AUTOANTIBODIES to thyroid peroxidase (TPO) arising

Transcription

1 X/99/$03.00/0 Vol. 84, No. 5 The Journal of Clinical Endocrinology & Metabolism Printed in U.S.A. Copyright 1999 by The Endocrine Society Cellular Thyroid Peroxidase (TPO), Unlike Purified TPO and Adjuvant, Induces Antibodies in Mice That Resemble Autoantibodies in Human Autoimmune Thyroid Disease* JUAN CARLOS JAUME, JIN GUO, YAN WANG, BASIL RAPOPORT, AND SANDRA M. MCLACHLAN Autoimmune Disease Unit, Cedars-Sinai Research Institute and University of California-Los Angeles School of Medicine, Los Angeles, California ABSTRACT Autoantibodies to several protein antigens in human autoimmunity interact with a restricted range of epitopes, whereas diverse epitopes are recognized by antibodies induced in animals using antigen and adjuvant. To examine the basis for this difference, we compared the qualitative nature of antibodies developing in AKR/N mice injected with purified thyroid peroxidase (TPO) and adjuvant or with TPO expressed on major histocompatibility complex (MHC) class II fibroblasts. Mice injected with purified TPO had higher TPO antibody levels than TPO /class II fibroblast-treated mice. Despite lower titers, recipients of TPO /class II cells developed very high affinity antibodies (K d M), comparable with those of human TPO autoantibodies and about 10-fold higher than those in purified Received October 26, Received January 21, Accepted February 1, Address all correspondence and requests for reprints to: Dr. Sandra M. McLachlan, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Suite B-131, Los Angeles, California * This work was supported by NIH Grants DK and EY Present address: Endocrinology Division, Veterans Administration Medical Center and University of California, San Francisco, California TPO plus adjuvant-immunized mice. Moreover, more than 90% of TPO antibodies in TPO /class II fibroblast-injected mice, compared with only approximately 50% in TPO plus adjuvant-immunized mice, were to the immunodominant region recognized by patients autoantibodies. Consistent with this epitopic restriction, TPO /class II fibroblast-injected mice had TPO antibody epitopic fingerprints similar to those of human autoantibodies. In conclusion, mice injected with TPO /class II fibroblasts, but not those injected with purified TPO and adjuvant, develop antibodies closely resembling autoantibodies in human disease. These observations indicate that some animal models based on conventional immunization may not be representative of human diseases with a major humoral component. (J Clin Endocrinol Metab 84: , 1999) AUTOANTIBODIES to thyroid peroxidase (TPO) arising spontaneously in human thyroid autoimmunity interact with a restricted set of epitopes (reviewed in Ref. 1). In contrast, monoclonal antibodies generated in mice by immunization with the same antigen and adjuvant interact with a wide range of epitopes on TPO (2). Similar observations on restricted recognition by spontaneously arising human autoantibodies vs. diverse epitopes in experimentally immunized animals have been made for autoantibodies to other autoantigens, including thyroglobulin (3), the acetylcholine receptor (myasthenia gravis) (4, 5), and mitochondrial dehydrogenase (primary biliary cirrhosis) (6). Moreover, mice immunized with a variety of adjuvants and purified recombinant TSH receptor (TSHR) preparations of bacterial or insect cell origin do not develop antibodies resembling the stimulatory TSHR autoantibodies responsible for Graves hyperthyroidism (reviewed in Ref. 7). The basis for this qualitative difference between antibodies arising spontaneously vs. antibodies induced by immunization with purified antigen and adjuvant is not known. However, the manner of antigen presentation can modulate the subsequent immune response. Evidence for this possibility is provided by a novel approach used to develop the first animal model of Graves hyperthyroidism. The Shimojo model was achieved by injecting mice with fibroblasts coexpressing syngeneic major histocompatibility complex (MHC) class II and the TSHR (8, 9). Aberrant MHC class II expression on thyroid cells from patients with autoimmune thyroid disease was first observed in 1983 (10). The Shimojo approach is based on these early observations, and its success supports the hypothesis that thyroid cells may function as antigen-presenting cells (APC) and initiate the autoimmune response (11). The approach used in the Shimojo model now provides the opportunity to examine the relationship between antigen presentation to the immune system and the