Transcript Metabolic/Subsystem Reconstruction

Metabolic/Subsystem
Reconstruction
And Modeling
Given a “complete” set of genes…
• Assemble a “complete” picture of the biology
of an organism?
• Gene products don’t generally function in
isolation
• The whole is greater than the sum of the
parts? Or can it also be less?
A few examples of higher order entities
(multiple gene products and even some additional components)
• Protein complexes (ribosomes, enyzmes,
secretion systems, etc.)
• Pathways
• Metabolism (linked pathways)
• Processes (chemotaxis, splicing, etc.)
• Cellular structures (membrane, cell wall, etc.)
Metabolic Reconstruction
• Determination of which metabolic pathways
are present in an organism based on the
genome content
• Can provide insight into organisms as well as
environments
• But, we can only reconstruct what we
recognize
KEGG (Kyoto Encyclopedia of Genes
and Genomes)
KEGG-Reference Pathway Overview
KEGG - Escherichia coli MG1655 overview
KEGG- Citrate Cycle
KEGG – Citrate Cycle (E. coli MG1655)
Mouse over EC number 1.3.99.1
(succinate dehydrogenase)
KEGG
• Other functionality
• Growing
• Automated annotation server for assigning
genes from a new genome to pathways
• Map subsets of genes to pathways
(enrichment analyses)
Bacterial chemotaxis – Pectobacterium
atrosepticum… but what about the other 33 receptors?
One size doesn’t fit all
• Specialized
pathways for
individual
organisms in
specialized
database
resources
• Allow for
variations on a
theme
The SEED - variants
Pathway holes can lead to discovery
Metabolic Model
• Computable metabolic reconstruction
Five uses:
1. Contextualization of high-throughput data
2. Guiding metabolic engineering
3. Directing hypothesis-driven discovery
4. Interrogation of multi-species relationships
5. Network discovery
Contraint-based modeling
-A stoichiometric matrix, S (M x N) is constructed for an organism, where M=metabolites (rows) &
N=reactions (columns)
r1 r2 ……..rk
Ex.
r1
m1+m2=> m3
r2
m3 <=> m1 + m4
m1 -1 1
m2 -1 0
m3 1 -1
… 0 1
mi
The dynamic mass balance equation
dmi/dt = Σ sik vk
k
-sik represent entries in S
- vk represents a reaction flux that produce and/or degrade
metabolite mi
-Concentration of a given metabolite: mi
dm/dt =Sv
m=a vector that represents a set of metabolites
v = flux vector
at steady-state there is no accumulation or depletion of metabolites in the network, so the rate of production=
rate of consumption, hence this balance of fluxes is represented mathematically as
Sv = 0
-bounds that further constrain individual variables can be identified, such as fluxes, concentrations, and kinetic
constants.
(vmin < v < vmax)
Irreversible reactions vmin=0, some metabolites such as O2 or CO2 have vmax=infinity, other metabolites are
constrained based on experimental measurements as determined for the biomass reaction for E. coli 1 gm dry cell
weight
There are normally more columns (reactions ~2,300) than
rows (metabolites ~1,100) there does not exist a single
solution but rather a steady-state solution space
containing all possible solutions.
(Thiele I. et al. 2009 PLOS Comp. Biol.)
Flux Balance Analysis (FBA): FBA calculates the flow of metabolites through this
metabolic network, thereby making it possible to predict the growth rate of an organism
or the rate of production of a biotechnologically important metabolite.
-With no constraints, the flux distribution of a biological network may lie at any point in a
solution space.
-When mass balance constraints imposed by the stoichiometric matrix S and capacity
constraints imposed by the lower and upper bounds (ai and bi) are applied to a network, it
defines an allowable solution space.
-Through optimization of an objective function, FBA can identify a single optimal flux
distribution that lies on the edge of the allowable solution space.
(Orth, Thiele, and Palsson Nat. Biotech 2010)
The Iterative reconstruction and history of the E. coli metabolic network
(Feist A.F. and B.O Palsson (2008) Nature Biotechnology)
Applications of the RMN of E. coli
Feist A.F. and B.O. Palsson (2008) Nature Biotechnology
Validation of metabolic models through comparison of in silico
vs. experimental data with or without oxygen
Comparison of carbon source utilization
Flux Balance Analysis (FBA)
Given an uptake rate for key nutrients (such as glucose and
oxygen), the maximum possible growth rate of the cells can
be predicted in silico.
0.3
10
0.25
8
0.2
6
0.15
4
0.1
2
0.05
0
0
0
1.5
3
4.5
6
7.5
9
10.5
12
12
(Becker SA, et al. (2007) Nature Protocols)
Time (h)
gDW/L
mmol/L
Comparison of batch growth
Glucose
(mmol/L)
Acetate
(mmol/L)
Formate
(mmol/L)
Lactate
(mmol/L)
Ethanol
(mmol/L)
Succinate
(mmol/L)
Biomass(gDW/L)
Carbon source utilization results
E. coli K-12
Strain
MG1655
E. coli O157:H7 (EHEC)
W3110
EDL933
E. coli (UPEC)
Sakai
CFTO73
Salmonella
UTI89
LT2
O2
No O2
O2
No O2
O2
No O2
O2
No O2
O2
No O2
O2
No O2
O2
No O2
Tested
compounds
included in
the model
In silico and
experimental
Agreement
76
76
76
76
76
76
76
76
76
76
76
76
55
55
70
66
71
64
69
63
68
64
67
63
71
65
52
48
Experimental = N
In silico = Y
False positives
4
1
1
0
2
2
2
1
3
0
1
0
2
0
Experimental = Y
In silico = N
False
negatives
2
9
4
12
5
11
6
11
6
13
4
11
1
7
In general good agreement of in silico vs experimental carbon source utilization for both
aerobic (>88% accurate) and anaerobic conditions (>83 % accurate).
Batch growth results in MOPS minimal media + 0.2 % glucose
anaerobic
0.4
0.35
Biomass (g/L)
0.3
0.25
0.2
0.15
0.1
0.05
0
E. coli K12 E. coli K12
MG1655
W3110
E. coli E. coli Sakai E. coli CFT
EDL933
O157:H7
(UPEC)
O157:H7
E. coli
UTI89
(UPEC)
S. typhi LT2
M. tuberculosis
• Built a genome-scale model
• Predicted essential genes using FBA and
compared to saturated transposon-based
characterization of essentiality (78%
accuracy/agreement)
• Compared flux through all pathways under
slow and fast growth by changing nutrient
uptake flux constraints
Major difference in isocitrate lyase and
glyoxylate shunt
Yeast deletion mutants
• Used quantitative image analysis to measure
growth of replica pinned cells on agar under 16
conditions (no growth, slow growth, wt growth)
• FBA to predict growth from yeast model
• 94% agreement
• Refined experiments based on model (checked
mutations, secondary mutations, unlinked
phenotypes)
• Gained insight into glycerol and raffinose
catabolism