Eur. J. Biochem. 207,915-922 (1992)

0FEBS 1992

Precise epitope mapping of monoclonal antibodies to the cytoplasmic side of the acetylcholine receptor a subunit Dissecting a potentially myasthenogenic epitope Socrates J. TZARTOS and Michael S. REMOUNDOS Department of Biochemistry, Hellenic Pasteur Institute, Athens, Greece (Received February 24/April24, 1992) - EJB 920250

The epitopes for twelve monoclonal antibodies against the cytoplasmic side of the acetylcholine receptor (AChR) a subunit were precisely mapped using over 300 continuously overlapping synthetic peptides attached on poly(ethy1ene) rods. mAb cross-reactive between Torpedo and human AChR generally bound to the homologous peptides from both species. Epitopes 4 - 10-residues long were identified. One mAb could bind to either arm on both sides of a j?-turn structure. Five mAb bound to a very-immunogenic cytoplasmic epitope on a373 - 380 (VICE-a). Three of the mAb against VICEM were earlier found to cross-react with non-AChR protein(s), present in thymomas from myasthenia gravis patients but absent in thymomas from non-myasthenics. Since VICE-a has a potentially crucial pathogenic role, the antigenic role of each residue within it was subsequently studied by 55 analogues, most having single amino acid substitutions. All the mAb against VICE-a bound similarly but not identically to the analogues, thus explaining their known binding heterogeneity. Lys373 proved indispensable for mAb binding. Ile376, Glu377, Gly378 and Lys380 were quite critical, while Ser374, Ala375 and Val379 seemed rather inactive. These data should prove instructive in searches for VICEa-like epitopes carrying autoantigens with potential involvement in myasthenia gravis and should further expand the applications of the anti-(AChR) mAb in AChR studies.

nates [lo, 111. The majority of these mAb are directed to the main immunogenic region (MIR), localized within residues 67-76 of the a subunit [9, 121. On the contrary, when the rats are immunized with SDS-denatured AChR subunits, the majority of the produced mAb are directed against the cytoplasmic side of the four subunits [13-151. Binding of such mAb to the Torpedo and human a subunit has been localized within a 60-residue segment (residues 330- 390) of the 437 residues of the a subunit [12, 15-17]. Non-MIR anti-(a subunit) mAb have proved valuable in the study of both the AChR and myasthenia gravis. In particular, they have been used to study the transmembrane orientation of several AChR segments [14, 151 and other aspects of AChR conformation [18- 201, to map the antigenic structure of the AChR [12, 15, 161, to study the B/T cell interactions in experimental myasthenia gravis [21, 221 and in several other studies in which detection and characterization of the expressed AChR or its a subunit was necessary (reviewed in [8, 91). Furthermore, three non-MIR mAb were found to crossreact with non-AChR protein(s) present in the great majority of thymomas from myasthenic patients, but not in thymomas from non-myasthenic patients [23]. One such protein, p153, has been isolated and characterized [24]. Based on several observations, it was proposed that thymoma-associated myasCorrespondence to S. J. Tzartos, Hellenic Pasteur Institute, 127, thenia gravis may be initiated by an immune reaction against Vas. Sofias Ave., Athens 11521, Greece p153 [24]. This protein could be the primary autoimmunogen Fax: +30 16423498. for triggering AChR-specific T cells [5, 241. Further investiAbbreviations. AChR, acetylcholine receptor; MIR, main immunogenic region; F,,,, N(9-fluorenylmethoxycarbonyl); VICE-GI, gation on this hypothesis requires detailed knowledge of the very immunogenic cytoplasmic epitope on ct subunit. common epitope.

The nicotinic acetylcholine receptor (AChR) from fish electric organs and vertebrate skeletal muscles is a transmembrane glycoprotein composed of five homologous subunits of the stoichiometry a&6 or aZj?&6,with known amino acid sequences. Acetylcholine binds to the two a subunits regulating the opening of the AChR ion channel (reviewed in 11- 31). Autoantibodies against the AChR cause AChR loss and/ or blockade of its function, resulting in impairment of neuromuscular transmission and development of the disease myasthenia gravis (reviewed in [4]). Despite extensive studies, the mechanism that triggers the autoimmune response against AChR is not yet understood and it is still uncertain whether, in addition to AChR, other cross-reactive molecule(s) participate in the initiation of the autoimmune response [5]. Monoclonal antibodies against AChR from various species and tissues have been produced by several groups and have proved to be excellent tools for studying both the molecule and the disease (reviewed in [6- 81). Among our mAb against the different AChR subunits, the anti-(a subunit) mAb have been the most extensively used [8, 91. When rats are immunized with intact AChR, production of mAb against the extracellular side of the AChR predomi-

