Conversion of Bioreactors to

Continuous Perfusion Using

Hollow Fiber Cell Separators
Practical Aspects

Spectrum Laboratories, Inc.

United States 800-634-3300

Europe 31-76-5719-419
Japan 81-77-552-7820
China 86-21-68810228

copyright 1989 all rights reserved

Contents

1. Introduction 2
1.1 Available Options for Conversion of a Bioreactor to
Perfusion Operation 3
1.2 Results Obtained Using CellFlo Hollow Fiber Cell Separators for
Converting Bioreactors to Continuous Perfusion 5

2. Basic Concepts of Crossflow Separation 6

3. Converting a Batch Bioreactor to Perfusion Using CellFlo Separators 7

3.1 Bioreactor With Steam Supply for Sterilization 8
3.1.1 The Basic System Startup 8
3.1.2 Component Parts 8
3.1.3 Bioreactor Modifications and Preparations 10
3.1.4 System Setup Procedure 11
3.2 Alternate System Configurations 12
3.2.1 Fewer Valves 12
3.2.2 Dual CellFlo Setup 12
3.2.3 Stainless Tubing Setup 13
3.3 Bioreactor Without Steam Supply for Sterilization 13
3.4 Pump and Tubing Recommendations 14

4. Operating Considerations 15
4.1 Perfusion Process Optimization 15
4.1.1 Daily Perfusion Rate 15
4.1.2 Recirculation Rate 17
4.1.3 Changing of the CellFlo Modules 19
4.1.4 Bioreactor Volume Control 19
4.1.5 Cell Harvesting 19
4.1.6 Serum 20
4.1.7 Antifoams and Surfactants 21
4.1.8 Backflushing and Optional Media Feeding Strategies 22
4.1.9 Sterility Protection 23

5. Sterilization and Sanitization 24

5.1 Sterilization 24
5.1.1 Autoclaving a Perfusion Loop Containing a CellFlo Module 24
5.1.2 Chemical Sanitization 25

6. Integrity Testing 25
6.1. Wetting the CellFlo Module 25
6.1. Pressure Hold Test 27
6.1. Diffusion and Bubble Point Tests 27

7. References 28

1
1. Introduction Historically, most pharmaceutical fermentation products have
been derived from bacteria and fungi grown in stirred tank
fermenters ranging in size up to tens of thousands of liters. With the
advent of recombinant DNA techniques in the mid 1970s and the
emergence of the biotechnology industry in the late 1970s, it was
assumed that production of the new class of pharmaceuticals would be
done in the same fermenters. As the first recombinant products were
being produced in bacteria, it was discovered that many of them
were not biologically active. The lack of activity is related to the
inability of procaryotic cells to post-translationally modify
mammalian derived proteins correctly.
At the same time, monoclonal antibodies were recognized as
having powerful diagnostic and therapeutic potential. The fused
hybridoma cells that produce monoclonal antibodies are also
mammalian derived cells. Thus, a large proportion of the new
generation of biopharmaceuticals for diagnostic and therapeutic
applications is now being produced in mammalian cells (1).
The long history of pharmaceutical production in fermenters and a
lack of bioreactors custom designed for mammalian cells, has led to
interest in converting existing stirred-tank bacterial fermenters into
mammalian cell bioreactors (2-4). Most microbial fermenters are
designed for batch operation. Batch operation of a microbial
fermenter is economical due to high cell densities (>109 cells/ml),
rapid doubling times (-20 min for many bacteria) and to the hardy
nature of the cells, which allows the use of aggressive agitation and
gassing. Typical microbial batch operations last 5-10 days. In this
short period of time, vast quantities of product can be produced.
Mammalian cells, on the other hand, grow to densities 10-1000
times lower than bacterial cells, even under the most optimal
conditions. They have doubling times in the range of 18-36 hours,
much more complex nutritional requirements than microbial cells
and they are far more sensitive to minor perturbations in their
environment. A mammalian cell culture operated in a 5-10 day
batch mode will only reach reasonable densities near the end of the
run, so product yields are generally low. Variations of batch operation
(such as fed-batch or semi-continuous batch) can improve the run
length, cell density and productivity, but are still far from optimal.
Therefore, optimization of bioreactor designs and culture processes
for mammalian cells is a high priority within the biopharmaceutical
industry (4). Optimization efforts share common goals; maximizing
cell density, cell viability, productivity per cell and culture length. To
achieve these goals, all aspects of the culture process must be examined
and carefully controlled. Factors to be considered include cell line
construction, culture media formulations, feeding strategies,
bioreactor system design and process control strategies (4,5).

2
 There are three basic reasons for converting bacterial fermenters
into mammalian cell bioreactors. First, there is a long history of use
of stirred tanks and their design and operation are well understood.
Second, there are a large number of fermenters being used for
bacterial production that can be converted to mammalian cell
production, resulting in enormous capital equipment savings. Third,
many custom designed mammalian cell culture bioreactors have
been reported as being non-scalable for use as production systems.
Modifications to existing fermenter designs to convert them to
mammalian cell bioreactors include tank design, impeller design,
impeller drive design, gas transfer system design and pH and DO
controller designs. In addition to hardware designs, changes are
being made to the culture process itself.
Optimization of mammalian cell production requires a high degree
of control over the extracellular environment and the ability to run
batches for several weeks. Control of the extracellular environment is
possible with continuous perfusion culture (6-11). In perfusion
culture, fresh media is continuously delivered to the cells and waste
products are removed. Under these conditions, mammalian cells
grow to densities as much as ten-fold higher than in batch or fed-
batch culture, leading to proportional increases in production. The
higher cell densities also result in reduced serum dependence, thus
lowering operating and purification costs (11).

1.1 Available Options for Conversion of a Bioreactor to

Perfusion Operation
A chemostat-like environment can be achieved by simply
pumping the contents of the bioreactor to a harvest bottle while also
feeding, but this technique removes cells and dilutes the culture,
severely limiting the cell density and thus, total production. The
optimal continuous perfusion process is one where the cells are
retained within the bioreactor, while fresh media is added and waste
media is removed, allowing higher cell densities and production.
Retention of cells within the bioreactor requires a separation
process of some type. Methods for converting fermenters to
continuous perfusion bioreactors include the use of disc filters,
spin filters, flat sheet filters and hollow fiber filters. Due to the cell
density and the presence of cell debris, filters used for continuous
perfusion clog or foul after a period of time. As a result, in order to
maintain continuous perfusion, the filter must be replaceable while
the system is operating.
There are two options for connecting a perfusion filtration system
to the bioreactor. It can be mounted inside or outside the bioreactor.

