Mapping the Neutralizing Specificities of HIV Antibodies

In a new study led by @CaelanRadford in collaboration with @FKlein_lab, we use an improved deep mutational scanning system to map the neutralizing specificities of broad human polyclonal serum antibodies from individuals living with HIV.https://t.co/GCeUHRyjlQ

— Bloom Lab (@jbloom_lab) March 25, 2023

As background, HIV infection sometimes elicits antibodies that broadly neutralize many strains.

Individual broadly neutralizing antibodies have been extensively characterized, but it is much harder to map the neutralizing specificity of the actual polyclonal antibody response.

— Bloom Lab (@jbloom_lab) March 25, 2023

Our group previously used deep mutational scanning to map mutations affecting HIV antibody neutralization by creating mutant libraries of Env in actual virus (https://t.co/j71nyo238b). While that work has been useful, it had 3 main limitations…

— Bloom Lab (@jbloom_lab) March 25, 2023

Those limitations, now all overcome, were:

1⃣Could only map effect of individual Env mutations, not combinations

2⃣Required work w actual HIV, which created biosafety considerations & made expts more difficult

3⃣Measured relative enrichment rather than absolute neutralization

— Bloom Lab (@jbloom_lab) March 25, 2023

To overcome limitations, we used new non-replicative lentiviral deep mutational scanning our group recently developed:https://t.co/7I9vAoFzNI

Makes it possible to safely measure how combos of mutations to many viral entry proteins, including Env, affect antibody neutralization.

— Bloom Lab (@jbloom_lab) March 25, 2023

We first created libraries of Env from transmitted/founder strain BF520.

Each library had ~40K Env mutants w avg of ~2.5 amino-acid mutations.

We measured how mutations affected viral entry: as expected, mutations found more in natural HIV alignment were better for viral entry pic.twitter.com/s7AK3mgnMc

— Bloom Lab (@jbloom_lab) March 25, 2023

We next measured how all the mutations affected antibody neutralization, using an “absolute neutralization standard” to convert relative deep sequencing counts into quantitative measures of infection neutralization (https://t.co/d1xP1FYaWa).

— Bloom Lab (@jbloom_lab) March 25, 2023

Initially we validated this approach with the well-characterized antibody PGT151. Our deep mutational scanning accurately mapped the epitope of PGT151 in a way highly consistent with extensive prior structural and functional data. pic.twitter.com/3I7J7f08qQ

— Bloom Lab (@jbloom_lab) March 25, 2023

Importantly, the high-throughput measurements of affect on PGT151 neutralization we make for 1000s of Env mutations correlate extremely well with traditional TZM-bl neutralization assays (much better than for our prior HIV deep mutational scanning methods). pic.twitter.com/PeUCyrO5qI

— Bloom Lab (@jbloom_lab) March 25, 2023

After validating approach, we started interesting part: mapping neutralizing specificity of human polyclonal serum antibodies to HIV.

First, we focused on individual from whom @FKlein_lab had isolated extremely broad & potent monoclonal antibody 1-18: https://t.co/wDqlcMy0fT

— Bloom Lab (@jbloom_lab) March 25, 2023

The serum from this individual (IDC561) had a neutralization escape map targeting the CD4 binding site that resembled but was not identical to 1-18. Suggests that 1-18 or very similar antibodies substantially contribute to polyclonal serum neutralization for this individual. pic.twitter.com/WzAN3bNIoq

— Bloom Lab (@jbloom_lab) March 25, 2023

Other serum targeting the CD4 binding-site had different neutralization specificities. For instance, serum IDC513 had an escape map similar to monoclonal antibody 3BNC117. Must be convergent evolution, as that antibody not from this individual. pic.twitter.com/vU0qXIJyoT

— Bloom Lab (@jbloom_lab) March 25, 2023

The existence of multiply mutated Envs in our libraries also enables us to identify serum where neutralization is contributed by antibodies targeting distinct epitopes.

For details of how that is done, see this thread / paper: https://t.co/5VgkGOjhwL

— Bloom Lab (@jbloom_lab) March 25, 2023

We do find one CD4-binding-site targeting serum that clearly has two neutralizing epitopes in the CD4 binding site, each of which resembles the sole epitope of a different serum. Suggests how there can variable number of targets in polyclonal neutralizing response. pic.twitter.com/rUZjayC8vZ

— Bloom Lab (@jbloom_lab) March 25, 2023

Overall, we think ability to accurately & directly interrogate neutralizing specificity of polyclonal serum can help inform vaccine design, particularly in conjunction with other new techniques (eg, emPEM; https://t.co/08AzoFnMng) for directly visualizing binding specificities.

— Bloom Lab (@jbloom_lab) March 25, 2023

All of the data and code from our study, including interactive visualizations of the antibody and serum escape maps, are available at https://t.co/dVzT7uiJze

— Bloom Lab (@jbloom_lab) March 25, 2023

Thanks to all the contributors to this study: @CaelanRadford, @SchommersP, @LutzGieselmann, @khdcrawford, @bdadonaite, @timcyuu, @AdamDingens, Julie Overbaugh, and @FKlein_lab.

— Bloom Lab (@jbloom_lab) March 25, 2023