic autoantibody willemerge as mutations accumulate within the CDR3 of thatautoantibody (i.e. increasing CDR3 R/S values).

Tolerance to TSHR is incomplete (42–44) and thus allowsthe possibility that stimulation of relevant B cells may leadto anti-TSHR Ab production. An earlier study demonstratedthat concentrated TSHR-specific autoantibodies fromhealthy individuals not only contain TBII activity but alsoshow the same TSHR N-terminal binding specificity as thatof pathogenic autoantibodies from GD patients (45). Thisraises the possibility that diversification of these Abs throughsomatic hypermutation may increase their affinity and leadto the production of pathogenic TSAbs.

Our CDR3 analysis focused on the antibody response gen-erated in the spleen after immunization with purified TSHRprotein. Unlike generating monoclonal TSAbs to TSHR or

TABLE 5. Associations between molecular characteristics of sequences and TSHR immunization status and T4 hormone levels

Molecular characteristicsTSHR-immunized vs. nonimmunized High T4 vs. TSHR-immunized

Finding P value Finding P value

V subgroup usage No NoD gene usage Yes (1SP2.9) 0.04 Yes (1Q52) 0.05J gene usage Yes (1J1, 2J4) 0.005, 0.0001 Yes (2J3) 0.004Amino acid group usage Yes (2S-containing) 0.0002 Yes (2S-containing) 0.02CDR3 length No NoCDR3 R mutations No 0.55 1Yes 0.002FR3 R mutations No 0.49 No 0.785CDR3 S mutations No 0.62 No 0.854FR3 S mutations 2Yes 0.048 No 0.06

Results reflect differences in frequencies of molecular characteristics among sequences derived from mice grouped by TSHR immunizationstatus and by T4 hormone levels. P values are based on the Wilcoxon signed-rank test.

Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies Endocrinology, February 2007, 148(2):752–761 759

separating out TSHR-specific B cells, potential anti-TSHRCD3 sequences were identified by certain molecular char-acteristics that were overrepresented (34). Despite this, wewere able to identify certain molecular characteristics asso-ciated with the TSAb activity. However, our ability to drawa more definitive conclusion is tempered by the paucity ofdata on monoclonal TSAb sequences. It should also be notedthat although a large number of studies have reported gen-eration of monoclonal antibodies to TSHR, there is verylimited information available on molecular characteristics ofTSAbs. As more monoclonal TSAb sequences become avail-able, we will have a greater ability to identify common patho-genic antibody molecular characteristics. Similarly, devel-opment of new techniques to isolate TSHR-specific B cellsshould allow a clearer characterization of the repertoire ofactivated TSHR-specific B cells. This later approach will notonly help detect the emergence of dominant specific B cellclones but also provide the context from which this popu-lation emerges.

Acknowledgments

We are very grateful to Veronique Giudicelli for her help in thesequence analysis with the newly improved IMGT/JunctionAnalysisprogram. We are also grateful to David Ucker and Prasad Kanteti forhelpful commentary.

Received August 11, 2006. Accepted October 17, 2006.Address all correspondence and requests for reprints to: Bellur S.

Prabhakar, Department of Microbiology and Immunology (M/C 790),Room E-709, Building 935, 835 South Wolcott Avenue, Chicago, Illinois60612. E-mail: [email protected].

This work was supported by National Institutes of Health Grant RO1DK047417.

Disclosure: The authors have nothing to disclose.

References

1. Prabhakar BS, Fan JL, Seetharamaiah GS 1997 Thyrotropin-receptor-medi-ated diseases: a paradigm for receptor autoimmunity. Immunol Today 18:437–442

2. Kim WB, Chung HK, Park YJ, Park DJ, Lee HK, Cho BY 2001 Clinicalsignificance of classification of Graves’ disease according to the characteristicsof TSH receptor antibodies. Korean J Intern Med 16:187–200

3. Ando T, Latif R, Pritsker A, Moran T, Nagayama Y, Davies TF 2002 Amonoclonal thyroid-stimulating antibody. J Clin Invest 110:1667–1674

4. Costagliola S, Franssen JD, Bonomi M, Urizar E, Willnich M, Bergmann A,Vassart G 2002 Generation of a mouse monoclonal TSH receptor antibody withstimulating activity. Biochem Biophys Res Commun 299:891–896

5. Sanders J, Jeffreys J, Depraetere H, Richards T, Evans M, Kiddie A, BreretonK, Groenen M, Oda Y, Furmaniak J, Rees Smith B 2002 Thyroid-stimulatingmonoclonal antibodies. Thyroid 12:1043–1050

6. Nordyke RA, Gilbert Jr FI, Miyamoto LA, Fleury KA 1993 The superiority ofantimicrosomal over antithyroglobulin antibodies for detecting Hashimoto’sthyroiditis. Arch Intern Med 153:862–865

