Christopher Voigt

1
Robustness and Evolvability of Biological Systems
Christopher Voigt
Recruitment Weekend 2003
2

• Wide catalytic range
• High specificity / selectivity
• Low energy input
• Green chemistry

• Low activities for non-natural reactions
• Marginally stable
• Harsh industrial conditions
• Low expression
• Ulterior motives

3
Directed Evolution
Wild-type DNA Sequence
ligate into plasmids
PCR
1,000 to 1,000,000 mutants
top mutant becomes parent of next generation
transform into cells
Less-fit mutants are discarded
screen for improvements
4
Evolvability

• In a dynamic environment, the lineage that
  adapts first, wins
• Fewer mutations means faster evolution
• Are some biosystems constructed to minimize the
  mutations required to find improvements?

5
Questions
6
Sequence Space
300 residues 20 amino acids
3 residues 2 amino acids
7
Folding, Inverse Folding, and Robustness

• GOAL Computationally determine the sequence
  space volume a protein structure occupies

8
Sequence Entropy

• How many sequences exist at energy E?

• Sum of residue entropies

• Amino acid probabilities

S(E)

• Probabilities are calculated using an energetic
  model and mean-field theory

energy E
Sequence Space
(Saven Wolynes, J. Chem. Phys. B, 1997)
9
T4 Lysozyme Mutagenesis Data
(Poteete, 1991)
10
Directed Evolution Experiments

• Low mutation rates are optimal for small
  libraries
• Single mutations confer improvements
• These mutations are biased towards tolerant
  residues

Subtilisin E

• Structural robustness improves functional
  evolvability

(Arnold and co-workers, 1993, 1994)
11
Sexual Recombination
12
Schema Theory

• Schema (or building blocks) are clusters of
  bits that interact

• Optimal crossover points minimize the number of
  schema that are disrupted

• Schema ARE preserved with this partition

• Schema ARE NOT preserved with this partition

13
Schema Disruption

• If a structural domain contains many internal
  interactions, a crossover within the domain is
  likely to be disruptive
• Count the number of broken interactions

Hybrid 2
Hybrid 1
Es 0
Es 3

• Two residues are considered interacting if
  their side-chains are within a distance cutoff
  (4.5 Å)

14
Beta-lactamase Library

• Two beta-lactamase variants (TEM-1 and PSE-4)
  are recombined
• Amino acid sequence alignment 40
• Structural alignment 0.97 RMS
• Constructed a library of 16384 hybrids using 13
  crossovers
• Can select for antibiotic resistance

15
Minimizing Disruption

• Unselected Hybrid

• Functional Hybrid

16
Evolution of Genetic Circuits
Genetic circuits control cellular behaviors
response pathways, differentiation, chemical
production, cell-cell interactions
Dynamical models of genetic circuits

• How do architectures confer robustness and
  evolvability?
• Predict attainable variability and evolution
  strategies
• Which nodes are the control points?
• How is information processed by the cell?

Develop experimental systems to test and refine
techniques
17
Stress response in B. subtilis
INPUTS
OUTPUTS
18
Control of Sporulation Initiation
Transcription Factor
concentration
R4
time
Stage II genes
SIN switch
R
R
R
R
R
R
R
R
spoIIA
R
R
spoIIE
P1
P3
P3
sinR
sinI
sinR
spoIIG
19
Control of Sporulation Initiation
Transcription Factor
Stage II
concentration
spoOAP
Bad Times
P
R4
time
Stage II genes
SIN switch
R
R
R
R
R
spoIIA
R
spoIIE
P1
P3
sinR
sinI
spoIIG
20
The Model
Exhaustive Parameter Sampling
Parameter Space
21
Contact Me CAVoigt_at_lbl.gov
Acknowledgements
Carlos Martinez Michelle Meyer Jonathan Silberg
Caltech
Prof. Zhen-Gang Wang Prof. Stephen Mayo
Prof. Frances Arnold
Prof. Adam Arkin
UC-Berkeley
Denise Wolf Naum Pfleger
Support
National Science Foundation Burroughs-Welcome The
Santa Fe Institute Sloan / Department of Energy