916 Table 1. Characteristics of the used mAb. All data are from [lo- 151, except for epitope-mapping data which were derived from the present study. Underlined residues denote that they have been substituted in human AChR. Small-size residue letters denote partial contribution to the epitope. Ha, human a subunit.

mAb

Ig class

Binding to intact AChR from ~-

6 149 3 5 142 147 152 153 155 157 164 61

IgG 1 IgM IgG2b IgG2b IgG2a IgG2a IgG2a IgG2a IgG2a IgG2a IgG2a IgG2a IgG2a

187

IgG2a

8

Torpedo

human

++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++

+-

-

+++ ++ ++ ++ ++ ++ -

One of the major reasons for these mAb being valuable in the above studies was the considerable characterization of their epitopes on the AChR, achieved by multiple approaches [S, 91. Nevertheless, for most of these epitopes, only their approximate localization on the AChR has been determined. In this paper, we present the fine localization of the epitopes for most of the available non-MIR anti(u subunit) mAb in order to further improve their capacity as probes for AChR and myasthenia gravis. In order to produce a large number of synthetic peptides, we employed the PEPSCAN technique of Geysen and collaborators [25, 261. Epitopes of various sizes were identified. Emphasis was placed on the mAb that bound to a very-immunogenic epitope on AChR and cross-reacted with potentially myasthenogenic thymoma proteins from myasthenic patients. The details of their epitope characteristics are expected to serve as a key to further investigations on the role of this epitope and related structures in the induction of myasthenia gravis. MATERlALS AND METHODS Peptide synthesis

Peptides were synthesized on the tips of small poly(ethy1ene) rods on which polymers of acrylic acid had been formed by radiation grafting [25]. Rods with attached N-(9fluorenylmethoxycarbony1)-protected j-alanine [N-(9-fluorenylmethoxycarbonyl, F,,,)], all F,,,-amino acids and hydroxybenzotriazole were obtained from Cambridge Research Biochemicals, Cambridge; piperidine was from Fluka; all other reagents for synthesis were from Sigma. Peptide synthesis was performed according to Geysen [25] and the manufacturer’s instructions in 96 well microtiter plates. In order to exclude the possibility of cross-contamination during synthesis, only 24 rodsfplate were used.

Subunit and region specificity

Presently identified epitope

a, MIR a, cytoplasmic

(a67 - 74) (WNPADYGG) a341 - 344 (IFAD) a351 -360 ( I S ~ Q V T G E V ) a351 -360 ( I S G K W ) ~ 3 58 363 (GEVIFQ) a365 - 376 (PLIKNPDVKSAI) a370 - 379 (PDVKSAIEG!) a373 - 380 (KsAIEGvK) a373 - 380 (KsAIEG~K) a373 - 380 (KsAIEGvK) a373 - 380 ( K S A I E G ~ K ) a373 - 380 (KsAIEG~K) a375 - 382 (AIEGVKYI) (+Ha379 - 386) ( F Y I A E T M )

a a, cytoplasmic

a, cytoplasmic a, cytoplasmic a

cc a, cytoplasmic a, cytoplasmic cc, cytoplasmic a a,

cytoplasmic

a, cytoplasmic

-

AChR, mAb 61 from a rat immunized with SDS-denatured AChR CI subunit from the electric organ of Electrophorus electricus and all the other mAb from rats immunized with SDSdenatured Torpedo. The preparations used in this study were 50% ammonium sulfate precipitates from hybridoma supernatants, dialyzed against 145 mM NaCI, 7.5 mM Na2HP04, 2.5 mM NaH2P04, pH 7.4, (NaCI/P,), containing 0.05% (mass/vol.) NaN,. Some of their characteristics are shown in Table 3 . ELISA assays

Each peptide-carrying rod was incubated with 200 p1 NaC1/Pi containing 0.1 YOTween-20,0.8% bovine serum albumin and 0.2% ovalbumin for 1 h at room temperature. The rods were then incubated overnight with 175 p1 of a dilution of the test mAb (ljl000 to 1/5000)in the above solution. After incubation, the rods were washed three times with NaCI/Pi containing 0.05% Tween-20, for 10 min each wash, followed by incubation with peroxidase-labeled rabbit anti-(rat yglobulin) (DACO, ljl000 to lj2000 dilution) for 1 h. Then the rods were washed again three times in the same buffer. The presence of bound mAb was detected by reaction for 10 min with a substrate solution (0.05?40 azino-di-3-ethylbenzthiazodinsulphonate and 0.03% H z 0 2 in 0.1 M Na2HP04, 0.08 M citric acid, pH 4). The rods were then removed and the absorbance of the substrate was measured at 405 nm. Subsequently, the bound antibodies were released by sonicating the rods for 30 min in a water bath with 0.1 M NaH2P04, 1% SDS and 0.1% 2-mercaptoethanol, at 60°C. The rods were then washed twice with water, once with methanol and air dried. Peptides retained their antibody binding capacity for several assays. For each specific sequence, about 3-6 copies were synthesized, i.e. 3 - 6 rods with the same residue sequence, in 2 or 3 independent synthesis cycles.

