- Email: [email protected]
Autoantigenic determinants on human thyroglobulin
CLINICAL
IMMUNOLOGY
AND
Autoantigenic
54, 64-75 (i%o)
lMMUNOPATHOLOGY
Determinants
I. Determinant Specificities
on Human Thyroglobulin
of Murine Monoclonal
Antibodies
HERBERTS.BRESLER,' C. LYNNE BUREK, AND NOEL R.ROSE Deportment of Immunology Hygiene und Public
and Infectious Diseusrs. Health, 615 North Wolje
The Johns Hopkins liniversity, School Street, Baltimore, Mapland 21205
(!j
To map the antigenic determinants, a panel of 20 mouse monoclonal antibodies (mAbs) to native human thyroglobuhn (Tg) was generated. Four criteria were established for distinguishing the determinants recognized by the various mAbs: ti) reactivity to Tgs from eight different species; (ii) reactivity to oxidized and reduced human Tg; (iii) the ability of thyroxine to inhibit binding of mAbs; and (iv) the pattern of antigenic determinant reactivity as determined in reciprocal competitive inhibition binding assays. Of 20 mAbs examined, 12 bound only to human Tg. 3 bound to all eight species tested. and 5 bound to human Tg and to Tg from at least one of the other seven species. Eight mAbs bound oxidized/reduced and native human Tg equally well, 2 bound to the oxidized/ reduced protein better than to the native, 7 showed virtually no reactivity to oxidized/ reduced human Tg, and 3 others reacted to the oxidized/reduced protein significantly less than they bound to the native. The binding of 5 mAbs to native human Tg was inhibited by thyroxine. The pattern of shared determinant specificities revealed that at least I2 determinant clusters were defined by the panel of mAbs. Among the 12 determinant clusters were 3 designated as immunodominant; X of 20 mAbs defined these immunodominant clusters. The four criteria taken together indicate that at least 19 different epitopes on human Tg could be distinguished by this panel of mAbs. These mAbs are useful for the study of determinant specificities of Tg autoantibodies in humans. ‘1 Iwo Academic
Pre?\.
Inc
INTRODUCTION
Autoantibodies to Tg are found in most patients with chronic thyroiditis ( I. 2), but also appear in many apparently normal individuals without any clinical or biochemical evidence of thyroid disease (2-4). Several investigations have been recently conducted (5-8) to discover the possible differences between the autoantibodies to Tg which appear in diseased individuals and the autoantibodies which appear in healthy individuals, Some of these studies have used panels of mAbs to characterize the antigenic sites on the human Tg molecule in order to distinguish the determinants recognized by human autoantibodies to Tg. A precise characterization of the determinants recognized by the autoantibodies from patients, normal individuals, and individuals with known subclinical disease will increase our understanding of the pathogenesis of the disease. This study presents a new and expanded panel of mAbs to human Tg. The methods used for the detailed characterization of this large panel are described, and the distribution and salient properties of antigenic determinants on human Tg, as defined by these mAbs, are examined. The use of this panel to study the ’ Present address: Biotronic Systems Corp.. Kockville, MD 20850. 64 OOYO-1229190$1.50 Copyright 0 1990 by Academic Press. Inc. All nghts of reproduction in any form reserved.
AUTOANTIGENIC
DETERMINANTS
ON
HUMAN
THYROGLOBULIN
65
determinant specificities of autoantibodies from humans is the subject of a companion paper which compares the autoimmune response to Tg in normal and diseased individuals. Each of the determinants recognized by the mAbs was also recognized by human autoantibodies. MATERIALS
AND METHODS
Thyroids. Normal human thyroid tissue was obtained at autopsy. Mouse thyroid glands were obtained from freshly killed BALB/c mice; thyroids from rats, guinea pigs, rabbits, cattle, and swine were purchased frozen from Pel-Freez Biologicals (Rogers, Arkansas); lesser (red) panda thyroid was kindly provided frozen by Dr. S. Kennedy-Stoskopf (Division of Comparative Pathology, JHU School of Medicine). All tissues were stored at -70°C until used. Thyroglobulin. Thyroglobulin was prepared with borate-buffered saline and purified by gel filtration on Sepharose CL4B (Pharmacia, Inc., Piscataway, NJ) (9-11). Protein determinations. The absorbance at 280 nm of the gel filtration eluate fractions was measured and protein concentrations were estimated based on ‘an extinction coefficient for Tg of 1.0 (mg/ml) (12, 13). Alternatively, a microplate adaptation of the bicinchoninic acid method (14, 15) was used to determine concentrations of Tg and monoclonal Ig in purified preparations. Monoclonal antibodies. Monoclonal antibodies were produced by standard methods (16) using the Sp210 mouse myeloma (17). Hybridomas were cloned by limiting dilution in 96-well culture plates, as described by Oi and Herzenberg (18). An alternative method for cloning was the soft agar technique. Production and purification of mAbs. The mAbs for biotinylation were either grown in culture in low serum (0.5% FCS)-containing medium (Biotain-MPS or KC2000, Hazelton Research Products, Lenexa, KS) or as ascites in female BALB/c mice. The mAb was precipitated with saturated ammonium sulfate according to the method of Herbert (19). Further purification was either by ionexchange chromatography or by affinity chromatography on protein A-Sepharose (MAPS-II preparative antibody purification kit, Bio-Rad, Rockville Center, NY). Biotinylation. Purified mAbs for use in ELISAs were labeled with biotin via their carbohydrate moieties by a modification of the method of O’Shannessay et al. (20). The biotinylation procedure deviated from that of O’Shannessay in that the antibody was oxidized with 10 mM sodium periodate for 30 min at O-XC and that the dry biotin hydrazide was mixed with the oxidized Ig samples by gently homogenizing in a Dounce apparatus. ELZSA. Antibody to human Tg obtained from hybridomas was measured by ELISA, using a modification of the method of Voller et al. (21). Human Tg solution, 50 ~1, was added to each well of Immulon-2 flat-bottom plates (Dynatech, Chantilly, VA) and incubated 22.5 hr in a humid chamber at 4°C. The plates were washed with phosphate-buffered saline (PBS)-Tween, and either used immediately or wrapped in plastic bags and stored at -20°C for up to 4 weeks. Antibody samples were diluted in 100 ~1 PBS-Tween containing 0.04% N, and incubated in human Tg-coated ELISA plate wells at room temperature in a humid
66
BRESLER,
BUREK,
AND
ROSE
chamber for 60 min, then washed. Affinity-purified, peroxidase-conjugated goat anti-murine Ig antibody (Jackson Immunoresearch, West Grove, PA) in PBSTween with 2% normal goat serum (NGS) was added to each well, incubated, and washed as before. The orthophenylene diamine (OPD) substrate (100 pg/ml OPD [predissolved in 2-3 ml methanol] + 3 mg/ml Hz02) in PBS was added and incubated at room temperature in the dark for 45 min, at which time 3 M sulfuric acid was added to stop the reaction. Optical density (OD) at 495 nm was measured spectrophotometrically. Species reactivity of the mAbs. Cross-reactivities of the anti-human Tg mAbs with Tg from other species were assessed in ELISAs using Tg from the other species. Reactivity to oxidized/reduced human 7g. Human Tg-coated ELISA plates, warmed to room temperature, were incubated with 1 mM sodium periodate in PBS at room temperature for 30 min. This was followed by 1 mlM sodium cyanoborohydride in PBS incubated at room temperature for 30 min. ELISAs were then performed as described above. Subtyping of the mAbs. Isotypes of mAbs were determined in another ELISA using anti-murine Ig subclass-specific rabbit antibodies followed by peroxidaseconjugated anti-rabbit Ig antibody (both from Boehringer-Mannheim Biochemicals, Indianapolis, IN). Affinity determinations. To determine the affinity of each of the mAbs for native human Tg, the procedure of Friguet et al. (22) was used. Reactivity of the mAbs to thyroxine ( T4) and triiodothyronine ( Ts). Serial halflog,, dilutions of each T, and T, (Sigma) were made in PBS-Tween-NGS-N,. Each mAb was added to a set of tubes containing T, and to a set containing T,. Inhibition of a mAb by T, or T, was detected by a decrease in OD of at least 25% compared to that of the same mAb without T, or T, present. Determinant reactivities of the mAbs. Biotinylated mAbs were used in ELISAs to determine whether the epitopes recognized by the various mAbs were different The ability of each unlabeled monoclonal antibody to inhibit the binding of the same or other labeled monoclonal antibody was determined by incubating human Tg-coated plastic microwells with various amounts of unlabeled monoclonal antibody, followed, without washing or removing the first antibody, by a constant amount of the same or other biotin-labeled monoclonal antibody. All mAbs were diluted in PBS-Tween-NGS-N,. After washing, the wells were incubated with avidin-peroxidase conjugate (Vector Laboratories, Burlingame, CA), in PBSTween-NGS, for 1 hr at room temperature. The wells were washed again, and the OPD substrate was then added. The mAbs were arranged so that the controls necessary for proper interpretation of results were included on each plate. RESULTS
Isotype and Reactivity
to OxidizedlReduced
Human
Tg
The 20 mAbs, their isotypes, and their reaction with oxidized/reduced Tg relative to native human Tg are shown in Table 1.
