Structure of a Neutralizing Antibody Bound Monovalently to Human Rhinovirus 2

Abstract

The structure of a complex between human rhinovirus 2 (HRV2) and the Fab fragment of neutralizing monoclonal antibody (MAb) 3B10 has been determined to 25-Å resolution by cryoelectron microscopy and three-dimensional reconstruction techniques. The footprint of 3B10 on HRV2 is very similar to that of neutralizing MAb 8F5, which binds bivalently across the icosahedral twofold axis. However, the 3B10 Fab fragment (Fab-3B10) is bound in an orientation, inclined at approximately 45° to the surface of the virus capsid, which is compatible only with monovalent binding of the antibody. The canyon around the fivefold axis is not directly obstructed by the bound Fab. The X-ray structures of a closely related HRV (HRV1A) and a Fab fragment were fitted to the density maps of the HRV2–Fab-3B10 complex obtained by cryoelectron microscope techniques. The footprint of 3B10 on the viral surface is largely on VP2 but also covers the VP3 loop centered on residue 3064 and the VP1 loop centered on residue 1267. MAb 3B10 can interact directly with VP2 residue 2164, the site of an escape mutation on VP2, and with VP1 residues 1264 to 1267, the site of a deletion escape mutation. Deletion of these residues shortens the VP1 loop, moving it away from the MAb binding site. All structural and biochemical evidence indicates that MAb 3B10 binds to a conformation epitope on HRV2.

Picornaviruses are small single-stranded RNA viruses, 300 Å in diameter, some of which exhibit great antigenic variation (26). Human rhinoviruses, (HRVs), medically important members of the picornavirus family, are the major cause of the common cold. Their capsid is composed of 60 copies each of four viral coat proteins, VP1, VP2, VP3, and VP4, on a T=1 icosahedral lattice (25). The HRVs are classified into a major group and a minor group based on their specificities for cell receptors: intercellular adhesion molecule 1 for the major group (see, for example, reference 11) and members of the low-density lipoprotein receptor family for the minor group (15). The structures of several HRVs representing both groups are known (e.g., HRV14 [25], HRV1A [16], HRV16 [12, 21], and HRV3 [39]). The study of escape mutants to neutralization by monoclonal antibodies (MAbs) has led to the definition of four neutralizing immunogenic (NIm) sites (IA, IB, II, and III) for the major-group virus HRV14 (29) and three such sites (A, B, and C) for the minor-group virus HRV2 (1). Reviews of picornavirus antigenicity and its relation to virus structure are found in references 6 and 18, respectively.

Antibodies play an important role in combating viral infection, and a number of mechanisms for antibody-mediated neutralization of viruses have been proposed. It is possible that each antibody is capable of invoking more than one mechanism; however, the relative importance of these mechanisms in vitro, and more importantly in vivo, remains uncertain. The proposed mechanisms include viral aggregation as a result of the interlinking of particles (3), inhibition of virus receptor binding, and inhibition of virus uncoating (20). Antibodies also mark invading particles for destruction by the complement or other pathways of the immune system. Viral aggregation and inhibition of receptor binding can be detected biochemically in vitro and have been shown to occur for selected neutralizing MAbs. Observations of large pI changes upon antibody binding have led to the hypothesis that antibody-mediated modification of the virus capsid may be involved (8); however, the lack of correlation between pI change and neutralizing strength (4) and the absence of any change in the structure of HRV14 upon binding of a strongly neutralizing MAb, as seen in the X-ray structure of the HRV14-Fab complex (33), argue against neutralization induced by capsid modification upon antibody binding. In the crystallographic structures of Fabs complexed with peptides that mimic the viral epitope for a poliovirus (38) and HRV2 (13, 36), the conformation of the peptide differs from its homolog on the virus. Taken at face value, these results imply that antibody binding induces change in capsid conformation (38); however, since the inherent flexibility of a short peptide allows it to adopt different conformations to suit its environment, further confirmation is required. At present there is insufficient information to say to what extent modification of the virus capsid plays a role in antibody-induced virus neutralization.