Monoclonal antibodies

The production and characteristics of the mAb used have been described earlier [lo-13, 15, 161. mAbs 6, 3, 5 and 8 have been derived from rats immunized with intact Torpedo

Structural profiles

Hydrophilicity, flexibility and p-turn propensity profiles of residues 330 - 400 of the Torpedo u subunit were analyzed

91 7 according to the methods of Hopp and Woods [27], Karplus and Schulz [28] and Chou and Fasman [29], respectively.

1l167-74

RESULTS

0

Most of the mAb tested in the present study have been derived from rats immunized with SDS-denatured Torpedo electric organ AChR a subunit. They bind well to the denatured a subunit of Torpedo (and some of human muscle) AChR, as well as to the intact AChR [lo, 11, 131 (Table 1). The approximate location of the epitopes for these mAb has been previously determined by the use of conventional synthetic peptides [12,15,16] and for three of them by PEPSCAN peptides [17]. Therefore, we focused our efforts mainly to the parts of the a subunit which contain the predetermined segments. All possible overlapping octapeptides covering residues 308 -437 of the Torpedo a subunit and residues 330 388 of the human a subunit of AChR, as well as peptides of various lengths within these sequences, were synthesized on poly(ethy1ene) rods and were tested for mAb binding by ELISA.

1-

Monoclonal antibody binding to peptides of the Torpedo a subunit All fourteen mAb in Table 1 were studied for their binding activity to all possible overlapping octapeptides of residues 308 - 437 of the Torpedo a subunit as well as to the control MIR peptide a subunit residues 67 -74. Fig. 1 shows the binding patterns of eleven of these mAb. The majority of the mAb exhibited a quite specific binding pattern. Two antibodies (mAb 152 and the IgM mAb 149), in addition to their major binding sites, also exhibited binding to several other peptides, although of much lower level and probably of non-specific nature. IgM mAb often bind non-specifically to several peptides [12].The anti-(MIR) mAb 6, which was used as negative control, as expected bound only to residues 67 - 74 of the ci subunit. mAb 149 and 142 each bound to a group of several adjacent octapeptides, suggesting that a smaller segment common to all the peptides of the group may be the actual epitope. mAb 147 clearly bound to two groups of partially overlapping but distinct sites, suggesting that these sites comprise two different parts of the whole epitope, each sufficient for antibody binding (Fig. 1). mAb 152, 153, 155 and 164 (andmAb 157, not shown) bound to a single octapeptide, residues 373-380 of the a subunit, named the very immunogenic cytoplasmic epitope on the a subunit (VICE-a). Binding to a single octapeptide suggests that these mAb require the whole eight-residue segment in order to bind. mAb 61 bound primarily to residues 375-382 of the a subunit. mAb 8 and 187, earlier mapped to bind approximately to residues 366 - 378 and 333 - 349 respectively [15, 161, did not exhibit significant binding (Fig. 1). Finally, mAb 3 and 5 did not bind significantly to any of the peptides of Fig. 1. The above mAb were further mapped by the use of selected overlapping peptides of various sizes (Fig. 2). The use of the decapeptides generally confirmed the results of Fig. 1 and localized the epitopes for mAb 8,3 and 5. mAb against VICEci (mAb 153 and 164, shown in Fig. 2; mAb 152,155 and 157, not shown) bound to the three decapeptides which contain the above identified octamer composed of residues 373 - 380 of the CI subunit. They did not bind to any peptide smaller than eight residues, confirming that the minimum size of their

rnAb 6 320

340

360

380

420

400

0

m A b 149

rnAb 142

Y

rnAb 147

1-

mAb8

'1

I

mAb 1521

0 320

340

360

380

I

400

420

m A b 153

373-380

I

0

rnAb 155 I

11

1373-380

I

I 1-

0. 1-

I

-

m A b 164 373-380

mAb 61

0,

m A b 187

Fig. 1. Epitope analysis of 11 anti-AChR mAb. Monoclonal antibodies were tested for binding to all possible continuously overlapping octapeptides derived from the Torpedo CI subunit sequence 308 -437 (122 peptides), plus the control MIR a-subunit peptide 67- 74. Peptides were synthesized, remained attached to poly(ethy1ene) rods and were tested for mAb binding by ELISA using peroxidase-labelled anti-(rat y-globulin) antiserum. Numbers on the abscissa denote the amino-terminal residue numbers of the octapeptides.

epitope is residues 373 - 380 of the a subunit. mAb 149 bound to a tetrapeptide (residues 341 - 344) and mAb 142 bound to a pentapeptide (residues 359 - 363). mAb 147 again exhibited the two-site-binding pattern. The latter three mAb (mAb 149,

91 8 mAb 142

1-

mAb 1491

1

mAb8

mAb 147

369-778

1-

I

362-37 1

I

0-I

1-

LLAL--

0 mAb m A b88

370-379

0

-

7

I

L

!E CI

373-380

II L

rnAb 155

1-

0.

