Kinetic and Functional Mapping of Viral Epitopes Using Biosensor Technology

Kinetic and Functional Mapping of Viral Epitopes Using Biosensor Technology

VIROLOGY 213, 462–471 (1995) Kinetic and Functional Mapping of Viral Epitopes Using Biosensor Technology HE´LE`NE SAUNAL and MARC H. V. VAN REGENMOR...

116KB Sizes 0 Downloads 12 Views

VIROLOGY

213, 462–471 (1995)

Kinetic and Functional Mapping of Viral Epitopes Using Biosensor Technology HE´LE`NE SAUNAL and MARC H. V. VAN REGENMORTEL1 Institut de Biologie Mole´culaire et Cellulaire, Centre National de la Recherche Scientifique, 15, rue Descartes, 67084 Strasbourg Cedex, France Received June 16, 1995; accepted September 6, 1995 Some monoclonal antibodies (Mabs) that react with the extremity of the tobacco mosaic virus (TMV) particle containing the 5* end of the RNA are able to block the disassembly of TMV by ribosomes while others are totally devoid of such activity. No correlation could be established between the binding kinetics and affinity of the Mabs and their inhibitory capacity. An epitope map of the Mab binding sites was constructed on the basis of kinetic two-site binding assays with the viral monomeric protein (TMVP) performed using biosensor technology (BIAcore). Mabs possessing inhibitory activity were found to bind to the part of the TMVP surface closest to the central axis in the polymerized particle. As this part of the subunit is known to interact with the viral RNA, it seems that inhibitory Mabs act by sterically preventing the interaction between virus and ribosomes. This study illustrates the advantages of the biosensor technology for locating conformational epitopes in viral proteins. q 1995 Academic Press, Inc.

INTRODUCTION

units. Metatopes are present in both dissociated and polymerized forms of the viral protein. It has been possible to locate some cryptotopes in the TMVP molecule because the corresponding anticryptotope antibodies reacted with natural or synthetic fragments of the protein (Benjamini, 1977; Trifilieff et al., 1991; Van Regenmortel, 1986, 1990). This type of epitope mapping was not successful in the case of neotopes and metatopes because none of the corresponding antibodies reacted with any linear fragments of TMVP. Some information on the location of a limited number of neotopes and metatopes on TMV particles was obtained by immunoelectron microscopy (Dore et al., 1988, 1990). All examined antimetatope Mabs were found to react with the virion extremity containing the 5* end of the RNA, whereas the antineotope Mabs reacted along the entire length of the virus particle. In a recent study the location of neotopes and metatopes in TMV was further investigated by means of a new biosensor technology (BIAcore, Pharmacia AB, Uppsala) using all available Mabs to TMV and TMVP (Saunal and Van Regenmortel, 1995). Double antibody capture assays with viral subunits and binding stoichiometry calculations with virus particles made it possible to demonstrate the presence of neotope and metatope specificities on parts of the viral surface where they had not been found before. Some metatopes were found to be present on the entire surface of the virion, indicating that these epitopes were located on the external surface A of the subunit (see Fig. 1). Some neotopes were identified at the extremity of the viral rod, i.e., on the polymerized surface E* of the subunit. In earlier studies, this surface had been found to harbor only metatopes (Dore et al., 1988). The biosensor analysis thus made it possible to distinguish two types of antimetatope Mab reacting with surfaces A and E of the subunit, respectively, as well as two types of antineotope

It has been demonstrated that in an in vitro translation system, disassembly of tobacco mosaic virus (TMV) particles can be induced by ribosomes that reach the 5* leader sequence of RNA. These ribosomes initiate translation and dislodge the coat protein subunits during their translocation on the viral RNA (Wilson, 1984; Wilson and Shaw, 1985). This cotranslational disassembly was also shown to take place in vivo upon inoculation of plants with TMV (Shaw et al., 1986). In an earlier study monoclonal antibodies (Mabs) raised against TMV coat protein (TMVP) were tested for their ability to block the in vitro disassembly of virions and the translation of viral RNA (Saunal et al., 1993). The majority of the antibodies that were tested reacted with the extremity of the virion that contained the 5* end of the RNA and about half of these antibodies were found to inhibit the translation of the viral RNA. Since this inhibitory activity may be related to the property possessed by many viral antibodies to neutralize virus infectivity, it seemed of interest to investigate why some TMV antibodies inhibited RNA translation and others did not. Antibodies raised against TMV or dissociated TMVP have been classified according to their ability to react with three different types of epitopes known as cryptotopes, neotopes, and metatopes (Van Regenmortel, 1966, 1990). Cryptotopes are epitopes that are buried inside the assembled virus particle and become accessible to antibodies only after dissociation of the virion. Neotopes are specific for the quaternary structure of the virion and are thus not present at the surface of monomeric sub1 To whom correspondence and reprint requests should be addressed. Fax: 33 88 61 06 80.

0042-6822/95 $12.00

462

Copyright q 1995 by Academic Press, Inc. All rights of reproduction in any form reserved.

/ m4171$7543

10-11-95 12:11:42

viral

AP-Virology

EPITOPE MAPPING USING BIOSENSORS

463

al., 1982, 1985). The Mabs were purified from ascitic fluid by affinity chromatography on Hi Trap Protein A (Pharmacia 17-0402-01) and dialyzed against 0.1% (w/v) NaN3 in 0.02 M phosphate buffer, pH 7.0. Measurement of inhibition of in vitro disassembly of TMV by Mabs

FIG. 1. Schematic model of TMV protein subunits in monomeric form and in the virus particle, with an indication of which protein surface is recognized by each of the 23 Mabs. This allocation was done on the basis of BIAcore mapping results (Saunal and Van Regenmortel, 1995). It should be noted that earlier mapping attempts by ELISA had led to some erroneous conclusions. For instance, Mabs 42P, 67P, and 249P had previously been considered to be antimetatope antibodies (Al Moudallal et al., 1985; Saunal et al., 1993).

