Transcript HG-6-64-1 in A375, HCT-116, HT-29

LINCS Consortium Meeting
10/27/2011
LINCS joint project: Linking the phosphoproteome and
transcriptome in oncogenic signaling pathways.
Mario Niepel and Peter K. Sorger
Harvard Medical School
Aravind Subramanian and Todd R. Golub
Broad Institute
Collaboration of the Broad and HMS LINCS Centers
Slide 1/17
7/7/2015
LINCS joint project: a subset of perturbations.
Collaboration of LINCS centers means that these
perturbations are explored in great detail.
• Time-points: series instead of one single time point
• Doses: saturating and (multiple) subsaturating concentration.
• Cell contexts: extensively characterized set of breast cell lines.
• Multiple readouts: HTM, ELISA, RPLA, L1000, live-cell microscopy.
Resulting data can be integrated into the larger LINCS effort:
Cell lines are a subset of a set of 50+ breast cancer cell lines screened by HMS.
Drugs are a subset of thousands of drugs screened by Broad Institute.
Slide 2/17
7/7/2015
Growth factors play important role in disease.
Ligands cause diverse phenotypes.
Ligand signaling targeted by therapeutics.
BT474 / Lapatinib
Ligands protect from drug treatments.
Slide 3/17
Nature Reviews Molecular Cell Biology 2, 127-137/Nature Reviews Drug Discovery 8, 627-644
7/7/2015
Selected subset of kinase inhibitors.
Extensive in vitro characterization.
Many used as cancer therapeutics.
Cause detectable phenotypes.
Slide 4/17
Nature Reviews Molecular Cell Biology 2, 127-137/Nature Reviews Drug Discovery 8, 627-644
7/7/2015
Measure transcriptional response to ligands.
ligands(32)
x
cell lines (5)
x
doses(2)
x
time points (3)
x
replicates (4)
~4000 profiles
Slide 5/17
7/7/2015
Inferred ligand profiles match published data.
BT20 treated with EGF matches
published MCF10A treated with EGF:
All top matches are ‘inferred’ genes
rather than ‘landmark’ genes.
Growth factor treatments or breast
tissue make up only a tiny fraction of
data used for inference.
Slide 6/17
7/7/2015
Ligand responses can be connected to cMAP.
LCK: 1 / 5557
NTRK2: 3 / 5557
FRK: 5 / 5557
Top matches of genes upregulated by EGF treatment in MCF7:
Queries against perturbations already contained in the CMap database
uncover pathways that are preferentially activated by specific ligands in
different cell lines.
Slide 7/17
7/7/2015
Ligand families have specific response signatures.
Correlation matrix of MCF10A
Responses to ligand family
members fall into specific
clusters.
There are specific differences
in how a cell line responds to
different ligand families.
Slide 8/17
7/7/2015
Measure phosphorylation and expression responses.
Measured signals
cell lines / time points
Ligands / concentrations
Slide 9/17
7/7/2015
Similar immediate early signals.
pERK
Slide 10/17
pAkt
7/7/2015
Cell line specific differences in a pathway.
pGSK3b
Slide 11/17
pS6
7/7/2015
increasing
sustained
transient
Kinetic analysis shows significant differences
between ligands and cell lines.
Automated classification of pAkt in response to ligands: While most
ligands stimulate the phosphorylation of AKT, the kinetics of the
stimulation are dependent on the ligand and cell line used.
Slide 12/17
Evan Paull in Josh Stewart’s lab
7/7/2015
Computationally link proteome with transcriptome.
Predictions of transcription factor and kinase activity:
Using public databases, the activities of transcription factors and
upstream kinases are predicated from the L1000 profiles for specific
perturbations.
Slide 13/17
Ma’ayan Lab
7/7/2015
Measure transcriptional response to inhibitors.
Inhibitors (100)
x
cell lines (5)
x
doses(4)
x
time points (2)
x
replicates (4)
~16000 profiles
Slide 14/17
7/7/2015
Proliferation response to HMS inhibitor library.
Measurement of cell growth in response to drugs: High-throughput
live cell microcopy is used to analyze the phenotypic responses
(proliferation, apoptosis, migration, invasion) of cells to perturbagens.
Slide 15/17
Benes Lab
7/7/2015
Drugs with same ‘target’ have different specificities.
KiNativ
a
b
c
KINOMEscan
d
a
b
c
MRCPPU
d
a
b
c
d
Drug specificity profiles measured by multiple platforms:
This data defining the ‘off-target’ effects of drugs will be integrated
with the remaining data to better understand the actions of the
perturbagens.
Slide 16/17
Gray Lab
7/7/2015
How do we apply this to biological questions.
• How do changes in the proteome drive transcription?
• How do responses of the proteome and transcriptome link
perturbations with phenotypic outcomes?
• Are these links cell type or tumor subtype specific? Do they
correlate to drug responsiveness?
• What are the similarities and differences in ligand responses
across different ligands or different cells?
• Can we learn more about how exogenous or autocrine/paracrine
ligands effect drug responses?
Slide 17/17
Ma’ayan Lab
7/7/2015