A precise knowledge of the molecular details of virus-antibody interactions should contribute to our understanding of the mechanisms of antibody-mediated neutralization. The study of such large molecular complexes is not always feasible by X-ray crystallography alone; however, a combination of data from cryoelectron microscopy and X-ray crystallography is currently proving very fruitful: the picornaviruses, namely HRVs (e.g., HRV14 [31–33]; HRV2 [13], and foot-and-mouth disease virus [FMDV] [14]), are receiving particular attention. The structural study of a selected range of antibodies with different neutralization characteristics and NIm sites is a step toward understanding antibody-mediated virus neutralization. In this paper, we describe structural studies of the complex of the minor-group HRV2 and neutralizing MAb 3B10 directed against the NIm B site. We employed cryoelectron microscopy and three-dimensional reconstruction techniques combined with X-ray crystallographic data. The X-ray structures of the closely related HRV1A (16), which has 73% amino acid sequence similarity in its capsid proteins, and a Fab fragment were fitted to the density maps of the HRV2–Fab-3B10 complex obtained by cryoelectron microscope techniques. The atomic structure of HRV2, predicted by comparison with the known HRV1A structure, was also placed in the cryoelectron microscopy density map. The structure of neutralizing MAb 8F5 bound bivalently to NIm site B of HRV2 (13) is compared with that of MAb 3B10, which binds monovalently to the same NIm site.

Preparation and purification of HRV2, MAb 3B10, and Fab-3B10.

HRV2 was prepared essentially as described by Skern et al. (30) with the modifications detailed by Hewat and Blaas (13). MAb 3B10 was purified from tissue culture supernatants of hybridomas grown in roller bottles in RPMI containing 5% immunoglobulin-free serum (Gibco-BRL) by standard methods involving polyethylene glycol precipitation and ion-exchange chromatography. Fab fragments were prepared by cleavage with papain followed by Mono-Q column chromatography. MAb 3B10 was raised against the native HRV2 virus particle.

Neutralization experiments.

Five hundred 50% tissue culture infectious doses (TCID₅₀) of HRV2 were incubated for 1 h in 100 μl of purified antibodies at the final concentrations indicated in Fig. 4. The solutions were then transferred to 96-well plates with subconfluent HeLa-OHIO cells (Flow Laboratories). Incubation was at 35°C for 3 days, whereupon cells were stained with amido black and washed with phosphate-buffered saline. The stain was solubilized in 1 M NaOH, and the A₅₃₀ was determined for each well with a microplate reader.

FIG. 4.

Inhibition of virus infection by 3B10 IgG and Fab. Infection of HeLa cells by HRV2 is inhibited by 3B10 in a concentration-dependent manner. HeLa cell monolayers in 96-well plates were challenged with 500 TCID₅₀ of HRV2 in the presence of antibody at the concentrations indicated. Cells surviving after 3 days of incubation were stained, and the A₅₃₀ was measured in a microplate reader. The A₅₃₀ of the intact monolayer (not infected) was set to 100% protection, whereas the A₅₃₀ of infected monolayers in the absence of antibody was set to 0% protection. Error bars represent standard deviations of the means of 10 experiments carried out in parallel.

Preparation of HRV2-3B10 complexes.

Complexes were prepared as described previously (13). HRV2 (25 μg) and Fab-3B10 were incubated at a molar ratio of 1:400 in 50 μl of incubation buffer (50 mM Tris-HCl, 150 mM NaCl, 2 mM CaCl₂, 30 mM MgCl₂ [pH 7.4]) for 1 h at room temperature. Excess Fab was removed by passage through an exclusion column (Sephacryl S300 Spun; Pharmacia) which had been equilibrated with incubation buffer. Assuming no loss of HRV2, the complex was estimated to have a total protein concentration of 1 μg/μl. The HRV2–MAb-3B10 complex was prepared as described for the HRV2–Fab-3B10 complex but with a molar ratio of 1:200. Attempts to remove excess MAbs by passage through a Spun column resulted in loss of the specimen because of aggregation.

Preparation of frozen hydrated specimens.

Frozen hydrated specimens were prepared on holey carbon grids as described previously (13). The holey carbon film supported on 400 mesh grids was not glow discharged before use. Samples of virus suspension (4 μl) were applied to grids, blotted immediately with filter paper for 1 to 2 s, and rapidly plunged into liquid ethane cooled by nitrogen gas at −175°C. Specimens were observed at a temperature of approximately either −180°C with a Gatan 626-DH Cryoholder or −175°C with an Oxford CT3500 Cryoholder in a Phillips CM200 operating at 200 kV. Images were obtained under low-dose conditions (<10e/Ų) at a nominal magnification of ×27,500 at 1.8 and 2.5 μm underfocus.

