Polyfunctionality and breadth of HIV-1 antibodies are associated with delayed disease progression

Abstract

HIV-1 infection leads to chronic disease requiring life-long treatment and therefore alternative therapeutics, a cure and/or a protective vaccine are needed. Antibody-mediated effector functions could have a role in the fight against HIV-1. However, the properties underlying the potential beneficial effects of antibodies during HIV-1 infection are poorly understood. To identify a specific profile of antibody features associated with delayed disease progression, we studied antibody polyfunctionality during untreated HIV-1 infection in the well-documented Amsterdam Cohort Studies. Serum samples were analyzed from untreated individuals with HIV-1 at approximately 6 months (n = 166) and 3 years (n = 382) post-seroconversion (post-SC). A Luminex antibody Fc array was used to profile 15 different Fc features for serum antibodies against 20 different HIV-1 envelope glycoprotein antigens and the resulting data was also compared with data on neutralization breadth. We found that high HIV-1 specific IgG1 levels and low IgG2 and IgG4 levels at 3 years post-SC were associated with delayed disease progression. Moreover, delayed disease progression was associated with a broad and polyfunctional antibody response. Specifically, the capacity to interact with all Fc γ receptors (FcγRs) and C1q, and in particular with FcγRIIa, correlated positively with delayed disease progression. There were strong correlations between antibody Fc features and neutralization breadth and several antibody features that were associated with delayed disease progression were also associated with the development of broad and potent antibody neutralization. In summary, we identified a strong association between broad, polyfunctional antibodies and delayed disease progression. These findings contribute new information for the fight against HIV-1, especially for new antibody-based therapy and cure strategies.

Fig 1. Antibody features over time and associations with HIV-1 disease progression.

(A) 92BR020-specific antibodies targeting SOSIP and gp120 antigens in serum of IgG type, IgG1-4 subtype and interacting with FcγRIIa (high affinity allelic variant H131) (associated with antibody-dependent cellular phagocytosis), FcγRIIIa (high affinity allelic variant V158) (associated with antibody-dependent cellular cytotoxicity) and C1q protein (associated with complement activation and complement-dependent cytotoxicity). Antibody responses are compared between the time point 6 months after seroconversion (SC) and 3 years after SC using a Wilcoxon matched-pairs signed rank test. Only individuals with samples at both time points (n = 166) are included in this subfigure. The data is expressed as median fluorescence intensity (MFI) measured by Luminex assay. The eight features shown here are those we were most interested in based on literature, seven additional features are shown in S2A Fig. (B) Spearman correlation analysis between the antibody features analyzed in this study at 3 years post-SC and three clinical variables related to disease progression. Positive correlations between antibody features and time to AIDS or CD4+ T cell count at setpoint are related to delayed disease progression whereas positive correlations with viral load at setpoint are related to faster disease progression. Color indicates the rho Spearman correlation coefficient and significant results are indicated by asterisks. The results were corrected for multiple comparisons using the Bonferroni-Holm method. Antibody features shown in this figure are for responses specific for gp120 and SOSIP proteins of subtype B strain 92BR020. Abbreviations: AIDS: acquired immunodeficiency syndrome; cp/mL: viral copies per milliliter. * = P<0.05, ** = P<0.01, *** = P<0.001, **** = P<0.0001.

Fig 2. Polyfunctionality of antibody Fc features.

Polyfunctionality was assessed by calculation of a polyfunctionality score at 3 years post-seroconversion (SC). We only assessed activating FcγR (FcγRI, FcγRIIa, FcγRIIIa, FcγRIIIb) and C1q interaction, and for each Fc feature we determined for each participant if there was a response that was higher than the median response of all responders in the cohort. The resulting scores (amount of features above the responder median, between 0 and 5, per person) were plotted for the four groups based on time between SC and acquired immunodeficiency syndrome (AIDS) (less than 3 years (n = 28), between 3 and 7 years (n = 147), between 7 and 11 years (n = 66) and more than 11 years (n = 71)) with one plot for 92BR020 gp120-specific responses and one plot for 92BR020 SOSIP-specific responses. In addition, a polyfunctionality index was calculated for each group as a quantitative measure of polyfunctionality. This is a value between 0 and 100 calculated by a weighted addition of the percentage of individuals in each category based on the number of features, derived from Larsen et al. [54], see methods). This index is shown above each bar. Polyfunctionality of antibodies against these two strains at 6 months post-SC are shown in S4 Fig.

Fig 3. Breadth of gp120-specific antibody interaction with FcγRs and C1q.

Breadth is compared between four categories based on time between seroconversion and acquired immunodeficiency syndrome (AIDS) diagnosis: less than 3 years (n = 28), between 3 and 7 years (n = 147), between 7 and 11 years (n = 66) and more than 11 years (n = 71). Bar charts show the percentage of individuals within each group with antibody interaction with the plotted FcγR or C1q above the responder cut-off for the color-coded amount of strains. Above each bar the breadth index for that group is shown as a quantitative measure of breadth. This is a value between 0 and 100 calculated by weighted addition of the percentage of individuals in each category based on the number of features, derived from Larsen et al. [54] (see methods). Breadth of gp120-specific antibody interaction at 6 months post-SC and breadth of SOSIP-specific antibody interaction at 3 years and 6 months post-SC are shown in S6 Fig.

