82 — 4 Kinotles — finding quantities for bloprocess reactions,
4.4 The yield coefficient — combining substrate turnover
and growth
‘The question we are grappling with in this paragraph is how the cells ‘know’ how fast
they can grow. Many environmental parameters like temperature, pH, or salinity of
course play a role. Intracellular enzymatic steps also influence possible growth rates
differently for different groups of organisms. Even in the case that all environmental
parameters are at their respective optimum the cell is limited by intracellular steps
e.g., ifan enzyme isat its maximum turnover rate. This gives reason to define a maxi-
mum specific growth rate pmax [g/g/hl. Itis the task of metabolic modeling to find out
the intracellular reasons for the value of imax for different classes of microorganisms.
In the next paragraphs we focus on the role of substrate concentration.
From previous experience it is known that substrate concentration is one ma-
jor factor that can be influenced by process engineering means. In the discussion of
medium design (Chapter 3) we already used the relationship between initial substrate
and final biomass concentration. From the experiment (Figure 4.1) it becomes obvi-
ouss that with progressively increasing biomass concentration substrate concentration
also decreases faster, so we assume that growth and substrate uptake are coupled. In
the time interval where cs approaches zero growth is no longer possible, implying that
growth depends on the actual substrate concentration.
The first attempt to find a more or less unspecific relation between growth and
a limited environmental resource was formulated by Pierre-Francois Verhulst (1854).
The background was the observation that growth, e.g., as predicted by Fibonacci, can-
not go ad infinitum but has to stop somewhere. The imagination was a limited area
offering enough space or food for a couple of animals but with a limited carrying ca
pacity K of individuals. As this capacity becomes exhausted the animals have to mi-
grate or do not get so many offspring due to starvation. Food like grass could further
grow but is divided by more and more animals. Finally, each individual gets so little
food that only survival but no propagation is possible. In addition to the linear growth
model (4.29) Verhulst postulated that growth depends on the residual capacity (K-N)/K
(normalized from 0 to 1) of the habitat:
an
“Gp = imax N (4.37)
The solution of this differential equation with No being the initial population is:
K-No-etnmt
N= (4.38)
K¥No- (et 1)
This so called ‘logistic growth curve’ is a sigmoid function and a popular (as it has at
least a closed solution) first trial to fit growth data. Nevertheless, the assumptions do
not really represent the situation ina bioreactor. In our batch, for example, we see that
growth ceases only very late, when glucose is nearly exhausted.4.4 The yleld coefficient — combining substrate turnover and growth —— 83
Fig, 4.23: The concept of substrate Umitation; an enzymatic transport step (symbolized by the turn:
Ing enzyme) carries substrate Into the cell. This substrate flux determines the possible growth,
The first one to describe growth as a function of substrate concentration was Jacques
Monod in 1942 when he gave the famous Monod equation:
és
Wp GE «ss
Monod won the Nobel Prize (Physiology or Medicine) in 1965 for the operon theory, not
for this equation. Since then the Monod equation has been employed to describe myr-
iad bioprocesses. The equation shows structural similarity to enzyme kinetics, so the
assumption has been formulated that enzyme kinetics is an underlying mechanism.
Substrate uptake as an enzymatic step in metabolism is the first candidate to investi-
gate as shown in Figure 4.13. Understanding that the more biomass is present the more
substrate will be taken up, it is sensible to define the specific substrate uptake rate rs
in analogy to the specific growth rate:
dms,
~ dtmx
Ts (4.40)
as a measurable property of the cells. Following the attempt to understand the Monod
equation and in analogy to enzyme kinetics rs can be given asa function of the sub-
strate concentration cs: «
s
Kg ves
This is in accordance with the experiment in Figure 4.1 that at lower substrate concen-
trations the substrate uptake rate is reduced. Substrate availability inside the cell then
determines the possible growth rate. This concept is called ‘substrate limited growth’,
The substrate is then converted by hundreds of enzymes inside the cell. However,
growth is regarded as a thermodynamic process where the cell could make ‘the best
of it’ under the restrictions of energy and material balances, stoichiometric relations,
and other constraints. We have already discussed the elemental balance as such acon-
straint. Biology has provided different approaches to understand how much biomass
(441)
Ts = 1,max84 — 4 Kinetics — finding quantities for bloprocess reactions,
can be built up by a given amount of substrate taken up. Of specific interest is the
partitioning into anabolic pathways lumped as growth on the one hand and energy
producing metabolic pathways on the other. A fundamental relationship determining
the relation of growth and substrate uptake was found by S. John Pirt in 1965:
ty= ore ttm (4.42)
YES
To get a consistent system of terms the historic symbol y/ is replaced by rx, where r
is rate and X is biomass. To maintain a given specific growth rate the cell has to take
upa proportional rate of substrate. An important point is that the proportional factor,
interpreted as yield coefficient, is constant for many growth conditions. This has to be
understood as an optimized carbon allotment. Metabolic energy is provided by a part
of the substrate while the other part is used for growth under energy utilization. This
balance is kept stable by the cells in order to avoid waste of energy and carbon. As
Jong as the cell does not change this metabolic pattern this results in a constant yield.