human
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
TABLE LIST
OF THE
20
MONOCLONAL
ANTIBODIES
1 (mAbs)
Clone
Heavy chain” class
Relative reaction with oxidized/ reduced human Tgb
133A2A 133A2G 134C2 134c4 137B2 132C5 145A6 133c4 155A4 156A6 42C3 41c3 137Cl 154C6 151Cl 152D3 133Bl 42C4 121B2 41A5
A GI Gl Gl Gl Gl Gl Gl Gl Gl Gl M Gl G2b Gl G2b G2a M G2a Gl
99 97 96 105 106 38 6 2 94 2 108 229 64 79 9 7 2 127 87 17
AND
67
THYROGLOBULIN
THEIR
CHARACTERISTICS
KJM-') 1.14 9.49 2.62 7.04 5.28 5.21 1.62 4.55 1.03 2.52 8.31 1.23 2.39 2.34 1.66 3.84 7.73 3.33 1.06 2.74
x x x x x x x x x x x x x x x x x x x x
10-s 1o-9 10-i’ lo-i0 lo-’ IO-” lo-’ 10-8 1o-9 10m9 10-i” 1O-6 1O-9 1O-9 lo-’ lo-’ 10-7 1o-6 1O-9 lo-’
Reactivity with T, or T,c
T, + T, T, + T,
T.6 T, + T, T, + T,d
u Heavy chain subclass as determined by ELISA. All mAbs had K light chains. b Human Tg was oxidized with 1 mM sodium periodate for 30 min and reduced with 1 mM sodium cyanoborohydride for 30 min. Reaction with oxidized/reduced Tg given as optical density relative to the reaction with native Tg defined as 100%. c Each mAb that was inhibited by T, or T, is designated. d Indicates that the mAb was only inhibited at high concentrations of T, or T,.
AfJinity of the mAbs for Native Human
Tg
The dissociation constants for the 20 mAbs are shown in Table 1. The most avid mAb (132C5) bound human Tg with aK,of 5.21 x lo-” M-‘, while the least avid (145A6) had a I& of 1.62 x lop5 M-‘. Most K, were in the range of lo-‘. Correlation coeffkients for the regression lines were 0.987 to >0.999, with all but one greater than 0.990. Reactivity
to T4 and T3
Denoted in Table 1 are the five mAbs which were inhibited by free T4 or T, in solution. Three of the mAbs were inhibited by T, and T, at concentrations of 31.6 pg/ml or greater. One mAb, 41A5, was inhibited by T, and T, at 316 t&ml or greater. Another mAb, 133B1, was only inhibited at concentrations of T, greater than 100 pg/ml, and was not inhibited at all by T,. T, and T, became saturated at the higher concentrations used and began to form precipitates. Precipitation was seen above 316 pg/ml of T, and above 100 t&ml of T, despite the presence of dimethyl sulfoxide (DMSO). Additionally, since a significant amount of DMSO
68
BRESLER,
BUREK,
AND
ROSE
was present in the highest concentrations of T, and T, (11 and 21%~ DMSO, respectively), the DMSO may have inhibited the binding of mAbs to Tg. Therefore, inhibition of mAbs by the highest concentrations of T4 and T, were difficult to evaluate. Reactivity to Tg from Other Animal Species Figure I shows the reaction of the mAbs with human Tg and Tg from seven other animal species. Some of the epitopes recognized by the panel of mAbs are shared by all of the animal species tested (e.g., 41C3), while others are restricted to human Tg (e.g., 134C2 and 133Bl). Still other epitopes are present on human Tg and Tg of at least one of the other species tested, but not all species Tg (e.g.. 154C6 and 12lB2). Shared Determinant
Reactivities
Reciprocal competitive binding inhibition ELISAs were carried out. Data were used only from experiments in which all the controls in a series of nine plates performed satisfactorily. Representative data from experiments examining the competitive inhibition of four mAbs to native human Tg by one labeled mAb (156A6) are shown in Fig. 2. The greatest inhibition of biotin-labeled 156A6 by its unlabeled counterpart was taken as lOO%, and the inhibitions of other labeled mAbs by unlabeled 156A6 were evaluated relative to that amount. A mAb was considered to have blocked another when the maximum relative inhibition of the labeled mAb was greater than 25% and was decreased to approximately 0 upon dilution of unlabeled mAb (for example, mAb 155A4 in Fig. 2). In each case, unlabeled mAb blocked the binding of its labeled counterpart by at least 40% at a dilution of 1: 10. In the example shown in Fig. 2, unlabeled mAb 156A6 inhibited the binding of its labeled counterpart and labeled 155144, but mAbs 133A2A and 154C6 were not inhibited by mAb 156A6. Collation
of Inhibition
Data
A summary of the competitive inhibition experiments 12 apparent determinant clusters are designated.
is presented in Fig. 3. The
DISCUSSION
Hybridomas have proven to be useful and appropriate tools to analyze the antigenic sites on the surface of human Tg. The possible confounding effects of different affinities among mAbs could be mitigated by performing the inhibition tests with suitable dilutions of reactants. By allowing relatively high concentrations of the first mAb to react to the Tg on the ELISA plates for a long time and by diluting labeled second mAbs to relatively low concentrations and allowing them to react for a relatively short time, the confounding effects created by displacement of unlabeled mAbs by higher afftnity labeled mAbs were diminished. The validity of this technique was confirmed by the observation that mAb 145A6 with a Kd of 1.62 x 10e5 M-’ inhibited the binding of mAb 132C5 with a Kd of 5.21 X lo-‘i M-’ (see Table 1 and Fig. 3). In this way, the interpretation of results was made more straightforward. A similar
AUTOANTIGENIC 133A2A
DETERMINANTS
133AXi
ON
HUMAN
100
0
THYROGLOBULIN
134c2
~g,~~~~~~~~~~
0
100
0
100
0
100
0
100
0
relative l7Cl
154C6
blndlng
100
0
mouse rat I-” guinea pig = .-E rabbi 5
0
100
0
100
0
100
(%)
151Cl
152D3
3381
nrrr
human
100
42C4
2182
4lA5
1 1 1
bovine porcine red panda 0
100
0
100
0
Jr!-LL. 100
0
100
0
100
0
100
0
100
0
100
0
100
0
relative binding (%) 1. Bar graphs showing reactivity of 20 mAbs to Tg from seven mammalian species relative to their reactivity to human Tg. For each mAb (listed at the top of each bar graph), each horizontal bar represents the relative reactivity to Tg from the animal species listed at the left. The binding of the mAbs was measured at a dilution of 1:lO and the relative binding to Tg from animal species was calculated as follows: FIG.
OD animal Tg % relative binding = OD human Tg X 100.
approach has been successfully used by a number of investigators for characterizing the interrelationship of antigenic sites recognized by mAbs, and has been verified in a number of different antigen-antibody systems, including Tg-anti-Tg (5, 6, 23-29).
