Structured Models of Metabolism and Growth 225
SPECIFIC OXYGEN CONSUMPTION Ima/9-08"1
SPECIFIC GROWTH RATE (ORV)
Figure 3.21. Relationship between specific oxygen uptake rate relative to fresh weight and
the specific growth rate of the fresh weight [from D. Drapeau, H. Blanch and C. Wilke,
Biotech. Bioeng., 23, 1555 (1986)].
on
SPECIFIC OIOSGENIN RATE (mg/o-08M1
°
da
spectre ono mare far“
Figure 3.22. The specific diosgenin biosynthesis rate relative to fresh weight as a function
of the specific growth rate of fresh weight (from D. Drapeau, H. Blanch and C, Wilke,
Biotech. Bioeng., 23, 1555 (1986)).
3.5 Structured Models of Metabolism and Growth
Unstructured models do not recognize the complex set of metabolic reactions occurring
within the cell. Therefore we should not expect these simple models to be able to predict
the dynamic behavior of cells subject to changing external conditions. In addition,
unstructured models can predict intracellular concentrations only if we assume there is a
constant fraction of the particular metabolite in the cell; for example, that the fraction of226 Structured Models of Metabolism and Growth
RNA or DNA within a cell is constant. They thus have limited utility in guiding research
aimed at understanding cellular regulation and dynamics. Such studies and the models
derived from them are also required for implementation of sophisticated control of biological
reactors or biological processes.
A number of approaches have been developed to tackle the problem of developing
more sophisticated models of cell metabolism and growth. We shall examine some simple
approaches in the following sections. It is clear that attempting to model alll the reactions
within a cell would be an extremely complex undertaking, particularly since we know rel-
atively little about the kinetics and regulation of many of the enzyme-catalyzed steps
involved. We can, however, group many of these reactions based on their characteristic time
constants or relaxation times. The temporal organization within the cell can be described in
terms of the following systems:
Genetic system: the longest timescale, associated with changes in DNA composition due to
random mutagenesis, mutagenesis arising in response to changing environmental
parameters, or natural genetic transfer of new characteristics via a vector such as a
virus. This timescale is in multiples of the organism's generation time.
Epigenetic system: the intermediate timescale and is related to DNA transcription and
translation. Enzyme induction (synthesis) and repression occur over this timescale.
Metabolic system: the short timescale of individual enzyme reactions. Usually much shorter
than the epigenetic system. The behavior of the genetic and epigenetic systems may
be considered constant over the period of observation of metabolic events.
In most situations, there is sufficient difference between these timescales that the
behavior of other systems can be considered constant. For example, if we are examining
the substrate and product concentration changes of a particular enzyme-catalyzed reaction
within the cell, the enzyme concentration can be considered constant over the time period
of our observations. Changing enzyme concentrations involves transcription and translation
of DNA onatimescale of order tens of minutes or hours, while the enzyme-catalyzed reaction
may have a relaxation time of order seconds or minutes. On the other hand, if changes
occurring in the external environment are on the timescale of the epigenetic system, the
behavior of the metabolic system will be very fast and we can consider it to be in a quasi-
steady state. This approach greatly simplifies model development.
Models which incorporate the details of intracellular metabolism are referred to as
structured models. Such models attempt to account for unbalanced growth of microor-
ganisms, i.e., when the composition of the major cellular constituents, such as RNA, enzyme
concentrations etc. vary as a result of changing external conditions. Such conditions apply
in batch growth, in fed-batch growth, and in transient situations in well-stirred tank reactors.
The transient responses of cells to these changing external conditions can be modelledStructured Models of Metabolism and Growth 227
by analogy with classical reactor modelling using the transfer function approach. By using
an appropriate forcing function and determining the transient response of the cells, the
behavior of various cellular constituents can be modelled as first or higher order’. This
approach has some advantages in developing and analyzing strategies for process control,
but does not provide much insight into the factors that regulate metabolism. We shall now
turn our attention to models which incorporate more of the features of microbial metabolism.
3.5.1 Compartmental Models
The earliest attempts to include structure in models of cell growth and metabolism
generally subdivided the cell mass into various components on the basis of the function of
parts of the cell's internal machinery, We shall examine a simple two-compartment model
to illustrate the features of this approach.
The Model of Williams”
The model of Williams divides the cell into two compartments, a synthetic one (k-
compartment) that we can consider as consisting of RNA and pools of small metabolites,
and a genetic one (g-compartment) consisting of DNA and protein. The third component is
the external substrate concentration. A simple model based on these compartments can be
developed as follows. If K and G are the concentrations of the components in the k- and
g-compartments, written as mass per unit cell volume (V,), we can write mass balances for
aconstant reactor volume (V,) as follows.
A
HT hSKVe ie. ap Thx (3.78a)
dSV,_ 1 ds
i £ 7 hx¥e ie. a (3.786)
The rate of substrate uptake is assumed to be first order in substrate concentration S
and in total cell concentration X (both S and X being expressed as mass/reactor volume).
We assume that the structural-genetic compartment material is produced from the synthetic
compartment at a rate that depends directly on the concentration of species in each com-
partment. The mass balances are based on the reactor volume, thus the concentration per
cell must be multiplied by the total cell volume per unit reactor volume, X/p,, where p, is
the cell density (cell mass per unit cell volume).
(31) See for example Young, T.B., Bruley, D.F. and H.R, Bungay, Biotech. Bioeng. 12, 747 (1970).
(32) A review of some of the many compartment models is provided in Harder, A. and J.A. Roels, Adv. Bio-
chem, Engng., 21, 55 (1982).