Mab reacting with surfaces A* and E* of the virion (Fig. 1). The advantages of biosensor technology for analyzing viral epitopes have been described in a recent review (Van Regenmortel et al., 1994). Two-site BIAcore binding assays with TMVP (Saunal and Van Regenmortel, 1995) showed that the binding of a first antimetatope Mab on one face of the subunit (such as surface E) did not prevent the binding of a second antimetatope Mab on another face. Furthermore some pairs of Mabs specific for surface E were able to bind concurrently to the same monomeric subunit. However, no case was observed where two Mabs specific for surface A bound to the same TMVP molecule. In view of the more refined mapping of TMV epitopes made possible by the biosensor technology, we decided to investigate if there was any correlation between the location of metatopes on TMVP and the capacity of the corresponding antibodies to inhibit cotranslational disassembly. In addition, the binding affinity and kinetics of the different antibodies were measured with BIAcore in order to establish if there was any correlation between antibody affinity and inhibitory capacity. The results showed that the Mabs possessing the strongest inhibitory capacity bound to the part of the surface E closest to the central axis of polymerized TMVP. Since this portion of the TMVP surface is known to interact with the viral RNA, these results suggest that the Mabs inhibit disassembly by directly preventing ribosomes from dislodging viral subunits from the RNA. MATERIALS AND METHODS Purified TMV (common strain) and Mabs raised to TMV and TMVP were from laboratory stocks (Al Moudallal et

/ m4171$7543

10-11-95 12:11:42

viral

Inhibition tests were performed as previously described (Saunal et al., 1993) with slight modifications. After centrifuging TMV–Mab complexes it was difficult to remove completely the supernatants from the 200 ml TL centrifuge tubes (TL-100 Beckman ultracentrifuge). Small residual volumes (1–5 ml) remained in the tubes and diluted the medium of translation (15 ml), which led to variable results. In the present study, quantities of TMV, Mabs, and medium of translation used in the inhibition assays were doubled so that any small volume of supernatants remaining in the tubes became negligible in comparison with the 30 ml of translation medium. Results were found to be more reproducible. Experiments were performed at molar ratios of 50, 100, and 200 Mabs per virus particle. Biosensor measurements The biosensor instrument BIAcore (Pharmacia Biosensor AB, Uppsala, Sweden) allows real-time biospecific interaction analysis. This system uses the optical phenomenon of surface plasmon resonance which detects changes in optical properties at the surface of a thin gold film on the sensor chip (Lo¨fas and Johnsson, 1990). The sensor chip carries a dextran matrix on which one of the two reactants is covalently linked, while the other one is introduced in a flux passing over the surface. The resonance angle depends on the refractive index in the vicinity of the surface which changes as the concentration of molecules on the surface is modified. This concentration is expressed in resonance units (RU). A signal of 1000 RU corresponds approximately to a surface concentration change of 1 ng/mm2. The BIAcore system, sensor chip CM5, surfactant P20, and amine coupling kit containing N-hydroxysuccinimide (NHS), N-ethyl-N*-(3-diethylaminopropyl) carbodiimide (EDC), as well as ethanolamine-HCl were from Pharmacia Biosensor AB. Preparation of the sensor surfaces. Most antibodies were immobilized covalently on the dextran matrix. Many binding assays can be performed on the same chip containing immobilized antibodies since the bound antigen can be dissociated from the covalently linked Mab by injection of a few microliters of HCl. Usually regenerated Mab surfaces can be used at least 50 times. Some Mabs were found to lose binding activity after HCl regeneration and these were immobilized by means of a first layer of rabbit anti-mouse Fc globulins (RAM) covalently linked to the dextran matrix. In this case, the noncovalently trapped Mab was removed during each regeneration cy-

AP-Virology

464

SAUNAL AND VAN REGENMORTEL

cle and it had to be reinjected on the RAM surface in each successive binding assay. Anti-TMV Mabs 5V, 16P, 17V, 29V, 47P, 151P, and 173P were covalently bound to the matrix while Mabs 25P, 167P, 181P, and 188P were bound via RAM. Covalent immobilization of RAM or Mabs to the sensor chip was performed via primary amino groups using standard procedures described previously (Fa¨gerstam et al., 1990; Lo¨fas and Johnsson, 1990). The immobilization run was performed at a flow of 5 ml/min in HBS buffer, pH 7.4 (10 mM HEPES, 0.15 M NaCl, 3.4 mM EDTA, 0.05% surfactant P20). The carboxylated matrix was first activated with 35 ml of an EDC/NHS mixture. Thirty microliters of RAM Fc diluted in 10 mM acetate buffer, pH 5, at 100 mg/ml or 30 ml of Mabs diluted in 10 mM acetate buffer, pH 4, at 150 mg/ml was injected. Unreacted groups were blocked by the injection of 30 ml of ethanolamine–HCl, pH 8.5, followed by three 5-ml washes of 0.1 N HCl to remove noncovalently bound antibodies. Two-site binding assays with TMVP. After direct or indirect immobilization of capture Mab1 on the sensor surface, 30 ml of TMVP at 0.26 mg/ml was injected and trapped by this Mab. The amount of trapped TMVP was usually higher when the capture Mab was immobilized directly on the matrix (230 to 1360 RU) than when it was immobilized via a first layer of RAM (75 to 210 RU). The unoccupied sites of RAM Fc were blocked with 20 ml of a nonspecific Mab (ascitic fluid diluted 1/5) and Mab2 (30 ml, 150 mg/ml) was then injected and allowed to interact with TMVP. The concentration of molecules (expressed in RU) on the sensor chip surface is monitored continuously over time and is registered as a sensorgram. The y axis of the sensorgram is denoted as the RU signal, whereas the time (in sec) is represented on the x axis. Figure 2 shows two typical sensorgrams of two-site binding assays performed after direct immobilization of a capture Mab1. During two-site binding assays, continuous dissociation and reassociation of one or more of the binding partners occurs. When the kd of Mab1 is about 1003 sec01, approximately half of the trapped antigen will become dissociated during the 10 min of Mab2 injection. On the dissociated antigen molecule, the part of the surface previously occupied by Mab1 may then bind to Mab2 and this will prevent reassociation of the antigen to Mab1. This results in a lower RU response. In view of the dynamic competition between different Mabs that can be observed with the BIAcore, the type of epitope mapping used in this study has been termed kinetic mapping. The reversible nature of antigen–antibody interactions can be visualized if the kd of the immobilized first antibody lies in the region 1002 to 1004. If the kd value of Mab1 is larger than 1002 sec01, no meaningful data can be obtained since practically all antigen molecules will dissociate during the time frame of the experiment. Stoichiometry calculations. When injection of Mab2 led