Fig 4. Epitope specificity of antibodies interacting with FcγRs and C1q in different groups based on disease progression.

(A) Spearman correlation analysis between the antibody features analyzed in this study and time between seroconversion (SC) and AIDS diagnosis. The color indicates the rho Spearman correlation coefficient and significant results are indicated by asterisks. The results were corrected for multiple comparisons using the Bonferroni-Holm method. (B) Levels of antibodies were normalized by Z-score across the eight protein antigens to allow comparison of the specificity of the response of different types of antibodies and antibody interactions. This allows comparison of the different graphs even though the original magnitude of the different antibody types and functions was quite different. Therefore, the scale of the graph is not relevant, but the pattern and the ratio between different specificities can be compared between plots. In addition, Kruskall-Wallis tests were used to compare the responses to each protein antigen separately and the results are shown on the figure as stars. The plots are color-coded for four categories based on time between SC and AIDS diagnosis: less than 3 years (n = 28), between 3 and 7 years (n = 147), between 7 and 11 years (n = 66) and more than 11 years (n = 71). Protein antigens shown in this figure are gp120 and SOSIP proteins of subtype B strain JRCSF, V1/V2, V3 and gp41 epitope constructs were also based on the JRCSF sequence. Constructs comprising the CD4 binding site (CD4bs), the MPER and a covalently linked fusion of gp120 and CD4 (gp120-CD4) were previously described (see methods). * = P<0.05, ** = P<0.01, *** = P<0.001, **** = P<0.0001.

Fig 5. Further identification and confirmation of associations with HIV-1 disease progression.

A partial least squares regression analysis was performed with time between seroconversion (SC) and acquired immunodeficiency syndrome (AIDS) diagnosis in years as the independent variable. Twenty-five Fc features were included in the model as selected by elastic net feature selection. Data for the feature selection included antibody responses for all gp120, SOSIP and epitope antigens including antibody types, subtypes, interaction with all FcγRs and C1q, breadth scores, polyfunctionality scores, neutralization titers and neutralization breadth. Further properties of the model are shown in S8 Fig. (A) The model comprised two latent variables (LVs) and they are plotted against each other. (B) LV1 is displayed in a column plot with four groups based on time to AIDS (less than 3 years (n = 28), between 3 and 7 years (n = 147), between 7 and 11 years (n = 66) and more than 11 years (n = 71)) compared using a Kruskal-Wallis test followed by a Dunn’s multiple comparison test. (C) Loading and variable importance in projection (VIP) scores of all 25 included features are shown, with variables with a VIP score higher than 1 shown in pink (also indicated by the grey vertical line) and the remaining variables shown in dark green. (D) Spearman correlation analysis between antibody-dependent cellular phagocytosis (ADCP) of 92BR020 gp120-conjugated beads by THP-1 cells at 3 years post-SC and time between SC and AIDS. This analysis was done on a subset of 140 participants with a similar distribution of time between SC and AIDS as the full cohort. r = rho Spearman correlation coefficient, ** = P <0.01, *** = P<0.001, **** = P<0.0001.

Fig 6. Associations between antibody Fc features and broad HIV-1 neutralization activity.

(A) Spearman correlation analysis between the antibody features analyzed in this study at 6 months post-seroconversion (SC) and geometric mean (geomean) half-maximal infective dilution (ID₅₀) pseudovirus neutralization titers at 3 years post-SC (n = 166). Antibody features shown in this panel are for responses specific for gp120 and SOSIP proteins of subtype B strain 92BR020. Correlations with antibody Fc features at 3 years post-SC are shown in S9 Fig. The color indicates the rho Spearman correlation coefficient and significant results are indicated by asterisks. The results were corrected for multiple comparisons using the Bonferroni-Holm method. (B) Spearman correlation analysis between the antibody features analyzed in this study at 6 months post-SC and geomean ID₅₀ pseudovirus neutralization titers at 3 years post-SC (n = 166). Antibody features shown in this figure are for responses specific for gp120 and SOSIP proteins of subtype B strain JRCSF, V1/V2, V3 and gp41 epitope constructs were also based on the JRCSF sequence. Constructs comprising the CD4 binding site (CD4bs), the membrane-proximal external region (MPER) and a covalently linked fusion of gp120 and CD4 (gp120-CD4) were previously described (see methods). The color indicates the rho Spearman correlation coefficient and significant results are indicated by asterisks. The results were corrected for multiple comparisons using the Bonferroni-Holm method. (C) Partial least squares regression analysis geomean ID₅₀ pseudovirus neutralization titer as independent variable. Eleven Fc features were included in the model as selected by elastic net feature selection with alpha 0.5 and 16-fold cross-validation. The model comprised two latent variables (LVs), had a R² of calibration of 0.389, root mean square error of calibration of 0.267 and root mean square error of cross-validation of 0.336. (D) Loading and variable importance in projection (VIP) scores of all 11 included features, with variables with a VIP score higher than 1 shown in light blue (also indicated by the grey vertical line) and the remaining variables shown in dark blue. * = P<0.05, ** = P<0.01, *** = P<0.001, **** = P<0.0001.