3
 Spin filters and hollow fiber filters have the proper dimensional char-
acteristics to fit within most bioreactors. Flat sheet and hollow fiber
filters can be mounted outside the bioreactor and connected to the
bioreactor via an external recirculation loop. There are advantages
and disadvantages to both types of setups. Placement of the filter
inside the bioreactor eliminates the need for pumps to move the
bioreactor contents through the filter and return them to the
bioreactor. The disadvantage of this setup is that there is no access to
the filter once the culture is started. If there are any problems with the
filtration setup (such as clogging), the bioreactor must be shut down,
ending the production run. In addition, spin filters are specialized
devices that require modifications to the impeller and headplate of
the bioreactor which can require custom engineering. External
filtration setups on the other hand, connect to the bioreactor through
existing ports and do not require specialized hardware. They do
require that the cells be pumped through an external loop. An
external perfusion filter can be changed if it becomes clogged (or
other problems occur), allowing long term continuous operation.
Optimal performance of a perfusion filtration system requires
efficient retention of cells within the bioreactor, a compact and
convenient filtration system that can be readily changed when it
becomes clogged and a membrane porosity sufficiently large to
allow removal of the desired secreted product as part of the perfusion
process. Low product residence time minimizes exposure to pro-
teases, reduces the potential for feedback inhibition and improves
overall product stability (9).
Of the various filtration options available, microporous hollow
fiber filters best meet the requirements for optimal performance of a
perfusion filtration system. They have a high filter surface area to
volume ratio, are compact, easily incorporated into a fluid loop
configuration and can be changed during a run if necessary. A filter
with 0.2-0.45um pore size will allow high flux rates, large volume
throughput and the continuous harvest of spent media containing
cell-free product. Since the harvest is cell free, the initial cell separa-
tion step is eliminated. Thus, simplifying downstream purification. In
addition, a continuous cell-free harvest is compatible with
existing continuous affinity purification techniques.
Spectrum Labs has developed the CellFlo line of hollow fiber
cell separators specifically for the conversion of stirred or airlift
bioreactors to continuous perfusion. CellFlo modules utilize large
internal diameter membranes, which allow recirculation of cells
without damage and reduce clogging by cell debris. CellFlo modules
are available in a variety of surface areas from 800 cm2 to 3.3 m2.

4
 1.2 Results Obtained Using CellFlo Hollow Fiber
Cell Separators for Converting Bioreactors to
Continuous Perfusion
Recent comparisons of fed-batch versus continuous perfusion
operation of a 40L airlift bioreactor for the production of
monoclonal antibodies, showed several differences between the two
modes of operation (8). Both the product concentration and daily
production increased 5-7 fold in perfusion culture (Figures 1&2).
Cumulative production also increased proportionally (Figure 3). In
perfusion culture, the production per liter of media consumed was
2-5 fold higher than in batch culture, thus reducing the overall media
cost per gram of product produced (Figure 4). In addition, the ability
to produce 5-10 fold more product from the existing bioreactor
Figure 1 Daily Antibody Concentration reduced the length of the
production run required to
produce a given quantity of
product. Perfusion can also reduce
capital equipment costs by
extending the capacity of existing
bioreactors and reducing the size of
new bioreactors.
This manual is intended to be a
general guide to conversion of
stirred or airlift bioreactors to continuous perfusion using CellFlo
hollow fiber cell separators. Since there are a wide variety of
different bioreactors being used for mammalian cell culture, this
document will not address specific bioreactor models.
Instead, it will focus on suggestions that should be applicable to a
variety of different types of bioreactors. While bioreactors differ, the
setup and operation of a perfusion loop does not differ
significantly from one bioreactor type to another.
Daily Antibody Production

The use of CellFlo modules for

continuous perfusion of bioreactors is
essentially a modified cross-flow
filtration process. Thus, a brief discussion
of cross-flow filtration will be followed
by specific discussions concerning the
setup and operation of continuous
perfusion bioreactors.

5
2. Basic Concepts of Traditional filtration (or "dead-ended" filtration as it is often called),
Crossflow Separation is a pressure driven separation. It consists of forcing a solution
containing suspended solids directly through a filter material such as a
membrane or filter pad. Solids are retained and generally collect al or
near the surface of the filter, continually reducing the filtration rate and
eventually leading to filter plugging (Figure 5).
Crossflow (or tangential-flow filtration) is a far more efficient means
of separating suspended solids from a solution because the bulk of the
solution flow is parallel to the filter surface rather than perpendicular
to it. The most common forms of tangential flow filtration are stacked
plate and spiral wound flat sheet membrane devices and hollow fiber
devices. Hollow fiber membranes have been found to be more efficient
due to their uniform tubular flow geometry (see ref. 12 for review).

Figure 3
Cumulative Production During operation of the module, solution is
pumped in a recirculating loop through the inside
of the fiber (lumen)(Figure 6). Trans-membrane
pressure (TMP) is generated across the fiber
membrane walls. The filtration process is driven
by the TMP. In microfiltration, solids larger than
the rated pore size of the filter are quantitatively
held back by the membrane, while soluble
components pass through the membrane walls
along with the fluid. Filtration through the fiber
membrane's walls results in a concentration build
up of retained components near the fiber's internal
Figure 4 membrane surface. Ideally, these components are
Production Per Liter of carried down the length of the fiber and out the
Media Consumed end of the module by the sweeping action of the
recirculating fluid. However, under certain
conditions, a gel or cake-like layer can accumulate
on the internal surface of the membranes. This
boundary layer is composed of solids and/or solute
macromolecules which arc retained by the
membrane during the filtration process. This
phenomenon is termed "concentration
polarization" and its formation affects module
performance.
The extent of concentration polarization is
influenced by such variables as particle size

6
 and shape, degree of solid solubility, concentration and temperature
of the solution. Operational variables such as solution velocity along
the membrane surface and IMP also affect concentration polarization.
Control of these variables is important to maximize the flow rate
across the membrane (flux) and solute (product) passage.
Of the options available to minimize concentration polarization,
adequate solution velocity at the liquid/fiber wall interface is usually
the most effective. Fluid velocity is controlled by the system
recirculation rate.
If the passage of solute (e.g. proteins) through the membrane is a
requirement, particular attention should be paid to recirculation rate.
Generally, higher recirculation rates increase the efficiency of solute
passage across the membrane. Depending on the characteristics of the
retained solution components, some solutions can form a gel layer on
the walls of the fibers that effectively reduces the pore size of the
membrane. This gel layer can be such that it, and not the membrane,
controls passage of the solute. Higher recirculation rates and lower
transmembrane pressures minimize this gel layer formation. Setting of
the recirculation rate for a perfusion loop is discussed in detail in
Section 4.1.2.

3. Converting a Batch Virtually any stirred or airlift bioreactor can be converted to

Bioreactor to Perfusion perfusion operation using CellFlo cell separators. The conversion
Using CellFlo Cell process and the materials used differ slightly from one bioreactor type
Separators to another and with increasing bioreactor size. The essential
components required are; three fluid metering pumps, a CellFlo cell
separator, tubing, valves and connectors. The hardware setup and
connections differ somewhat between systems with and without a
steam source. There are several different options for operation, each of
which requires a slightly different system configuration. Several
configurations will be discussed.

7
3.1 Bioreactor With Steam Supply for Sterilization
3.1.1 The Basic System Setup
A basic system setup is illustrated in Figure 7. The essential com-
ponents are the bioreactor, tubing for recirculation, the CellFlo
module, sanitary valves for aseptic connection of the loop to the
bioreactor, a media feed line, a harvest line and three pumps.
The system illustrated is one of numerous possible setup
configurations, depending upon the needs and desires of the end-
user. This setup was designed to meet GMP requirements and to
simplify the setup and on-line changing of CellFlo modules. The
various individual components of the loop will be identified in
order going around the perfusion loop. An explanation of the setup
procedure will follow the discussion of component parts.

