Biotechnol. Prog. 1990, 6, 114-120
114
Shear Sensitivity of Hybridoma Cells in Batch, Fed-Batch, and Continuous Cultures Jonathan F. Petersen and Larry V. McIntire Department of Chemical Engineering, Rice University, P.O. Box 1892, Houston, Texas 77251-1892
Eleftherios Terry Papoutsakis* Department of Chemical Engineering, Northwestern University, Evanston, Illinois 60208
Previously, we observed that CRL-8018 hybridoma cells were more sensitive to welldefined viscometric shear during the lag and stationary phases than during the exponential phase of batch cultures. Some potential hypotheses for explaining the increase in shear sensitivity are (1)nutrient limitations that result in a decrease in production of specific cellular components responsible for the mechanical strength of the cell, (2) nutrient limitations that lead t o synchronization of the culture in a cell cycle phase that is more sensitive to shear, or (3) a link between cell growth and shear sensitivity, such t h a t slowly growing cells are more sensitive to shear. Here, the duration of the exponential phase was increased with use of fed-batch, and the effect on shear sensitivity of the cultures was measured with a viscometric technique. Extension of exponential growth resulted in a n increased period during which the cells were insensitive t o shear. Additionally, the shear sensitivity of the cells was constant over a wide range of growth rates and metabolic yields in chemostat cultures. These observations suggest t h a t as long as the cells are actively (exponentially) growing, their shear sensitivity does not depend on the growth rate or metabolic state of the cell as expressed by metabolic yields. Thus, hypothesis 3 above can be dismissed.
Introduction The large-scale processing of mammalian cells and their products has been given much attention in recent years. Many of the techniques that have been proposed for culture scale-up will unavoidably subject the cells to mechanical stresses. Two familiar examples of this are the growth of cells in stirred and air-lift fermenters. A variety of responses by cells to mechanical stresses have been reported. Moderate stress levels have caused increased production or secretion of urokinase (Stathopoulos and Hellums, 1985), tissue plasminogen activator (Diamond et al., 1989),and prostacyclin (Frangos et al., 1985; Frangos et al., 1988). Higher levels of mechanical stresses may result in damage to the cells. Damaging stresses are most likely a result of fluid stresses caused by agitation (Nelson, 1988; Tolbert et al., 1982),the motion and/or bursting of gas bubbles in the reactor (Handa et al., 1987; HandaCorrigan et al., 1989; Tramper et al., 1988; Tramper and Vlak, 1988), or a combination of these mechanisms. It is generally accepted that an understanding of how cultured mammalian cells are damaged by mechanical forces will be beneficial to the scale-up of these cultures. Shear stresses occur in stirred cultures primarily as a result of agitation, and their effects have been studied independent of other damage mechanisms with use of viscometers or other devices that generate well-defined shear fields. The viability of Spodoptera frugiperda insect cells was decreased by 15-550 dyn.cm-2 shear in a viscometer (Tramper et al., 1986). Exposure to turbulent
* Corresponding
author.
8756-7938/90/3006-0114$02.50/0
capillary flow at a wall shear stress of 1800 d p c m - 2 caused lysis in cultured myeloma cells (McQueen et al., 1987). The specific cell death rate of a hybridoma line increased with shear stresses of 0-160 dyn.cm-2 in a viscometer (Schurch et al., 1988). Though progress has been made in understanding cell damage mechanisms, the factors that affect a culture’s sensitivity to mechanical damage are not well understood. In previous work, we observed that the shear sensitivity of cells may be modulated by a number of factors, including the agitation history of the cells, the concentration of specific metabolites, and the age of the cells in batch culture (Petersen et al., 1988). The effect of culture age was particularly interesting. In batch cultures, the cells were more sensitive to shear both early and late in the culture and much less sensitive a t intermediate times. This observation is consistent with the increased sensitivity to agitation late in batch cultures reported later by Lee et al. (1988). The sensitivity early and late in the culture corresponded approximately with the lag and stationary phases, respectively. We propose to explain this behavior by a number of different hypotheses. One hypothesis is that during stationary phase, one or more nutrients may be present in limiting concentrations. This limitation may have a detrimental effect on cellular structures that mediate shear sensitivity. For instance, a limiting supply of an essential fatty acid may result in a less robust cell membrane. This limitation would be removed when the cells were supplied with fresh nutrients during subculture, but it would take some time for the cells to assimilate the nutrients into new cell structures. As a result, the shear sen-