development of antibodies that resemble those observed in human thyroid autoimmunity. In the present study, we compared the qualitative nature of antibodies developing in the same strain of mice immunized by conventional means and by injecting fibroblasts expressing both antigen and MHC class II on their cell surface. The use of TPO, rather than the TSHR, in these studies offers a number of advantages, including the availability of large amounts of purified, native mammalian antigen (12) as well as a panel of human monoclonal autoantibodies that define the TPO-immunodominant region (13). Remarkably, although both approaches elicited TPO antibodies, only mice receiving fibroblasts coexpressing TPO and MHC class II fibroblasts develop antibodies that closely 1651

2 1652 JAUME ET AL. JCE&M 1999 Vol 84 No 5 resemble patients autoantibodies in terms of their high affinity and predominant recognition of the TPO-immunodominant region. Materials and Methods Fibroblast lines coexpressing TPO and MHC class II RT4.15HP fibroblasts (14) (provided by Dr. Ron Germain, NIH, Bethesda, MD), were propagated in DMEM with high glucose, 10% FCS, and antibiotics (Life Technologies,Gaithersburg, MD). Cells were stably transfected using lipofectin (Life Technologies) with complementary DNAs (cdnas) for 1) human TPO (pece-htpo cotransfected with psv2-neo) (15), 2) human TSHR (psv2-neo-ece-tshr 5,3 ) (16), 3) both TPO and TSHR (as above), and, as a control, 4) vector alone (psv2-neo-ece). Selection with G418 (400 mg/ml) and cloning by limiting dilution were performed using standard techniques. Cells, detached by mild trypsinization, were monitored for cell surface TPO expression by flow cytometry using a human monoclonal TPO autoantibody Fab (WR1.7) (13) followed by phycoerythrin-conjugated antihuman (Caltag, Burlingame, CA), as described previously (17). Fluorescence was analyzed using the Becton Dickinson and Co. FAC- Scan-CellQuest system (Mountain View, CA). Chinese hamster ovary (CHO) cells overexpressing TPO (18) provided a positive control. Negative controls were RT4.15HP cells (14) and untransfected CHO cells (DG44; provided by Dr. Robert Schimke, Stanford University, Palo Alto, CA). Assays included cells treated with second antibody alone and normal mouse serum. TSHR expression in transfected RT4.15HP monolayers was monitored by [ 125 I]TSH binding, as described for CHO cells (16). Clones expressing TPO alone, TSHR alone, or TPO and TSHR or vector-transfected cells were expanded in hypoxanthine-aminopterinthymidine medium (HAT, Sigma Chemical Co., St. Louis, MO) to preserve MHC class II expression. Class II expression was assessed by flow cytometry using fluorescein isothiocyanate-conjugated anti-ia-k(a a k ) (PharMingen, San Diego, CA). A line of RT4.15HP cells, propagated without HAT selection medium, provided a negative control. Induction of TPO antibodies using fibroblasts coexpressing TPO and MHC class II Six-week-old female AKR/N mice (NCI, Bethesda, MD) received six ip injections at 2-week intervals of RT4.15HP fibroblasts (10 7 cells/ injection) expressing TPO, TSHR, or TPO and TSHR or vector-transfected fibroblasts (five mice per group). Before transfer, fibroblasts were pretreated with 50 g/ml mitomycin C (Sigma Chemical Co.). This protocol is similar to that described for induction of TSHR antibodies using RT4.15HP-TSHR cells (8). Blood was obtained from the tail vein 2 weeks after the fifth injection and by cardiac puncture when the animals were killed 2 weeks after the sixth injection. All animal studies were performed in accordance with the highest standards of humane care. Conventional immunization with purified TPO and adjuvant Recombinant human TPO was prepared as previously reported (12). The membrane-bound TPO protein (933 amino acids, including signal peptide) was converted into an 848-residue molecule by introduction of a stop codon at the ectodomain/plasma membrane junction followed by transgenome amplification with a dihydrofolate reductase minigene (18). TPO secreted by CHO cells was affinity purified from culture medium. Concentration was determined by spectrophotometry at 280 nm optical density (extinction coefficient 17.9), and purity was determined by PAGE. AKR/N female mice (6 weeks old) were injected ip with purified TPO (50 g/mouse) in complete Freund s adjuvant (Sigma Chemical Co.). Two weeks later, immunization was repeated with the same antigen dose in incomplete Freund s adjuvant (Sigma Chemical Co.). A similar protocol is used to immunize mice against a variety of purified protein antigens including