4sa Biotechnol. Prog.

1990, 6, 458-464

Inclined Sedimentation for Selective Retention of Viable Hybridomas in

a Continuous Suspension Bioreactor
Brian C. Batt, Robert H. Davis, and Dhinakar S.Kornpala'
Department of Chemical Engineering, University of Colorado, Boulder, Colorado 80309-0424

T h e continuous separation of nonviable hybridoma cells from viable hybridoma cells

by using a narrow rectangular channel t h a t is inclined from t h e vertical has been
investigated experimentally. The effectiveness of the settler in selectively retaining viable
hybridomas in the bioreactor while permitting the removal of nonviable hybridomas
has been shown to depend on the flow rate through the settler. Intermediate flow rates
through the settler have been found t o provide the highest removal of nonviable hy-
bridomas relative t o viable hybridoma retention. At high dilution rates through the
chemostat, over 95 % of the viable cells could be partitioned to the bottom of the settler
while over 50 5% of the nonviable cells are removed through the top of the settler. This
successful separation is due to the significantly larger size of the viable hybridomas than
the nonviable ones. A continuous perfusion experiment was performed in which an
external inclined settler was used to retain virtually all of the viable hybridomas in the
culture, while selectively removing from the culture approximately 20% of the nonvi-
able cells that entered the settler. A stable viable cell concentration of 1.0 X lo7 cells/
m L was achieved, as was a n antibody productivity of over 50 pg/(mL-day). These
represent 3- and 6-fold increases, respectively, over the values obtained from a chemo-
stat culture without cell retention.

Introduction sufficient to cause difficulty in filtering spent medium (11,

Early studies in batch hybridoma culturing showed that
monoclonal antibody production is proportional to the
number of viable cells in the culture (1-6). Furthermore,
it has recently been shown that nonviable or dead hybri-
domas do not lose significant amounts of antibody and,
consequently, contribute negligibly to the productivity of
a culture (7). Thus, high productivity of monoclonal
antibody in suspension culture can be attained not by
maximal cell growth rate but by maintaining high viable
cell concentrations.
High viable cell concentrations in suspension bioreac-
tors are currently achieved by using various cell retention
or recycle devices. These enable bioreactors to be per-
fused without cell washout a t dilution rates greater than
the maximum specific growth rate of the cells. A typical
cell retention device employs an internal spin filter, which
is attached to the reactor's impeller shaft and permits the
removal of cell-free culture medium (8-10). Cell recycle
devices typically employ an external tangential flow filter
in order to concentrate the cell suspension and recycle it
back to the bioreactor (11,12). Another alternative passes
the reactor effluent through a vertical sedimentation
column that is sufficiently long to allow suspended cells
to settle back into the bioreactor before they can be washed
out (13-15).
Filter devices can be detrimental to long-term culture
productivity because the hybridomas are subjected t o
excessive shear forces, resulting in higher cell death rates.
Moreover, all of t h e methods mentioned above are
characterized by the continuous accumulation of unpro-
ductive dead cells in the bioreactor, which forces the
intermittent removal of a culture fraction containing both
viable and nonviable cells. This is especially imperative
in the case of filtering devices when cell density becomes
* Author to whom correspondence should be addressed.
12). The accumulation of nonviable cells and the removal
of viable cells along with nonviable cells limit the culture
productivity.
It is conceivable that the limitations in culture produc-
tivity may be overcome by continuously removing only non-
viable hybridomas from the reactor while selectively
retaining all viable cells in the reactor, provided that a
method may be developed for accomplishing a selective
separation of these subpopulations. It is known that faster-
growing, viable mammalian cells are larger in average cell
volume (12) and specific mass (16) than slower-growing
and dead cells. Therefore, it is expected that the desired
selective cell separation may be accomplished by exploiting
the different sedimentation velocities of viable and non-
viable hybridomas by using inclined sedimentation.
An inclined settler is a long and narrow tube or channel
that is inclined from the vertical. Larger cells are removed
from suspension by settling onto the upward-facing surfaces
of the settler, where they form thin sediment layers that
slide down to be collected a t the bottom of the vessel.
Smaller cells, which do not settle as rapidly, are collected
from the top. A kinematic theory for inclined sedimen-
tation was developed more than a half-century ago (17,
28). This theory states that the volumetric production rate
of clarified fluid from an inclined channel due to particle
sedimentation is equal to the vertical settling velocity of
the particles multiplied by the horizontal projected area
of the channel surface available for sedimentation. For
a rectangular channel, this implies that
S(u) = uw(L sin 0 + b cos 0) (1)
where S(u) is the volumetric rate of production of fluid
clarified of particles with settling velocity u, w is the width
of the settler, L is the length of the settler, b is the spacing

