Stoichiometric Analysis of Animal

Cell Growth and Its Application

in Medium Design
Liangzhi Xie and Daniel I.C. Wang"
Department of Chemical Engineering, Biotechnolog y Process Engineering
Center, Room 2OA-207, Massachusetts Institute of Technology,
18 Vassar Street, Cambridge, Massachusetts 02 139

Received August 6, 1993/Accepted December 23, 1993

Animal cell cultivation in vitro has been studied for more

than 40 years. However, the culture medium composition has not been designed on the basis of the stoichiometric nutritional demands for animal cell growth.
In this article, a model was developed to study the
stoichiometric demands for nutrients (including glucose,
20 amino acids, and 10 vitamins) for the synthesis of cell
mass and product. The coefficients for these nutrients
in the stoichiometric equation governing animal cell
growth were determined based on cell composition.
In addition, a detailed analysis of the nutrients' roles
in the synthesis of cell mass and product was also
performed. Applications of the stoichiometric analysis in
animal cell cultivation, such as culture medium design,
supplemental medium formulation, and feeding strategy
will also be discussed. The stoichiometric analysis can
be potentially employed to analyze results from animal
cell cultures, to improve the performance of culture
processes, and to design new process rationally. It can
also help to provide a better understanding of animal
cell metabolism. Simplifications on the cellular energy
metabolism were made in order to simplify the model
and to provide the preliminary bases to test the process
performance. However, this could introduce inaccuracies
for the model and results in errors in the calculations of
glucose and glutamine concentrations when employed
in medium design. 0 1994 John Wiley & Sons, Inc.
Key words: stoichiometric analysis medium design
animal cell culture feeding strategy

INTRODUCTION
One of the most important considerations in animal cell
cultivation is in determining the composition of the culture
medium. Unlike microorganisms, animal cells are very
sensitive to their culture environment. Animal cells require
many essential nutrients, such as glucose, amino acids,
vitamins, inorganic salts, and serum components in order to
survive and grow in vit1-0.~The concentrations of glucose,
amino acids, and vitamins in the culture medium not only
affect the cell growth rate, but could also stoichiometrically
limit the maximum cell density obtainable in a batch
culture.'Q
The complex nutritional requirements of animal cells
impose real challenges to design of a proper medium needed
to achieve optimal cell growth and to attain high cell
* To whom all correspondence should be addressed

Biotechnology and Bioengineering, Vol. 43, Pp. 1164-1 174 (1994)

0 1994 John Wiley & Sons, Inc.

density. Animal cellular metabolism in vitro is not only

complicated, but also variable, depending on the culture
environment as well as differences in the cell line. This
renders the medium design more difficult. At the early stage
of medium development, the goal was to determine all of
the nutrients essential for in vitro cell
Ever
since Eagle developed a minimal essential medium (MEM)
in 1955,7many efforts have been devoted to kinetic studies
to improve the medium composition and to achieve a higher
cell growth rate.12,*5*22
The stoichiometry governing microbial growth is often
simpler and has undergone detailed
However, the stoichiometric nutritional demands for animal
cell growth have not yet been extensively studied. Thus,
the nutrient concentrations in the conventional media were
primarily determined experimentally through trial and error.
This is due to the following reasons: first, the nutritional
requirements for animal cell growth are much more complicated than those of microorganisms, which makes the
stoichiometric equation governing animal cell growth intractable by element balances; second, animal cells produce
protein as well as toxic byproducts, such as ammonia and
lactate. Factors affecting ammonia and lactate formation
need to be considered in medium design to decrease these
toxic byproduct formations.
Furthermore, animal cellular metabolism and growth
behavior are highly variable. Yields on nutrients calculated
from a standard batch culture depend greatly on the residual
concentrations of the nutrients. Due to the dependency of
the yields on the culture environment, these values often
vary from batch to batch. High ammonia and lactate formation rates and low nutrient efficiencies can thus result when
yields from conventional batch culture are employed in
medium design. Consequently, culture performance would
be difficult to improve significantly, compared with that of
the conventional batch culture from which the yields are
derived.
On the other hand, if detailed metabolic pathways and
intermediates were taken into account in the stoichiometric
analysis, the model would become overly complex. Without
experimental data, no unique solution could be obtained,
because the number of unknown variables is greater than
the number of equations governing cell m e t a b o l i ~ m The
.~~

CCC 0006-3592/94/0111164-11

sensitivity of the model to the experimental data, which

depend on culture conditions, makes the model useless
when employed under different conditions.
It is our goal to develop a medium design strategy
through the study of the stoichiometric nutritional demands
for animal cell growth. A stoichiometric equation governing
cell growth in animal cell cultivation will be determined.
The intricate cellular metabolism will be lumped into a
black box, therefore avoiding the intricate details to be
considered in the model. The stoichiometric coefficient for
each nutrient in animal cell culture will be determined by
a detailed analysis of the roles of nutrients in the synthesis
of cell mass and product. In effect, the stoichiometric
equation simplifies the complicated cell culture process into
a simplified but useful single equation. This model will
be helpful in understanding cell growth, designing culture
medium, and controlling the nutritional environment in
animal cell cultivation.

STOlCHlOMETRlC MODEL
Stoichiometric Equation
In animal cell cultivation, the desired product is protein
secreted by the cells during cell growth. The production of
cell mass is a necessary prerequisite to produce the desired
protein, because the product concentration is a function
of viable cell density, culture time, and specific product
synthesis rate. Therefore, the governing features in the
stoichiometric equation in animal cell growth are cell mass
and biological product secreted by the cells.
The synthesis of cell mass and product requires nutrients
such as glucose, amino acids, vitamins, inorganic salts, and
serum components. Animal cell mass contains a very small
fraction of inorganic salts, and these compounds are the
most abundant species in the conventional media. Thus, the
chance of any stoichiometric limitation of inorganic salts is
very slim. Inorganic salts are therefore not considered in
the model.
The essential roles of serum are to provide growth
factors, hormones, and binding proteins and to provide
hydrodynamic stress protection for cultured cell^.'^^^' Small
molecules present in serum, such as amino acids, glucose,
vitamins, and fatty acids, can be replaced by nutrients
present in the basal medium. In dialyzed serum, these small
molecules are removed before the serum is supplemented
to the cultured medium. Essential components in serum
are assumed to be sufficient for cell growth and are not
considered in the model.
By neglecting serum components and inorganic salts, the
overall stoichiometric equation for cell mass and product
synthesis becomes:

