Dependence of Morphology on Agitation

Intensity in Fed-Batch Cultures of

Aspergillus oryzae and Its Implications
for Recombinant Protein Production
A. Amanullah,1* L. H. Christensen,2 K. Hansen,3 A. W. Nienow,1
1 C. R. Thomas1
1

Centre for Bioprocess Engineering, School of Chemical Engineering,

The University of Birmingham, Edgbaston, Birmingham B 15 2TT, UK
2
Novo Nordisk, Novo Alle, DK-2880, Bagsvaerd, Denmark
3
Novozymes, Novo Alle, DK-2880, Bagsvaerd, Denmark
Received 26 March 2001; accepted 19 October 2001

Abstract: We previously reported that, although agitation conditions strongly affected mycelial morphology,
such changes did not lead to different levels of recombinant protein production in chemostat cultures of Aspergillus oryzae (Amanullah et al., 1999). To extend this
nding to another set of operating conditions, fed-batch
fermentations of A. oryzae were conducted at biomass
concentrations up to 34 g dry cell weight/L and three
agitation speeds (525, 675, and 825 rpm) to give specic
power inputs between 1 and 5 kWm 3. Gas blending was
used to control the dissolved oxygen level at 50% of air
saturation except at the lowest speed where it fell below
40% after 6065 h. The effects of agitation intensity on
growth, mycelial morphology, hyphal tip activity, and
recombinant protein (amyloglucosidase) production in
fed-batch cultures were investigated. In the batch phase
of the fermentations, biomass concentration, and AMG
secretion increased with increasing agitation intensity. If
in a run, dissolved oxygen fell below 40% because of
inadequate oxygen transfer associated with enhanced
viscosity, AMG production ceased. As with the
chemostat cultures, even though mycelial morphology
was signicantly affected by changes in agitation intensity, enzyme titers (AGU/L) under conditions of substrate
limited growth and controlled dissolved oxygen of >50%
did not follow these changes. Although the measurement of active tips within mycelial clumps was not
considered, a dependency of the specic AMG productivity (AGU/g biomass/h) on the percentage of extending
tips was found, suggesting that protein secretion may be
a bottle-neck in this strain during fed-batch fermentations. 2002 John Wiley & Sons, Inc. Biotechnol Bioeng 77:
815826, 2002; DOI 10.1002/bit.10181

2 Keywords: fed-batch fungal fermentations; agitation intensity; mycelial morphology; hyphal tip activity; recombinant protein

Correspondence to: A. Amanullah

*Present address: Fermentation and Cell Culture, Bioprocess R&D,
Merck Research Laboratories, Merck & Co. Inc., WP26C1-108, P.O.
Box 4, Westpoint, PA 19486, USA; Tel: +1-215-652-0423; fax: +1215-652-8691; e-mail: ashraf_amanullah@merck.com

2002 John Wiley & Sons, Inc.

INTRODUCTION
Filamentous fungi are being used increasingly as
eukaryotic hosts for foreign gene expression because of
their capability of secreting large quantities of proteins
(Peberdy, 1994). In addition, post-transcriptional modications of proteins, such as glycosylation, are important capabilities oered by these hosts (Mackenzie et al.,
1993), whereas many species are generally regarded as
safe by regulatory authorities. Despite the widespread
industrial use and potential of fungal strains for heterologous protein production, relatively little is known
about the inuence of engineering variables such as
agitation conditions on the morphology of such organisms in submerged cultures and, in turn, whether productivity itself is aected. In many fungal fermentations,
the high apparent viscosities and the non-Newtonian
behavior of the broths necessitate the use of high agitation speeds to provide adequate mixing and oxygen
transfer. However, mycelial damage at high stirrer
speeds (or power input) can limit the acceptable range of
speeds and, hence, the oxygen transfer capability and
the volumetric productivity of the fermenter. The eects
of mechanical forces (``shear'') on fungal physiology, in
particular branching, and tip extension are poorly understood.
Although many studies have been conducted to investigate the eects of mechanical forces on mycelial
morphology and productivity (Ayazi Shamlou et al.,
1994; Dion et al., 1954; Konig et al., 1981; Makagiansar
et al., 1993; Metz et al., 1981; Nielsen et al., 1995;
Reuss, 1988; Smith et al., 1990; Ujcova et al., 1980; van
Suijdam and Metz, 1981), they generally have from two
limitations. First, because of the lack of suitable
methods for characterizing clumps (Tucker et al.,
1992), only the freely dispersed form has been considered, although it may only account for only a small
fraction of the biomass (Justen et al., 1996; Tucker
et al., 1992). However, it should be noted that the

fraction of freely dispersed mycelia in a culture depends

on the type of the strain, its physiological state, and the
operating conditions of the bioreactor. In addition,
although the two-dimensional measurements of the
projected area of mycelial clumps are amenable by
image analysis, physiological measurements are generally limited to the freely dispersed class because of the
diculty of making measurements on hyphae within
the three-dimensional structure of clumps. Second, the
inuence of agitation has generally not been dissociated
from mass transfer eects. In addition, it should be
noted that product formation rates have been shown to
depend on agitation intensity (often referred to as
``shear'') for a wide variety of lamentous fungi (Markl
et al., 1991; Smith et al., 1990; Ujcova et al., 1980;
Vardar and Lilly, 1982).
We previously reported on the inuence of agitation
intensity in constant mass chemostat cultures of an industrial recombinant strain of Aspergillus oryzae at a
concentration of 2 g/L, with dissolved oxygen controlled
at 75% of air saturation (Amanullah et al., 1999). The
eects of agitation intensity on mycelial morphology
and protein production were investigated independently
of the eects of dissolved oxygen and gas holdup. The
proteins produced were a-amylase (homologous protein) and amyloglucosidase (AMG) (heterologous protein), an exo-attacking enzyme hydrolyzing 1,4-a- as
well as 1,6-a-linkages in starch, converting it to sugars,
syrups, and dextrins (Harvey and McNiel, 1994). Protein production was found to be independent of agitation intensity in the power per unit volume (P/V) range
of 2.212.6 kWm)3, although signicant changes in
mycelial morphology could be measured for similar
changes in agitation conditions. This nding suggested
that mycelial morphology did not aect protein production directly. Although there is very limited use of
continuous culture systems in industry, they are extremely useful research tools because they can give
precise information on the inuence of a single variable. However, it is important to verify the results in
fed-batch fermentations at industrially realistic conditions of biomass concentration and specic power input.
The present study reports on the inuence of agitation
conditions on the morphology and recombinant protein
production capability of the same genetically modied
industrial strain of Aspergillus oryzae but this time in
fed-batch cultures. This is an important issue, because it
has been suggested that protein secretion in fungi occurs
only at the tips (Wessels, 1990; 1993), so that there might
be a strong dependence of the secretion rate on mycelial
morphology if secretion is a bottleneck. The aim of the
experiments was to discover whether impeller induced
fragmentation occurred in fed-batch cultures, and if so,
whether recombinant protein production depended on
the change in morphology.