In addition, energy is partly used for nongrowth-associated purposes like movement,
keeping the water balance, or maintaining membrane potential. This part of energy
utilization is lumped into the ‘maintenance* term ry, (m in the original reference). In-
deed, very often measurement data fit quite well to the linear relation between growth
and substrate uptake. For clarity it is noted that yy.s = dry/drs isa kinetic parameter,
while ¥x,s = Amx/Ams is a measured process characteristic over a longer cultivation
period. About the inner connection between substrate uptake rate and growth rate
more background is given in the next chapter.
A first idea to describe biomass formation and substrate uptake simultaneously is
simply combining the Monod and the Pirt equation ((4.33) and (4.37)). A short inspec-
tion of this system of equations suggests that it is not a good idea. For cs = 0a positive
substrate uptake rate is predicted, whichis physically not possible. Turning the cause-
effect chain we set the substrate uptake asa primary limiting step and reformulate the
resulting growth rate following Pirt’s linear relationship:
Fonw es
a re = Hames (oa
te = Yas ts—tm 5)
This is a first simple physiological model of microbial growth and will be used as an
example throughout this book. According to the essence of kinetics these equations
hold irrespective of the way cs changes during the process.
Besides biomass formation and substrate consumption, product formation is of
interest. As was done before we define a specific product formation rate:
dmp
tp (4.45)
dt-mx4.4 The yleld coefficient — combining substrate turnover and growth —— 85
In addition, a formal relation with growth and/or substrate turnover is required. As
the cell is highly stoichiometrically determined again a linear relation in analogy to
Pirt’s equation can be a good start:
rp =yox-re + kp (4.46)
This equation was introduced to the field by Luedeking and Piret and bears their
names. Itis a flexible tool to fit different kinds of product formation patterns. The yield
coefficient yp. = drp/dry describes the ratio of product formed per biomass growth.
Often product formation is coupled with catabolism, meaning substrate turnover. At
least the carbon comes from there. For technical applications the ratio of product
produced to substrate consumed yp,s = Amp/(ms is of interest. This is considered in
the reformulation:
rp = Yp.s°TS—7P,m (447)
Here product formation is coupled to substrate uptake where a small constant part of
the substrate is reserved for other purposes. This holds only for rs and rx over a given
threshold, which is also the case for the Luedeking-Pirt equation. In the next chapter
and some case studies the coupling of product formation to catabolism and anabolism
(rs, rx) and the complexity of control patterns of the cells is analyzed in more detail.
Before trying out the kinetic equations for our standard batch process good
guesses for reasonable values of the kinetic parameters are required. Table 4.4 gives
some typical values.
Some trends can be observed from Table 4.4. Aerobic microorganisms can reach a
yield of 0.5 g/g growing on glucose. Sometimes it is a bit lower, but never higher. The
yield also depends on the substrate. Witha higher degree of reduction (ethanol) higher
yields are obtained. Anaerobic growth on glucose results in a low yield of about 0.03-
0.05 g/g. Here, the product is always formed in a high yield product per substrate of
e.g., 0.5g/g. As substrate uptake is often higher for anaerobic growth, similar growth
rates as for aerobic growth are obtained.
‘Tab. 4.4: Typical specific growth rates of different microorganisms and substrates.
