Regulatory Networks in the Drosophila Blastoderm

1
Regulatory Networks in the Drosophila Blastoderm

• John Reinitz (USB group)
• Maria Samsonova (SPb group)
• David H. Sharp (LANL group)

2
Blastoderm Systems Biology Three Central Problems
1. Determination of a Morphogenetic Field
Cells interact to create blueprints for body
parts (leg, wing, body segments and so on).
Building these body parts takes place later
(Differentiation). Determination requires
differential gene expression in space. Hence
understanding determination leads to
consideration of 2. The fundamental problem of
metazoan molecular genetics How do cells
containing the same genetic material come
to express different genes in a precise spatial
pattern? The above question concern pattern
formation in space. Implementation at the
molecular level implies a third question 3. How
is transcription of key developmental
genes controlled by binding sites, groups of
which form modular enhancers?
3
(No Transcript)
4
(No Transcript)
5
Maternal Coordinate Genes
bicoid (bcd) caudal (cad) hunchback (hb)
hunchback (hb) Krüppel (Kr) knirps (kni) giant
(gt) tailless (tll)
Gap Genes
even-skipped (eve) odd-skipped (odd) hairy
(h) runt (run) fushi-tarazu (ftz) paired (prd)
Pair-Rule Genes
En
Segment Polarity Genes
engrailed (en) wingless (wg)
6
14 segmentation genes are expressed in the
blastoderm. Each has a distinct pattern
of expression.
7
In addition, each expression pattern
changes over time.
8
The Gene Circuit Approach

• Formulate a theoretical model
• Obtain gene expression data
• Perform optimization to fit model to data
• Learn some new biology

9
1. Formulate a theoretical model
10
(No Transcript)
11
Synthesis
Transport
Decay
12
Synthesis
The regulation-expression function g(u)
1.0
Transport
Relative Activation g(u)
Decay
0.0
-6.0
0.0
6.0
Total Regulatory Input u
13
Synthesis
Genetic Interconnectivity Matrix (T)
b
Gene

1
2
N
a
T parameters

1
T11
T12
T1N

positive negative zero
activation repression no interaction
2
T22
T21
T2N





N
TN1
TN2
TNN
14
2. Obtain gene expression data
15
We start with many confocal scans of embryos.
(of which these are the first 509)
16
(No Transcript)
17
Integrated Patterns of Segmentation Gene
Expression (Time Classes 3 and 8)
Myasnikova et al (2001), Bioinformatics 173-12
18
A surprise posterior domains move to the
anterior.
Computation of shifts 1. calculate the
average position of a given characteristic
feature over the embryos of each time class. 2.
compute the difference between average positions
of the feature in different time classes
19
3. Perform optimization to fit model to data
20
Chu et al (1999) J.C.P. 148646
21
4. Learn some new biology
22
Analysis of the Gap Gene System
Using the new dataset and the model with
nuclear divisions, we ask How are gap gene
domains maintained and refined?
Why do they move?
Nature (2004) 430368 Genetics (2004) 1671721
23
Gap Gene Expression
Data
Model
255
T1
T1
Hb
Kr
Rel Prot Conc
Kni
0
24.225
Kr
Time (min)
Kni
Gt
67.975
255
T8
T8
Gt
Rel Prot Conc
Tll
0
40
50
60
70
80
90
40
50
60
70
80
90
A-P Position ()
A-P Position ()
24
Constraints on the T Matrixbased on analysis of
gn58c13_goodies
Activation
No Interaction (cutoff 0.01)
Repression
Weak Constraint
N 10
25
Synthesis
Transport
Decay
26
Synthesis
Dynamic dissection
We can look at individual parts of this sum to
dissect the various regulatory contributions
on a specific gene. For example Thb Kr vKr
represents Krs regulatory input on hb
27
Kr Activation
Kr
Bcd
Cad
250
T8
200
150
Rel Prot Conc
100
50
0
4.0
0.0
Reg Contrib
-10.0
-17.0
90
40
50
60
70
80
AP Position ()
u
Bcd
Cad
Kr
28
Kr Repression
Kr
Gt
250
T8
200
Rel Prot Conc
150
100
50
0
4.0
0.0
Reg Contrib
-14.0
90
40
50
60
70
80
AP Position ()
u
Gt
29
Hb
Hb
Kni
Tll
Gt
Gt
Kr
Bcd
Cad
30
Domain Shifts Mechanism
Kr
Kni
Gt
21.1
10
8
30.0
6
40.0
Time (min)
4
50.0
2
0
60.0
-2
71.1
90
50
60
70
80
40
90
50
60
70
80
40
90
50
60
70
80
40
A-P Position ()
A-P Position ()
A-P Position ()
31
Spatial Dynamics Domains of Synthesis and Decay
Kr
Kni
Gt
21.1
10
8
30.0
6
40.0
Time (min)
4
50.0
2
0
60.0
-2
71.1
90
50
60
70
80
40
90
50
60
70
80
40
90
50
60
70
80
40
A-P Position ()
200
T5
3.0
100
Rel Prot Conc
Kr (prot conc)
0
0.0
Kr (dv/dt)
-3.0
90
50
60
70
80
40
AP Position ()
32
Gap Domains RNA vs Protein
Kr
kni
gt
30
50
70
45
60
60
75
80
100
A-P Position ()
A-P Position ()
A-P Position ()
RNA
RNA
Protein
Protein
33
Domain shift conclusions

• Posterior domains shift because of regulative
  cross-interactions.
• Anterior domains form in place under the
• control of maternal positional information.
• Thus, a mosaic mode prevails in the anterior and
  a regulative mode in the posterior.