TPO (19). Blood was obtained from the tail vein 7 days after boosting to assess TPO antibody levels, and a final sample was collected by cardiac puncture at death 4 weeks after priming. Flow cytometric analysis of TPO antibodies Mouse sera (diluted 1:10) were analyzed using TPO-expressing CHO cells (18) and fluorescein isothiocyanate-conjugated, affinity-purified goat antimouse IgG (Caltag, South San Francisco, CA) as described previously (17). Assays included cells treated with second antibody alone and normal mouse serum. TPO antibody binding of [ 125 I]TPO and affinity for TPO The assay was performed as previously described (20). In brief, duplicate aliquots of mouse sera (diluted 1:20 unless otherwise specified) were incubated with [ 125 I]TPO (20,000 cpm; labeled using iodogen to a specific activity of 50 Ci/ g). To precipitate the antigen-antibody complex, protein A (Pansorbin; Calbiochem, La Jolla, CA) or antimouse IgG coupled to a solid phase (Sac Cel, IDS, Boldon, Tyne and Wear, U.K.) was added, and the incubation was continued. After the addition of assay buffer [0.1 m NaCl, 10 mm Tris-HCl (ph 7.5), 0.1% Tween-20, and 0.5% BSA], the mixture was vortexed and centrifuged, supernatants were removed, and radiolabeled TPO remaining in the pellets was counted. Nonspecific [ 125 I]TPO binding by normal mouse serum ( 3% of the total counts per min) was subtracted in calculating the percentage of [ 125 I]TPO bound by antibodies in mouse serum. Antibody affinities were determined by Scatchard analysis (21) from binding values obtained in the presence of increasing concentrations of unlabeled TPO. Interaction between mouse antibodies and the TPOimmunodominant region recognized by human autoantibodies [ 125 I]TPO binding by mouse serum antibodies was examined in the absence and presence of four recombinant human TPO-specific autoantibody Fab (SP1.4, WR1.7, TR1.8, and TR1.9) (13) that define the TPO-immunodominant region recognized by patients TPO autoantibodies (reviewed in Ref. 1). For simplification, the subdomains recognized by these Fab were previously renamed A1, A2, B1, and B2, respectively (22). Duplicate aliquots of serum were incubated with [ 125 I]TPO (20,000 cpm) alone or with the four Fab pools (each Fab at m). Recombinant Fab were prepared as previously described in detail (23). After 1hatroom temperature, complexes were precipitated using protein A, and radiolabeled TPO remaining in the pellets was counted. Fab lack the CH2 domain of the Fc region and are not precipitated by protein A. Preliminary experiments were performed to determine the serum antibody dilutions required to provide binding values of about 15% in the absence of TPO Fab. Such dilution is necessary to attain maximal inhibition of TPO binding by the addition of an excess concentration of Fab. Nonspecific [ 125 I]TPO binding ( 3% of the total counts per minute) was subtracted to calculate the percent inhibition by the TPO-specific Fab. Epitopic profiles of the mouse TPO antibodies were determined using the same assay, except competition for [ 125 I]TPO binding was performed in the presence of the individual human TPOspecific Fab that define the immunodominant region. Results Coexpression of TPO and MHC class II on RT4.15HP fibroblasts The RT4.15HP fibroblast line expresses a recombinant MHC class II molecule essentially identical to MHC class II (I-Ak) of AKR/N mice (14). TPO was readily detectable on MHC class II-expressing RT4.15HP cells transfected with the cdna for human TPO (Fig. 1A). However, as anticipated, the magnitude of fluorescence for the highest expressing clone was far less than that for CHO cells overexpressing TPO consequent to transgenome amplification (18) (Fig. 1B). Stable transfection with TPO cdna did not alter class II levels compared with those in the parent line (Fig. 1, C vs. D). Of the clones doubly transfected with TPO and TSHR cdna, TPO expression for the best expressing clone was slightly lower than that for the best expressor among the clones

