Molecular and Cellular Enabcrinologv, 13 (1990) 15-26 Elsevier Scientific Publishers Ireland, Ltd.

15

MOLCEL 02354

Carbohydrate-dependent

epitope mapping of human thyrotropin

Marie-Jeanne Papandreou ‘, Isabelle Sergi ‘, Moufid Benkirane 2 and Catherine Ronin’ ’ Laboratoire de Biochimie, URA II 79 CNRS, Fact& de Mkdecine-Seeteur Nerd 13326 Marseilles Cedex 15, France, and 2 Immunotech, 13288 Marseilles, France (Received 16 March 1990; accepted 28 June 1990)

Key work

Thyrotropin,

human; Epitope mapping; Immunoassay

To probe possible effects of carbohydrate chains in the conformation of pituitary glycoprotein hormones, two radiolabeled derivatives of human thyroid-stimulating hormone (hTSH), either partially deglycosylated in the #!-subunit or fully deglycosylated in both the (Y-and /3-subunits, were compared to the native hormone for binding to monoclonal as well as polyclonal antibodies. Monoclonal antibodies were screened for their ability to bind the intact hormone (anti-hTSH), hTSH and its free a-subunit (anti-a) or its free Bsubunit (anti-p). A panel of 14 monoclonal antibodies directed against at least eight out of the 12 epitopes known to be present in the hormone was tested in solid-phase assays for their capacity to bind intact and deglycos lated forms of hTSH. All of them displayed identical recognition of 7 native and partially deglycosylated 251-hTSH. In contrast, binding of fully deglycosylated 12’I-hTSH to anti-hTSH and anti+3 antibodies was dramatically lost while that of anti-cw was preserved. This clearly indicates that most of the epitopes specific for subunit association as well as those present on the /3-subunit are glycosylation dependent. No alteration was found in antibody recognition following deglycosylation of free individual subunits, indicating that the carbohydrate effect can only occur in the combined dimer. Using polyclonal antisera raised against the International Reference Preparations, we found that the deglycosylated hormone could be bound by the anti-/3 antiserum although at a much lower dilution than the native antigen, suggesting the presence of at least one glycosylation-independent epitope in the P-subunit. Competitive binding assays revealed that deglycosylated hTSH is 5 times less immunoreactive toward the anti-8 compared to the anti-a antiserum. The current data thus demonstrate the presence of the glycosylation-independent epitopes in the a-subunit of hTSH and the localization of most of the glycosylation-dependent domains in the /?-subunit.

Address for correspondence: Catherine Ronin, Laboratoire de Biochimie, URA 1179 CNRS, Faculte de MkkcineSecteur Nord, Bd. P.-Dramard, 13326 MarseilIe Cedex 15, France. Abbreviations: mAb, monoclonal antibody; hTSH, human thyroid-stimulating hormone; p@)DG-hTSH, human TSH par&By degIycosylated in the /?-subunit; DG-h’ISH, human TSH totally degIycosylated in both a- and /3-subunits; IRP, International Reference Preparation; endo F, endogIycosidase F type II; PBS, phosphate buffer saline; BSA, bovine serum albumin; hCG, human chorionic gonadotropin

0303-7207/90/.$03.50

Q 1990 EIsevier Scientific Publishers Ireland, Ltd.

16

Introduction Very recent work from a number of laboratories has demonstrated that monoclonal antibodies can be raised against human pituitary glycoprotein hormones, thyrotropin (TSH), lutropin (LH) and follitropin (FSH) and thereby considerably increase the specificity of recognition of each member of this hormone family (Schwartz et al., 1986; Thotakura and Bahl, 1986; Griffin and Odell, 1987; Spencer et al., 1987). Indeed placental and pituitary gonadotropins are closely related glycoprotein dimers, sharing with TSH the same a-subunit as well as high homologies in the primary sequence of their /3-subunits. Although the overall pairing of the numerous cysteine residues (10 in (Y and 12 in /?) has not been so far totally attributed for any of them, it is widely admitted that all these hormones must exhibit very similar three-dimensional structures (Ryan et al., 1987) and this may account in part for the undesired cross-reactivity of most polyclonal antisera observed so far. Epitope mapping of human chorionicgonadotropin (hCG) using 21 monoclonal antibodies revealed no more than nine distinct antigenic domains for this 38 kDa glycoprotein, a number strikingly small compared to that of 18 found for insulin which is a 6 kDa unglycosylated polypeptide (Schwartz et al., 1986). Three epitopes are localized on the a-subunit, four on the P-subunit and two are expressed in the holoprotein only. As expected, all three a-epitopes were also found within the same topographical relationship to each other on the respective International Reference Preparations for hFSH, hLH and hTSH (Berger et al., 1988). Recently, one of us estimated as 12 the number of epitopes present in human TSH distributed as two on (Y, six on /I and four specific for the dimer (Benkirane et al., 1987a, b, c). As expected, the anti-a-antibodies and several of the others were also found to recognize gonadotropins. While this work was in progress, it was reported that immunization with chemically deglycosylated hCG resulted in the production of antibodies directed to determinants common to the native and deglycosylated antigen but also in some others specific for the protein deprived of sugars, suggesting that glycosylation may modulate pro-