100
70
BRESLER,
BUREK,
AND
ROSE
loo 3 & ul
90
z E
*
0 z!
m
3
6o 50
6 5
40
..tr P E .-c
3o
: '5 m z
a
20 10 0 -10 lb
Reciprocal
dilution
of unlabeled
mAb 156A6
FIG. 2. Results of representative inhibition binding experiments showing that unlabeled mAb 156A6 inhibits its labeled counterpart and labeled 15SA4, but does not inhibit labeled mAbs 133A2A or 154C6. Therefore, mAbs 156A6 and ISA4 share determinant specificities, whereas 133A2A and 154C6 do not share specificities with 156A6.
The inability of some mAbs to bind periodate/cyanoborohydride-treated human Tg provided information on the nature of the epitopes recognized by these mAbs. It is likely that this oxidation/reduction procedure affected carbohydrate moieties on the Tg molecule by cross-linking them to the polypeptide portion of the molecule (20). The mAbs which had diminished or completely lost the ability to bind Tg oxidized and reduced by this method may be reacting via carbohydrate or carbohydrate-dependent epitopes, or via epitopes dependent upon one or more of the cross-linked amino acids. The two mAbs that reacted significantly better to oxidized/reduced human Tg were the only two IgM antibodies in the panel (41C3 and 42C4). There are several likely explanations for these observations, two of which are presented here. First, these IgM mAbs may represent “natural” autoantibodies from the normal antibody repertoire of the immunized mouse. Natural autoantibodies are likely to be low affinity and cross-reactive, qualities both of these mAbs exhibit. Therefore, it is quite likely that natural autoantibodies may react better with a distinctly foreign, denatured antigen than with a native one. This observation has been made by other investigators when looking at the reactivity of mAbs to denatured versus native antigens (30). A second explanation for greater reactivity to a denatured antigen involves the size of the IgM molecule relative to IgG. Other mAbs in this
AUTOANTIGENIC
DETERMINANTS
Unlabeled l,,,,,,,I
ON
HUMAN
71
THYROGLOBULIN
Monoclonal I I
Antibody I I I I I I
;5 0
ev a z
Y m A
137Cl _ 154C6 15lCl . 152D3 1 133611 42C4 12182
; (
I
I
I
I I I I I I I I I I I..... I I I I >($C
FIG. 3. Compiled results of reciprocal competitive inhibition binding assays of 20 mAbs showing proposed determinant clusters of human Tg. Each of 20 mAbs was tested in competition assays against all of the others for binding sites on human Tg (see Materials and Methods). When an unlabeled mAb blocked the binding of a labeled mAb, the appropriate square in this figure was filled in. After all the appropriate squares were tilled in, the columns and rows were rearranged (while being kept internally intact) to bring as many filled squares together as possible. The order in which the mAbs were listed on the top was made the same as the order on the left. The 12 apparent determinant clusters were then designated.
panel (e.g., 42C3) reacted slightly more with oxidized/reduced human Tg than with native. This slight increase in OD could have been exaggerated due to the large mass of pentameric IgM protein bound relative to monomeric IgG bound, and subsequently detected by a conjugated antiserum to murine Igs, even though in molar terms the amount of bound IgG and IgM could have been approximately equal. Information concerning the conservation of epitopes in other mammalian species was obtained by testing the reactivity of the mAbs with Tg from several other mammalian species. Most mAbs (16/20) recognized epitopes that are not widely conserved among animal species; and even though no effort was made to select for mAbs which recognized uniquely human epitopes, 12 of 20 mAbs bound only human Tg. Of the four mAbs which recognized epitopes widely shared among several animal species (including the evolutionarily remote lesser panda), three could be inhibited by T, and T,. These three were the only mAbs which bound to Tg from all species tested. This finding suggests that Tg may vary considerably in
72
BRESLER,
BUREK,
AND
ROSE
its antigenic properties from one species to another, while the thyroxine itself. and the hormonogenic regions from which they arise, are largely conserved. Evidence that the regions neighboring the thyroxines are not identical comes from the reactivity patterns of two mAbs (133Bl and 41A5). The pattern of shared determinant reactivities presented in Fig. 3 indicates that there are topologically continuous antigenic regions on the native human Tg molecule. Determinant clusters 1, 2, and 3 represent a continuum of partially overlapping epitopes. These three overlapping determinant clusters have been designated as “immunodominant” since 8 of the twenty mAbs in the panel define sites in these clusters. All of the mAbs in the immunodominant clusters react with human Tg, and not with Tg from seven other species. Comparison of the reactivities of the mAbs to oxidized/reduced human Tg (Table l), with their arrangement in Fig. 3, reveals further evidence that this topological relationship exists. All of the mAbs in clusters 1 and 2 reacted equally well with oxidized/reduced and native human Tg, except 132C5, which reacted about half as well with oxidizedi reduced as the native protein. Thus, 132C5 represents a transition from cluster 2 to cluster 3 in which all of the other epitopes were destroyed by oxidation/ reduction. Relationships among mAbs within determinant clusters were rein forced by the observation that within given clusters the mAbs shared the same profiles of reactivity to Tg from different species (compare Figs. I and 3). Since each mAb blocked the binding of its labeled counterpart and the order of labeled mAbs was kept the same as the order of unlabeled mAbs, the resulting pattern in Fig. 3 was a solid diagonal of filled squares. The proposed determinant clusters also arrange about this diagonal due to the fact that most inhibitions (50/84) are two-way; that is, if mAb, blocked mAb,, in most cases, mAb, blocked mAb,. However, one-way inhibitions are not uncommon (34184) and could represent special topological relationships between epitopes. A likely explanation for a one-way inhibition in which mAb, blocked mAbz, but mAb, failed to block mAb,, is that the epitope defined by mAb, resides at the top of a cleft in the surface of the molecule and the epitope defined by mAb, resides lower in the same cleft. Another possible explanation of one-way inhibition is that the binding ot mAb, alters the conformation of the epitope recognized by mAbZ. but not vice versa. Some topologically related epitopes are revealed by the high number of oneway inhibitions exhibited by a few mAbs. The best example is 133A2G, which was inhibited by 12 mAbs while it only inhibited 3 mAbs. The mAb 133A2G defines an epitope with a unique position on human Tg. There are at least two explanations for these observations. Epitope 133A2G may reside on or near the junction between the two monomers of the dimeric human Tg, or may be situated in the bottom of a large cleft, or both. Such an epitope would be in a prime position for the mAb which binds it to be blocked by many other mAbs, while itself not being able to block the other mAbs. Another set of one-way inhibitions which may reveal a cleft in the Tg molecule is seen in determinant cluster 12 (Fig. 3). The inhibition pattern of this cluster is consistent with the notion that the epitopes defined by 1370, 154C6, and 15lCl are inside a cleft which is “guarded” at its top by epitopes 152D3 and 133Bl.
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
THYROGLOBULIN
13
When the different reactivities of the mAbs are taken together, 19 different epitopes can be discerned. The only two mAbs that apparently shared the same epitope specificity were 134C2 and 134C4. Each of the other epitopes defined by these mAbs could be distinguished from the others in the panel by one or more of the above-described criteria (Figs. 1 and 3, Table I). This analysis of antigenic sites on human Tg is more extensive than previous ones (5-8, 28, 29, 31-33) in two important ways: (i) this panel of mAbs defines a greater diversity of determinants than any previously published panel of mAbs to human Tg; and (ii) the characteristics of the epitopes defined by this panel are better defined as a result of the combination of studies used. No one of the four criteria used to distinguish the nature or interrelationships of the epitopes would have given the insights that were gained through the use of this combination of criteria. Detailed analysis of this new panel of mAbs by a variety of means has enabled us to distinguish 19 different epitopes on the human Tg molecule. The purpose of the creation and detailed characterization of this panel of mAbs is to be able to use the mAbs in competition assays with human sera to probe the epitope specificities of autoantibodies from patients with thyroid disease. This ail-important first step will enable us to understand better the nature of autoantibodies found in humans. ACKNOWLEDGMENTS We are pleased to acknowledge the advice of Drs. Sharon Krag and Howard Dint& in the performance of these experiments. This work was supported in part by NIH Grants AR31632. AG04362, and AR35383.