20

10

B,,i[amino acidIi

13~~,[glucose]

i=l

[cell mass]

Energy Production
In general, both glucose and glutamine can be utilized for
energy production. Branched-chain amino acids, such as
leucine, isoleucine, and lysine, have also been suggested to
contribute to energy production in some cases.9,21,23,24
The

Glucose

lAla

(1)

G l u m p r o Ser

\t J

Carbohydrates 3 c l e o t i d e s
and Lipids

Energy
Production

j=1

6ihAsn

+ 1B,,j[vitamin]j

+ B,[productl + BATP[ATP]

where Bglc is the stoichiometric coefficient for glucose, ea,i

is the stoichiometric coefficient for the ith amino acid, B,,,
is the stoichiometric coefficient for the jth vitamin, 0, is
the stoichiometric coefficient for the product, and BATp is
the stoichiometric coefficient for ATP.
Glucose, 20 amino acids, and 10 vitamins are reactants,
and cell mass, product, and ATP are products in the above
reaction governing animal cell growth. Note that the ATP
term included in the stoichiometric equation is not a net
production of ATP but is the amount consumed in the
synthesis of cell mass and product. Two major byproducts,
lactate and ammonia, are produced while cell mass and
product are synthesized. However, these two byproducts
are omitted from the equation for simplicity and will be
discussed later (see Discussion).
The cell mass is a complex mixture of cellular proteins,
lipids, carbohydrates, DNA, RNA, water, and inorganic
Because inorganic salts are neglected in the stoichiometric equation, water and inorganic salts in the wet
cell mass will also be omitted in the later analysis.
Even though Eq. (1) appears simple, the determination of
the unknown stoichiometric coefficients is a real challenge.
There are 33 unknown variables but only three chemical
elements involved in the stoichiometric equation. Therefore, the equation cannot be solved by elemental balances.
The roles of nutrients in the synthesis of cell mass and
product are formulated in Figure 1 in order to determine
the stoichiometric coefficients. In the stoichiometric model,
the pathways for cell mass and product synthesis are
classified as pathways for energy production and pathways
for the synthesis of carbohydrates, lipids, nucleotides, cellular proteins, and p r o d ~ c t . ~ , As
, ~ ~ shown in Figure l,
glucose is required for energy production and for the
synthesis of carbohydrates, lipids, and sugars in nucleotides
(DNA and RNA). All of the 20 amino acids are required
for the synthesis of cellular proteins and product protein.
Glutamine, aspartate, and glycine are also consumed for
the synthesis of nucleotides (DNA and RNA). The detailed
analysis of the above pathways will be discussed in the
subsequent sections.

1 Glutamine

Other Essential
Amino Acids

Cellular

Cell M a s s
Cell

Product

Figure 1. Schematic pathways of the synthesis of cell mass and product.

XIE AND WANG: ANIMAL CELL GROWTH ANALYSIS

1165

metabolism of glucose and glutamine for energy production

is interrelated depending on the concentration of glucose
and glutamine, and the pH of the culture
At a very low glucose concentration, the consumption of
glutamine increases due to its role in energy prod~ction.'~
Generally, at high pH, glucose is the major energy source.
At low pH, glutamine may contribute a significant amount
of energy for animal cell growth.
Energy production from glutamine is not preferred and is
not included in the model based on the following reasons.
First, the consumption of glutamine for energy production
generally found in conventional batch culture is not an
essential role of glutamine in animal cell growth. The
essential roles of glutamine are to provide nitrogen source
or nucleotide synthesis and to provide precursor for protein
synthesis. Second, both energy and ammonia are byproducts in the synthesis of nonessential amino acids from
glutamine. It is well known that ammonia is very toxic to
animal cell^.^^^'^,^' In order to decrease ammonia formation,
glutamine consumption needs to be minimized. Hence, the
nonessential role of glutamine in energy production needs
to be minimized if possible. Third, it has been shown that
ammonia production from glutamine can be significantly
reduced by providing a limited amount of glutamine to
cell^.'^,^^ Under this condition, the utilization of glutamine
for energy production will be low. Fourth, neglecting the
contribution of glutamine to energy formation can greatly
simplify the stoichiometric model and thus make the model
of greater utility. However, this simplification will tend
to underestimate the glutamine requirement, which will
render a glutamine-limited environment (this will help to
decrease ammonia formation). Based on the above analysis,
glucose is assumed to be the only nutrient source for ATP
production in this mode, even though this simplification
will tend to overestimate the requirement for glucose.
The calculation of the exact number of ATP molecules
produced per glucose molecule consumed for energy production is very complicated in animal cell cultivation due
to the complex glucose metabolic pathways. Only two
ATP molecules are produced if glucose is incompletely
oxidized into lactate. However, more ATP can be generated
if glucose is completely oxidized into CO2 and water via
the tricarboxylic acid (TCA) cycle. In addition, because
lactate is known to inhibit both cell growth and product
secretion," incomplete oxidation of glucose needs to be
minimized. It was found that the conversion of glucose to
lactate varies and depends on the culture conditions." At a
low glucose concentration, lactate formation from glucose
can be significantly d e ~ r e a s e d . ~Therefore,
~ , ~ ~ high lactate
formation is not necessary for cell growth and is only a
result of high glucose concentration in the culture medium.
Two cases will be studied in the model. In the first case,
glucose is assumed to be completely oxidized for ATP
production. This assumption will tend to underestimate the
requirement for glucose (this might be corrected partially
by the overestimation resulting from neglecting the contribution of glutamine for energy production). In the second

1166

case, a model including lactate formation will be analyzed

and discussed. Stoichiometrically, one glucose can generate
36 ATPs if one assumes a P/O ratio of 3.
C6H1206

__*

36 ATP

(2)

Synthesis of Carbohydrates
Cellular carbohydrates are a mixture of sugars and their
polymers.' Most of the sugars are interconvertible with
glucose and are derived directly from the pentose phosphate
cycle of glucose metabolism. In addition to glucose, most
cell lines can use alternative sugars such as galactose and
fructose as their energy and carbon s o u r ~ e . However,
~,~~
for most cell lines, cells grow most optimally with glucose.
Hence, glucose is usually supplied in culture media as an
energy and carbon source.
Carbohydrates have a similar overall chemical composition and can be expressed as (CHzO),. Based on a
carbon balance, one glucose can provide 6 carbons for the
carbohydrate synthesis, as shown in Eq. (3).
C6H1206