816

MATERIALS AND METHODS

The recombinant strain of Aspergillus oryzae was supplied by Novo Nordisk A/S (Baegsvard, Denmark) now
known as Novozymes A/S. A host strain of Aspergillus
oryzae, IFO 4177, obtained by Novozymes A/S from the
Institute of Fermentation (Osaka, Japan) was used to
create the transformant. The donor strain was Aspergillus niger BO-1, isolated by Novozyme. The base
sequence of the functional part of the insert is described
in Boel et al. (1984). The function of the genetic modication was to express A. niger AMG, which primarily
catalyzes the hydrolysis of a(1-4) glycolysidic linkages in
polysaccharides such as starch.
Fermentations were conducted in a 6-L bioreactor
(LSL Biolatte, Luton, United Kingdom) with a working volume of 5.2 L, vessel diameter T 150 mm, impeller
diameter D 75 mm. Agitation was provided by two
Rushton turbines with D/T = 0.5. The lower impeller
was 0.39 T from the bottom of the vessel, and the impeller spacing was adjusted to equal D, with aeration at 1
vvm through a pipe sparger. The broth temperature was
maintained at 32 0.2C, and the pH at 5.0 0.05 by
using 10% w/v H3PO4 and 10% w/v NaOH solutions. It
was important to maintain good pH control because the
activities of the expressed enzymes are known to be pH
sensitive (Carlsen et al., 1996). Thus, the broth pH was
conrmed by two independent in situ pH probes. The
readings of the probes never diered by more than 0.05
pH units. Feed, acid, base, and antifoam additions were
via a branched tting terminating in a single tube in the
zone of high-energy dissipation close to the lower impeller. Dissolved oxygen tension (DOT) was measured
by a polarographic probe (Ingold Messtechnik AG,
Urdorf, Switzerland) and, where possible, was controlled
at 50 2% of air saturation by initially blending nitrogen with air, and later oxygen with air. Thus, the
agitation speed could be varied over a wide range of
conditions without aecting the dissolved oxygen concentration in the broth. The agitation speeds (525, 675,
and 825 rpm) were chosen to give specic gassed power
inputs in the range of 15 kWm)3 and are shown in Table
I. The gassed power input at these speeds was measured
separately in water at 1.0 vvm by using a frictionless air
bearing dynamometer (Nienow and Miles, 1969). The
rheological characteristics of the broth was measured
only at the end of the fermentation, in part because of the
requirement of a large volume of broth for the measurement. A Rheomat 30 (Contraves, Zurich, Switzerland) tted with a six-blade Rushton turbine (53 mm in
diameter) was submerged in 400 mL of broth contained in
a 600-mL (88-mm diameter) glass beaker at 24 0.2C.
Only results from the laminar ow regimen were used.
The composition of the inlet and exit gases from the
bioreactor was measured by using a VG M8-80 mass
spectrometer (VG Gas Analysis, Middlewich, United

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002

Table I. Fermentation conditions for fed-batch cultures.

Fermentation

Final liquid volume (L)

Agitation speed (rpm)

P/V (W/kg)

(P/kD3)(1/tc)(kWm)3s)1)

Maltodextrin feed rate

0.50g/L/h from t = 16 h
1.0 g/L/h from t = 19 h
0.50g/L/h from t = 16 h
1.0 g/L/h from t = 19 h
0.50g/L/h from t = 16 h
1.0 g/L/h from t = 19 h

FB1

5.2

825

5.0

340

FB2

5.2

675

2.5

140

FB3

5.2

525

1.0

40

Kingdom). A SETCIM (AspenTech, Boston, MA)

process management system was used for monitoring
and control.
Medium
The medium used was as follows (in g/L): maltodextrin,
12.0; citric acid, 2.0; MgSO4 7H2O, 2.0; KH2PO4, 2.0;
K2SO4, 2.0; NH4SO4, 3.0; CaCl2, 0.8; yeast extract, 5.0;
trace metal solution (citric acid, 3.0; ZnSO4, 0.29;
Fe2SO4, 0.28; CuSO4, 0.25; MnSO4, 0.25; NiCl, 0.05) 0.5
mL/L; Pluronic P6100, 1.0 mL/L. To avoid catabolite
repression of a-amylase and b-glucosidase production
(Harvey and McNiel, 1994), maltodextrin, which has a
low content of mono- and disaccharides, was used as a
limiting carbon source to control the steady-state biomass concentration. Maltodextrin is a starch hydrosylate containing approximately 0.5% glucose and 99.5%
oligo- and polysaccharides. The AMG produced by the
mycelia degrades the latter to glucose, maltose, and
smaller oligosaccharides. The feed solution (1500 mL)
contained maltodextrin, 360 g; ammonium sulphate, 112
g; citric acid, 1.2 g; and Pluronic P6100, 10 mL. The feed
rate and time of feed initiation is shown in Table I.
Samples were taken periodically for measurements of
biomass concentration, hyphal morphology, and AMG
activity. Mycelial morphology using image analysis and
biomass concentration were measured as described
earlier (Amanullah et al., 1999). Standard errors for
biomass measurements did not exceed 3%. The mean
projected area of all hyphal elements was taken as a
measure of the total biomass (Packer and Thomas,
1990). Hyphal tip activity was measured by using uorescent probe coupled to a semiautomatic image analysis
method, which is briey described here.
Staining Protocol
A solution of Calcouor White (uorescent brightner
no. 28 from Sigma Chemical Company, St. Louis, MD)
was prepared by dissolving 0.1 g in 1 L of distilled water.
Broth samples were diluted to approximately 1 g/L by
using 0.9% NaCl, and 1 mL of the diluted broth was
mixed with 4 mL of Calcouor White solution. Twenty
microliters of the incubated sample (10 min) was placed
on a slide, with a coverslip positioned centrally. The
sample was sealed with nail varnish to reduce evapora-

tion during the course of the analysis. Growing tips also

referred to in the following as ``active tips'' appeared
intensely bright (Fig. 1a). This brightness only extended
to typically 68 lm from the apex of the tips. In addition,
septation was clearly visible. Nonextending tips (``inactive tips'') did not uoresce (Fig. 1b) nor did broken ends
of hyphae (Fig. 1c). A semiautomatic image analysis
method using uorescence microscopy was used to determine the fraction of active (growing) mycelial tips.