8756-7938/90/3006-0458$02.50/00 1990 American Chemical Society and American Institute of Chemical Engineers
Eiofechnol. Prw...1990, Vol. 6, No. 6
between the inclined walls of the settler, and 0 is the angle
of inclination of the settler from the vertical.
For a channel of fixed dimensions and inclination angle,
whether a cell of a given Stokes velocity settles out of
suspension or is instead washed out of the channel is
controlled by the volumetric flow rate through the channel,
which determines the residence time of the suspended cells.
Differences in Stokes velocity between cells of equal density
are determined primarily by differences in cell diameter,
as indicated by Stokes’ law:
u = d%, - p ) g / 1 8 ~ (2)
where d is the cell diameter, pc is its density, p is the fluid
density, p is the fluid viscosity, and g is the gravitational
acceleration constant. Thus, by adjusting the settler
overflow rate so that the residence time of the suspension
in the settler is in between the settling times of two sub-
populations of different sizes, a selective cell separation
may be accomplished. The benefit of inclined sedimen-
tation is that cells need to settle only a distance on the
order of the narrow spacing between the inclined walls of
the settler in order to be removed from suspension, rather
than a distance on the order of the settler height as in
vertical sedimentation.
Inclined sedimentation has been applied successfully in
a variety of bioreactor designs for selective cell retention:
maintenance of a stable continuous culture containing a
mixed bacteria/yeast population (19), selective recycle of
flocculant yeast (20)and bacteria (21), and the separation
of nondividing and dividing yeast (22). The advantages
of applying an inclined settler as a cell retention device
are the elimination of a high shear environment in the cell
retention process, faster cell sedimentation rates than
vertical settlers, and reduced accumulation of nonviable
cells in the bioreactor. Furthermore, by continuously
removing nonviable cells while selectively retaining viable
cells in the culture, a perfusion bioreactor can in principle
be operated as a steady-state process.
This paper describes a study of the feasibility of using
inclined sedimentation to selectively separate nonviable
from viable hybridomas in a suspension hioreactor. The
performance of an inclined settler in promoting high viable
cell concentration and antibody productivity, while
preferentially removing nonviable cells from a perfusion
hybridoma bioreactor, is also described.
Materials a n d Methods
Cell Line and Medium. Cell line AB2-143.2 is a mouse
hybridoma derived from Sp2/0 myeloma that produces
IgC2a antibodies against benzenearsonate (23). Cells were
grown in Dulbecco’s modified Eagle’s medium (DMEM)
supplemented with 10% fetal bovine serum and 1%each
lOOX MEM nonessential amino acids and 10 mM sodium
pyruvate solution. The glucose concentration was 22 mM,
and extra glutamine was added to give a concentration of
4.8 mM. No antibiotics were used in the medium. Cell
size measurements (see below) indicated that the nonvi-
able cells have an average diameter of approximately 8 pm,
whereas that of the viable cells is approximately 12-14 pm,
with the latter depending on the cell growth rate. The cell
density was not determined exactly, but a crude neutral
buoyancy measurement indicated that it is approximately
1.06g/cm3. From eq 2, the average sedimentation velocity
of the nonviable cells (d = S pm) is 1.1cm/h, while that
of the viable cells (d = 13 pm) is 2.9 cm/h.
Cell Culture Reactor a n d Inclined Sedimentation
Channel. A 1.5-L reactor (Celligen, New Brunswick
Scientific) equipped with a floating surface aerator (24,
Figure 1. Schematic of the hioreactor confwation for studying
the effect of the settler overflow rate on viable and nonviable hy-
bridoma separation. During perfusion operation, the settler
overflow stream is the reactor effluent instead of recycling it back
to the reactor.
25) and marine impeller was used in all experiments.
*
Agitation was controlled at 120 1rpm, and the tempera-
ture was maintained at 37.0 f 0.1 OC. An interactive, four-
gas control system designed for use with the Celligen
reador was employed to control the dissolved oxygen (DO)
concentration and pH. During chemostat operation, DO
was maintained at 50% of air saturation by gas transfer
from the headspace. During perfusion reactor operation,
*
DO was maintained at 10% 3% of air saturation by
sparging oxygen periodically into the culture medium as
a supplement to gas overlay in the headspace. The pH
*
was maintained a t 7.20 0.01 by regulating the flow of
COz into the reactor headspace and by adding 0.5 M
sodium bicarbonate.
Two inclined sedimentation channels were used in this
study. Each was made of glass and had the same rectan-
gular dimensions of 5 cm in width and a 0.5-cm separation
between the two inclined surfaces. One sedimentation
channel had a length of 37 em, while the other was 23 cm
long. From eq 1, the clarification rates for the average-
sized nonviable and viable hybridoma cells are predicted
to he S = 100 and 270 mL/h, respectively, for the longer
settler and S = 64 and 170 mL/h, respectively, for the
shorter settler. Each settler was tapered in width at the
lower end to facilitate the return of all settling cells to the
bioreactor. Each settler was also equipped with a water
jacket through which 37 OC water was circulated during
the experiment in order to maintain the cells in the settler
a t the same temperature as in the reactor. To minimize
the adherence of settling cells to the glass surfaces of the
channel, each settler was siliconized with dimethyldi-
chlorosilane prior to use.
The bioreactor configuration for this study is shown in
Figure 1. The external inclined sedimentation channel WBS
kept a t a constant angle of 30° from the vertical, in order
to provide sufficient area for sedimentation while allowing
the sediment to easily slide down the inclined wall. Flow
through the inclined settler was generated by a peristaltic
pump a t the settler outlet. Cells that settle completely
from suspension are returned to the bioreactor by gravity
flow of the sediment layer. Smaller cells that do not have
460
sufficient time to be removed from suspension are washed
out in the settler overflow stream.
Two different studies were conducted, and the operation
of the bioreactor was dependent on the particular study.
The first experiment was to assess the separation of non-
