Design of Pathway-Level Bioprocess Monitoring and Control Strategies Supported by Metabolic Networks - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 197

participate in only one or two reactions, while a few, such as adenosine triphos-
phate (ATP) or pyruvate, are metabolic hubs participating in dozens of metabolic
reactions [ 26 ]. In the context of bioprocess control it is particularly relevant to
analyze how metabolic networks interact with the extracellular environment.
Bernhardsson et al. [ 27 ] have analyzed the metabolic networks of 134 bacterial
species and concluded that common reactions are found at the center of the net-
work and decrease as we move to the periphery of the metabolic network, i.e.,
closer to the metabolites that cross the cellular membrane. Borenstein et al. [ 4 ]
have determined the seed set compounds (i.e., exogenously acquired compounds)
for each of the 478 prokaryotic species with metabolic networks available in the
KEGG database. They found that about 8-11 % of the compounds in the whole
metabolic network correspond to the seed set and that each organism possesses a
characteristic seed set. Moreover, comparing the seed set of the different organ-
isms enabled them to trace the evolutionary history of both metabolic networks
and growth environments across the tree of life, supporting the ''reverse ecology''
principle. These structural features are pivotal for the design of process control
strategies based on metabolic networks. On the one hand, given the high number
and specificity of metabolites that cross the cellular membrane, the measurement
of the metabolic footprint, i.e., the complete set of extracellular metabolites, might
carry sufficient information to reconstruct a large number of intracellular metabolic
processes. On the other hand, the concentrations of many such extracellular
metabolites can be manipulated in order to control intracellular processes linked to
product yield and quality.

2.2 Material Balances

The list of metabolic reactions identified in a genome-scale reconstruction project
can be translated into a stoichiometric matrix, A, with dim(A) = m 9 q, where
m is the number of intracellular metabolites and q is the number of metabolic
reactions. The material balances over the intracellular metabolites take the fol-
lowing general form:

dc i
dt ¼ A v l c i ;

ð 1 Þ

where c i is the vector of intracellular concentrations [dim(c i ) = m], v is the vector
of intracellular fluxes, and l is the specific growth rate. Under the pseudo-steady-
state hypothesis, intracellular metabolites do not accumulate and the dilution term
is much smaller than the net turnover of metabolites, thus Eq. ( 1 ) simplifies to

0 ¼ Av
v j 0

ð 2 Þ