Exploring gene regulatory networks

1
Exploring gene regulatory networks
Keith Shearwin Molecular and Biomedical
Sciences University of Adelaide

• - What numbers do we need to know
• How do we measure these numbers

2
The CII protein of phage 186 is essential for
entry into the lysogenic state. What experiments
can be done to (i) understand and (ii)
quantitate the simple CII interactions.
3
186 CII how does it work
Purified protein required
4
186 CII self association
Monomer-dimer-tetramer equilibrium. Potentially,
each species could bind DNA.
Shearwin Egan, JBC 2000
5
186 CII - stoichiometry of binding
Sedimentation analysis of CII-DNA mixture
(chromophore-labelled DNA) shows that the CII
dimer is the major species to bind DNA. (Could
also use gel filtration chromatography coupled to
a light scattering detector to determine
stoichiometry)
6
186 CII DNA binding energies (a)
Electrophoretic mobility shift assay (EMSA)
The DNA is labelled, and present at low
concentration. The protein-DNA complex moves more
slowly through the gel matrix than unbound DNA.
Titration with increasing amounts of protein.
Shearwin Egan, JBC 2000
7
186 CII DNA binding energies (a)
Electrophoretic mobility shift assay (EMSA)
Obtain values for binding energies
Shearwin Egan, JBC 2000
8
186 CII DNA binding energies (b) DNAseI
footprinting
Protein (CII) binding protects the DNA from
cleavage by the enzyme DNAseI. Titration with
increasing CII allows determination of binding
energies.
Shearwin Egan, JBC 2000
(c) Isothermal titration calorimetry (ITC)
Surface plasmon resonance (SPR) gt binding rates
9

• Determining the number of molecules in a cell
• Western blotting uses antibodies to detect
  specific molecules amongst total cell extracts.
• In a bacterial cell, one molecule 1nM.
• In a bacterial cell lysogenised by phage 186,
  there are about 700 molecules of CI repressor.

Dodd et al., Genes and Development 15, 3013 (2001)

• 186 CII concentration unknown. CII is very
  unstable with a half life of 2 min, measured
  using antibodies.

10

• How many of these parameters do we need to know ?
• Dependent on the modelling approach and the
  desired output of modelling exercise.
• Caution In vitro vs in vivo values.
• Bacteriophage Lambda
• Extremely well studied system, yet connections
  missing
• Long range DNA looping
• Simulation of promoter reporter data

11
2.4Kb
OL
OR
12
Lambda switch textbook model
lysogeny

• two sets of three operators bind CI or cro
• bistable switch
• CI protein binds as a dimer to each operator
• Cooperative interactions give CI tetramers on the
  DNA
• -Puzzle
• 1. Why have OL3?

13
In vivo single copy reporter assay
Controlled CI expression from a single copy
plasmid.
WLU wild type lysogenic units
14
Modified model of lambda switch
CI
Repressors at OR and OL interact to form an
octamer gt DNA loop. The octamer positions OR3
and OL3 to be occupied by a tetramer, repressing
pRM -gt extra level of cooperativity.
15
Simulation of reporter data
4
0
0
3
5
0
-OL
pRM lacZ activity
3
0
0
2
5
0
2
0
0
OL
1
5
0
1
0
0
5
0
0
0
.
0
1
.
0
2
.
0
3
.
0
CI WLU
Free energies for CI binding to all six operators
are known (DNAse1 footprinting). Pairwise
cooperativities, repressor dimerisation constant
known. Unknowns are (a) net free energy of
octamer formation (DGoct) (b) net free
energy of tetramer formation (DGtet)
Aim of simulations 1. Obtain estimates of DGoct
and DGtet. 2. Simulate effect of operator mutants
- consistent with model? 3. Predict order of
site filling.
16
Example repressor binding to OR
Eight possible states
-11.1
1. Free operator 2. Bound to OR1 3. Bound to
OR2 4. Bound to OR3 5. Bound to OR1 OR2 6.
Bound to OR1 OR3 7. Bound to OR2 OR3 8. Bound
to OR1, OR2, OR3
-3.0
-13.2
-10.7
-10.2
Free energies in kcal/mol Can calculate fraction
of each state as a function of CI concentration.
17
OR and OL - 64 possible states, 9 of these can be
either looped or unlooped, so 73 conformations. -
unknowns are free energies for OR1-OR2-OL1-OL2
octamer and OR3-OL3 tetramer. These are the
energies required to go from the unlooped to the
looped state. - compare simulations to
experimental data to obtain estimates of the two
unknowns.
18
1. Simulate in vivo pRM reporter data
DGoct -0.5 kcal/mol DGtet -3.0 kcal/mol
Requires adjustment of concentration scale to
align the in vitro and in vivo data, to account
for non-specific binding.
Octamer pays the cost of forming the loop and
allows the OR3-OL3 tetramer to form more easily.
19
2. Simulations can correctly predict effects of
mutations
Change only DGOR3
20
3. Simulations allow calculation of the order of
site filling
DGoct -0.5 kcal/mol DGtet -3.0 kcal/mol
In a lysogen, the system is poised to respond to
fluctuations in repressor levels.
21

• Summary
• Modelling can be extended to include effect of
  other proteins (Cro, RNA polymerase) on the
  lambda switch. (1200 states)
• Modelling whole phage network frequency of entry
  into each state, stability of states, switching
  between states. (Stochastic)
• Question How many parameters do we need to
  know, and which are the most important.

22
(No Transcript)
23
(c) Isothermal titration calorimetry
Measures directly the amount of heat (q) absorbed
or released upon the mixing of two components. q
(area under peak) is proportional to amount of
ligand bound. Can obtain binding affinity,
stoichiometry, enthalpy, entropy in one
experiment.
24

• (c) Surface plasmon resonance (Biacore)
• - rates of interaction gt binding energies

25
Size exclusion chromatography with light
scattering detection
26
Looping cooperativity is critical for negative
feedback
Dodd et al. (2004) Genes and Development 18,
344-354.
27
Thermodynamics of looping
DGoct represents a balance between the favourable
association of DNA-bound protein (DGprotein) and
unfavourable DNA looping (DGDNA).
DGoct DGprotein DGDNA
Estimates of values DGprotein (in solution)
-9.1 kcal/mol DGDNA for 3.8kb loop (calculated)
11.3 kcal/mol (mostly entropic) DGoct 2.2
kcal/mol compared to 0.5 kcal/mol from
simulations So either (i) CI forms octamers from
tetramers more favourably when bound to DNA than
in solution or, (ii) forming the DNA loop is not
quite so unfavourable in vivo as the in vitro
calculations suggest (or both). Supercoiling
28
Intact cI Binding to Adjacent Operator Sites
29
Predictions from modified model
(1) Mutations at either OR3 or OL3 should reduce
repression of pRM, and so increase lysogenic CI
levels.
CI levels are 3 fold higher in OR3- (or OL3-)
lysogens (Western blots)
30
Predictions
(2) Higher CI levels make the switch from
lysogeny to lytic development less efficient
(there is more CI to inactivate)
OR3
Repression of pRM by CI is important for the
phage
Dodd et al. (2001) Genes and Development 15, 3013
3022.