viable from viable hybridomas in the inclined settler as
a function of the overflow rate from the settler. It was
essential that the cells in the reactor be maintained at a
constant growth r a t e t o ensure a uniform cell size
distribution for comparison purposes as the overflow rate
was varied. Therefore, the reactor was operated as a
conventional chemostat with the settler overflow returned
to the reactor via a separate port to comprise an internal
recycle loop for maintenance of a constant culture volume
(Figure 1). The settler was allowed to operate a t each
overflow rate studied for a minimum of 3 h before sampling
to assure steady-state conditions. A range of overflow rates
was chosen to span the clarification rates predicted by eq
1for the nonviable and viable cells. Two different dilution
rates were used: 0.39 day-' (for the 37-cm-long settler) and
0.89 day-' (for the 23-cm-long settler). T h e second
experiment was to operate the system as a perfusion
culture. Instead of recycling the settler overflow stream
as shown in Figure 1, the overflow stream became the
reactor effluent. The feed rate was adjusted to exceed the
overflow rate slightly, with the effluent tube employed in
chemostat operation used as a level control to maintain
constant culture volume.
Sample Analyses. Cell concentrations were determined
by counting in a hemacytometer, and viability fractions
were determined by trypan blue staining (26). During the
study of cell separation versus settler overflow rate, samples
were taken from both the reactor and overflow stream at
each overflow rate. A t least two samples were taken from
the overflow stream a t each overflow rate so that the
reported concentrations of both viable and nonviable cells
are the average of two cell counts. During the perfusion
culture, samples were taken initially twice a day, and then
once a day after the dilution rate was set at its final value,
from both the reactor and settler overflow stream. After
the cells were counted, the perfusion reactor sample was
centrifuged and the supernatant was analyzed for antibody
concentration by using a sandwich ELISA procedure in
which alkaline phosphatase was used as the conjugated
enzyme and the absorbance was read at 405 nm (27). The
standard error for each reported antibody concentration
was propagated from the point of half-maximal absor-
bance of the dilution curve of both the standard antibody
and sample.
Determination of Cell Size Distributions. Cell size
distributions and mean cell diameters were determined by
using an Elzone 180XY particle size analyzer (Particle
Data, Inc.). Reactor and settler overflow samples were
diluted to a concentration of 3000-5000 cells/mL in
phosphate-buffered saline (PBS) and then counted in the
size analyzer with a 120-pm orifice.
Size and Viability Measurements with Flow Cy-
tometry. The distributions of viable and nonviable hy-
bridomas in reactor and overflow samples from the per-
fusion culture were determined by using an EPICS 541
flow cytometer with a Coherent argon ion laser. Samples
were centrifuged and resuspended in cold PBS containing
0.5 pg of propidium iodide/mL and incubated for 30 min
(28). Samples were then filtered through 62-pm nylon
mesh just prior to analysis. Propidium iodide does not
enter the viable cells but passes readily into the nonvia-
ble cells and stains their DNA. The dye in the nonvia-
ble cells is excited by the laser at 488 nm and fluoresces
Biotechnol. Prog., 1990, Vol. 6, No. 6
?; 1.0,
0 viable cells
0 nonviable cells 0
D I 0.59 day-' d o
0
1 0.64 0
p 0.5 - 0
0.4 - 0
- 0' 0
2
0
0.3 - 0
:0.2 -
0 .
'0 0.1 -I 0 I
0 100 200 300 400 500 600 700
overflow rote from settler (ml/hr)
Figure 2. Effect of settler overflow rate on the concentrations
of viable and nonviable cells in the overflow stream relative to
those in the reactor at a chemostat dilution rate of 0.39 day1.
An inclined settler 37 cm in length at an angle of inclination of
30" was used in the study.
at 540 nm. Two discrete cell populations (viable and non-
viable) were resolved in a two-dimensional forward-
angle light scatter versus log integrated red fluorescence
histogram for the reactor sample, and contour maps were
drawn around the viable and nonviable populations in
order to determine the separate distributions for the two
cell populations in both the reactor and overflow samples.
Results and Discussion
Effect of Overflow Rate on Cell Separation. The
effect of the overflow rate from the inclined sedimentation
channel on the removal of nonviable hybridomas from, and
the retention of viable hybridomas within, a suspension
chemostat culture was assessed by determining the
concentration of each in the overflow stream over a range
of overflow rates. Because cell size is dependent on specific
growth rate (12),the cell separation achievable by inclined
sedimentation was determined at two dilution rates (in
separate cultures). The two dilution rates chosen, 0.39 and
0.89 day-', provided specific growth rates of 0.65 and 0.97
day-', respectively, that represent low and high values
relative to the maximum specific growth rate of 1.10 day-l
determined for these cells in batch culture. At a low growth
rate there is less difference in size between viable and non-
viable cells, which implies that a cell separation is more
difficult because there is a smaller difference in the Stokes
settling velocity of each cell type. The converse is expected
to be true a t a high growth rate, for which the size
difference between viable and nonviable hybridomas is
greater.
At a dilution rate of 0.39 day-', the total cell concen-
tration in the reactor was 3.9 X 106 cells/mL, and the viable
fraction was 0.60, when the inclined settling experiments
were initiated. The concentrations of viable and nonvi-
able hybridomas in the settler overflow stream from the
longer settler, relative to those in the feed stream from the
bioreactor, are shown in Figure 2 over a range of overflow
rates from 16 to 603 mL/ h. All cells were retained in the
settler for overflow rates below 55 mL/h, and only non-
viable cells were detected in the overflow for overflow rates
between 55 and 107 mL/h. From eqs 1 and 2, this range
corresponds to diameters between 6 and 8 pm for the
largest cells predicted to reach the overflow. As the
overflow rate was increased further, the relative concen-
tration of both cell types in the settler effluent increased,
with that for the nonviable cells generally being higher
Biotechnol. Prog., 1990, Vol. 6, No. 6 461

