UCSD-Bioinformatics & Systems Biology Group
Networks and Context:Identifying differences between Macrophage and Macrophage
Derived Cell Types
Benner, Subramaniam and Glass. 2003
UCSD-Bioinformatics & Systems Biology Group
Macrophage Cell Types
• RAW – Macrophage cell line
• TM - Thioglycolate elicited macrophages
• BM – Bone Marrow derived macrophages
• ES – Embryonic Stem Cells (not macrophage)
UCSD-Bioinformatics & Systems Biology Group
Chromosome 1
UCSD-Bioinformatics & Systems Biology Group
High Throughput Analysis
Gene Ontology
Biocarta.org
KEGG, others…
15160161711694822329
10421520482
2323141214213169717042166520912679211910611630223291298523872205291249416859664031969
15160161711694822329104215204822323141214213169717042166520912679211910611630223291298523872205291249416859664031969
15160161711694822329104215204822323141214213169717042166520912679211910611630223291298523872205291249416859664031969
BM resposive Groups (vs. TM) P-ValueMouse Breast Cancer and Estrogen Receptor Signaling 2.23E-34Mouse TGFb BMP Signaling Pathway 2.37E-31Common Diseases 3.28E-29Mouse JAK / STAT Signaling Pathway 3.05E-28Mouse Angiogenesis 3.71E-28Growth Factors and Receptors 4.52E-28Mouse Hypoxia Signaling Pathway 4.82E-28Molecular Markers for Stem and Differentiated Cells 5.49E-28Mouse Extracellular Matrix & Adhesion Molecules 5.86E-28Extracellular Matrix and Adhesion Molecules 5.86E-28Cancer 4.07E-25Growth factors and Cytokines 5.11E-25Mouse cAMP/Ca2+ PathwayFinder 5.11E-25Mouse Nitric Oxide 5.11E-25Stem Cell 8.31E-25
TM reposive Groups (vs. BM) P-ValueSmall Inducible Cytokines 1.19E-11Molecules Regulating Inflammation 8.38E-11NFkB Responsive Genes 1.01E-10Interleukins & Receptors 5.29E-10Endothelial cell injure (apoptosis) 1.24E-09Cytokine Production 2.62E-09Chemokines 2.87E-09Cytokines 3.15E-09Smad target gene 4.80E-09Cell-cycle control 6.39E-09CD4+T cell markers 1.24E-08G1 Phase 6.35E-08Other Related Genes 7.69E-08Genes induced by Stat proteins 8.46E-08
Significant Categories are
Assigned
Inspect Results
UCSD-Bioinformatics & Systems Biology Group
Differential Response to LPS• Example: NF-kB Responsive Genes (possible Acquired Immunity Response)
BM TM
UCSD-Bioinformatics & Systems Biology Group
Investigating DifferencesCell Cycle Genes
Common Transcription Factors
UCSD-Bioinformatics & Systems Biology Group
Common cellular and pathway phenotypes lead to distinct
regulatory networks in primary B-cells
Mock and Subramaniam. 2003
UCSD-Bioinformatics & Systems Biology Group
ALLIANCE FOR CELULAR SIGNALING
UCSD-Bioinformatics & Systems Biology Group
Cell Lab in Dallas
Produces Cells
Treats Cellswith Ligands
Molecular Biology Lab
Microarray analysis
Protein Lab
P-proteins
Antibody Lab
P-proteins
Lipid Lab
Lipid analysis
cAMPCalcium
Ligand Screen: Perturbing Cells
UCSD-Bioinformatics & Systems Biology Group
Group Ligand Calcium Cyclic AMP P-Protein Blots Genes Other
Gi S1P ++ + E/A/R SomeSDF ++ + E/A/R SomeBLC ++ + E/A/R SomeELC ++ + E/A/R SomeSLC ++ + E/A/R Some
Gs Terbutaline 0 ++++ 0 SomePGE2 0 ++ (sust) 0 Some