3.1.2 Component Parts

1. Bioreactor. Any stirred or airlift will do. Bioreactor size will affect
the size of the tubing and the CellFlo module used but not other
components.
2. Diaphragm valve. Sanitary valve with access port for steam
inlet.
3. Small diaphragm valve. Valve for steam inlet or outlet at the
connection point.
4. Diaphragm valve. Second valve to serve as a sterile disconnection
point for the entire perfusion loop.
5. Small diaphragm valve. For steam inlet or outlet.
6. Flexible pump tubing (suitable for extended operation).
7. Flowmeter with digital readout (optional). To monitor the
recirculation rate.
8. Diaphragm valve. For on-line sterile connection of
replacement CellFlo modules.

8
9. Small diaphragm valve. Steam inlet/outlet for sterile connection.
10. Diaphragm valve. Second valve for sterile connection of
replacement CellFlo modules.
11. Small diaphragm valve. Second steam inlet/outlet valve.
12. CellFlo Module. The size of the CellFlo module will depend
upon the size of bioreactor, the serum content and desired module
life.
13. Product harvest line.
14. Pressure Transducer (optional). For monitoring transmembrane
pressure. This transducer is useful for determining when the filter is
fouling and needs to be changed.
15. Diaphragm valve. For sterile connection of the top end of the
CellFlo module.
16. Small diaphragm valve. Steam inlet/outlet for sterile connection.
17. Diaphragm valve. Second valve for sterile connection of the top
end of the CellFlo module.
18. Small diaphragm valve. Second steam inlet/outlet for sterile
connection.
19. Flexible pump tubing.
20. Diaphragm valve. For sterile connection of the upper end
(bioreactor inlet)of the perfusion loop.
21. Small diaphragm valve. Steam inlet/outlet for sterile connection.
22. Diaphragm valve. Second valve for sterile connection of
perfusion loop.

23. Small diaphragm valve. Second steam

inlet/outlet for sterile connection.
24. Media inlet tubing line.

The setup detailed in Figure 7 provides

maximum flexibility in connection of the
perfusion loop and for changing of
CellFlo modules during a continuous
culture. The dual valve sets at the outlet
and inlet to the bioreactor permit sterile
changing of the entire perfusion loop,
including all tubing. The dual valve sets
at the inlet and outlet of the CellFlo
module permit sterile replacement of just
the CellFlo module. The two valve sets
allow

9
sterile connections to be made by allowing the fluid path to be shut
off by closing two valves (such as valves 2 & 4 in Figure 7). Once the
valves are closed, the loop is disconnected between the two valves.
When the new CellFlo with valves on both ends is connected, the two
ends of the valves can be steamed to sterilize the connection before
opening. All dual sets of valves work on the same principle. Not all
four dual sets of valves are required for setting up a perfusion loop
that allows sterile changes of CellFlo modules. The four sets allow
maximum flexibility in that you can change the whole perfusion
loop, the CellFlo module, or individual segments of the connecting
tubing. If less flexibility is acceptable, the two dual valve sets at the
outlet and inlets of the bioreactor may be eliminated. Initial
sterilization of the perfusion loop can be done using the bioreactor
steam supply. However, under no conditions should non-supported,
flexible tubing be steamed directly. In this type of setup, changes
after initial sterilization involve only the CellFlo module.
All connections throughout the perfusion loop are via either 3/4"
or 1-1/2" Tri-Clamp* sanitary connections.

3.1.3 Bioreactor Modifications and Preparations

Some slight modifications and plumbing preparations are
necessary to prepare a bioreactor for continuous perfusion opera-
tion. These changes, while minor, may require some custom work.
The first modification will be to add inlet and outlet ports to the
bioreactor. The two ports will connect to the diaphragm valves, so the
end fittings must be compatible. We recommend that any new or
existing connector be fitted so it ends in a 3/4" or 1-1/2" sanitary
connector and that the diaphragm valves end in the same size
connector. In addition, the outlet port needs to be mounted at the
bottom of the bioreactor, or have an internal tube extending
approximately one-half way down into the bioreactor connected
to it.
The second modification will be the addition of up to four steam
inlet and outlet lines. These lines will be located at positions 3, 9, 16
and 21 as shown in Figure 7. The steam lines should end at either a
diaphragm valve or a ball valve modified to have a 3/4" or 1-1/2"
sanitary connector on one end. The sanitary ends of the steam
lines will connect to the diaphragm valves at positions 2, 8, 15 and
20 (Figure 7), after the perfusion loop has been assembled on the
bioreactor.
The third modification will be a media feed port entering the
bioreactor. Many bioreactors have ports that are already suitable.
The connector on the media feed port should be compatible with

10
whatever is being used for the media feed line. A dual valve setup like
the rest of the ports may be desirable to allow for sterile changes of
the media feed line.

3.1.4 System Setup Procedure

The exact setup procedure used will depend upon the configura-
tion of the perfusion loop. The procedure outlined here will assume a
configuration similar to that in Figure 7.
1. Assemble a CellFlo module with a diaphragm valve connected
to both ends, with a pressure transducer and harvest line connected
to the filtrate ports as shown in Figure 7. See section 5 for additional
information concerning preparation and sterilization of CellFlo
modules.
2. Assemble the rest of the perfusion loop. This will require two
lengths of tubing with diaphragm valves at both ends. The tubing
should be long enough to permit connection of the CellFlo to the inlet
and outlet ports on the bioreactor.
3. Connect diaphragm valves to the inlet and outlet ports on the
bioreactor.
4. Autoclave the CellFlo with the two valves and the two tubing
segments attached (#s 4,6,8,10,12,13,14,15,17,19,20 in Figure 7).
All valves should be covered but open.
5. After autoclaving, remove all components to a laminar flow
hood, allow them to cool, then close all valves.
6. Once the components have all cooled to room temperature, the
perfusion loop can be connected to the bioreactor.
7. Once the perfusion loop is connected, all connection points
must be sterilized. Open the steam inlet and outlet valves at all
connections (3, 9, 16 and 21 in Figure 7) and steam the connections
for 30-60 min.
8. Shut off all steam valves (3, 9, 16 and 21 in Figure 7). Allow all
valves to cool to room temperature.
9. Prepare the perfusion loop for operation. Load the tubing into
the recirculation pump. Open all valves within the recirculation loop
(2, 4, 8, 10, 15, 17, 20 and 22 in Figure 7).
10. Start the recirculation pump, media feed and harvest pumps.
If the extracapillary space of the CellFlo does not fill with media,
restrict the outlet end of the recirculation loop briefly, or close valve
#15 until the extracapillary space fills, then re-open the restriction or
the valve.