6- (CHzO)-carbohydrates

(3)

Synthesis of Lipids
The most important carbon source for the synthesis of
cellular lipids in animal cell cultivation is glucose, even
though lipids can be synthesized directly from intermediates
other than from glucose. Cellular lipids are a mixture
of fats, phospholipids, and steroid^.^ Both phospholipids
and steroids are important components of the animal cell
membranes. Fats and phospholipids are synthesized from
two major precursors, glycerol and fatty acids.33 Glycerol
is derived mainly from the intermediates of the glycolytic
pathway and the pentose phosphate cycle. Fatty acids can be
utilized from the culture medium or can be synthesized from
a ~ e t y l - C o A . ~ 'Because
,~~
fatty acids are absent in dialyzed
serum, and are not added to the medium, they must be
synthesized from acetyl-CoA. Lipids carried by albumin
and lipoproteins in serum can be metabolized by cultured
cells. However, it is very difficult to quantify this factor due
to their unknown compositions in the serum. Hence, the
fatty acid contribution in serum is neglected in the model.
Cholesterol, which is one of the most abundant steroids,
is an important component of the animal cell membranes
as well as a precursor for other steroids. Cholesterol is also
synthesized via a ~ e t y l - C o A .Hence,
~~
steroids are assumed
to be derived from acetyl-CoA, either directly or indirectly.
Acetyl-CoA can be derived from pyruvate, which is an
important intermediate in the glycolytic pathway and the
TCA cycle. Because no glutamine is assumed to enter the
TCA cycle for energy production, pyruvate is assumed to
be derived from glucose only.
The uptake of leucine in animal cell cultivation is usually greater than the uptake of most of the other amino
acids.I7 It has been suggested that leucine is converted into
acetyl-CoA for the synthesis of lipids.I8 When leucine is

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 11, MAY 1994

converted into acetyl-CoA, the amino group in leucine is

transaminated to glutamate and could be further hydrolyzed
to produce a r n m ~ n i a Hence,
.~
the synthesis of lipid from
leucine is not preferred if ammonia formation is to be
decreased. It is assumed that, at a low concentration of
leucine, the formation of acetyl-CoA from leucine is negligible compared with the formation of acetyl CoA from
pyruvate. Therefore, the synthesis of lipids from leucine
was neglected in the model. Glucose is consequently considered as the only nutrient source for acetyl-CoA. From
the above analysis, all of the precursors required for lipid
synthesis are therefore assumed to be directly or indirectly
derived from glucose.
A n empirical formula of the lipid mixture is assumed
to be (CHzOxPy)rn. Phospholipids have a small fraction
of oxygen and phosphate. Fats and steroids also have
a very small fraction of oxygen. The total fraction of
oxygen and phosphate in cellular lipids is usually small
and unknown. It is assumed that the fraction of oxygen
and phosphate is negligible compared with the fractions of
carbon and hydrogen, and therefore can be neglected. Thus,
the empirical formula of lipids is simplified to be (CH2), .
Stoichiometrically, one glucose can provide 6 carbons for
lipid synthesis as shown in Eq. (4).
C6H1206

6- (CHz)-lipids

(4)

Synthesis of Nucleotides
The deoxyribose in DNA and ribose in RNA are synthesized from the pentose phosphate c y ~ l e . The
~ ~ ,nutrient
~~
source for these pentoses is glucose. Stoichiometrically,
1 glucose can provide 6 carbons for the synthesis of these
pentoses as shown in Eqs. (5) and (6).
C6H1206
C6H1206

6
--(X-C5H70-P04)-DNA
5

(5)

6
--(X-CgH702-P04)-RNA

(6)

+ Asp + Gly
2Gln + Asp + 2Gly

guanine

+ 3Glu

(7)

adenine

+ 2Glu

(8)

In the synthesis of pyrimidine, cytosine is first synthesized from glutamine and asparatic acid. Uracil is then
synthesized from cytosine. Thymine is further derived
from u r a ~ i l . ~Therefore,
,~~
the stoichiometric coefficients
for glutamine and asparatic acid in the equations for the
pyrimidine synthesis are assumed to be the same and are
shown in the following equations:
2Gln
2Gln

+ Asp

+ Asp

+ 2Glu
+ 2Glu + NH,'

cytosine

uracil

+ Asp-

thymine

+ 2Glu + NH;

(11)

Synthesis of Proteins
The synthesis of cellular and product protein requires all
of the 20 amino acids. Cellular proteins are a mixture of
enzymes, membrane proteins, structural proteins, etc. The
amino acid sequences for most of the proteins in cells are
unknown. To simplify this problem, the synthesis of the
protein mixture in the cells is stated to be a single equation.
20

6$[amino acidsIi = proteins

(12)

i=l

The stoichiometric coefficient for amino acid in Eq. (12),

6i,i, is thus the overall amino acid composition of the
cellular proteins and could be measured experimentally.
In our model, the product in animal cell cultivation is
a monoclonal antibody. In many cases, the amino acid
sequences of the product are often unknown. This problem
can be solved if the amino acid composition of the product
protein can be determined experimentally. To avoid the
detailed purification and protein characterization needed,
our approach, as a first approximation, was the following.
In many instances, the protein product from animal cell
cultivation is only a very small fraction of the total cellular
proteins. When the amino acid composition of the product
is unavailable, the amino acid composition of the product
protein is assumed to be similar to the overall cellular
protein composition without encountering significant errors.
Hence, the amino acid compositions of cellular proteins and
product are assumed to be the same in the model.