-

0

rn

mAb 164

I-

Q 0,-

I

-

7

'1

375-3821

rnAb 61 I

Fig. 3. Monoclonal antibody binding to overlapping human AChR asubunit synthetic peptides. Binding to two decapeptides, fiftytwo octapeptides and six hexapeptides is presented. Experimental conditions and symbols as in Figs 1 and 2.

-

H 1 OP

H

H 7P

H 6P

5P

4P

Fig. 2. Monoclonal antibody binding to selected Torpedo AChR a-subunit synthetic peptides of varying lengths. Experimental conditions and symbols as in Fig. 1 . Symbols lop, 7p etc., denote decapeptides, heptapeptides etc. Gaps in the horizontal axis of each mAb denote that the corresponding peptides were not tested.

142 and 147) have recently been mapped by Das and Lindstrom with PEPSCAN peptides to similar sites (residues 341 - 347, 359 - 365 and 368 - 375, respectively) [17], however, the peptides used by these authors did not allow them to identify either the minimum epitopes for the first two mAb or the two-site-binding pattern of mAb 147. The binding pattern of the latter mAb is peculiar and is thus analyzed in the Discussion. mAb 61 bound well to the a-subunit heptapeptide 376-382. mAb 3 and 5 both bound to residues 351 -360 of the a subunit (summarized in Table 1) which was included in a subsequently synthesized group of decapeptides (from residues 343 - 352 to residues 358 - 367). Monoclonal antibody binding to human a-subunit peptides

Subsequently, the mAb were tested for binding to all possible overlapping octapeptides (as well as to selected decapeptides and hexapeptides) of the human muscle AChR CY subunit sequence of residues 330-388 (Fig. 3). mAb 142 (Fig. 3) and 149,147,187,3 and 5 (not shown) are specific for Torpedo AChR, thus, expectably, they did not bind to any human peptide. mAb 8, as with Torpedo peptides, did not bind

to any octapeptide, but it did bind to the human decapeptide of residues 370 - 379, i.e. homologous to the Torpedo decapeptide to which the mAb bound. The group of mAb against VICE-a (mAb 153,155 and 164, as well as mAbs 152 and 157 which are not shown) bound to the human octapeptide of the a subunit composed of residues 373 - 380 (homologous to the Torpedo VICE-a), as well as to the decapeptide that contains this segment. mAb 61, in addition to binding to residues 375 382 of the human a subunit as expected (and to a lower extent to its five surrounding peptides), also bound very well to peptide 379 - 386 (Fig. 3). Structural features of the epitopes

The structural profiles of the identified epitopes are shown in Fig. 4. Hydrophilicity, flexibility and high fl-turn propensity have been proposed as characteristics of high antigenicity [26, 30 - 321. mAb 3,5,147 and 8 bound to hydrophilic and flexible epitopes. The epitope for mAb 147 also contains a putative fl turn which is part of each of the two binding subsites of this epitope. Interestingly, no characteristic profiles could be detected for VICE-a, nor for the epitopes for mAb 61 and 142 (Fig. 4). Determination of the antigenic role of each residue within the VICE-a

VICE-a is likely to play a critical role in the induction of the autoimmune response to AChR in thymoma-associated

919 A

'1

330

Hydrophilicity

'

340

'

350

'

360

'

370

'

360

'

390

'

Tor-a K R A S K E K Q E N K I F A D D I D I S D I S G K Q V T G E V I F P T P L I K N P D V K S A I E G V K Y I A E H M K S D E E S S N A A E E W K Hum-a P R DK TE PGPPPMO Y S H E I I Q N A

C

Fig. 4. Comparison of the determined epitope locations with structural profiles of the corresponding a-subunit sequence. (A) Hydrophilicity [27], flexibility [28] and /?-turn propensity profiles [29] of the Torpedo a-subunit residues 330-400. Segments of the three plots above the dashed lines represent hydrophilic, flexible or probable /?-turn sites, respectively. (B) Amino acid sequence of Torpedo (Tor) and human (Hum) sequence 330-400 of the a subunit [I]. Only the substituted residues in the human a subunit are shown; non-conservative substitutions are underlined. (C) The presently determined epitopes are summarized; () at the ends of some epitopes denote partial involvement of the corresponding residues; dashed lines represent the epitopes on the human a subunit; numbers represent the various mAb.