tein antigenicity (Rebois and Liss, 1987; Sairam et al., 1988). This finding correlates well with previous work from this laboratory showing that thyroglobulin secreted from primary cell culture exhibited a reduced immunoreactivity towards polyclonal antibodies compared to the thyroid antigen (Fenouillet et al., 1986). We therefore proposed that the structure of carbohydrate chains or more likely their bulky conformation may modulate the expression of antigenic polypeptide domains at the surface of a glycoprotein hormone. Likewise, human prolactin displayed a 5-fold decrease in immunoreactivity towards a polyclonal antiserum compared to nonglycosylated prolactin while some monoclonal antibodies exhibited the same recognition (Pellegrini et al., 1988). A recent report claimed that monoclonal antibodies can be screened further to specifically measure non-glycosylated prolactin of porcine origin (Scott et al., 1988). Altogether, these observations indicate that there may very well be glycosylation-dependent and independent epitopes at the surface of glycoproteins. In a report to be published elsewhere r, we describe the enzymatic deglycosylation of hTSH and the isolation of hTSH form deglycosylated in the /3-subunit as well as that of a form fully deglycosylated on both (Y-and P-subunit. We show here that compared to the native glycosylated hormone, both deglycosylated derivatives are differentially recognized by various monoclonal antibodies. Since most of the clinical measurements of circulating hTSH are standardized with the International Reference Preparations, we also examined the behavior of each form of hTSH toward polyclonal antisera directed against such preparations. Materials and methods Materials Hormones (hTSH) and subunits (hTSH-a and hTSH-P) were purchased from UCB Bioproducts (Brussels, Belgium). The International Reference Preparations (NIADDK-hTSH-RP-1, AFP-3290-

’ Papandreou, M.J., Sergi, I., Kopeyan, C., Canonne, C. and Ronin, C., submitted for publication.

17

were B-hTSH-ar, and AFP-2808-C-hTSH-~~) kindly provided by the National Pituitary Agency (NIH, Bethesda, MD, U.S.A.). Polyclonal antibodies raised in rabbit against hTSH (NIADDKanti-hTSH-3), hTSH-a (anti-hTSH-cr,) and hTSHj3 anti-hTSH-~,) were also a gift of the National Pituitary Program. Monoclonal antibodies were kindly provided by Immunotech (Marseilles, France). Endo F type II was from Boehringer (Marmheim, France). All the chemicals were of analytical grade. Munoclunal anti-hTSH antibodies Monoclonal antibodies were raised in mouse using hCG or hTSH, with or without hTSH-8, as antigen and in rat, using mAb-hTSH complex as antigen. Their characterization and specificity have been described elsewhere (Benkirane et al., 1987a, b, c). Briefly, they were screened and classified as anti-o when they were able to bind hTSH and free LY,anti+ when they bound hTSH and free /I and anti-hTSH when they recognized the dimer only. They were purified on protein A-Sepharose and biotinylated as described in Benkirane et al. (1987b). Polyclonal anti-hTSH antibodies Anti-IRP antisera were raised in rabbits using hTSH as antigen for anti-hTSH antibodies, hTSH/3 subunit for anti-p antibodies or hTSH-a subunit for anti-a: antibodies. In our hands, half-maximal binding of “‘1-hTSH was obtained for each antiserum at the following final dilutions: 1:2,500,000 for anti-hTSH (IRP), 1:60,000 for ~ti-hTSH-~ (IRP) and 1:2~,~ for anti-hTSH-a! (IRP). Hormone deglycosylation, iodination and isolation of the tracers and subunits hTSH (approx. 50 &i/pg) (hTSH-ar = 60 &i/r.lg, hTSH-fi = 40 $i/pg} were iodinated with Iodogen as described elsewhere (Ronin et al., 1987). The deglycosylation procedure and isolation of partially and fully deglycosylated labeled hTSH were as follows: aliquots of hTSH (5 pg) were incubated with 300 mU of endo F in a 50 mM PBS buffer pH 7.5 containing 0.05% Triton X-100 for 3 days at 30 o C under a toluene atmosphere to avoid bacterial

