Relationships between Neutralization, Binding, and ADCC of Broadly Neutralizing Antibodies against Reservoir HIV

LETTER

HIV-specific antibodies (Abs) may contribute to the elimination of HIV reservoirs by binding to reactivated cells and targeting them for antibody-dependent cell-mediated cytotoxicity (ADCC) (1, 2). Broadly neutralizing antibodies (bNAbs) may be particularly efficacious, by mediating virus neutralization alongside ADCC for multiple HIV strains (in the context of untreated simian-human immunodeficiency virus [SHIV] infection, neutralization rather than ADCC appears to dominate the antiviral effect of bNAb infusion [3, 4]). Few studies have assessed the activities of bNAbs against reservoir-derived viruses. The relationships between neutralizing activity, ADCC function, and binding to reservoir virus in infected primary CD4⁺ T cells have not been comprehensively studied. In a study previously published by Ren et al. (5), we tested a panel of 14 bNAbs against 36 reservoir-derived viral isolates, demonstrating that susceptibility to neutralization by broadly neutralizing antibodies generally correlated with infected-cell binding for a panel of clade B HIV reactivated from latent reservoirs.

Our original study also tested ADCC with 2 viral isolates using 9 bNAbs and two types of effector cells, comprising primary NK cells and a haNK cell line, and demonstrated highly significant and direct correlations with infected-cell binding (5). Here, we applied the same methods to extend these results to all 36 viral isolates and the full panel of all 14 bNAbs, using haNK cells as effectors in in vitro assays. We report relatively broad and moderately potent ADCC of infected cells treated with CD4 binding site (CD4bs)-targeted bNAbs, as well as with the V3 glycan-targeted bNAbs PGT121 and 10-1074 (Fig. 1). The V1/V2 bNAb PG9 also exhibited substantial ADCC activity but with somewhat less coverage of viral isolates. The remaining bNAbs, including those targeting the membrane-proximal external region (MPER), exhibited sparse ADCC activity against this virus panel (Fig. 1).

FIG 1.

Breadths and potencies of ADCC with a panel of bNAbs against reactivated reservoir virus-infected cells. The heat map shows each antibody-dependent cellular cytotoxicity (ADCC) efficiency value determined by the use of percent ADCC values, which were calculated using the following formula: percent ADCC = {[(percent Gag⁺ cells among viable CD3⁺ cells under with-NK-cell no-Ab conditions) − (percent Gag⁺ cells among viable CD3⁺ cells under test conditions)]/(percent Gag⁺ CD3⁺ cells under with-NK cell no-Ab conditions)} × 100%. Negative values were set equal to zero. All bNAbs were used at 10 μg/ml, and the effectors were haNK cells. The heat map is scaled from green to red for values from 0% to 100%, respectively. pos, positive; ^a, the cutoff for a positive ADCC value was set at 24.1%, where the 10-1074-GASDALIE Ab showed no binding but an ADCC value.

Considering all the bNAbs together, we observed moderate correlations between ADCC and infected-cell binding (Fig. 2A, Spearman’s r = 0.49, P < 0.0001) and between neutralization 80% inhibitory concentration (IC₈₀) and ADCC (Fig. 3A, Spearman’s r = 0.46, P < 0.0001). At the level of individual bNAbs, 7/15 showed significant correlations between ADCC and infected-cell binding (Fig. 2B). Despite the overall significant correlation between ADCC and neutralization, only the 10-1074 Ab displayed a significant correlation on the individual bNAb level (Fig. 3B, Spearman’s r = 0.66, P < 0.0001), though several other bNAbs displayed trends. Grouped by binding site, each class of antibodies showed significant correlations between neutralization and ADCC, though these ranged from moderately strong for V3 glycan antibodies to marginal for MPER antibodies (Table 1). A similar gradient was observed in the multiway correlations between binding, ADCC, and neutralization (Table 2). However, since the strengths of these correlations tracked with overall potency, this gradient is likely a reflection of the relative presence or absence of a detectable ADCC signal. We therefore do not interpret this as evidence of fundamentally different relationships between these effector activities across antibody classes (6). An additional caveat is needed for the V3 glycan antibodies, since PGT121 and 10-1074 are clonal relatives derived from the same individual and thus are not independent representatives of this antibody family.

FIG 2.

Correlations between ADCC and bNAb binding to the virus-infected cells (matched). (A) Correlation for all antibodies tested together. (B) Correlations for bNAbs tested independently. The infected-cell binding data were completely matched to the ADCC condition, with the same infected cells assessed for both readouts following 7 h of culture with 10 μg/ml bNAbs. Each virus-bNAb combination is indicated by a symbol, and each color represents one study participant. Correlations were analyzed by determining the Spearman correlation coefficient (r), with statistical significance (P) values highlighted in red. MFI, median fluorescent intensity.

FIG 3.

Correlations between ADCC and virus neutralization (IC₈₀). (A) Correlation for all antibodies tested together. (B) Correlations for bNAbs tested independently. Each virus-bNAb combination is indicated by a symbol, and each color represents one study participant. Correlations were analyzed by determining the Spearman correlation coefficient (r), with statistical significance (P) values highlighted in red. Calculation of the neutralization IC₈₀ values in the 10-1074 GASDALIE graph used data from 10-1074 IgG1.

TABLE 1.

Spearman correlations between ADCC and neutralization (IC₈₀) by bNAb class^a

bNAb class	Spearman correlation coefficient	P value	n
CD4bs	0.285	0.001	141
V3 glycan	0.688	<0.001	106
V1/V2	0.201	0.038	106
MPER (gp41)	0.170	0.046	139

^aNote that 10-1074 GASDALIE data were excluded from analysis.

TABLE 2.

Multiway correlations between binding, ADCC, and neutralization (IC₈₀) by bNAb class^a

bNAb class	“Taylor” correlation coefficient
CD4bs	0.405
V3 glycan	0.608
V1/V2	0.204
MPER (gp41)	0.204

^aData were calculated by the method of Taylor (9). Note that 10-1074 GASDALIE data were excluded from the analysis.

Our results confirm previous reports indicating a correlation between infected-cell binding and ADCC (for bNAbs with a shared Fc domain) (5, 7, 8) and extend this to a broader panel of bNAbs and primary reservoir virus isolates. While the results are less robust on the individual bNAb level, we also interpret our overall neutralization-versus-ADCC data as demonstrating a general correlation between these important effector functions. To reiterate the note of caution in our original report, however, the directionality of the relationship is critical. While we showed that for antibodies known to have neutralizing activity (bNAbs), this activity was reasonably predictive of ADCC activity, the converse is known not to be true (refer to reference 5).

Together with our initial report, we provide a uniquely comprehensive analysis of the three-way relationships between neutralization, infected-cell binding, and ADCC using virus derived from primary CD4⁺ T cells and bona fide infected CD4⁺ T cells as targets. We feel that this will provide a valuable resource for clinical trials which aim to use bNAbs to achieve remission or cure of HIV infection.