Viral phylodynamics of

host adaptation

& antigenic evolution

Sidney Bell

Bedford Lab

Molecular & Cellular Biology

car

cat

Reconstructing evolution is a lot like "telephone"

zebra

....

car

cat

Reconstructing evolution is a lot like "telephone"

zebra

Reconstructing evolution is a lot like "telephone"

ACG

AGG

AGT

Evolution provides context

to viral outbreaks

Where did this virus come from?

When did interesting traits change?

ACG

AGG

AGT

Contextualizing the HIV pandemic

Understanding dengue antigenic evolution

Outline

Contextualizing the HIV pandemic

Cross-species transmission can be devastating

Ebola

Swine flu

HIV

HIV has entered human populations 16 times

x2

x2

x12

45+ primate species carry a unique SIV

?

?

?

How often do

SIVs switch

host species?

Viral phylogenies show

history and migration

Color = host species

Bell & Bedford, PLoS Pathog., 2017

0.5 sub/site

Cross-species transmission looks

like color changes

Caveat: recombination complicates evolutionary history

+

Recombination

Solution: assess each genome segment separately

 

12 segments x ~2,000 trees = 24,000 trees

 

Assessing many hypotheses

Bell & Bedford, PLoS Pathog., 2017

Modeling host species as a discrete trait

P(tree, rates | data) = \frac{P(tree, rates) * P(data | tree, rates)}{P(data)}
P(tree,ratesdata)=P(tree,rates)P(datatree,rates)P(data)P(tree, rates | data) = \frac{P(tree, rates) * P(data | tree, rates)}{P(data)}

From host

To host

A B C

...

A

B

C

...

SIVs have been adapting to new hosts for millions of years

Bell & Bedford, PLoS Pathog., 2017

Understanding dengue antigenic evolution

Dengue is a mosquito-borne virus

that kills 15,000 children annually

Dengue has recently expanded geographically

There are four serotypes of dengue

DENV1

DENV3

DENV4

DENV2

Each serotype is genetically diverse

DENV1

DENV3

DENV4

DENV2

ACTG

ACTT

AGTT

Antigenic variation

Genetic

variation

Brief detour: genetic variation can

cause "antigenic variation"

Antigenic variation changes what the virus looks like to your immune system

Measles

Antigenically uniform

Flu

Antigenically

diverse

Lifelong protection

Temporary cross-protection

Antigenic variation is

why you need a new flu shot every year

Measles is

antigenically uniform

Flu is

antigenically diverse

Dengue serotypes are antigenically distinct

Measles

Antigenically uniform

Flu

Antigenically

diverse

Within

a dengue

serotype

Antigenically uniform (?)

Between dengue serotypes

Antigenically diverse

Lifelong protection

Temporary

cross-protection

Initial

response

For dengue, antigenic differences

can cause severe disease

After

1-3 yrs

 

Increased risk

Dengue vaccine efficacy varies by genotype

DENV1

DENV3

DENV4

DENV2

Juraska et al., PNAS 2018

DENV antigenic relationships

are poorly understood

Serotypes are genetically distinct

Clades are genetically distinct

Serotypes are antigenically distinct

Are clades antigenically distinct?

?

How does

dengue evolve

antigenically?

Measuring antigenic relationships

Neutralization

+

+

Replication

Neutralizing antibodies block virus replication

Titers measure antigenic distance

Antigenic similarity = how much sera is required to neutralize viral replication?

Serum

Test viruses

1x

2x

4x

(relative to autologous)

X

X

X

X

X

X

--

Log2

0

1

2

Titer

Titers measure antigenic distance

\propto
\propto

log2(titer)

antigenic

distance

cross

immunity

\propto
\propto
0 1 2
1 0 2
2 2 0

Serum

Test viruses

Dengue titers are confusing

Non-human primate sera; 3 months post primary infection

Katzelnick et al, Science 2015

Antigenic distance

(less cross immunity)

Cartography "maps" antigenic relationships

0 1 2
1 0 2
2 2 0

1

2

2

Antigenic Distance

\propto
\propto

log2(Titer)

Distance

Dengue serotypes look (maybe?) antigenically diverse

= 2-fold change in titer

Katzelnick et al, Science 2015

Where did this variation come from? Does it matter?

Genetic vs antigenic distance

Homotypic

Heterotypic

Phylogenies can describe both genetic and antigenic relationships

Quantify antigenic change along each branch, 

Neher et al, PNAS 2016

0 1 2
1 0 2
2 2 0
\approx \sum{d_b}
db\approx \sum{d_b}

Titer

between

&

+         avidity +     potency

d_b
dbd_b
d_{1}
d1d_{1}
d_{2}
d2d_{2}
d_{3}
d3d_{3}
d_{4}
d4d_{4}

1.0

0.5

0.5

0.5

\hat{T_{ij}} = \sum{d_b} + v_i + s_j
Tij^=db+vi+sj\hat{T_{ij}} = \sum{d_b} + v_i + s_j
\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2
i,j(Tij^Tij)2+λbdb+γivi2+δjsj2\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2