2-MA (?) 0 + 0 Some
Gq & Gs LPA ++ (fast) ++ 0 Some
BCR Antigen ++++ 0 E/A/R Lots Proliferation
IL4 IL4 0 0 Stat 6 Lots
CD40 Anti-CD40 0 0 Delayed E/A/R Lots Proliferation
IL10 IL10 0 0 Stat 3 Yes
BAFF BAFF 0 0 0 Yes Viability
TNF TNFalpha 0 0 0 YesIGF IGF1 0 0 0 YesBOMB BOM 0 0 0 YesfMLP fMLP 0 0 0 YesIFN IFG 0 0 0 YesLeukotriene LTB4 0 0 0 YesNeurokinin NKB 0 0 0NGF NGF 0 0 0 Yes
Summary of Ligand Screen Responses
UCSD-Bioinformatics & Systems Biology Group
Reconstructing Networks
Legacy Data AfCS Data
Protein Interactions Microarray Data
Biochemical Pathways Yeast Two-Hybrid Data
RNAi Data
Protein Data
Perturbation Data
Microscopy Data
Published Literature
Reconstructing Networks
UCSD-Bioinformatics & Systems Biology Group
from Downward, Nature, August (2001)
Signal Transduction in a Cell
UCSD-Bioinformatics & Systems Biology Group
• B cell samples prepared by Cell Lab (Dallas).• Cultured for different time periods (.5, 1, 2, and 4 hr) in
the presence or absence of ligands before harvesting for total RNA isolation.
• Treated and untreated time-course samples hybridized against a spleen reference.
• After removing the common spleen denominator, comparison to 0 time point data reflects the changes in mRNA levels due to ligand treatment and/or time in culture.
• One of the largest mammalian array sets (33 ligands).• All of the experiments were done in triplicate. Including in controls >450 arrays (Caltech)
Ligand Screen Transcript Analysis
UCSD-Bioinformatics & Systems Biology Group
The mitogenic response from the ligands AIG, 40L, I04, LPS, CPG dominate at the center of the plot. This is too dense for a clear view (see histogram to the left).
IF, GRH, CGS, PAF, TGF, M3A, 2MA also showed a significant gene response.
Graph association map (4hr)
Differentially expressed genes for ligands vs UNTREATED @ 4hr [ SAM ; False Discovery Rate ( ) ]
ligand (4hr)
40L
(1%
)
LPS
(1%
)
AIG
(1%
)
IL4
(1%
)
CP
G (1
%)
IFB
(1.5
%)
GR
H (1
%)
2MA
(18%
)
LPA
(17%
)
CG
S (2
.9%
)
BOM
(35%
)
IGF
(8%
)
S1P
(38%
)
PAF
(2.4
%)
70L
(6%
)
NP
Y (1
0%)
DIM
(9%
)
LB4
(23%
)
M3A
(3.5
%)
FML
(11%
)
TGF
(2.5
%)
TER
(35%
)
IL10
(20%
)
ELC
(26%
)
PG
E (1
1%)
BAFF
(11%
)
BLC
(57%
)
NG
F (4
2%)
TNF
(33%
)
SDF
(20%
)
IFG
(25%
)
NE
B (2
5%)
SLC
(NA
)
num
ber o
f gen
es (p
robe
s)di
ffere
ntia
lly e
xpre
ssed
0
50
100
150
200
500
600
700
800
900
1000
1100
down-regulated up-regulated
UCSD-Bioinformatics & Systems Biology Group
Similarity measures between genes under different conditions with respect to expression levels for…
… groups of genes clustering methods
… pairs of genes correlation methods
Linear correlation (x – xmean) (y – ymean) [ (x – xmean)2 (y- ymean)2 ]½
Partial correlation
= r2 xy
= r xy.z
r xy - r
xz r yz
[(1- r2xz
) (1- r2yz
)]½
“marginal” global correlation (for ligand j ) r2
all xy - r2 all xy except ligand j
UCSD-Bioinformatics & Systems Biology Group
Two-way hierarchical cluster:
mean ratio (vs control) of phosphoprotein levels and ligand
Several ligands that elicit an ERK response (chemokines + AIG, CD40L) clustered together.