3 MOUSE TPO ANTIBODIES RESEMBLING HUMAN AUTOANTIBODIES 1653 FIG. 1. Fibroblasts stably transfected with the human TPO cdna (RT4.15HP-TPO cells) express TPO and retain expression of MHC class II. Flow cytometry with TPO-specific Fab WR1.4 (followed by antihuman phycoerythrin) was used to assess cell surface TPO expression (solid histograms) on RT4.15HP-TPO cells (A) and CHO-TPO cells (B) overexpressing TPO consequent to transgenome amplification (29). MHC class II expression [fluorescein isothiocyanate-conjugated anti-ia-k(a a k )] was determined on RT4.15HP-TPO cells (C) and the original RT4.15HP cell line (D) (14). Controls (open histograms) are RT4.15HP cells (A), untransfected CHO cells (B), and RT4.15 HP cells negative for class II after propagation in the absence of hypoxanthine-aminopterin-thymidine medium (C and D). transfected with TPO cdna alone (median, 15.7 vs fluorescent units, respectively). Similarly, TSHR expression, assessed by [ 125 I]TSH binding, was lower in the optimal double transfectant than in the optimal TSHR-only transfected clone (14.9% vs. 16.2%, respectively; 4.3% binding by untransfected RT4.15HP cells). Eliciting TPO antibodies by injecting fibroblasts coexpressing TPO and MHC class II As assessed by flow cytometry with TPO-expressing CHO cells, sera from four of five AKR/N mice injected with TPO /class II RT4.15HP fibroblasts produced high levels of IgG class TPO antibodies relative to those in the five mice injected with TSHR /class II fibroblasts (t 4.44; P 0.002; Fig. 2A). These observations were confirmed by protein A precipitation of [ 125 I]TPO-IgG antibody complexes, an assay routinely used to detect patients serum TPO autoantibodies (24). Thus, sera from all five mice injected with TPO /class II fibroblasts bound significantly more TPO than sera from mice injected with TSHR /class II fibroblasts (t 7.55; P 0.001; Fig. 2B). Titers and affinities of TPO antibodies in AKR/N mice immunized conventionally or in conjunction with MHC class II As measured by flow cytometry and [ 125 I]TPO binding, AKR/N mice immunized with purified TPO and adjuvant developed higher levels of TPO antibodies than mice injected with TPO /class II fibroblasts. Thus, the median fluorescence was (mean sem; sera diluted 1:20) for five mice immunized with purified TPO compared with in similarly diluted sera from five mice injected with TPO fibroblasts. Similarly, [ 125 I]TPO binding was significantly greater in sera from conventionally immunized mice than in sera from mice injected with class II fibroblasts coexpressing TPO alone or with the TSHR (by ANOVA: P 0.001; F 21.29, 24.89, and for sera diluted 1:50, 1:100, and 1:500;

4 1654 JAUME ET AL. JCE&M 1999 Vol 84 No 5 FIG. 2. TPO antibodies are induced in AKR/N mice injected with TPO /class II, but not TSHR /class II, RT4.15HP cells. A, Flow cytometry of IgG class antibodies in mouse sera using TPO-expressing CHO cells. Each point depicts the median immunofluorescence units for serum (diluted 1:5) from an individual mouse. B, Direct binding by mouse IgG of [ 125 I]TPO as determined by precipitation with protein A. Each point indicates specific binding by serum (diluted 1:20) from an individual mouse. Solid circles, Mice injected with TPO /class II RT4.15HP cells; open circles, mice injected with TSHR /class II RT4.15HP cells; solid triangle, mouse injected with vector-only transfected RT4.15HP cells. *, P 0.002; **, P (by t test). Fig. 3). Coexpression of the TSHR with TPO had no significant effect on the titers of TPO antibodies. Antibody affinities were determined by Scatchard analysis (21) based on competition for radiolabeled TPO binding by unlabeled TPO. Despite lower titers, antibodies in AKR/N mice injected with TPO /class II fibroblasts were of higher affinity for TPO than antibodies in mice immunized by the conventional approach (representative data for one mouse in each group illustrated in Fig. 4). The mean affinity for TPO antibodies in mice injected with TPO /class II cells was m K d ( sd; n 3) vs m K d (n 3) in mice immunized with purified TPO and adjuvant (t 6.45; P 0.003). The higher antibody affinities in the former group of mice were comparable with those of autoantibodies in patients sera, namely K d of approximately m (24). Antibodies arising in response to TPO, class II cells are predominantly to the immunodominant region Because of the foregoing resemblance to human autoantibodies, we wished to determine whether the antibodies arising in response to TPO /class II fibroblasts shared other characteristics with human TPO autoantibodies, in particular recognition of the TPO-immunodominant region. This region of overlapping domains on native TPO is defined by four human monoclonal TPO autoantibodies in the form of recombinant Fab (13) (Fig. 5, upper panel). The concept that the majority of TPO autoantibodies within an individual patient s serum are to the immunodominant regions has been confirmed in a three-way collaborative comparison of the epitopes of the four original Fab with the epitopes of 1) a panel of murine monoclonal antibodies to TPO (2) and 2) FIG. 3. TPO antibody titers are higher in AKR/N mice immunized