Vaccine-Induced Antibodies Mediate Higher Antibody-Dependent Cellular Cytotoxicity After Interleukin-15 Pretreatment of Natural Killer Effector Cells

The secondary analyses for correlates of risk of infection in the RV144 HIV-1 vaccine trial implicated vaccine-induced antibody-dependent cellular cytotoxicity (ADCC) responses in the observed protection, highlighting the importance of assessing such responses in ongoing and future HIV-1 vaccine trials. However, in vitro assays that detect ADCC activity in plasma from HIV-1 infected seropositive individuals are not always effective at detecting ADCC activity in plasma from HIV-1 vaccine recipients. In vivo, ADCC-mediating antibodies must operate at the site of infection, where effector cells are recruited and activated by a local milieu of chemokines and cytokines. Based on previous findings that interleukin 15 (IL-15) secretion increases during acute HIV-1 infection and enhances NK cell-mediated cytotoxicity, we hypothesized that IL-15 pretreatment of NK effector cells could be used to improve killing of infected cells by vaccine-induced antibodies capable of mediating ADCC. Using the HIV-1 infectious molecular clone (IMC)-infected target cell assay along with plasma samples from HIV-1 vaccine recipients, we found that IL-15 treatment of effector cells improved the ability of the vaccine-induced antibodies to recruit effector cells for ADCC. Through immunophenotyping experiments, we showed that this improved killing was likely due to IL-15 mediated activation of NK effector cells and higher intracellular levels of perforin and granzyme B in the IL-15 pretreated NK cells. We also found that using a 4-fold dilution series of plasma and subtraction of pre-vaccination responses resulted in lowest response rates among placebo recipients and significant separation between treatment groups. This represents the first attempt to utilize IL-15-treated effector cells and optimized analytical approaches to improve the detection of HIV-1 vaccine-induced ADCC responses and will inform analyses of future HIV vaccine clinical trials.

equal to 10% in one of the first two dilutions. The positivity definition was derived to primarily to control the false positive rate among placebo recipients, while accounting for the fact that response curves in vaccine recipients are often non-monotonic, due to a prozone effect. Together, this suggested that if there was detectable ADCC activity, we would expect to see some signal in the most concentrated dilutions, even if the peak activity is detected at lower concentrations. As responses in negative control wells did not exceed 10% in any sample assayed with IL-15 pretreated NK cells ( Figure 2 in the primary manuscript), we set the positivity threshold to 10%.
In Figure S1 we plot baseline-subtracted ADCC dilution curves by the dilution at which the peak response is observed and by vaccination group (note this is the IL-15-pretreated data in Figure 6 of the primary manuscript). Among placebo recipients, ADCC responses rarely exceeded 10%.
In the one curve that does exceed 10%, it does so at the 1:800 dilution and reaches its peak at the 1:3200 dilution, however the first two dilutions have no detectable ADCC activity. Figure S1. Baseline-subtracted %killing ADCC dilution curves plotted by the dilution at which the peak response is observed and by vaccination group for all IMCs combined.

Partial area under the dilution curve (pAUC)
To determine which dilutions should be included in the computation of the partial area under the baseline-subtracted dilution curve (pAUC), we compare the dilution-specific distribution of baseline-subtracted ADCC responses by vaccine group in Figure Figure 1 The dilution curves shown in Figure 1 in the main manuscript were repeated in two additional independent experiments, in which we tested more dilutions so that plasma from the seropositive sample would reach endpoint. Thus, we have results from 3 independent experiments for the first six plasma dilutions, and two independent experiments for the last two plasma dilutions. Each independent experiment included two technical replicates; average % killing is reported. The results are summarized in Figure S3, where we plot the average % killing across all three of the independent experiments and include error bars for 1 standard deviation. Pilot study to determine the dilution scheme for detecting vaccine-induced ADCC activity

Additional independent replicates of the dilution curves shown in
The purpose of the pilot study was to determine what dilution scheme could capture complete dilution curves for vaccine recipients while potentially yielding better dilution curves for placebo recipients.
Using samples from 15 HVTN 100 participants (n= 6 placebo recipients and n = 9 vaccine recipients), 2-, 3-, and 4-fold dilution curves were plotted for ADCC responses measured in the presence of IL-15. The average baseline-subtracted percent killing in responders and nonresponders for each of the four sets of assay conditions is shown in Figure S4. For samples from vaccine responders, the 2-and 3-fold dilution curves were not sufficient to capture the entire response curve and compute a titer. In fact, the titer could be determined for only 20% of the 5 positive two-fold dilution curves, and 60% of the 5 positive curves from the 3-fold dilution. In contrast, a titer was able to be determined for all positive 4-fold dilution curves. From this small study, we concluded that the 2-and 3-fold dilution series would not be appropriate for capturing complete dilution curves for HVTN studies. Figure S4. Comparison of average baseline-subtracted ADCC response dilution curves for HVTN 100 responders and non-responders. The assay used target cells infected with the TV1 IMC. All tests were performed in the presence of IL-15.