REFERENCES 1. Roitt. 1. M., Doniach, D., Campbell, P. N., and Hudson, R. V., V: Autoantibodies in Hashimoto’s disease. Lance? 2, 820-821. 1956. 2. Bigazzi, P. E., and Rose, N. R., Autoimmune thyroid disease. In “The Autoimmune Diseases” (N. R. Rose and I. R. Mackay, Eds.), pp. 161-199, Academic Press, New York, 1985. 3. Aho, K.. Gordin. A., Sievers, K.. and Takala, J.. Thyroid autoimmunity in siblings: A population study. Acta Endocrinol. (Copenhagen) Suppl. 251, 11-15, 1983. 4. Scherbaum, W. A., Stockle, G., Wichmann, J., and Berg, P. A., Immunological and clinical characterization of patients with untreated euthyroid and hypothyroid autoimmune thyroiditis. antibody spectrum, response to TRH and clinical study. Acta Endocrinol. (Copenhagen) 100, 373-381, 1982. 5. Ruf, J., Carayon. P., Sarles-Philip, N., Kourilsky, F., and Lissitzky, S., Specificity of monoclonal antibodies against human thyroglobulin; comparison with autoimmune antibodies. EMBO. J. 2, 1821-1826, 1983. 6. Ruf, J., Carayon. P., and Lissitzky , S., Various expressions of a unique anti-human thyroglobulin antibody repertoire in normal state and autoimmune disease. Eur. J. Immunol. 15, 268-272. 1985. 7. Piechaczyk, M., Bouanani, M., Salhi, S. L., Baldet, L., Bastide, M.. Pau, B., and Bastide, J.-M., Antigenic domains on the human thyroglobulin molecule recognized by autoantibodies in patients’ sera and by natural autoantibodies isolated from the sera of healthy subjects. C&n. Zmmuno/. Immunopathol. 45, 114-121, 1987. 8. Chan, C. T. J., Byfield, P. G. H., Himsworth, R. L., and Shepherd, P., Human autoantibodies to thyroglobulin are directed towards a restricted number of human specific epitopes. Clin. i&p. Immunol. 69, 516-523, 1987. 9. Chemoff, S. B.. and Rawitch, A. B., Thyroglobulin structure-function: Isolation and character-
74
10.
11. 12. 13.
BRESLER,
BUREK.
AND
ROSE
ization of a thyroxine containing polypeptide from bovine thyroglobulin. J. Biol. Chem. 256, 9425-9430, 1981. Kong, Y., David, C. S.. Giraldo, A. A., Elrehewy, M.. and Rose. N. R., Regulation of autoim mune response to mouse thyroglobulin: Influence of H-2D-end genes. J Immunol. 123, 15-18, 1979. Mishell, B. B., and Shiigi, S. M.. Appendix A, In “Selected Methods in Cellular Immunology“ (B. B. Mishell and S. M. Shiigi, Eds.), pp. 443-456, Freeman, San Francisco. 1980. Haeberli, A.. Salvatore, G., Edelhoch, H., and Rail, .I. E., Relationship between iodination and the polypeptide chain composition of thyroglobulin. J. Biol. Chem. 250, 7836-7841. 1975. Rose, N. R.. and Styles, W. A., Splitting of human thyroglobulin. I. Reduction and alkylation. Clin.
Exp.
Immunol.
5, 129-140.
1969.
14. Sorensen. K.. and Brodbeck, U.. A sensitive protein assay method using micro-titer plates. Experientiu 42, 161-162, 1986. 15. Redinbaugh, M. Cl., and Turley. R. B., Adaptation of the bicinchoninic acid protein assay for use with microtiter plates and sucrose gradient fractions. Anal. Biochem. 153, 267-271, 1986. 16. Fazekas de St. Groth, S., and Scheidegger, D.. Production of monoclonal antibodies: Strategy and tactics. .I. Immunol. Methods 35, l-21, 1980. 17. Shulman, M., Wilde , C. D., and Kohler, G., A better cell line for making hybridomas secreting specific antibodies. Nature (London) 276, 269-270, 1978. 18. Oi, V. T., and Herzenberg, L. A., Immunoglobulin-producing hybrid cell lines. In “Selected Methods in Cellular Immunology” (B. B. Mishell and S. M. Shiigi, Eds.), pp. 351-372, Freeman. San Francisco, 1980. 19. Herbert, G. A., Ammonium sulfate fractionation of sera: Mouse, hamster, gumea pig, monkey. chimpanzee, swine. chicken, and cattle. Appl. Microbial. 27, 389-393, 1974. 20. O’Shannessy, D. J., Dobersen, M. J.. and Quarles, R. H., A novel procedure for labeling immunoglobulins by conjugatione to oligosaccharide moieties. Immunol. Lett. 8, 273-277, 1984. 21. Voller, A., Bidwell. D. E., and Burek. C. L., An enzyme-linked immunosorbent assay @LISA) for antibodies to thyroglobulin. Proc. Sot. Exp. Biol. Med. 163, 402-405. 1980. 22. Friguet, B., Chaffotte, A. F., Djavadi-Ohaniance, L., and Goldberg, M. E.. Measurements of the true affinity constant in solution of antigen-antibody complexes by enzyme-linked immunosorbent assay. J. Immunol. Methods 77, 305-319, 1985. 23. Heinz. F. X., Tuma, W., Guirakhoo, F., Berger, R., and Kunz, C.. lmmunogenicity of tick-borne encephalitis virus glycoprotein fragments: Epitope-specific analysis of the antibody response. .1. Gen. Virol. 65, 1921-1929, 1984. 24. Heinz, F. X., Berger, R., Tuma, M., and Kunz, C., A topological and functional model of epitopes on the structural glycoprotein of tick-borne encephalitis virus defined by monoclonal antibodies. Virology 126, 525-537, 1983. 25. Anderson, L. J., Hierholzer, J. C.. Stone, Y. O., Tsou, C., and Fernie. B. F., Identification of epitopes on respiratory syncytial virus proteins by competitive binding immunoassay. J. C/in. Microbial. 23, 4754180, 1986. 26. Rebai, N., Malissen, B., Pierres, M., Accolfa, R. S.. Corte, G., and Mawas, C., Distinct HLA-DR epitopes and distinct families of HLA-DR molecules defined by 15 monoclonal antibodies (mAb) either anti-DR or allo-anti-Iak. I. Cross-inhibition studies of mAb cell surface fixation and differential binding of mAb to detergent-solubilized HLA molecules immobilized to a solid phase by a first mAb. Eur. J. Immunol. 13, 106-111. 1983. 27. Pierres, M., Kourilsky, F. M., Rebouah, J.-P., Dosseto, M., and Caillol, D., Distinct epitopes on Ik gene products identified by monoclonal antibodies. Eur. J. Immunol. 10, 950-957, 1980. 28. DeBaets, M., Janssens, A., Vlek, L., Nakamura, R., Weigle, W. 0.. Vermuelen, A., and van Breda Vriesman, P., Structure of human thyroglobulin studied using monoclonal antibodies. Ann. Endocrinol. (Paris) 43, Al 18, 1985. [Abstract] 29. Kotani, T., Maeda, M.. Ohtaki, S., Kobayashi, S., and Okuyama. H., Analyses of the specificity of mouse monoclonal antibodies to syngeneic and allogeneic thyroglobulins. Clin. Immunol. fmmunopathol. 34, 147-157, 1985.