/ m4171$7543

10-11-95 12:11:42

viral

FIG. 2. Sensorgrams corresponding to two-site binding assays. (A) The RU level at position a corresponds to the amount of capture Mab1 (47P) covalently immobilized on the dextran matrix. Phase a to b corresponds to the binding of 949 RU of TMVP. Phase b to c corresponds to the injection of Mab2 (151P). The positive response (1637 RU) indicates that Mab 151P does not overlap the epitope occupied by Mab 47P. Phase c to d corresponds to the injection of 10 ml of HCl 100 mM for the regeneration of the Mab1-immobilized surface by dissociation of noncovalently bound molecules. (B) The RU level at position a corresponds to the amount of capture Mab1 (47P) remaining linked to the matrix after the regeneration phase described above. Phase a to b corresponds to the binding of 892 RU of TMVP. Phase b to c corresponds to the injection of Mab2 (16P). The negative response (0104 RU) indicates that Mab 16P overlaps the epitope occupied by Mab 47P. Phase c to d corresponds to the regeneration phase. A flow rate of 5 ml/min was used during phases a to d.

to a positive RU response (Fig. 2A), the molar ratio (MR) of Mab2/TMVP was calculated from MR Å (D RU Mab2/ D RU TMVP) 1 (MW TMVP/MW Mab) (Daiss and Scalice, 1994). MR values range from 0, for an interfering interaction in which Mab1 and Mab2 fully overlap, to 1.0, for a noninterfering interaction in which Mab2 and Mab1 recognize totally nonoverlapping epitopes. When injection of Mab2 led to a negative RU response (Fig. 2B), the interaction was considered as interfering. Binding affinity and kinetics of antimetatope Mabs of type E reacting with TMVP. Since TMVP tends to aggregate at the concentration needed to achieve sufficient immobilization levels, it was not possible to immobilize directly monomeric viral protein on the sensor surface (Saunal and Van Regenmortel, 1995). TMVP was there-

AP-Virology

EPITOPE MAPPING USING BIOSENSORS

fore trapped by a covalently immobilized first Mab. Mab 17V was used as capture Mab because it recognizes surface A and was unlikely to influence the binding kinetics of Mabs of type E added as second antibody. Moreover the reactivity of Mab 17V was shown not to decrease after several HCl regeneration cycles. TMVP was trapped on the Mab 17V by injecting 30 ml of viral protein (0.26 mg/ml). Since no TMVP molecules were found to dissociate from the capture Mab 17V under mass transport conditions (Karlsson et al., 1994), the observed kinetics of Mab2 are unlikely to be influenced by the fact that TMVP was immobilized via a first Mab. After injection of TMVP, a flow of HBS buffer was maintained for 500 sec to remove free TMVP. Mabs at concentrations ranging from 5 to 35 nM in HBS were then allowed to interact with the trapped TMVP. Runs were performed at 257, at a flow rate of 5 ml/min, taking report points (for dRU/dt and R values) every 0.5 sec. After the association phase which lasted 10 min, HBS buffer was injected during the dissociation phase which extended over 3 min. The BIAevaluation software 2.0 was used to process dRU/dt and RU values that were measured automatically during the association and dissociation phases. The processing of these data consists in correcting the possible background and in plotting values according to different equations describing the mass action law. Our data were interpreted according to a nonlinear regression method (O’Shannessy et al., 1993; Malmqvist, 1993). According to this method dissociation rate constant (kd) and association rate constant (ka) values are obtained from the dissociation and association phase, respectively, for each binding experiment. Statistical calculations carried out with the BIAevaluation software 2.0 indicated that the simplest interaction model A / B B AB was suitable and plots were shown to fit this model. Each apparent rate constant corresponds to the mean of values calculated from 10 runs of interactions (two independent experiments for five different Mab concentrations). RESULTS Inhibition of TMV RNA translation by Mabs The inhibitory capacity of all Mabs whose epitope location had been established with BIAcore (Saunal and Van Regenmortel, 1995) was measured at different molar ratios of Mabs per virus particle, using slightly different conditions from the ones used previously (Saunal et al., 1993). Mabs 249P and 236P were found to possess less inhibitory capacity than reported previously, whereas the results with all other Mabs were the same. Three additional Mabs that had not been analyzed previously were included in the present study (167P, 188P and 175P). All antimetatope and antineotope Mabs of type A (Fig. 1) were found to inhibit virus disassembly at high molar ratio of 200 Mabs/virion and only small differences in inhibitory capacity were observed between the different

