Download Asymmetric relationships between metabolic reactions

Asymmetric relationships between
metabolic reactions shape genome evolution
Philip R. Kensche*, Richard A. Notebaart*, Martijn A. Huynen
, Bas E. Dutilh
*authors contributed equally
C B
m
i
Center for Molecular and Biomolecular Informatics
Radboud University Nijmegen Medical Center
Nijmegen Center for Molecular Life Sciences
The Netherlands
Results
Summary
Often, dependency relations between proteins are asymmetric: one protein (A) depends for its function on another, second
protein (B), but that second protein does not depend on the
first.
Stoichiometric modelling of complete metabolisms suggests
that most of the reaction pairs with coupled fluxes are
asymmetrically rather than symmetrically coupled: 82% in S.
cerevisiae and 67% in E. coli.
This asymmetry in the coupling of fluxes is reflected in the
expression of the catalyzing enzymes, the impact of their
gene’s knock-outs and the evolution of gene content.
Figure 1
reaction A is active also reaction B carries a flux. Conversely, B can be
than the random expectation (two-sided, onesample Wilcoxon test; McNemar test).
S. cerevisiae
E. coli
E. coli and S. cerevisiae
0.0
0.2
0.4
).
B
| Analysis outline. As activity of reaction A is not necessary for
reaction B, A could be considered “dispensable”. Various other genomic
data can be interpreted in terms of dispensability: “dispensable” genes are
not expressed in a condition, do not affect growth when knocked-out or
are absent from a species’ genome.
flux-coupling
analysis
|
growth/
expression essentiality
directional
coupling
A→B
1.0
Strength of asymmetry is independent of network distance.
essentiality
expression
growth
occurrence
B
phylogenetic
occurrence
dispensability profiles
avg. asymmetry
Figure 2
0.8
A
Figure 5
A
0.6
average asymmetry
active without A because of alternative converging or diverging fluxes
(
essentiality
expression
growth
occurrence
losses
gains
contingent gain A
maintenance
Asymmetry in various genomic
All asymmetry values are significantly larger
| Examples of asymmetric relations between reactions. Nodes
represent metabolites, edges reactions. For a reaction pair A→B, if
|
Figure 4
data sets is consistent with model predictions.
gains
losses
contingent gain A
maintenance
1.0
0.5
S. cerevisiae
0.0
Figure 6
3
4
5
6
7
E. coli
1
2
8 9+ 1 2 3
network distance
4
5
6
7
8
|
The asymmetric relationship [ThrB → Asd] conserved
9+
between E. coli and S. cerevisiae is reflected in their phylogenetic
asymmetry measure f01=n01/(n01+n10)
distributions: thrB is almost never present in a genome without asd
because ThrB’s activity depends on Asd, but not vice versa.
Aspartate
semialdehyde
dehydrogenase
(Asd)
Is there asymmetry?
H0: fAB:=f01=0.5
Figure 3
Homoserine
kinase
(ThrB)
Threonine
synthesis
| Four measures of asymmetry extracted from the parsimony
reconstruction of ancestral gene content (STRING database & PAUP). The
type of change across branches that is more expected in asymmetric pairs
A→B is denoted in pink. E.g. “contingent gain A”: If A’s function depends on
Lysine
synthesis
Methionine
synthesis
B, then A should be gained if B is already present but not if B is absent.
asd
#gain
00
#loss
11
10
01
10
#contingent gain A
00 01
01
10
11
#maintenance
10 01
10
01
thrB
present
absent
present
absent
184
1
129
59
Thermophilum pendens
Crenarcheote that feeds on peptides
Conclusions
Genomic data contain not only information about whether
proteins interact but also about the characteristics of the
interaction. Asymmetric patterns could be used to predict
characteristics of known and new functional relations.
Contact
[email protected]