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
THYROGLOBULIN
75
30. Friguet, B., Djavadi Ohaniance, L., and Goldberg, M. E., Some monoclonal antibodies raised with a native protein bind preferentially to the denatured antigen. Mol. Zmmunol. 21, 673-677, 1984. 31. Piechaczyk, M., Chardes, T., Cot, M.-C., Pau, B., and Bastide, J.-M., Production and characterization of monoclonal antibodies against human thyroglobulin. Hybridoma 4, 361-367, 1985. 32. Kurata, A., Ohta, K., Mine, M., Fukuda, T.. Ikari, N., Kanazawa, H., Matsunaga, M., Izumi, M., and Nagataki, S., Monoclonal antihuman thyroglobulin antibodies. J. Clin. Endocrinol. Metab. 59, 573-579, 1984. 33. Bellet, D., Schlumberger, M., Bidart, J. M., Assicot, M., Caillou, B., Motte, P., Vignal, A., and Bohuon, C., Production and in vitro utilization of monoclonal antibodies to human thyroglobulin. J. Clin. Endocrinol. Metab. 56, 530-533, 1983. Received May 15, 1989; accepted with revision August 8, 1989
IMMUNOLOGY
AND
Autoantigenic
54, 64-75 (i%o)
lMMUNOPATHOLOGY
Determinants
I. Determinant Specificities
on Human Thyroglobulin
of Murine Monoclonal
Antibodies
HERBERTS.BRESLER,' C. LYNNE BUREK, AND NOEL R.ROSE Deportment of Immunology Hygiene und Public
and Infectious Diseusrs. Health, 615 North Wolje
The Johns Hopkins liniversity, School Street, Baltimore, Mapland 21205
(!j
To map the antigenic determinants, a panel of 20 mouse monoclonal antibodies (mAbs) to native human thyroglobuhn (Tg) was generated. Four criteria were established for distinguishing the determinants recognized by the various mAbs: ti) reactivity to Tgs from eight different species; (ii) reactivity to oxidized and reduced human Tg; (iii) the ability of thyroxine to inhibit binding of mAbs; and (iv) the pattern of antigenic determinant reactivity as determined in reciprocal competitive inhibition binding assays. Of 20 mAbs examined, 12 bound only to human Tg. 3 bound to all eight species tested. and 5 bound to human Tg and to Tg from at least one of the other seven species. Eight mAbs bound oxidized/reduced and native human Tg equally well, 2 bound to the oxidized/ reduced protein better than to the native, 7 showed virtually no reactivity to oxidized/ reduced human Tg, and 3 others reacted to the oxidized/reduced protein significantly less than they bound to the native. The binding of 5 mAbs to native human Tg was inhibited by thyroxine. The pattern of shared determinant specificities revealed that at least I2 determinant clusters were defined by the panel of mAbs. Among the 12 determinant clusters were 3 designated as immunodominant; X of 20 mAbs defined these immunodominant clusters. The four criteria taken together indicate that at least 19 different epitopes on human Tg could be distinguished by this panel of mAbs. These mAbs are useful for the study of determinant specificities of Tg autoantibodies in humans. ‘1 Iwo Academic
Pre?\.
Inc
INTRODUCTION
Autoantibodies to Tg are found in most patients with chronic thyroiditis ( I. 2), but also appear in many apparently normal individuals without any clinical or biochemical evidence of thyroid disease (2-4). Several investigations have been recently conducted (5-8) to discover the possible differences between the autoantibodies to Tg which appear in diseased individuals and the autoantibodies which appear in healthy individuals, Some of these studies have used panels of mAbs to characterize the antigenic sites on the human Tg molecule in order to distinguish the determinants recognized by human autoantibodies to Tg. A precise characterization of the determinants recognized by the autoantibodies from patients, normal individuals, and individuals with known subclinical disease will increase our understanding of the pathogenesis of the disease. This study presents a new and expanded panel of mAbs to human Tg. The methods used for the detailed characterization of this large panel are described, and the distribution and salient properties of antigenic determinants on human Tg, as defined by these mAbs, are examined. The use of this panel to study the ’ Present address: Biotronic Systems Corp.. Kockville, MD 20850. 64 OOYO-1229190$1.50 Copyright 0 1990 by Academic Press. Inc. All nghts of reproduction in any form reserved.
AUTOANTIGENIC
DETERMINANTS
ON
HUMAN
THYROGLOBULIN
65
determinant specificities of autoantibodies from humans is the subject of a companion paper which compares the autoimmune response to Tg in normal and diseased individuals. Each of the determinants recognized by the mAbs was also recognized by human autoantibodies. MATERIALS
AND METHODS
Thyroids. Normal human thyroid tissue was obtained at autopsy. Mouse thyroid glands were obtained from freshly killed BALB/c mice; thyroids from rats, guinea pigs, rabbits, cattle, and swine were purchased frozen from Pel-Freez Biologicals (Rogers, Arkansas); lesser (red) panda thyroid was kindly provided frozen by Dr. S. Kennedy-Stoskopf (Division of Comparative Pathology, JHU School of Medicine). All tissues were stored at -70°C until used. Thyroglobulin. Thyroglobulin was prepared with borate-buffered saline and purified by gel filtration on Sepharose CL4B (Pharmacia, Inc., Piscataway, NJ) (9-11). Protein determinations. The absorbance at 280 nm of the gel filtration eluate fractions was measured and protein concentrations were estimated based on ‘an extinction coefficient for Tg of 1.0 (mg/ml) (12, 13). Alternatively, a microplate adaptation of the bicinchoninic acid method (14, 15) was used to determine concentrations of Tg and monoclonal Ig in purified preparations. Monoclonal antibodies. Monoclonal antibodies were produced by standard methods (16) using the Sp210 mouse myeloma (17). Hybridomas were cloned by limiting dilution in 96-well culture plates, as described by Oi and Herzenberg (18). An alternative method for cloning was the soft agar technique. Production and purification of mAbs. The mAbs for biotinylation were either grown in culture in low serum (0.5% FCS)-containing medium (Biotain-MPS or KC2000, Hazelton Research Products, Lenexa, KS) or as ascites in female BALB/c mice. The mAb was precipitated with saturated ammonium sulfate according to the method of Herbert (19). Further purification was either by ionexchange chromatography or by affinity chromatography on protein A-Sepharose (MAPS-II preparative antibody purification kit, Bio-Rad, Rockville Center, NY). Biotinylation. Purified mAbs for use in ELISAs were labeled with biotin via their carbohydrate moieties by a modification of the method of O’Shannessay et al. (20). The biotinylation procedure deviated from that of O’Shannessay in that the antibody was oxidized with 10 mM sodium periodate for 30 min at O-XC and that the dry biotin hydrazide was mixed with the oxidized Ig samples by gently homogenizing in a Dounce apparatus. ELZSA. Antibody to human Tg obtained from hybridomas was measured by ELISA, using a modification of the method of Voller et al. (21). Human Tg solution, 50 ~1, was added to each well of Immulon-2 flat-bottom plates (Dynatech, Chantilly, VA) and incubated 22.5 hr in a humid chamber at 4°C. The plates were washed with phosphate-buffered saline (PBS)-Tween, and either used immediately or wrapped in plastic bags and stored at -20°C for up to 4 weeks. Antibody samples were diluted in 100 ~1 PBS-Tween containing 0.04% N, and incubated in human Tg-coated ELISA plate wells at room temperature in a humid
66
BRESLER,
BUREK,
AND
ROSE
chamber for 60 min, then washed. Affinity-purified, peroxidase-conjugated goat anti-murine Ig antibody (Jackson Immunoresearch, West Grove, PA) in PBSTween with 2% normal goat serum (NGS) was added to each well, incubated, and washed as before. The orthophenylene diamine (OPD) substrate (100 pg/ml OPD [predissolved in 2-3 ml methanol] + 3 mg/ml Hz02) in PBS was added and incubated at room temperature in the dark for 45 min, at which time 3 M sulfuric acid was added to stop the reaction. Optical density (OD) at 495 nm was measured spectrophotometrically. Species reactivity of the mAbs. Cross-reactivities of the anti-human Tg mAbs with Tg from other species were assessed in ELISAs using Tg from the other species. Reactivity to oxidized/reduced human 7g. Human Tg-coated ELISA plates, warmed to room temperature, were incubated with 1 mM sodium periodate in PBS at room temperature for 30 min. This was followed by 1 mlM sodium cyanoborohydride in PBS incubated at room temperature for 30 min. ELISAs were then performed as described above. Subtyping of the mAbs. Isotypes of mAbs were determined in another ELISA using anti-murine Ig subclass-specific rabbit antibodies followed by peroxidaseconjugated anti-rabbit Ig antibody (both from Boehringer-Mannheim Biochemicals, Indianapolis, IN). Affinity determinations. To determine the affinity of each of the mAbs for native human Tg, the procedure of Friguet et al. (22) was used. Reactivity of the mAbs to thyroxine ( T4) and triiodothyronine ( Ts). Serial halflog,, dilutions of each T, and T, (Sigma) were made in PBS-Tween-NGS-N,. Each mAb was added to a set of tubes containing T, and to a set containing T,. Inhibition of a mAb by T, or T, was detected by a decrease in OD of at least 25% compared to that of the same mAb without T, or T, present. Determinant reactivities of the mAbs. Biotinylated mAbs were used in ELISAs to determine whether the epitopes recognized by the various mAbs were different The ability of each unlabeled monoclonal antibody to inhibit the binding of the same or other labeled monoclonal antibody was determined by incubating human Tg-coated plastic microwells with various amounts of unlabeled monoclonal antibody, followed, without washing or removing the first antibody, by a constant amount of the same or other biotin-labeled monoclonal antibody. All mAbs were diluted in PBS-Tween-NGS-N,. After washing, the wells were incubated with avidin-peroxidase conjugate (Vector Laboratories, Burlingame, CA), in PBSTween-NGS, for 1 hr at room temperature. The wells were washed again, and the OPD substrate was then added. The mAbs were arranged so that the controls necessary for proper interpretation of results were included on each plate. RESULTS
Isotype and Reactivity
to OxidizedlReduced
Human
Tg
The 20 mAbs, their isotypes, and their reaction with oxidized/reduced Tg relative to native human Tg are shown in Table 1.
