Analysis of immunoglobulin transcripts and hypermutation following SHIV(AD8) infection and protein-plus-adjuvant immunization

Abstract

Developing predictive animal models to assess how candidate vaccines and infection influence the ontogenies of Envelope (Env)-specific antibodies is critical for the development of an HIV vaccine. Here we use two nonhuman primate models to compare the roles of antigen persistence, diversity and innate immunity. We perform longitudinal analyses of HIV Env-specific B-cell receptor responses to SHIV(AD8) infection and Env protein vaccination with eight different adjuvants. A subset of the SHIV(AD8)-infected animals with higher viral loads and greater Env diversity show increased neutralization associated with increasing somatic hypermutation (SHM) levels over time. The use of adjuvants results in increased ELISA titres but does not affect the mean SHM levels or CDR H3 lengths. Our study shows how the ontogeny of Env-specific B cells can be tracked, and provides insights into the requirements for developing neutralizing antibodies that should facilitate translation to human vaccine studies.

Figure 1. Serum neutralization breadth and viral diversity during SHIV_AD8 infection.

(a) Plasma neutralization of selected viruses 100–102 weeks post infection with SHIV_AD8. Eight animals were segregated into good and poor neutralizers on the basis of the potency and breadth of their responses. ID₅₀ values are shown; colours indicate potency: 40–99, green; 100–999, yellow; ≥1,000, red. (b–d) Viral sequencing was performed on plasma from SHIV-infected NHP at week 6, 26, 54 and 99, except where indicated. (b) Phylogenic trees of Env sequences rooted to the inoculum sequence for good and poor neutralizers. Colours indicate time points; scale bar indicates 0.3% divergence. The mean divergence (c) or diversity (d) was calculated for each animal at the indicated time points. Horizontal bars indicate group medians; *, significant discovery by multiple t-test comparison. Sequence data were not available for DBVC wk 54, DCV9 wk 26 and DCF1 wk99 because of low viral titres. (e) Consensus Env sequences at the indicated time points highlighting the V2 and V3 loops. Conserved mutations at amino-acid positions 167 and 309 are shown (orange) compared with the original residue (green) in the good and poor neutralizer animals.

Figure 2. Next-generation sequencing of antigen-specific B cells after SHIV_AD8 infection.

Data are shown from four good neutralizers, red; and four poor neutralizers, blue. (a,b) SHM for each animal over time; each symbol represents the average percent divergence from germline for Env-specific (gp120+) (a) or nonspecific (gp120−) (b) sequences from a given animal; error bars indicate s.e.m. (c) Histogram representation of SHM distribution from all sequences from 90–110 weeks post infection; binning averaged in 2% increments. (d) CDR H3 length distribution of Env-specific sequences from individual animals for all time points combined. Colour/symbol scheme as in a; binning averaged in 3-aa increments; arrow indicates population of sequences with long CDR H3 regions. (e) Proportion of unique sequences with long CDR H3 regions from DCF1 at the indicated time points. χ²-test with Yates correction was used to calculate P values between proportions from Env+ or − samples at each time point; n.s., not significant; *P<0.0001. (f) Long CDR H3 regions from DCF1 may be derived in at least two unique ways. Example 1 depicts the CDR H3 region of sequence 4594, arising from V(DD)J recombination. Example 2 depicts the CDR H3 region of sequence 107, arising from N-addition. Bold text indicates mature antibody sequence; red stars indicate predicted tyrosine sulfation; CDR H3 charge is shown. (g) Divergence over time of long CDR H3 antibodies from the parent lineage that includes sequence 4594. A phylogenic tree was constructed by maximum likelihood and rooted to the IGHV4D*01 allele and is colour-coded by time point. (h,i) Two-dimensional plots depicting Env-specific sequences from animal DCF1 at the indicated time points. Sequences are plotted by their V_H divergence from germline and their identity to the 4594 H_C (h); or by their CDR H3 length (i). Red triangle indicates sequence 4594; magenta triangles indicate sequences related to 4594; shaded boxes indicate reads with long CDR H3 regions.