DETERMINATION OF STOlCHlOMETRlC
COEFFICIENTS
Stoichiometric Coefficient for Product

Three amino acids, glutamine, aspartic acid, and glycine,

are involved in the synthesis of n u ~ l e o t i d e s . ~Based
,~~
on the nitrogen balance, the following equations show
the stoichiometric relationship between amino acids and
purine:
3Gln

2Gln

(9)

(10)

The stoichiometric coefficient for product (6,) is defined

by
gram of product secreted
6, =
(13)
number of cells produced
The cell growth kinetics can be expressed by Eqs. (14),
(15), and (16). In the stationary phase of the animal cell
culture, the specific growth rate decreases with increasing
culture time. This is due to the decreasing nutrient concentrations and the increasing waste product concentrations. In
contrast, a constant specific growth rate is usually observed
in the exponential phase. If the nutritional environment can
be well controlled, such as that in continuous culture, a high
or constant specific growth rate can also be maintained. We
will assume that this can be achieved in order to eliminate
the time variable from the model. The viable and the total
cell densities are then derived by integrating Eqs. (14) and
(16) to yield Eqs. (17) and (18).
dxv- - ( p -

dt

XIE AND WANG: ANIMAL CELL GROWTH ANALYSIS

1167

dxd
--

Stoichiometric Coefficient for Glucose

ax,

dt
dxt- -

px,

dt

The product formation rate can be expressed by Eq. (19).

By integrating Eq. (19), the product concentration is expressed as a function of the viable cell density, the specific
product synthesis rate, and culture time [Eq. (20)].

dP
dt - 4PXV

- -

PO +

lt

q p X v dt

PO +

(19)

qpxvo
~

P - a

[e@-(I)' - 1

(20)
A general model for the specific product synthesis rate is
assumed to consist of two terms: a growth-associated term,
and a non-growth-associated term:
qp

= ap

+b

(21)

According to Eqs. (13), (18), (20), and (21), the stoichiometric coefficient for product can be obtained as:

Stoichiometric Coefficient for ATP

The stoichiometric coefficient for ATP can be expressed by
its definition as:

OATP =

Jof4ATPxvdt
mole of ATP consumed number of cells produced
Xr - Xt0
(23)

According to a maintenance model, ATP is required

for the synthesis of macromolecules and also for
m a i n t e n a n ~ e . ~ ~ ,The
~ ' ? ~specific
'
ATP consumption rate is
thus defined as:

qATP

= - + mATP

(24)

YATP
The cellular yield on ATP, YATP, can be estimated
from the energy needed for adding monomer units to
macromolecules in the synthesis of cell mass. The ATP
requirement for maintenance, m A v , can be estimated
experimenta~ly.~~,~~
Because the specific growth rate is assumed to be constant, the specific ATP consumption rate will also be
constant, according to Eq. (24). Therefore, by inserting
Eqs. (17) and (18) into Eq. (23), OATp can be simplified
as:

The determination of the stoichiometric coefficients for

nutrients requires the cell composition. In our model, we
have assumed that the average dry cell composition (dry
weight percentage) is composed of the following: proteins
(Zp), DNA ( Z D ) ,RNA (ZR),lipid (Z L),and carbohydrate
(ZC). The average total dry cell weight is denoted W
(mg/cell).
Based on the stoichiometric analysis, glucose is used for
the synthesis of lipids, carbohydrates, pentoses in DNA
and RNA, and ATP production. Consequently, the stoichiometric coefficient for glucose is the sum of these five
components.
The total moles of carbon in cellular carbohydrates per
cell is the weight of carbohydrates per cell, Z c W , divided
by the molecular weight of the carbohydrate unit, CH20.
The stoichiometric coefficient for glucose in carbohydrate
synthesis is derived from Eq. (3):
,gcarb =
ZCW/[MCH20
g lc

61

(26)

The total moles of carbon in cellular lipids per cell is the

lipid weight per cell, ZLW, divided by the unit molecular
weight (CH2). According to Eq. (4), the stoichiometric
coefficient for glucose in the synthesis of lipids is:

The calculation of the moles of pentoses in DNA and

RNA requires the base compositions (molar percentage).
Assume that DNA has the base composition (molar percentage) of adenine (DA), guanine (DG), cytosine (Dc),
and thymine (DT); and RNA has the base composition
(molar percentage) of adenine (RA),guanine (RG),cytosine
(Rc), and uracil (Ru). The average molecular weights of
the basic units of DNA, X-CSH~O-PO~ ( M D ) , and
RNA, X-C5H702-P04
( M R ) ,are derived from the base
compositions of DNA and RNA.
MD = MP04

+ MC5H70 + DAMCSH~NS+ DGMCsHlONs

-k

MR

D C M C ~ H ~+ODNT~ M c ~ H ~ o(28)
~N~

Mm4 + M C ~ H , +
O ~R A M c ~ H+
~N
R~G M C ~ H ~ O N ~

+ RCMC4HdON3 + RUMC4H302N2 (29)

The number of moles of deoxyribose in DNA is readily
calculated by the DNA weight per cell, Z D W , divided by
the average molecular weight of the basic unit. According
to Eqs. (5) and (6), the stoichiometric coefficients for
glucose in the synthesis of DNA and RNA are expressed
in Eqs. (30) and (31), respectively.

According to Eqs. (1) and (2), the stoichiometric coefficient of glucose in ATP production is the stoichiometric
1168

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 11, MAY 1994

coefficient for ATP, eATp, divided by the number of ATP

produced per glucose oxidized for ATP production:

eiYc = @ A T P / ~ ~

(32)
Based on the above analysis, the stoichiometric coefficient for glucose in the stoichiometric equation governing
cell growth will be the sum of the above five stoichiometric coefficients for glucose in the synthesis of the cell
components and energy:

ZR w
6
MR X -

ATP
+ -~ 36

The stoichiometric coefficients for glutamine, glycine,

aspartate, and glutamate are calculated by adding the stoichiometric coefficients in the synthesis of DNA and RNA,
and the stoichiometric coefficients in the synthesis of cellular proteins and the product. The stoichiometric coefficients
for the other amino acids are simply the stoichiometric
coefficient in the synthesis of cellular proteins and product.

(45)

(33)

Stoichiometric Coefficients for Amino Acids

Stoichiometric Coefficients for Vitamins

The total moles of nucleotides in DNA (NDNA)and RNA

(NRNA)can be calculated from the cell composition and
the average molecular weights of DNA and RNA. The
number of moles for each base per cell is then calculated
by multiplying the total moles of nucleotides by the base
compositions of DNA and RNA.

