Mapping viral antibody escape to inform vaccines

 

Jesse Bloom

Fred Hutch Cancer Center / HHMI

 

These slides at https://slides.com/jbloom/dms-for-vax

 

@jbloom_lab

Disclosures

I am on the scientific advisory boards of Apriori Bio, Aerium Therapeutics, Invivyd, and the Vaccine Company

 

I am an inventor on Fred Hutch licensed patents related to deep mutational scanning of viral proteins

Outline

  • Conceptual principles of viral antigenic evolution
  • Deep mutational scanning of SARS-CoV-2 RBD
  • Generalized deep mutational scanning of viral entry proteins

Outline

  • Conceptual principles of viral antigenic evolution
  • Deep mutational scanning of SARS-CoV-2 RBD
  • Generalized deep mutational scanning of viral entry proteins

The Faroe Islands

"Measles had not prevailed on the Faroes since 1781, then it broke out early in April 1846."

"Of the 7782 inhabitants, about 6000 were taken with measles."

"Of the many aged people still living in the Faroes who had measles in 1781, not one was attacked the second time."

Panum is describing immune memory, which provides lifelong protection from measles.

But we are repeatedly infected by some other viruses. Typical person infected by H3N2 influenza ~5 years. Why?

History offers natural experiment with influenza like Panum's study of measles

In 1977, old H1N1 strain from ~1954 was inadvertently re-released and caused a global pandemic.

"One boy from Hong Kong had a transient febrile illness from 15 to 18 January. On Sunday 22 January, three boys were in the college infirmary… 512 boys (67%) spent between three and seven days away from class."

"Of about 130 adults who had some contact with the boys, only one, a house matron, developed similar symptoms."

Influenza infection also elicits multi-decade immunity, but only if virus is evolutionarily frozen

Some human RNA respiratory viruses evolve to escape immunity

Why some viruses evolve to escape immunity while others don't is a deep question outside scope of this talk. See here for some possible explanations. 

Rate of viral antigenic evolution

Measles

Influenza

Rate of viral antigenic evolution

Measles

Where do human coronaviruses fall on this spectrum?

Coronaviruses

???

???

Influenza

Rate of viral antigenic evolution

Measles

Coronaviruses

???

???

Initially, people speculated SARS-CoV-2 would evolve slowly. But we weren't so sure...

Influenza

We decided to study another human coronavirus: CoV-229E causes common colds and has been circulating in humans for a long time.

Reconstructing evolution of CoV-229E spike

We experimentally generated CoV-229E spikes at ~8 year intervals so we could study them in the lab:

- 1984

- 1992

- 2001

- 2008

- 2016

Note "ladder-like" shape of tree

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. 

Viral evolution erodes antibody immunity of different people at different rates

Ideally vaccines would elicit more evolution-resistant sera as on the right.

Rate of viral antigenic evolution

Measles

CoV-229E

CoV-OC43

SARS-CoV-2

Most human coronaviruses undergo rapid antigenic evolution

Influenza

Phylogenetic tree shape & vaccine strategy

CoV-229E has ladder-like tree:

  • new variants displace old ones
  • new variants descend from recent successful ones

Human influenza A evolves this way too. It's theoretically possible to pick single well-matched vaccine strain.

Phylogenetic tree shape & vaccine strategy

CoV-229E has ladder-like tree:

  • new variants displace old ones
  • new variants descend from recent successful ones

Human influenza A evolves this way too. It's theoretically possible to pick single well-matched vaccine strain.

CoV-OC43 split into two ladder-like lineages. Influenza B evolves this way too. It's theoretically possible to pick well-matched bivalent vaccine.

Phylogenetic tree shape & vaccine strategy

CoV-229E has ladder-like tree:

  • new variants displace old ones
  • new variants descend from recent successful ones

Human influenza A evolves this way too. It's theoretically possible to pick single well-matched vaccine strain.

CoV-OC43 split into two ladder-like lineages. Influenza B evolves this way too. It's theoretically possible to pick well-matched bivalent vaccine.

In non-ladder-like tree, next variant not descended from recent successful one. Makes picking vaccine strains difficult.

Outline

  • Conceptual principles of viral antigenic evolution
  • Deep mutational scanning of SARS-CoV-2 RBD
  • Generalized deep mutational scanning of viral entry proteins

Strongest evolutionary selection is in RBD

Sites of evolutionary change in the spike of CoV-229E over the last four decades

Strongest evolutionary selection is in RBD

Sites of evolutionary change in the spike of CoV-229E over the last four decades

Sites of mutations in SARS-CoV-2 Omicron BQ.1.1 spike relative to Wuhan-Hu-1

Majority of neutralizing antibody response in vaccinated/infected humans targets RBD

Importance of RBD

Human CoVs, which evolve to escape transmission-blocking immunity, show strongest selection in RBD.

 

So virus is telling us RBD antibodies matter most for blocking transmission. Other antibodies and T-cells may reduce disease severity while putting less selection on virus.