human
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
TABLE LIST
OF THE
20
MONOCLONAL
ANTIBODIES
1 (mAbs)
Clone
Heavy chain” class
Relative reaction with oxidized/ reduced human Tgb
133A2A 133A2G 134C2 134c4 137B2 132C5 145A6 133c4 155A4 156A6 42C3 41c3 137Cl 154C6 151Cl 152D3 133Bl 42C4 121B2 41A5
A GI Gl Gl Gl Gl Gl Gl Gl Gl Gl M Gl G2b Gl G2b G2a M G2a Gl
99 97 96 105 106 38 6 2 94 2 108 229 64 79 9 7 2 127 87 17
AND
67
THYROGLOBULIN
THEIR
CHARACTERISTICS
KJM-') 1.14 9.49 2.62 7.04 5.28 5.21 1.62 4.55 1.03 2.52 8.31 1.23 2.39 2.34 1.66 3.84 7.73 3.33 1.06 2.74
x x x x x x x x x x x x x x x x x x x x
10-s 1o-9 10-i’ lo-i0 lo-’ IO-” lo-’ 10-8 1o-9 10m9 10-i” 1O-6 1O-9 1O-9 lo-’ lo-’ 10-7 1o-6 1O-9 lo-’
Reactivity with T, or T,c
T, + T, T, + T,
T.6 T, + T, T, + T,d
u Heavy chain subclass as determined by ELISA. All mAbs had K light chains. b Human Tg was oxidized with 1 mM sodium periodate for 30 min and reduced with 1 mM sodium cyanoborohydride for 30 min. Reaction with oxidized/reduced Tg given as optical density relative to the reaction with native Tg defined as 100%. c Each mAb that was inhibited by T, or T, is designated. d Indicates that the mAb was only inhibited at high concentrations of T, or T,.
AfJinity of the mAbs for Native Human
Tg
The dissociation constants for the 20 mAbs are shown in Table 1. The most avid mAb (132C5) bound human Tg with aK,of 5.21 x lo-” M-‘, while the least avid (145A6) had a I& of 1.62 x lop5 M-‘. Most K, were in the range of lo-‘. Correlation coeffkients for the regression lines were 0.987 to >0.999, with all but one greater than 0.990. Reactivity
to T4 and T3
Denoted in Table 1 are the five mAbs which were inhibited by free T4 or T, in solution. Three of the mAbs were inhibited by T, and T, at concentrations of 31.6 pg/ml or greater. One mAb, 41A5, was inhibited by T, and T, at 316 t&ml or greater. Another mAb, 133B1, was only inhibited at concentrations of T, greater than 100 pg/ml, and was not inhibited at all by T,. T, and T, became saturated at the higher concentrations used and began to form precipitates. Precipitation was seen above 316 pg/ml of T, and above 100 t&ml of T, despite the presence of dimethyl sulfoxide (DMSO). Additionally, since a significant amount of DMSO
68
BRESLER,
BUREK,
AND
ROSE
was present in the highest concentrations of T, and T, (11 and 21%~ DMSO, respectively), the DMSO may have inhibited the binding of mAbs to Tg. Therefore, inhibition of mAbs by the highest concentrations of T4 and T, were difficult to evaluate. Reactivity to Tg from Other Animal Species Figure I shows the reaction of the mAbs with human Tg and Tg from seven other animal species. Some of the epitopes recognized by the panel of mAbs are shared by all of the animal species tested (e.g., 41C3), while others are restricted to human Tg (e.g., 134C2 and 133Bl). Still other epitopes are present on human Tg and Tg of at least one of the other species tested, but not all species Tg (e.g.. 154C6 and 12lB2). Shared Determinant
Reactivities
Reciprocal competitive binding inhibition ELISAs were carried out. Data were used only from experiments in which all the controls in a series of nine plates performed satisfactorily. Representative data from experiments examining the competitive inhibition of four mAbs to native human Tg by one labeled mAb (156A6) are shown in Fig. 2. The greatest inhibition of biotin-labeled 156A6 by its unlabeled counterpart was taken as lOO%, and the inhibitions of other labeled mAbs by unlabeled 156A6 were evaluated relative to that amount. A mAb was considered to have blocked another when the maximum relative inhibition of the labeled mAb was greater than 25% and was decreased to approximately 0 upon dilution of unlabeled mAb (for example, mAb 155A4 in Fig. 2). In each case, unlabeled mAb blocked the binding of its labeled counterpart by at least 40% at a dilution of 1: 10. In the example shown in Fig. 2, unlabeled mAb 156A6 inhibited the binding of its labeled counterpart and labeled 155144, but mAbs 133A2A and 154C6 were not inhibited by mAb 156A6. Collation
of Inhibition
Data
A summary of the competitive inhibition experiments 12 apparent determinant clusters are designated.
is presented in Fig. 3. The
DISCUSSION
Hybridomas have proven to be useful and appropriate tools to analyze the antigenic sites on the surface of human Tg. The possible confounding effects of different affinities among mAbs could be mitigated by performing the inhibition tests with suitable dilutions of reactants. By allowing relatively high concentrations of the first mAb to react to the Tg on the ELISA plates for a long time and by diluting labeled second mAbs to relatively low concentrations and allowing them to react for a relatively short time, the confounding effects created by displacement of unlabeled mAbs by higher afftnity labeled mAbs were diminished. The validity of this technique was confirmed by the observation that mAb 145A6 with a Kd of 1.62 x 10e5 M-’ inhibited the binding of mAb 132C5 with a Kd of 5.21 X lo-‘i M-’ (see Table 1 and Fig. 3). In this way, the interpretation of results was made more straightforward. A similar
AUTOANTIGENIC 133A2A
DETERMINANTS
133AXi
ON
HUMAN
100
0
THYROGLOBULIN
134c2
~g,~~~~~~~~~~
0
100
0
100
0
100
0
100
0
relative l7Cl
154C6
blndlng
100
0
mouse rat I-” guinea pig = .-E rabbi 5
0
100
0
100
0
100
(%)
151Cl
152D3
3381
nrrr
human
100
42C4
2182
4lA5
1 1 1
bovine porcine red panda 0
100
0
100
0
Jr!-LL. 100
0
100
0
100
0
100
0
100
0
100
0
100
0
relative binding (%) 1. Bar graphs showing reactivity of 20 mAbs to Tg from seven mammalian species relative to their reactivity to human Tg. For each mAb (listed at the top of each bar graph), each horizontal bar represents the relative reactivity to Tg from the animal species listed at the left. The binding of the mAbs was measured at a dilution of 1:lO and the relative binding to Tg from animal species was calculated as follows: FIG.
OD animal Tg % relative binding = OD human Tg X 100.
approach has been successfully used by a number of investigators for characterizing the interrelationship of antigenic sites recognized by mAbs, and has been verified in a number of different antigen-antibody systems, including Tg-anti-Tg (5, 6, 23-29).
100
70
BRESLER,
BUREK,
AND
ROSE
loo 3 & ul
90
z E
*
0 z!
m
3
6o 50
6 5
40
..tr P E .-c
3o
: '5 m z
a
20 10 0 -10 lb
Reciprocal
dilution
of unlabeled
mAb 156A6
FIG. 2. Results of representative inhibition binding experiments showing that unlabeled mAb 156A6 inhibits its labeled counterpart and labeled 15SA4, but does not inhibit labeled mAbs 133A2A or 154C6. Therefore, mAbs 156A6 and ISA4 share determinant specificities, whereas 133A2A and 154C6 do not share specificities with 156A6.