7. Costagliola S, Vassart G 2002 Monoclonal antibodies with thyroid stimulatingactivity, at last. Thyroid 12:1039–1041

8. Weetman AP, Yateman ME, Ealey PA, Black CM, Reimer CB, Williams Jr RC,Shine B, Marshall NJ 1990 Thyroid-stimulating antibody activity betweendifferent immunoglobulin G subclasses. J Clin Invest 86:723–727

9. MacNeil S, Munro DS, Metcalfe R, Cotterell S, Ruban L, Davies R, WeetmanAP 1994 An investigation of the ability of TSH and Graves’ immunoglobulinG to increase intracellular calcium in human thyroid cells, rat FRTL-5 thyroidcells and eukaryotic cells transfected with the human TSH receptor. J Endo-crinol 143:527–540

10. McLachlan SM, Rapoport B 2004 Thyroid stimulating monoclonal antibodies:overcoming the road blocks and the way forward. Clin Endocrinol (Oxf)61:10–18

11. Sanders J, Evans M, Premawardhana LD, Depraetere H, Jeffreys J, RichardsT, Furmaniak J, Rees Smith B 2003 Human monoclonal thyroid stimulatingautoantibody. Lancet 362:126–128

12. Kabat EA, Wu TT, Reid-Miller M, Perry HM, Gottesman KS 1987 Sequences

of proteins of immunological interest. Bethesda, MD: National Institutes ofHealth

13. Lefranc MP, Lefranc G 2001 The immunoglobulin factsbook. London: Aca-demic Press

14. Ruiz M, Lefranc MP 2002 IMGT gene identification and Colliers de Perles ofhuman immunoglobulins with known 3D structures. Immunogenetics 53:857–883

15. McIntosh R, Watson P, Weetman A 1998 Somatic hypermutation in autoim-mune thyroid disease. Immunol Rev 162:219–231

16. Dogan RN, Vasu C, Holterman MJ, Prabhakar BS 2003 Absence of IL-4, andnot suppression of the Th2 response, prevents development of experimentalautoimmune Graves’ disease. J Immunol 170:2195–2204

17. Kaithamana S, Fan J, Osuga Y, Liang SG, Prabhakar BS 1999 Induction ofexperimental autoimmune Graves’ disease in BALB/c mice. J Immunol 163:5157–5164

18. Shimojo N, Kohno Y, Yamaguchi K, Kikuoka S, Hoshioka A, Niimi H, HiraiA, Tamura Y, Saito Y, Kohn LD, Tahara K 1996 Induction of Graves-likedisease in mice by immunization with fibroblasts transfected with the thyro-tropin receptor and a class II molecule. Proc Natl Acad Sci USA 93:11074–11079

19. Valente WA, Vitti P, Yavin Z, Yavin E, Rotella CM, Grollman EF, ToccafondiRS, Kohn LD 1982 Monoclonal antibodies to the thyrotropin receptor: stim-ulating and blocking antibodies derived from the lymphocytes of patients withGraves disease. Proc Natl Acad Sci USA 79:6680–6684

20. Kohn LD, Suzuki K, Hoffman WH, Tombaccini D, Marcocci C, Shimojo N,Watanabe Y, Amino N, Cho BY, Kohno Y, Hirai A, Tahara K 1997 Charac-terization of monoclonal thyroid-stimulating and thyrotropin binding-inhib-iting autoantibodies from a Hashimoto’s patient whose children had intra-uterine and neonatal thyroid disease. J Clin Endocrinol Metab 82:3998–4009

21. Shimojo N, Arima T, Yamaguchi K, Kikuoka S, Kohn LD, Kohno Y 2000 Anovel mouse model of Graves’ disease: implications for a role of aberrant MHCclass II expression in its pathogenesis. Int Rev Immunol 19:619–631

22. Kikuoka S, Shimojo N, Yamaguchi KI, Watanabe Y, Hoshioka A, Hirai A,Saito Y, Tahara K, Kohn LD, Maruyama N, Kohno Y, Niimi H 1998 Theformation of thyrotropin receptor (TSHR) antibodies in a Graves’ animalmodel requires the N-terminal segment of the TSHR extracellular domain.Endocrinology 139:1891–1898

23. Giudicelli V, Chaume D, Lefranc MP 2004 IMGT/V-QUEST, an integratedsoftware program for immunoglobulin and T cell receptor V-J and V-D-Jrearrangement analysis. Nucleic Acids Res 32:W435–W440

24. Lefranc MP, Giudicelli V, Kaas Q, Duprat E, Jabado-Michaloud J, ScavinerD, Ginestoux C, Clement O, Chaume D, Lefranc G 2005 IMGT, the interna-tional ImMunoGeneTics information system. Nucleic Acids Res 33:D593–D597