 

Most neutralizing activity from RBD antibodies (although antibodies to other domains including NTD can also be neutralizing).​

SARS-CoV-2 variants have many mutations in RBD

RBD mutations in Omicron BA.1

How does Omicron even bind ACE2 with so many mutations?

To answer this question, we measure how RBD mutations affect ACE2 affinity

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.

Deep mutational scanning: measurements for a library of all RBD mutations

Library of yeast each expressing a different RBD mutant. Click here for details on how library is made.

Interactive heatmaps are available here, and are from Starr et al, 2022 and related work.

ACE2 affinity-enhancing mutations buffer antibody-escape mutations

RBD will not run out of evolutionary space

25 of 31 residues in CoV-229E RBD that contact receptor varied during virus's evolution in humans over last ~50 years (Li et al, 2019)

 

There are lots of mutations to SARS-CoV-2 RBD that retain (and sometimes even enhance) ACE2 affinity (Starr et al, 2020; Starr et al, 2022)

 

We can also map RBD mutations that escape antibody binding

RBD

fluorescently labeled antibody

yeast

fluorescent tag on RBD

Experiments combine flow cytometry and deep sequencing of a library of yeast expressing all RBD mutants

Escape map from one antibody (LY-CoV555, ie bamlanivimab). Peaks indicate sites where mutations escape binding.

484

452

490

Infection / vaccination elicit polyclonal antibodies 

How do mutations affect polyclonal antibodies? First, consider an equal mix of three monoclonal antibodies.

LY-CoV555 is escaped at both sites 484 and 490, so mutating either site has same overall effect

Average escape across all antibodies

Mutating site 484 or 490 eliminates neutralization by antibody LY-CoV555, as reflected in thick black line showing average

Antibody-escape calculator extends this principle to deep mutational scanning data for ~3,000 different human antibodies

Escape calculator is described in Greaney et al (2022), and is available at https://jbloomlab.github.io/SARS2_RBD_Ab_escape_maps/escape-calc/

36 antibodies mapped by Tyler Starr & Allie Greaney in Bloom lab, from early SARS-CoV-2 strains

Antibody-escape calculator extends this principle to deep mutational scanning data for ~3,000 different human antibodies

Escape calculator is described in Greaney et al (2022), and is available at https://jbloomlab.github.io/SARS2_RBD_Ab_escape_maps/escape-calc/

36 antibodies mapped by Tyler Starr & Allie Greaney in Bloom lab, from early SARS-CoV-2 strains

3,000 (!) antibodies mapped by Yunlong Cao et al at Peking University. From early strains, BA.1, & patients with prior SARS-CoV-1 infection. See here.

Antibodies elicited by early SARS-CoV-2 that neutralize Wuhan-Hu-1 heavily focus on sites like 484, 417, 446

417

446

484

417

446

484

Omicron BA.1 is mutated at many of these sites, which is why it was neutralized substantially less well by current vaccines

417

446

484

Omicron BA.2 has some different RBD mutations than BA.1, but similar overall escape from antibodies from early SARS-CoV-2

Can identify mutations that mediate further escape. In Dec 2021 we predicted 486 as likely site of future evolution--in April 2022, mutation F486V was identified in Omicron BA.4 and BA.5.

486 is largest site of escape for antibodies not already escaped by mutations in BA.2

Outline

  • Conceptual principles of viral antigenic evolution
  • Deep mutational scanning of SARS-CoV-2 RBD
  • Generalized deep mutational scanning of viral entry proteins

Limitations of RBD yeast-display deep mutational scanning

  • Only examines mutations in RBD, which is just part of spike
  • Measures antibody binding, not neutralization. They are unequal in polyclonal serum.
  • Only works for viral entry proteins with domains amenable to yeast display.

Lentiviral pseudotyping

  • Many viruses have entry proteins amenable to lentiviral pseudotyping.
  • However, traditional pseudotyping does not create genotype-phenotype link.

Two-step method to create genotype-phenotype linked spike-pseudotypes

Example: RBD antibody LY-CoV1404

Example: S2 antibody CC67.105

We can also directly map serum escape

Direct mapping of neutralizing specificity of sera could inform vaccine design.

Lentiviral deep mutational scanning extensible to many viral entry proteins

Applicable to entry proteins from other coronaviruses, influenza viruses, Ebola virus, Nipah virus, Lassa virus, RSV, rabies virus, HIV, etc

 

Safe way to do high-throughput mutational studies

 

There is potential information hazard: as we can increasingly map the functional and antigenic effects of viral mutations, we need to ensure such information is not used to inform high-risk gain-of-function studies using actual human pathogens.

Crowe lab (Vanderbilt)

Chu lab (Univ Wash)

Veesler lab (Univ Wash)

King lab (Univ Wash)

Li lab (Brigham & Women's)

Boeckh lab (Fred Hutch)

Alex Greninger (Univ Wash)

Nussenzweig lab (Rockefeller)

Bjorkman lab (Caltech)

Katie Kistler (Fred Hutch)

Tyler Starr

Allie Greaney

Rachel Eguia

Bloom lab (Fred Hutch)

Bernadeta Dadonaite

Kate Crawford

Caelan Radford