SB2_1_Balakrishnan

Download Report

Transcript SB2_1_Balakrishnan

INCOB, Singapore, September 11, 2009
Alternative paths in HIV-1
targeted human signal
transduction pathways
Judith Klein-Seetharaman
Associate Professor
Department of Structural Biology
University of Pittsburgh
&
Language Technologies Institute
Carnegie Mellon University
Human Immunodeficiency Virus-1 (HIV-1)
• Causative agent of AIDS
– Destroys the immune system
– Leads to opportunistic infections & malignancies
Global Summary of AIDS epidemic, December 2007
Number of people living
with HIV in 2007
Total
Children under 15 years
33 million
2 million
AIDS related deaths
in 2007
Total
Children under 15 years
2.0 million
270 000
• Current antiviral therapy
– Not accessible to everyone
– Cannot eradicate HIV from the body
– Drug resistance problems
– Side effects
HIV-1 drug discovery needed
• No vaccine
HIV-1 Life Cycle
Peterlin and Trono Nature Rev. Immu.(2003) 3: 97-107
Communication between HIV-1 and
human host is essential
Outline
• Aim 1. Define interactome
– Predictions of HIV1,human protein
interactions (Background)
• Aim 2. From interactions to function
– This paper
Aim 1: Define Interactome
• Identify network of interactions between
HIV-1 and human proteins
– Rank-order / stratify known interactions
– Predict new interactions
Our approach: supervised learning
• HIV-1 human protein pair is described with a
feature vector and a class label :
( xi , y)
y  {'Interact','Not Interact'}
Each feature summarizes a biological information
Given data learn a function that would map
feature space into one of the two classes:
f : X Y
Tastan, O., Qi, Y., Carbonell, J. and Klein-Seetharaman (2009) Prediction of Interactions Between HIV-1
and Human Proteins by Information Integration, Proc. Pacific Symp. Biocomputing 14, 516-527
The Data Source
http://www.ncbi.nlm.nih.gov/RefSeq/HIVInteractions
• NIAID database curated from literature
Fu W, Sanders-Beer et al. (2009) Nucleic Acids Res. 37, D417-22.
Types of Interactions Reported
Keywords: “Nef binds hemopoietic cell kinase isoform p61HCK”
• Group 1: more likely direct
acetylated by, acetylates, binds, cleaved by, cleaves,
degraded by, dephosphorylates, interacts with,
methylated by, myristoylated by, phosphorylated by,
phosphorylates, ubiquitinated by
1063 interactions, 721 human proteins, 17 HIV-1 proteins
• Group 2: could be indirect
activated by, activates, antagonized by, antagonizes,
associates with, causes accumulation of, co-localizes
with, competes with, cooperates with ...
1454 interactions, 914 human proteins, 16 HIV-1 proteins
www.hivppi.pitt.edu
HIV-1 protein
Human protein
Training and Testing Data
The ‘interaction’ class:
Group 1, the more likely direct interactions
1063 interactions, 721 human proteins, 17 HIV-1 proteins
The ‘non-interaction’ class:
Select randomly from the pairs that are not
reported in NIAID database
100:1 interacting vs. non-interacting pairs
Human Interactome Features
• Making use of human protein protein interaction
knowledge: Mimicry of human interaction partners
NAP-22/CAP-23
Calmodulin
f neigh (i, j ) 
max
kS j {k1,k2 }
The N-termini resemble
and are both myristoylated
Sequence
Nef Post translational
modification
Cellular location
Molecular process
Molecular function
f pairwise (i, k )
Human Interactome Features
• Making use of human protein protein interaction
knowledge: Human protein’s topological properties
in the human protein interaction network
Degree
Number of neighbors
kv
Clustering coefficient
The extent the neighbors are
connected with each other
2nv
kv (kv  1)
Betweenness Centrality
The fraction of shortest paths
u , wV
pass through the node
 uw (v)
 uw

u , w v
Features (35)
• Differential gene expression
in HIV infected vs uninfected
cells (4)
• HIV-1 protein type (17)
• Motif-ligand feature (1)
• Human protein expression
in HIV-1 susceptible tissues
(1)
• Human PPI interactome
features (8)
• Similarity of the two proteins in
terms of (4)
– Cellular location
– Molecular process
– Molecular function
– Sequence
Feature Importance
Prediction of specific interactions
www.cs.cmu.edu/~HIV/hivPPI.html
Aim 2: From Interactions to Functions