12 1 .o --b 1.0 I I
:f: -
0 viable cells
0 0 - 0.9 0.9
t
h
11 1
0 0 -c nonviable cells
* . - 0.8 #j 0.8 -
E : c
a .
e
.o a - 0.7 - 0.7 -
D = 0.89 day -'
- -
E
10-
- C
20 ,o 0.6
.
- a .

:I,
V
-0.6
t - V
f T 0.5 2
:
9-
a

1
6
, :, , ;, , , ,
T

,
0 cells in reactor

T
, ,
,a , cells in
~, , ,
overflow
, ,

viable fraction in overflow

v viable fraction in reoctor
, , , , ,j:_
10.4

0.1

0.0
.5

-
-
e 0.0 ,: ,?.0,4
o
. .,.
,
o o
1 ,
0

., . , , ., , , , , , .
> , I

0 100 200 300 400 500 600 700 0 50 100 150 200 250 300
overflow r o l e f r o m settler (ml/hr) overflow rote from settler ( m l / h r )

Figure 3. .Mean cell diameter (circles) and viable fraction

(triangles)in reactor and overflow stream samples over a range Figure 4. Effect of settler overflow rate on the concentrations
of settler overflow rates. The chemostat dilution rate was 0.39 of viable and nonviable cells in the overflow stream relative to
day-1 and the inclined settler was 37 cm long at an angle of those in the reactor at a chemostat dilution rate of 0.89 day1.
inclination of 30° throughout the study. An inclined settler 23 cm in length at an angle of inclination of
30' was used in the study.
because they settle more slowly than the viable cells. These In the second culture, the chemostat was operated at
results are as expected because, as the overflow rate is a dilution rate of 0.89 day-' and had a total cell concen-
increased, cells in suspension have a shorter residence time tration of 2.5 x 106 cells/mL with a viable fraction of 0.92.
in the settler, and thus a smaller fraction has sufficient The relative concentrations of viable and nonviable hy-
time to settle out of suspension. A t very high overflow bridomas in the overflow stream as a function of the
rates, the residence time is small and the concentrations overflow rate are plotted in Figure 4. The shorter settler
of both subpopulations in the settler overflow are seen in (23 cm long) was used in order to reduce the required
Figure 2 to approach their respective concentrations in the throughput of culture medium and also because the viable
bioreactor. These results show t h a t operation a t an cell sizes, and hence settling velocities, were larger for the
intermediate overflow rate provides the best separation higher dilution rate. Hybridomas began to wash out of
of nonviable cells from the culture. the culture at an overflow rate of 31 mL/h. As the overflow
The mean cell diameter was calculated from the cell size rate was increased, the cell concentration in the settler
distribution generated by the particle size analyzer for effluent increased also, as was observed in the previous
samples from both the reactor and overflow stream at four experiments. At overflow rates less than 123 mL/h, the
overflow rates. These results along with the viable cell nonviable cell concentration in the overflow stream was
fraction in each overflow and reactor sample are plotted greater than that for viable cells even though the viable
in Figure 3. The 99 76 confidence interval for each mean cell concentration in the reactor was nearly 12 times greater.
was less than f0.06 pm because the number of cells counted The nonviable cells were being removed preferentially while
in each sample was greater than 10 000. The results show viable cells were retained by the inclined settler a t these
that chemostat operation maintained the cells in the reactor overflow rates. From Figure 4, it may be concluded that
a t a relatively constant mean cell diameter of 11.2-11.3 operation at intermediate overflow rates between about
pm that was not disrupted by the recycle loop through the 70 and 120 mL/h gives very good separation in which the
inclined settler. However, the mean diameter and viable settler retains more than 95% of the viable cells while
fraction of cells removed in the settler overflow increased removing through its overflow more than 50% of the non-
with increasing overflow rate. Because the Stokes settling viable cells. From eqs 1 and 2, this range corresponds to
velocity of a cell is determined primarily by its diameter, diameters between 8 and 11 pm for the largest cells
only the smallest cells in the culture failed to settle predicted to reach the overflow. This separation is better
completely during their residence time in the inclined than that achieved in the previous experiment at the lower
settler at low overflow rates. As seen in Figure 3 for an dilution rate (see Figure 2 for comparison) because the
overflow rate of 107 mL/h, these small cells are primarily viable cells were larger, on average, at the higher dilution
nonviable. The reduction in residence time caused by rate and were therefore more easily separated from the
increasing the overflow rate allowed larger, viable cells to smaller nonviable cells by differential sedimentation.
be washed out so that, at the highest overflow rate studied, The results a t the two different dilution rates demon-
the mean cell diameter and the viable fraction in the strate that selective removal of nonviable hybridomas from
overflow were nearly the same as those in the reactor. a mixed culture is achievable with inclined sedimentation.
In addition to the increase in cell size in the settler The cell separation is affected by varying the overflow rate
effluent with overflow rate, the viable fraction in the through the inclined settler, which controls the residence
overflow stream increased as well. The viable fraction in time of suspended cells in the settling channel. The degree
the settler effluent remained less than that in the reactor of nonviable cell removal is constrained by the difference
at all but the highest overflow rate studied. The viable in size distribution between viable and nonviable cells.
fraction in the reactor decreased slightly as the overflow Therefore, it is desirable to maintain the cells at a high
rate increased, presumably because larger viable cells were specific growth rate, which maximizes the size of the viable
increasingly retained as sediment in the settler and cells.
returned more slowly to the reactor compared to nonvi- Having demonstrated t h a t nonviable cells can be
able cells, which were returned quickly to the reactor in preferentially removed even at a relatively low growth rate
the recycled overflow. is significant for applications to perfusion culture. The
462 Biotechnol. Prog., 1990,Vol. 6,No. 6

0 viable cells

.. I
0 viable cells * *

C
0
0
0
o o oo
0 0 0
L

e
1 1 0.1
j .
8

o j
e I begin
, A ,
perfusion

I , , I , ,
-
..
0

-
0
.
.
bepin
.*
p.rf"rlon
.
0 2 4 6 8 10 12

time (day)