/ m4171$7543

10-11-95 12:11:42

viral

465

Mabs (Table 1). In contrast, the inhibitory efficiency of antimetatope and antineotope Mabs of type E was very variable, some being able to inhibit disassembly at a molar ratio of 50 or 100 Mabs/virion and others being totally devoid of any activity (Table 1). Kinetics measurements The binding affinity and kinetics of seven antimetatope Mabs of type E were measured in order to establish if there was a correlation between the affinity and the inhibitory capacities of the antibodies. These particular Mabs were analyzed because they exhibited the greatest variability in inhibition capacity. In view of the low binding stoichiometry of these Mabs to TMV particles, the kinetics of binding were measured using the viral protein as antigen. As shown in Table 1, the ka values of the different Mabs ranged from 1 1 105 to 5 1 105 M01 sec01 and the kd values ranged from 3 1 1004 to 10 1 1004 sec01. Deviations from the mean for apparent rate constants calculated in independent experiments were below 20%. The KA values ranged from about 1 1 108 to 11 1 108 M01. The small variations between the affinity of Mabs were not correlated with their inhibitory capacity. For instance, Mabs 16P and 25P which exhibited totally different inhibitory capacities were found to have the same apparent rate constants, whereas Mabs 25P and 151P which were both strong inhibitors exhibited a sixfold difference of affinity. Two-site binding assays with TMVP Since antineotope Mabs are unable to trap monomeric TMVP, only antimetatope Mabs were used in two-site binding assays with TMVP. Similar patterns of reactivity were obtained with antimetatope Mabs 79P, 236P, 17V, 19V, 5V, and 6V recognizing surface A and results will only be presented for two of these Mabs, i.e., 5V and 17V. A necessary control in this type of experiment consists in using the same antibody as both 1st and 2nd antibody in order to check that no oligomers are trapped by the Mab1. For the 11 Mabs listed in Table 2, this control gave a zero response with 2 Mabs (25P and 29V) and a negative RU response with the 9 other Mabs. A negative response indicates that TMVP actually dissociated from the capture Mabs. This phenomenon illustrates the reversible nature of the antigen–antibody interaction. After dissociating from the immobilized antibody, the TMVP molecules are able to bind to the same free antibody introduced in the flow cell as Mab2 and this prevents TMVP–Mab2 complexes from reassociating to the immobilized antibody. In many of the tests in which different antibodies were used as Mab1 and Mab2, zero or negative RU values were also obtained. When this result was observed independently of the binding order tested, it was assumed that the two Mabs recognized overlapping epitopes.

AP-Virology

466

SAUNAL AND VAN REGENMORTEL TABLE 1 Kinetic Binding Constants and Capacity of Mabs to TMV or TMVP to Inhibit Cotranslational Disassembly Inhibitory capacity (Molar ratio Mabs/virion) Mabs

Antimetatope type E

Antimetatope type A

Antineotope type E

Antineotope type A

Kinetic constants

200

100

50

25P 151P 181P 167P 188P 16P 47P

NDb ND ND /// /// 0 0

/// /// /// /// /// 0 0

/// /// /// / / 0 0

29V 79P 236P 17V 19V 5V 6V 175Pa

/// /// /// // // / / /

// // // / // / / 0

/ / / / / 0 0 0

4P 67P 42P

ND ND 0

/// /// 0

/// /// 0

18V 249P 107P 7V 253P

/// /// /// // /

// / / / /

/ / / / /

ka (105)

1.7 4.9 1.1 1.4 1.6 1.7 1.9

{ { { { { { {

0.2 0.4 0.1 0.2 0.3 0.2 0.1

kd (1004)

KA (108)

{ { { { { { {

1.8 { 0.8 10.2 { 1.2 1.5 { 0.6 4.1 { 1.0 2.2 { 1.0 1.7 { 0.8 4.4 { 0.6

9.5 4.8 7.3 3.4 7.2 9.9 4.3

0.2 0.2 0.2 0.3 0.2 0.2 0.2

Note. The inhibitory capacity of Mabs was assessed by examining radiolabeled TMV RNA translation products analyzed by electrophoresis and autoradiography as described by Saunal et al., 1993. ///, strong inhibition (the band corresponding to the 126-kDa viral protein was not detectable); //, noticeable inhibition (the 126-kDa band was barely detectable); /, slight inhibition (the intensity of the 126-kDa band was slightly weaker than the one obtained without Mab); 0, no effect (the intensity of the 126-kDa band was similar to that obtained in the control test). a Mab 175P is specific for surface A of TMVP although it was found to bind to the extremity of TMV particles, presumably to the last row of subunits. b ND, not determined.

Pairs of Mabs that bound to overlapping sites are listed in Table 3. In most of these cases the binding of Mab2 to a TMVP molecule that had dissociated from Mab1 prevented the reassociation of the antigen to Mab1 and this contributed to a negative RU value. When the capture Mab was immobilized by means of a first layer of rabbit anti-mouse globulins (Mabs 25P, 167P, 181P, and 188P), negative values may also partly be due to dissociation of the capture Mab from RAM. This was certainly the case, for instance, with Mab 25P, since the injection of Mab 167P as second antibody led to value of 0186 RU, whereas the amount of bound TMVP was only 129 RU. When combinations of Mabs led to a MR value (Mab2/ TMVP) higher than 0.1, it was assumed that the two Mabs bound to nonoverlapping sites even if the reaction in the reverse order was not possible. MR values lower than 0.1 were considered to be equivocal. Pairs of Mabs that bound unequivocally to either overlapping or nonoverlapping sites are listed in Table 3. A three-dimensional

/ m4171$7543

10-11-95 12:11:42

viral

drawing representing surfaces A and E of TMVP is presented in Fig. 3A. The results presented in Table 3 combined with previously obtained information regarding which surface, A or E, was recognized by each Mab (Saunal and Van Regenmortel, 1995) were used to establish the relative position of the epitopes on the map (Fig. 3B). The combining sites of the antibodies were assumed to cover a minimum surface of 600 A˚2 (Braden and Poljak, 1995) in a circular footprint. We also assumed that two Mabs are able to bind simultaneously to a TMVP molecule even if the corresponding epitopes have a common border. A minimal distance separating two epitopes was thus not considered necessary to permit simultaneous binding of two Mabs. On the schematic epitope map presented in Fig. 3B, three clusters of nonoverlapping epitopes can be distinguished. The first one corresponds to epitopes recognized by Mabs of type A (5V, 17V, 175P). The second one, which adjoins face A, corresponds to epitopes on

AP-Virology

/ m4171$7543

10-11-95 12:11:42

viral

AP-Virology

125/099 132/11 123/095 111/9 129/0186 138/0100 128/061 127/437 138/441 114/491 122/370