Long-Term Goal: Drug Discovery
• Functionally relevant human proteins are not
always direct interactors:
Recall
Precision
Pairs
in Virion
0.51
0.37
0.26
0.18
0.13
0.09
0.20
0.29
0.36
0.41
0.47
0.47
3372
1942
1440
1085
622
279
246
101
48
17
8
4
Brass et al. siRNA screen
Genes
Interactors
46
1054
14
435
5
208
2
97
1
49
0
25
•
•
•
•
Konig et al. screen
Genes
Interactors
77
422
21
182
11
99
7
54
4
28
2
14
Zhou et al. screen
Genes
Interactors
73
1101
26
456
9
210
5
98
0
51
0
23
304 cellular proteins in Ott Rev Med Bio (2008) 17: 159-75
273 genes in Brass et al, Science (2008) 319: 921-6
295 genes in Konig et al. Cell (2008) 1: 49-60
291genes in Zhou et al. Cell Host Microbe.(2008)4:495-504
• Link interactions to functions
– Identify which signal-transduction pathways HIV-1
targets
Opportunity
Number of human partners
HIV-1 targets human hub proteins
HIV-1 human interactions
Randomly paired interactions
Degree,d
• Epstein–Barr virus targets high degree human proteins
Calderwood et al., PNAS (2007) 104: 7606-11
• Pathogens tend to interact with host proteins with high
degrees and betweenness centrality
Dyer et. al. PLoS Pathog (2008) 4, e32
New Pathway Analysis Approach
• Opportunity:
Ligands
Receptors
– HIV-1 has to be
minimalistic:
HIV protein
Hubs
a lot of work
with just 9
Effectors
genes
– Human host signal transduction pathways are
robust: many proteins are redundant
• Idea:
– Identify alternate pathways
Approach
1. Identify potentially HIV-1 targeted pathways
2. Define paths: going from a start point (i.e.
no edges going in to the node) to an end
point (i.e. no edges leaving the node).
3. Find simple paths, HIV-1 targeted paths, and
alternate paths to the end points.
4. Supplement with functional information
– Drug targets from DrugBank: www.drugbank.ca
– siRNA genes: Brass et al., Science (2008) 319:
921-6; Konig et al., Cell (2008) 1: 49-60; Zhou et al.
Cell Host Microbe.(2008)4:495-504)
Signal Transduction Pathway Data
• Find the points at which HIV-1 targets human
signal transduction pathways
1Matthews
Reactome1
NCI PID2
Initial number of pathways
823
132
Pathways discarded:
a) Insufficient detail
b) Subsets of other pathways
150
393
0
9
Final number of pathways
330
123
et al. (2009) “Reactome knowledgebase of human biological pathways and
processes” Nucleic Acids Research. http://www.reactome.org/
2Schaefer et al. (2008) “PID: the pathway interaction database” Nucleic acids research.
http://pid.nci.nih.gov/
HIV-1 targeted pathways
• The larger the
pathways, the
more proteins
targeted
• HIV proteins
target small &
large
pathways
• Top-ranked
degradation
pathways
• 225 of 453
pathways
targeted by >1
interaction
• 277 of 453
pathways
targeted by at
least one
Group 1
interaction
sorted by
known
interactions
(Group 1)
Alternative paths
SiRNA
G1
G2
Oznur
Gold1
Gold2
NT + SiRNA genes
Number of drug targets
NT
17
34
10
16 No
30 No
8 No
5
12
4
1
2
1
1
7
1
1
7
2
1
9
2
0
2
0
0
1
0
4
2
2
2
3
1
2
2
3
2
10
2
3
8
4
6
11
13
4
11
No
No
No
No
No
No
Yes
No
No
No
No
No
No
Yes
7
78
4
3
20
17
57
1
12
1
1
5
2
13
1
4
2
2
10
4
43
7
5
1
1
7
2
0
4
10
3
3
13
5
54
0
1
2
0
1
0
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
2
1
1
1
1
10
8
Proteasome?
Number of paths
6
8
3
3
2
2
5
9
4
1
No
No
No
Ubiquitin?
Number of end points
Pathway Name
Activated_AMPK_stimulates_fatty_acid_oxidation_in_
muscle
Generation_of_second_messenger_molecules
Phosphorylation_of_Emi1
Activation_of_BAD_and_translocation_to_mitochondri
a_
Mitotic_Prometaphase
Polo_like_kinase_mediated_events
Notch_HLH_transcription_pathway
Global_Genomic_NER__GG_NER_
Intrinsic_Pathway
Orc1_removal_from_chromatin
Number of proteins
• Example pathways with alternate paths
that contain at least one HIV-1 target, at
least one drug target and at least one siRNA target:
Alternative paths
SiRNA
G1
G2
Oznur
Gold1
Gold2
NT + SiRNA genes
Number of drug targets
NT
17
34
10
16 No
30 No
8 No
5
12
4
1
2
1
1
7
1
1
7
2
1
9
2
0
2
0
0
1
0
4
2
2
2
3
1
2
2
3
2
10
2
3
8
4
6
11
13
4
11
No
No
No
No
No
No
Yes
No
No
No
No
No
No
Yes
7
78
4
3
20
17
57
1
12
1
1
5
2
13
1
4
2
2
10
4
43
7
5
1
1
7
2
0
4
10
3
3
13
5
54
0
1
2
0
1
0
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
2
1
1
1
1
10
8
Proteasome?