11
3.2 Alternate System Configurations

3.2.1 Fewer Valves

If the number of valves and steam lines required for the setup
described above is more than desired, an alternative setup can be used.
While it is less expensive and simpler, it is also less flexible. This
configuration is illustrated in Figure 8. It differs from Figure 7 in that
the valves on either end of the CellFlo have been eliminated. Thus,
the entire perfusion loop is a single entity.
When installing this unit, the complete loop
is assembled and sterilized intact. It is
connected to the valves at the inlet and
outlet of the bioreactor. When changing
Cell-Flo modules, the entire loop is
removed. When this method is used, it is
convenient to have a second fully
configured perfusion loop available to
minimize the time required for CellFlo
changes. Without a second loop, the first
will need to be cleaned, the CellFlo
changed and the complete loop re-sterilized
while the recirculation pump is shut off.

3.2.2 Dual CellFlo Setup

While several types of configurations are

possible and will work equally well, we will
discuss one other configuration which is
particularly useful. This configuration is
useful in several situations; for larger
bioreactors where the throughput is too
great for a single module, for changing of
modules without shutting off the perfusion
loop, or for both feeding and harvesting
through CellFlo modules. The system
illustrated in Figure 9 contains two CellFlo
modules. Additional components required
are two "Tee" fittings, two dual valve set-
ups connected to a CellFlo module and a
second pressure transducer

12
 (optional). In this configuration, both CellFIo
modules may be operated individually or
simultaneously. Alternatively, one CellFIo may
be used for feeding media into the bioreactor
while the other is used for harvesting (Figure
10). With the addition of a programmable timer,
the feed and harvest CellFIo can be alternated
such that each module is used for feeding for a
period of time, then it can be switched automati-
cally to harvesting using the timer and electronic
valves (not shown). This operation option serves
to backflush the CellFIo, potentially extending
its useful life (see section 4.1.8 for additional
discussion of backflushing).

3.2.3 Stainless Tubing Setup

As a variation to the setups illustrated in Figures 7-10, most of the

perfusion loop can also be hard plumbed with stainless steel tubing so
that the only segment of flexible tubing is the one passing through the
peristaltic pump. In this configuration, one or both of the diaphragm
valves at the inlet and outlet (#s 2, 4, 20, 22 in Figure 7) of the
bioreactor can be eliminated. Instead, the steam supply to the
bioreactor can be used to sterilize the stainless lines. The steam outlet
in this case becomes the steam out on the valve set on the CellFIo (#11
in Figure 7). This configuration requires valve sets at either end of the
Cellflo for changing modules. One complication to this setup is the
pump tubing. The tubing must be suitable for steam service and
adequately shrouded to protect personnel. Otherwise, a bypass must be
incorporated for steaming the stainless tubing. In addition, when a
bypass is used, an alternate method to sterilize the flexible tubing must
be used.

3.3 Bioreactor Without Steam Supply for Sterilization

Most bioreactors up to about 20L in size are not connected to steam.

With these types of bioreactors, connections must be made aseptically.
We have had success using the configuration illustrated in Figure 11.
The perfusion loop has all the same basic components as the loop for
larger systems. However, the loop for a non-steam system generally
uses only flexible tubing, it is smaller in diameter and most, if not all,
connections are hose barb, rather than sanitary fittings. Since this type
of perfusion loop cannot be sterilized after

13
making connections, all CellFlo changes must be made aseptically. We
have also had success using stainless steel Swagelok quick connects
with double end shutoffs at all connection points (Figure 11}. The
bodies of the quick disconnects are connected via bulkhead Swagelok
tube connectors to the inlet and outlet ports of the bioreactor. The
sterns are connected to the recirculation loop tubing via a hose barb
connector double end shutoff stem. Sterile connection of the
recirculation loop is accomplished by flame sterilization of both ends
of the quick disconnect just prior to assembly. Replacement of a
recirculation loop requires disconnecting the old loop, flame
sterilization of the old body and the new stem simultaneously, as the
two are being connected. This procedure allows connections to be
made in an open laboratory environment with a relatively high degree
of confidence.
Setup of this system consists of assembly of all the components of
the perfusion loop and then connection to the bioreactor. The entire
bioreactor and perfusion loop arc autoclaved together at the same time.
CellFlo changes occur at the Swagelok connectors.
As an alternative, two CellFlo modules can be assembled into the
perfusion loop in parallel using "Y" fittings (Figure 12}. One CellFlo
can be clamped off while the other is used. Once the first unit fouls,
the second unit can be opened and the first clamped off. A dual CellFlo
setup like this will often allow the perfusion culture to operate for
several weeks, which is sufficient for most production runs.

3.4 Pump and Tubing Recommendations

Recirculating cells through an external loop requires the use of a

pump. Many types of pumps are available for the movement of
fluids. We recommend a low-shear pump whenever possible (for a dis-
cussion of shear effects sec section 4.1.2). While
peristaltic pumps are not necessarily low shear,
they are common and readily available. If
peristaltic pumps are used, we recommend
Watson-Marlow peristaltic pumps for all
bioreactors up to 100 liters. These pumps have
spring loaded rollers, which allow adjustment of
the tension on the tubing. Watson-Marlow pumps
are available in a variety of sizes, with different
maximum flow rates. Selection of pump size will
depend upon the size of the bioreactor and the
desired flow rate.

14
 We recommend Marprene brand tubing from Watson-Marlow or an
equivalent long life tubing for all flexible lines. Marprene is a high
durability, long life tubing which is ideal for the continuous pumping
used in perfusion. Marprene also passes U.S.R XXI class VI testing for
cytotoxicity. More than one tubing diameter may be required for
optimal operation of a perfusion setup. Since the recirculation rates
used are generally much higher than the feed and harvest rates,
different tubing diameters are usually used for the two types of lines. It
is useful to determine the approximate recirculation, feed and harvest
rates before selecting pumps and tubing. Watson-Marlow publishes
tables of recommended tubing diameters and flow rates for each size
pump they sell. The tables make an excellent reference when selecting
pumps and tubing.

4. Operating 4.1 Perfusion Process Optimization

Considerations 4.1.1 Daily Perfusion Rate

The primary goal of continuous perfusion culture is to maximize the

density and thereby maximize the productivity of mammalian cells
over an extended culture life. The use of CellFlo hollow fiber cell
separators for the conversion of stirred or airlift bioreactors to
continuous perfusion operation will help attain that goal.
Mammalian cells require both glucose and glutamine for catabolic
reactions that produce the ATP and NADPH used for biosynthesis of
cellular material and for the production of proteins of commercial
interest (5, 13). The end products of glucose and glutamine
metabolism; lactate, ammonium and CO2 are secreted into the culture
medium. Both ammonium and lactate have been shown to inhibit
growth of many types of cells (13-17).
Thus, maximizing the cell
density and product formation is
dependent upon optimizing the
extracellular environment.
Typically, this means maintaining
nutrient and waste product
concentrations within specific
ranges. Maintenance of pH,
glucose, glutamine, lactic acid and
ammonia levels is essential to
optimizing the extracellular
environment.
Batch or fed-batch cultures
generally do not maintain the
various metabolic parameters
within narrow enough limits to
achieve high cell densities and
productivity. Continuous perfusion
using CellFlo cell separators allows complete flexibility in setting the
media feed rate. Since the media is usually the single most expensive