myasthenia gravis [23,24], therefore the antigenic role of each residue within this segment was then determined. A series of 50 different analogues of residues 373 - 380 of the a subunit was synthesized. Each residue was substituted with various amino acids representative of the inert (Ala, Gly), positively charged (Lys or Arg), negatively charged (Glu) and hydrophobic (usually Leu) amino acid groups. They were also selected such as to represent both conservative and non-conservative substitutions for each residue (Fig. 5). Overall, the contribution of each residue to the antigenicity of the epitope was rather similar among all the fivemAb. However, a few significant differences in their binding patterns could be identified, especially for mAb 157 and 164, which were generally more sensitive to substitutions (Fig. 5). The average values of mAb binding to all 50 analogues, expressed as a percentage of their binding to the intact VICE-a, were 54%, 53%, 47%, 37% and 39%, for mAb 152,153, 155,157 and 164, respectively. The relative contribution of the different VICE-a residues to the average mAb binding was quite different from residue to residue (Figs 5 and 6). Lys373 was the most indispensable, any of the six substitutions, including its conservative substitution by Arg, completely inactivated the epitope. In contrast, Ser374, Ala375 and Val379 do not seem to contribute much to the antigenicity of the epitope. Overall, neither conservative nor non-conservative substitutions at these residues significantly affected mAb binding. Only substitution of Val379 by Glu decreased mAb binding below 50%; this was apparently due to the negative charge of Glu, since its amide derivative, Gln, had practically no effect on mAb binding. Residues Ile376, Glu377 and Gly378 offered moderate contributions to the antigenicity of the epitope. The binding pattern of Gly378 analogues was unexpected; while its non-

conservative substitution by the large aromatic amino acids Tyr and Phe (or by Ser) had no dramatic effect, other substitutions, including the conservative Ala substitution, abolished mAb binding. We suggest that Gly378 (essentially the backbone of the peptide at this site) is critical to mAb binding; thus, addition of the side chains of Ala, Leu, Lys or Glu abolishes binding. The fact that its substitution by Ser, Tyr or Phe does not abolish peptide function may mean that their side chains are oriented away from the antibody-binding surface of the epitope. The requirements for epitope integrity on residue 380 of the sequence 373 - 380 seemed to be the positive charge and the large size of the amino acid. Thus substitution of Lys by the positively charged Arg had no effect whatsoever, whereas the negatively charged Glu completely abolished mAb binding. The uncharged, but large amino acids, Gln and Leu exhibited, on average, only moderate effects (varying from weak to strong, depending on the tested mAb), whereas the small amino acids Gly and Ala abolished binding (Figs 5 and 6). Finally, we investigated the effect of simultaneous substitutions of two or three residues (Fig. 7). mAb bound well to the Asn374/Ile379 analogue which corresponds to the chick muscle AChR CI subunit sequence. Interestingly, simultaneous substitution of all three residues 374, 375 and 379, that were shown above not to be critical for mAb binding, only weakly affected binding. In contrast, simultaneous substitution of the three residues in the middle (residues 376 - 378), each of which when tested alone moderately affected mAb binding (Fig. 5), resulted in practically complete loss of the antibody binding capacity of the peptide (Fig. 7). This effect (on the average 4% binding) is not very different from the theoretical value calculated as if each substitution acted independently (on the average 15% binding).

920 100

50 A

g o ii)

A

G

R

H

Q

I-< 373

I

L

A

G

K

E

s

L

374

r)

..r

.-

G

a

K

E

0 A0

-".

L

Y

A

G

K

A375

%

E

Q

I376

I

I

I

1001

D

.c _

50

U

.a c

n "

A

G

il

d

g

K

L

E377

A

K

E

L

G

S

Y

F

378

I

I

100

50

A

G

K

E v379

Q

L

I

A

G

R

E

Q

L

K380

a373-580 analogues Fig. 5. Binding of mAb, specific for VICE-a, to single residue analogues of the residues 373 - 380 of the a-subunit octapeptide. Letters just below the bars represent the amino acids used for each substitution, whereas the larger-size letters represent the original amino acids. Each group of five bars represents binding of mAb 152 (first bar), 153, 155, 157 and 164 (last bar) to the same analogue, each given as a percentage of the binding value of the corresponding mAb to the original peptide.