Image analysis.

Preliminary selection of defocus pairs of micrographs, digitalization, and preparation of virus particle images for analysis were performed as described previously (28). The pixel size of 12.5 μm on the micrographs corresponds to a nominal pixel size of 4.5 Å/pixel for the specimen. Images were reinterpolated to a pixel size of 1.5 times the original so that 128 by 128 files could be used. Further image analysis was performed on a DEC Alpha by using modified versions of the MRC icosahedral programs supplied by S. D. Fuller (9, 10). The orientations and origins of each particle were first determined and refined by the method of common lines (5) for the higher defocus image. All subsequent refinement was performed with model-based programs (2), to determine the particle origin and orientation, and cross-common lines (Simplex program) to refine these parameters. The three-dimensional reconstructions were made by Fourier Bessel inversion. The reconstruction of the high defocus image was used as a starting model for the low defocus image, and several cycles of refinement were performed. The best reconstruction used 54 particles and included information to 25-Å resolution. The phase residual went to 90° at 24 Å⁻¹. All inverse eigenvalues were less than 1.0, and 99.8% were less than 0.1. Isosurface representations of the reconstructed density were visualized by Explorer on a Silicon Graphics computer.

Fitting the HRV and Fab X-ray structures to the cryoelectron microscope reconstructed density map.

The cryoelectron microscope reconstructed density map of the HRV2–Fab-3B10 complex was scaled to the X-ray data by comparing the HRV capsid density only. The spherically averaged density within a spherical shell from a radius of 115 to 145 Å was compared by cross-correlation. This gave a pixel size of 6.3 (± 0.05) Å/pixel. No correction for the effect of the contrast-transfer function on the cryoelectron microscopy reconstruction was made. Since the X-ray structure of Fab-3B10 has not yet been resolved, the structure of another Fab, Fab-SD6 (37), was employed. The criterion for the choice of Fab was that the elbow angle should be close to 180°, as seen in the three-dimensional reconstruction of 3B10. The X-ray structure of Fab-SD6 was then fitted by eye to the electron microscope density by using program “O” on a Silicon Graphics computer. The new coordinates of the Fab were output from “O”, and all other Fab fragments were generated by icosahedral symmetry operations. The structure of HRV2, as predicted by SwissMod with the HRV1A X-ray structure as a basis, was also placed in the cryoelectron microscopy density map.

Cryoelectron microscopy of HRV2–MAb-3B10 and HRV2–Fab-3B10 complexes.

Cryoelectron microscope images of HRV2–MAb-3B10 complexes revealed a very high degree of particle aggregation, in keeping with previous biochemical observations indicating that MAb 3B10 is bound monovalently to HRV2. These large three-dimensional aggregates are not suitable for image analysis. Cryoelectron microscope images of complexes between the Fab fragment of 3B10 and HRV2 show a homogeneous population of particles with a knobbly appearance and no appreciable aggregation (Fig. 1). It may be deduced from these images that the Fab fragments do not project radially from the capsid surface, in contrast to Fab-SD6 on FMDV, which does project radially (14). As in the native HRV2 specimen, a small proportion of the capsids were empty and they were also decorated with Fab fragments. While it has been shown that the aggregation of virus-monovalent MAb complexes can be practically eliminated (24), we chose to work exclusively on the virus-Fab complex since it facilitates reconstruction to a higher resolution.

FIG. 1.

Electron micrographs of frozen hydrated HRV2–Fab-3B10 complex at a defocus of −2.5 (A) and −1.8 μm (B). Bar, 1,000 Å.

Reconstructed density of the HRV2–Fab-3B10 complex.

Analysis of the orientations of the HRV2–Fab-3B10 complexes seen in images of untilted specimens revealed a highly preferred orientation along the fivefold axis. There is apparently a strong interaction of the complexes with the air-water interface. This does not allow a reconstruction with isotropic resolution; hence, images of specimens tilted by 10° were obtained. The three-dimensional reconstruction made from a zero-tilt image (not shown) was, however, remarkably similar to that made from a 10°-tilted image (Fig. 2) despite the missing information. Both reconstructions show HRV2 decorated with 60 bilobed Fab fragments which are tilted at an angle of approximately 45° to the viral surface. The major difference between the two reconstructions is the presence of unexpected density linking the Fabs around the threefold axis in the untilted case. In both reconstructions, the maximum density in the Fab is lower than in the viral capsid, indicating a Fab occupancy of 70 to 80%.