(33) Williams, F.M., J. Theoret. Biol. 15, 190 (1967)228 Structured Models of Metabolism and Growth
dG(X/p.)
dt
Expanding the left hand side yields
=kGK(X/p,)
a(Xip.)
dt
(Xip.) So +G =k,GK(X/p,)
dG 1dX
ie k,GK-G Xa
and combining this result with Eq. (3.78a)
dG
pT hGK ~ (kG (3.79)
We have assumed that the density of the cell, p, , is constant. The synthetic portion of the
biomass is produced at a rate that is first order in substrate concentration and depends on
the cell density p, (equal to the sum of K and G):
dKX/p,
Go k,S(G +K) (Xip.) —kGK(X/p,)
dK Lax . Lax |
GF TRSG+K)-KGK-KYF and noting that ¢ 7 =kS
IK
wk $0,-hGK-kKS (3.80)
We note that equations (3.79) and (3.80) can be added to show that the relationship below
holds:
dK 4G i 5p.-kS{K+G)=0 (as K+G=p,) 81)
dt de BSP e .
Equations (3.78) and (3.79) can be added to eliminate X
d(X +¥,8)
ease a0) thus X4Y,5 =Xq4 ¥5p (3.82)
dt
Equation (3.78) can now be solved for $
+38)
ma
Lara (3.83)
1 +7 exp{k (Xo VsSuV Ys)
If we assume that the cell number depends only on the amount of genetic component present
(ie., G), then the cell number will be proportional to GX/p, cells/reactor volume. The cell
volume will change as a reflection of the changing amounts in the genetic and synthetic
compartments, hence the cell sizé will be proportional to (K+G)/G, ie. p,/G. The behavior
of the model is shown in Figure 3.23.Structured Models of Metabolism and Growth 29
batch growth
2.000
1.500! coll size
5
g
2
: 1.000
5
€
3 cell mass
0.500
substrate
0.000 +
0 20 40 60 80 100
Time
Figure 3.23. Simulation of the two-compartment model of Williams for batch growth with
the following initial conditions: X,= 0.05 mg/ml, S>= 1.0 mg/ml, Ky= 0.6 mg/ml, Gy= 0.4
mg/ml. Time is expressed in hours, concentrations in mg/ml of cell (for g-mass) or reactor
volume (for cell mass and substrate). Values of the constants are k,= 0.0125 (mg/ml)'.hr',
k,= 0.025 (mg/ml)''.hr', ¥=0.5 mg/mg, p, = 1.0 gm/ml. Cell number is proportional to the
g-mass.
The compartment model of Williams illustrates some important properties of cell
growth, It predicts the existence of a initial lag phase (see the cell mass curve in Figure
(3.23) and, if the inoculum is not fully adapted (ie. Ky and Gy do not correspond to values
found in exponential growth), cell mass will increase (due to an increase in the K com-
partment), while the cell number will not change initially (the cell density p, remains
constant). During the growth phase, both cell mass and cell numbers increase exponentially
The stationary phase is attained with respect to cell mass prior to cell number (as illustrated
in Figure (3.23) by the curves of g-mass and cell mass). These phenomena are typically
observed in batch culture experiments with bacteria. The model can be refined by changing
the linear dependence of the rate of substrate uptake and growth from first order to a Monod
type in equation (3.78). As can be seen from the batch simulations above, the concentration
of the g-compartment approaches the cell density pat long times, while that-of the
k-compartment approaches zero. Neither of these predictions is consistent with experimental
observation. The inclusion of maintenance in the formulation of the model would change
this inconsistent result and improve the model.230 Structured Models of Metabolism and Growth
The Model of Ramkrishna et al.
An analogous compartment model to that of Williams has been developed by Ram-
krishna et al. The cell is divided into two compartments, G-mass, comprising RNA and
DNA, and D-mass, which mostly consists of proteins. An inhibitor (T) is produced during
growth which converts both G- and D-mass to inactive forms of biomass. The reactions
assumed are the following:
G+a,S+D32G+D+a,T
G+a,/5+D 52D+G+a,'T
GtTNgt(It+a,)T
D+T Ny +(1+ay/)T
In the first reaction, D-mass catalyzes the formation of G-mass, consuming a, units
of substrate and producing a units of an unidentified inhibitor T. In the second reaction
G-mass catalyzes the formation of D-mass. The last two equations show that both G- and
D-mass are deactivated by inhibitor (producing inactive forms Ng and Np). The rate
expressions assumed in the model for production of G- and D-mass are of the double substrate
form of equation (3.44), while those for the deactivation reactions are assumed to be first
order in each reactant. Mass balance equations for batch and continuous stirred tank reactors
can be written and the model predictions can be compared with experimental data. This
model can predict oscillatory behavior about a steady state.
3.5.2 Models of Cellular Energetics and Metabolism
As we move beyond the compartmental models of the type above, the level of metabolic
detail required increases. At this point, it will be useful to consider the types of reactions
that occur within the cell. Metabolic pathways can be distinguished as catabolic and ana-
bolic. In catabolism, energy-containing molecules, such as carbohydrates, hydrocarbons
and other reduced carbon-containing compounds are degraded to CO, or other oxidized
end-products and the energy is stored in ATP, GTP, and other energy-rich compounds. In
anabolism, intermediates and end-products formed from catabolism are incorporated into
cell constituents (such as DNA, RNA, lipids, carbohydrates, etc.) and their intermediate
precursors (amino acids, purines and pyrimidines, simple sugars etc.). Anabolic reactions
generally require energy, which is supplied via ATP and other high-energy phosphates
generated during catabolism. As the concentration of these high-energy intermediates within
the cell is rather small, anabolism is linked to catabolism and ATP is rapidly turned over.
This implies that energy producing and energy consuming processes must be tightly regu-
(34) Ramkrishna, D., Fredrickson, A.G. and H. Tsuchiya, Biotech, Bioeng., 9, 129 (1967),