The Role of Antibody Concentration and Avidity in Antiviral Protection

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2024-2027
DOI: 10.1126/science.276.5321.2024


Neutralizing antibodies are necessary and sufficient for protection against infection with vesicular stomatitis virus (VSV). The in vitro neutralization capacities and in vivo protective capacities of a panel of immunoglobulin G monoclonal antibodies to the glycoprotein of VSV were evaluated. In vitro, neutralizing activity correlated with avidity and with neutralization rate constant, a measure of on-rate. However, in vivo, protection was independent of immunoglobulin subclass, avidity, neutralization rate constant, and in vitro neutralizing activity; above a minimal avidity threshold, protection depended simply on a minimum serum concentration. These two biologically defined thresholds of antibody specificity offer hope for the development of adoptive therapy with neutralizing antibodies.

Antibody responses against chemically defined haptens, proteins, and pathogens have been well characterized, and the properties of polyclonal sera and monoclonal antibodies (mAbs) specific for these antigens have been studied in detail in vitro. Increased avidities and on-rates of antibodies have been postulated to provide increased in vivo effectiveness and protection (1). However, such a correlation has only rarely been analyzed for antibodies specific for, and protective against, infectious agents in vivo. Low-avidity (105M–1) opsonizing antibodies can protect against bacteria (2), and some studies have correlated in vitro virus neutralization titers with in vivo protection (3), whereas others have found no such relation (4). Avidity, on-rate, neutralizing activity, or antibody concentration have not previously been analyzed with respect to protective activity in vivo. We used a panel of mAbs (5-7) and polyclonal antibodies derived from high-titer secondary and hyperimmune responses to test whether characteristics of antibodies in vitro can predict protective efficiency in vivo—that is, whether increased avidity of immunoglobulin G (IgG) provides protection at lower serum concentrations.

VSV is a rhabdovirus closely related to rabies virus. It is highly neurotropic and may cause neurological disease and death in mice. Recovery of mice from primary infections or resistance against reinfection depends on neutralizing IgG antibody responses; CD8+ T cells are not involved, whereas mice lacking B cells always die (8, 9). The surface envelope of VSV contains ∼1200 identical glycoprotein molecules that form a regular and densely ordered pattern of spike tips; these tips are the only sites accessible to neutralizing antibodies (10). Neutralization of rhabdoviruses is mediated by the prevention of docking of the virus to cellular receptors. This requires a minimum of 200 to 500 IgG molecules bound per virion (11). The Fc portions of antibodies are not crucial for antiviral protection in vivo or in vitro (8, 12).

We previously described a set of virus-neutralizing mAbs derived from mice immunized with VSV serotype Indiana (VSV-IND) (6, 7). Virtually all of a collection of 62 mAbs that neutralize VSV bind to a single antigenic site on VSV-G comprising three overlapping subsites with avidities ranging from 2 × 107 M–1to 9 × 109 M–1 (average avidity 2 × 109 M–1). This panel of defined neutralizing antibodies was a useful tool to determine which of the known parameters—avidity, neutralization rate constant, specific in vitro neutralizing capacity, or serum concentration—was critical for antiviral protection in vivo. Avidities, neutralization rate constants [a direct measure of the on-rate (13)], and capacities to neutralize VSV in vitro are shown in Table 1 for the mAbs analyzed in this study. The in vitro neutralizing capacity correlated closely with avidity and with neutralization rate constant (Fig.1, A and B). Intravenous VSV infection of SCID (severe combined immunodeficiency disease) mice, which lack B and T cells, reproducibly leads to a lethal encephalitis; this lethal infection can be prevented by passive immunization with intact neutralizing antibodies of various Ig classes and subclasses, or with antibody fragments (9, 12).

Figure 1

Correlation of in vitro and in vivo parameters of mAbs. (A and B) Avidity (A) and neutralization rate constant (B) of mAbs, correlated with their in vitro neutralizing capacity. (C to E) Avidity (C), neutralization rate constant (D), and in vitro neutralizing capacity (E) of the same mAbs did not correlate with their in vivo protective concentration (see Table 1). Linear regression revealed correlation coefficients r of 0.86 (A), 0.93 (B), and <0.4 [(C) to (E)].

Table 1

Protective capacity of mAbs against VSV-IND. For details of the in vivo protection assay, see (19). The serum antibody concentration required for protection of 50% of passively immunized SCID mice (50% protective concentration) was identified by graphical extrapolation as illustrated in Fig. 2A, lower left panel, and division by the diffusion volume of a mouse (3 ml). For details of the in vitro analysis of antibodies, see (13). ND, could not be determined.