1233/3155 1225/3179 1359/0138 1216/2052

25P

1193/0184 950/1229 1196/0378 848/1096 1225/0528 444/268 510/44

16P

920/1952 924/2126 915/068 916/1189

892/0104 1020/1760 922/0150 949/1637 664/1031 936/807 518/848

47P

840/3756 762/3834 847/2849 756/3167

781/477 798/417 775/76 789/0110 791/0119 790/098 814/36

151P

ND ND 110/379 115/495

125/054 170/102 122/053 149/051 129/055 131/056 130/10

167P

187/483 175/603 147/477 155/585

170/054 216/131 163/049 203/27 182/0108 192/061 175/09

181P

ND ND 75/258 76/327

82/049 130/75 80/048 120/041 100/043 110/047 87/022

188P

237/045 243/046 231/13 282/0200

278/958 519/1795 269/697 286/1358 NDa 266/1043 ND

5V

797/0373 820/0327 786/0351 851/0551

866/2729 928/3849 829/2163 891/3541 ND 608/2098 ND

17V

29V

520/08 574/06 472/12 758/013

840/56 1014/2309 683/04 904/2067 1045/3604 619/1678 850/2932

Type A

350/0239 322/0140 328/085 333/0115

304/1050 347/1499 297/770 408/1760 253/655 407/1403 367/951

175P

Note. The two values represent RU levels corresponding to bound TMVP/bound Mab2. Mabs covalently bound to the dextran matrix are 16P, 47P, 151P, 175P, 5V, 17V, and 29V. Mabs bound via a first layer of RAM are 25P, 167P, 181P, and 188P. a ND, not determined.

Type E 16P 25P 47P 151P 167P 181P 188P Type A 5V 17V 29V 175P

Mab2

Type E

Mab1

Results of Two-Site Binding Assays between Antimetatope Mabs and TMVP Measured in BIAcore

TABLE 2

EPITOPE MAPPING USING BIOSENSORS 467

468

SAUNAL AND VAN REGENMORTEL TABLE 3 Epitope Mapping Using Unambiguous Pairwise Interaction Data Mabs

TMVP surface

Overlapping Mabs

16P

E

47P, 167P, 188P, 29V.

25P

E

47P

E

16P, 29V.

151P 167P

E E

167P, 181P, 188P. 16P, 151P, 181P, 188P.

181P 188P

E E

151P, 167P, 188P. 16P, 151P, 167P, 181P.

5V

A

175P, 17V, 29V.

17V

A

175P, 5V, 29V.

29V

A

16P, 47P, 175P, 5V, 17V.

175P

A

5V, 17V, 29V.

Nonoverlapping Mabs 25P (0.15), 151P (0.15), 175P (0.2–0.4), 5V (0.3–0.4), 17V (0.3–0.4). 16P (0.15), 47P (0.2), 175P (0.4–0.5), 5V (0.4–0.5), 17V (0.4–0.5), 29V (0.4–0.3). 25P (0.2), 151P (0.2), 167P (0.18), 175P (0.2–0.3), 181P (0.1), 188P (0.2), 5V (0.24–0.3), 17V (0.25–0.3). 16P (0.15), 47P (0.2), 175P (0.5–0.5), 5V (0.5–0.6), 17V (0.5–0.6), 29V (0.4–0.3). 47P (0.18), 175P (0.5–0.3), 29V (0.4–0.4). 47P (0.1), 175P (0.4–0.4), 5V (0.3–0.5), 17V (0.4–0.4), 29V (0.4–0.3). 47P (0.2), 175P (0.5–0.3), 29V (0.4–0.4). 16P (0.4–0.3), 25P (0.5–0.4), 47P (0.3–0.25), 151P (0.6– 0.5), 181P (0.5–0.3). 16P (0.4–0.3), 25P (0.5–0.4), 47P (0.25–0.3), 151P (0.5– 0.6), 181P (0.4–0.4). 25P (0.3–0.4), 151P (0.3–0.4), 167P (0.4–0.4), 181P (0.3– 0.4), 188P (0.4–0.4). 16P (0.4–0.2), 25P (0.5–0.4), 47P (0.3–0.2), 151P (0.5–0.5), 167P (0.3–0.5), 181P (0.4–0.4), 188P (0.3–0.5).

Note. Values in parentheses indicate stoichiometry of binding [MR values corresponding to (RU Mab2/RU TMVP) 1 (MW TMVP/MW Mab)].

surface E recognized by Mabs 16P and 47P. The third one corresponds to epitopes recognized by Mabs 151P, 167P, 181P, and 188P. Mab 29V was found to overlap two of the antibody clusters recognizing surfaces A and E. Mab 29V was the only Mab specific for surface A that

FIG. 3. (A) Schematic model of surfaces A and E of TMVP. Dimensions were calculated from data described by Namba and Stubbs (1986) (the virus particle is 300 nm long, 18 nm in diameter with a central hole of 0.4 nm, and possesses 2130 identical protein subunits that form a helix of pitch 2.3 nm with 16 31 subunits in every turn). (B) Schematic epitope map constructed from pairwise interaction data (see Table 3). Each antibody-combining site was assumed to cover about 600 A˚2 in a circular footprint.