NDNA= Z D W / M D

(34)

NRNA

(35)

zRw/MR

NA = NDNADA+ NRNARA

(36)

NG

NDNADG+ NRNARG

(37)

Nc

NDNADC+ NRNARC

(38)

NT

(39)

NDNADT

Nu = NRNADU
(40)
The requirements for glutamine, glycine, and aspartate
for synthesis of DNA and RNA can be calculated according
to Eqs. (7)-(11). This is done by multiplying the number
of moles for each base by the stoichiometric coefficient for
the corresponding amino acid in Eqs. (7)-(11).

e,"g = ~

N G
+ ~ [ N A+ N c

+ NT + N U ]

+ 2N*
ea"s"p'= NG + NA + N~ + NT + N~
eg

NG

(41)
(42)

(43)
In the synthesis of nucleotides, only the amide group in
glutamine is incorporated into nucleotides, and glutamate
is produced. The amount of glutamate produced in the
synthesis of nucleotides is exactly the amount of glutamine
consumed. A minus sign is used to denote production rather
than consumption.

ez

= - eglr

(44)

All of the 20 amino acids are needed for the synthesis of

cellular proteins and the product protein. The stoichiometric
coefficient for amino acid in the synthesis of cellular
proteins and the product is calculated according to Eqs. (1)
and (12) as e$[Z,W + e,].

Vitamins are essential nutrients for animal cell survival and

growth in vitro, even though the quantities required are
much smaller compared with amino acids and g l u ~ o s e . ~ , ' ~
The cell components synthesized from vitamins are mainly
enzymes and cell membrane components. The amounts of
these components are difficult to measure experimentally
due to the larger number of species involved. Therefore,
the stoichiometric coefficients for vitamins cannot be determined through the same approach devised for amino acids
and glucose.
The method used here to estimate the stoichiometric
coefficients for vitamins is experimental. Due to the small
amount consumed, the vitamin concentrations in the culture
medium is usually very low. Depletion of any essential
vitamins can result in cell death, but high concentrations
of vitamins show no harm to animal cells. Therefore, the
most important condition here is to ensure that no vitamin
depletion will occur during the cultivation process. The
above analysis results in the use of the vitamin yields (Yv,;)
to determine the stoichiometric demands for cell growth.
By definition, the stoichiometric coefficients for vitamins
are given by:

ev,; = l/Yv,,

(vitamins)

(50)

APPLICATIONS IN MEDIUM DESIGN FOR

ANIMAL CELL CULTURES
One of the most promising applications of the stoichiometric study is in the design of culture media. Once the
stoichiometric equation governing animal cell growth is
solved, the concentration of each nutrient required for the
synthesis of cell mass and product can be determined.
According to Eq. (l),the stoichiometric coefficient for
the kth nutrient in the group of glucose, amino acids, and
vitamins, e k , is defined by:
moles of nutrient consumed
ek =
cell number produced

XIE AND WANG: ANIMAL CELL GROWTH ANALYSIS

1169

C, is the total concentration of glucose, amino acids,

and vitamins in the medium. This is an extremely important parameter and it must be determined experimentally.
The molar fraction of the kth nutrient is defined by the
concentration of the kth nutrient divided by Cf.
By definition, the sum of the molar fractions of glucose,
amino acids, and vitamins should equal 1. By adding all
of the stoichiometric coefficients for glucose, amino acids,
and vitamins together, the following equation is obtained:

The above analysis allows the determination of relative

requirements for each nutrient. The molar fraction of the
kth nutrient is solved from Eqs. (51) and (52). The concentration of each nutrient required can then by determined by
multiplying the molar fraction by the total concentration of
the nutrients.
Xk =

ek/p

(53)
(54)

The stoichiometric coefficients are constants if the average

cell composition, the specific growth rate, and the specific product synthesis rate are assumed to be constant.
Therefore, the model developed using Eqs. (51)-(54) is
constrained by these assumptions. However, it is our opinion that this is still a useful first approximation toward the
design of medium and its subsequent feeding strategy to
maximize cell density and product concentration.

Batch Culture Medium Design

In conventional culture media, the relative concentrations
of glucose, amino acids, and vitamins are often suboptimal.
Usually, the maximum cell density in batch culture is
limited either by the accumulation of toxic waste-products
and/or nutrient depletion. When nutrient depletion occurs,
often only one of the essential nutrients is stoichiometrically
limiting. It is perceived that maximum cell density and
product concentration in conventional batch culture can
be increased to the point when toxic waste accumulation
becomes limiting to cell growth. This requires that nutritional composition is properly designed according to the
stoichiometric needs of the cells.
According to Eqs. (53) and (54), in order to determine
the nutrient concentration, the only unknown parameter is
C,,the total concentration of glucose, amino acids, and
vitamins. It is apparent that the maximum cell density in
the conventional batch culture increases with the increase
of C,. However, the maximum cell density will no longer
increase with the increase of the total nutrient concentration
when the toxic waste-products reach a certain inhibitory
level. Therefore, the optimal C, can be determined experimentally simply by increasing Cfto achieve maximum cell
density. The concentrations of the other components in the
culture medium (except NaCl) can be set to be the same

1170

as those in the basal media. The concentration of NaCl

needs to be adjusted to maintain a proper osmolality for
cell growth.

Supplemental Medium Design in

Fed-Batch Culture
The most exciting application of this stoichiometric study
is that it can be employed to design a supplemental medium
and to determine a proper feeding strategy to control the
nutritional environment. In conventional fed-batch cultivation, the process is initiated with a conventional medium.
Glucose and/or glutamine are the most commonly employed
supplemental nutrient^.^'^'^,'^ Concentrated amino acids
from the conventional medium are also employed in some
fed-batch systems.20 Although feeding cells with these
empirically selected nutrients can eliminate the stoichiometric limitation of these nutrients, the culture performance
could still be stoichiometrically limited by other essential
nutrients.
The stoichiometric demands for nutrients in animal cell
growth can be properly satisfied by formulating a supplemental medium, which should contain all of the possible
limiting nutrients using the above stoichiometric analysis.
The idea is to maintain the nutritional environment at a
relatively constant level to ensure optimal and stable cell
growth. The consumption of nutrients for cell mass and
product synthesis is compensated by feeding supplemental
medium into the fed-batch reactor.
The molar fractions can be determined by Eq. (51).
Again, the problem here is to determine Cf,the total
concentration of nutrients in the supplemental medium.
Generally, animal cells show a very narrow osmolality
range of optimal cell growth. The osmolality in the reactor
is affected by the following factors: feed flow due to
dilution; and cell growth due to the consumption as well
as production of low molecular weight species. In order to
maintain a constant osmolality in the reactor, the osmolality
of the supplemental medium needs to be designed to adjust
for the osmolality changes during cell growth. Usually, the
increase in osmolality resulting from lactate and ammonia
formation is compensated by the decrease in osmolality resulting from the incorporation of amino acids into cell mass.
Therefore, the osmolality of the supplemental medium can
be selected to be the same as the osmolality in the reactor
(usually in the range of 270 to 330 mOsm/kg H20). The
total concentration of nutrients in the supplemental medium
is partially related to the osmolality. The formulation of the
supplemental medium can be determined by Eq. (54).
The feeding strategy employed in the conventional fedbatch is to maintain the concentration of a key nutrient
such as glucose or glutamine above a certain level. Usually
a kinetic model or an empirical yield on the key nutrient is
employed to predict its utilization. A corresponding feeding
strategy is then determined by estimating consumption
of the key nutrient. Therefore, only one nutrient can be
controlled in the conventional fed-batch culture.