View this table:

SCID mice were treated with graded doses of mAbs. The measurable neutralizing antibody titer in serum revealed a calculated diffusion volume of ∼2 to 3 ml for a mouse weighing 25 g (9). Five hours later, when a steady state of antibody concentration had been reached, mice were intravenously infected with 108plaque-forming units (PFU) of VSV. Four days later, mice were killed and VSV titers in the brain were assessed. Surviving SCID mice that lacked detectable virus titers in the brain (<103 PFU per brain) were scored as protected. Representative experiments are shown in Fig. 2A. Using the 3-ml diffusion volume of mice, we determined the serum concentration needed for protection of 50% of the mice (Table 1). The minimal antibody concentration in serum necessary for in vivo protection was largely independent of subclass and of in vitro characteristics of the antibodies, namely, avidity, neutralization rate constant, and neutralizing capacity determined in balanced salt solution (Fig. 1, C and E). However, a minimal range of antibody concentration in serum of 0.3 to 7 μg/ml was needed for mAbs to be protective in vivo against 108 PFU of VSV-IND. As expected, protection was specific because a polyclonal serum against the New Jersey serotype of VSV (VSV-NJ) was not protective (Table2).

Figure 2

Protective capacity of mAbs in mice. (A) SCID mice were reconstituted with graded doses of antibodies; 5 hours later, neutralizing titers were determined from 1:40 prediluted and 1:2 serially diluted sera (upper panel). Mice were intravenously infected with 108PFU of VSV-IND, and 4 days later brains of surviving mice were assessed for the presence of virus (19). Mice without detectable virus were scored as protected (lower panel). One example of graphical extrapolation of the 50% protective antibody concentration is shown in the lower left panel (dotted arrow). (B) IFNαβR−/− mice were reconstituted with graded amounts of mAbs VI22 and VI41 (25, 2.5, or 0.25 μg), and neutralizing titers were determined from 1:40 prediluted and 1:2 serially diluted serum 5 hours later (upper panel). Mice were intravenously infected with 104 PFU of VSV-IND, and 3 days later the number of surviving mice was scored (lower panel) (16).

Table 2

Protective capacity of polyclonal sera against VSV-IND and VSV-NJ.

View this table:

One mAb (VI42) with an avidity of 2 × 107M–1 was not able to protect mice against intravenous VSV infection, even at doses of up to 100 μg per mouse; this suggested that antibodies must have a minimal avidity of >2 × 107 M–1 for effectiveness at the antibody doses tested. To analyze this avidity threshold further, we selected a VSV-IND escape mutant in vitro that was not neutralized by mAbs VI22 and VI41; this mutant was denoted VSV-TF (14). The panel of VSV-IND–specific antibodies was then tested for binding to this mutant virus by enzyme-linked immunosorbent assay (ELISA), and low-avidity antibodies were selected (Table 3). As in the first series of experiments, SCID mice were reconstituted with graded amounts of antibody and infected with VSV-IND or VSV-TF. None of the antibodies with avidities of ≤107 M–1, given at up to 100 μg per mouse, were protective after infection of mice with 108 PFU of VSV-TF. One mAb (VI32) with an avidity of 5 × 107 M–1 was protective against VSV-TF only at the very high dose of 100 μg/ml, confirming the range of the avidity threshold of ∼2 × 107 to 5 × 107 M–1 necessary for protection. Previous experiments documented in vivo protection by recombinant single-chain Fv antibody fragments and showed that complement and Fc receptors were not essential for mice to survive infection with VSV (12). Our findings add support to these results, because saturation of unoccupied Fc receptors of SCID mice by antibodies with normal mouse serum before the experiment did not change the protective capacity of the specific mAbs (15).

Table 3

Protective capacity of mAbs against the virus variant VSV-TF. For details of the generation of VSV-TF, see (14). ND, not determined.

View this table:

In normal mice, VSV does not replicate outside neurons. However, in mice lacking a functional interferon α/β system (IFNαβR–/–), a low dose of VSV causes a generalized infection and virus replication in all tissues examined, leading to death within 2 to 3 days (12, 16). Therefore, IFNαβR–/– mice offered a sensitive system to assess in vivo protection against a low dose of VSV. IFNαβR–/– mice, which have B and T cells and normal Ig concentrations in serum, were reconstituted with graded amounts of mAbs VI22 (9 × 109 M–1) and VI41 (2 × 108 M–1), yielding either high or barely measurable neutralizing titers in sera (Fig. 2B). These mice were then intravenously infected with 104 PFU of VSV-IND, and survival was scored (Fig. 2B) (16). Although reconstitution with the low-avidity mAb VI41 led to a low neutralizing titer in serum (Fig. 2B), mice were protected, as were those that received the high-avidity mAb (VI22) and exhibited high neutralizing titers in serum (Fig. 2B).

