Immunobiology 2017 Christiaan H. van Dorp Rutger G ...

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

CD8+ T-cell epitopes in the influenza A virusImmunobiology 2017

Christiaan H. van Dorp Rutger G. WoolthuisJeffrey H. C. Yu Can Kesmir Rob J. de Boer

Michiel van Boven

Universiteit Utrecht

RIVM

22/06/2017

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

MAIN QUESTIONS

1. What is antigenic drift and how can we visualize it?2. Are T-cell epitope escapes of influenza under positive selection?3. Does (T-cell) epitope escape lead to more flu infections?

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

Introduction

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

INFLUENZA A VIRUS (IAV)

I Orthomyxoviridae (Influenza A, B, C,...)I Hosts: bird, human, horse, swine,...I Respiratory virus: infects epithelial cellsI ssRNA virus, genome consists of 8 segmentsI Surface proteins (naming: H1N1, H3N2, H7N1):

I Haemagglutinin: binding to target cells (bind sialic acid)I Neuraminidase: release from host cell (cleave sialic acid)

I Other proteins (more conserved):I Nucleoprotein (NP): encapsidates RNAI Matrix protein 1 & 2 (M1, M2)I Polymerase, non-structural proteins: transcription,

replication

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

IAV IMMUNOLOGY

I IgA Abs against HA canprovide neutralizingimmunity

I Abs against NA alsoeffective

I CD8+ T-cell response: nosterilizing immunity, buthelps to clear infection;

I asymptomatic infectionand reducedinfectiousness.

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

IAV EVADES THE IMMUNE SYSTEM

Antigenic Drift/Shift

Reassortment

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

IAV SEASONAL EPIDEMICS (DRIFT/SHIFT)

I Data: Influenza-like illness (ILI) reported by generalpractitioners (GPs), collected by NIVEL.

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

2009 H1N1 PANDEMIC (RE-ASSORTMENT)

I 2009 ‘swine flu’: reassortment of pig, bird and human flu.I 1918 ‘Spanish flu’ (H1N1): 500 mln infections, 50–100 mln

deaths

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

HISTORY OF IAV

I multiple re-assortment eventsI co-circulation of H1N1 and H3N2

van de Sandt et al., J. Virol. (2016)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

Antigenic cartography

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

ANTIGENIC MAP SHOWS DRIFT

I H3N2 subtypeI Colored eplipses: IAV strainsI Open elipses: Ab-responses

I How does one draw such a map?

Smith et al. Science (2004)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

HEMAGGLUTINATION INHIBITION (HI) ASSAYI HA binds sialic acid on red blood cells (RBCs)I Adding Influenza to blood causes clustering of RBCsI Addition of different concentrations of Abs inhibits

clustering.I compute IC50 value.

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

MULTI-DIMENSIONAL SCALING

I interpret IC50 values form HI assay as ‘distances’I 2-dimensional representation of the strain-spaceI minimize an error functionI (compare with PCA)

Strains represented as points xv on a map (responses by yw).

E(x1, . . . , xN, y1, . . . , yN) =∑v,w

(D(v,w)− ‖xv − yw‖2)2 (1)

Find x1, . . . , xN, y1, . . . , yN such that E is as small as possible.

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

ANTIGENIC CARTHOGRAPHY

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

RELATION TO EPIDEMIOLOGY?

I Large distance between strains = Large epidemic?I Only very little evidence!

Bedford et al. eLife (2014)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

T-cell Antigeniccartography

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

CTL ANTIGENIC MAP FOR THE LPF EPITOPE

I Map for one very variable epitope on NPI Map involving all CD8+ T-cell epitopes?

Boon et al., J Immunol (2004)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

IAV EPITOPES IN IEDB

I HLA-I restricted epitopes (8, 9, 10, 11-mers)I experimentally confirmed CTL-responses (160)I and ‘non-immunogenic’ HLA-I binders (59)

www.iedb.org

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

IAV PROTEOMES FROM GISAID

I H1N1, H2N2, H3N2 whole-proteome strains (≈ 6100)I smaller representative set (314)I from the years 1930–2014

platform.gisaid.org

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

SUMMARY OF THE DATA

I characteristic virus setI epitope loci can have multiple alleles

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

SUMMARY OF THE DATA

I full virus setI color: frequency of the allele in a year

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

DEFINING A METRICI What do we want to measure? cross reactive (memory)

responsesI NB: simplification w.r.t. LPF map...I (adjusted/scaled) Jaccard distance.

