Pseudovirus deep mutational to forecast SARS-CoV-2 evolution

Jesse Bloom

Fred Hutch Cancer Center / HHMI

These slides at https://slides.com/jbloom/sars2dms

Most viral vaccines induce neutralizing antibodies to the viral entry protein

All viruses have one or more entry proteins that bind receptor and then fuse with the cell membrane:

SARS-CoV-2 spike
influenza hemagglutinin
HIV envelope protein
Lassa virus glycoprotein
Nipah virus G and F proteins
RSV G and F proteins

Mutations to viral entry proteins affect antigenicity and transmissibility

For SARS-CoV-2 spike, we know following molecular phenotypes are important:

Cell entry: if spike cannot mediate cell entry, virus will not be fit
ACE2 binding: if spike does not bind ACE2 strongly enough, virus will not be fit
Antibody escape: if spike is resistant to antibodies, virus will be more fit

Viruses constantly acquiring mutations that potentially affect these phenotypes

How can we rapidly and safely measure how mutations affect these phenotypes?

General: can work for many viral entry proteins
Comprehensive: measure mutations throughout spike, not just RBD
High-throughput: can be applied new variants and antibodies
Safe: we want to characterize mutations with making novel pathogenic viruses
Applicable directly to sera, to capture heterogeneity due to imprinting

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

Dadonaite et al, Cell (2023)

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

Dadonaite et al, Cell (2023)

Sequencing measures relative amounts, so include neutralization standard

Dadonaite et al, Cell (2023)

Example: RBD antibody LY-CoV1404

Dadonaite et al, Cell (2023)

Validates very well

Dadonaite et al, Cell (2023)

Example: S2 antibody CC67.105

Dadonaite et al, Cell (2023)

days before present

fraction of USA sequences

Estimates of clade growth from multinomial logistic regression

Experimentally measured spike phenotypes correlate with SARS-CoV-2 clade growth

OLS regression combining spike phenotypes predicts clade growth quite well

We can measure how mutations affect key phenotypes that actually shape SARS-CoV-2 evolution

What could we hope to do next?

Develop a sustainable approach to keep generating these phenotypic data as the virus and population immunity evolve.

Share the data so they can be easily visualized and used by other scientists and variant trackers.

Partner with groups developing models to leverage these data for more sophisticated predictions and countermeasure development.

Overall, the last decade has seen tremendous advances in viral surveillance. What we need now is advances in interpreting the results of this surveillance.

Bloom lab

Bernadeta Dadonaite

Kate Crawford

Caelan Radford

University of Washington

David Veesler

Karolinska Institute

Ben Murrell

Penn

Raiees Andrabi

Scripps

Dennis Burton