UCSD-Bioinformatics & Systems Biology Group
ERK-MAPK p38 JNK-SAPK
STAT1 P53
ETS.v5
N.MYC1 NFATC1ETS.v6
MEF2C Gadd45aH3F3A
CHOP Gadd45bCREB1
Max Gadd45gC.FOS
Bcl2l11 Egr1H3F3B
Bcl2l2 CHOPSocs3MaxCREB3
JUN JUNSTAT1Egr1 C.FOSN.MYC1
Bcl2l11Bcl2l2
Diagrams are from …
“Mitogen-Activated Protein Kinase Pathways Mediated by ERK, JNK, and p38 Protein Kinases”
G. L. Johnson and R. Lapadat Science 2002 December 6; 298: 1911-1912. (in Review)SRF
Three main pathways of MAPK and their respective target genes and transcription factors.
UCSD-Bioinformatics & Systems Biology Group
Level plots “Marginal” correlation of genes in MAPK pathways
UCSD-Bioinformatics & Systems Biology Group
“marginal” global correlation (for ligand j )
difference in correlation = r2
all xy - r2 all xy except ligand j
Red indicates positive influence on the gene upon removing ligand j
Green indicates negative influence on the gene upon removing ligand j
Highly responsive genes from MAPK-ERK pathway
B cells respond to AIG through the MAPK-ERK pathway.
UCSD-Bioinformatics & Systems Biology Group
We see the correlation results of removing ligands CD40L (40L) and interleukin 4 (I04) separately from the pool of 33 ligands. The colors red and green refer to decreases/increases in the subsequent correlation similarity matrix respectively. The absolute differential effects are almost uniform across CD40L (with a slightly smaller marginal difference from the ERK related genes h3f3b, ets-v6,c-fos), in contrast to interleukin 4 which shows darker shades, with the color black showing no differences, except for a few p38 (chop, jun) and JNK-SAPK (gadd45q) related genes.
lesser effect in ERK pathway than AIG
cytokine stress-related genes
UCSD-Bioinformatics & Systems Biology Group
B cells do not show any response to NGF but respond to LPS. Note: LPS has more response genes in p38 & JNK-SAPK than ERK.
No marginal changes in the pairwise gene correlations in the MAPK pathways from the addition or subtraction of this ligand NGF.
UCSD-Bioinformatics & Systems Biology Group
Positive pairwise correlation was more positive by the
additional ligand
Negative pairwise correlation was less negative by the
additional ligand
Positive pairwise correlation was more negative by the
additional ligand
Negative pairwise correlation was less positive by the
additional ligand
Marginal Correlations Connection Maps for MAPK Pathways
40L
Legendtarget genes only
transcription factors
This shows the marginal changes [eg edge threshold =0.1] in the significant pairwise correlation [95% confidence interval for the Fisher transformed distribution] between genes after the addition of the four timepoints of a particular ligand [40L] to the low, intermediate-response ligands (n =112, 28 ligands).
UCSD-Bioinformatics & Systems Biology Group
Positive pairwise correlation was more positive by the
additional ligand
Negative pairwise correlation was less negative by the
additional ligand
Positive pairwise correlation was more negative by the
additional ligand
Negative pairwise correlation was less positive by the
additional ligand
Marginal Correlations Connection Maps for MAPK Pathways
Legend
AIG
target genes only
transcription factors
This shows the marginal changes [eg edge threshold =0.1] in the significant pairwise correlation [95% confidence interval for the Fisher transformed distribution] between genes after the addition of the four timepoints of a particular ligand [AIG] to the low, intermediate-response ligands (n =112, 28 ligands).
UCSD-Bioinformatics & Systems Biology Group
Positive pairwise correlation was more positive by the
additional ligand
Negative pairwise correlation was less negative by the
additional ligand
Positive pairwise correlation was more negative by the
additional ligand
Negative pairwise correlation was less positive by the
additional ligand
Marginal Correlations Connection Maps for MAPK Pathways
Legend
LPS
target genes only
transcription factors
This shows the marginal changes [eg edge threshold =0.1] in the significant pairwise correlation [95% confidence interval for the Fisher transformed distribution] between genes after the addition of the four timepoints of a particular ligand [LPS] to the low, intermediate-response ligands (n =112, 28 ligands).
UCSD-Bioinformatics & Systems Biology Group
Cell Cycle
Kohn Map
UCSD-Bioinformatics & Systems Biology Group
UCSD-Bioinformatics & Systems Biology Group
Subcluster of“mitogenic” ligand(late time periods)
Myc box genes (cell cycle)
UCSD-Bioinformatics & Systems Biology Group
MYC Connection Map
Ap2a1Ap2a2Arhgef6
Cdc16
Cdc42ep3
Cdc42ep4.
Cdc42ep5
Cdgap
Cdh11
Cks1
Clk
Clk2Clk3 Clk4 Mad4
Max
Mga
Mina.
Mycbp
Rb1
Rbbp9.
Rbl1
Spec1
Srprb..
Tcfap2aTcfap2b
abc
d
e
f
g
h
i
j
kl
m n op
q
r
s
t
u
v
w
xy
z
Genetic regulatory module generated by partial correlations critical value = 10-6
UCSD-Bioinformatics & Systems Biology Group
Syk A000040Lyn
A001441
CD22 A000542
Immunoglobulin B-cell receptor
PIP3 A003330
CaM A000452
NFATA000024
cytosol
nucleus
some known Immediate-early
genes /transcription
factors
c-Jun A001300Max
A001480
NFB A002936-7
Bclxl A000373
Oct-2 A002281
Bfl-1
some known transcription regulators
SMAD3 A002175
p53 A001721
c-Fos A000404
PPARA001888 ETS-1 A000889
ER-a A000884
Bcl-6 A000369
Egr-1A003269
ATF-2A000347
CHOP A000655MEF2c
A001503
c-Myc A001568
CREB A003216
Stat1 A002230
SRF A002147
NFATc1A000024
Elk-1 A000830
phosphorylated
Ras A000004
Grb2 A001088Sos
A002030-1
PLC-2 A001809-10
Blnk A000381
DAG A000764-7
PKC A001919-
Endoplasmicreticulum IP3
A000077
Ca++
Rap A002022
Btk A000038
Shc A002150-3
ERK 1-2 A000874-5
p38 A001717
JNK1 A001296
Vav A002360-2
Rac1 A002001
calcineurin A000420-4
Rho
FcRIIB
Shp A2156-7
CD45 A000567
Akt A000249
PTEN A001941
GSK-3 A001105-6
PKC - A001920
IKK A001170
TRAF3 A002309
TRAF2 A002308
TRAF5 A002311
dimer
CD40 Toll-like 4receptor
MyD88 A003535
dimerJak1
A001290
Interleukin 4R
dimer
STAT6 A002236
Plasma membrane
Shp A2156-7
MEK1/2 A001512-3
Raf1 A002008
Growth proliferation signals
Cyclin D1 /cdk 4,6 A000721 A000608,10
Rb A002035
p130 A002038
Rb - P p130 - P
E2F1 A002962
E2F2 A002963
E2F3a A002964
Cyclin EA000724 / Cdk2 A000602
S Phase genes
CKIs
E2F4,5-p130 A002965-6 E2F3b-Rb A002964
p19ARF
A001711
Mdm2 A001498
p53
Cytochrome Crelease
DNA damage
ATMA000349 /ATR
Chk2
Apaf1 A000302
p73 A002935 Apoptosis
Caspase 9
Survival signals
Bad
A000358
MDM2 A001498
Bclxl A000373
GDP
s
GTP
iGTP
G-protein-coupled receptor
PYK2
cAMP
qGTP
PKA A0
0
GTP
PI3K A0
GDP
PLC- A0
+others
Dennis Mock-UCSD
Connection matrix cytosol only
Signaling pathways of primary B cell (mouse)