The inability of some mAbs to bind periodate/cyanoborohydride-treated human Tg provided information on the nature of the epitopes recognized by these mAbs. It is likely that this oxidation/reduction procedure affected carbohydrate moieties on the Tg molecule by cross-linking them to the polypeptide portion of the molecule (20). The mAbs which had diminished or completely lost the ability to bind Tg oxidized and reduced by this method may be reacting via carbohydrate or carbohydrate-dependent epitopes, or via epitopes dependent upon one or more of the cross-linked amino acids. The two mAbs that reacted significantly better to oxidized/reduced human Tg were the only two IgM antibodies in the panel (41C3 and 42C4). There are several likely explanations for these observations, two of which are presented here. First, these IgM mAbs may represent “natural” autoantibodies from the normal antibody repertoire of the immunized mouse. Natural autoantibodies are likely to be low affinity and cross-reactive, qualities both of these mAbs exhibit. Therefore, it is quite likely that natural autoantibodies may react better with a distinctly foreign, denatured antigen than with a native one. This observation has been made by other investigators when looking at the reactivity of mAbs to denatured versus native antigens (30). A second explanation for greater reactivity to a denatured antigen involves the size of the IgM molecule relative to IgG. Other mAbs in this
AUTOANTIGENIC
DETERMINANTS
Unlabeled l,,,,,,,I
ON
HUMAN
71
THYROGLOBULIN
Monoclonal I I
Antibody I I I I I I
;5 0
ev a z
Y m A
137Cl _ 154C6 15lCl . 152D3 1 133611 42C4 12182
; (
I
I
I
I I I I I I I I I I I..... I I I I >($C
FIG. 3. Compiled results of reciprocal competitive inhibition binding assays of 20 mAbs showing proposed determinant clusters of human Tg. Each of 20 mAbs was tested in competition assays against all of the others for binding sites on human Tg (see Materials and Methods). When an unlabeled mAb blocked the binding of a labeled mAb, the appropriate square in this figure was filled in. After all the appropriate squares were tilled in, the columns and rows were rearranged (while being kept internally intact) to bring as many filled squares together as possible. The order in which the mAbs were listed on the top was made the same as the order on the left. The 12 apparent determinant clusters were then designated.
panel (e.g., 42C3) reacted slightly more with oxidized/reduced human Tg than with native. This slight increase in OD could have been exaggerated due to the large mass of pentameric IgM protein bound relative to monomeric IgG bound, and subsequently detected by a conjugated antiserum to murine Igs, even though in molar terms the amount of bound IgG and IgM could have been approximately equal. Information concerning the conservation of epitopes in other mammalian species was obtained by testing the reactivity of the mAbs with Tg from several other mammalian species. Most mAbs (16/20) recognized epitopes that are not widely conserved among animal species; and even though no effort was made to select for mAbs which recognized uniquely human epitopes, 12 of 20 mAbs bound only human Tg. Of the four mAbs which recognized epitopes widely shared among several animal species (including the evolutionarily remote lesser panda), three could be inhibited by T, and T,. These three were the only mAbs which bound to Tg from all species tested. This finding suggests that Tg may vary considerably in
72
BRESLER,
BUREK,
AND
ROSE
its antigenic properties from one species to another, while the thyroxine itself. and the hormonogenic regions from which they arise, are largely conserved. Evidence that the regions neighboring the thyroxines are not identical comes from the reactivity patterns of two mAbs (133Bl and 41A5). The pattern of shared determinant reactivities presented in Fig. 3 indicates that there are topologically continuous antigenic regions on the native human Tg molecule. Determinant clusters 1, 2, and 3 represent a continuum of partially overlapping epitopes. These three overlapping determinant clusters have been designated as “immunodominant” since 8 of the twenty mAbs in the panel define sites in these clusters. All of the mAbs in the immunodominant clusters react with human Tg, and not with Tg from seven other species. Comparison of the reactivities of the mAbs to oxidized/reduced human Tg (Table l), with their arrangement in Fig. 3, reveals further evidence that this topological relationship exists. All of the mAbs in clusters 1 and 2 reacted equally well with oxidized/reduced and native human Tg, except 132C5, which reacted about half as well with oxidizedi reduced as the native protein. Thus, 132C5 represents a transition from cluster 2 to cluster 3 in which all of the other epitopes were destroyed by oxidation/ reduction. Relationships among mAbs within determinant clusters were rein forced by the observation that within given clusters the mAbs shared the same profiles of reactivity to Tg from different species (compare Figs. I and 3). Since each mAb blocked the binding of its labeled counterpart and the order of labeled mAbs was kept the same as the order of unlabeled mAbs, the resulting pattern in Fig. 3 was a solid diagonal of filled squares. The proposed determinant clusters also arrange about this diagonal due to the fact that most inhibitions (50/84) are two-way; that is, if mAb, blocked mAb,, in most cases, mAb, blocked mAb,. However, one-way inhibitions are not uncommon (34184) and could represent special topological relationships between epitopes. A likely explanation for a one-way inhibition in which mAb, blocked mAbz, but mAb, failed to block mAb,, is that the epitope defined by mAb, resides at the top of a cleft in the surface of the molecule and the epitope defined by mAb, resides lower in the same cleft. Another possible explanation of one-way inhibition is that the binding ot mAb, alters the conformation of the epitope recognized by mAbZ. but not vice versa. Some topologically related epitopes are revealed by the high number of oneway inhibitions exhibited by a few mAbs. The best example is 133A2G, which was inhibited by 12 mAbs while it only inhibited 3 mAbs. The mAb 133A2G defines an epitope with a unique position on human Tg. There are at least two explanations for these observations. Epitope 133A2G may reside on or near the junction between the two monomers of the dimeric human Tg, or may be situated in the bottom of a large cleft, or both. Such an epitope would be in a prime position for the mAb which binds it to be blocked by many other mAbs, while itself not being able to block the other mAbs. Another set of one-way inhibitions which may reveal a cleft in the Tg molecule is seen in determinant cluster 12 (Fig. 3). The inhibition pattern of this cluster is consistent with the notion that the epitopes defined by 1370, 154C6, and 15lCl are inside a cleft which is “guarded” at its top by epitopes 152D3 and 133Bl.
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
THYROGLOBULIN
13
When the different reactivities of the mAbs are taken together, 19 different epitopes can be discerned. The only two mAbs that apparently shared the same epitope specificity were 134C2 and 134C4. Each of the other epitopes defined by these mAbs could be distinguished from the others in the panel by one or more of the above-described criteria (Figs. 1 and 3, Table I). This analysis of antigenic sites on human Tg is more extensive than previous ones (5-8, 28, 29, 31-33) in two important ways: (i) this panel of mAbs defines a greater diversity of determinants than any previously published panel of mAbs to human Tg; and (ii) the characteristics of the epitopes defined by this panel are better defined as a result of the combination of studies used. No one of the four criteria used to distinguish the nature or interrelationships of the epitopes would have given the insights that were gained through the use of this combination of criteria. Detailed analysis of this new panel of mAbs by a variety of means has enabled us to distinguish 19 different epitopes on the human Tg molecule. The purpose of the creation and detailed characterization of this panel of mAbs is to be able to use the mAbs in competition assays with human sera to probe the epitope specificities of autoantibodies from patients with thyroid disease. This ail-important first step will enable us to understand better the nature of autoantibodies found in humans. ACKNOWLEDGMENTS We are pleased to acknowledge the advice of Drs. Sharon Krag and Howard Dint& in the performance of these experiments. This work was supported in part by NIH Grants AR31632. AG04362, and AR35383.