Figure 3. Vaccination with adjuvants results differential effects on humoral and B cellular responses.

(a) Vaccination project overview. Nine vaccines were given to 53 NHP in a homologous prime-boost manner at 0, 4, 12 and 24 weeks (black arrows). PBMC sampling was performed before vaccination (pre-vax) and 6, 26 or 36 weeks after the prime (red arrows). (b) Env-specific IgG-binding titres at week 26. (c) Plasma neutralization of tier 1A MW965.26 Env-bearing pseudovirus at week 26. Horizontal bars indicate medians. *P<0.05; **P<0.01 compared with Env+alum group by the Kruskal–Wallis test. (d–f) Antigen-specific B cells were identified from PBMCs at the indicated time points by binding to a gp140 protein probe followed by flow cytometry. (d) Representative flow cytometry plots of antigen-specific memory B cells at the indicated time points after prime. Antigen-specific memory (e) or naive (f) cells were enumerated at the indicated time points. Data points represent vaccine group means and are depicted as a percent of all IgG⁺ cells. *P<0.05 compared with pre-time point by two-way analyses of variance.

Figure 4. Next-generation sequencing of antigen-specific B cells after protein and adjuvant vaccination.

(a) SHM for each animal (n=51) at week 26; each symbol represents the average percent divergence from germline for sequences from a given animal. Horizontal bars indicate the medians. (b) SHM for each vaccine at week 26; each symbol represents the percent divergence from germline for a unique sequence. Horizontal yellow bars indicate the geometric means. (c) Histogram representation of SHM at week 26. The distribution is shown as a composite for all unique sequences from a given vaccine; binning averaged in 2% increments. (d) Effect of boosting on SHM. Percent divergence of antigen-specific cells sorted at week 6 or 26 is depicted for seven animals. Data points represent the mean SHM for each animal; error bars show s.e.'s; vertical dashed lines represent immunization points; n.s., not significant; **P<0.01; ****P<0.0001 compared with week 6 time point by two-way ANOVA. (e) CDR H3 length distribution as a composite from all sequences from a given vaccine; binning averaged in 2-aa increments. Inset highlights the low frequency of reads with long (≥28 aa) CDR H3 regions. Colours denote vaccines as in c.

Figure 5. Characteristics of animals and sequences with high levels of SHM at week 26 after vaccination.

(a) Histogram representation of SHM distribution from animals with an average >10% divergence from germline, compared with all vaccinated animals. Binning averaged in 2% increments. (b) V_H gene composition for sequences from animals with high SHM compared with the whole data set, all animals. The frequency of reads mapping to each V_H gene are represented as a fraction of the total sequences. (c) V_H gene composition for sequences >20% divergent from germline from a subset of samples taken pre-vaccination (bulk IgG reads) or from Env-specific B cells sorted from all animals post-vaccination. (d) V_H gene composition for sequences >20% divergent from germline from sorted Env-specific (gp120+) or nonspecific (gp120−) B cells from SHIV-infected animals. P values are derived from the Fisher's exact test for the proportion of IGHV3Q. The number of sequences in each population subset is indicated under the corresponding pie chart. Sequences mapping to IGHV4L (blue), IGHV3Q (yellow) and IGHV3J (red) are highlighted.

Figure 6. V_H gene correlations between Env/adjuvant vaccination and SHIV_AD8 infection models.

The composition of individual V_H genes is graphed as the percentage of sequences mapping to each V_H gene within each data set. Each dot represents an individual V_H gene; diagonal lines indicate the position of genes lacking a preference between data sets. (a) HIV Env-specific compared with nonspecific sequences combined from SHIV_AD8 infection and Env/adjuvant vaccination data sets. (b) Env-specific sequences from the SHIV data set compared with the Env/adjuvant vaccination data set. V_H genes with enriched composition in the SHIV data set are shown in green. (c) Env-specific sequences from SHIV_AD8 good neutralizers compared with poor neutralizers. V_H genes with enriched composition in the good neutralizer data set are shown in red; those enriched in the poor neutralizer data set are shown in blue.