HCV Evolution: Diversification and Convergence

Yury Khudyakov

Division of Viral HepatitisCenters for Disease Control and Prevention, Atlanta, GA

Public Health

Reduction of morbidity and mortality

- Diagnostics- Treatment- Prevention

Medicine

Introduction

A high rate of mutation defines rapid HCV evolution

HCV genome continuously changes

“Arms Race”

Pervasive coevolution

Opportunity for convergence

Introduction

PHYLOGENETIC ANALYSISIntra-Host HCV Variants

Patient 1

Patient 2

Patient 3

Patient 4

Network of coordinated substitutions in the HCV polyprotein

K-Core Decomposition of HCV Network

Bayesian Network of HCV Polyprotein Site Interactions and Therapy Outcome

Bayesian Network associating the HVR1 sites with IFN response and host demographic factors

HVR1-BN

Intra-Host Evolution over

Many Years

15.991aIDUFemaleUnknownD

15.961aUnknownMaleBlackC

18.121aIDUFemaleBlackB

8.841bTransfusionMaleWhiteA

YearsGenotypeTransmissionGenderRacePatient ID

Sentinel County HCV Follow-up Study

0.0 5.0 10.0 15.0

0.00

10.00

20.00

Div

erge

nce

Time

DBA C

HCV Quasispecies Divergence During Long-Term Chronic Infection

Patients

HVR1: patient A

Time-pints (Yrs)

0

8.8

2.8

7.92.3

0-2.8 yr

7.9-8.8 yr

HVR1: patient B

0

17.218.2

11.2

15.3

12.213.4

10.19.07.93.33.02.8

16.2

Time-points (Yrs) 0 - 3.3 yr

7.9-13.4

7.9-17.2

15.3-18.2

HVR1: patient DTime-points (Yrs)

0

16.0

1.5

2.52.74.8

2.3

1.62.0

11.612.114.015.0

1.5-2.7 yr

4.8-16.0 yr

HVR1: patient CTime-points (Yrs)

0

9.0

16.0

7.1

15.0

8.1

14.6

10.5

0.3

11.1

0-0.3 yr

7.1-8.1 yr

9.0-16.0 yr

HVR1 Phylogenetic Trees

DBA C

0.0 5.0 10.0 15.0

0.00

0.50

1.00

1.50

2.00

dN/d

S

Time

Changes in Selection Pressures Over Time

dN/dS vs. TimeHVR1 – R=-0.58, p=0.0001NS5A – R=-0.61, p=0.0001

Titer vs. TimeR=0.585, p=0.0001

Titer vs. dN/dSR=-0.383, p=0.012

Patients

Genetic Linkage to Viral and Host Factors

Genomic StructureQS diversity

HCV QS SEQUENCE HOST

Viral titer

dN/dS

HCV QS SEQUENCE Factors

I. Molecular Epidemiologic Data• NHANESIII:

• 106 patients• 1384 HVR1 quasispecies; Genotypes 1 – 6• HVR1: positions 1491 to 1577nt (polyprotein 488 to 516)• 5’UTR: positions 127 to 340nt• NS5B: positions 8290 to 8589nt (polyprotein 2651 to 2749)

II. Quantitative Structure Relationships• Probabilistic Graphical Models:

• Bayesian Networks (BN)

III. Predictions• Causal models:

• BN classifiers

Bayesian Network Model Associating Sequences of HCV HVR1 Quasispecies to Viral and Host Parameters

Bayesian Network Model Associating Sequences of HCV HVR1 Quasispecies to Viral and Host Parameters

Bayesian Network Model Associating Sequences of HCV HVR1 Quasispecies to Viral and Host Parameters

Bayesian Network Model Associating Sequences of 3 HCV Genome Regions to Viral and Host Parameters

Target classes 10-fold-CV ‡

(%) Acc.randTest †

(10-fold-CV ‡)TestSet**

Genotype 99.9% 0.3286 100%

dN/dS^^ (3-bin) (2-bin)

94.4%92.2%

0.4020 0.5120

70.3%82.7%

NQSaa-hvr1 88.0% 0.3887 70.3%

NQSnt-hvr1 87.7% 0.3978 72.4%

Viral Titer 97.2% 0.6031 52.40%

‡ Avg. accuracies† Random assignment of class labels** 10 NHANES-3 patients; 5M and 5F; Genotypes 1a and 1b; 185nt/96aa HVR1 QS^^ Based on dNdS 3 class or 2 class grouping

Quantitative Validation of Models

Predictions: Classification Modeling

Five physicochemical properties (Atchley et al, 2005):

Polarity, α-helix , Size, aa frequency, Charge

Multiple sequence alignment

Euclidean distance between every pair of sequences

Visualization of distance matrix Pathfinder network (r = ∞, q = n-1)

Methods

PFNET

Inter-genotype convergence

Genotype 1

Genotype 2

Inter-genotype convergence

• 24.3% of all links are between different genotypes.

Genotype convergence

• We immunized mice with 102 HVR1 peptides covering all high-density regions of the sequence space.

• We tested the reactivity of each sera against 262 peptides (in yellow), a total of 26724 reactions

Cross-reactivity experiment

• There were 5039 positive reactions (blue links), which correspond to 18.85% of all tested.

• Three peptides (yellow) were found that collectively reacted with all 262 antigens.

Relationship between the reduction of selection pressure and cross-immunoreactivity among HCV intra-host variants

Patient Correlation between DN/DS and ACR

p-value

Patient B -0.6262 0.0166Patient C -0.9101 0.0003Patient D -0.476 0.1001

60*ACR

DN/DS

Patient C

Public Health: Reduction of morbidity and mortality

- Diagnostics- Treatment- Prevention

Medicine:

Conclusion

Many viral phenotypic traits with significant medical and public healthimplications are convergent rather than ancestral

Thank you!