Image Analysis Method

A Quantimet 570 image analyzer (Leica, Cambridge
Ltd., United Kingdom) connected to a Polyvar optical
microscope (Reichert Jung Werke AG, Austria) was
used. A uorescent lamp was used in combination with
a Leica UV1 lter block containing an excitation lter
(maximum excitation at 350 nm) and a Leica barrier
lter with a cuto wavelength of 418 nm. A magnication of 120 was used, allowing both sucient visual
detail as well as a relatively large eld of view.
An image analysis program was developed for the
manual selection and automatic analysis of uorescing
tips. The image analysis program allowed the calculation of the mean grayness level that was used for determining whether tips were active (extending) or not.
Program initialization was performed after setting the
slide on the microscope. This program required the
correction of camera controls to produce a suitably
contrasted image. Acquisition of a typical image from
the microscope, multiple image capture, and averaging
(to reduce variation in illumination intensity) was then
performed. This allowed for binary mask detection
levels to be set, as well as the calibration for pixel size.
These parameters were maintained constant from one
sample analysis to another. The rst step involved a gray
processing of the image, which was a simple delineation
of the image to sharpen the edges of the observed hyphae. Archiving of the image was then performed, with
each image given a by numerically incremented novel le
name. Tips were cut from the acquired image of the
hyphae by using a mouse, and the bulk of the hyphae
was removed. Measurements were then performed on
the binary mask of the tips and the underlying grayness
values of the isolated tips. At this stage, the grayness
values of both active and nonactive tips could be
measured, allowing the setting of a user-dened cuto

AMANULLAH ET AL.: MORPHOLOGY AND RECOMBINANT PROTEIN PRODUCTION IN ASPERGILLUS ORYZAE

817

(threshold) value of tip grayness representing active tips.

Although this cuto value was xed for each sample
analysis, allowing a consistent method to distinguish
between active and nonfactive tips, it could nevertheless,
be veried by the user for consistency. A method based
on the measurement of grayness level was needed to
discriminate active and nonactive tips from one another.
The grayness level threshold was achieved by measurement of the grayness level of previously chilled mycelia in ice, followed by Calcouor White staining.
Figure 1a shows an example of bright uorescing tips
representing actively extending tips obtained by staining
mycelia with Calcouor White at room temperature.
The same sample when rst chilled to render the mycelia
metabolically inactive and then stained did not show any
bright tips. The grayness level of the chilled mycelial tips
never exceeded 125 (a grayness level of 0 represented a
black image, whereas a grayness level of 256 represented
a white image). Allowing for a small margin of error, a
threshold value below 130 corresponded to metabolically inactive tips. Grayness levels of 400 hyphal tips were
measured for each sample taken from the fermentation,
and the percentage of active tips was determined. It was
found that the percentage of uorescing (or active) tips
did not change signicantly after a count of 300 tips.
Freely dispersed mycelial tips as well as those on the
outer edges of clumps (and in some instances those on
the inside of loose clumps) were manually cut from the
images of the mycelia by using a mouse. The highest
magnitude of the grayness level corresponded to the
brightest uorescing tips. In the case of uorescing tips,
care was also taken to ensure that the tips were cut
where the region of uorescence terminated. For inactive (nonextending) tips, the distance from the tip apex
to the cuto point was maintained approximately constant and equivalent to that for active tips.
AMG Activity
The assay activity of AMG was based on a manual
method reported in (Amanullah et al., 1999). This
method is based on the spectrophotometeric analysis of
p-nitrophenol (yellow solution under alkaline conditions) formed by the reaction of amyloglucosidase with
the colorless p-nitrophenyl-alpha- glycoside, pNPG
(Merck, Darmstadt, Germany). Samples were appropriately diluted, and 1 mL was used to react with 2 mL
of 0.1% pNPG at pH 4.34.4 (using 1M acetate buer).
After incubation for 20 min at 30C, the reaction was
stopped by using 3 mL 0.1 M borax solution. The ab-

b
Figure 1. Examples of tips stained with Calcouor White showing (a)7
growing tips (b) newly emerging growing tip and an inactive tip. Images shown are at 400 magnication and the width of the hyphae is
approximately 2.3 0.1 lm. (c) Image of fragmented hyphae (image
shown is at 120 magnication and the width of the hyphae is approximately 2.1 0.1 lm).

818

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002

sorbance of the resulting yellow solution was measured

immediately at 400 nm. The contribution of both pNPG
and borax was subtracted as blanks from this measurement. The activity of this enzyme is expressed in an
arbitrary unit, AGU. This unit is dened by Novozyme
A/S as the amount of enzyme that hydrolyzes l lmol of
maltose per minute at 30C. The analysis was calibrated
by using an AMG standard from Novozyme A/S. A
linear correlation (r2 > 0.99) was established between the
absorbance at 400 nm and the AMG concentration, in
the range of 0.0051.696 AGU/mL. Analysis of repeated
samples did not show variations in excess of 0.03
AGU/mL. The AMG titer was expressed as AGU/L.
RESULTS AND DISCUSSION
Fermentation Reproductivity
The results of two fermentations at an agitation speed of
825 rpm are shown in Figure 2 by way of an example of
fermentation reproducibility. In these initial experiments, the maltodextroin feed rate used was higher than
those listed in Table I and was also gradually increased
with fermentation time (1619 h: 0.75 g/L/h; 1924 h

1.10 g/L/h and 24 h onward: 1.50 g/L/h). The rationale

for the graduated incremental feed rates was to prevent
the formation of excessively large clumps during the
rapid growth phase, which could lead to nutrient and
oxygen depletion within the core of the clumps. This
could be prevented by selecting lower feed rates during
the rapid growth phase, thus limiting the size of the
clumps. Higher feed rates were used after 19 h following
clump fragmentation as a result of nutrient limitation,
which subsequently allowed improved mixing and mass
transfer. Figure 2 compares the carbon dioxide production rates (CPR) (Fig. 2a), the biomass concentration and specic growth rate (Fig. 2b) and the AMG
titer and active tips (Fig. 2c). Figure 2a is only shown for
the rst 45 h because dissolved oxygen (DOT) became
limiting (DOT <40%) after that time and are overall
very similar as they are in Figure 2b for the whole of the
120 h of the fermentation. The biomass concentration
increased to a maximum of 58 g/L and resulted in an
extremely high apparent viscosity, which in turn led to
the inability to adequately control the DOT. Similarly,
Figure 2c also shows that the measurements of AMG
titer and active tips were very reproducible, but the
AMG titer did not increase signicantly above 7000
AGU/L beyond 40 h. The reason for this is believed to
be due to the fall in DOT; further support for this reasoning is given later in this article.
These results clearly showed the reproducibility of the
fermentation results but also highlighted the need to
reduce the maltodextrin feed rates to conduct fermentations where the eects of agitation could be separated
from dissolved oxygen and such considerations form the
basis of the experiments described in Table I.
General

Figure 2. Reproducibility of two fermentations at 825 rpm at high

maltodextrin feed rates (1619 h: 0.75 g/L/h; 1924 h: 1.10 g/L/h; 24 h
onward: 1.50 g/L/h): (a) carbon dioxide production rate, (b) biomass
and specic growth rate, and (c) active tips and AMG titer.