15
item used in the production of proteins from mammalian cells, mini-
mizing media usage is essential for economical production. The
single biggest challenge in optimizing production of mammalian
cell-derived biologicals is trying to balance the need to maintain a
proper nutrient environment within the bioreactor with the need to
minimize media usage.
Traditional methods of maintaining adequate levels of glucose,
glutamine and pH, while simultaneously minimizing ammonia and
lactate levels, involve simple dilution with media. More recent
attempts at optimization involve the selective addition of glucose,
glutamine and base to cultures. The goal in this case is to maintain
proper levels of energy sources and pH while minimizing lactate and
ammonia production (1, 8, 18, 19). Addition of these metabolites is
supplemental to a regular media feed. Effective optimization of
media usage thus requires very close monitoring of all essential
metabolites.
When setting up a continuous perfusion culture, a media feed rate
needs to be established. The optimal rate depends upon a number of
factors such as; the cell line, the media, the percentage of serum used
and the tolerance level of the cell line to ammonium and lactate. It
also depends upon whether glucose, glutamine, ammonium and
lactate will be monitored and whether specific addition of glucose,
glutamine and base is feasible. A simple way to discuss the media
feed rate used in continuous perfusion cultures is to equate the rate
to bioreactor volumes per day. The daily perfusion rate has been
shown to have a significant effect on both the viable cell count and
the percent viability of a culture (14). These studies were done using
media perfusion only, without specific addition of glucose, glutamine
or base. The results suggest that perfusion rates of 0.8 bioreactor
volumes per day or greater are necessary to maintain both the
percent viability and specific cell growth rate. If specific metabolite
additions are used successfully, it can be assumed that lower
perfusion rates will probably be effective.
Determining optimal perfusion and metabolite addition rates
requires considerable experimentation with a specific cell line. Is
there a perfusion rate that can be used that will work for most cell
lines? Yes. Most cell lines will increase 5-10 fold in cell density and
productivity if a perfusion rate of one bioreactor volume per day is
used. Higher or lower rates may be needed for certain cell lines, but
one volume/day is a good place to start. If glucose levels are found to
be too low or lactic acid and ammonia levels are too high, higher
perfusion rates may be necessary. If glucose levels are high and
lactic acid and ammonia levels are low, lower perfusion rates can
probably be used.

16
4.1.2 Recirculation Rate
The recirculation rate of the cells and media through the CellFlo
separator is dependent upon the size of the bioreactor. As discussed
in Section 2, optimal crossflow filtration generally occurs at high
recirculation rates. It is also desirable to maximize the life of the cell
separator so that the number of CellFlo changes during the culture
run is minimized. These factors must be balanced with the desire to
maintain a high cell viability within the culture. Generally, high
recirculation rates are detrimental to cells. Thus, the recirculation
rate used is a compromise between two considerations.
The limiting factor in setting the recirculation rate is the shear
sensitivity of the cell line. All cell lines have different sensitivities to
shear. Three major sources of shear are present within stirred or
airlift cultures operated in perfusion mode using peristaltic pumps.
They are; the agitation system, the gassing system and the pump/
perfusion loop. In stirred cultures, the agitation system is usually an
impeller of some type. Most mammalian cell bioreactors have
impellers designed to impart low shear stress, but the actual shear
generated varies considerably. Airlift bioreactors rely on a column of
rising air to provide agitation and gassing. Air bubbles in a culture
generate shear at the air/liquid interface. Most stirred and airlift
bioreactors utilize some degree of sparging for oxygenation of
cultures. Sparging also contributes to shear on the cells. In addition,
higher density cultures require proportionally more oxygen. If
sparging is the main source of oxygen to the culture, then higher
densities require increased sparging, which increases the shear on
the cells. The third source of shear in a perfusion culture is the
external perfusion loop. The shear stress in the perfusion loop is
directly proportional to the recirculation rate of the culture fluid. It is
also influenced by the diameter of the tubing used for the loop and
the RPM of the pump. The use of larger diameter tubing and lower
RPM settings results in lower shear at any given flow rate. We rec-
ommend relatively large diameter tubing and high capacity pumps
for setting up perfusion loops.
Recent studies have examined the effects of shear on mammalian
cells being pumped through an external recirculation loop setup
containing a hollow fiber cell separator (20). The setup was similar to
that used for continuous perfusion of stirred or airlift bioreactors.
These studies showed that cells can tolerate wall shear rates up to
3000 sec ' before loss of viability occurs. For time-independent,
laminar, newtonian fluid flow in a circular tube (fiber), the shear rate
{) can be expressed as:

17
 = 4Q/; r3 = 8Vav/D (21 -23) (Eq. 4.1,2-A)
where:
Q = flow rate
r - radius of the fiber
V.iv = average fluid velocity
D = diameter of the fiber since:
Vav = 4Q/7D2N (Eq. 4.1.2-B)
where;
N = number of fibers then:
= 32Q/D3N (Eq.4.1.2-C)

Which simplifies to:

=1.7X105(Q)/D3N (Eq.4.1.2-D)

When:

= sec-1
Q = liters per minute
D = millimeters
r = millimeters

From equation 4.1.2-D, it is fairly simple to calculate the shear rate

within the CellFlo module at any given recirculation rate. Knowing
t h e shear rate is useful when setting recirculation rates for
continuous perfusion using CellFlo modules. Recirculation rates
resulting in shear rates exceeding 3000 sec-1 are not recommended.
Thus, for each of the CellFlo products we can recommend a
maximum recirculation rate, depending upon the size of the
module. These are:

Catalog Number Number of Fibers Surface Area Maximum Recirculation Rate

(at 3000 sec-1)
C22M-011-01N 100 0.06 m2 1.8 LPM
C22M-051-01N 350 0.35 m2 6.2 LPM
C22M-101-01N 1575 1.00 m2 28.0 LPM
C22M-201-01N 1575 2.10 m2 28.0 LPM
C22M-30J-01N 1575 3.30 m2 ' 28.0 LPM

These flow rates assume an internal fiber diameter of 1,0 mm. They
also represent an UPPER limit for recirculation rate. Resistance to
shear varies greatly from cell line to cell line and under different
culture conditions. Recirculation rates as much as 10 fold lower may
be necessary with certain cell lines and culture conditions.

18
 Shear stress on cells also occurs as they pass from one diameter tube
to another as they are pumped through the recirculation loop. Small
orifices such as those found in sample ports of some bioreactors should
be avoided. It is important to keep the diameter of the tubing and
connectors used to construct the recirculation loop as constant as possible
in order to avoid unnecessary shear.

