Biology Reference
In-Depth Information
FIGURE 13.1
Metabolic genotypes
and phenotypes. Panel a) shows the
metabolic genotype of a genome-scale
metabolic network. It can be represented
in a simplified form as a binary string.
The entries of this string correspond to
one biochemical reaction in a 'universe'
of known reactions. Panel b) shows one
of many possible representations of
a metabolic phenotype, a binary string
representation whose entries correspond
to individual carbon sources. This
representation contains a one for every
carbon source from which a metabolic
network can synthesize all biomass
precursors. (Figure and caption adapted
from
[10]
. Used with permission from
Oxford University Press.)
(b)
Metabolic phenotype
(viability on carbon source)
(a)
Metabolic genotype
(network of enzymatic reactions)
1
⎛
⎞
Glucose + ATP → Glucose 6-phosphate + ADP
Fructose 1,6-bisphosphate → Fructose 6-phosphate + P
i
⎜
⎟
0
⎛
⎞
Alanine
Glucose
1
⎜
⎟
⎜
⎟
1
⎜
⎟
⎜
⎟
0
⎜
⎟
⎜
⎟
Isocitrate → Glyoxylate + Succinate
Acetoacetyl-Co + Glyoxylate → CoA + Malatet
⎜
⎟
1
Ethanol
1
⎜
⎟
⎜
⎟
⎜
⎟
⎜
⎟
⎜
⎟
0
1
Melibiose
Xanthosine
Oxaloacetate + ATP → Phosphoenolpyruvate + CO
2
+ ADP
Pyruvate + Glutamate ↔ 2-Oxoglutarate + Alanine
1
⎝
⎠
0
⎝
⎠
sole carbon
sources
>5000 biochemical reactions
network or vice versa; and a distance of D
¼
1 if they differ
sources of carbon or other elements. Note that even for
a modest number of 100 different potential carbon sources,
the number of possible metabolic phenotypes is already
2
100
or
in every single reaction
[54,56,57]
.
There are as many ways to define a metabolic phenotype
as there are tasks of metabolism. Metabolism detoxifies
waste, synthesizes molecules for defense and communi-
cation, and manufactures all small precursor molecules for
biomass synthesis. The latter task is the most fundamental,
and I will therefore focus on metabolic phenotypes related
to this task. For free-living organisms such as E. coli there
are of the order of 60 small biomass precursors
[58]
. These
include all proteinaceous amino acids, DNA nucleotide
precursors, RNA nucleotide precursors, as well as multiple
lipids and enzyme cofactors. A network's ability to
synthesize all these molecules will depend on the nutrients
that are available in an environment. Some organisms, such
as E. coli, can survive in very simple, minimal chemical
environments. These environments contain only one kind
of molecule that provides each chemical element; at least
one of these molecules also provides energy. Most free-
living organisms can use multiple different sources of
chemical elements and of energy.
These observations give rise to the following definition
of a metabolic phenotype, which is focused on sources of
carbon and energy but can be easily extended to sources of
other chemical elements
[54,57]
. Consider a given number
of molecules that could serve as sources of carbon energy to
some organism. Write these molecules as a list, as shown in
Figure 13.1
b. If the metabolism of a given organism can
synthesize biomass
10
30
.
This representation of metabolic phenotypes lends itself
to the systematic study of new metabolic phenotypes.
Consider a genotypic change that causes the addition of
chemical reactions to a metabolic network by horizontal
gene transfer. If these new reactions allow an organism to
survive on a carbon source that it had not been previously
able to utilize, a metabolic innovation has arisen. In an
environment where other carbon sources limit growth, or
where they are absent, this ability can make a life-changing
qualitative difference to its carrier.
I will now discuss analogous definitions of genotypes
and phenotypes for regulatory circuits. The evolution of
regulatory circuits, and especially of transcriptional regu-
lation circuits, is difficult to study experimentally. Part of
the reason is that regulatory DNA can occur far away from
the genes it regulates; also, such DNA can change very
rapidly on evolutionary timescales
[36,59
>
66]
. In addition,
to understand the relationship between genotype and
phenotype requires an analysis of many circuit genotypes
and their phenotypes. For these reasons, computational
models of regulatory circuits are still
e
indispensable to
understand genotype
phenotype relationships in such
circuits. The evidence discussed below stems from well-
studied models of
e
transcriptional
regulation circuits
that is, if it can sustain life on any one
of these carbon sources (that is, the organism needs to be
able to use this carbon source as its only carbon source)
[67
71]
. Variants of these models have been used
successfully to understand the development of specific
organisms, such as the early fruit fly embryo, and to predict
the developmental changes
e
e
e
write a 1 next to the carbon source. Otherwise, write a 0. In
this way one can define a metabolic phenotype as a binary
string that reflects an organism's viability on different
in mutant embryos
[67,
72
75]
. In addition, they have helped us understand
a variety of evolutionary phenomena,
e
such as how