~nta~ation and then the whole mixture was submitted to iodination before purification by gel filtration on a Bio-Gel AcA column and concanavalin A-Sepharose chromatography ‘. Tryptic and 12SI-DGmapping of ’ ‘I-p(P)DG-hTSH hTSH revealed that these tracers display a very similar specific activity close to that of the native hormone, i.e. 50 pCi/pg ‘. Free subunits (1 pg), radiolabeled (50 x lo3 cpm) or not, were deglycosylated with 50 mU endo F in a final volume of 50 ~1 containing 0.5% Triton in 50 mM PBS pH 7.5 according to Ronin et al. (1987). In contrast to the intact hormone, free subunits can be easily deglycosylated and when radioiodinated, were treated with endo F under the above conditions and further used as such for immunoassays. To prepare enough de~ycosylat~ hTSH for displacement studies, several aliquots (2 ,ug) of hTSH were incubated with 120 mU of endo F in the presence of 0.5% Triton for 3 days at 35°C as described above. The enzymatic digests were then pooled and c~omatograph~ on concanavalin ASepharose column (1 X 0.5 cm) equilibrated in 50 mM PBS buffer pH 7.5 containing 0.5% BSA and precycled with 0.5 M cr-methyl-marmoside in 50 mM PBS. The samples were run in 50 mM PBS containing 0.1% BSA and the glycosyiated products eluted with 0.3 M ~-methyl-m~oside (Fenouillet et al., 1986). The yield of the column was approximately 15% for unlabeled DG-hTSH as already found for the corresponding radiolabeled material ‘. Unbound and bound fractions were assayed for their capacity to inhibit binding of 1251-hTSH to anti-o ~dilution 1:2~,~), antihTSH-p (1:60,000) and anti-hTSH (1:2,500,000) IRP antiserum. Solid-phase binding immunoassays Biotinylated mAbs were adsorbed onto tubes precoated with avidin (I~unot~h, patent No. 84430010.3). The various iodinated ligands (50 x lo3 cpm) were incubated with coated monoclonal antibodies, in a final volume of 250 ~1 50 mM PBS pH 7.5 containing 0.1% BSA for 24 h at 4O C. The content of the tubes was then eliminated, the tubes washed with 0.9% NaCl and the bound radioactivity counted in a multidetector Gamma Packard Counter. All assays were done in dupli-

18 N

cates and the nonspecific binding was about 0.2% that of the total radioactivity in the assay. The results are given as the ratio of specific binding (B) to the total radioactivity (T) multip~ed by 100. In competitive experiments, the final concentration of the antibody was chosen to be 3050% of the maximal binding capacity and displacement of the tracer obtained by addition of increasing amounts of the appropriate subunit. Liquid-phase

0h

binding immunoassays

Iodinated ligands (50 X lo3 cpm) were incubated with polyclonal antisera with or without unlabeled hormone components in a final volume of 300 ~1 of 50 mM PBS buffer pH 7.5 cont~~ng 0.1% BSA overnight at 4’ C. Nonimmune rabbit serum (100 ~1, dilution 1 : 100) and horse anti-rabbit immunoglobulins (100 ~1, dilution 1: 4) were then added for 2 h at 4” C. Immune complexes were finally precipitated with 6% polyethylene glyco1 6000 in a final volume of 1 ml and recovered by centrifugation at 10,000 X g for 15 min. All assays were performed in duplicates and the nonspecific binding was about 0.5% of the total radioactivity in the assay. In displacement studies, the final concentration of the antibody was taken as to give 30% of the maximal binding capacity.

TABLE

1

SPECIFICITY OF THE MONOCLONAL USED IN SOLID-PHASE ASSAYS Immunization

Results Screening of the monoclonaf antibodies and deg~y&osy~a~ed h TSH

s Fig. 1. Epitope map of human TSH. Each antigenic region is arbitrarily represented as a circle of identical surface as presented elsewhere (Benkirane, 1987a). North represents the OLsubunit and South the &subunit while E-W reflects the o-b interface.

with native

As many as 12 epitopes could be identified in hTSH using a panel of 28 different mAbs as shown in the epitope map of hTSH presented in Fig. 1. Most of them are spatially related to the &3 interface and only four appeared to be present in the holo-ho~one (Benkirane et al., 1987a, b, c). Some clusters are defined by a single antibody while the so-called pi determinant is recognized by as many as eight mAbs and thus is likely to constitute the major antigenic determinant of hTSH. In the current study, we have used the 14 mAbs directed against eight distinct epitopes of hTSH presented in Table 1 and analyzed their capacity to bind i2?-hTSH, 1251-p(/3)DG-hTSH and 125I-DC-hTSH in an ultrasensitive solid-phase binding assay. Each mAb was systematically coated at the concentration for which it maximally

ANTIBODIES

Specificity

Animal

Antigen

Recognition

Mouse

hCG hTSH

a+hTSH n+hTSH

a, LYE

hTSH

hTSH hTSH hTSH

4%

530

42

490

4

210

p+hTSH

&

B+hTSH B+hTSH B+hTSH B+hTSH

Br 8, 81 86

27 17 1120 182 x2

j3 + hTSH

p (other than & and &)

hTSH j + hTSH Rat

mAb 27-hTSH complex

a Epitopes b

mAb number 6E4 326

68 3:: 779

a The screening procedure was as described in Benkirane et al. (1987c) and briefly summarized under Materials and Methods. b Epitopes were designed as in Fig. 1 and identified from two-site binding assays with radiolabeled hTSH and/or free subunits as described in Benkirane et al. (1987a).