/ m4171$7543

10-11-95 12:11:42

viral

did not bind concurrently with some of the Mabs specific for surface E (16P or 47P). Mab 25P presents a special problem since when used as Mab2 in combination with Mabs 151P, 167P, 181P, or 188P, it produced an equivocal MR value of 0.06–0.07. As surface E is unlikely to be able to accommodate more than two Mabs and as Mab 25P bound clearly to a different epitope from those recognized by Mabs 16P and 47P, it was considered to overlap with Mabs 151P, 167P, 181P, and 188P. This location is also consistent with the negative RU values observed with three of these antibodies when Mab 25P was used as Mab1 (Table 2). An examination of the epitope map (Fig. 3B) reveals that Mabs devoid of inhibitory capacity (Mabs 16P, 47P) bound to the part of surface E that adjoins surface A while the inhibitory Mabs bound to the part of surface E closest to the central axis of polymerized TMVP. From the epitope map constructed on the basis of pairwise interaction data, it seems that three molecules of Mabs should be able to be accommodated simultaneously on surfaces A and E of each subunit. This was verified by conducting multiple site binding assays taking note of the proposed positions of nonoverlapping Mabs. The results of these assays showed that several combinations made it possible to accommodate three Mabs on the combined surfaces A and E of a subunit. For instance, TMVP trapped by Mab 47P was able to accommodate both Mab 17V and Mab 151P (results not shown). Nonreciprocal results were obtained only when two Mabs of type E were tested for concurrent binding on TMVP. In all these cases of nonreciprocity, a positive result was obtained when Mab1 was a noninhibitory Mab

AP-Virology

EPITOPE MAPPING USING BIOSENSORS

and Mab2 was an inhibitory one. For instance, the combination Mab 47P/TMVP/25P was found to be possible, whereas a negative result was obtained when the antibodies were tested in the reverse order. MR values corresponding to combinations that involved pairs of Mabs able to bind concurrently to surface E were low (0.15–0.2). Combinations involving Mabs that bound to the part of surface E closest to surface A (47P) with Mabs specific for face A (17V, 5V) were found to give MR values of about 0.25–0.3. Higher MR values (0.5–0.6) were obtained when Mabs that bound to the part of the surface E closest to the central axis of polymerized TMVP (25P, 151P, and 181P) were combined with Mabs specific for surface A (17V, 5V). MR values are thus clearly correlated with the distance between epitopes. DISCUSSION The aim of this study was to explain why some Mabs to TMVP had the capacity to block the in vitro uncoating of the virus and others had not. One possibility was that the binding affinity or kinetics of the Mabs was correlated with their inhibitory capacity but biosensor measurements showed that this was not the case. Epitope mapping by means of BIAcore was then undertaken to establish if there was a link between the location of epitopes on the viral subunit and the inhibitory capacity of the corresponding Mabs. In a previous study, anti-TMVP Mabs whose specificity had been determined by ELISA (Al Moudallal et al., 1985) were tested for their ability to inhibit TMV RNA translation (Saunal et al., 1993). Many antimetatope Mabs were found to possess inhibitory activity. In a subsequent study (Saunal and Van Regenmortel, 1995), the location of the epitopes recognized by 14 antiTMVP Mabs and 7 anti-TMV Mabs was further investigated using biosensor technology. Several of these Mabs were found to have a specificity (neotope or metatope) different from the one that was suggested by ELISA results. Not all anti-TMVP Mabs whose inhibition capacity had previously been measured (Saunal et al., 1993) were analyzed in BIAcore because insufficient quantities were secreted by the hybridomas. Which surface of the TMVP molecule is recognized by each Mab is indicated in Fig. 1. The inhibitory capacity of the various anti-TMVP and anti-TMV Mabs was more precisely measured at different molar ratios of Mabs/virion (Table 1). The results showed that half of the tested antimetatope and antineotope type E Mabs strongly inhibited disassembly at a molar ratio of 50 or 100 Mabs/virion while the other half inhibited weakly or not at all. In contrast all antimetatope and antineotope type A Mabs possessed a low inhibitory capacity. Two-site binding assays were performed with 7 antiTMV Mabs and 16 anti-TMVP Mabs in order to determine

/ m4171$7543

10-11-95 12:11:42

viral

469

the locations of the corresponding epitopes. In two-site binding assays, a monomeric antigen trapped by a first antibody is tested for its ability to accommodate a second antibody. The localization of epitopes in this type of blocking assay is based on the assumption that a Mab binding to a specific site hinders the attachment of another antibody to the same or an overlapping site. The resolution is of the order of the footprint of the antibody. In comparison with ELISA, the BIAcore technology provides several advantages for epitope mapping (Daiss and Scalice, 1994; Van Regenmortel et al., 1994; Saunal and Van Regenmortel, 1995). The stoichiometry of binding of the second antibody to the viral antigen can be expressed in terms of molar ratio of bound Mab2/TMVP. Since the amount of TMVP trapped by different Mabs varied significantly, the stoichiometry of binding rather than the absolute RU response corresponding to bound Mab2 was used to infer epitope location. When the MR value was higher than 0.1, it was assumed that the two Mabs bound to nonoverlapping epitopes. When the same antibody was used as first and second antibody a lowered RU response corresponding to the amount of dissociated TMVP that was removed from the matrix during Mab2 injection was obtained. When different Mabs2 were tested in conjunction with the same Mab1, various levels of negative RUs were obtained. For instance, the combination 16P/TMVP/167P led to a RU response of 0528 while a response of 0184 was obtained with the combination 16P/TMVP/16P (Table 2). This difference may be due to differences in the kd values of Mab 16P (1003 sec01) and Mab 167P (3 1 1004 sec01). Since Mab 167P dissociated more slowly from TMVP than Mab 16P, it was more effective in preventing the reassociation of dissociated TMVP to the immobilized antibody. Pairwise interaction data give information only on the relative positions of epitopes since it is difficult to determine the orientation and the distances separating two nonoverlapping epitopes. As a result epitope maps are usually interpreted as functional ‘‘surface-like’’ maps (Daiss and Scalice, 1994; Johne et al., 1993; Nice et al., 1993; Tosser et al., 1994). In our system, several factors made it possible to construct a schematic map depicting the actual physical location of the epitopes on the antigen surface. Use was made of the known specificity for surface A or E of each Mab. The small size of the protein together with the observation that one Mab specific for surface A (29V) interfered with two Mabs specific for surface E (16P and 47P) allowed the unambiguous positioning of Mabs on the TMVP surface. Mab 175P recognized a particular type of epitope that seemed to be located on surface E since the maximum molar ratio of bound Mab 175P/TMV was about 10. Results of two-site binding assays indicated that Mab 175P overlapped all Mabs reacting with surface A but failed to interfere with any of the Mabs reacting with surface E (Table 3). These results suggest that Mab 175P binds