Figure 5. Viable and total hybridoma concentrations in the

reactor during the perfusion culture. Perfusion was initiated on
day one at a dilution rate of 0.89 day-', which was increased
incrementallyto 1.65 day1 over the next 2'12 days. The dilution
rate was increased to 1.70 day-' on day 5 and stayed there for
the remainder of the study.
perfusion rate may be high relative to the washout dilution
rate, but the growth rate could still be submaximal due
to increased utilization of substrate caused by the greater
viable cell concentration reached in a perfusion culture.
Since monoclonal antibody production is non-growth-
associated, however, it is desirable to maximize the viable
cell concentration rather than the growth rate. The next
phase of the study was to demonstrate the utility of
inclined sedimentation in perfusion culture to promote a
high viable cell concentration and monoclonal antibody
productivity by selectively removing nonviable cells.
Perfusion Culture with Inclined Sedimentation.
The perfusion culture was initiated approximately 3 days
following the previous cell separation study to ensure a
stable baseline in the cell concentration. During the
interim, in which the bioreactor was operated as a chemo-
stat with a dilution rate of 0.89 day-l, the total cell
concentration reached 3.1 X lo6cells/mL, the viability was
92%,and the antibody concentration was 10.4 pg/mL. A t
the onset of perfusion, the dilution rate was maintained
at 0.89 day-1 because this corresponded to an overflow rate
of 37.8 mL/h, which in the previous study was found to
retain practically all viable cells in the culture. The
concentrations of total and viable hybridomas in the biore-
actor are plotted over the duration of the perfusion culture
in Figure 5, whereas Figure 6 shows the relative viable and
nonviable cell concentrations in the settler effluent stream.
At this low dilution rate, 100% of the viable cells, and
approximately 95 7; of the nonviable cells, were returned
to the bioreactor by the settler. Consequently, both the
viable and nonviable cell concentrations increased in the
bioreactor. Since no viable cells were observed in the
effluent, the dilution rate was increased to 1.10 day-l
(corresponding to an overflow rate of 46.7 mL/h) after
almost 2 days in order to increase the number of nonvi-
able cells removed. Over the next 2'/2 days, the dilution
rate was increased incrementally to 1.65 day-'. The non-
viable cell concentration in each overflow sample progres-
sively increased with increasing dilution rate so that 20 %
of the nonviable cells entering the settler were removed
through the overflow a t the end of this period. In contrast,
at most one viable cell was counted in any of these samples,
indicating t h a t the viable cells were being retained
preferentially in the culture.
On the fifth day, the dilution rate was raised to 1.70 day-'
(corresponding to a settler overflow rate of 72.2 mL/h and
Biotechnol. Prog., 1990,Vol. 6,No. 6
0
1
x
I I
NONVIABLE CELLS
I
I I
i l
W
VIABLE CELLS
Figure 7. Forward-angle light scatter histograms of the viable
and nonviable cell populations in the bioreactor (solid lines) and
the nonviable cell population in the settler overflow stream
(dashed line) on day 8 of the perfusion culture at a dilution rate
of 1.70 dayl. There were so few viable cells in the overflow sample
that the histogram corresponds to the bottom axis line. The peak
in the reactor viable cell histogram is on the right in a higher-
numbered channel than the peak in the reactor nonviable cell
histogram at left. The peak in the overflow nonviable cell
histogram is slightly to the left of that for the reactor nonvia-
ble cell population.
Cell Size a n d Viability i n Perfusion C u l t u r e by
Flow Cytometry. The samples for analysis by flow cy-
tometry were taken on the eighth day of the perfusion
culture, after the dilution rate was set a t 1.70 day-l. Data
were collected by gating around the separate viable and
nonviable cell populations to permit later comparison of