4.1.3 Changing of the CellFlo Modules

As previously discussed, continuous perfusion using CellFlo

modules is essentially a filtration operation. Like all filters, CellFlo
hollow fiber cell separators have a finite lifetime. The operating life of
the CellFlo can vary considerably due to the volume filtered per day, the
cell density, the serum content of the media and the recirculation rate
within the perfusion loop. Deciding when to change the CellFlo can be
difficult. When the filtration rate declines due to fouling, the harvest rate
drops. If the media feed rate has been set equal to the harvest rate, the
media level within the bioreactor will rise. If level detectors are used to
control the feed and harvest pumps (see section 4.1.4), the feed pump
will slow or stop as the filtrate rate declines. The volume within the
harvest tank will also level off and stop. Without level detectors, the
possibility of overfilling the bioreactor exists. Under similar running
conditions, CellFlo modules will have a predictable operating life. Thus,
after some experience is gained using CellFlo in a process, the modules
can be safely changed at fixed intervals. This interval is usually slightly
shorter than the normal maximum life of the CellFlo.
If a more precise measure of the CellFlo life is desired, a pressure
transducer can be installed on the second filtrate port of the CellFlo. The
transducer will display the backpressure on the filtrate. In the early stages
of the life of a CellFlo, the pressure reading will be positive or zero. As
the filter begins to foul, the reading will drop and may become negative
as the harvest pump begins to pull filtrate. After monitoring several
changes of CellFlo modules using the pressure transducer, fouling will
correlate with certain pressure readings. Once a particular threshold
pressure has been determined, it can be used as the signal for changing of
the module. Operation at a negative pressure should be avoided as it
increases the likelihood of a contamination.

4.1.4 Bioreactor Volume Control

As part of the perfusion process, some of the volume of the bioreac-

tor is continuously removed using a pump. The removed volume is
replaced at the same rate using a separate pump. Due to differences in the
pumps and in tubing, it is difficult to precisely match the flow rates of
two different pumps. In addition, (as stated in section 4.1.3),

19
filter fouling can reduce the removal rate from the bioreactor. Thus,
the volume within the bioreactor can fluctuate. If the volume gets too
high or too low, problems can occur. Several options are available to
help maintain the volume of the bioreactor. As mentioned above, a
pressure transducer can help predict when the CellFlo needs to be
changed and will help prevent volume increases due to fouling of the
CellFlo. For routine operation, level detectors within the bioreactor
work quite well. Many equipment manufacturers offer level
detector/controllers for their systems. The best setup for using level
detectors is one where the detectors are connected to the feed and
harvest pumps so that when the level gets too high, the feed pump is
shut off and when the level gets too low, the harvest pump is shut off.
For smaller bioreactors or those without level detectors, regular
inspection of the level and appropriate manual adjustment of the
pumps can be effective. A good schedule is to check the fluid level in
the bioreactor in the morning and again at the end of the day.

4.1.5 Cell Harvesting

Periodic removal of cells from a continuous perfusion culture may
be beneficial. If the percentage viability is lower than desired,
periodic cell removal can increase it. Removal of small volumes
frequently, rather than larger volumes infrequently is preferred.
Removal of 5-10% of the bioreactor volume daily is suggested.

4.1.6 Serum
Many mammalian cell lines are cultured in media containing
animal serum of one type or another, primarily fetal bovine serum
(FBS). Cultures are usually supplemented with 1-10% serum. Serum
can cost 200-400 dollars per liter. Thus, the use of serum can
dramatically affect the overall cost of production of biologicals from
mammalian cells. One of the beneficial effects of continuous
perfusion culture is increased cell densities. Mammalian cells grown at
high densities require less serum for growth and production of
proteins (24). However, until the cells reach high density, they require
serum or serum components (11). One of the benefits of using CellFlo
cell separators for continuous perfusion is the ability to eliminate
serum from the culture once a sufficient cell density has been reached.
Serum removal is possible through two basic methods, gradual
removal over several days, or complete removal in one step. Gradual
removal is frequently used for long term cultures producing secreted
products. One step removal works well for processes where product
is collected only for short periods, such as lytic virus production or
where the secretion must be induced and cannot be maintained for
long periods. The advantages of serum removal from l o n g term
cultures are two-fold; cost reduction of the production

20
and simplification of purification (also a cost reduction). In short
term applications the advantage is simplification of purification.
With some products, the presence of serum completely prevents
purification.
Gradual serum removal is done by reducing the percentage of
serum contained in the daily feed media. For optimal results, the
gradual removal should be done over a period of 5-7 days. Total
removal of serum too quickly can result in loss of cell viability.
Immediate serum removal is done by using the harvest line to
nearly drain the bioreactor, then refilling the bioreactor with serum
free media. To do this, the recirculation pump is left on, the media
feed pump is shut off and the harvest pump is left on. Once the
volume in the bioreactor reaches about 10% of the starting volume,
the harvest pump is shut off and the feed pump is started and run
until the volume in the bioreactor returns to normal. This process
reduces the serum content 10 fold. If greater reductions are desired,
the process can be repeated. Caution must be exercised when
reducing the volume of the culture so that monitoring probes
contained within the bioreactor are not exposed to air for any length of
time. Exposure to air can damage monitoring probes.

4.1.7 Antifoams and Surfactants

Mammalian cells grown in stirred and airlift bioreactors arc subject to
foaming due to gas sparging in the presence of serum (proteins in
general). The foam tends to accumulate at the top of bioreactors and
can be the source of considerable cell debris. Frequently, this debris
will break off and float throughout the bioreactor. When a bioreactor is
being run in continuous perfusion using CellFlo modules, this
debris is pumped through the inside of the fibers. The debris can
physically occlude the walls of the membrane, leading to shortened
module life.
Antifoams work well to reduce the foaming a n d cell debris.
However, many antifoams, especially silicone based ones, have severe
detrimental effects on the filtration capacity of membranes,
including Spectrum Labs membranes. Recent reports have
suggested that one type of surfactant, Pluronic F-68 (BASF
corporation, Parsippany, NJ) can reduce the damage caused by gas
sparging in insect cells grown in sparged and airlift bioreactors (25).
Other studies have discussed growth of mammalian cells in medium
supplemented with Pluronic F-68 (26, 27). Pluronic F-68 is a
nonionic block copolymer with an average molecular weight of
8400, consisting of a center block of polyoxypropylene (20% by
weight) and blocks of polyoxyethylene at both ends (25). It is
typically used in concentrations

21
ranging from 0.02-0.2%. Pluronic F-68 does not affect the filtration
capacity of Spectrum Labs hollow fiber membranes. In fact, preliminary
data suggests that it actually increases filtration capacity by as much as
30-50% (unpublished observations). When Pluronic F-68 was used in a
40L airlift bioreactor in continuous perfusion using CellFlo cell
separators, a dramatic reduction was seen in the quantity of foam and cell
debris present. It also appeared to improve CellFlo life by 30-50%
(unpublished observations),

4.1.8 Backflushing and Optional Media Feeding Strategies

One routinely used method for improving filter life in crossflow

filtration applications is to periodically backflush the filter. Backflushing
helps to remove the gel or cake layer that builds up on the inside walls of
the hollow fibers (see section 2; Basic Concepts of Crossflow Separation;
Figure 4). Reducing the cake layer improves both the flux through the
membrane and filter life.
Several options exist for backflushing of CellFlo modules and
feeding the bioreactor through the CellFlo. Some of the different
variations that have been used include:

Using the second permeate port as the fresh media inlet and
cycling the feed and harvest using a timer so that only one or the
other operates at a time.
Backflushing by reversing the harvest pump in order to extend
filter life.