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 11, M A Y 1994

The stoichiometric analysis provides a rational feeding

strategy for fed-batch cultivation. The composition of the
supplemental medium is determined according to the demand for cell growth. Therefore, all of the nutrients are
collectively fed and assumed to be consumed in an identical
pattern. This in turn allows the possibility to control the
entire nutritional environment as a single input variable.
The amount of the supplemental medium needs to be
fed can be derived from Eq. (52). The total concentration
and the total stoichiometric coefficient can be determined
from the stoichiometric analysis. Therefore, the amount to
be fed is linearly correlated with the increase of the total
cell number in the reactor, and can be estimated from the
cell growth kinetics by Eq. (55).

AV

PANJC,

(55)

DISCUSSION
In the proposed stoichiometric model, the detailed pathways
of animal cellular metabolism were not considered. The
detailed processes of cell mass and product synthesis have
been formulated into a single parameter. Because the energy
metabolism is complicated in animal cells, simplifications
were made in order to solve the model without introducing
too many empirical or adjusted parameters. These simplifications make it possible to solve the complex problem
and make the model relatively simple and useful in culture medium design. This work represents our first step
in designing culture medium rationally according to the
stoichiometric demands for nutrients in animal cell culture.
Sufficient details of the processes of cell mass and
product formation have been taken into account in the
model. However, the production of lactate and nonessential
amino acids have been omitted. These will be discussed in
the next two subsections.

glucose is not included in the model because the molar ratio

of lactate to glucose varies from case to case depending
on the glucose and glutamine concentrations in the culture
medium.
However, lactate formation can be accounted for in the
application of the stoichiometric model by adjusting the
number of ATP molecules produced per glucose molecule
oxidized according to the ratio of lactate to glucose. The
following equation represents the formation of ATP and
lactate from incomplete glucose oxidation.

C6H1206
2ATP + 2C3H603 Lactate
(56)
Assuming that the molar fraction of glucose consumed
for lactate formation in the total glucose consumed for ATP
production is y , the average number of ATP molecules
produced per glucose molecule consumed for ATP production is 2y + 36(1 - y). The stoichiometric coefficient
for glucose in the production of ATP can be derived from
Eqs. (1) and (2):
gen
glc = Q A T P / [ ~ Y

+ 36(1

711

(57)
where e A T p is the stoichiometric for ATP in Eq. (1) and can
be calculated from Eq. (25). Glucose consumed for lactate
formation is then given by Y@ATP/[2y + 36(1 - y ) ] . According to Eq. (56), the moles of lactate produced per cell
is 2yOATP/[2y + 36(1 - y ) ] . The total stoichiometric coefficient for glucose is then adjusted according to Eqs. (33)
and (57):

We further define

Lactate Production
In conventional batch cultivation of animal cells, a significant amount of glucose is converted into lactate often due to
the high glucose concentration in the culture m e d i ~ m . ' ~ , * ~and
een =
~ATP
Therefore, a significant portion of the energy is directed
glc
toward the formation of lactate. In some cases, the molar
2y
36(1 - y )
ratio of lactate to glucose can be greater than 2, especially
By definition, the average molar ratio of lactate to
at the beginning of conventional batch cultivation. In this
glucose Ylac/glc is:
case, glutamine may have contributed to pyruvate formation
and then to lactate and energy production. In most cases,
the overall molar ratio of lactate to glucose can vary
from 0.7 to 1.5 which suggests that 35% to 75% of
Ylac/glc can be measured experimentally. By combining
glucose has been converted into l a ~ t a t e . " , ' ~ , In
~ ~ the
, ~ ~ Eqs. (60) and (61), the fraction of glucose converted to
process of lactate formation, only two ATP molecules are
lactate can be solved as:
generated per glucose molecule consumed. Therefore, the
overall stoichiometric coefficient for ATP formation from
glucose is lower than 36, and the demand for glucose
Therefore, lactate formation can be taken into considcould be greater than the theoretical value calculated from
eration
in the stoichiometric model if the average molar
the stoichiometric analysis. The formation of lactate from

XIE AND WANG: ANIMAL CELL GROWTH ANALYSIS

1171

ratio of lactate to glucose Ylaclglc can be experimentally

determined. The above analysis is based on the assumption
that lactate and energy productions are derived only from
glucose. Therefore, the average molar ratio of lactate to
glucose must be less than or equal to 2 in order to obtain
a reasonable solution from the above equations. In the
case in which Ylaclglc is greater than 2, the contribution
of glutamine to the energy and lactate production has to be
considered, but it will not be discussed here. Fortunately,
Ylaclglc is usually less than 2. Hence, the above analysis is
usually applicable.