these distinct cell populations. Figure 7 shows the forward-
angle light scatter (FALS) histograms of the viable and
nonviable cell populations in the reactor sample (solid
lines). The extent of forward-angle light scatter from a
particle or cell is correlated to the square of its diameter
such that the larger the particle, the higher the channel
number in which it falls in the forward-angle light scatter
measurement (29). Determining the size distribution in
this way is not as quantitatively accurate as the Elzone
180 XY particle size analyzer, b u t it does provide
qualitative cell size distributions of the two separate sub-
populations in a mixture of viable and nonviable cells. As
seen in Figure 7, the viable cells in the perfusion culture
were typically larger than the nonviable cells, although
there was some overlap in cell size between the two
populations. Thus, complete removal of nonviable cells
by t h e inclined settler is not attainable without a
considerable loss of viable cells, which is consistent with
the results in Figures 2 and 4. However, it is possible to
obtain partial removal of the nonviable cells while retaining
virtually all of the viable cells, provided that the overflow
rate is chosen properly. This was the case for t h e
conditions of Figure 7. There were very few viable cells
in the overflow sample, and the viable cell histogram is
nearly identical with the ordinate axis on this scale. This
result concurs with the cell count data and provides further
evidence that a negligible number of viable cells were lost
in the settler effluent. The FALS histogram of the non-
viable cell population in the overflow sample is shown as
the dashed line in Figure 7. The overflow sample was also
analyzed in the Elzone 180 XY particle size analyzer to
determine the nonviable cell size distribution. The mean
cell diameter was calculated to be 7.9 pm, which is in
contrast to the mean diameter of 11 pm for the mixture
of 50% viable and 50% nonviable cells in the bioreactor.
A comparison of the FALS histograms of the nonviable
cell populations in the reactor and settler overflow stream
shows that the two populations are similar, as expected.
The peak in the overflow distribution is shifted slightly
to the left of the peak for the reactor distribution because
viable cell retention in the culture necessitated the use of
483
40 10
I 0
5 1 , A ,A . , , , , !2 n
"
0 2 4 6 0 10 12
time (day)
Figure 8. The antibody concentration (open triangles) and
specific antibody productivity (solid triangles) in the reactor
during the perfusion culture. The error bars show the standard
error for each antibody concentration.
a perfusion rate that allowed retention of larger nonvia-
ble cells that were similar in size to the smaller viable cells.
Antibody Concentration in Perfusion Culture. The
antibody concentration in the reactor during the perfu-
sion culture is shown in Figure 8. The antibody concen-
tration increased from 10.4 pg/mL in the chemostat to 30.4
pg/mL on the final day of the perfusion culture. The
standard error associated with each concentration
determination (shown as error bars) varied between 6.5 %
and 20%. This result demonstrates that the reactor
antibody concentration can be enhanced nearly 3-fold in
this perfusion bioreactor over a chemostat because of the
greater viable cell concentration achieved by selective
retention.
The increase in antibody concentration during the per-
fusion culture reflects not only the increased viable cell
concentration but also an increase in specific antibody
productivity per viable cell as well. The variation in
specific antibody productivity also is plotted in Figure 8.
The specific antibody productivity nearly doubled over the
duration of the perfusion culture. This observed increase
in specific antibody productivity over time is consistent
with an earlier experimental finding that the antibody
productivity of this cell line decreases as the specific growth
rate increases (30). The total antibody productivity per
culture volume increased from 8.6 pg/(mL.day) in the