FIG. 2.

(A) Stereo view of the isosurface representation of the reconstructed HRV2–Fab-3B10 complex viewed down a twofold axis. HRV2 is shaded in grey (radius, <160 Å), and Fab-3B10 is shaded in blue (radius, >160 Å). Three separate Fab fragments are depicted in shades of magenta. Note that these are identical symmetry-related Fabs. Only the front half of the virus-Fab complex is shown. (B) Comparison of the reconstruction of HRV2–Fab-3B10 (blue) and HRV2-MAb 8F5 (yellow) (13) viewed down a twofold axis. (C and D) Thick sections from panel B cut parallel to the plane of the page. The density corresponding to the single-stranded RNA has been set to zero. A twofold axis, perpendicular to the plane of the page, lies at the center of each figure. (C) The differences in binding orientations of 3B10 and 8F5 are visible. (D) The footprints of MAbs 3B10 and 8F5 on HRV2 (arrowheads) are very similar. Icosahedral two-, three-, and fivefold axes are indicated. Bars, 50 Å.

The known surface topology of HRVs (13, 31, 32), i.e., a pentameric dome on each of the icosahedral fivefold axes surrounded by the “canyon” with a raised triangular plateau centered on each of the icosahedral threefold axes, allowed determination of the hand of the electron microscope reconstruction. Bound Fab-3B10 does not obstruct the canyon, which, by analogy with the major-group viruses, may contain the receptor binding site. Comparison of the three-dimensional reconstructions of the HRV2-MAb 8F5 (13) and HRV2–Fab-3B10 complexes (Fig. 2B, C, and D) shows that, to a very good first approximation, both MAbs 8F5 and 3B10 bind to the same epitope on HRV2 (NIm site B in HRV2, which is equivalent to NIm site II of HRV14 [29]); however, the Fab fragments are oriented very differently. The Fab fragments of 8F5 project almost radially from the viral surface, facing each other across the twofold axis and giving rise to bivalent binding of the MAb. In contrast, the Fab fragments of 3B10 tilt away from the twofold axis. Even taking into account the flexibility of the Fab elbow, it is not possible to join a 3B10-Fab to any of its neighbors. Thus, MAb 3B10 binds monovalently to HRV2. In order to transform the orientation of Fab-8F5 to that of Fab-3B10, the Fab must be rotated by approximately 150° about its long pseudo-twofold axis and then tilted towards the viral surface by 45°.

In both the HRV2-MAb 8F5 and HRV2–Fab-3B10 reconstructions, the viral surfaces are similar (Fig. 2B). However the canyon in the HRV2–Fab-3B10 reconstruction is slightly deeper (by a few angstroms only). We believe this difference is not significant at 25-Å resolution. (The receptor binding site on minor-group viruses is not known at present; while it is thought to differ from that of major-group viruses, it is probably also in the canyon area [7]).

Fit of X-ray structures of a Fab and HRV1A to the cryoelectron microscope reconstructed density map.

The atomic structure of a Fab usually fits almost equally well into a medium-resolution density map in two orientations related by a 180° rotation about the long (pseudodyad) Fab axis. This ambiguity can generally be lifted by observing the orientation of the Fab elbow angle. (The Fab elbow angle is defined as the angle between the two pseudodyad axes relating the heavy and light chains in the variable and constant modules). Unfortunately, the elbow angle of Fab-3B10 visible in the reconstructed maps is too close to 180° to be of use. However, at 25-Å resolution, the constant domain of the Fab has an asymmetric shape, which allows the ambiguity to be resolved (Fig. 2).

The X-ray structures of HRV1A and Fab-SD6 fit the reconstructed density map of the HRV2–Fab-3B10 complex very well (not shown). The structure of HRV2 predicted by SwissMod (22, 23) fits equally well (Fig. 3). Although positions of the important antigenic loops in this predicted structure cannot be considered very reliable, particularly in the hypervariable regions, they at least have the correct residues and numbers of residues in these loops.

FIG. 3.