As already pointed out, if the eect of mechanical stress

on culture performance is to be analyzed, it is necessary
to decouple that aspect of agitation from other agitation-related parameters. Hence, the gas blending is used
to enable the level of DOT to be controlled independently of agitation intensity. However, in addition, it is
necessary to ensure broth homogeneity so that, for example, nutrient concentrations, DOT, and pH levels are
spatially the same everywhere. In that case, any change
in culture performance cannot be attributed to the biological species passing through zones of dierent concentration as found, for example, on the large scale
(Oosterhuis and Kossen, 1984; Bylund et al., 1998;
Langheinrich and Nienow, 1999). At the bench scale,
homogeneity is usually achieved because the ow is fully
turbulent, and mixing times are very short [< 10 s
(Nienow, 1998)]. However, in these mycelial fermentations, the broth becomes increasingly viscous and shear
thinning with time, this change is mainly correlated with
increasing biomass (Riley et al., 2000). In this study,
broth rheology was not a parameter of particular in-

AMANULLAH ET AL.: MORPHOLOGY AND RECOMBINANT PROTEIN PRODUCTION IN ASPERGILLUS ORYZAE

819

terest, but it was measured at the end of the fermentation, which gave the highest biomass concentration
(FBI; see below for details). For this broth, the ow
behavior index, n, and consistency index, K, were 0.18
and 2.4 Nm)2 sn, respectively.
Because, in the fermentations FB2 and FB3, the
biomass concentration was slightly lower, it is reasonable to assume that the broth was slightly less viscous
(Riley et al., 2000). However, even assuming that it had
the same rheological properties, then even at the lowest
speed of 525 rpm, the Reynolds number [calculated by
using the method of Metzner and Otto (1957)] would
always have been >900*). For such Reynolds numbers
covering the high transitional to fully turbulent region,
the mixing time at the bench scale is short and essentially independent of viscosity (Nienow, 1998). Here,
even at the lowest speed of 525 rpm (FB3), both the pH
probes indicated a pH dierence <0.05 units throughout the fermentation (Fig. 3c), suggesting that indeed
good mixing was maintained. After about 65 h, the RQ,
as discussed below, became less stable (i.e., deviated
further from 1, and the DOT could not be maintained at
50% even with gas blending because of insucient oxygen transfer and fell steadily to 30%.
Fermentation Parameters
Values of DOT, carbon dioxide production rate (CPR),
and pH for the three runs are shown together in Fig. 3a,
b, and c. At 825 rpm (FB1; P/V 5.0 W/kg), CPR peaked
at 46 mmol/L/h at the end of the batch phase and remained relatively constant throughout most of the fedbatch phase of the fermentation. The respiratory quotient (RQ) remained at approximately 1 throughout the
run (data not shown). The proles of DOT, CPR, pH,
and RQ at 675 rpm (FB2; P/V = 2.5 W/kg) were similar to those for FB1, except in two respects. First, the
peak in CPR was lower at 35 mmol/L/h and second,
although on average RQ was still 1, the noise in the data
was a little larger. These trends were also continued at
525 rpm (FB3; P/V = 1.0 W/kg), where the peak in
CPR was still lower at 29 mmol/L/h. At this latter speed
after about 65 h, the RQ, as discussed below, became
less stable (i.e., deviated further from 1), perhaps indicating some local variations in dissolved oxygen due to
inadequate mass transfer, although the gas blender
control dynamics also contributed to this noise. In addition, the DOT could not be maintained at 50% because of insucient oxygen transfer and fell steadily to
30%. However, because until at least 65 h, all the three
fermentations were conducted by using identical procedures and the DOT was maintained at 50% with a high
degree of homogeneity, it is reasonable to assume that
any changes in culture performance up until this time
* Because on the industrial scale, Reynolds numbers are >1000 even
in high biomass concentration broths (Nienow, 1998), this choice of
Reynolds number range seems apporiate for this study.

820

Figure 3. Proles of (a) dissolved oxygen tension (DOT), (b) carbon

dioxide production rate (CPR), and (c) pH in fed-batch fermentations
at 825, 675, and 525 rpm (details of experiments are shown in Table I).

were a direct consequence of the dierent agitation

conditions on biomass morphology alone.
Biomass and Specic Growth Rate
Figure 4a shows the biomass concentration as a function
of time for each of the three fermentations. The highest
biomass concentration at the end of the batch phase was
obtained at the highest agitation speed. The rather limited number of dry cell weight measurements in the batch
phase did not allow for accurate calculations of the
specic growth rate. Given the low biomass concentrations and high concentration of carbon source during the
rst 12 h of fermentation, it is reasonable to assume that
the respiratory machinery operated at their maximum
value, suggesting that the measured CPR value was a
linear function of the biomass concentration. Consequently, for a limited time of the batch phase, changes in
the specic growth rate can be estimated from the slope
of a semilogarithmic plot showing the CPR values versus
time (Fig. 4d). Based on this plot it was not possible to
detect major changes in the specic growth rate as a
function of the agitation intensity. For all three fermentations, a distinct lowering of the specic growth rate
was measured between 8 and 10 h of fermentation.