Let Ev, Ew = epitopes in strain v and w, respectively, definedistance

D(v,w) =

(1− #Ev ∩ Ew

#Ev ∪ Ew

)· E[#Ev ∪ Ew] (2)

D(v,w) = (1− 23 + 2 + 2

) · scale ≈ 3 + 2

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

T-CELL ANTIGENIC MAP

I Antigenic map thatcontains all subtypes(unlike H3N2 Ab-map)

I Antigenic drift/shiftacross subtypes (H1N1→H2N2→ H3N2).

I compare avian strains withpre-existing immunity

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

Evolution of CTL epitopes

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

THE NUMBER OF EPITOPES DECREASES

I trend for H3N2 stronger than for H1N1 (−0.38 vs. −0.11e/y). M1 and NP largest contributers

I Why would the number of epitopes decrease?

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

THE NUMBER OF NON-IMMUNOGENIC PEPTIDES IS

CONSTANT

I Compare epitopes with non-immunogenic peptides (HLAbinders) from IEDB

I Observational bias unlikely

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

MORE NON-SYNONYMOUS SUBSTITUTIONS AT

EPITOPE SITES?

Problem: HLA typically binds more conserved peptides (HLAtargeting efficiency)

Machkovech et al. JVI (2015); Hertz et al., PNAS (2013)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

COMPARE WITH PARALLEL SWINE FLU LINEAGE

I tree for subtype A / H3N2I red: Human flu; blue: Swine fluI Swine flu not targeted by human T-cells!

Machkovech et al. JVI (2015)

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

CTL ESCAPE EPITOPES UNDER POSITIVE SELECTION?I amino-acid cites indexed by rI Er number of epitopes that contain rI Sr number of non-synonymous substitutions at site rI

F =

∑r Er × Sr∑

r Sr

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

Relation to epidemiology

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

EPITOPE GAIN AND LOSS

I How many epitopes (dis)appear each year?I Is there a relation of T-cell antigenic drift (loss) to the sizes

of influenza epidemics?

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

ESTIMATE SUSCEPTIBILITY FROM ILI DATA

I SIR model: Susceptible, Infected, RecoveredI infection rate β, recovery rate γI start of epidemic t0

I fraction susceptible at start of epidemic: S0

IdSdt

= −βSI , S(t0) = S0

dIdt

= βSI − γI

dRdt

= γI

I Fit model to data...I Age stratified version: age classes 0-4, 5-10, 10-20, 20-45,

45-65, 65+

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

DATA AND MODEL FIT

black line: data. red line: fitted model

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

ESTIMATE SUSCEPTIBILITY FROM ILI DATA

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

COMPUTE EPITOPE LOSS FOR RELEVANT VIRUSES

I only use viruses that circulated during Dutch epidemics(estimated with SIR model)

I Missing data: month and day unknown...

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

RELATION TO LOSS OF EPITOPES?

I Correct for susceptibility of very young children (0-4 years)I Only significant correlation for 45-64 year age class with

naive timing

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

SUMMARY

1. IAV evades the immune system by antigenic drift/shiftand re-assortment.

2. IAV evolves in a 2-dimensional antigenic space.3. T-cell escape epitopes are under positive selection.4. Some evidence for effect of drift on epidemics, but not at

all conclusive

Introduction Antigenic cartography T-cell Antigenic cartography Evolution of CTL epitopes Relation to epidemiology Summary

REFERENCES

I D.J. Smith et al., Mapping the Antigenic and Genetic Evolution ofInfluenza Virus, Science (2004)

I H.M. Machkoveck et al., Positive Selection in CD8+ T-Cell Epitopesof Influenza Virus Nucleoprotein Revealed by a Comparative Analysisof Human and Swine Viral Lineages, JVI (2015)

I A.C.M. Boon et al., Recognition of homo- and heterosubtypic variantsof influenza A viruses by human CD8+ T lymphocytes, J. Immunol(2004)

I C.E. van de Sandt et al., Differential Recognition of Influenza AViruses by M158-66 Epitope-Specific CD8+ T Cells Is Determined byExtraepitopic Amino Acid Residues J. Virol. (2016)

I T. Hertz et al., HLA targeting efficiency correlates with human T-cellresponse magnitude and with mortality from influenza A infectionPNAS (2013)

I T. Bedford et al. Integrating influenza antigenic dynamics withmolecular evolution eLife (2014)