Introduction
Much of what researchers know about antibody protection against HIV comes from studies using passively administered broadly neutralizing human monoclonal antibodies or monospecific neutralizing polyclonal antibodies in animal challenge models5, 6, 7, 8, 9, 10, 11, including intravenous (i.v.), vaginal and rectal challenge in macaques. The hallmark of most of these studies is that protection, in the form of sterilizing immunity, is achieved at relatively high serum neutralization titers corresponding to high antibody concentrations. The most quantitative of these studies suggest that sterilizing immunity requires serum antibody concentrations at least two orders of magnitude greater than in vitro neutralizing concentrations10, 11. However, this estimate is quite approximate and dependent upon, among other parameters, the neutralization assay used. Even so, the data have convinced many researchers that achieving sterilizing immunity via antibodies alone is extremely challenging, and a more realistic goal for vaccine-induced antibodies has been blunting infection and relying on vaccine-induced cellular immunity to clear, or, failing that, control, infection. However, as noted above, a limitation of macaque protection studies is the use of high viral challenge doses to ensure that all control macaques become infected with a single challenge. Yet it is well established that the average probabilities for heterosexual transmission in human exposures are low and dependent upon the viral burden in the donor and susceptibility factors associated with the donor and recipient, such as the presence of sexually transmitted diseases. Transmission frequencies on the order of 1 per 1,000 coital acts have been reported in chronic infection of the donor2, 3, 4, 12, increasing by about an order of magnitude in acute infection2, 3, 12, 13. The amount of virus contained in a typical macaque challenge, albeit estimated by quantitative PCR rather than infectivity, is much higher than what would be found, for example, in the semen ejaculate of an acutely infected man12, 13, 14, 15. Indeed, viral inoculums average 5 
105 copies per ejaculate, with a reported maximum of about 2 107 copies12, whereas we measured the inoculum of 300 median tissue culture infectious doses (300TCID50) of SHIVSF162P3, as used in a typical high-dose macaque challenge experiment, to contain about 1 108 viral copies.
To investigate antibody protection against viral challenge doses that may better represent those encountered in human heterosexual exposure, we used a low-dose repeated mucosal challenge model14, 16 in which a low virus dose requires several challenges to infect untreated macaques but eventually infects all animals. With this model, we would expect to observe benefit provided by antibody if the number of challenges required for infection in treated macaques was greater than the number of challenges required for infection in controls.
The human monoclonal antibody b12 neutralizes a broad range of HIV isolates from a variety of clades17, 18 through recognition of a conserved epitope overlapping the CD4-binding site of gp120 (ref. 19). A high serum concentration of b12, corresponding to about 75-fold the concentration required to inhibit 90% of viral replication in vitro (IC90) in a peripheral blood mononuclear cell (PBMC) assay and 3,000-fold the IC50 in a pseudotyped virus assay, provided 90% protection against a high-dose vaginal challenge with SHIVSF162P320. In the same study, the importance of the interaction of b12 with Fc receptors for protection was established by comparison of b12 and engineered b12 variants20.
Here we explored the question of whether a relatively low b12 serum neutralizing antibody titer could provide benefit to macaques in the low-dose repeated challenge model and simultaneously compared protection by the effector function–deficient b12 variant LALA. On the basis of earlier studies14, 16, 21, we began the experiment with repeated vaginal challenges using an inoculum dose of 3TCID50 of SHIV162P3. With only a single macaque infected after 11 challenges, we increased the viral dose to 10TCID50. This dose corresponds to approximately 2.65 106 viral RNA (vRNA) copies, an amount somewhat larger than typically found in human semen during acute infection12, 13 but only a few percent (2–3%) of traditional high-dose challenges with SHIVSF162P320.
The study involved a total of 14 macaques, consisting of four isotype control–treated macaques, five macaques receiving wild-type b12 and five macaques receiving the LALA variant, which has similar neutralizing activity as b12 but does not mediate Fc effector functions20. To maintain serum levels of antibody, we treated macaques i.v. weekly with 1 mg of b12 or LALA per kg body weight, on the basis of previously reported half-lives20. This dose of b12 antibody is far less than the 25 mg per kg body weight dose that provided 90% protection against high-dose challenge with SHIVSF162P320. Further, 1 mg per kg body weight of b12 provided negligible protection against high-dose challenge with SHIVSF162P410. Intravaginal challenges were administered twice weekly, and blood was drawn regularly to monitor viral infection, passively transferred antibody levels and serum neutralizing activity. Supplementary Figure 1 details the entire treatment course for each animal and Supplementary Table 1 summarizes antibody treatments, number of viral challenges, day of viremia detection and day-of-peak viremia in plasma.