(A) Stereo view of the footprint of 3B10 on HRV2. Residues 2164 (P [in magenta, top left]) and 1264 to 1267 (KIED [in black, bottom left]) are sites of escape mutations. (Note that the first digit denotes the viral capsid protein, here VP2 and VP1, respectively). An icosahedral twofold axis is just above the top of the figure. (B) View of part of the HRV2 surface indicating the region shown in panel A. Icosahedral two-, three-, and fivefold axes are marked. (C) Visual fit of the X-ray crystallographic structure of Fab-SD6 (37) into the cryoelectron microscope reconstruction of the HRV2–Fab-3B10 complex. The section shown was chosen to pass through the two regions of close contact between the Fab fragments. The C_α backbones of three Fab fragments are shown in shades of magenta and correspond to the similarly shaded Fab fragments in Fig. 2A. The side chains of neighboring Fab fragments approach to within 5 Å (arrows). The approximate position of a twofold axis is indicated. In both panels A and B, the C_α backbones of one copy of VP1, VP2, and VP3 of HRV2 (SwissMod prediction) are depicted in blue, green, and red, respectively. The electron microscope map is displayed in dark blue.

The 3B10 footprint is largely on VP2 and includes the immunogenic loop centered on residue 2153. However, it also covers the VP3 loop centered on residue 3064 and the VP1 loop centered on residue 1267. MAb 3B10 can interact directly with VP2 residue 2164, the site of an escape mutation on the VP2 immunogenic loop. It probably also interacts with VP1 residues 1264 to 1267, the site of a deletion escape mutation (Fig. 3A and B; Table 1). Deletion of residues 1264 to 1267 shortens the VP1 loop, moving it away from the MAb binding site.

TABLE 1.

Mutations which lead to loss of neutralization or binding by MAbs 8F5 and 3B10

Escape from binding by:
8F5a		3B10b
Amino acid position	Mutation	Amino acid position	Mutation
2159	E→G	1266	E→K
2161	R→T	1264	De1264–1267c
2163	N→Y	2164	P→L
2163	N→S	2164	P→S
2164	P→S

^aData from reference 36. 

^bData from reference 28. The mutations which allow escape from 3B10 binding were produced not by challenge from 3B10 but by other MAbs. 

^cDe, deletion. 

In contrast to 8F5 (36), 3B10 does not bind to any viral protein on Western blots. Amino acid residues other than those from the VP2 loop thus contribute to the binding of 3B10. The existence of an escape mutant on VP1 also indicates a conformation epitope. In the absence of the HRV2 and Fab-3B10 structures at atomic resolution, a more detailed analysis of the Fab-3B10–HRV2 interface is not possible.

The neighboring Fab fragments lie very close to one another. The side chains approach to within 5 Å across the twofold axis and around the threefold axis (Fig. 3C). Thus, it is not surprising that these two points of near contact are not dissociated at 25-Å resolution. Whereas Fab fragments of 3B10 form a complex with the virus with an occupancy of about 80%, the Fab-8F5 complex shows less than 10% occupancy, indicating that in this case bivalent binding of 8F5 increases the avidity (13). For 3B10, the distance between the C-terminal C_α of the heavy chains in Fab fragments on either side of the twofold axis is 105 Å; it is 52 Å for neighboring Fab fragments around the threefold axis. Even taking into account the flexibility of the Fab elbow and hinge, the distance between the heavy-chain carboxy termini of neighboring Fab fragments is too great to allow bivalent binding of MAb 3B10.

Neutralization experiments.

Infection inhibition experiments showed that complete 3B10 immunoglobulin G (IgG) had a much stronger inhibition capacity than Fab fragments (Fig. 4). The combination of this low virus concentration (500 TCID₅₀/0.1 ml) with the large molar excess of antibodies (in the range of 10⁶:1 to 10⁷:1) reflects the in vivo situation. We consider 3B10 to be a very weakly neutralizing antibody, as defined by Mosser and colleagues (20). Also, the relatively low occupancy (70 to 80%) of Fab 3B10 measured in the reconstruction indicates low binding affinity.

Discussion of the binding of MAb 3B10 on HRV2 and comparison with other antipicornavirus MAbs.

MAb 3B10 binds monovalently to a conformation epitope on immunogenic site B (equivalent to NIm site II on HRV14), and the bound Fab does not obscure the canyon, although it is possible that the unbound Fab and the Fc fragment do cause some steric hindrance. The affinity of Fab-3B10 is high enough to allow formation of complexes with HRV2, with a Fab occupancy of about 80%.