Synthesis of Nonessential Amino Acids

Eagle' showed that animal cells require only 13 essential
amino acids to survive in vitro. The other seven amino
acids needed for the synthesis of cellular proteins can be
synthesized from glutamine and therefore are regarded as
nonessential amino acid^.^,^ The demands for nonessential
amino acids in cell mass and product synthesis are about
45% of the total demands for amino acids, based on the
stoichiometric analysis. However, only 10% to 20% of the
total amino acids are nonessential amino acids in most
basal media.
Because nonessential amino acids are not sufficiently
present in the basal medium, the requirements for nonessential amino acids for cell mass and product synthesis have
to be met by glutamine. According to the stoichiometric
analysis, the requirement for glutamine is therefore very
high. Consequently, the consumption rate of glutamine is
much greater than for the other amino acids. Hence, in conventional culture media, glutamine concentration is usually
quite high, which results in excess ammonia formation. It
is estimated that about 80% of ammonia is derived from
glutamine metabolism in the conventional batch culture
when 4 mM glutamine is used in the medium.26 The
accumulation of ammonia is usually one of the major
factors causing a low maximum cell density and a low
product concentration.
Therefore, the consumption of glutamine for the synthesis of nonessential amino acids is not desired due to the
accompanying formation of ammonia. The key strategy for
decreasing ammonia formation is to increase the incorporation of extracellular nonessential amino acids into cell
mass and product, and thus decrease ammonia formation
from glutamine.
However, under actual culture conditions, the minimum
requirement of glutamine for the synthesis of nonessential amino acids is unknown. The stoichiometric analysis
can only determine the total demands for glutamine and
nonessential amino acids as a group.
In order to minimize ammonia formation, the concentrations of glutamine and nonessential amino acids need
to be studied further using stoichiometric analysis as a
guide. Without an exact model to determine the concentrations of glutamine and nonessential amino acids, empirical
methodologies will have to be employed.

1172

CONCLUSIONS
A stoichiometric model has been formulated in which
the stoichiometric demands for glucose, amino acids, and
vitamins in cell mass and product synthesis have been
analyzed. The stoichiometric coefficients for these nutrients
have also been determined by studying their roles in animal
cell growth. Though many details have been considered
in the model, the complicated energy metabolism had
to be simplified. These simplifications make the model
solvable, but also introduce errors in the estimation of the
requirement for glucose, glutamine, and nonessential amino
acids. This problem needs to be refined further. The new
approach can nevertheless be employed in medium design
as well as in the design of a feeding strategy for fed-batch
systems. The application of this approach is discussed in
the article on page 1175 of this issue,
The authors acknowledge the financial support from the National
Science Foundation (EEC 88-03014) under their Engineering Research Center initiative.

NOMENCLATURE
a

M
mATP

MD
MR
n

NA. N G , Nc, N T . Nu

NDNA,NRNA

t
V

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 11, MAY 1994

growth-associated constant in specific

product synthesis rate (mg/cell)
non-growth-associated constant in specific product synthesis rate (mg/cell .
h)
total concentration of glucose, amino
acids, and vitamins in medium (mM)
concentration of the kth nutrient in
medium (mM)
DNA base composition of adenine, guanine, cytosine, and thymine, respectively (mol %)
molecular weight
maintenance ATP consumption rate
(mmol ATP/cell . h)
average molecular weight of DNA basic unit (X-CSH~O-PO~)
average molecular weight of RNA hasic unit (X-CsH702-PO4)
total specie number of glucose, amino
acids, and vitamins
total moles of adenine, guanine, cytosine, thymine, and uracil per cell,
respectively (mmol/cell)
in
total moles of X-C5H70-P04
DNA and X-C5H702-P04
in RNA
per cell (mmol/cell)
total cell number in the reactor (cells)
product concentration at time t and
time t = 0, respectively (mg/L)
moles of ATP generated per mole
NADH oxidized
specific ATP consumption rate (mmol
ATP/cell . h)
specific product synthesis rate (mg/
cell . h)
RNA base composition of adenine, guanine, cytosine, and uracil, respectively
(mol %)
culture time (h)
medium volume (L)

X
Xd

xk

average dry cell weight (mg dry cell

mass/cell)
stands for DNA and RNA bases (G, A,
T, C, or U)
dead cell density (cells/L)
molar fraction of the kth nutrient in
the group of glucose, amino acids, and
vitamins
total cell density at time t and time
t = 0, respectively (ceIls/L)
viable cell density at time f and time
t = 0,respectively (ceIIs/L)
cellular yield on ATP (cells/mmol ATP)
molar ratio of lactate produced over
total glucose consumed (mol/mol)
yield on the jth vitamin (cell/mmol)
dry weight percentage of protein, DNA,
RNA, lipids, and carbohydrates in cell
mass, respectively

Greek letters
P
ff

P
Y

e
8a.i

specific growth rate (l/h)

specific death rate (l/h)
total stoichiometric coefficient for glucose, amino acids, and vitamins (mmol/
cell)
molar ratio of incompletely oxidized
glucose over total glucose oxidized
(mol/mol)
stoichiometric coefficient (mmol/cell)
stoichiometric coefficient for the ith
amino acid (mmol/cell)
composition of glutamine, glycine, aspartate, and glutamate, respectively, in
cellular proteins (mmol/mg proteins)
amino acid composition of cellular proteins (mmol of the ith amino acids/mg
proteins)
stoichiometric coefficient for glucose in
Eq. (I) (mmol/cell)
stoichiometric coefficient for the kth
nutrient (mmol/cell)
stoichiometric coefficient for ATP in
Eq. (I) (mmol/cell)
stoichiometric coefficient for glucose in
energy production, in the synthesis of
the cell mass excluding energy (mmol
glucose/cell)
stoichiornetric coefficient for glucose in
the synthesis of carbohydrates, DNA,
lipids, and RNA (mmol glucose/cell)
stoichiometric coefficient for glutamine,
glycine, aspartate, and glutamate, respectively (mmol/cell)
stoichiometric coefficients for glutamine, glycine, aspartate, and glutamate in the synthesis of DNA and RNA
(mmol/cell)
stoichiometric coefficient for product
(mg/celI)
stoichiometric coefficient for the jth
vitamin (mmol/cell)

amino acid
aspartate
glucose
glutamine
glutamate

glY
I

i
k
lac
P
V

glycine
index of amino acids
index of vitamins
index of glucose, amino acids, and
vitamins
lactate
product
vitamin