AP-Virology

470

SAUNAL AND VAN REGENMORTEL

to an epitope located on surface A that is accessible only on the last row of subunits in the polymerized particle, presumably because of steric hindrance arising from the next row of subunits along the length of the particle. The epitope map showed that overlapping conformational antigenic sites formed a continuum on the surface of the TMVP molecule. Mabs that were unable to block virus disassembly were found to bind to the part of surface E that adjoins surface A. In contrast, inhibitory Mabs were found to bind to the region of surface E closest to the central axis of polymerized TMVP which is known to interact with the viral RNA. Since the complete 5* leader of TMV RNA was shown to be uncoated before the initiation of cotranslational virus disassembly by ribosomes (Mundry et al., 1991), it seems that inhibitory Mabs act by preventing ribosomes from translocating on the viral RNA and dislodging the viral subunits. Our mapping results demonstrate the specificity of the recognition involved in this type of viral neutralization since epitopes corresponding to inhibitory and noninhibitory Mabs of type E were located very close to each other. Since inhibitory Mabs probably prevent the ribosomes from interacting with the virus, this mode of action resembles the neutralization mechanism operating when antibodies to animal viruses prevent viruses from interacting with cellular receptors (Dimmock, 1993). Mabs that bound to the entire length of the viral particle (surface A) were all found to possess a low inhibitory capacity. It is likely that these Mabs inhibited disassembly by a mechanism different from the one that operates when Mabs react with the extremity of the virus rod (surface E). Mabs specific for surface A may prevent disassembly of TMV by bridging viral subunits as in the case of poliovirus (Wetz, 1993). Since many of the subunits on the viral surface would have to be linked in order to prevent ribosomes from dislodging the last row of subunits presenting surface E, this may explain the high concentration of Mabs of type A that is required to observe inhibition of virus disassembly. Nonreciprocal interactions have been observed by many authors in competitive binding experiments performed with BIAcore (Daiss and Scalice, 1994; Johne et al., 1993; Karim et al., 1994) or in ELISA (Brown et al., 1990; Soos et al., 1992). In some cases these results could be explained by large differences in antibody affinity when the Mabs recognized overlapping sites. When the interaction involved a pair of Mabs that recognized distinct epitopes, it was suggested that the steric hindrance could occur at certain angles of attachment of the Mabs (Brown et al., 1990) especially if the binding sites were adjacent (Yewdell and Gerhard, 1981). Another interpretation is that the binding of the first Mab induces a conformational change which prevents it from being recognized by the second Mab. According to Karim et al. (1994), this could be due to modifications of the microenvironment after fixation of the first Mab on the BIAcore matrix.

/ m4171$7543

10-11-95 12:11:42

viral

The binding of two nonoverlapping Mabs on TMVP led to MR values ranging from 0.1 to 0.6. These MR values were correlated with the degree of proximity of epitopes since Mabs that bound to adjacent epitopes led to the lowest MR values. In contrast, the low MR values observed by Daiss and Scalice (1994) were not dependent on the close proximity of the epitopes and these authors suggested that the binding of a second antibody is an all or none phenomenon. They suggested that low stoichiometry of binding could arise from bivalent binding of Mabs to two antigen molecules or from incomplete saturation of the antigen because of too low affinity or too low concentration of Mabs. In some cases, neither an increase in concentration or volume injected nor the use of Fab fragments changed the result, and other factors such as steric hindrance within the matrix were then invoked. In our study, we used purified Mabs at high concentration (150 mg/ml) and all antimetatope Mabs had similar, high affinity constants (Table 1). In addition we chose a long time of injection in order to achieve maximum saturation of the epitopes. The correlation we observed between low MR values and proximity of epitopes may be explained by the steric hindrance that is likely to occur when two Mabs bind to epitopes located very close to each other. TMVP may be trapped by an antibody according to various angles of attachment and only particular angles may allow the Mab2 to bind concurrently to the same TMVP molecule. It is likely that this correlation between low MR values and proximity of epitopes was not found by Daiss and Scalice (1994) because the Mabs they used did not recognize epitopes located close together on the same surface of the antigen molecule. REFERENCES Al Moudallal, Z., Briand, J. P., and Van Regenmortel, M. H. V. (1982). Monoclonal antibodies as probes of the antigenic structure of tobacco mosaic virus. EMBO J. 1, 1005–1010. Al Moudallal, Z., Briand, J. P., and Van Regenmortel, M. H. V. (1985). A major part of the polypeptide chain of tobacco mosaic virus protein is antigenic. EMBO J. 4, 1231–1235. Benjamini, E. (1977). Immunochemistry of the tobacco mosaic virus protein. In ‘‘Immunochemistry of Proteins’’ (M. Z. Atassi, Ed.), pp. 265– 310. Plenum, New York. Braden, B. C., and Poljak, R. J. (1995). Structural features of the reactions between antibodies and protein antigens. FASEB J. 9, 9–16. Brown, L. E., Murray, J. M., White, D. O., and Jackson, D. C. (1990). An analysis of the properties of monoclonal antibodies directed to epitopes on influenza virus hemagglutinin. Arch. Virol. 114, 1–26. Daiss, J. L., and Scalice, E. R. (1994). Epitope mapping on BIAcore: Theoretical and practical considerations. Methods: Companion Methods Enzymol. 6, 143–156. Dimmock, N. J. (1993). ‘‘Neutralization of Animal Viruses.’’ SpringerVerlag, Berlin. Dore, I., Ruhlmann, P., Oudet, P., Cahoon, M., Caspar, D. L. D., and Van Regenmortel, M. H. V. (1990). Polarity of binding of monoclonal antibodies to tobacco mosaic virus rods and stacked disks. Virology 176, 25–29. Dore, I., Weiss, E., Altschuh, D., and Van Regenmortel, M. H. V. (1988). Visualization by electron microscopy of the location of tobacco mosaic virus epitopes reacting with monoclonal antibodies in enzyme immunoassay. Virology 162, 279–289.