19

v

326

530

BE4

490

218

mAb anti-hTSH

mAb anti-u

50

40 30 20 10 n ”

27

17

,120

182

x

75

779

BB

358

mAb anti- ’ I,

Fig. 2. Comparative binding of monoclonal antibodies to various deglycosylated forms of human TSH. Monoclonal antibodies were used in solid-phase assays at concentrations corresponding to maximal binding of the native tracer. Top panels: Binding of **‘I-hTSH (solid bars), ‘*%p(P)DG-hTSH (hatched (dashed bars) to anti-a (left) and to bars) and ‘*?-DG-hTSH anti-hTSH (right) mAbs. Bottom panel: Binding of the three tracers to anti-8 mAbs.

bound 12?-hTSH as well as half of the tracer. The data were comparable and therefore only the former are presented as histograms in Fig. 2. Comparing 1251-p(@DG-hTSH (Fig. 2, hatched bars) to 12?-hTSH (Fig. 2, solid bars), we found no major alteration in binding among the series of mAbs. Especially for the three anti-hTSH mAbs but also for anti-p mAb 27, we even noticed a reproducible lo-20% increase in binding of 12?p( @DG-hTSH compared to the intact tracer. This finding might be explained by a loss of some unknown interaction between the /3-linked CHO chain and the peptide backbone that could unmask the corresponding epitopes following deglycosylation in a significant portion of the tracer molecules. In no instance did we find a major modification in antibody recognition when we compared the native 1251-hTSH hormone to 1251p(/3)DG-hTSH. Nor did we observe any difference when native 1251-hTSH was treated with

endo F to generate 1251-p(B)DG-hTSH (data not shown). In contrast, the recognition of the totally deglycosylated hormone (Fig. 2, dashed bars) was significantly altered when compared to the native tracer. No major change occurred with mAb 326 (anti-o,) while the mAb 6E4 (ant&) as well as mAb 210 (anti-@3) displayed an inhibition of binding close to 50%. This implies that among the antigenic regions present on cu and those specific of the dimer, at least one and, to a certain extent, two others are not directly sensitive to removal of a-linked CHO chains. This also demonstrates that 1251-DG-hTSH was still present as an apoprotein closely resembling the native glycosylated antigen with a similar if not identical interaction between (Y-and P-subunit. Interestin y, a dramatic loss of !Y antibody binding toward l2 I-DG-hTSH was observed concerning two of the determinants specific for intact hTSH and, more importantly, all of those specific for the B-subunit. Indeed, while mAbs anti-p or anti-hTSH were poorly affected by the deglycosylation of the /.&subunit itself as in 1251-p(p)DG-hTSH, their recognition was singularly abolished when the a-subunit was further deglycosylated as it is the case in ‘251-DG-hTSH. Binding of native and deglycosylated hTSH to monoclonal antibodies We further analyzed the recognition of the various native and deglycosylated forms of 1251-hTSH as a function of antibody concentration. Fig. 3 shows the binding of ‘251-hTSH, 1251-p(/3)DGhTSH and 1251-DG-hTSH to those of the mAbs that exhibit the highest binding for the native tracer (25-658). Again, no significant difference in binding could be observed between native hTSH and its counterpart missing the CHO chain on 8. As noticed above, the major alteration was found to reside in the recognition of the hormone totally deglycosylated on its a-subunit. In this regard, the two a-specific epitopes behave differently: while mAb 326 recognized the three tracers equally well, mAb 6E4 gradually lost virtually half of its binding as a function of CHO removal (Fig. 3, left panels). This may indicate that the ai determinant is more sensitive to ar-subunit deglycosylation than the a2 epitope. It was also observed that the recognition of the dimer was severely impaired

20

(Fig. 3, middle panels): two out of three antibodies anti-hTSH (mAbs 530 and 490) were further unable to bind 12?-DG-hTSH while the avidity of the third (mAb 210) appeared decreased by a factor of 3. Therefore, some conformational changes probably should occur following cleavage of the cu-linked CHO chains that involve two out of the three epitopes specific for subunit association, namely Q1 and (Y&. Again, the binding of ‘251-DG-hTSH to antibodies (mAbs 27 and X2) directed toward the P-subunit was totally abolished over the range of concentrations tested (Fig. 3, right panels), indicating a dramatic loss of two of the epitopes highly specific for the hTSH hormone. This also indicates that the major antigenic domain of this hormone is not dependent upon the presence of its own CHO chain but rather upon those of the adjacent a-subunit.

Recognition of native and deglycosylated free subunits by monoclonal antibodies In order to determine whether or not alteration in antibody recognition observed with fully deglycosylated hTSH was dependent on subunit association, we compared the binding of monoclonal antibodies to the native and deglycosylated free subunits. Using solid-phase assays and monoclonal antibodies (Fig. 4) we did not observe any major modification of either LY-or P-subunit following CHO removal (Fig. 4, top panels). Rather, we even reproducibly noticed a slight increase in the binding of “‘I-DG-hTSH-/3 to anti-p antibodies. Therefore, the main antigemc region present in p is under the influence of CHO chains in the combined dimer but not in the dissociated subunit, indicating that this effect is clearly related to subunit association. Since the monoclonal

60 50 40 30 30

mAb 6E4

50

20

mAb 27

. 0

1 40

10

o

0.2

0.4

0.6

0.6

1

1.2

1.4

50

20

40

10

30

300 z:l_, 0

0.2

0.4

0.6

0.6

1

1.2

1.4

0

0.2

0.4 mAb

0.6 0.6 1 anti-p(pg/ml)

1.2

1.4

20 10 70

04...;.,5,.:.:.