MAb 8F5 binds bivalently to HRV2 on a contiguous epitope at immunogenic site B, and again the canyon is not obscured. The affinity of Fab-8F5 for HRV2 is so low that it forms a complex with only very low Fab occupancy. MAb 3B10 binds more strongly and is a more efficient neutralizer than 8F5, but neither MAb is a very strong neutralizing agent. Due to the topology of its binding site, MAb 8F5 may cause inhibition of virus uncoating.

While both MAbs 3B10 and 8F5 bind to the same immunogenic site, the large difference in binding geometry provides an insight into what determines monovalent or bivalent binding of a MAb. These observations support the idea (20) that it is not only the spanning distance but also the orientation which determines the nature of binding. It should be noted, however, that bivalent binding of a MAb to a virus can occur in the absence of a twofold axis. For example, bivalent binding across a threefold axis was seen in the case of a neutralizing MAb bound to a calicivirus (rabbit hemorrhagic disease virus) (35).

Both MAbs 3B10 and 8F5 bind to the NIm B site on HRV2, which does not overlap the putative receptor binding site; thus, it is interesting to compare our results with those for MAbs against a site which does overlap the receptor binding site (Table 2). Smith and colleagues (17, 31–34) have made extensive studies of three MAbs directed against the NIm 1A site of HRV14 with cryoelectron microscopy and three-dimensional reconstruction and with X-ray crystallography. Two of the MAbs (12-1A and 17-1A, which differ by only 5 amino acid residues) bind bivalently and are very strong neutralizers; the third MAb (1-1A) binds monovalently and is a weak neutralizer. MAbs 12-1A and 17-1A bind strongly to a very large area epitope, which involves framework residues of the heavy chain. It is also relevant to consider the very strongly neutralizing MAb SD6, which binds monovalently to the picornavirus FMDV-C (14). SD6 binds to a contiguous epitope on the long flexible GH loop of VP1, and its Fab forms a complex with 100% occupancy. Fab-SD6 neutralizes FMDV-C almost as well as IgG-SD6.

TABLE 2.

Characteristics of neutralizing MAbs against picornaviruses with known three-dimensional structures

Antibody (reference)	Antigen (epitope)	Receptor obstructed	Binding valence	Neutralization
17-1A (32)	HRV14 (1A)	Yes	Bi-	Strong
12-1A (34)	HRV14 (1A)	Yes	Bi-	Strong
1-1A (34)	HRV14 (1A)	Yes	Mono-	Weak
8F5 (13)	HRV2 (B)	No	Bi-	Weak
3B10	HRV2 (B)	No	Mono-	Weak
SD6 (14)	FMDV-C (GH loop)	Yes	Mono-	Strong

What conclusions can be drawn from these six picornavirus-MAb complexes for which the structures are known? In this rather limited collection of MAbs, there are examples of both strongly and weakly neutralizing MAbs which (i) obstruct receptor binding (e.g., 17-1A and 1-1A), (ii) bind monovalently to the virus (e.g., SD6 and 3B10), and (iii) bind bivalently to the virus (e.g., 17-1A and 8F5). The three strongly neutralizing MAbs, 17-1A, 12-1A, and SD6, each inhibit receptor binding and bind strongly to the virus (they each form a virus-Fab complex with 100% occupation, as seen by three-dimensional reconstruction). Thus, it appears that a MAb which binds strongly to the receptor binding site will be a strong neutralizer. The cooperativity of bivalent MAb binding can give an affinity constant of 100- to 1,000-fold higher than for the Fab alone; however, bivalent binding is neither necessary (e.g., SD6) nor sufficient (e.g., 8F5) to ensure strong neutralization. It is the affinity (or avidity) of the antibody which matters. Modification of the virus capsid upon antibody binding leading to inhibition of one or more steps in the infection cycle has been evoked, particularly in the case of polioviruses (8, 38), but there is no evidence for this in any of the six MAbs considered here; in fact, the X-ray structure of the HRV14–Fab-17-1A complex (33) shows no modification of the virus structure. MAbs with high affinity, irrespective of their mode of binding, will probably also be the most effective in vivo in marking the virus for destruction by other pathways of the immune system (19).