Biology Reference
In-Depth Information
BOX 12.2 Cofactor Specificity
Organism specificity is always a concern during manual
curation. For example, the isocitrate dehydrogenase (IDH)
reaction in E. coli's core metabolism uses NADP/NADPH as
cofactors rather than the NAD/NADH more commonly
associated with the reaction [181] , as a result of evolutionary
adaption to permit growth on acetate. In fact, organisms with
an isocitrate lyase gene (essential for growth on acetate), will
always use NADP for the IDH reaction.
BOX 12.4 Transport Reactions
Experimentally, E. coli (among others) is known to have
glucose and fructose import capabilities and export capa-
bilities for metabolites such as lactate and ethanol. Thus,
transport reactions are added for these metabolites. Twenty-
one transport reactions (including diffusion processes for
H 2 O and CO 2 ) have been added to the E. coli core metabolic
network, thereby increasing the total reaction count to 74.
possible, since cofactor specificity may vary between
organisms ( Box 12.2 ). If literature is unavailable, data-
bases such as KEGG [13] and BRENDA [20] can be used.
The localization of the reaction (i.e., the compartment in
which each reaction occurs) can be determined using
algorithms such as PSORT [24] or PASUB [25] if litera-
ture or experimental data is unavailable; if the algorithms
are indeterminate, enzymes are often assumed to be in the
cytosol. Care should be taken in assigning the correct
localization for each reaction, since localization affects
metabolic network functionality as in vivo metabolite
availability may be limited due to transport costs. Accu-
rate compartmentalization of reactions has been shown to
increase the predictive power of metabolic reconstructions
[26] . The subsystem (e.g., glycolysis or pyrimidine
metabolism) for each reaction is also often established
basedontextbookorKEGGassignments [13] . Finally the
gene e protein e reaction (GPR) association for each reac-
tion (i.e., the flow from gene to enzyme to reaction cata-
lyzed) should be found ( Box 12.3 ). These are described as
Boolean relationships between genes or proteins, and
include information on isozymes, secondary or promis-
cuous enzyme activities, and protein complex participa-
tion. Primary literature and organism-specific databases
are rich sources for this information. Moreover, careful
curation of this information can greatly affect the reli-
ability of simulations, such as gene deletion simulations.
In a similar manner, each metabolite must be carefully
curated. For each metabolite the formula should be found.
Multiple sources exist, e.g. KEGG [13] , BRENDA [20] ,or
PubChem [27] . It is preferable that multiple sources be
used to ensure that the formula is accurate, since databases
can vary in reported molecular formulae (e.g., protonation
state). From these formulae the charged formula is usually
determined, since physiological pH often results in non-
neutral compounds (e.g., deprotonation of many acids). To
aid in this, various software packages have been developed
(e.g., Pipeline Pilot, ChemAxon's pK a plugin, or ACD/Labs
pK a DB). Once all metabolite formulae are established,
tests must be conducted to verify that all reactions are mass
and charge balanced. Many reactions require the addition
of water and protons to balance reaction oxygen and
hydrogen. Any mistakes in mass balance (even with
protons) can block normally functioning pathways or
predict infinite ATP generation.
After reaction stoichiometry, gene association and
metabolite information is verified, a confidence score is
often provided as a semi-quantitative assessment of reac-
tion quality. For example, reactions in many reconstruc-
tions are assigned a confidence score (from 0 to 4, 4 being
the best) that reflects the amount of information available
for the reaction. Guidelines for assigning confidence scores
have been previously suggested [9] .
Incorporation of Spontaneous and Transport
Reactions
Not all reactions occurring in a cell are enzyme catalyzed,
and some hydrophobic metabolites readily diffuse through
cell membranes. Thus, spontaneous reactions should be
added to the network to reflect these processes. Many such
reactions are catalogued in databases such as KEGG [13]
and BRENDA [20] , or are characterized in the literature.
Transport reactions (i.e., the transport of a metabolite
across a membrane) can be either spontaneous or enzyme
facilitated, depending on each individual metabolite. If
evidence suggests that certain metabolites can be taken up
from or secreted into the medium, or transported between
cellular compartments, the associated transport reactions
should be added accordingly.
BOX 12.3 GPR Associations
Data for GPR associations in E. coli are readily available. For
example, pykA and pykF both encode proteins (PykA and
PykF, respectively) that can catalyze the pyruvate kinase
reaction (thus pykA OR pykF are required for the reaction).
Similarly, sdhA, sdhB, sdhC, and sdhD are required for
formation of the protein Sdh, which catalyzes the irreversible
succinate dehydrogenase reaction in oxidative phosphory-
lation (thus sdhA AND sdhB AND sdhC AND sdhD are
required for the reaction).
Search WWH ::




Custom Search