25. Monod MY, Giudicelli V, Chaume D, Lefranc MP 2004 IMGT/JunctionAn-alysis: the first tool for the analysis of the immunoglobulin and T cell receptorcomplex V-J and V-D-J JUNCTIONs. Bioinformatics 20(Suppl 1):I379–I385

26. Underhill GH, Kolli KP, Kansas GS 2003 Complexity within the plasma cellcompartment of mice deficient in both E- and P-selectin: implications forplasma cell differentiation. Blood 102:4076–4083

27. Waldmann TA, Strober W 1969 Metabolism of immunoglobulins. Prog Al-lergy 13:1–110

28. Slifka MK, Matloubian M, Ahmed R 1995 Bone marrow is a major site oflong-term antibody production after acute viral infection. J Virol 69:1895–1902

29. Manz RA, Lohning M, Cassese G, Thiel A, Radbruch A 1998 Survival oflong-lived plasma cells is independent of antigen. Int Immunol 10:1703–1711

30. McHeyzer-Williams MG 1997 Immune response decisions at the single celllevel. Semin Immunol 9:219–227

31. Sze DM, Toellner KM, Garcia de Vinuesa C, Taylor DR, MacLennan IC 2000Intrinsic constraint on plasmablast growth and extrinsic limits of plasma cellsurvival. J Exp Med 192:813–821

32. Jack HM, Wabl M 1988 Immunoglobulin mRNA stability varies during Blymphocyte differentiation. EMBO J 7:1041–1046

33. Chen-Bettecken U, Wecker E, Schimpl A 1987 Transcriptional control of �-and �-gene expression in resting and bacterial lipopolysaccharide-activatednormal B cells. Immunobiology 174:162–176

34. Delassus S, Gey A, Darche S, Cumano A, Roth C, Kourilsky P 1995 PCR-based analysis of the murine immunoglobulin heavy-chain repertoire. J Im-munol Methods 184:219–229

35. Dattamajumdar AK, Jacobson DP, Hood LE, Osman GE 1996 Rapid cloningof any rearranged mouse immunoglobulin variable genes. Immunogenetics43:141–151

36. Lefranc MP, Pommie C, Ruiz M, Giudicelli V, Foulquier E, Truong L,Thouvenin-Contet V, Lefranc G 2003 IMGT unique numbering for immu-noglobulin and T cell receptor variable domains and Ig superfamily V-likedomains. Dev Comp Immunol 27:55–77

37. Gu H, Tarlinton D, Muller W, Rajewsky K, Forster I 1991 Most peripheralB cells in mice are ligand selected. J Exp Med 173:1357–1371

38. Haines BB, Angeles CV, Parmelee AP, McLean PA, Brodeur PH 2001 Germ-line diversity of the expressed BALB/c VhJ558 gene family. Mol Immunol38:9–18

39. Dildrop R, Krawinkel U, Winter E, Rajewsky K 1985 VH-gene expression inmurine lipopolysaccharide blasts distributes over the nine known VH-genegroups and may be random. Eur J Immunol 15:1154–1156

760 Endocrinology, February 2007, 148(2):752–761 Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies

40. Freitas AA, Andrade L, Lembezat MP, Coutinho A 1990 Selection of VH generepertoires: differentiating B cells of adult bone marrow mimic fetal devel-opment. Int Immunol 2:15–23

41. Costagliola S, Bonomi M, Morgenthaler NG, Van Durme J, Panneels V,Refetoff S, Vassart G 2004 Delineation of the discontinuous-conformationalepitope of a monoclonal antibody displaying full in vitro and in vivo thyro-tropin activity. Mol Endocrinol 18:3020–3034

42. Akkaraju S, Canaan K, Goodnow CC 1997 Self-reactive B cells are not elim-inated or inactivated by autoantigen expressed on thyroid epithelial cells. J ExpMed 186:2005–2012

43. Akkaraju S, Ho WY, Leong D, Canaan K, Davis MM, Goodnow CC 1997 Arange of CD4 T cell tolerance: partial inactivation to organ-specific antigenallows nondestructive thyroiditis or insulitis. Immunity 7:255–271

44. Pichurin PN, Pichurina O, Marians RC, Chen CR, Davies TF, Rapoport B,McLachlan SM 2004 Thyrotropin receptor knockout mice: studies on immuno-logical tolerance to a major thyroid autoantigen. Endocrinology 145:1294–1301

45. Chazenbalk GD, Pichurin P, Chen CR, Latrofa F, Johnstone AP, McLachlanSM, Rapoport B 2002 Thyroid-stimulating autoantibodies in Graves diseasepreferentially recognize the free A subunit, not the thyrotropin holoreceptor.J Clin Invest 110:209–217

Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving theendocrine community.

Martinez et al. • Diversity in CDR3 of Anti-TSHR Antibodies Endocrinology, February 2007, 148(2):752–761 761