chemostat to 52 pg/(mL.day) at the time that the perfu-
sion culture was terminated.
Conclusions
The results of this study clearly show that selective
retention of viable cells and removal of nonviable cells from
a suspension hybridoma culture is practical with inclined
sedimentation. For a given cell culture growing a t a
particular rate, the degree of cell separation was found to
depend on the overflow rate through the settler, which
determines the residence time of suspended cells in the
sedimentation channel. The cell separation was also found
to be affected by the available area for sedimentation, since
the overflow rate at which hybridomas were first washed
out was lower with the 23-cm-long settler than with the
37-cm-long settler. The specific growth rate of the culture
also determined the degree of cell separation. T h e
enhanced viable cell retention a t higher specific growth
rate was due to the associated increase in size of the viable
cells relative to the generally smaller diameter of the non-
viable cells.
464
Perfusion culture results showed that the inclined settler
performed very well as a viable cell retention device, as
the maximum viable cell density reached in this study was
comparable to those found by other investigators (11-
15). However, inclined sedimentation was shown to be
advantageous over total cell retention devices in that
approximately 20 % of the nonviable cell concentration
entering the inclined settler was continuouslyremoved from
the culture, while less than 0.1% of the viable cell
concentration was lost. Thus, continuous removal of non-
viable cells from the culture by inclined sedimentation
allows the possibility of o erating a perfusion reactor a t
steady state. This woulcf eliminate interuption of the
culture to purge the accumulated cells necessitated by
clogging of filters used to retain cells (12) and would
possibly prevent the decrease in antibody productivity
observed in conventional perfusion cultures of long
duration (14). Because selective removal of nonviable cells
was shown to be possible a t low as well as high specific
growth rates, erfusion cultures using inclined sedimen-
tation could Yl e operated a t low growth rates, where
antibody productivity has been found to be greater (30).
Acknowledgment
We acknowledge the support of Grants NBSRAHS-
OH130 from the Department of Commerce and BCS-
8857719 from the National Science Foundation.
Literature Cited
(1) Fazekas de St. Groth, S. Automated Production of Mono-
clonal Antibodies in a Cytostat. J. Immunol. Methods 1983,
57,21-36.
(2) Boraston, R.; Thompson, P.; Garland, S.; Birch, J. Growth
and Oxygen Requirements of Antibody Producing Mouse Hy-
bridoma Cells in Suspension Culture. Deu. Bid. Stand. 1984,
55,103-111.
(3) Reuveny, S.; Velez, D.; Riske, F.; Macmillan, J.; Miller, J.
Production of Monoclonal Antibodies in Culture. Deu. Biol.
Stand. 1985, 60, 185-197.
(4) Birch, J.; Thompson, P.;Lambert, K.; Boraston, R. The Large
Scale Cultivation of Hybridoma Cells Producing Mono-
clonal Antibodies. In Large-Scale Mammalian Cell Culture;
Tolbert, W., Feder, J., Eds.; Academic Press: New York, 1985;
pp 1-16.
(5) Luan, Y.; Mutharasan, R.; Magee, W. Strategies to Extend
Longevity of Hybridomas in Culture and Promote Yield of
Monoclonal Antibodies. Biotechnol. Lett. 1987,9 (lo), 691-
696.
(6) Renard, J. M.; Spagnoli, R.; Mazier, C.; Salles, M. F.; Man-
dine, E. Evidence that Monoclonal Antibody Production
Kinetics is Related to the Integral of the Viable Cells Curve
in Batch Systems. Biotechnol. Lett. 1988, 10 (2), 91-96.
(7) Reddy, S.; Miller, W. M. Effects of Environmental Stress on
Hybridoma Antibody Production and Metabolism. AIChE
Annual Meeting, November 1989.
(8) van Wezel, A,; van der Veldon-de Groot, C.; de Haan, H.;
van den Heuvel, N.; Schasfoort, R. Large Scale Animal Cell
Cultivation for Production of Cellular Biologicals. Deu. Biol.
Stand. 1985,60,229-236.
(9) Reuveny, S.;Velez, D.; Macmillan, J.; Miller, L. Comparison
of Cell Propagation Methods for their Effect on Monoclonal
Antibody Yield in Fermentors. J. Immunol. Methods 1986,