34
Part 2 TranscriptionOutline of the problem,
and perhaps a partial solution

• The Big Picture
• Classic CRMs (enhancers)
• Anomalous Behavior of CRMs
• Quantitative Expression Data
• 4. A Computational Model

35
Prokaryotes

• Transcription is two reactions
• RNAP Prom ?? RNAP.Prom
• RNAP.Prom ? elongating RNAP
• Regulation involves no more than
• 1 to 6 additional proteins/binding
  sites.
• Regulation can be faithfully reconstituted in
• an assay using ONLY pure substances
• (e.g.) Hawley and McClure (1982)
  J.Mol. Biol. 157493

36
Eucaryotes/Metazoa

• Biochemistry The transcription complex
• contains over 100 polypeptides (!)
• DNA Genes have control regions of up
• to 50-100 kb, with 1000s of binding sites
• Chromatin The DNA is embedded in an
• active (dissipative) cellular organelle
• which hides or exposes these sites.
• Each of these mechanisms is best studied
• in a different experimental system.

37
In the Drosophila blastoderm
All genes expressed before gastrulation are in a
uniform state of chromatin competence in
all nuclei. Chromatin regulation can be
neglected. Only a small number of transcription
factors vary spatially. They are the same
ones that regulate development.
This is an excellent system to test what is now
the central unifying hypothesis of metazoan gene
regulation The control regions of metazoan
genes are divided into separable
cis-regulatory modules (CRMs).
38
CRMs (cis-regulatory modules, or enhancers)

• Are short contiguous segments of DNA.
• Contain clusters of binding sites for
• transcription factors.
• 3. Can act at a distance from the basal complex.
• 4. Can act in normal or reversed orientation.
• 5. Can act independently of one another.

CRMs are assayed by splicing different fragments
of the control region to a constant basal
promoter region and physiologically inert gene,
typically E. coli lacZ.
39
CRMs (cis-regulatory modules, or enhancers)

• Are short contiguous segments of DNA.
• Contain clusters of binding sites for
• transcription factors.
• 3. Can act at a distance from the basal complex.
• 4. Can act in normal or reversed orientation.
• 5. Can act independently of one another.

HYPOTHESIS (Gray et al., Phil. Trans. Royal
Society, 349257, 1995) Independence of CRMs is
ensured by short range (150 bp) repression
repressors bound to binding sites in a given CRM
are too far away to affect activators in a
separate CRM
40
even-skipped genomic region
-3.9
-3.3
1
eve
MSE3
MSE2
ftz-like
Stripe 46
Stripe 1
Stripe 5
Stripe 7
-3.9
-3.3
-1.5
-1.1
1.5
2.6
4.5
5.2
6.6
7.4
8.2
-1.7
-0.9
41
Anomalous Behavior 1 Nonlocality of CRMs
(Incomplete regulatory sequences)
MSE3 lacZ in wild type embryo (Small, Blair
Levine 1992, EMBO J. 114047)
MSE3 lacZ in knirps mutant embryo (Small, Blair
Levine 1992, EMBO J. 114047)
Native eve in knirps mutant embryo (Frasch and
Levine 1987, GD 1981-995)
42
Anomalous Behavior 2 Juxtaposing2 CRMs Gives a
Novel Pattern
43
What do we need in order to understandthis
behavior?

• Quantitative expression data is needed to
  understand
• quantitative changes.
• We need a method for calculating the repressive
  and activating effects of transcription factors
  bound at
• all sites, not just minimal elements.
• The calculation method must interface with
  available
• experimental manipulations (insertion,
  deletion, site-directed mutagenesis).

44
c14 T2
c13
lacZ mRNA
Eve protein
Rel Prot Conc
c14 T3
c14 T5
c14 T4
Rel Prot Conc
c14 T6
c14 T7
c14 T8
250
Rel Prot Conc
A-P Position ()
45
Cartoon of a Model
Activation. Scale 10kb - Repression by
Competition. Scale 20bp - Repression by
Quenching. Scale 100-150bp - Direct Repression
of the Basal Promoter. Scale 100-150bp
46
The Model Can Represent (and Predict?) the
Effects Of Mutation, Deletion, and Insertion.
47
Acknowledgments
Stony Brook
St. Petersburg
John Reinitz Carlos Alonso Jean Cadet Lucas
Carey King-Wai Chu Lorraine Greenwald Dave
Kosman Manu Alexander Spirov
Maria Samsonova Alexander Samsonov Maxim
Blagov Vitaly Gursky Konstantin Kozlov Dimitry
Malashonok Ekaterina Myasnikova Andrei
Pisarev Ekaterina Pustelnikova Anastassia
Samsonova Svetlana Surkova
Los Alamos
Dave Sharp Shuling Hou
http//flyex.ams.sunysb.edu/flyex http//urchin.sp
bcas.ru/flyex