Learn these

Minimize

this

Predict titers

\hat{T_{ij}} = \sum{d_b} + v_i + s_j
Tij^=db+vi+sj\hat{T_{ij}} = \sum{d_b} + v_i + s_j
\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2
i,j(Tij^Tij)2+λbdb+γivi2+δjsj2\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2

dTiter on each branch

avidity

potency

Predict titers

\hat{T_{ij}} = \sum{d_b} + v_i + s_j
Tij^=db+vi+sj\hat{T_{ij}} = \sum{d_b} + v_i + s_j
\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2
i,j(Tij^Tij)2+λbdb+γivi2+δjsj2\sum_{i,j}(\hat{T_{ij}} - T{ij})^2 + \lambda \sum_{b}d_b + \gamma \sum_{i} v_i^2 + \delta \sum_{j} s_j^2

Squared Error

Minimize

this

L1 norm on branch effects

L2 norm on virus avidity & serum potency

dTiter on each branch

avidity

potency

Predict titers

Interpolate across the tree

to estimate antigenic relationships

Data

Model

Does within-serotype genetic diversity change antigenic phenotypes?

Interserotype hypothesis

Full tree hypothesis

Pearson R 0.79
Abs. Error 1.02
0.86
0.86

Between-serotype variation explains most of

dengue antigenic phenotypes

Within-serotype variation substantially contributes to dengue antigenic phenotypes

Test Error

Interserotype model

Full tree model

Bell et al., bioRxiv 2018

Each serotype contains antigenic diversity

Bell et al., bioRxiv 2018

Vaccine efficacy varies within serotypes

Juraska et al., PNAS 2018

How does

antigenic diversity

impact dengue

population dynamics?

Serotypes cycle through populations

Genotypes

Bell et al., bioRxiv 2018

Does antigenic novelty contribute to dengue clade turnover?

Population susceptibility:

Previously circulating:

Population

immunity:

Clade growth:

Does antigenic fitness

predict clade growth & decline?

Lukzsa and Lassig, Nature 2014

\propto e^{(f_i(t) * dt)}
e(fi(t)dt)\propto e^{(f_i(t) * dt)}

Growth rate @

next 5 years

How antigenically distant

is          from what has recently circulated?

Does antigenic novelty contribute

to dengue fitness?

f_i(t) = f_{i0} - \beta P_i(t)
fi(t)=fi0βPi(t)f_i(t) = f_{i0} - \beta P_i(t)

Relative frequency of j

Waning immunity

P_i(t) \propto \sum_{t-n}^{t} \gamma(n) \sum_{j} x_{j}(t) * C(D_{ij})
Pi(t)tntγ(n)jxj(t)C(Dij)P_i(t) \propto \sum_{t-n}^{t} \gamma(n) \sum_{j} x_{j}(t) * C(D_{ij})

Antigenic distance between i and j

Lukzsa and Lassig, Nature, 2014

Does antigenic novelty contribute

to dengue fitness?

f_i(t) = 1 - P_i(t)
fi(t)=1Pi(t)f_i(t) = 1 - P_i(t)

Relative frequency of j

Waning immunity

Probability of protection from i

given exposure to j

Lukzsa and Lassig, Nature, 2014

P_i(t) \propto \sum_{t-n}^{t} \gamma(n) \sum_{j} x_{j}(t) * C(D_{ij})
Pi(t)tntγ(n)jxj(t)C(Dij)P_i(t) \propto \sum_{t-n}^{t} \gamma(n) \sum_{j} x_{j}(t) * C(D_{ij})

Serotype frequency

Predict clade growth from antigenic fitness

Antigenic novelty drives

viral fitness and serotype turnover

Pearson's r = 0.79

Bell et al., bioRxiv 2018

Measuring genotype antigenic novelty

Serotype resolution

Clade resolution

Serotype resolution

Clade resolution

Antigenic fitness contributes

to genotype turnover

Pearson's r = 0.56

Pearson's r = 0.53

Bell et al., bioRxiv 2018

Summary

2: Serotype-level                    antigenic relationships      drive dengue
    population dynamics

1: Dengue serotypes        are moderately            antigenic diverse

An evolutionary perspective

helps us understand how viruses...

....Enter populations

....Adapt to their hosts

....Spread through populations

+

Acknowledgements

Bedford Lab

Trevor Bedford, Alli Black, Gytis Dudas, James Hadfield, John Huddleston, Katie Kistler, Maya Lewinsohn, Colin Megill, Louise Moncla, Barney Potter, Tom Sibley

Collaborators & Colleagues

Sarah Cobey, Michael Emerman, Jeremy Freeman, Ted Gobillot,  Leah Katzelnick, Richard Neher, Molly O'hainle, Duncan Ralph, David Shaw, Frank Wen, Janet Young

Committee

Jesse Bloom, Julie Overbaugh, Adam Geballe, Erick Matsen, & Daniela Witten

Computational Biology

Melissa Alvendia, Sam Distel, Lindsey Mwoga, An Tyrrell, Scientific Computing

Graduate Program

Rich Gardner, Katie Peichel & Nina Salama

Andrea Brocato, Denise Barnes, Tierra Johnson, Maia Lowe, Laura Masserman, Teresa Swanson

Outreach

Lori Blake, Jeanne Chowning, Sarah Hilton, Emma Hoppe, Shannan Moran, Liza Ray, Beverly Torok-Storb, Kim Wells

Funding

NSF Graduate Research Fellowship, NIH Interdisciplinary Training Grant, ARCS & Vassar Scholar Awards

Slides: RevealJS & The Noun Project

Questions?

@sidneymbell

slides.com/sidneymbell/dissertation