References
1 Campbell, N. A. 1990. Biology. Benjamin/Cummings, New York.
2 Barton, M. E. 1971. Effect of pH on the growth cycle of HeLa cells in
batch suspension culture without oxygen control. Biotechnol. Bioeng.
13: 471-492.
3 Batt, B. C., Kompala, D. S. 1989. A structured kinetics modeling
framework for the dynamics of hybridoma growth and monoclonal
antibody production in continuous suspension cultures. Biotechnol.
Bioeng. 34: 515-531.
4. Bender, D.A. 1985. Amino acid metabolism, 2nd edition. Wiley,
New York.
5. Darnell, J., Lodish, H., Baltimore, D. 1990. Molecular cell biology.
2nd edition. Scientific American Books, New York.
6. Duval, D., Demangel, C., Munier-Jolain, K., Miossec, S., Geahel, I.
1991. Factors controlling cell proliferation and antibody production
in mouse hybridoma cells: 1. influence of the amino acid supply,
Biotechnol. Bioeng. 38: 561-570.
7. Eagle, H. 1955. Nutrition needs of mammalian cells in tissue culture.
Science 122: 501-504.
8. Eagle, H. 1959. Amino acid metabolism in mammalian cell cultures.
Science 130: 432-437.
9. Eagle, H., Barban, S., Levy, M., Schulze, H.O. 1958. The utilization of carbohydrates by human cell cultures. J. Biol. Chem. 233:
551-558.
10. Glacken, M.W., Fleischaker, R.J., Sinskey, A.J. 1986. Reduction of waste product excretion via nutrient control: possible strategies for maximizing product and cell yields on
serum in cultures of mammalian cells. Biotechnol. Bioeng. 28:
1376-1389.
11. Glacken, M. W. 1987. Development of mathematical descriptions of
mammalian cell culture kinetics for the optimization of fed-batch
bioreactors, Ph.D. thesis, MIT, Cambridge, MA, USA.
12. Glacken, M. W., Huang, C., Sinskey, A. J. 1989. Mathematical descriptions of hybridoma culture kinetics. 111. Simulation of fed-batch
bioreactors. J. Biotechnol. 10: 39-66.
13. Griffiths, J.B. 1987. Serum and growth factors in cell culture media-an introductory review. Devel. Biol. Stand. 66: 155- 160.
14. Hu, W. S., Dodge, T. C., Frame, K. K., Himes, V.B. 1987. Effect of
glucose on the cultivation of mammalian cells. Devel. Biol. Stand.
66: 279-290.
15. Jayme, D. W., Blackman, K. E. 1985. Culture media for propagation
of mammalian cells, viruses, and other biologicals. Adv. Biotechnol.
Proc. 5: 1-30,
16. Kelley, W.N. 1972. Purine and pyrimidine metabolism of cells
in culture, pp. 211-256. In: G.H. Rothblat and V.J. Cristofalo
(eds.), Growth, nutrition, and metabolism of cells in culture, vol. 1.
Academic Press, New York.
17. Lambert, K., Pirt, S. J. 1975. The quantitative requirements of human
diploid cells (strain mrc-5) for amino acids, vitamins, and serum. J .
Cell Sci. 17: 397-411.
18. Lazo, P.A. 1981. Amino acids and glucose utilization by different
metabolic pathways. Eur. J. Biochem. 117: 19-25.
19. Ljunggren, J., Haggstrom, L. 1992. Glutamine limited fed-batch
culture reduces the overflow metabolism of amino acids in myeloma
cells. Cytotechnology 8: 45-56.
20. Luan, Y.T., Mutharasan, R., Magee, W. E. 1987. Strategies to extend the longevity of hybridomas in culture and promote yield on
monoclonal antibodies. Biotechnol. Lett. 9: 691 -696.

XIE AND WANG: ANIMAL CELL GROWTH ANALYSIS

1173

21. McKeehan, W. L. 1986. Glutaminolysis in animal cells, pp. 111-150.

In: M. J. Morgan (ed.), Carbohydrate metabolism in cultured cells.
Plenum Press, New York.
22. Miller, W. M., Blanch, H. W., Wilke, C. R. 1988. A kinetic analysis
of hybridoma growth and metabolism in batch and continuous suspension cultures: effect of nutrient concentration, dilution rate, and
pH. Biotechnol. Bioeng. 32: 947-965.
23. Miller, W. M., Wilke, C. R., Blanch, H. W. 1989. Transient responses
of hybridoma cells to nutrient additions in continuous culture: I.
Glucose pulse and step changes. Biotechnol. Bioeng. 33: 477-486.
24. Miller, W.M., Wilke, C.R., Blanch, H. W. 1989. The transient
responses of hybridoma cells to nutrient additions in continuous
culture: 11. Glutamine pulse and step changes. Biotechnol. Bioeng.
33: 487-499.
25. Morgan, M. J., Faik, P. 1986. The utilization of carbohydrates by
animal cells: an approach to their biochemical genetics, pp. 29-75.
In: M. J. Morgan (ed.), Carbohydrate metabolism in cultured cells.
Plenum Press, New York.
26. Nissim, I., States, B., Yudkoff, M., Segal, S. 1987. Characterization
of amino acid metabolism by cultured rat kidney cells: study with
"N. Am. J. Physiol. 253 (Renal Fluid Electrolyte Physiol. 22):
F1243 - 1252.

1174

27. Paul, J. 1965. Carbohydrate and energy metabolism, pp. 239-276. In:
E. N. Wilmer (ed.), Cells and tissues in culture. Academic, New York.
28. Reitzer, L. J., Wice, B.M., Kennell, D. 1980. The pentose cycle:
control and essential function in HeLa cell nucleic acid synthesis.
J. Biol. Chem. 255: 5616-5626.
29. Savinell, J. M., Palsson, B. 0. 1992. Network analysis of intermediary
metabolism using linear optimization. I. Development of mathematical formation. J. Theor. Biol. 154: 421-454.
30. Stouthamer, A.H. 1973. A theoretical study on the amount of ATP
required for the synthesis of microbial cell material. Antonie van
Leeuwenhoek 39: 545-565.
31. Stouthamer, A.H. 1979. The search for correlation between
theoretical and experimental growth yields, pp. 1-47. In: J. R. Quayle
(ed.), Microbial biochemistry. University Park Press, Baltimore,
MD.
32. Wang, H. Y. N. 1977. Computer control of yeast fermentations, Ph.D.
thesis, MIT, Cambridge, MA, USA.
33. White, A,, Handler, P., Smith, E. L. 1968. Principles of biochemistry.
4th edition. McGraw-Hill, New York.
34. Zielke, H. R., Ozand, P. T., Tildon, J. T., Sevdalian, D. A,, Cornblath,
M. 1976. Growth of human fibroblasts in the absence of glucose
utilization. Proc. Natl. Acad. Sci. USA 73: 4110-4114.

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 11, MAY 1994