by the conventional approach using purified ( soluble ) TPO and adjuvant than in mice injected with fibroblasts coexpressing TPO and class II. Binding of [ 125 I]TPO was measured in sera diluted 1:50, 1:100, and 1:500. Data shown are the mean SEM (n 5). The hatched line represents the 95% confidence limits (upper range) for background binding by sera from three vector-immunized mice. Purified TPO vs. TPO /class II fibroblasts or purified TPO vs. TPO /TSHR / class II, fibroblasts; *, P 0.05 (by ANOVA). a panel of human Fab isolated independently by a different laboratory (20). Remarkably, competition studies with the pool of four recombinant Fab revealed that almost all antibodies arising in response to TPO /class II cells, unlike TPO antibodies in mice induced by conventional immunization, were to the TPO-immunodominant region. Thus, the four-fab pool inhibited almost completely ( 90%) TPO binding by serum antibodies in mice injected with class II fibroblasts expressing TPO, either alone or in combination with TSHR (Fig. 5, lower panel). In contrast, inhibition by the four Fab pool was significantly lower (only 50%) for TPO antibodies in the conventionally immunized mice (P 0.001; F 30.04; by ANOVA). Epitopic fingerprinting of TPO antibodies in immunized mice When used separately, the four human autoantibodies to the TPO-immunodominant region (Fab A1, A2, B1, and B2) can be used to determine a quantitative epitopic fingerprint for TPO autoantibodies in an individual patient (23; reviewed in Ref. 1). The epitopic fingerprints of mice injected with TPO /class II fibroblasts were similar to those observed in human autoantibodies (Fig. 6, upper and middle panels). For example, TPO antibodies in mouse 6 were markedly inhibited by Fab A1 and A2 ( 80%) and to a lesser extent by Fab B2 and B1 (50% or less), indicating preferential recognition of the A domain. In four mice (no. 5, 8, 9, and 10), Fab B2 was the dominant inhibitor ( 80%), whereas in three mice (no. 5, 4, and 10) TPO binding was reduced more than 75% by Fab B1. Comparable inhibition by A and B domain Fab was observed in three mice (no. 2, 3, and 7). The fingerprints were unrelated to percent inhibition by the four- Fab pool.

5 MOUSE TPO ANTIBODIES RESEMBLING HUMAN AUTOANTIBODIES 1655 FIG. 4. Higher affinity of TPO antibodies in mice injected with TPO /class II fibroblasts than in mice immunized by the conventional approach using purified TPO and adjuvant. A, Inhibition by unlabeled TPO of [ 125 I]TPO binding by IgG antibodies in a mouse injected with TPO /class II fibroblasts (solid symbols) and a mouse immunized with purified TPO plus adjuvant (open symbols). B and C, Scatchard analyses of the binding data shown in A. The data shown are for a representative mouse in each group. Mean affinities were M K d (n 3) in mice injected with TPO /class II cells and M K d (n 3) in mice immunized with purified TPO and adjuvant. In contrast, in mice immunized conventionally using purified TPO and adjuvant, the extent of antibody inhibition by individual Fab was lower, ranging from a maximum of 60% to less than 10% (Fig. 6, lower panels). Such low inhibition values reflect overlap with only about 50% of epitopes in the TPO-immunodominant region (Fig. 5). Not surprisingly, we have rarely observed epitopic profiles like those of purified TPO-immunized mice 13, 14, and 15 in human thyroid autoimmunity. Discussion Major differences were observed between TPO antibodies induced in AKR/N mice by injecting fibroblasts coexpressing TPO and MHC class II vs. immunization with purified TPO in adjuvant. Mice immunized by the conventional approach develop high titers of TPO antibodies, but these antibodies do not have the characteristic features of human TPO autoantibodies. Thus, the affinities of these induced TPO antibodies (K d 10 9 m) are about 10-fold lower than the affinities of autoantibodies in patients sera (24). Moreover, only 50% of the epitopes recognized by antibodies in purified TPO-immunized mice overlap with the epitopes in the immunodominant region recognized by the majority of TPO autoantibodies in all patients sera (reviewed in Ref. 1). In contrast, mice injected with TPO /class II fibroblasts develop lower TPO antibody titers. However, these murine antibodies closely resemble patients autoantibodies in terms of their high affinities (K d m), restricted epitopes (predominant recognition of the immunodominant region), and TPO epitopic fingerprints. It is intriguing and perhaps contrary to expectation that a heterologous antigen such as human TPO (or human TSHR, see