86,61-69.
(10) Tolbert, W.; Lewis, C.; White, P.; Feder, J. Perfusion Culture
Systems for Production of Mammalian Cell Biomolecules. In
Large-Scale Mammalian Cell Culture; Tolbert, W., Feder, J.,
Eds.; Academic Press: New York, 1-98; pp 97-119.
(11) Brennan, A,; Shevitz, J.;Macmillan, J. A Perfusion System
for Antibody Production by Shear-Sensitive Hybridoma Cells
in a Stirred Reactor. Biotechnol. Tech. 1987, 1, (3), 169-
174.
Biotechnol. Prog., 1990, Vol. 6,No. 6
(12) Goebel, N. K.; Kuehn, R.; Flickinger, M. C. Methods for
Determination of Growth-Rate-Dependent Changes in Hy-
bridoma Volume, Shape, and Surface Structure During
Continuous Recycle. Cytotechnology, in press.
(13) Sato, S.; Kawamura, K.; Hanai, N.; Fujiyoshi, Production
of Interferon and Monoclonal Antibody Using a Novel Type
of Perfusion Vessel. In Proceedings o f the International
Symposium on Growth and Differentiation of Cells in Defined
Environment;Murakami, H., Yamane, I., Barnes, D., Mather,
J., Hayashi, I., Sato, G., Eds.; Springer-Verlag: New York, 1085;
pp 123-127.
(14) Kitano, K.; Shintani, Y.; Ichimori, Y.; Tsukamoto, K.; Sa-
sai, S.; Kida, M. Production of Human Monoclonal Antibodies
by Heterohybridomas. Appl. Microbiol.Biotechnol. 1986,24,
282-286.
(15) Takazawa, Y.; Tokashiki, M.; Hamamoto, K.; Murakami,
H. High Cell Density Perfusion Culture of Hybridoma Cells
Recycling High Molecular Weight Components. Cytotech-
nology 1988, 1, 171-178.
(16) Tovey, M.; Brouty-Boy6, D. Characteristics of the Chemo-
stat Culture of Murine Leukemia L1210 Cells. Exp. Cell Res.
1976,101, 346-354.
(17) Ponder, E. On Sedimentation and Reouleaux Formation.
Q. J . Exp. Physiol. 1925, 15, 235-253.
(18) Nakamura, N.; Kuroda, K. La Cause de l’hcceleration de
la Vitesse de Sedimentation des Suspensions sans les Recipients
Inclines. Keijo J. Med. 1937, 8 , 256-296.
(19) Davison, B. H.; San, K.-Y.; Stephanopoulos, G. Stable
Competitive Coexistence in a Continuous Fementor with Size-
Selective Properties. Biotechnol. Prog. 1985,l (4), 260-269.
(20) Davis, R. H.; Parnham, C. S. Competitive Yeast Fermen-
tation with Selective Flocculation and Recycle. Biotechnol.
Bioeng. 1989, 33, 767-776.
(21) Henry, K. L.; Davis, R. H. Continuous Recombinant
Bacterial Fermentation Utilizing Selective Flocculation and
Recycle. Biotechnol. Prog. 1990, 6, (l),7-12.
(22) Lee, C. Y.; Davis, R. H.; Sclafani, R. A. Cell Separations
of Nondividing and Dividing Yeast Using an Inclined Settler.
In Proceedings of the 19th Annual Biochemical Engineering
Symposium; Bajpai, R., Ed.; University of Missouri: Columbia,
MO, 1989; pp 123-132.
(23) Hornbeck, P. V.; Lewis, G. K. Idiotype Connectance in the
Immune System 11. A Heavy Chain Variable Region Idio-
type that Dominates the Antibody Response to the p-Azoben-
zenearsonate group is a Minor Idiotype in the Response to Tri-
nitrophenyl Group. J. Exp. Med. 1985, 161, 53-71.
(24) de Bruyne, N. A. The Design of Bench-Scale Reactors. Anim.
Cell Biotechnol. 1988, 3, 141-176.
(25) Hu, W. S.; Meier, J.; Wang, D. I. C. Use of Surface Aerator
to Improve Oxygen Transfer in Cell Culture. Biotechnol.
Bioeng. 1986,28, 122-125.
(26) Phillips, H. J. Dye Exclusion Tests for Cell Viability. In
Tissue Culture Methods and Applications; Kruse, P. F., Patter-
son, M. K., Eds.; Academic Press: New York, 1973; pp 406-
408.
(27) Catty, D.; Raykundalia, C. ELISA and Related Enzyme Im-
munoassays. In Antibodies: A Practical Approach; Catty, D.,
Ed.; IRL Press: Oxford, England, 1988; Vol. 11, pp 97-154.
(28) Sasaki, D. T.; Dumas, S. E.; Engleman, E. G. Discrimination
of Viable and Non-Viable Cells Using Propidium Iodide in Two
Color Immunofluorescence. Cytometry 1987,8, 413-420.
(29) Shapiro, H. M. Practical Flow Cytometry, 2nd ed.; Alan
R. Liss, Inc.: New York, 1988.
(30) Miller, W. M.; Blanch, H. W.; Wilke, C. R. A Kinetic Analysis
of Hybridoma Growth and Metabolism in Batch and Contin-
uous Suspension Culture: Effect of Nutrient Concentration,
Dilution Rate, and pH. Biotechnol. Bioeng. 1988, 32, 947-
965.
Accepted October 2, 1990.