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002

sity in the batch phase of the fermentation was not due

to dierent limitations, because the CPR values peaked
at approximately the same time. One possibility for the
results is that the clumps were oxygen limited in the
batch phase, even though the DOT was maintained at
50% with a high degree of culture homogeneity. However, conservative calculations using pellet diusion
models (Metz and Kossen, 1977) show that even pellets
with core diameters of <140 lm would not be oxygen
limited. It is, therefore, unlikely that intra-particle diffusional limitations will be signicant in clumps, especially because the morphological measurements
indicated that the clumps had loose structures with
fullness ratios (Tucker et al., 1992) generally <0.50 (a
fullness ratio of 1 reects a structure of a pellet core).
This has also been conrmed by direct measurements
using microprobes (Wittler et al., 1986; Cronenburg3
et al., 1994). Therefore, it is unlikely that local DOT
limitations existed within the microenvironment of
clumps in the batch phase of the fermentations.
Another explanation for the dierences in biomass
concentration with agitation intensity is that maltodextrin needs to be enzymatically degraded by AMG to
smaller sugars before it can be used by the mycelia and
may, therefore, be related to the availability of secreted
AMG. Table II shows the AMG liters measured at the
end of the batch phase of the fermentations. The AMG
titres were 370, 230, and 110 AGU/L at 825, 675, and
525 respectively. Although the AMG titers were low
because of the repressive eects of high maltodextrin
levels as discussed below, there was clearly a dependence
on agitation intensity. This was probably due to a larger
number of active tips at the higher agitation speeds.
Although the percentage of active tips was approximately constant for all three runs (Fig. 8), the total
number of active tips was higher at higher agitation
speeds as a result of increased mycelial fragmentation.
Therefore, this explanation suggests that mycelial fragmentation and AMG secretion was the greatest at the
highest agitation speed of 825 rpm, which in turn resulted in greater availability of smaller sugars and,
hence, higher growth rates and biomass concentrations.
Because the initial concentration of maltodextrin was
the same in each fermentation, the explanation discussed
above would result in a greater residual concentration of
maltodextrin at the lowest agitation speed at the end of
the batch phase. The second CPR peak between 16 and
22 h for the fermentation at 525 rpm (Fig. 3b) may
Figure 4. Biomass (a), AMG titer (b), and specic AMG activity (c)
versus time in fed-batch fermentations at 825, 675, and 525 rp. (d)
Semi-logarithmic plot of CPR versus time in fed-batch fermentations
at 825, 675, and 525 rpm.

Despite the lack of distinct dierences in the specic

growth rate, the peak CPR values shown in Figure 3b
reected the dierences in biomass concentration. The
increase in biomass concentration with agitation inten-

Table II. AMG titers at the end of the batch phase for fermentations
at 825, 675, and 525 rpm.
Fermentation

AMG titer at the end of the

batch phase (AGU/L)

FB1 (825 rpm)

FB2 (675 rpm)
FB3 (525 rpm)

370 30 (t =11 h)
230 30 (t = 11.5 h)
110 30 (t = 11 h)

AMANULLAH ET AL.: MORPHOLOGY AND RECOMBINANT PROTEIN PRODUCTION IN ASPERGILLUS ORYZAE

821

Figure 5. Mean projected area of mycelia in fed-batch fermentations

at 825, 675, and 525 rpm.

indicate use of such residual maltodextrin, especially

because this coincided with signicant mycelial fragmentation (see Fig. 5). Table III shows the integrated
CPR values for the three fermentations as a function of
time, with the values for the 675 and 525 rpm runs
normalized to the corresponding time values for the 825
rpm run. The calculations at the end of the batch phase
show that the integrated CPR values were only 49 and
78% in the 525 and 675 rpm runs compared to that at
825 rpm. However, these values increased to 89 and 91%
by 20 h, and the dierences of the integrated CPR values
between the three runs decreased with increasing time
(<5% at 60 h). It appears that the rate of maltodextrin
use was delayed with decreasing agitation intensity. A
carbon mass balance was not conducted in this study
because of the diculty of making precise measurements
of maltodextrin concentration (because AMG readily
degrades maltodextrin to smaller sugars). However, in a
related study on P. chrysogenum (Justen et al., 1998),
where very similar eects of agitation intensity on biomass concentration and carbon dioxide production rate
were reported, a carbon balance showed that typically
only 5% of the carbon remained unaccounted.
AMG Production
Figure 4b also shows that very little AMG was produced
in the batch phase of the fermentations, because of repressive eects of easily metabolizable sugars. Such a

result has also been reported previously (Carlsen, 1994;

Morkeberg et al., 1995; Sphor et al., 1998). The AMG
titer from the fermentation at 825 rpm (FB1) increased
steadily with time, eventually reaching an activity of
11,400 AGU/L. Although the AMG litre at 675 rpm was
slightly higher than that at 825 rpm between 20 and 60
h, the activities produced in both runs were, overall, very
similar until 100 h. After that, the volumetric AMG
activity at 675 rpm leveled o for the last 20 h. The
AMG titer at the lowest agitation speed of 525 rpm was
similar to the two higher speed runs until 60 h, whereas
as noted above, the DOT probe actually measured a
value of 50% until 65 h after which it fell (Fig. 3a). A
tentative conclusion from the results is that at high
biomass concentrations, a condition was reached at
which initially the speed was not high enough to prevent
some local loss of DOT essentially (e.g., during the last
20 h at 675 rpm and between about 6065 hours at 525
rpm). Subsequently, at 525 rpm, the fall in DOT was
recorded by the probe (Fig. 3a). These changes suggest
that AMG production is strongly sensitive to DOT even
though biomass growth (Fig. 4a) is not. A similar result
was reported by Wongwicharn et al. (1999), who found
higher levels of both heterologous (hen egg white lysozyme) and native (glucoamylase) enzymes with increasing dissolved oxygen levels in chemostat cultures of
Aspergillus niger, whereas biomass concentration remained unaected.
Chemostat studies at constant agitation speed have
shown that the specic a-amylase production by another
A. oryzae strain increased linearly with specic growth
rate of the culture (Agger et al., 1998). Here, it is shown
that specic enzyme productivity also depends on agitation speed. The specic AMG productivity (AGU/g
biomass/h) for the three runs (Fig. 4c) were calculated
by rst, tting fourth-order polynomial functions
(curves shown in Fig. 4b) to the AMG titer data points.
These polynomials were then dierentiated and divided
by the biomass concentration. The specic AMG productivity at 825 rpm (Fig. 4c) was signicantly lower
compared to those in the 675 and 525 rpm runs. The
specic AMG productivity at 525 rpm was slightly
higher than at 675 rpm until 45 h. Beyond this time, it
decreased at 525 rpm probably because of DOT limitations as discussed earlier. Although the AMG titer
remained independent of changes in agitation speed in
the absence of DOT limitation, the specic AMG pro-

Table III. Integrated CPR values for the three fermentations shown as a function of time.
Fermentation

End of batch phase

(peak CPR)

20 h

40 h

60 h

80 h

FB1 (825 rpm)

FB2 (675 rpm)
FB3 (525 rpm)

100
78
49

100
89
91

100
95
94

100
96
96

100
97
98

The values for the 675 and 525 rpm runs have been normalized to the corresponding time values for the 825 rpm run (end of batch phase: 168.4
mmol/L; t = 20 h: 289.1 mmol/L; t = 40 h: 541.6 mmol/L; t = 60 h: 785.0 mmol/L; t = 80 h: 1047.3 mmol/L).

