Metabolic Engineering With RetroPath
Antti Koistinen, Nguyen Hoang Vu
1. RetroPath and XTMS server Review
2. Design a pathway for Rosmarinic acid
Retropath Review
- Based on Retrosynthesis: do a reverse transformation from target to precursor.
- Involve 2 main steps: Enumerate and Rank pathways
Biologist: Hey dude, tell your algorithm to find a pathway from target to precursor for me!
CS: Ok then I will do a graph search, with nodes are chemical compounds and vertex are reactions.
Biologist: Enough talking. Here is the reaction, stuff it into your algorithm!
CS: Ok my algorithm would not understand your drawing, I am going to present it by molecular
signature, with parameter height. How big do you want your height to be?
Biologist: Well I am 1m7. Would that fit into your algorithm?
Reaction
vs
Molecular signature
Retropath Review
1. Enumeration
- Different h values correspond to different
specificity.
- By tuning h, Retropath achieve reasonable tradeoff between complexity and specificity.
Retropath Review
2. Ranking:
- Thermodynamic feasibility: assess Gibbs free energy of reactions.
- Enzyme promiscuity: Find the enzyme for each reaction, if no known enzyme is
available → Propose putative ones.
- Enzyme performance: using kinetics information from BRENDA database
- Gene compatibility with host chassis
- Compound/intermediates toxicity
- Nominal fluxes: estimate maximize production while keeping cell growth
NOTE: Experimental data are generally lacking → Machine learning to the rescue.
Retropath Review
3. Weight optimization: which score is most important?
- Unweighted average of every score does not make sense
→ Determine weights for each score by optimization based on annotated pathways
→ Total score is used to rank pathway
4. Validation:
- Also based on annotated pathways
Rosmarinic acid
• Secondary metabolite produced in plant cell cultures
• 19% of the cell dry weight
• Many beneficial biological activities
•
•
•
•
Prevention of development of Alzheimer’s disease
Cancer chemopreventation
Antioxidant
Anti-inflammation
=> Promising applications in medicine
• Not produced in large scale
Pathways
• Totally 42 proposed pathways
• Most pathways involves
aromatic amino acid metabolism
• Differences
• Evaluated points
• Precursors/substrates
• Many pathways are same but
with different reaction order or
with additional steps
Choosing the pathway
Rank 1/ ”The best pathway”
Rank 4
• Precursors: L-tyrosine + L-Dopa
• Precursors: Trans-cinnamate + 4-Hydroxyphenylpyruvate
• Products: Rosmarinic acid + 3,4-dihydroxyphenylpruvate • Product: Rosmarinic acid
Advantages
- High total score
- Higher yield
Advantages
- All the other scores bigger except yield
- Less steps
Disadvantages
Disadvantages
- Program assumes that L-tyrosine is transported to cell => - Trans-cinnamate not naturally produced in E. coli => Needs
Must be included into medium
additonal enzyme for that
Phenylalanine
pathway
Transcinnamate
1
2
4
5
4
4-hydroxyphenlypyruvate
Tyrosine pathway
3
6
Choosing the construct
1) Superoxide dismutase
• Bradyrhizobium diazoefficiens (bacterium)
2) Isoleucyl-tRNA synthetase
• Rhizobium leguminosarum (bacterium)
3) Glutamate synthase
• Acinetobacter sp (bacterium)
4) Alanine acetyltransferase
• Methanosarcina barkeri (archea)
5) Phenol hydroxylase
• Aliivibrio fischeri (bacterium)
6) Superoxide dismutase
• Bradyrhizobium diazoefficiens (bacterium)
Choosing the construct
• Some suggested enzymes were odd and isoenzymes are found in E.
coli
• Glutamate synthase
• Testing needed to validate functionality
Final thoughts
Pros:
- Good trade-off in complexity and specificity
- Some useful suggestions
Cons:
- Many pathways are essentially the same, just different order of reactions.
- Still need extended human expertise, esp. for examining enzyme constructs and
making sure calculated scores make senses.