introduction) expressed on the surface of class II fibroblasts, but not as purified antigen combined with adjuvant, is capable of inducing antibodies in recipient mice with properties similar to those of autoantibodies in humans. It should be emphasized that we do not know the basis for the differences among the titers, affinities for TPO, and antibody epitopes that develop using these immunization protocols. However, a number of factors may be involved including 1) immunization with monomeric purified TPO vs. dimeric TPO expressed on the cell surface, 2) use of Freund s adjuvant, 3) differences in antigen processing, and 4) differences in protein concentration. The first two parameters can probably be excluded. First, like patients autoantibodies (17), sera from mice immunized using either approach interact with both purified, monomeric radiolabeled TPO as well as with dimeric TPO expressed on the cell surface. Second, after immunization and screening with conformationally intact purified TPO, Ruf et al. (2) isolated a panel of mouse monoclonal antibodies that resembled serum antibodies in our adjuvant immunized mice in recognizing diverse epitopes on TPO (20) despite the use of a different adjuvant (Bordetella pertussis vs. Freund s adjuvant in our study). Two other factors are potentially important in eliciting the difference between the two forms of immunization. One possibility is that, unlike in purified TPO- and adjuvantimmunized mice, the class II, TPO fibroblasts are the major antigen presenting cells (APC) and present a different spectrum of peptides than conventional APC in T cell activation. This concept has been suggested with respect to the Shimojo model for inducing Graves disease-like stimulatory TSHR antibodies (see introduction). Stimulatory TSHR antibodies have also been generated by intramuscular immunization with naked TSHR cdna in mice pretreated with cardiotoxin to induce local muscle damage (25). Because human myoblasts incubated with cytokines are able to function as APC (26), it is conceivable that myoblasts are the major APC in this model. The possibility that professional and nonprofessional APC (such as thyroid cells or TPO-expressing fibroblasts) present different epitopes is consistent with the ability of some T cell clones to recognize TPO endogenously processed by autologous thyroid cells (27) or by TPO-transfected B cells (28, 29), but not exogenous, purified TPO presented by conventional APC (29). On the other hand, evidence against a role for nonprofessional APC in our study is the development of TSHR antibodies in mice receiving fibroblasts expressing the

6 1656 JAUME ET AL. JCE&M 1999 Vol 84 No 5 FIG. 5. Antibodies arising in response to cells expressing TPO and class II are predominantly to the immunodominant region recognized by human autoantibodies. Competition assays for [ 125 I]TPO binding were performed between individual mouse sera and four human Fab that define the immunodominant region (see Materials and Methods). Data are presented as the percent inhibition of TPO binding by the four-fab pool for sera from mice injected with TPO /class II fibroblasts (n 5) or TPO /TSHR /class II fibroblasts (n 5) and mice immunized with purified TPO in adjuvant. Data shown are the mean SEM. The difference between groups (by ANOVA) was P The difference between the purified TPO-immunized group and the other two groups was P TSHR in the absence of class II (30). However, few of these mice had TSHR antibodies with the stimulatory characteristics of Graves IgG, suggesting that coexpression of class II and TSHR influences induction of antibodies with an epitope(s) required for TSHR stimulation. Nevertheless, we cannot exclude the possibility that conventional APC are responsible for processing TPO in mice injected with a population of TPO fibroblasts that is alive but rendered incapable of dividing and destined to die in the peritoneum. Turning to antigen concentration, mice injected with TPOexpressing fibroblasts receive an extremely small dose of TPO. Based on the relative fluorescence of TPO-transfected fibroblasts and CHO cells (the latter expressing about 10 6 molecules/cell), the course of six injections of the former cells delivers, at most, 0.5 g TPO/mouse. In contrast, we and others (2, 19, 31) have used a total of g (generally in two doses) of purified human TPO for conventional immunization that has, intentionally or otherwise, resulted in TPO FIG. 6. Epitopic fingerprinting of TPO antibodies in immunized mice. AKR/N mice were injected with TPO /class II fibroblasts (upper panel), TPO /TSHR /class II fibroblasts (middle panel), and purified TPO in adjuvant (lower panel). Binding of [ 125 I]TPO to serum from individual mice was measured in the absence and presence of individual Fab (Fab A1, A2, B1, and B2) that define overlapping epitopes in the TPO-immunodominant region (see inset, C). Data are presented as the percent inhibition by Fab A1, A2, B1, or B2, with shading corresponding to that in the inset. Above each fingerprint is the percent inhibition obtained with the pool of four Fab for serum from each mouse. antibody production. Indeed, a lower dose of purified TPO ( 40 g) was ineffective for TPO antibody induction (19). We cannot be sure of the concentration of TPO available to the immune system in vivo using either of the two immunization protocols. However, the higher affinities of TPO antibodies arising in mice receiving TPO-transfected fibroblasts (vs. conventionally immunized mice) are consistent with a lower

7 MOUSE TPO ANTIBODIES RESEMBLING HUMAN AUTOANTIBODIES 1657 dose of antigen for the following reason. The progression from a primary to a secondary antibody response involves selection of B cells with surface Igs of progressively higher affinity (reviewed in Ref. 32). When antigen levels are limiting (as is likely to be the case with TPO /class II fibroblasts), only the highest affinity B cells can engage antigen and become activated. It should be emphasized that unlike T cells, which only recognize processed peptides, B cells usually bind conformationally intact protein antigens. As mentioned above, TPO antibodies arising in mice injected with TPO-expressing fibroblasts recognize TPO on transfected cells. In contrast to the present data on TPO antibodies, neither the epitopic repertoire nor the affinities of antibodies have been compared in mice immunized with fibroblasts expressing the TSHR and MHC class II (8, 9) vs. purified TSHR and a variety of adjuvants (reviewed in Ref. 7). Our present observations for TPO suggest that TSHR antibodies generated using the Shimojo protocol may have a restricted epitopic repertoire and higher affinities for the TSHR than antibodies in conventionally immunized mice. Moreover, it is conceivable that the unusual characteristic of TSHR antibodies that only develop in mice injected with TSHR-expressing fibroblasts, namely their ability to stimulate the thyroid, may be related to affinity and/or epitopic restriction. In conclusion, injection of AKR/N mice with fibroblasts coexpressing TPO and class II, but not conventional immunization with purified antigen, induces antibodies that closely resemble patients serum TPO autoantibodies. These observations in the same strain of mice indicate that some animal models based on conventional immunization using high doses of purified antigen may not be representative of human autoimmune diseases with a major humoral component. Acknowledgments We thank Dr. Ronald N. Germain, Laboratory for Immunology, Lymphocyte Biology Section, NIAID, NIH, for generously providing us with the RT4.15HP fibroblast line. References 1. McLachlan SM, Rapoport B Genetic and epitopic analysis of thyroid peroxidase (TPO) autoantibodies: markers of the human thyroid autoimmune response. Clin Exp Immunol. 101: Ruf J, Toubert M, Czarnocka B, Durand-Gorde J, Ferrand M, Carayon P Relationship between immunological structure and biochemical properties of human thyroid peroxidase. Endocrinology. 125: Nye L, Pontes de Carvalho LC, Roitt IM Restrictions in the response to autologous thyroglobulin in the human. Clin Exp Immunol. 41: Tzartos SJ, Seybold ME, Lindstrom JM Specificities of antibodies to acetylcholine receptors in sera from myasthenia gravis patients measured by monoclonal antibodies. Proc Natl Acad Sci USA. 79: Heidenreich F, Vincent A, Roberts A, Newsom-Davis J Epitopes on human acetylcholine receptor defined by monoclonal antibodies and myasthenia gravis sera. Autoimmunity. 1: Rowley MJ, Maeda T, MacKay IR, Loveland BE, McMullen GL, Tribbick G, Bernard CCA Differing epitope selection of experimentally-induced and natural antibodies to a disease-specific autoantigen, the E2 subunit of pyruvate dehydrogenase complex (PDC-E2). Int Immunol. 4: Rapoport B, Chazenbalk GD, Jaume JC, McLachlan SM The thyrotropin receptor: interaction with thyrotropin and autoantibodies. Endocr Rev. 19: Shimojo N, Kohno Y, Yamaguchi K, et al Induction of Graves-like disease in mice by immunization with fibroblasts transfected with the thyrotropin repector and a class II molecule. Proc Natl Acad Sci USA. 93: Kita M, Ahmad R, Marians RC, Vlase H, Unger P, Graves P, Davies TF Regulation and transfer of a murine model of thyrotropin receptor antibody mediated Graves disease. Endocrinology. 