DISCUSSION Precise epitope determination of mAb is a prerequisite for exploiting their full potential as tools in the study of the corresponding antigen. In the present study, we have used overlapping small synthetic peptides to determine the exact sites of the sequential epitopes for several mAb to the cytoplasmic side of the AChR a subunit. The epitope that may be critical for the induction of the autoimmune response in the thymoma-associated myasthenia gravis was characterized in depth in order to provide the basis for subsequent investigations on the elucidation of the myasthenia-inducing mechanisms. Use of synthetic peptides for mAb mapping may not always reveal the whole epitope as it is in the intact molecule [33], but when properly used, they may identify at least critical parts of the actual epitope. For example, combined data from different approaches (peptide mapping, antibody competition for binding to the intact molecule and site-directed mutagenesis) have unequivocally shown the validity of the peptide mapping techniques for anti-(AChR) mAb mapping [9, 16, 17, 341. The PEPSCAN technique allows the production, in a short time, of many peptides, ready for use in ELISA since they are bound on an appropriate insoluble matrix [25, 261. In addition, the mode of peptide attachment to the matrix allows the whole molecule to be accessible for

binding to the antibody, as opposed to techniques in which ELISA plates are coated with soluble peptides. This technique has been used successfully in epitope mapping for several antigens [25, 26, 35, 361 as well as for the mapping of other ligand binding sites [37]. In most cases, the present results were generally consistent with earlier approximate epitope localization of the same mAb [12, 15, 161. mAb 147 bound to two discontinuously overlapping groups of peptides (Figs 1 and 2). Amongst the first group of positive peptides, it bound only to those ending at positions 371 and 372 of the a subunit, regardless of their size. Since it bound to the two heptamers of residues 365 - 371 and 366 372 of the a subunit, it should also bind to all larger peptides that contained these sequences. However, peptides extending over position 372 of the ct subunit were inactive. A likely explanation for this peculiar binding behavior is the following. The complete epitope formAb 147 may be the dodecamer 365 - 376, whose putative strong /I turn at its middle (positions 369 - 372) seems to be fundamental for mAb binding. Thus, the intact /I turn, together with any of its two arms, may be sufficient for mAb binding. However, when the positively charged Lys373 was added to the first binding segment (residues 366 - 372), it probably interacted with Asp371, thus interfering with its interaction with Asn369, resulting in unfolding of the fl turn and inactivation of the peptide. Subsequent addition of residues 374- 376 apparently again allowed the formation of the p turn, resulting in the second group of mAb-147-binding peptides. The present results explain to a considerable extent the species cross-reactivities of the studied mAb. Generally, mAb that cross-react with intact Torpedo and human AChR bound to the corresponding peptides of both AChR; these peptides had only small differences between them. Thus, mAb 152164, which bind well to both Torpedo and human AChR, bound to both Torpedo and human sequences 373 - 380 of the a subunit (VICE-a), which differ only by one conservative substitution (Val379 to Ile379). However, Torpedo-specific mAb generally bound only to Torpedo peptides which were very different from the corresponding human segments. These correlations further suggest that the identified epitopes represent the major parts of the actual mAb epitopes on the intact AChR. Nevertheless, the binding patterns of mAb 147 and 8 to the human peptides were not exactly as expected. mAb 147 bound only to Torpedo peptides, whereas it cross-reacts with intact human AChR (Table 1) and was earlier found to bind to the 17-amino-acid human ct subunit 364-380 [12]. The human homologue of the presently identified 12-residue whole epitope (residues 365 - 376 of the ct subunit), has two substitutions (at positions 369 and 371), which abolish the putative /I-turn structure (not shown). This fact may explain the absence ofmAb 147 binding to the present human peptides and is in accordance with our above-described speculations. However, when the whole epitope was present, as in the earlier studies, the contribution of the other ten residues apparently compensated for the two substitutions. mAb 8, contrary to mAb 147, does not cross-react with intact human AChR, but it did bind to a human decapeptide, homologous to its Torpedo epitope (Fig. 3), as well as to an overlapping human 18-amino-acid peptide [12]:Apparently, its inability to bind to the intact human AChR is due to steric hindrance induced by substitutions on neighboring residues on the intact AChR. In-depth characterization of the epitope(s) for the three anti-(AChR) mAb that cross-react with thymoma proteins from myasthenic patients [23, 241 was also performed. This

921

;=! -

0

‘ m

50

0

E

AGRHQL

AGKEL

GKELY

AGKEQ

K

S

A

I

x

v

AGKL

E

AKELSYF

G

AGKEQLI

V

AGREQL

K

a373-580 analogues Fig. 6. Average values of mAb binding to the VICE-a analogues. Each bar represents the mean of the percentages of binding of mAb 152, 153, 155, 157 and 164. 2373 - 380, residues 373 - 380 of the cc subunit.

0

.N.. .I.

.AG.....

.AG...A.