An IgG concentration in serum of 0.3 to 7 μg/ml is equivalent to ∼1012 to 2.6 × 1013 molecules/ml, and 108 PFU of VSV-IND is equivalent to 1 × 109 to 3 × 109 virus particles displaying ∼1.3 × 1012 to 4 × 1012 antigenic determinants, which indicates an almost equimolar ratio of antibody and antigenic determinants. It was therefore possible that in the protection assays, the effective antiviral antibody concentration in serum was reduced below the protective concentration immediately after the virus inoculum by absorption of the antibody by the virus. To exclude this possibility, we reconstituted mice with protective and subprotective concentrations of mAb VI10 (10 μg/ml and 1 μg/ml, respectively), infected them with VSV-IND, and determined virus neutralizing titers in serum before, 2 days after, and 4 days after infection. At a protective concentration of antibody, injection of 108 PFU of VSV-IND did not affect neutralizing titers in the sera of SCID mice; at subprotective concentrations of antibody, neutralizing titers decreased slightly during the course of the infection, by 1.8 dilution steps on average (Fig.3). Similarly, neutralizing titers in the sera of IFNαβR–/– mice reconstituted with protective doses of antibody did not decrease upon injection of 104PFU of VSV. Because of the generalized nature of the VSV infection, in IFNαβR–/– mice with subprotective concentrations of antibody, neutralizing titers in serum decreased by 3.5 dilution steps within 4 days (Fig. 3). However, no drop in titers could be observed when sera were analyzed 12 hours after infection. Thus, the lethal outcome of a VSV infection was not the result of extensive antibody consumption immediately after injection of VSV.

Figure 3

Antibody consumption after VSV infection of passively immunized mice. SCID mice (open symbols) and IFNαβR−/− mice (solid symbols) were injected intraperitoneally with protective (squares, 30 μg per mouse) and subprotective doses (circles, 3 μg per mouse) of VSV neutralizing mAb VI10. SCID mice and IFNαβR−/− mice were infected with 108 and 104 PFU of VSV-IND, respectively. Mice were bled before, 2 days after, and 4 days after infection. Neutralizing titers were determined in a standard neutralization assay in serum. Titers of uninfected SCID mice exhibiting protective antibody concentrations are shown as controls (open diamonds).

These data confirm that the avidity and neutralization rate constant of antibodies correlate well with the neutralizing activity assessed in balanced salt solution in vitro (Fig. 1, A and B). However, no obvious correlation was found between minimal protective serum concentration in vivo and the in vitro parameters of avidity, neutralization rate constant, or neutralizing capacity (Fig. 1, C to E). This discrepancy cannot be attributed to IgG subclass differences because the eight IgG2a antibodies analyzed alone showed the same correlation (Fig. 1). Above a minimal avidity threshold of ∼2 × 107 to 5 × 107 M–1, the antibody concentration (>1 to 10 μg/ml) seems to limit the protective effectiveness of antibodies in vivo. This avidity threshold corresponds to a dissociation constant of ∼2 × 10–8 to 5 × 10–8 M, which is close to the measured protective IgG concentration of 0.3 to 7 μg/ml (2 μg/ml ≈ 10–8 M IgG). These IgG antibody concentrations must be reached within 5 to 6 days to protect mice against VSV (9). Antibodies below this avidity threshold seem to require very high in vivo concentrations for effectiveness (>30 μg/ml); such concentrations probably cannot be reached within the critical few days available for mice to survive. It is nevertheless interesting that the behavior of these low-avidity (but not of high-avidity) antibodies is apparently correctly described by the law of mass action, that is, low avidity can be compensated for by high concentrations. Collectively, our data reveal concrete threshold numbers for a protective immunity unit for antibodies (17).

It had been predicted that higher avidity of antibodies, with consequently lower concentrations necessary for protection, should improve antibody-dependent memory and might render it more economical (1). Our results suggest that maturation of avidity beyond the threshold of 107 to 108M–1 may not improve protective capacity. Thus, there is a discrepancy between parameters defining the in vitro and in vivo activities of neutralizing antibodies. This discrepancy cannot be readily explained by uncertainties regarding the mechanism by which antibodies neutralize VSV (18). It may be that the physicochemical properties of mouse serum and tissues in vivo are drastically different from the buffered saline conditions usually used in vitro. In particular, the kinetics of virus neutralization may be considerably slower in vivo than in vitro, because of complex diffusion kinetics of antibodies in blood and tissue lesions. Because the host-virus interaction is essentially a nonequilibrium system, these complex kinetics and their changes may drastically alter the net outcome of infection.

  • * Present address: Department for Medical Biophysics, Ontario Cancer Institute, Toronto, Ontario M5G 2M9, Canada.

  • Present address: Department of Biomedicine, University of Pisa, I-S6127 Pisa, Italy.

  • Present address: Department of Pathology and Microbiology, University of Bristol, Bristol BS8 1TD, UK.

  • § To whom correspondence should be addressed. E-mail: RZI{at}


View Abstract

Navigate This Article