60

o

0.2

0.4

0.6

0.6

1

1.2

1.4

50 50

40 30

40

20 30 10 20

0 0

0.2

0.4 0.6 0.6 mAb anli-d&ml)

1

1.2

1.4 10 0 0

0.2

0.4 0.6 0.6 1 mAb anti-hTSH @g/ml)

1.2

1.4

Fig. 3. Binding of native and deglycosylated hTSH to monoclonal antibodies as a function of antibody concentration. Increasing amounts of anti-a (left panels), anti-hTSH (middle panels) and anti-P (right panels) mAbs were coated to plastic and assayed for their binding of 12’I-hTSH (o), 1251-p(P)DG-hTSH (m) and ‘251-DG-hTSH (A).

21

did not find any change in the affinity of mAb 326 to recognize its epitope whether the a-subunit was native or deprived of CHO (Fig. 4, right). Using mAb 27, the displacement curves of the tracer with native and deglycosylated /I-subunit were also found superimposable (Fig. 4, left), suggesting that the epitopes present in the B-subunit become insensitive to glycosylated following hormone dissociation.

antibodies studied here were selected according to their capacity to bind equally well the hormone and the free subunit, it also suggests that the conformational change induced by sugar removal can only occur when the subunits are combined and cannot be reproduced after dissociation. As full deglycosylation was easily achieved with free subunits in contrast to intact hTSH, we performed competitive binding experiments using the monoclonal antibodies and each native or deglycosylated free subunits as tracers and/or competitors (Fig. 4, bottom panels). No difference ‘could be observed between the native tracer and the deglycosylated tracer, for the cw-subunit as well as for the P-subunit (data not shown). Similarly, we

Binding of native and deglycosylated hTSH to polyclonal antibodies Since the monoclonal antibodies used in the

present study have been obtained very recently, there has not been yet any report on their use in

mAb antI+

mAb anti-Q 60 -I

6E4

326

27

17

1120

182

T

mAb 27

mAb 326 u 9

z :: z E

100-r

loo-

n .

80-

80-

60 -

60-

40 -

40 -

20 -

20 -

0

1

0 0

0.1

100

1000

0

1

.

I 100

.

-I 1000

Fig. 4. Binding of individual free subunits to monoclonal antibodies. Top panels: Anti-a (left) and anti-/3 (right) monoclonsl antibodies were tested for their binding of native and deglycosylated radioiodmated subunits. Bottom panels: Unlabeled native (0) and deglycosylated (W) a-subunit at varying concentrations were competed with ‘251-hTSHs for binding to mAb 326 (left). Displacement curves of 1251-hTSH-/3 with native (0) or deglycosylated (0) B-subunit were performed using mAb 27 (right).

22

clinical trials. It was therefore of interest to compare their behavior toward the various forms of hTSH to that of the conventionally used polyclonal antisera raised against the International Reference Preparations. In each case, the tracers were ‘251-hTSH, “‘1-p( /3)DG-hTSH and 12’1DG-hTSH (Fig. 5). When their binding was tested as a function of antibody concentration, again no difference between 12?-hTSH and 1251-p( P)DGhTSH was observed among the various antibodies (Fig. 5A-C). In contrast to monoclonal antibodies, all the antisera were found to bind fully degly-

Anti

cosylated hTSH. The anti+ antiserum recognized 12?-DG-hTSH (Fig. 5A) as did monoclonal antibodies, thereby further confirming that the underlying epitopes were CHO independent. Of particular interest, anti-hTSH antibodies were found to significantly bind the fully deglycosylated form of hTSH in the same dilution range as that required for the native antigen (Fig. 5B) whereas antisera of similar specificities from other sources totally failed (data not shown). Furthermore, the polyclonal antiserum directed against the free fisubunit appeared also capable to bind ‘251-DG-

hTSH u antiserum

100

A P

80

F

100

.

ma

60

2 2

40

40

20

20

I?

60

04

I

0

80

P a m E 8 k a

60 40 20 0

G

80

20

1

r u 40 60 1 I Ab dilulion x 10-6

E

80

"m ”

0.01

.

*

- hTSH

(ng)

IO

100

40 60 1 I/b dilution x lo-'

80

.

