Interpreting the evolution

of SARS-CoV-2

 

Jesse Bloom

Fred Hutch Cancer Research Center / HHMI

 

Slides at https://slides.com/jbloom/sars-cov-2_activ

 

@jbloom_lab

Only sometimes does virus evolution erode antibody immunity

  • Measles virus: Does not evolve to escape immunity. People infected at most once in their lives. Vaccine developed in the 1960s still works today.

 

  • Influenza virus: Evolves to escape immunity. People infected every ~5 years. Vaccine updated annually.

Evolution of CoV-229E spike erodes neutralization by human antibody immunity

Serum collected in 1985 neutralizes virus with spike from 1984, but less effective against more recent viruses. 

Antibody immunity recognizes CoV-229E strains that circulated during a person's life

For complete data on all sera, see

Viral evolution erodes antibody immunity of different people at different rates

We are studying basis of differences, as ideally vaccines would elicit evolution-resistant sera as on the right.

Most mutations in RBD and to lesser extent NTD

Plot of sequence variability across CoV-229E spike taken from Eguia, ..., Bloom, bioRxiv (2020). See also Wong, ..., Rini, Nature Communications (2017) and Li, ..., Rini, eLife (2019) for detailed structural studies of evolution in CoV-229E receptor-binding loops.

We decided to focus on prospectively mapping effects of mutations in SARS-CoV-2 RBD

Most neutralizing activity of most convalescent human sera is from RBD-binding antibodies

Data shown here from Greaney et al (2021); similar results were obtained by Piccoli et al (2020).

 

Note that neutralizing antibodies can also target other spike domains such as NTD (e.g., McCallum et al, 2021).

We wanted to map how each mutation affects key biochemical phenotypes

Impossible to measure true viral fitness in the lab, so we focus on three biochemical phenotypes that crucially contribute to fitness:

 

1) Does RBD fold properly?

 

2) Does RBD bind ACE2 with high affinity?

 

3) Is RBD bound by anti-viral antibodies?

 

Impossible to measure true viral fitness in the lab, so we focus on three biochemical phenotypes that crucially contribute to fitness:

 

1) Does RBD fold properly?

 

2) Does RBD bind ACE2 with high affinity?

 

3) Is RBD bound by anti-viral antibodies?

 

Evolutionary pressure is to maintain these two phenotypes...

 

 

... while changing this phenotype.

We wanted to map how each mutation affects key biochemical phenotypes

We use yeast display to enable high-throughput experiments

RBD

fluorescent ACE2

yeast

fluorescent tag on RBD

Importantly, we use ACE2 titrations to measure true affinities, not just relative FACS binding signal; see here for details.

Library of yeast, each expressing different RBD mutant with identifying 16 nt barcode

Click here for details on how library is made.

Phenotypic maps of how mutations affect RBD folding & ACE2 affinity

We map escape mutations by sorting for RBD variants that don't bind antibody

We map escape mutations by sorting for RBD variants that don't bind antibody

We map escape mutations by sorting for RBD variants that don't bind antibody

We map escape mutations by sorting for RBD variants that don't bind antibody

We map escape mutations by sorting for RBD variants that don't bind antibody

In maps, tall letters indicate strong escape mutations

Escape map for one antibody (COV2-2499)

Interactive structural view of escape map (click here)

To test if phenotypic maps explain evolution, we grew virus with antibody

Spike-expressing VSV (Case*, Rothlauf*, ..., Whelan. Cell Host & Microbe, 2020) in a real-time cell analysis assay (Gilchuk, ..., Crowe. Immunity, 2020).

See here for more details.

Antibody COV2-2499 selected escape mutations in 5 of 16 replicates 

mutation

G446D

Q498R

count

3

2

mutation

G446D

Q498R

count

3

2

Escape map has many mutations, why were these two selected?

mutation

G446D

Q498R

count

3

2

Only some mutations accessible by single-nucleotide changes

mutation

G446D

Q498R

count

3

2

Many mutations deleterious for ACE2 binding, but ones selected in virus aren't

ACE2 binding

weak                         strong

Selected mutations are single-nt changes that escape antibody, retain ACE2 binding

effect on escape

ACE2 affinity

effect on escape

ACE2 affinity

Phenotypic maps also explain mutations selected by other antibodies

Antibody COV2-2165 did not select any escape mutations in repeated trials

0 escape mutants in 56 attempts

Initially appear to be many escape mutations from this antibody

But only a few are accessible by single-nucleotide changes

Accessible mutations impair folding or ACE2 binding, explaining why virus can't escape

RBD expression

ACE2 binding

RBD expression /

ACE2 binding

weak                         strong

We now have escape maps for many antibodies, including lead clinical ones

Infection / vaccination elicit polyclonal antibodies that can bind many epitopes

Monoclonal antibodies bind one epitope, so can usually be escaped by single mutation

Polyclonal antibodies can bind many epitopes, so often more resistant to escape

Extent of focusing of polyclonal antibodies shapes antigenic evolution

  • For influenza virus, human polyclonal antibody immunity often highly focused, such that single mutations can reduce neutralization by >10 fold. (Lee et al, eLife, 2019)

 

  • For measles virus, polyclonal antibodies target many epitopes with similar potency, so no single mutation has a large effect. (Munoz-Alia, bioRxiv, 2020)

We mapped how all RBD mutations affect binding by convalescent human sera

For more maps see

The maps define main epitopes as receptor-binding ridge and secondarily 443-450 loop

Most important site is E484, which is mutated in 501Y.V2 and 501Y.V3 lineages

501Y.V2 also known as B.1.351, originally identified in South Africa.

 

501Y.V3 also known as P1, originally identified in Brazil.

But as for CoV-229E, people affected differently by viral mutations

E484 mutations have huge effect

G446/G447 mutations have large effect

No mutations have a large effect

These differences between people validate in neutralization assays

For instance, E484K greatly reduces neutralization by subject C. Subject G is unaffected by E484K, but affected by G446V.

We can use maps to assess which current mutations have biggest antigenic effect 

Chu lab (Univ Wash): Helen Chu, Caitlin Wolf

Crowe lab (Vanderbilt): James Crowe, Seth Zost, Pavlo Gilchuk

King lab (Univ Wash): Neil King, Dan Ellis

Veesler lab (Univ Wash): David Veesler, Alexandra Walls, Ale Tortorici

Whelan lab (Wash U)

Boeckh lab (Fred Hutch): Terry Stevens-Ayers

Alex Greninger (Univ Wash)

Janet Englund (Seattle Children's)

Andrea Loes

Tyler Starr

Allie Greaney

Rachel Eguia

Bloom lab (Fred Hutch)

Sarah Hilton

Kate Crawford

sars-cov-2_ACTIV

By Jesse Bloom

sars-cov-2_ACTIV

Presentation to ACTIV group, Feb 2021

  • 1,077