AP-Virology

EPITOPE MAPPING USING BIOSENSORS Fa¨gerstam, L. G., Frostell, A., Karlsson, R., Kullman, M., Larsson, A., Malmqvist, M., and Butt, H. (1990). Detection of antigen–antibody interactions by surface plasmon resonance. Application to epitope mapping. J. Mol. Recognit. 3, 208–214. Johne, B., Gadnell, M., and Hansen, K. (1993). Epitope mapping and binding kinetics of monoclonal antibodies studied by real time biospecific interaction analysis using surface plasmon resonance. J. Immunol. Methods 160, 191–198. Karim, B., Beliard, R., Huart, J. J., and Bourel, D. (1994). Four epitopes on tumor necrosis factor-alpha defined by murine anti-tumor necrosis factor-alpha monoclonal antibodies. Immunol. Lett. 41, 139–145. Karlsson, R., Roos, H., Fa¨gerstam, L., and Persson, B. (1994). Kinetic and concentration analysis using BIA technology. Methods: Companion Methods Enzymol. 6, 99–110. Lo¨fas, S., and Johnsson, B. J. (1990). A novel hydrogel matrix on gold surfaces in surface plasmon resonance sensors for fast and efficient covalent immobilization of ligands. J. Chem. Soc. Chem. Commun. 21, 1526–1528. Malmqvist, M. (1993). Surface plasmon resonance for detection and measurement of antibody–antigen affinity and kinetics. Curr. Opinion Immunol. 5, 282–286. Mundry, K. W., Watkins, P. A., Ashfield, T., Plaskitt, K. A., Eisele, W. S., and Wilson, T. M. (1991). Complete uncoating of the 5* leader sequence of tobacco mosaic virus RNA occurs rapidly and is required to initiate cotranslational virus disassembly in vitro. J. Gen. Virol. 72, 769–777. Namba, K., and Stubbs, G. (1986). Structure of tobacco mosaic virus at 3.6 A˚ resolution: Implications for assembly. Science 231, 1401–1406. Nice, E., Layton, J., Fabri, L., Hellman, U., Engstrom, A., Persson, B., and Burgess, A. W. (1993). Mapping of the antibody- and receptorbinding domains of granulocyte colony-stimulating factor using an optical biosensor. Comparison with enzyme-linked immunosorbent assay competition studies. J. Chromatogr. 646, 159–168. O’Shannessy, D. J., Brigham-Burke, M., Soneson, K. K., Hensley, P., and Brooks, I. (1993). Determination of rate and equilibrium binding constants for macromolecular interactions using surface plasmon resonance: Use of nonlinear least-squares analysis methods. Anal. Biochem. 212, 457–468. Saunal, H., and Van Regenmortel, M. H. V. (1995). Mapping of viral

/ m4171$7543

10-11-95 12:11:42

viral

471

conformational epitopes using biosensor measurements. J. Immunol. Methods 183, 33–41. Saunal, H., Witz, J., and Van Regenmortel, M. H. V. (1993). Inhibition of in vitro cotranslational disassembly of tobacco mosaic virus by monoclonal antibodies to the viral coat protein. J. Gen. Virol. 74, 897– 900. Shaw, J. G., Plaskitt, K. A., and Wilson, T. M. A. (1986). Evidence that tobacco mosaic virus particles disassemble cotranslationally in vivo. Virology 148, 326–336. Soos, M. A., Field, C. E., Lammers, R., Ullrich, A., Zhang, B., Roth, R. A., Andersen, A. S., Kjeldsen, T., and Siddle, K. (1992). A panel of monoclonal antibodies for the type I insulin-like growth factor receptor. Epitope mapping, effects on ligand binding, and biological activity. J. Biol. Chem. 267, 12955–12963. Tosser, G., Delaunay, T., Kohli, E., Grosclaude, J., Pothier, P., and Cohen, J. (1994). Topology of bovine rotavirus (RF strain) VP6 epitopes by real-time biospecific interaction analysis. Virology 204, 8–16. Trifilieff, E., Dubs, M. C., and Van Regenmortel, M. H. V. (1991). Antigenic cross-reactivity potential of synthetic peptides immobilized on polyethylene rods. Mol. Immunol. 28, 889–896. Van Regenmortel, M. H. V. (1966). Plant virus serology. Adv. Virus Res. 12, 207–271. Van Regenmortel, M. H. V. (1986). Tobacco mosaic virus. Antigenic structure. In ‘‘The Plant Viruses. The Rod-Shaped Plant Viruses’’ (M. H. V. Van Regenmortel and H. Fraenkel-Conrat, Eds.), pp. 79– 104. Plenum, New York. Van Regenmortel, M. H. V. (1990). Structure of viral B-cell epitopes. Res. Microbiol. 141, 747–756. Van Regenmortel, M. H. V., Altschuh, D., Pellequer, J. L., Richalet-Secordel, P., Saunal, H., Wiley, J. A., and Zeder-Lutz, G. (1994). Analysis of viral antigens using biosensor technology. Methods: Companion Methods Enzymol. 6, 177–187. Wetz, K. (1993). Attachment of neutralizing antibodies stabilizes the capsid of poliovirus against uncoating. Virology 192, 465–472. Wilson, T. M. A. (1984). Cotranslational disassembly of tobacco mosaic virus in vitro. Virology 137, 255–265. Wilson, T. M. A., and Shaw, J. G. (1985). Does TMV uncoat cotranslationally in vivo? Trends Biochem. Sci. 10, 57–60. Yewdell, J. W., and Gerhard, W. (1981). Antigenic characterization of viruses by monoclonal antibodies. Annu. Rev. Microbiol. 35, 185– 206.

AP-Virology