140: Hanafusa T, Pujol-Borrell R, Chiovato L, Russell RCG, Doniach D, Bottazzo GF, Feldmann M Aberrant expression of HLA-DR antigen on thyrocytes in Graves disease: relevance for autoimmunity. Lancet. 2: Bottazzo GF, Pujol-Borrell R, Hanafusa T, Feldmann M Role of aberrant HLA-DR expression and antigen presentation in induction of endocrine autoimmunity. Lancet. 2: Guo J, McLachlan SM, Hutchison JS, Rapoport B The greater glycan content of recombinant human thyroid peroxidase of mammalian than on insect cell origin facilitates purification to homogeneity of enzymatically active protein remaining soluble at high concentration. Endocrinology. 139: Chazenbalk GD, Portolano S, Russo D, Hutchison JS, Rapoport B, McLachlan SM Human organ-specific autoimmune disease: molecular cloning and expression of an autoantibody gene repertoire for a major autoantigen reveals an antigenic dominant region and restricted immunoglobulin gene usage in the target organ. J Clin Invest. 92: Germain RN, Ashwell JD, Lechler RI, Margulies DH, Nickerson KM, Suzuki G, Tou JYL Exon-shuffling maps control of antibody- and T-cellrecognition sites to the NH 2 -terminal domain of the class II major histocompatibility polypeptide A. Proc Natl Acad Sci USA. 82: Magnusson RP, Chazenbalk GD, Gestautas J, Seto P, Filetti S, DeGroot LJ, Rapoport B Molecular cloning of the complementary deoxyribonucleic acid for human thyroid peroxidase. Mol Endocrinol. 1: Kakinuma A, Chazenbalk G, Filetti S, McLachlan SM, Rapoport B Both the 5 and 3 non-coding regions of the thyrotropin receptor messenger RNA influence the level of receptor protein expression in transfected mammalian cells. Endocrinology. 137: Nishikawa T, Rapoport B, McLachlan SM Exclusion of two major areas on thyroid peroxidase from the immunodominant region containing the conformational epitopes recognized by human autoantibodies. J Clin Endocrinol Metab. 79: Kaufman KD, Foti D, Seto P, Rapoport B Overexpression of an immunologically-intact, secreted form of human thyroid peroxidase in eukaryotic cells. Mol Cell Endocrinol. 78: McLachlan SM, Atherton MC, Nakajima Y, Napier J, Jordan RK, Clark F, Rees Smith B Thyroid peroxidase and the induction of autoimmune thyroid disease. Clin Exp Immunol. 79: Guo J, McIntosh RS, Czarnocka B, Weetman A, Rapoport B, McLachlan SM Relationship between autoantibody epitopic recognition and immunoglobulin gene usage. Clin Exp Immunol. 111: Scatchard G The attractions of proteins for small molecules and ions. Ann NY Acad Sci. 51: Jaume JC, Burek CL, Hoffman WH, Rose N, McLachlan SM, Rapoport B Thyroid peroxidase autoantibody epitopic fingerprints in juvenile Hashimoto s thyroiditis: evidence for conservation over time and in families. Clin Exp Immunol. 104: Nishikawa T, Costante G, Prummel MF, McLachlan SM, Rapoport B Recombinant thyroid peroxidase autoantibodies can be used for epitopic fingerprinting of thyroid peroxidase autoantibodies in the sera of individual patients. J Clin Endocrinol Metab. 78: Beever K, Bradbury J, Phillips D, et al Highly sensitive assays of autoantibodies to thyroglobulin and to thyroid peroxidase. Clin Chem. 35: Costagliola S, Rodien P, Many M-C, Ludgate M, Vassart G Genetic immunization against the human thyrotropin receptor causes thyroiditis and allows production of monoclonal antibodies recognizing the native receptor. J Immunol. 160: Goebels N, Michaelis D, Wekerle H, Hohlfeld R Human myoblasts as antigen-presenting cells. J Immunol. 149: Dayan CM, Londei M, Corcoran AE, Grubeck-Loebenstein B, James RFL, Rapoport B, Feldmann M Autoantigen recognition by thyroid-infiltrating T cells in Graves disease. Proc Natl Acad Sci USA. 88: Martin A, Magnusson RP, Kendler DL, Concepcion E, Ben-Nun A, Davies TF Endogenous antigen presentation by autoantigen-transfected Epstein-Barr virus-lymphoblastoid cells. J Clin Invest. 91: Mullins RJ, Chernajovsky Y, Dayan C, Londei M, Feldmann M Transfection of thyroid autoantigens into EBV-transformed B cell lines. J Immunol. 152: Yamaguchi K, Shimojo N, Kikuoka S, et al Genetic control of antithyrotropin receptor antibody generation in H-2k mice immunized with thyrotropin receptor-transfected fibroblasts. J Clin Endocrinol Metab. 82: Kotani T, Umeki K, Hirai K, Ohtaki S Experimental murine thyroiditis induced by porcine thyroid peroxidase and its transfer by the antigen-specific T cell line. Clin Exp Immunol. 80: Berek C, Ziegner M The maturation of the immune response. Immunol Today. 14: