Deciphering genetic regulatory networks

1
Deciphering genetic regulatory networks

• Tzachi Pilpel
• Department of Molecular Genetics
• http//longitude.weizmann.ac.il

January 2005
2
Mapping DNA sequence onto RNA expression patterns
.ACGTTGGATTGATACGAGCAGTGACAGATCAGACGATAGACAGATACA
GATACACCCCAGAGTGACAGATCAGACGATAGACAGATTGACAGATCAGA
CGATAGACAGATTGACAGATCAGACGATAGACAGATTGACAGATT..
mRNA abundance
Time
3
Transcription regulation
Pol2
RNA
4
Recent news RNAs also regulate genes
Sense transcript
Anti-sense transcript
5
Overlapping sense and anti-sense transcripts
regulatory region
Protein-coding (sense)
non-protein-coding (anti-sense)
regulatory region
From EST libraries and coding capacity analysis
we gathered 1634 partially overlapping sense
(coding) anti-sense (non-coding) pairs In the
human genome
Ophir Shalem
6
Upstream of antisense is promoter-like
Sense
Anti-sense
66
50
GC
GC
48
44
500
-1000
1
500
-1000
1
7
A computational scheme for identify regulation
of sense and anti-sense
regulatory region
Protein-coding (sense)
non-protein-coding (anti-sense)
regulatory region
8
A co-regulation matrix
Regulators (ordered by similarity)
STAT5A PITX2 T3R IRF1 Sp1 Nkx6-2 AP-2 E2F C/EBP HO
XA4 EGR TBP GCM p53
P-value on the hypothesis that regulator i and j
co-regulate sense anti sense pairs
Regulators (ordered by similarity)
9
The p53-mdm2 switch
p53
mdm2
Protein degradation
10
Oren, Biderman, Minsky
11
A proposed twits on the switch topology
p53
mdm2
Anti(mdm2)
Sense-anti sense inhibition/degradation
Protein degradation
12
Differential sense and anti-sense binding
efficiencies
Predicted difference between sense and
anti-sense binding
Ordered list of transcription factors
Merble, Shen-Orr, Shalem, Shalgi
13
A possible buffering role of the anti-sense
transcript
Sense transcription
Transcript concentration
Anti-sense level
Time
14
A network of regulators and regulatees
Sense-transcripts
Anti sense-transcripts
Micro-RNAs
Transcription factors
RNA-inhibitory interactions
15
Background and Motivation

• Most genes in yeast and worm are non-essential
• Non-essentiality is larger for genes that have
  duplicates
• Yet duplicates are inherently unstable
• Stable duplicate maintenance may require
  divergence

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Background and Motivation

• Most genes in yeast and worm are non-essential
• Non-essentiality is larger for genes that have
  duplicates
• Yet duplicates are inherently unstable
• Stable duplicate maintenance may require
  divergence

10
90
20
Open questions

• How could backup by paralogs evolve?
  (evolutionary time scale)
• What controls the utilization of backup?
  (individual life time scale)

21
Ancient paralogs backup if dissimilarly
expressed Recent one - if similarly expressed
1
0.9
0.8
0.7
0.6
Backup capacity
0.5
0.4
0.3
Ran Kafri Arren Bar-even
Mean expression similarity
22
Reprogramming in Acs1/Acs2
D Acs2
Wild-type
Glucose
Glucose
Acs2
Acs1
Acs1
Acs1
Acs2
23
Partial motif content overlap is optimal for
backup
1
0.8
Proportion of dispensable genes
0.6
0
0.2
0.4
0.6
m1n m2
Motif content overlap (O)
O
m1 U m2
Motifs sources Harbison (Nature 2004) Kellis
(Nature 2003) Pilpel (Nature Genet. 2002)
24
Models prediction highly flexible pairs should
be up-regulated following counterparts deletion
10
9
8
7
6
Fold change
(Hughes et al. Cell 2000)
5
4
3
2
1
0
lt0.35
gt0.45
0.35 0.45
Partial co-regulation (predicted backup capacity)
25
The reprogramming switch
T concentration of transcription factor G
Transcription factor activity E concentration
of enzymes K binding constants
26
The reprogramming switch (cont)
G1 knockout
M2
E2
E1
Kafri, Bar-Even, and Pilpel, Nature Genetics
March 2005 Hurst and Pál, Nature Genetics March
2005 'Highlights section Nature Reviews Genetics
March 2005
27
Acknowledgments

• Backup and reprogramming
• Ran Kafri
• Arren Ben-even
• Ilya Vegner
• Shachar Shachmon
• Sense anti-sense regulation
• Ophir Shalem
• Shai Shen-Orr
• Yifat Merble
• Reut Shalgi
• Moshe Oren
• Neri Minsky
• Lynn Biderman

• Motif dictionaries
• Michal Lapidot
• Reut Shalgi
• Control of cellular transformation
• Yuval Tabach
• Varda Rotter
• Eytan Domany

28
Dispensability quantification on yeast genes
.
.
Rich medium
High salt
Low pH
mean
DG1
2.5
DG2
3.8
Non-essential in rich medium
DG3
.
0
Essential in rich medium
.
.
mean
Steinmetz et al., Nat Genet 2002
29
The labs high performance linux cluster
x10x2
x6x2
x4
36 processors
Shai Kaplan
30
Decomposing the mapping problem
Rubinstein
Barkai
Rotter, Domany
Oren
31
Responsive circuit design
glucan syntheases substrate regulation
Hexose transporters end-product regulation
32
Myogenesis

• MyoD deletion - 4 increase Myf-5
• MRF4 deletion ? 4 fold increase of myogenin

Zhang W, Behringer RR, Olson EN. Genes Dev. 1995
Jun 19(11)1388-99.
33
Responsive Backup Circuits Function
Sum Gate
Alon U. Kalir S. Cell. 117, 713-720 (2004)
34
Responsive circuit Acs1 Acs2
D Acs2
Wild-type
Glucose
Glucose
Acs2
Acs1
Acs1
35
Responsive circuit
Deletion of Hxt1 ? up-regulation of Hxt2 (in high
glucose)
Hxt1
Hxt2
Mutation in Hxt1
ref
36
Responsive circuit
Deletion of Hxt1 ? up-regulation of Hxt2 (in high
glucose)
Hxt1
Hxt2
ref
37
Extracellular glucose
Hxt2
Hxt1
Snf3
intracellular glucose
38
Direction of asymmetric backup can be inferred
from its regulation
0.2
0
n. of regulators
-0.2
-0.4
D
-0.6
-0.8
0
0.2
0.4
0.6
0.8
1
1.2
D
dispensibility
39
gene1
gene2
40
(No Transcript)
41
(No Transcript)
42
Backup reprogramming among transcription factors
MyoD
Myf5
genes
Mammalian myogenesis
E.Coli nucleic acids metabolism
43
Two evolutionary regimes
Sub-functionalization
Neo-functionalization
1
0.9
0.8
0.7
0.6
Backup capacity
0.5
0.4
0.3
Mean expression similarity
44
Genes with duplicates are less essential
0.6
0.4
proportion
0.2
0
strong effect
lethal
moderate effect
Weak effect
Gu, Z. et al. Nature (2003)
45
Single gene deletion experiments show that most
genes are dispensable
Yeast (knockout)
Worm (RNAi knockdown)

Essential
Essential
10
27
Dispensable
Dispensable

• (Winzeler nature 1999, Kamath Nature 2003)

46
A naked eye view of a genome
CAGAATATAGGAGAGGAGTAGATCAGATACGGGATACAGATAGGATCAGA
GAGAGGGCATAGGGCAGCGGGTATAGAGACAGATAGACAGTTAGAGTAGA
CAGAATATAGGAGAGGAGGGGGATATAGTAGC
GATCAGAGAGAGGGCATAGGGCAGCGGGTATAGAGATCAGAGAGAGGGCA
TAGGGCAGCGGGTATAGAGACAGATAGACAGTTAGAGTAGACAGAATATA
GGAGAGGAGGACAGATAGACAGTTAGAGTAGACAGAATATAGGAGAGGAG
TAGATCAGATACGGGATACAGATAGGATCAGAGAGAGGGCATAGGGCAGC
GGGTATAGAGACAGATAGACAGTTAGAGTAGACAGAATATAGGAGAGGAG
GGGGATATAGTAGC
47
Assessment of regulatory motif content overlap
Assessment of regulatory motif content overlap
Motif Set1
Motif Set2
m1n m2
O
m1 U m2
48
Dictionaries of functional sequence motif
mRNA decay profiles
EC0.05
expression level
Mean ½ life 26 min
Mean ½ life 6 min
mRNA level
time
time
Naama Barkai
Protein Coding
Gene 1 TCATTGAAAGCTTCCCTTATCCGTGCCAGene 2
TCGAATACAACGCCTGAGGAGGACCTTTGene 3
GCACCATCCCTCCTACAATAACCTTCAGGene 4
TGAGCTCATTAAGCTTCCCAGCACACTT
Gene 5 GCACCATCCCTCCTACAATAACTACACGGene 6
TGAGCTCATTAAGCTTCCCAGCACAACT
Michal Lapidot Reut Shalgi
Jeff Gerst
49
mapping tumor-suppressive signals onto gene
expression
NFY CHR
0.3
0.25
0.2
expression level
0.15
0.1
0.05
L
H
p16
0
H
L
Rotter, Domany, Tabach
50
ATGCTAGTCAGTCAGATCAGGACAGTGCTGACTAGACGATGACAGATGAC
A
Define backup circuits. Study evolution their
and dynamics