... KLY.

x

v

a.373-380 (KSAIEGVK) analogues

Fig. 7. Binding of mAb to five VICE-a analogues with multiple-residue substitutions. Just below the bars are shown the substituted residues, whereas the unchanged residues are represented by dots. The first analogue corresponds to the chick residues 373 - 380 of the ci subunit (a373 - 380). The other analogues contain substitutions selected from the single substitutions of Fig. 5.

epitope (VICE-a on residues 373 - 380) is not characterized by hydrophilicity, flexibility and high B-turn profiles, considered as a propensity for high antigenicity, but rather has the characteristics of an amphipathic structure. In fact, it is located in the amphipathic region which was considered in the past as the ‘M5 helix’, supposed to form the ion channel of the AChR. It turns out that this structure seems to be the most immunogenic structure of the denatured a subunit of the AChR. The fact that VICE-a has proved to be a highly antigenic B-cell epitope and also has the characteristics of Tcell epitopes (i.e. amphipathic character [38]) is in favor of a potential myasthenia-triggering role of the proteins that carry this epitope. The three thymoma-binding mAb bound to the cross-reactive proteins with significantly different efficiency (mAb 155 being the best and mAb 152 the less well binding mAb) [23]. This fact, together with differences in the relative binding efficiency of these mAb for AChR from various species [13], suggested that their binding sites are not identical. Study of the antigenic role of each residue within VICE-a, using peptide analogues, showed that, despite the similarities in the binding patterns of these mAb, a number of characteristic binding differences also exist. This indicates that, although these mAb bind to the same epitope, their antigen-binding sites differ significantly. The contribution of each VICE-a residue to mAb binding varied a lot among the different residues. Overall, the core three residues 376 - 378 (having at their center the negatively charged Glu377) were quite critical; these were surrounded by three rather inactive residues (Ser374, Ala375 and Va1379) which, in turn, were surrounded by the most critical residues

at the two ends (the positively charged Lys373 and Lys380). Even simultaneous substitutions of the inactive residues Ser374, Ala375 and Val379 in the same peptide only weakly affected mAb binding. Furthermore, with the exception of Lys373, any other critical residue could be substituted in single-substitution analogues by selected amino acids without dramatic loss of the mAb-binding efficiency. The detailed characterization of VICE-a is expected to provide the key to further studies on the pathogenetic role of this epitope and of the proteins that carry it. Analysis of DNA and RNA from myasthenia-associated thymomas by a probe based on an earlier estimated location of VICE-a (residues 371 - 378 instead of residues 373 - 380) revealed the presence of a non-AChR RNA containing the base sequence corresponding to residues 371 - 378 of the a subunit (presumably coding for the p153 protein) and three non-AChR DNA restriction fragments [39]. A direct application of the present results will be to allow the construction of probes coding for the actual epitope, as well as for its analogues that retain the basic antigenic characteristics. These probes may identify additional genes with potentially critical involvement in myasthenia gravis. Another use of the obtained data will be the investigation of VICE-a and its active analogues as potential T-cell epitopes. This work was supported by grants from the Greek General Secretariat of Research and Technology, the United Nations Industrial Development Organization and the Association FranGaise contre les Myopathies. We thank Ch. Dosiou for excellent assistance, I. Papadouli and A. Kokla for valuable help and discussions, H. Loutrari and Drs R. Matsas, A. Mamalaki, E. Patsavoudi, K. Soteriadou for valuable suggestions, and E. Tzartos for secretarial assistance.

REFERENCES 1. Numa, S. (1987) Chemica Scripta 27B, 5-19. 2. Maelicke, A. (1988) Handb. Expl. Pharrnacol. 86, 267-3313, 3. Changeux, J. P. (1990) Fidia Research Foundation Neuroscience Award Lectures vol. 4, pp. 21 - 168, Raven Press, New York. 4. Lindstrom, J., Shelton, D. & Fugii, Y. (1988) Adv. Immunol. 42, 233-284. 5. Hohlfeld, R . (1990) Lab. Invest. 62, 241 -243. 6. Fuchs, S., Souroujon, M. C. & Mochly-Rosen, D. (1984) in Receptors and Recognition B17 (Greaves, M. F., ed.) pp. 163-199, Chapman and Hall, London. 7. Lindstrom, J:, Schoepfer, R. & Whiting, P. (1987) Molec. Neurohiol. 1 , 283 -337. 8. Tzartos, S. J. (1991) Current aspects in neurosciences (Osborne, N., ed.) vol. 3, pp. 195-226, Macmillan Press, Houndmills.

