Assigning numbers to the arrows: Parameterizing a gene regulation network by using accurate expressi

1
Assigning numbers to the arrows Parameterizing a
gene regulation network by using accurate
expression kinetics

• Michal Ronen, Revital Rosenberg, Boris I.
  Shraiman, and Uri AlonPNAS Aug. 6, 2002 Vol. 99
  p 10555-10560

BioNetworks Journal Club -- Sept. 9, 2002 Slides
prepared by Alison Hottes
2
Assigning Numbers to Arrows

• Need to go from the connectivity and topology of
  a network to its dynamics

• Dynamics
• How does the concentration of each transcription
  factor change temporally?
• How does a transcription factors level affect
  each of the promoters that it controls?

• Connectivity and Topology
• What genes are involved in each process?
• What transcription factor(s) controls each gene?

Protein
Promoter Activity
time
3
Basic Setup
4
DNA Damage Repair System

• Want to understand dynamics
• In what order do promoters turn on and off?
• When on, how active is a promoter?
• How do LexA protein levels vary?

5
Experimental System

• Create strains with GFP under the control of
  different promoters.
• Stable protein
• Assume get the same number of proteins from every
  transcript.
• Is GFP bleaching is a problem?
• How rapidly is GFP folding?
• Grow cells. Irradiate. Measure GFP and OD
  levels every 3 minutes for 2 cell cycles (90 min).

6
Data

• Smooth raw OD and GFP measurements
• Subtract off background fluorescence levels.
• Find promoter activity
• Smooth again

7
Finding Parameters

• Model the response as
• A(t) is the concentration of LexA (M timepoints).
  ß and k are parameters. There are N genes each
  with M timepoints.
• Measured NM elements. Need to fit 2NM
  variables.
• Measured error as
• Genes that dont respond to LexA have large
  errors. Most genes in process had errors of
  10-25.

8
Extensions

• Once ß and k have been determined for each gene,
  one can calculate the response of all of the
  promoters to new experimental conditions by
  measuring just one.

9
Measured and Modeled Profiles
Curves were predicted using the urvA promoter
only.

• Turn off order makes biological sense.
• Genes involved in early repair processes turn off
  before those involved in late processes

10
LexA Protein Levels

• Model, especially early on, agrees well with
  independent measurements.
• Authors were able to estimate protein levels.

11
Thoughts about method

• Could this be extended to a case where multiple
  transcription factors are active?
• How would the system be perturbed?
• What equation would be used as the model?
• Would this work for all transcription factors?
• Will the promoter on the low copy plasmid suck
  up so much transcription factor so as to affect
  the system?
• If a promoter is normally on, will too much GFP
  accumulate?
• Does the exact DNA sequence of a promoter give
  any hints about its kinetic parameters?
• Could this method be used to screen for promoters
  in a transcription factors regulon?
• Need to perturb the transcription factors levels

12
Comparison to cDNA Microarrays

• Reporter GFP strains must be synthesized.
• Must build system for making fluorescence
  measurements.
• Suited for timecourses
• Measure promoter activity (i.e. mRNA synthesis
  rates)
• Extra timepoints can be gathered with little
  additional work.
• Normalization and data processing issues.

• Must amplify DNA for each ORF.
• Need hardware for printing and scanning.
• Suited to comparing snapshots.
• Measure steady state mRNA levels
• Extra timepoints require a lot of additional
  work.
• Normalization and data processing issues.

13

• We need YOU to volunteer to lead a discussion.