Setting up two CellFlo modules in parallel, using one for feeding

and one for harvesting. Set up on a timer for automatic switching of
feed and harvest modules {Figure 10).

Using the second permeate port for

feeding fresh media (Figure 13) has two
potential advantages. First, the media passes
through an additional filter before entering the
bioreactor, reducing the risk of contamination.
Second, it will also backflush the filter,
improving filter life. This setup requires a
programmable timer, which can be purchased
relatively inexpensively. One disadvantage of
this setup is that it is also a discontinuous
process. However, the time cycles for feed
and harvest can be kept short, thereby
minimizing the swings in metabolite
concentration.

22
 A regular backflush cycle can also be
used even if feeding is not done
through the permeate port. The harvest
pump can he reversed either manually
or automatically so that media is
returned to the bioreactor for a short
period. Manual operation involves
shutting off the media feed line long
enough to reduce the volume of the
reactor equal to that which will be
returned while backflushing. Then the
harvest pump is reversed long enough
to re-fill the bioreactor. Backflushing
should be done at a high flow rate in
order to maximize the cleaning effect.
Once a backflush cycle is complete, the feed and harvest pumps are
returned to normal. Operation of a backflush cycle on a timer with
frequent, short backflushes will be more efficient than occasional
cycles.
The use of two modules in parallel offers several advantages; it
doubles the filtration capacity, it helps maintain sterility and both
filters are backflushed regularly, improving filter life.

4.1.9 Sterility Protection

In any mammalian cell culture, maintaining sterility is an issue.

Mammalian cell cultures are very susceptible to contamination with
bacteria or fungi. Maintenance of sterility is an issue with short term
batch cultures. It is a more important issue with long term continuous
cultures. The most frequent source of contamination in perfusion
cultures is usually the media feed source. The media being fed into the
system can become contaminated in several ways; during changes of
feed bottles, through ineffective sterilization of the media, or from
commercially prepared liquid media that becomes contaminated before
use.
One effective means of preventing contamination from a feed source
is through the use of sterility protection filters. Placement of the
sterility protection filter in the line going from the media feed bottle to
the bioreactor (Figure 14} effectively prevents contaminants
originating in the feed bottle from reaching the bioreactor. For
maximum effectiveness, the filter should be placed in the feed line
before sterilization. The filter should also be between any

23
 connection point and the bioreactor. For example, if the perfusion run
lasts longer than the media from a single container, the media containers
will need to be switched, which requires aseptic connections. All such
connections should be made "upstream" of the filter. If a contamination
occurs in the media feed, it will be stopped by the sterility protection
filter. However, once detected, both the media container and the filter
must be removed and replaced. Without a second filter, replacement
requires an unprotected aseptic connection. However, the dual filter
arrangement shown in Figure 15, ensures that all connections are made
"upstream" of a sterilizing filter.
Sterility protection filters work best with serum free media only. The
presence of serum in media reduces the filtration capacity of the filters.
This reduction with serum may shorten the filter life below the life
expectancy of the culture. In cases where the filter life is less than
desired, the filter may be changed aseptically or the original system may
be configured with more than one filter in parallel. When one filter
becomes clogged, another can be used.

5. Sterilization and 5.1 Sterilization

Sanitization
CellFlo modules can be sterilized by autoclaving, radiation, or
ethylene oxide. They may be sanitized by chemical flushing.

5.1.1 Autoclaving a Perfusion Loop Containing a CellFlo Module

1. Pre-wet the CellFlo module with clean water.

2. Assemble a complete perfusion loop including all tubing,
connectors and the CellFlo module (see section 3 for loop
configuration). Do not fully tighten the sanitary clamps.
Note: If the module is to be diffusion or bubble point tested, refer to
section 6.1 for wetting procedure.
3. Wrap the ends of the loop, making sure to leave any valves or
connectors open.
4. Autoclave the complete perfusion loop at 121C for 30 minutes
on a slow exhaust cycle. Note: Do not use a temperature cycle above
124C as damage to the CellFlo module may occur.

24
 5. Allow the system to cool to room temperature and finger-tighten
all clamps to assure a proper seal.
6. Check that all tubing connections are tight and close any valves or
connectors on the ends of the loop.
Note: Spectrum Labs warrants CellFlo Modules to be auioclavable
one time. Due to the potential variability of use and storage
beyond our control no warranty is given, nor implied, for
repeated, autoclavability.

5.1.2 Chemical Sanitization

If you wish to sanitize the entire perfusion loop including the
CellFlo module in place, it may be chemically sanitized by flushing
one of the following solutions through the assembled loop under the
following conditions:

1. Sodium Hypochlorite - 200 ppm for 20 minutes.

2. Formaldehyde - 0.5% for 20 minutes.
3. Hydrogen Peroxide - 2.0% for 20 minutes.
4. Hot Water - 180F (82C) for 1-2 hours.

6. Integrity Testing Integrity testing may be performed by several methods:

Pressure Hold Test
Diffusion Test
Bubble Point Test
If your validation scheme requires diffusion or bubble point
testing of the module, temporary modifications will be necessary to
the perfusion loop.
Prior to initial integrity testing, dry modules must be primed to
ensure complete fiber wetting. Priming and integrity testing of
modules requires modification of the basic perfusion loop setup. To
modify the setup, a "Tee" connector is inserted in the perfusion loop at
the inlet and outlet ends of the CellFlo as shown in Figure 16. A
length of tubing is connected to each Tee. Clamps to restrict the
tubing arc required as shown in Figure 16.

6.1 Wetting the CellFlo Module (Figure 16)

Wetting of a Cellflo module that has already been assembled ir\t a
perfusion loop and sterilized is difficult. If an integrity test is
required, the CellFlo module should be wetted before sterilization.

25
1. Assemble the perfusion loop as described in Section 3.1.1 or 3.3.
2. Fill a 2 liter reservoir with 30% (by volume) isopropyl alcohol in
water.
3. Install the added length of tubing at the inlet end of the CellFlo into
the recirculation pump. Place the end of the tubing in the IPA/water
reservoir. Also place the end of the added length of tubing at the
outlet end of the CellFlo and the open end of the filtrate line into the
reservoir.
4. Close the two clamps at the inlet and outlet end of the CellFlo.
5. Turn the recirculation pump on to forward flow and set the speed
controller to about 2-3 liters per minute.
6. Flush the system for 1-2 minutes. During this time it is advisable to
partially restrict the tubing on the outlet of the recirculation loop so
that the solution is flushed through the fiber walls. Also be sure that
the extracapillary space (the space between the outside wall of the
fibers and the shell of the module) is filled with solution so that all
of the fibers are immersed.
7. After 1 -2 minutes of flushing, with the pump still on, remove the
inlet end of the recirculation loop from the reservoir and wait until
air bubbles flow out of the end of the outlet line, then turn off the
pump.
8. Drain the extracapillary space (HCS) of solution. If only one of
the two filtrate ports is connected to a tubing segment, the port will
need to be at the bottom of the module to allow for complete
draining. However, the harvest line should be on the top during a
normal perfusion run.
9. Refill the reservoir with clean, pyrogen-free water (preferably
distilled or deionized and pre-filtered with a 0.2urn filter) and set up
the system as described in step 3 above.
10. Follow steps 4 through 6 above, except the flush time should be 3-4
minutes.
11. Repeat steps 8 and 9 a second time.