The results show overall that notably more challenges were required to infect b12-treated macaques compared to control macaques (Fig. 1), and they also suggest that somewhat fewer challenges may be required to infect LALA variant–treated macaques compared to wild-type b12-treated macaques. Macaques in the isotype control group became virus positive after a maximum of four challenges of 10TCID50 (Fig. 1 and Supplementary Table 1). A cumulative total number of ten challenges of 10TCID50 was sufficient to infect all four control macaques (Supplementary Table 1). In the b12-treated group, macaque BK10 was infected after six challenges of 3TCID50 (Supplementary Fig. 2). After we switched to the 10TCID50 challenge regime, three b12-treated macaques became virus positive after 6, 23 or 38 viral challenges, and one macaque (BF68) remained virus negative after 40 challenges (Fig. 1 and Supplementary Table 1). Thus three of these four b12-treated macaques were infected by a cumulative total of 107 challenges of 10TCID50. In the LALA-treated group, plasma virus was observed after 6, 8, 12, 17 or 23 viral challenges (Fig. 1 and Supplementary Table 1). The five LALA-treated macaques were thus infected by a cumulative total of 66 challenges of 10TCID50 virus. We suspended viral challenges and i.v. antibody treatments after positive detection of virus in plasma, but we monitored the course of infection for several weeks.
Figure 1: Viral loads in antibody-treated macaques during vaginal low-dose repeated challenge with SHIVSF162P3.
(a–c) Plasma viral loads assessed by the measurement of SIV viral RNA using a quantitative reverse-transcription PCR (QRT-PCR) assay. The quantity of SIV viral RNA genomic copies (vRNA copies per ml) is plotted against the total number of viral challenges at a dose of 10TCID50. All macaques in each antibody treatment group were treated with 1 mg per kg body weight and are shown separately. (a) Macaques treated with the isotype control antibody. (b) Macaques treated with b12. One b12-treated macaque (BF68) remained virus negative after 40 challenges. One b12-treated macaque (BK10) was infected after six challenges of 3TCID50 and was therefore not presented in these data (Supplementary Fig. 2). (c) Macaques treated with b12 LALA variant. Viral challenges and i.v. antibody treatments were suspended after positive detection of virus in plasma, but the course of infection was monitored for several weeks. The SIV viral RNA (vRNA) assay detection limit is 125 copies per ml (log 2.1). The number of challenges is depicted here; the time course of challenge was interrupted in some instances as depicted in detail in Supplementary Figure 1.
Full size image (48 KB)We investigated the magnitude of protection using three approaches. First, we used an adapted Kaplan-Meier analysis (Fig. 2) in which the percentage of macaques remaining uninfected is plotted against the number of 10TCID50 viral challenges. To prevent positive bias, the macaque (BK10), who was infected in the 3TCID50 challenge series, was included in the analysis as if infected by the first 10TCID50 challenge (Supplementary Fig. 2). The three survival curves are significantly different (P = 0.0377; Fig. 2). A comparison of the individual pairs of Kaplan-Meier curves reveals that LALA is significantly different from the control in preventing infection (P = 0.0027). Owing to the strong penalty incurred by including BK10 in the analysis (Fig. 2), the difference between b12 and control did not reach significance (P = 0.056). The same analysis excluding BK10 yielded a significant difference (P = 0.0058). The LALA and b12 groups did not differ significantly from each other (Fig. 2). Second, we calculated hazard ratios for b12- and LALA-treated macaques with a Cox proportional hazard model that estimates the relative risk of infection for each of the treatment groups versus the controls. We found that treatment with b12 reduced the infection risk at each challenge by a factor of 21 (Table 1a) and for LALA treatment by a factor of 10 (Table 1a). Third, we calculated the reduction in infection susceptibility as described previously22 by tallying the total number of 10TCID50 virus challenges required to infect all macaques within each group (within the limits of the experiment). b12-treated macaques (P = 0.0016) as well as LALA-treated macaques (P = 0.0145) became infected after a significantly larger number of challenges compared to the control group (Table 1b). Of note, this number is underestimated for b12 in this type of analysis, as one b12-treated macaque remained uninfected at the end of the experiment. Overall, our analyses suggest that there is a significant difference in the protection afforded by the repeated administration of 1 mg per kg body weight of both b12 antibody and LALA variant as compared to treatment with the isotype control antibody. The approximately twofold difference in b12 and LALA hazard ratios and the observation that b12-treated macaques resisted nearly twice as many challenges as LALA-treated macaques (104 versus 61) reflects the trend, previously described in a high-dose virus challenge, for the effector function–crippled LALA variant to be less effective in protection than the fully effector function–competent wild-type b12 antibody20. An analysis of peak viremias suggests a trend toward lower peak viremias in the b12-treated group compared to controls, although this difference does not achieve significance (Supplementary Fig. 3). However, there is a significant difference (P = 0.016), about two orders of magnitude, between peak viremias in the b12- and LALA-treated macaques, again consistent with an impact of effector function on antiviral activity.
