Outline HIV coreceptors Importance of coreceptor usage - - PowerPoint PPT Presentation

geno2pheno[CORECEPTOR]

Alexander Thielen

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 2

Outline HIV coreceptors Importance of coreceptor usage Determination and prediction geno2pheno[coreceptor] Outlook of new features Summary

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 3

HIV coreceptors HIV gains entry into cells that express:

• 1. CD4 and
• 2. one of two coreceptors:

a) C-C chemokine receptor 5 (CCR5) b) C-X-C chemokine receptor 4 (CXCR4)

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 4

Three modes of coreceptor usage

(Pfizer, 2006)

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 5

Infection of a cell Infection steps:

• 1. gp120 binds to CD4
• 2. structural alterations in Env

exposing the chemokine binding domains of gp120

• 3. gp120 engages coreceptor
• 4. gp41 penetrates the cell

membrane

• 5. Membrane fusion and entry
• f the viral capsid

CD4 gp120 gp41 coreceptor

(Pfizer, 2006)

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 6

Importance of viral tropism R5 viruses dominate in early phase of infection X4 variants are associated with a more rapid CD4 cell depletion and progression to AIDS

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 7

New class of drugs: coreceptor antagonists Maraviroc (Pfizer): Expanded access program since 4/1/2007 Vicriviroc (Schering-Plough): Phase II concerns about the risk

• f cancer

Failures: Aplaviroc (Glaxo): proven toxic in mid-stage trials

(Nature, 2007)

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 8

Determination of coreceptor usage Two commercially available assays: Trofile – Monogram Biosciences Tropism Recombinant Test (TRT) - VIRalliance use patient plasma-derived viral envelope sequences to construct either replication-competent or replication-defective viruses, respectively Problems: Expensive Not always available

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 9

Prediction methods 11/25 rule X4, if positively charged amino acid at position 11 or 25 of the V3 loop PSSMs Scores sequences according to their frequencies at the different positions in relation to R5-viruses Decision Trees Sequence of criteria deciding the phenotype Neural Networks Simulates a network of communicating nerve cells Support Vector Machines Machine learning methods based on separating hyperplanes

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 10

geno2pheno[coreceptor] Based on Support Vector Machines Trained on 1110 clonal samples Best performance among all tested methods Over 15000 predictions since 6/1/2004

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 11

geno2pheno[coreceptor]

DEMO

http://geno2pheno.org/

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 12

Outlook – usage of clinical markers Predictions of population based sequences generally worse Approach: Incorporation of clinical markers into the prediction model Data: HOMER cohort, coreceptors determined with Trofile Results: Significant improvements with clinical markers such as CD- cell count, viral loads

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 13

Outlook – structure based predictions Homology model of the V3-loop backbone Models side-chains of the loop Computes structural descriptors based on the interactions of the amino acids

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 14

Outlook – structure based predictions 432 different g/p pairs from clonal data set no indels relative to consensus 11/25 specificity 94.6% sequence-based SVM: sensitivity increase: 11.6% sequence + structure: sensitivity increase: 18.6%

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 15

Summary HIV needs coreceptors for infection of cells Two coreceptors, use of one is correlated with rapid disease progression Determination of coreceptor usage is important for new therapies geno2pheno[coreceptor] performs very well New insights, predictions can be improved with clinical markers Predictions can be improved with structure based descriptors

Thielen, Alexander geno2pheno[coreceptor] 26/04/2007 16

Thanks to… Max-Planck-Institute for Informatics Tobias Sing Oliver Sander André Altmann Achim Buech Thomas Lengauer University of Cologne Rolf Kaiser University of British Columbia Richard Harrigan