822

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002

ductivity did not, with more ecient protein secretion at

lower agitation speeds during the rst 45 h.
The proles of the mean projected area representing
the whole of the biomass for each agitation speed are
shown in Figure 5. The mean projected area increased
rapidly during the exponential phase with the maximum
values being found after 16 h. The values were strongly
inuenced by the agitation intensity, being signicantly
lower at the highest speed. This increase was followed by
a rapid decline until after about 3050 h, a constant
value with time was reached, which again was lower at
higher speeds. The choice of 16 h as the time to initiate
fed-batch operation was in part based on the fact that
signicant mycelial fragmentation was starting, allowing
improved mixing and mass transfer. Typical images reecting these changes in morphology as a function of
time for FB3 (N = 525 rpm) are shown in Figure 6. The
percentage of clumps (expressed as a fraction of the total
projected area) also changed with time (data not
shown), with a maximum value of 80% at 16 h decreasing to 30% at the end of the fermentation. As with
the chemostat experiment (Amanullah et al., 1999),
morphological analysis conrmed that the high percentage of mycelial in the samples warranted the inclusion of clumps for full characterization of the biomass.
The rapid increase in mean projected area during the
exponential phase can be explained by considering that
in this high specic growth rate phase, mycelial growth
dominated over mycelial fragmentation (Amanullah
et al., 1999). Therefore, the highest mean projected area
was obtained for the lowest fragmentation conditions.
The rapid decrease in mean projected area after the batch
phase would then be due to the dominance of fragmentation over growth. It is probable that fragmentation was
determined primarily by the agitation intensity, although
changes in mechanical strength of mycelia were almost
certainly of importance. After 30 h, low specic growth
rates (l < 0.02 h)1) after about 30 h), growth and
fragmentation processes appeared to have reached a
balance that was dependend on the agitation intensity.
This balance of growth and fragmentation and their
impact on mycelial morphology was recently accounted
for successfully by the energy dissipation/circulation, (P/
kD3)(1/tc), or EDC function (Justen et al., 1996; 1998;
Amanullah et al. 1999; 2000). This function can be
considered as the specic energy dissipation rate in the
impeller swept volume, P/kD3, where P is the power
input and D the diameter of the impeller and the value
of the factor k depends on the impeller used. The term 1/
tc also allowed for the dierent impeller ow numbers
and, hence, the frequency of mycelial circulation
through the impeller swept volume. In particular, the
EDC function was applied to fragmentation of nongrowing cultures of A. oryzae and P. chrysogenum
(Amanullah et al., 2000), fed-batch fermentations of P.
chrysogenun (Justen et al., 1996; 1998) and to chemostat
cultures of Aspergillus oryzae (Amanullah et al., 1999).

Figure 6. Changes in morphology (40) as a function of time in FB38

(N = 525 rpm) Width of image = 1350 lM.

Here, the EDC function was used to correlate the

mean projected area at specic growth rates of 0.05 h)1
(biomass concentrations between 9 and 12 g/L) and 0.02
h)1 (biomass concentrations between 18.2 and 22.4 g/L).
EDC values were calculated by using the gassed power
input for each impeller assuming that the total power
input could be divided equally by the two impellers in
use (Nienow and Lilly, 1977). The slope of the mean
projected area versus EDC was approximately )0.50
(with r2 0.93) both for l = 0.05 h)1 and 0.02 h)1 (Fig.
7). It is important when such correlations are developed
that the morphological data are obtained at similar
specic growth rates and not fermentation times (where
specic growth rates may vary). This approach ensures
that errors in the fragmentation analysis are not introduced because of dierences in growth behavior. It is
interesting to note that with A. oryzae (Amanullah et al.,
1999), the slope of a similar plot for chemostat cultures

AMANULLAH ET AL.: MORPHOLOGY AND RECOMBINANT PROTEIN PRODUCTION IN ASPERGILLUS ORYZAE

823

Figure 7. Dependence of mean projected area on the energy dissipation/circulation function at specic growth rates of 0.05 h)1 (biomass concentrations between 9 and 12 g/L) and 0.02 h)1 (biomass
concentrations between 18.2 and 22.4 g/L).

at a dilution rate of 0.05 h)1, where the biomass concentration was 2 g/L, was also approximately the same
()0.55). Therefore, the EDC function appears to be able
to correlate mycelial fragmentation, at a given specic
growth rate, at very dierent biomass concentrations.
Measurements of hyphal tip activity (number of active or growing tips expressed as a percentage of the
total number of measurable tips) generally showed a
maximum value of approximately 80% towards the end
of the exponential phase, decreasing to a constant value
that depended on the agitation speed (Fig. 8). The
highest value of percentage active tips was obtained at
the lowest agitation speed and, therefore EDC. However, one inherent limitation of the image analysis
technique used for the measurement of hyphal tip activity is that tips within the three-dimensional structures
of clumps cannot be observed and thus taken into
consideration. Generally, most workers (e.g., Nielsen
et al., 1995) have assumed that the tips on the freely
dispersed entities also represent those associated with

Figure 8. Hyphal tip activity in fed-batch fermentations at 825, 675,

and 525 rpm.

824

clumps. Here, tips on both freely dispersed entities and

the edge of clumps (wherever they were observed) were
also considered. In both cases, the assumption is that
these data apply to all the biomass, even though a certain proportion of it within the clumps is not accessible
to the analysis.
Although there did not appear to be a direct relationship between hyphal tip activity (Fig. 8) and AMG
titer (Fig. 4b), a plot of hyphal tip activity versus specic
AMG productivity (Fig. 9) for the rst 60 h of the fermentations (i.e., only considering the results that were
not aected by DOT limitations) shows that there appears to be a relationship between these variables. Despite the scatter in the data (a linear regression yielded
an r2 value of 0.65), the relationship may be acceptable
given the diculty in making the measurements of active
tips and estimating the specic AMG productivity. Peberdy (1994) proposed that the secretory capacity of
lamentous fungi may be increased with increasing
numbers of active tips per unit biomass. This hypothesis
assumes rst that the total area of the extending tip is
also increased when the number of tips per unit biomass
increases, which may not be the case if the tip extension
rates of the hyphae are reduced without a substantial
increase in hyphal diameter, and second, that protein
secretion is the rate limiting factor.
Overall, the idea that protein secretion occurs at hyphal tips (Mackenzie et al., 1993; Wessels, 1990; 1993;
Peberdy, 1994) and therefore, depends on the number of
active tips appears to be supported by this study. This is
despite the fact that the measurement of tips within
clumps is not possible even with use of state-of-the-art
image analysis methods. Therefore, this analysis does
not account for the physiological state of hyphal tips
within clumps, which may potentially bias the conclusion of this result. Such problems also render it dicult
to correlate the numbers of active tips per unit biomass
and specic AMG productivity.