0.01

0.1

1 Subunil

(ng)

lb

100

antiserum

100

100

80

80

60

60

40

40 20

o-l 20

1 Subunit

20

.a

0

0.1

Anti I.5

m

0.001

..-7

*

0.01

0.1

Hormone I

...-1

4

1 Subunit

10

100

( ng)

0 0.001

0.01

1 0.1 Subunit (r-g)

IO

100

Anti - hTSH 0 antiserum 100

100

F

80

80

60

60

40

40

20

20 m

”

0

0.05

0.5

1

1 I Ab dilution x 10-a

2

I

0.01

0.1

1 Subunit

Ing)

IO

100

0m 0.01

I

0. 00 i

0 I .".

0.1

1 Subunit

(ngj

IO

100

Fig. 5. Binding of native and deglycosylated forms of hTSH to rabbit antisera. Left panels: Anti-IRP antisera directed against hTSH-a (A), hTSH (B) or hTSH-j? (C) were analyzed in liquid-phase assays for their ability to bind 12’I-hTSH (o), ‘2SI-p(p)DGMiddle and right panels: Anti-hTSH-a (D and G), hTSH (m) and 1251-DG-hTSH (A) as a function of antibody concentration. anti-hTSH (E and H) and anti-hTSH-P (F and I) were used at a final dilution of 1 : 200,000, 1 : 2,500,OOO and 1: 60,000 respectively and 1: 60,000, 1: 300,000 and 1: 1000 respectively when binding ‘251-DG-hTSH (G-Z) in liquid-phase for 1251-hTSH binding (D-F) immunoassays. Displacement by unlabeled hormone material in middle and right panels was performed with native (0) or deglycosylated (m) hTSH-a, hTSH (+), native (0) and deglycosylated (0) hTSH-P as appropriate.

23

hTSH although at a lo4 lower dilution compared to the native hormone (Fig. SC). This finding indicates that at least one epitope present on the ~-subunit is not under the influence of CHO chains and that it could be revealed by an heterogeneous population of antibodies such as polyclonals. To elucidate further the identity of the different epitopes contributing to the recognition of deglycosylated hTSH by the various antisera, we analyzed the competition of ‘*‘I-hTSH (Fig. 5DF) and ‘*‘I-DG-hTSH (Fig. 5G-I) with both individual free subunits, either in their native or deglycosylated forms. It can be seen that for both tracers the same displacement curves were obtained with native or deglycosylated individual OLor /&subunits, indicating no major alteration of the corresponding epitopes upon CHO removal in free subunits. Again no major difference could be detected with anti-a antibodies (Fig. 5D and G). With respect to the anti-hTSH antiserum, it is worth noting that while the glycosylated ‘*‘I-hTSH tracer is equally well displaced by the unlabeled RP-1 preparation, free IRP a- or fi-subunit (Fig. 5E), the cy-subunit appeared 17 times more potent than the P-subunit. in displacing ‘251-DG-hTSH (Fig. 5H). This indicates that within this antiserum the native and deglycosylated hormone are recognized by antibodies of different specificity, more likely to be directed toward the /&subunit for the former but toward the cu for the latter. Conversely, despite a rather poor binding of 1251DG-hTSH by the anti-p antiserum, it was clearly based on a /3 specificity (Fig. 51). Displacement curves for the two tracers were not parallel with a much higher half-dis lacement concentration for P ‘*?-hTSH than for ‘* I-DG-hTSH suggesting that the recognition of both forms of hTSH might occur through different anti-/3 antibody subpopulations (Fig. 5F and 1). Accordingly, the antibodies recognizing the glycosylation-independent pepitope appeared to be of low affinity compared to those working in the routine assay. To further ascertain unambiguously that dramatic alteration occurred in hTSH as a result of carbohydrate removal, we examined the ability of unlabeled DG-hTSH to inhibit the binding of “*I-hTSH to the various anti-IRP antisera. As it can be seen in Fig. 6, antibody recognition of the

a0 5 cl" 60 u

E 8 k E

40 20

Sample Dilution

100

(

2

a0

s m

60

f 8 ki a

40 20 0

-l---v-..,

0.001 0.01

. -...-.a

0.1

.

I

10

7 .--i

100

hTSH (n;) Fig. 6. Competitive binding assays of endo F-treated hTSH. Unlabeled h’ISH was incubated with endo F under optimized conditions and further chromatographed on ConcanavaIin ASepharose. Unbound (A) and bound (B) fractions were tested for their capacity to inhibit binding of 12?-hTSH to anti-a (A), anti-b @) and anti-hTSH (0) IRP antisera. Panel C shows the corresponding standardization curves using the original preparation of hTSH.

concanavalin A fractions was significantly dependent upon the specificity of the antiserum. The concanavalin A-unbound hTSH-like material appeared approximately 5 times less potent in displacing the glycosylated tracer from anti-/3 antibodies than from anti-a! or hTSH antisera (Fig. 6A). In contrast, the concanavahn A-bound material (Fig. 6B) largely behaved like intact hTSH