Number of paths
6
8
3
3
2
2
5
9
4
1
No
No
No
Ubiquitin?
Number of end points
Pathway Name
Activated_AMPK_stimulates_fatty_acid_oxidation_in_
muscle
Generation_of_second_messenger_molecules
Phosphorylation_of_Emi1
Activation_of_BAD_and_translocation_to_mitochondri
a_
Mitotic_Prometaphase
Polo_like_kinase_mediated_events
Notch_HLH_transcription_pathway
Global_Genomic_NER__GG_NER_
Intrinsic_Pathway
Orc1_removal_from_chromatin
Number of proteins
• Example pathways with alternate paths
that contain at least one HIV-1 target, at
least one drug target and at least one siRNA target:
Generation of Second Messenger
Pathway in NCI PID
An HIV targeted path
•
•
•
•
•
•
•
•
•
•
CD3E: CD3epsilon -TCR complex
– DT: P07766
– HIV target according to G1/pred: P07766
CD3D: CD3delta -TCR complex
– HIV target according to G1/pred: P04234
CD3G: CD3gamma -TCR complex
– HIV target according to G1/pred: P09693
ZAP70: zeta-chain (TCR) associated protein kinase
70kDa
– DT: P43403
– HIV target according to G1/pred: P43403
ITK: IL2-inducible T-cell kinase
– DT: Q08881
CD4: CD4 molecule
– HIV target according to G1/pred: P01730
– siRNA: P01730
LCK: lymphocyte-specific protein tyrosine kinase
– HIV target according to G1/pred: P06239
CD247 CD247 molecule
– HIV target according to G1/pred: P20963
LCP2: lymphocyte cytosolic protein 2 (SH2 domain
containing leukocyte protein of 76kDa)
– HIV target according to Oznur: Q13094
– siRNA: Q13094
PLCG1: phospholipase C, gamma 1
– HIV target according to Oznur: P19174
HIV-1 targets (Group 1)
Drug targets
siRNA gene
HIV-1 and drug target
HIV-1 target and siRNA gene
Cholesterol Biosynthesis Pathway
HIV target according to our predictions:
P37268 / FDFT1
Description: farnesyl-diphosphate
farnesyltransferase 1
siRNA: Q01581 / HMGCS1
Description: 3-hydroxy-3-methylglutarylCoenzyme A synthase 1 (soluble)
Drug target: P48449 / LSS
Description: lanosterol synthase (2,3oxidosqualene-lanosterol cyclase)
Drug target: Q14534 / SQLE
Description: squalene epoxidase
Cholesterol Biosynthesis: A new antiHIV Drug Discovery Pathway?
• AIDS patients are at
increased risk for
arthrosclerosis
• HIV Nef inhibits
cholesterol exporter
• Cholesterol
accumulates in HIVinfected cells
Mujawar Z, Rose H, Morrow MP,
Pushkarsky T, Dubrovsky L, et al. (2006)
Human immunodeficiency virus impairs
reverse cholesterol transport from
macrophages. PLoS Biol 4: e365.
Summary
• Aim 1. Define Interactome
– Collected data from multiple biological information sources and
encoded as features
– Developed a model to predict HIV-1,human protein interaction
network. Predictions available at:
http://www.cs.cmu.edu/~oznur/hiv/hivPPI.html
www.hivppi.pitt.edu
• Aim 2. From Interactions to Function – Drug Discovery
– HIV-1 targets human hubs
– HIV-1 targets many interaction partners of functionally relevant
(siRNA) genes
– Mapped known and predicted interactions to signal
transduction pathways: HIV-1 targets many pathways
– Combining path identification, drug target, siRNA and
HIV-1 target information yields experimentally testable
hypotheses on putative anti-HIV intervention routes
Acknowledgements
Oznur
Tastan
Jaime G.
Carbonell
Pittsburgh Center for HIV Protein Interactions
www.hivppi.pitt.edu
Sivaraman
Balakrishnan