26
 6.2 Pressure Hold Test (Figure 17)

We recommend this procedure be done after sterilization of the

system. Care must be taken to maintain sterility of the line coming off
the Tee at the inlet to the CellFlo and the filtrate line. We recommend
that hydrophobic vent filters be installed on both lines after wetting of
the CellFlo but before sterilization (Figure 17).
1. Close off the line at the outlet end of the CellFlo, either by using
a clamp or by closing the valve at the outlet end of the CellFlo
module.
2. Remove the inlet line from the flush/water solution.
3. Place the end of the filtrate line in a liquid container.
4. Momentarily start the recirculation pump. Slowly build a 5 to
10 psig (0.3 to 0.7 kg/cm^) pressure reading on the feed gauge
then shut oft the pump.
5. Rapid bubbling through the filtrate line into the liquid container
indicates a loss of integrity. Some bubbling may occur immedi-
ately upon pressurization, but will stop after a short time. This
bubbling is due to normal expansion of the membranes and does
not indicate loss of integrity.
6. No or very slow bubbling is normal diffusional flow and
indicates an integral module.
7. If the pressure drops quickly and there is no bubbling in the
module, either a connection is loose or there is leakage of air
back through the pump. Clamp off the inlet line between the
CellFlo module and the pump and cheek all connections.

6.3 Diffusion and Bubble Point Tests

Diffusion and bubble point tests can be performed on CellFlo units

but the procedures require extensive modification of the perfusion loop
setup. For information on these procedures, contact Spectrum Labs
Technical Services at 310-885-4600

27
7. References

1. Ratafia, M. 1987. Mammalian cell culture: worldwide activities and markets. Biotechnology
6(7); 692-694.
2. Griffiths, J.B. 1988. Overview of cell culture systems and their scale-up, In: Animal Cell
Biotechnology, Vol. III, chapter 8. Spire, R.E. and Griffiths, J.B. (Eds.) Academic Press,
London.
3. Van Brunt, J. 1987. A closer look at fermenters and bioreactors. Biotechnology 5(11); 1134-
1138.
4. Knight, P. 1988. Fermentation special report. Biotechnology 6(5); 506-516.
5. Glacken, M.W. 1988. Catabolic control of mammalian cell culture. Biotechnology 5(11); 1134-
1138.
6. Parkinson, G. 1988. The next wave of bioprocessing arrives. Chem. Eng. 95(18); 37-44.
7. Tolbert, W.R., Srigley, W.R. and Prior, C.P. 1988. Perfusion culture systems for large scale
pharmaceutical production. In: Animal Cell Biotechnology, Spire, R.E. and Griffiths, J.B.
(Eds.) Vol. III, Chapter 17, Academic Press, London.
8. Murphy, T.J. 1988. Continuous perfusion in a 40L airlift bioreactor. Microgon Technical
Report.
9. Prior, C.P., Prior, G.M. and Hope, J.A. 1989. Perfusion technology; a production method for
human rtPA. Amer. Biotech. Lab. 6(3); 25-31.
10. MacMichael, G.J. 1989. The use of perfusion in mammalian cell culture. Amer. Biotech. Lab.
6(3); 34-42.
11. Prior, C.P., Doyle, K.R., Duffy, S.A., Hope, J.A., Moellering, B.J., Prior, G.M., Scott, R.W.
and Tolbert, W.R. 1989. The recovery of highly purified biopharmaceuticals from perfusion
cell culture bioreactors. J. Parent. Sci. and Technol., 43(1); 15-23.
12. Tutunjian, R.S. 1984. Cell separations with hollow fiber membranes. Developments in
Industrial Microbiology 25: 415-4356.
13. Glacken, M.W., Fleishaker, R.J. and 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. Biotech. And Bioeng. 28; 1376-1389.
14. Miller, W.M., Blanch, H.W. and Wilke, C.R. 1988. A kinetic analysis of hybridoma growth and
metabolism in batch and continuous suspension culture: effect of nutrient concentration,
dilution rate and pH. Biotech. And Bioeng. 32; 947-965.
15. Glacken, M.W., Adema, E. and Sinskey, A.J. 1988. Mathematical descriptions of hybridoma
culture kinetics: I. Intial metabolic rates. Biotech. And Bioeng. 32; 491-506.
16. Reuveny, S., Velez, D., Macmillan, J.D. and Miller, L. 1986. Factors affecting cell growth and
monoclonal antibody production in stirred reactors. J. Immunol. Methods 86; 53-60
17. Thorpe, J.S., Murdin, A.D., Sanders, P.G. and Spier, R.E. 1987. The effect of waste products of
cellular metabolism on growth and protein synthesis in a mouse hybridoma cell line. Paper
presented at the 194th ACS National Meeting, New Orleans.
18. Miller, W.M., Wilke, C.R. and Blanch, H.W. 1989. Transient responses of hybridoma cells to
nutrient additions in continuous culture: I. Glucose pulse and step changes. Biotech and Bioeng.
33; 477-486.
19. Miller, W.M., Wilke, C.R. and Blanch, H.W. 1989. Transient responses of hybridoma cells to
nutrient additions in continuous culture: II. Glucose pulse and step changes. Biotech and
Bioeng. 33; 487-499.
20. Maiorella, B., Dorin, Glenn, Harano, D., Carion, A. and Keutzer, B. 1988. Design
considerations for aseptic crossflow microfiltration harvesting of viable animal cells. Paper
presented at the AICHE National Meeting, Washington, D.C.
21. Cheryan, M. 1986. Ultrafiltration Handbook, pg 112. Technomic Publishing, Lancaster, PA.
22. Bird, R.B., Stewart, W.E. and Lightfoot, E.N. 1960. Transport Phenomena, pg 46. John Wiley
and Sons, New York.
23. Bliem, R. and Katinger, H. 1988. Scale-up engineering in animal cell technology: Part II.
TIBTECH 6(9); 224-230.
24. Lauffenburger, D. and Cozens, C. 1989. Regulation of mammalian cell growth by autocrine
growth factors: Analysis of consequences for inoculum cell density effect. Biotech and Bioeng.
33; 1365-1378.
25. Murhammer, D.W. and Goochee, C.F. 1988. Scaleup of insect cell cultures: protective effects
28
 of Pluronic F-68. Biotechnology 6(12); 1411-1418.
26. Handa, A., Emery, A.N. and Spier, R.E. 1987. On the evaluation of gas-liquid interfacial effects
on hybridoma viability in bubble column bioreactors. Develop. Biol. Standard. 66; 241-252.
27. Handa, A., Emery, A.N. and Spier, R.E. 1987. Effect of gas-liquid inter faces on the growth of
mammalian cells: mechanisms of cell damage by bubbles. Presented at the 8th meeting of Eur.
Soc. for Animal Cell Technology, Israel.

29