Figure 2: Kaplan-Meier analysis and magnitude of protection by b12 and LALA treatment in low-dose (10TCID50) repeated challenge.
The percentage of macaques remaining uninfected is plotted against the number of 10TCID50 viral challenges. A single macaque (BK10; b12-treated) became infected during the initial repeat 3TCID50 challenge. To allow inclusion of this macaque in the analysis, we included as if it was infected in the first 10TCID50 challenge. The Kaplan-Meier survival curves are significantly different from each other (P = 0.0377; log-rank (Mantel-Cox) test).
Full size image (29 KB)
We determined antibody serum concentrations throughout the course of the experiment by ELISA (Supplementary Fig. 4). We found considerable variations in individual serum concentrations, but we did not find significant correlation between average concentration and the number of challenges to infection. Likewise, the appearance of infection did not correlate with the magnitude of the antibody concentration at the estimated time of infection (10–17 d before detection of virus). We assessed neutralizing antibody titers in sera in a pseudovirus assay, and they were as expected from previous studies10, 20, given the antibody concentrations measured by ELISA (Supplementary Table 2). Average b12 concentrations for challenges not resulting in infection were relatively low, about 40 
g ml-1, corresponding to an average serum IC50 titer of 1:200 in a pseudovirus assay (that is, serum at a 1 in 200 dilution produced 50% neutralization in the assay) and to an estimated serum IC90 titer of 1:5 in a PBMC assay (Table 2). Major histocompatibility complex (MHC) genotyping revealed that there was no apparent correlation with the allelic profiles of the macaques in this study that would account for any unusual ability to resist infection (Supplementary Table 3).
In summary, we have shown that neutralizing antibody can provide a clear benefit against repeated low-dose SHIV challenge in the macaque model at low serum antibody concentrations corresponding to modest neutralization titers. There is a concern that low-dose challenge models may be 'lowering the bar' too much in terms of the requirements for protection. In this context, we note that oral chemoprophylaxis is possibly less, and certainly not more, protective against SHIVSF162P3 challenge in the low-dose repeated challenge model, arguing that the model is not intrinsically and universally more susceptible to protective intervention21. If translated into protection against HIV infection in humans, the findings are a promising development for HIV vaccine design. Serum neutralizing antibody titers in the approximate range of 1:200 IC50 values in a pseudovirus assay, corresponding to about 1:5 IC90 values in a PBMC assay, increased the number of low-dose challenges necessary to achieve infection here by at least an order of magnitude. If vaccination in humans were to lead to a similar decrease of transmission rate, then one might expect a major impact on the pandemic. Neutralizing titers achieved in this study are near or below those described in the sera of a substantial proportion of HIV-infected donors against multiple isolates from various clades23, 24, 25, 26, 27, suggesting that such titers may be achieved with appropriate immunogens. Finally, the data further support the contribution of effector function in antibody resistance to HIV infection, underscoring the notion that the ability of an immunogen to elicit extra-neutralizing antibody activities in addition to neutralization should be assessed in vaccine evaluation.
Note: Supplementary information is available on the Nature Medicine website.
) that recognizes an epitope overlapping the CD4 binding site of gp120
