Repeated exposure to heterologous hepatitis C viruses associates with enhanced neutralizing antibody breadth and potency

Abstract

A prophylactic hepatitis C virus (HCV) vaccine that elicits neutralizing antibodies could be key to HCV eradication. However, the genetic and antigenic properties of HCV envelope (E1E2) proteins capable of inducing anti-HCV broadly neutralizing antibodies (bNAbs) in humans have not been defined. Here, we investigated the development of bNAbs in longitudinal plasma of HCV-infected persons with persistent infection or spontaneous clearance of multiple reinfections. By measuring plasma antibody neutralization of a heterologous virus panel, we found that the breadth and potency of the antibody response increased upon exposure to multiple genetically distinct infections and with longer duration of viremia. Greater genetic divergence between infecting strains was not associated with enhanced neutralizing breadth. Rather, repeated exposure to antigenically related, antibody-sensitive E1E2s was associated with potent bNAb induction. These data reveal that a prime-boost vaccine strategy with genetically distinct, antibody-sensitive viruses is a promising approach to inducing potent bNAbs in humans.

Keywords: Adaptive immunity; Antigen; Hepatitis; Immunology; Virology.

Figure 1. Representative graphs demonstrating the history of viremia of study participants.

Representative viral load graphs of HCV-infected participants with (A) cleared reinfection, (B) persistent reinfection, (C) persistent infection with a strain switch, or (D) persistent infection with 1 strain. The study was designed for monthly viral load testing, with more than 8 years of follow-up in some individuals. Dashed line indicates limit of detection (LOD) of the HCV RNA assay. Infections are shaded with different colors based on the HCV subtype of the infecting virus, with subtype indicated, determined by sequencing of Core-E1 genes.

Figure 2. Neutralizing breadth and potency of Abs in longitudinal plasma samples.

(A) Percentage neutralization of 19 HCVpp by plasma of (A) reinfection clearance, reinfection persistence, persistence strain switch subjects, and (B) persistence 1 strain subjects. Subjects from each group are arranged from highest to lowest neutralizing breadth. Negative percentage neutralization values were converted to 0. DOV, calculated by counting viremic periods and excluding periods of aviremia between infections; Infxn # (gt), number of genetically distinct infections the subject has experienced (genotype of the current infection); breadth, number of HCVpp neutralized at least 25% by plasma at 1:100 dilution; potency, highest percentage of neutralization across the panel of 19 HCVpp by plasma at 1:100 dilution. Percentage neutralization values are the average of 2 independent experiments performed in duplicate.

Figure 3. Plasma deconvolution reveals mAb types contributing to plasma-neutralizing breadth and potency.

(A) mAb types identified by deconvolution of neutralizing activity of each plasma sample. Neutralization profiles entered into the deconvolution algorithm were averaged from 2 independent neutralization experiments performed in duplicate. For each plasma sample, mAb types with deconvolution values exceeding the true positive cutoff are indicated, with a different color assigned to each mAb type. Reference mAbs were designated narrow spectrum or bNAbs based on neutralization of less than 50% or more than 50% of the HCVpp panel. Deconvolution was performed only for plasma samples with neutralizing breadth greater than or equal to 4. This breadth 4 or greater cutoff was determined using control mAb “spike-in” experiments to determine the minimum neutralizing activity necessary for accurate Ab deconvolution. Subjects with neutralizing breadth of less than 4 for all samples were excluded from this analysis. Plasma samples are grouped by subject outcome. P values were calculated for Pearson’s correlation between the plasma sample neutralization profile and the best fit combined reference mAb neutralization profile. Breadth, number of HCVpp neutralized at least 25% by plasma at 1:100 dilution; potency, highest percentage neutralization across the panel of 19 HCVpp by plasma at 1:100 dilution. (B) Number of mAb types detected per subject after 1 infection (n = 17), after 2 infections (n = 8), and after more than 2 infections (n = 4). Median with IQR is shown. (C) The proportion of subjects with each mAb type (or 0 mAb-types) after 1 infection, after 2 infections, and after more than 2 infections (n = 4).

Figure 4. Duration of viremia and number of distinct infections are associated with increased neutralizing breadth and potency.

(A) Quasi-Poisson regression analysis for the association of number of infections and duration of infection with neutralizing breadth. Curve and 95% prediction intervals (shaded areas) for each number of infections are indicated with different colors. The regression model equation and table with the estimated coefficients and 95% CI for each variable are shown (right). (B) Linear regression analysis for the association of number of infections and duration of infection with neutralizing potency. Curves and 95% prediction intervals (shaded areas) for each number of infections are indicated with different colors. The linear model equation and table with the estimated coefficients and 95% CI for each variable are shown (right).

Figure 5. Greater genetic distance between infecting viruses is not associated with increased neutralizing breadth or potency.

(A) Table illustrating divergence of each reinfecting virus E1E2 sequence (amino acid p distance) from that subject’s infection 1 T/F virus E1E2 sequence. (B) Quasi-Poisson regression (left) and linear regression (right) analyses for the association of divergence, number of infections, and DOV with neutralizing breadth (left) or neutralizing potency (right). Curves and 95% prediction intervals (shaded areas) for each number of infections are shown in different colors. Model equations and tables with the estimated coefficients and 95% CI for each variable are shown (below).

Figure 6. Longitudinal E1E2 isolates from reinfection subjects cluster in antigenically distinct clades, and infection with viruses from antigenic clade 1 is associated with increased neutralizing breadth and potency.

(A) Heatmap illustrating ELISA binding of a panel of mAbs recognizing conformational epitopes to longitudinal E1E2 proteins from reinfection subjects. Each value is the average of 2 replicates and is normalized for binding of HCV-1, a control mAb recognizing a linear epitope that is 100% conserved across all isolates. Gray cells indicate missing data. E1E2 proteins were clustered based on mean squared distance between binding profiles. Asterisks indicate mAb types or ruas of mAb types identified in broadly neutralizing plasma after multiple infections (Figure 3). (B) Number of infections with viruses from antigenic clade 1 is significantly associated with greater neutralizing breadth, but number of infections with viruses from clades 2 through 4 are not (P > 0.05). Normality of data was tested by Shapiro-Wilk test. Kruskal-Wallis test (clades 1 and 4) and Mann-Whitney nonparametric test (clades 2 and 3) were conducted. Horizontal lines indicate medians. (C and D) Quasi-Poisson regression (C) and linear regression (D) analyses for the association of number of infections with viruses from antigenic clade 1, total number of infections, and DOV with neutralizing breadth (C) or neutralizing potency (D). Each total number of infections from 1 to 4 is illustrated on a separate graph. Curves and 95% prediction intervals (shaded areas) for each number of antigenic clade 1 infections are shown in different colors. Model equations and tables with the estimated coefficients and 95% CI for each variable are shown (below).