REFERENCES 1. Roitt. 1. M., Doniach, D., Campbell, P. N., and Hudson, R. V., V: Autoantibodies in Hashimoto’s disease. Lance? 2, 820-821. 1956. 2. Bigazzi, P. E., and Rose, N. R., Autoimmune thyroid disease. In “The Autoimmune Diseases” (N. R. Rose and I. R. Mackay, Eds.), pp. 161-199, Academic Press, New York, 1985. 3. Aho, K.. Gordin. A., Sievers, K.. and Takala, J.. Thyroid autoimmunity in siblings: A population study. Acta Endocrinol. (Copenhagen) Suppl. 251, 11-15, 1983. 4. Scherbaum, W. A., Stockle, G., Wichmann, J., and Berg, P. A., Immunological and clinical characterization of patients with untreated euthyroid and hypothyroid autoimmune thyroiditis. antibody spectrum, response to TRH and clinical study. Acta Endocrinol. (Copenhagen) 100, 373-381, 1982. 5. Ruf, J., Carayon. P., Sarles-Philip, N., Kourilsky, F., and Lissitzky, S., Specificity of monoclonal antibodies against human thyroglobulin; comparison with autoimmune antibodies. EMBO. J. 2, 1821-1826, 1983. 6. Ruf, J., Carayon. P., and Lissitzky , S., Various expressions of a unique anti-human thyroglobulin antibody repertoire in normal state and autoimmune disease. Eur. J. Immunol. 15, 268-272. 1985. 7. Piechaczyk, M., Bouanani, M., Salhi, S. L., Baldet, L., Bastide, M.. Pau, B., and Bastide, J.-M., Antigenic domains on the human thyroglobulin molecule recognized by autoantibodies in patients’ sera and by natural autoantibodies isolated from the sera of healthy subjects. C&n. Zmmuno/. Immunopathol. 45, 114-121, 1987. 8. Chan, C. T. J., Byfield, P. G. H., Himsworth, R. L., and Shepherd, P., Human autoantibodies to thyroglobulin are directed towards a restricted number of human specific epitopes. Clin. i&p. Immunol. 69, 516-523, 1987. 9. Chemoff, S. B.. and Rawitch, A. B., Thyroglobulin structure-function: Isolation and character-
74
10.
11. 12. 13.
BRESLER,
BUREK.
AND
ROSE
ization of a thyroxine containing polypeptide from bovine thyroglobulin. J. Biol. Chem. 256, 9425-9430, 1981. Kong, Y., David, C. S.. Giraldo, A. A., Elrehewy, M.. and Rose. N. R., Regulation of autoim mune response to mouse thyroglobulin: Influence of H-2D-end genes. J Immunol. 123, 15-18, 1979. Mishell, B. B., and Shiigi, S. M.. Appendix A, In “Selected Methods in Cellular Immunology“ (B. B. Mishell and S. M. Shiigi, Eds.), pp. 443-456, Freeman, San Francisco. 1980. Haeberli, A.. Salvatore, G., Edelhoch, H., and Rail, .I. E., Relationship between iodination and the polypeptide chain composition of thyroglobulin. J. Biol. Chem. 250, 7836-7841. 1975. Rose, N. R.. and Styles, W. A., Splitting of human thyroglobulin. I. Reduction and alkylation. Clin.
Exp.
Immunol.
5, 129-140.
1969.
14. Sorensen. K.. and Brodbeck, U.. A sensitive protein assay method using micro-titer plates. Experientiu 42, 161-162, 1986. 15. Redinbaugh, M. Cl., and Turley. R. B., Adaptation of the bicinchoninic acid protein assay for use with microtiter plates and sucrose gradient fractions. Anal. Biochem. 153, 267-271, 1986. 16. Fazekas de St. Groth, S., and Scheidegger, D.. Production of monoclonal antibodies: Strategy and tactics. .I. Immunol. Methods 35, l-21, 1980. 17. Shulman, M., Wilde , C. D., and Kohler, G., A better cell line for making hybridomas secreting specific antibodies. Nature (London) 276, 269-270, 1978. 18. Oi, V. T., and Herzenberg, L. A., Immunoglobulin-producing hybrid cell lines. In “Selected Methods in Cellular Immunology” (B. B. Mishell and S. M. Shiigi, Eds.), pp. 351-372, Freeman. San Francisco, 1980. 19. Herbert, G. A., Ammonium sulfate fractionation of sera: Mouse, hamster, gumea pig, monkey. chimpanzee, swine. chicken, and cattle. Appl. Microbial. 27, 389-393, 1974. 20. O’Shannessy, D. J., Dobersen, M. J.. and Quarles, R. H., A novel procedure for labeling immunoglobulins by conjugatione to oligosaccharide moieties. Immunol. Lett. 8, 273-277, 1984. 21. Voller, A., Bidwell. D. E., and Burek. C. L., An enzyme-linked immunosorbent assay @LISA) for antibodies to thyroglobulin. Proc. Sot. Exp. Biol. Med. 163, 402-405. 1980. 22. Friguet, B., Chaffotte, A. F., Djavadi-Ohaniance, L., and Goldberg, M. E.. Measurements of the true affinity constant in solution of antigen-antibody complexes by enzyme-linked immunosorbent assay. J. Immunol. Methods 77, 305-319, 1985. 23. Heinz. F. X., Tuma, W., Guirakhoo, F., Berger, R., and Kunz, C.. lmmunogenicity of tick-borne encephalitis virus glycoprotein fragments: Epitope-specific analysis of the antibody response. .1. Gen. Virol. 65, 1921-1929, 1984. 24. Heinz, F. X., Berger, R., Tuma, M., and Kunz, C., A topological and functional model of epitopes on the structural glycoprotein of tick-borne encephalitis virus defined by monoclonal antibodies. Virology 126, 525-537, 1983. 25. Anderson, L. J., Hierholzer, J. C.. Stone, Y. O., Tsou, C., and Fernie. B. F., Identification of epitopes on respiratory syncytial virus proteins by competitive binding immunoassay. J. C/in. Microbial. 23, 4754180, 1986. 26. Rebai, N., Malissen, B., Pierres, M., Accolfa, R. S.. Corte, G., and Mawas, C., Distinct HLA-DR epitopes and distinct families of HLA-DR molecules defined by 15 monoclonal antibodies (mAb) either anti-DR or allo-anti-Iak. I. Cross-inhibition studies of mAb cell surface fixation and differential binding of mAb to detergent-solubilized HLA molecules immobilized to a solid phase by a first mAb. Eur. J. Immunol. 13, 106-111. 1983. 27. Pierres, M., Kourilsky, F. M., Rebouah, J.-P., Dosseto, M., and Caillol, D., Distinct epitopes on Ik gene products identified by monoclonal antibodies. Eur. J. Immunol. 10, 950-957, 1980. 28. DeBaets, M., Janssens, A., Vlek, L., Nakamura, R., Weigle, W. 0.. Vermuelen, A., and van Breda Vriesman, P., Structure of human thyroglobulin studied using monoclonal antibodies. Ann. Endocrinol. (Paris) 43, Al 18, 1985. [Abstract] 29. Kotani, T., Maeda, M.. Ohtaki, S., Kobayashi, S., and Okuyama. H., Analyses of the specificity of mouse monoclonal antibodies to syngeneic and allogeneic thyroglobulins. Clin. Immunol. fmmunopathol. 34, 147-157, 1985.
AUTOANTIGENIC
DETERMINANTS
ON HUMAN
THYROGLOBULIN
75
30. Friguet, B., Djavadi Ohaniance, L., and Goldberg, M. E., Some monoclonal antibodies raised with a native protein bind preferentially to the denatured antigen. Mol. Zmmunol. 21, 673-677, 1984. 31. Piechaczyk, M., Chardes, T., Cot, M.-C., Pau, B., and Bastide, J.-M., Production and characterization of monoclonal antibodies against human thyroglobulin. Hybridoma 4, 361-367, 1985. 32. Kurata, A., Ohta, K., Mine, M., Fukuda, T.. Ikari, N., Kanazawa, H., Matsunaga, M., Izumi, M., and Nagataki, S., Monoclonal antihuman thyroglobulin antibodies. J. Clin. Endocrinol. Metab. 59, 573-579, 1984. 33. Bellet, D., Schlumberger, M., Bidart, J. M., Assicot, M., Caillou, B., Motte, P., Vignal, A., and Bohuon, C., Production and in vitro utilization of monoclonal antibodies to human thyroglobulin. J. Clin. Endocrinol. Metab. 56, 530-533, 1983. Received May 15, 1989; accepted with revision August 8, 1989