Figure 9. Hyphal tip activity versus specic AMG productivity in

fed-batch fermentations at 825, 675, and 525 rpm.

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002

As with continuous cultures (Amanullah et al., 1999),

the results of the present study show a dependence of
mycelial fragmentation on agitation intensity, and results from the fed-batch phase of the fermentations indicate that mycelial morphology does not inuence
AMG titer with this A. oryzae strain. We did, however,
nd a dependence of protein secretion on agitation intensity in the batch phase of the fermentations, which
appeared to strongly aect the growth characteristics of
the culture. Bocking et al. (1999) using A. oryzae strains
of dierent morphologies derived by mutagenesis reported that mycelial morphology did not inuence
protein secretion in either continuous, batch, or fedbatch cultures, but given the dierences in strains used
in the two studies, it is dicult to draw direct comparisons. The results presented in this study are also in
contrast to those obtained with fed-batch P. chrysogenum
where agitation intensity was found to strongly inuence
penicillin production (Justen et al., 1996; 1998), There, it
was suggested that this interrelationship is due to
breakage of the relatively weaker vacuolated regions of
hyphae (i.e., regions where penicillin synthesis is located).
In contrast, although breakage still occurs at the weaker
vacuolated hyphal regions in A. oryzae (data not shown),
it does not aect AMG titer under conditions of substrate
limited growth. It is postulated that this lack of a relationship is due to the fact that protein secretion in fungi
appears to occur only at the tips (Wessels, 1990; 1993).
Therefore, the results found in this study are not generally applicable to all mycelial fermentations, and each
system of interest should be judged on its own merits.
Operation at Higher Biomass Concentration
Attempts were also made to repeat this study with
higher biomass concentrations of up to 58 g/L when
mycelial interactions should be pronounced. These
biomass concentrations were achieved, but it became
impossible to mix the broth adequately because of its
highly shear thinning, viscous nature. This led to a much
lower Reynolds number at the small scale and very long
mixing times (Nienow, 1998), and possibly even complete stagnation due to yield stresses (Nienow, 1998). A
similar biomass and broth rheology on the large scale at
equal specic power input would give relatively higher
Reynolds numbers and better mixing. The problem of
maintaining suitable scale-down conditions with respect
to both uid dynamics and biomass concentration is a
particularly dicult one. Further work is still needed to
resolve this problem.
CONCLUSIONS
The eects of agitation intensity on growth, mycelial
morphology, hyphal tip activity, and recombinant protein (amyloglucosidase) production in fed-batch cultures

of Aspergillus oryzae were investigated. Dissolved oxygen was controlled at 50% of air saturation by using gas
blending, and suciently high agitation speeds were
used to give good spatial homogeneity for at least 60 h.
These operating conditions also gave realistic parameters in terms of biomass concentration as well as specic
power inputs (1 to 5 kWm)3). The ``energy dissipation/
circulation'' function could be successfully used to correlate mycelial fragmentation at similar specic growth
rates but very dierent biomass concentrations. In the
batch phase of the fermentations, biomass concentration
increased with increasing agitation intensity and appeared to be related to higher AMG secretion. The total
AMG titer remained independent of changes in EDC as
long as the broth was well mixed with DOT controlled at
over 50%, even though the specic AMG productivity
was found to be dependent. In addition, we found a trend
suggesting a dependency of specic AMG productivity
on hyphal tip activity. Overall, as with chemostat cultures of A. oryzae (Amanullah et al. 1999), agitation intensity signicantly aected mycelial morphology, but
these changes did not inuence enzyme titers.
Presuming secretion to be dependent on the number
of actively growing hyphal tips, these results also suggest
that protein secretion may be a bottleneck in this strain
in fed-batch fermentations, even though mesurement of
hyphal tips within clumps was not considered.
In the latter stages of the fed-batch fermentation
(after 65 h), the DOT fell below 50% even with gas
blending at 525 rpm due to the increasing viscosity of
the broth. Beyond this time, no further AMG was
produced. At 675 rpm, beyond 100 h, AMG production virtually ceased. Again, it is proposed that this
arises because of loss of DOT control due to increasing
viscosity associated with increased biomass concentration. Preliminary experiments at a higher maltodextrin
feed rate also showed that when DOT fell below 40%,
AMG production ceased.
The practical and important implication of this study
is that the agitation intensity in this fungal fermentation
must be manipulated to meet process requirements in
terms of dissolved oxygen levels and bulk mixing, and
possibly to control broth rheology by changing mycelial
morphology. Such a strategy will ensure that though
morphology may change, recombinant protein production will not be compromised under conditions of
substrate limited growth. In the batch phase of the fermentation, such changes can lead to profound changes
in the growth characteristics of the culture with maltodextrin as the carbon source. It is also important to
point out that this is not a general conclusion for all
mycelial cultures but is specically applicable to this
strain. Therefore, such studies should be conducted with
the strain of interest.
The nancial support of the Biotechnology and Biological
Sciences Research Council, United Kingdom, and the En-

AMANULLAH ET AL.: MORPHOLOGY AND RECOMBINANT PROTEIN PRODUCTION IN ASPERGILLUS ORYZAE

825

zyme division of the former Novo Nordisk A/S, and now

known as Novozymes A/S, Denmark, is gratefully acknowledged.