922 9. Tzartos, S. J., Cung, M. T., Demange, P., Loutrari, H., Mamalaki, A,, Marraud, M., Papadouli, I., Sakarellos, C. & Tsikaris, V. (1992) Mol. Neurobiol. 5, 1-29. 10. Tzartos, S. J. & Lindstrom, J. L. (1980) Proc. Natl Acad. Sci. USA 77,755 -759. 11. Tzartos, S. J., Rand, D. E., Einarson, B. E. & Lindstrom, J. M. (1981) J . Biol. Chem. 256, 8635-8645. 12. Tzartos, S. J., Kokla, A,, Walgrave, S. & Conti-Tronconi, B. (1988) Proc. Natl Acad. Sci. USA 85,2899-2903. 13. Tzartos, S., Langeberg, L., Hochschwender, S., Swanson, L. & Lindstrom, J. (1986) J . Neuroimmunol. 10, 235-253. 14. Sargent, P., Hedges, B., Tsavaler, L., Clemmons, L., Tzartos, S. J. & Lindstrom, J. (1984) J . Cell Biol. 98, 609-618. 15. Ratnam, M., Le Nguyen, D., Rivier, J., Sargent, P. & Lindstrom, J. (1986) Biochemistry. 25, 2633 -2643. 16. Tzartos, S. J., Loutrari, H. V., Tang, F., Kokla, A,, Walgrave, S. L., Milius, R. P. & Conti-Tronconi, B. M. (1990) J . Neurochem. 54, 51 -61. 17. Das, M. K. & Lindstrom, J. (1991) Biochemistry 30,2470-2477. 18. Tzartos, S. J. & Changeux, J.-P. (1984) J . Biol. Chem. 259, 11 512-11 519. 19. Merlie, J. P. &Smith, M. M. (1986) J . Membr. Biol. 91, 1 - 10. 20. Blount, P. & Merlie, J. P. (1988) J . Biol. Chem. 263, 1072- 1080. 21. Hohlfeld, R., Toyka, K., Tzartos, S. J., Carson, W. & ContiTronconi, B. (1987) Proc. Natl Acad. Sci. USA 84,5379 - 5383. 22. Zhang, Y., Tzartos, S. J. & Wekerle, H. (1988) Eur. J. Zmmunol. 18,211 -218. 23. Kirchner, T., Tzartos, S., Hoppe, F., Schalke, B., Wekerle, H. & Muller-Hermelink, H. K . (1988) Am. J . Pathol. 130,268-279. 24. Marx, A., O’Connor, R., Geuder, K. I., Hoppe, F., Schalke, B., Tzartos, S., Kalies, I., Kirchner, T. & Muller-Hermelink, H. K. (1990) Lab. Invest. 62, 279-2286, 25. Geysen, H. M., Rodda, S. J., Mason, T. M., Tribbick, G. & Schoofs, P. G. (1987) J . Zmmunol. Meth. 102,259-274.

26. Geysen, H. M, Tainer, J. A., Rodda, S. J., Mason, T. J., Alexander, H., Getzoff, E. & Lerner, R. A. (1987) Science 235, 1184 - 1190. 27. Hopp, T. P. &Wood, K. R. (1981) Proc. Nail Acad. Sci. USA 78,3824-3828. 28. Karplus, P. A. & Schulz, G. E. (1985) Naturwissenschaften 72, 212 - 213. 29. Chou, P. Y. & Fasman, G. D. (1979) Biophys. J . 26,367-384. 30. Westhof, E., Altschuh, D., Moras, D., Bloomer, A. C., Mondragon, A., Klug, A. & VanRegenmortel, M. H. V. (1984) Nature 311, 123-126. 31. Tainer, J., Getzoff, E., Patterson, Y., Olson, A. & Lerner, R. A. (1985) Annu. Rev. Zmmunol. 3, 501 -535. 32. Krchnak, V., Mach, 0.& Maly, A. (1990) Methods Enzymol. 178, 586-611. 33. Laver, W. G., Air, G. M., Webster, R. G. & Smith-Gill, S. J. (1990) Cell61, 553-556. 34. Saedi, M. S., Anand, R., Conroy, W. G. & Lindstrom, J. (1990) FEBS Lett. 267, 55- 59. 35. Tan, X. H., Ratnam, M., Huang, S. M., Smith, P. L. & Freisheim, J. H. (1990) J. Biol. Chem. 265, 8022-8026. 36. Soteriadou, K. P., Remoundos, M., Katsikas, M. C., Tzinia, A. K., Tsikaris, V., Sakarellos, C. & Tzartos S. J. (1992) J . Biol. Chem., in the press. 37. Tzartos, S. J. & Remoundos, M. S. (1990) J . Biol. Chem. 265, 21 462 - 21 467. 38. DeLisi, C. & Berzofsky, J. A. (1985) Proc. Natl Acad. Sci. USA 82,7048 - 7052. 39. Geuder, K. I., Schoepfer, R., Heinemann, S., Witzemann, V., Schalke, B., Kirchner, T., & Muller-Hermelink, H. K. (1992) Lab. Invest., in the press.