Organisms ‘Substrate r5.max Pama ks Yes Remarks
fe/(g- te/e-hy)—tg/__ta/et
Sacch. cerev. Glucose 2.52 60 0.180 0.485 Aerobie
Ethanol 14 0.753 Anaerobic
Glucose 0.53-0.63,
E.coli Glucose 05-10 0.004 0.5 — Aerobic
Aspergillus oryzae Glucose 0.005 Acroble
Klebsiella aerogenes Glucose 0.009 Aoroble
Glycerol 0.009
Virt. generic Glucose 1.0 10 0.5 Aerobic
Virt. generic Glucose 1.0 1.0 0.05 — Anaerobic86 — 4 Kinoties — finding quantities for bloprocess reactions,
4.5 The batch process — the simplest form of a bioprocess
Now alll the information can be assembled to get a simple model to describe a batch
process. Remembering the physiological model
(4.48)
(a9)
(4.50)
the concentrations cx and cg have to be calculated to havea complete set of equations.
This has to be done by setting up the material balances for biomass and substrate. For
the batch process the balance equations are in the form of simple differential equa-
tions of accumulation and reaction terms:
ax ex (451)
aes = is cy (4.52)
ft =p ex (4.53)
These reactor equations together with the physiological equations form a complete
model to represent a batch cultivation. Note the minus sign for substrate to get 2
decreasing substrate concentration for a positive turnover rate. Even for more com-
plicated cases the biological submodel and the reactor submodel should always be
strictly separated during building up and for writing them down. Unlike in setting
formal kinetics, e.g., only for biomass, these models cannot be solved explicitly but
must be numerically integrated. A simulation is shown in Figure 4.14.
Batch processes are usually divided into several phases. During the lag phase the
cells have to adapt to the environmental conditions inside the reactor, which may be
different from the parameters in the preculture. The specific growth rate increases only
after some time, e.g., one hour, to the value predicted by kinetics. This can happene.g.,
in cases where the medium in the preculture contains supplementary compounds that
are not present in the technical medium or vice versa. Adaptation is not represented
in the model equations. As the lag phase is unproductive it should be kept as short as
possible e.g,, by adjusting the conditions in the shaking flasks or the cultivation in the
stage before.
During the ‘exponential phase’, p is close to but not equal to pfmax for several hours
as substrate concentration is far in the saturation range of Michaelis-Menten kinetics.
Do not use the outdated term “logarithmic phase’. This is actually the state of the reac-
tor we want to achieve as long as possible. This is no longer the case in the ‘stationary
phase’, where substrate concentration falls to the order of magnitude of the limitation
constant and becomes therefore the decisive value for growth. In many cultivationsac-
cumulation of an (possibly unknown) inhibiting product could also contribute to the4,5 The batch process ~ the simplest form of a bloprocess —— 87
120
g 04
we Z| & fos
: = 3 3
a o . “ 02 6s
8 e S
a = . o1 2
2 = ce pee &
2 5 . 00
B ee
.
ee? “04
0 $3 8s * *
at
04234567 8 8 01 1213 14 15
Cultivation time [h]
Fig, 4.24: Simulation of a batch cultivation with the parameter set for the model organism Virtuella
generica (Table 4.4) showing the course of the concentrations and the specific turnover rates.
stationary phase. After the substrate is completely exhausted a decrease of cell dry
mass is observed, which is also visible by the negative specific growth rate. That does
not mean that the cells really perish in the ‘clecay phase’. In fact, they fuel their main-
tenance demand by respiration on intracellular materials like storage compounds, a
process called ‘endogenous respiration’. The optimum harvesting point is shortly be-
fore entering the decay phase.
We started this chapter by evaluating a batch process in terms of macroscopically
observable volumetric rates R and yields Y (capital letter). These characteristic values
have to be understood as integral values over a longer time interval. Enzyme kinetics
could be deduced from the mass action law being a basic principle in chemical re-
actions. Starting from the biological observation of propagation by cell division and
doubling, the specific growth rate ry was introduced. Like the other specific rates r this
is understood as a differential value being valid at all time instances. The same holds
for the model parameters k and y (lowercase letter). Kinetics in combination with ma-
terial balances are the key to formulate descriptions of a growth process with at least
some mechanistic justification. The basic assumption was the translation of enzyme
kinetics to growth kinetics. In the next chapters we will see how far we can go with
this daring idea.