References
Agger T, Spohr AB, Carlsen M, Nielse J. 1998. Growth and product
formation of Aspergillus oryzae during submerged cultivations:
verication of a morphologically structured model using uorescent probes. Biotechnol Bioeng 57:321329.
Amanullah A, Blair R, Nienow AW, Thomas CR. 1999. Eects of
agitation intensity on mycelial morphology and protein production in chemostat cultures of recombinant Aspergillus oryzae.
Biotechnol Bioeng 62:434446.
Amanullah A, Justen P, Davies A, Paul GC, Nienow AW, Thomas
CR. 2000. Agitation induced mycelial fragmentation of Aspergillus
oryzae and Penicillium chrysogenum. Biochem Eng J5:109114.
Ayazi Shamlou P, Makagiansam HY, Ison AP, Lilly MD. 1994.
Turbulent breakage of lamentous micro-organisms in submerged
culture in mechanically stirred bioreactors. Chem Eng Sci 49:16,
26212631.
Bocking SP, Wiebe MG, Robson GD, Hansen K, Christiansen LH,
Trinci ABJ. 1999. Eect of branch frequency in Aspersillus oryzae
on protein secretion and culture viscosity. Biotechnol Bioeng
65:638648.
Boel E, Hansen MT, Hjort L, Fillo NP. 1984. Two dierent types of
intervening sequences in the glucoamylase gene from Aspergillus
niger. EMBO J3:15811585.
Bylund F, Collet E, Enfors SO, Larsson G. 1998. Substrate gradient
formation in the large-scale bioreactor lowers cell yield and increases by-product formation. Bioproc Eng 18:171180.
Carlsen M. 1994. PhD thesis. Department of Biotechnology, Technical
University of Denmark.
Carlsen M, Nielsen J, Villadsen J. 1996. Kinetic studies of acid-inactivation of alpha-amylase from Aspergillus oryzae. Chem Eng Sci
51:1, 3743.
Cronenburg CCH, Ottengraf SPP, van den Heuvel JC, Pottel F, Sziele
D, Schugerl K, Bellgardt K-H. 1994. Inuence of age and structure
of Pencillum chrysogenum pellets on the internal concentration
4
proles. Bioproc Eng 10:209216.
Dion WM, Carilli A, Sermonti G, Chain EB. 1954. The eect of mechanical agitation on the morphology of Penicillium chrysogenum
Thorn in stirred fermenters. Rend 1st. Super de Sanita 17:187205.
Harvey LM, McNeil B. 1994. Liquid fermentation systems and
product recovery of Aspergillus. In: Smith JE, editor. Aspergillus.
New York; Plenum Press, p 141176.
Justen P, Paul CG, Nienow AW, Thomas CR. 1996. Dependence of
mycelial morphology on impeller type and agitation intensity.
Biotechnol Bioeng 52:672684.
Justen P, Paul GC, Nienow AW, Thomas CR. 1998. Dependence of
Penicillium chrysogenum growth, morphology vacuolation and
productivity in fed-batch fermentations on impeller type and agitation intensity. Biotechnol Bioeng 59:762775.
Konig B, Seewald Ch, Schugerl K. 1981. Process engineering investi5
gations of penicillin production. Eur J Appl Microbiol Biotechnol
12:205211.
Langheinrich C, Nienow AW. 1999. Control of pH in large-scale, free
suspension animal cell bioreactors: alkali addition and pH excursions. Biotechnol Bioeng 66:171179.
Mackenzie DA, Jeenes DJ, Belshaw NJ, Archer DB. 1993. Regulation
of secreted protein production by lamentous fungi: recent development and perspectives. J Gen Microbiol 139:22952307.
Makagiansar HY, Ayazi Shamlou P, Thomas CR, Lilly MD. 1993.
The inuence of mechanical forces on the morphology and penicillin production of Penicillium chrysogenum. Bioproc Eng 9:8390.

826

Markl H, Bronnenmeier R, Wittek B. 1991. The resistance of microorganisms to hydrodynamic stress. Int Chem Eng 31:2:185.
Metz B, Kossen NWF. 1977. The growth of molds in the form of
pelletsa literature review. Biotechnol Bioeng 19:781799.
Metz B, de Bruijn EW, van Suidjamp JC. 1981. Method for quantitative representation of the morphology of molds. Biotechnol
Bioeng 23:149163.
Metzner AB, Otto RE. 1957. Agitation of non-Newtonian uids. AIChE J 3:310.
Morkerberg R, Carlsen M, Nielsen J. 1995. Induction and repression
of alpha-amylase production in batch and continuous cultures of
6
Aspergillus oryzae. Microbiol 141:10, 24492454.
Nielsen J, Johansen OL, Jacobsen M, Krabben P, Villadsen J. 1995.
Pellet formation and fragmentation in submerged cultures of
Penicillium chrysogenum and its relation to penicillin production.
Biotechnol Prog 11:9398.
Nienow AW, Miles D. 1969. A dynamometer for the accurate measurement of mixing torque. J Sci Inst 2:994995.
Nienow AW, Lilly MD. 1979. The power drawn by multiple impellers
in sparged, agitated vessels. Biotechnol Bioeng 21:23412345.
Nienow AW. 1998. Hydrodynamics of stirred bioreactors. Appl Mech
Rev 51:332.
Oosterhuis NMG, Kossen NWF. 1984. Dissolved oxygen concentration proles in a production scale bioreactor. Biotechnol Bioeng
26:546550.
Packer HL, Thomas CR. 1990. Morphological measurements on lamentous micro-organisms by fully automatic image analysis.
Biotechnol Bioeng 35:870881.
Peberdy JF. 1994. Protein secretion in lamentous fungitrying to
understand a highly productive black box. Trends Biotechnol
12:5057.
Reuss M. 1988. Inuence of mechanical stress on the growth of
Rhizopus nigricans in stirred bioreactors. Chem Eng Technol
11:178187.
Riley GL, Tucker KG, Paul GC, Thomas CR. 2000. Eect on fermentation broth rheology of biomass concentration and mycelial
morphology. Biotechnol Bioeng 68:2, 160172.
Smith JJ, Lilly MD, Fox RL. 1990. Morphology and penicillin production of Penicillium chrysogenum. Biotechnol Bioeng 35:1011
1023.
Spohr A, Carlsen M, Nielsen J, Villadsen J. 1998. Alpha-amylase
production in recombinant Aspergillus oryzae during fed-batch
and continuous cultivations. J Ferment Bioeng 86:1, 4956.
Tucker KG, Kelly T, Delgrazia P, Thomas CR. 1992. Fully automatic
measurement of mycelial morphology by image analysis. Biotechnol Prog 8:353359.
Ujcova E, Fencl Z, Musilkova M, Seichert L. 1980. Dependence of
release of nucleotides from fungi on fermenter turbine speed.
Biotechnol Bioeng 22:237241.
van Suidjam JC, Metz B. 1981. Inuence of engineering variables upon
the morphology of lamentous molds. Biotechnol Bioeng 23:111
148.
Vardar F, Lilly MD. 1982. Eect of cycling of dissolved oxygen on
product formation in penicillin fermentations. Eur J Appl Microbiol Biotechnol 14:203211.
Wessels JGH. 1990. Role of cell wall architecture in fungal tip growth
generation. 129. In: Heath IB, editor. Tip growth in plant and
fungal walls. San Diego: Academic Press.
Wessels JGH. 1993. Wall growth, protein excretion morphogenesis in
fungi. New Phytol 123:397413.
Wittler R, Baumgart H, Luebbers DW, Schuegerl K. 1986. Investigations oxygen transfer into Penicillium chrysogenum pellets
by microprobe measurements. Biotechnol Bioeng 28:10241036.
Wongwicharm A, McNiel B, Harvey LM. 1999. Eect of oxygen enrichment on morphology, growth and heterologous protein production in chemostat cultures of Aspergillus niger B1-D.
Biotechnol Bioeng 65:416424.

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 77, NO. 7, MARCH 30, 2002