24

(Fig. 6C) toward anti-a, anti-/3 as well as antihTSH antibodies. This further confirms that loss of hTSH immunoreactivity is caused by deglycosylation independently of the radioi~ination process. Altogether, our data suggest that carbohydrate removal indeed is able to induce a profound alteration of the conformation of hTSH, especially of its hormone-specific antigenic domains. Discussion Although the immunochemistry of glycoprotein hormones has been extensively investigated over the past two decades, there still appears a large discrepancy conce~ng the body of expe~mental work available for each of them. Most studies have been directed to the understanding of hCG whereas TSH, especially of human origin, received considerably less attention. In all instances, the contribution of glycosylation to hormone immunoreacti~ty was not addressed and, to our knowledge, very little information is available in this field. This was the purpose of the current study to investigate whether or not in human TSH the presence of CHO chains may influence antibody recognition. We show that indeed in the hormone dimer most of epitopes present in the P-subunit are glycosylation dependent and that this clearly occurs through the participation of the a-subunit as if the whole dimer contributes to the integrity of these antigenic domains. Most of the monoclonal antibodies used in the present study are likely to be directed against conformational determinants since they recognize very poorly a denatured form of the hormone. Especially those designated as anti-hTSH appear directed toward determinants present only in the combined subunits, either involving a few amino acids of both subunits spatially in contact or specifically located on one subunit but dependent of the interaction with the other. Among the three distinct regions of this type that we have investigated in this study, we observed that mAb anti-c& was less sensitive to CHO removal than the two others, mAb cy& and mAb api. This observation was reproduced with two antisera raised against the RP-1 hTSH preparation since they appeared to recognize unlabeled DG-hTSH through its flsubunit. These findings indicate that extensive de-

glycosylation of the hormone altered part of the three-dimensional conformation of the hormone probably through long-range interactions and in this process, the a-linked CHO chains appear more critical than that of the P-subunit. However, this conformational change is probably limited to portions of the protein surface since it does not involve the epitopes present in the a-subunit and thus may very well differ from that observed with chemically deglycosylated gonadotropins using circular dichroism (Ryan et al., 1988). It nevertheless demonstrates that the influence of CHO on antigenic domains occurs through subunit association and suggests that it may be specific for the hormone. This finding is in total agreement with a previous proposal based on the immunochemical analysis of gonadotropin recombinants suggesting that the conformation of the common a-subunit may be modulated by its interaction with the P-subunit (Strickland and Puett, X982). Since we found that the removal of the a-linked CHO chains did not alter the recognition of the underlying epitopes with anti-a antibodies, CHO may not be as important in the mature form of this subunit as they are during biosynthesis to maintain a functional folding and assembly of the hormone dimer. Unglycosylated TSH-f~ subunit synthesized by mouse thyrotropic tumors in the presence of tunicamycin failed to combine with TSH-P (Weintraub et al., 1983) although it appears to possess all the information to fold to a proper structure while the same has not been found for the P-subunit (Strickland and Pierce, 1985). None of the P-subunits studied have been observed to reform a totally native structure following reduction and denaturation. Our current findings may very well have clinical relevance in assaying TSH in human blood samples as all the ultrasensitive immunoassays are currently based on the use of P-specific monoclonal antibodies. In this regard, it is of importance to note that among the glycosylation-dependent epitopes present in B-subunit is included the major antigenic determinant defined as the /3, cluster in the epitope map of hTSH (see Fig. 1 and Table 1). This may explain in part why similar findings could be obtained with monoclonal and polyclonal antibodies as the immune response in rabbits as well as in mice is likely to be prefer-

25

entially directed toward this region. It is worth noting also that this epitope together with the so-called & are the two domains selected for the two-site immunoassays used in routine clinical investigations. Apparently, both corresponding polypeptide areas are under the influence of CHO structure and it would be of immediate interest to investigate whether or not they may be susceptible to changes in CHO structures in the intact hormone. At least in the rat, intrapituitary TSH forms are distinct from those secreted in vitro with respect to their glycosylation (Gesundheit et al., 1986). Furthermore, these structures are under the control of at least both the thyroid status (Taylor and Weintraub, 1985) and the hypothalamic secretion of TRH (Taylor et al., 1986; Gesundheit et al., 1987). Whether such changes in glycosylation affect in turn antigenic domains in native TSH, is still unknown but may very well occur. Preliminary data from our laboratory suggest that several isoforms of intrapituitary hTSH escape to the detection of anti-p antibodies (Sergi et al., manuscript in preparation). The physiologic regulators of hTSH have been shown to alter the biological activity of hormone. Indeed antagonistic as well as antagonistic forms of TSH could be isolated from thyrotropic mouse tumors and both were found immunoreactive using an anti-bovine TSH and a rat TSH tracer (Joshi et al., 1983). However, in the experiment the antibody recognition was far too heterologous to infer that both isoforms displayed similar immunological potency. If glycosylation modulates the expression of p-specific polypeptide domains other than the antigenic determinants evidenced here in human TSH, it is conceivable that changes in CHO structures may result in lack of receptor binding. It was found that in central hypothyroidism, circulating hTSH exhibited a decreased binding to human thyroid membranes compared to normal subjects (Beck-Peckoz et al., 1985). In this prospective, it is of interest that the most potent inhibitors of hTSH stimulatory activity towards human cells in culture were found to be mAbs directed against either the a2, c& 8, and & epitopes (Costagliola et al., 1988). Using synthetic peptides covering the entire sequence of the human a-subunit, it was recently shown that two separate sites (a 21-6, a 81-92) within the a-sub-

unit represent binding sites for the hormone in an heterologous receptor assay based on the inhibition of bovine TSH-mediated cyclic AMP production in rat FRTL-5 cells (Morris et al., 1988). Further work is required to identify the functional as well as the antigenic domains of hTSH and to ascertain whether or not both types of domains are under a similar control of hormone carbohydrate chains. Acknowledgments

This work has been supported by a grant of the Association pour la Recherche contre le Cancer (A.R.C.). The authors thank Dr. D. Bon-Poizat for help in the purification and derivatization of monoclonal antibodies and Prof. P. Jaquet for helpful discussions. The typing of C. Roussarie is gratefully acknowledged. References Beck-Peccoz, P., Amr, S., Menezes-Ferreira, N.M. and Weintraub, B.D. (1985) New Engl. J. Med. 312,1085-1090. Benkirane, M.M., Bon, D., Costaghola, S., Paolucci, F., Darbouret, B., Prince, P. and &rayon, P. (1987a) Endocrinology 121,1171-1177. Benkirane, M., Bon, D., Cordeil, M., Delori, P. and Delaage, M.A. (1987b) Mol. Immunol. 24, 1309-1315. Benkirane, M., Bon, D., Bellot, F., Prince, P., Hassoun, J., Carayon, P. and Delori, P. (1987~) J. Immunol. Methods 98,173-101. Berger, P., Panmoung, W., Khaschabi, D., Mayregger, B. and Wick, G. (1988) Endocrinology 123, 2351-2359. Costaghola, S., Madec, A.M., Benkirane, M.M., Orgiazzi, J. and Carrayon, P. (1988) Mol. Endocrinol. 2, 613-618. Fenouillet, E., Fayet, G., Hovsepian, S., Bahraoui, E.M. and Ronin, C. (1986) J. Biol. Chem. 261, 15153-15158. Gesundheit, N., Magner, J.A., Chen, J. and Weintraub, B.D. (1986) Endocrinology 119,455-463. Gesundheit, N., Fink, D.L., Silverman, D.A. and Weintraub, B.D. (1987) J. Biol. Chem. 262, 5197-5203. Griffin, J. and Odell, W.D. (1987) J. Immunol. Methods 103, 275-283. Joshi, L. and Weintraub, B.W. (1983) Endocrinology 113, 2145-2154. Morris, III, J.C., Jiang, N.-S., Charlesworth, M.C., McCornick, D.J. and Ryan, R.J. (1988) Endocrinology 123, 456-462. Pellegrini, I., Gunz, G., Ronin, C., Fenouillet, E., Peyrat, I.P., Delori, P. and Jaquet, P. (1988) Endocrinology 122, 26672674. Rebois, R.V. and Liss, M.T. (1987) J. Biol. Chem. 262, 38913896.

26 Ronin, C., Papandreou, M.J., Canonne, C. and Weintraub, B.D. (1987) Biochemistry 26, 5845-5853. Ryan, R.J., Keutman, H.T., Charlesworth, M.C., McCormick, D.J., Mill&, R.P., Calvo, F.O. and Vutgavanich, T. (1987) Recent Prog. Hot-m. Res., 43, 383-431. Ryan, R.J., Charlesworth, M.C., McCormick, D.J., Millins, R.P. and Keutmann, H. (1988) FASEB J. 2, 2661-2669. Sairam, M.R., Linggen, J. and Bhargavi, G.N. (1988) Biosci. Rep. 8, 271-278. Schwarz, S., Berger, P. and Wick, G. (1986) Endocrinology 118, 189-197. Scott, M.G., Lin, W.H., Lyle, L.R., Atkinson, P.R., Seely, J.E. and Markoff, E. (1988) Biochem. Biophys. Res. Commun. 151, 1427-1443.

Spencer, C., Eigen, A., Shen, D., Duda, M., Qualls, S., Weiss, S. and Nicoloff, J. (1987) Clin. Chem. 33, 1391-1396. Strickland, T.W. and Pierce, J.G. (1985) J. Biol. Chem. 260, 5816-5819. Strickland, T.W. and Puett, D. (1982) Endocrinology 111, 95-100. Taylor, T and Weintraub, B.D. (1985) Endocrinology 116, 1535-1542. Taylor, T., Gesundheit, N. and Weintraub, B.D. (1986) Mol. Cell. Endocrinol. 46, 253-261. Thotakura, N.R. and Bahl, 0. (1986) Endocrinology 119, 1887-1894. Weintraub, B.D., Stannard, B.S. and Meyers, L. (1983) Endocrinology 112,1331-1339.