@ 1990 American Chemical Society and American Institute of Chemical Engineers
115
Biotechnol. Prog., 1990, Vol. 6, No. 2
sitivity of the cells would remain higher during the next lag phase. A second hypothesis to explain the increased shear sensitivity due to nutrient limitation is based on the possibility that cells in a particular phase in the cell cycle may be more shear sensitive than other cells. For instance, mitotic cells may be more sensitive to shear than cells in S phase. Nutrient limitations late in the culture can cause synchronization of the cells (Ashihara and Baserga, 1979), leading to changes in shear sensitivity during different stages of the culture. Another possible explanation for the variable shear sensitivity is that this sensitivity is a growth-related phenomenon. Cells that are growing slowly or not a t all may be more sensitive to shear than fast-growing cells. Culture growth is slowest during stationary and lag phases, when shear sensitivity was greatest. In this paper, we examine the effects of culture age on shear sensitivity in more detail to investigate whether the last hypothesis is indeed true. Specifically, we employed batch, fed-batch, and continuous cultures to change the duration of exponential growth and the specific growth rate of cultures and then used a viscometric technique to determine how the shear sensitivity changed. We also examined whether the shear sensitivity of the cells depends on the metabolic state of the cells as characterized by a combination of the growth rate and the various metabolic yields.
Materials and Methods Cells and Medium. CRL-8018 hybridoma cells (ATCC) were used in all experiments. This hybridoma line produces an IgM monoclonal antibody to hepatitis B surface antigen (HBsAg). T h e experiments were performed with cells grown in a serum-containing growth medium (CDME). CDME consisted of Dulbecco's modified eagle's medium (DME; Sigma Chemical Co., St. Louis, MO) supplemented with 1% (v/v) fetal bovine serum (FBS; Hyclone Lot No. 1111533, Hyclone Laboratories, Logan, UT), 1% (v/v) Nutridoma-NS (Boehringer Mannheim Biochemicals, 100115, Indianapolis, IN), 3.7 g/L NaHCO,, and 50 units/mL penicillin, 0.05 mg/mL streptomycin, and 0.1 mg/mL neomycin (PSN; Sigma). The medium was adjusted to pH 7.2 by addition of 1 N HC1 or NaOH and then sterilized by filtration through 0.2pm filters before use and stored at 4 "C. The cell cultures were routinely maintained in 75-cm2 tissue culture flasks (T-flasks, Corning 25110, Corning Glass Works, Corning, NY). The cultures were grown in an incubator a t 37 "C in an atmosphere of 9% CO, and 95% relative humidity. Shear Sensitivity of Cells. In this study, the shear sensitivity of cultured cells was determined in the welldefined shear field of a couette viscometer. Detailed descriptions of the technique have been given earlier (Petersen et al., 1988; Petersen, 1989). Briefly, 5 mL of cells suspension from a culture of interest was injected into the viscometer. The motor speed controller was set to 340 rpm, which produces a shear rate of 5000 s-'. The viscosity of the culture medium is 1 g-cm-h-' a t 25 "C, which results in a shear stress of 50 dyn-cm-, in the sample. After the sample had been sheared for 10 min, the viscometer motor was stopped. The cell sample was then removed from the viscometer and assayed for cell damage. Assays for Cell Damage. Culture viability and cell lysis were used as measures of cell damage. Viable cell density was determined by trypan blue exclusion (Phillips, 1973). Basically, the cell suspension was diluted with
0.1% try an blue solution to bring the cell density to about 10F/mL. The number of viable cells and nonviable cells were determined by counting in a hemacytometer (Absher, 1973). At least 200 cells were counted in each determination to reduce statistical errors to below 15%. All cell counts were made immediately upon sampling, except for sheared samples, which were counted 20 min after shearing. Viability data from shear experiments were expressed as normalized viability (NV). This is defined as number of viable cells after shearing number of viable cells before shearing (1) This normalization results in a number between 0.0 and 1.0. I t is useful for comparing the shear sensitivity of cultures with different numbers of cells. The extent of cell lysis was determined by measuring lactate dehydrogenase (LDH) release from the cells. LDH was quantitatively determined with a commercially available assay kit (Sigma UV-340). For shear experiments, LDH results were expressed as fractional lysis (FL): NV =
FL =
LDH, - LDHo LDHl- LDHo
(2)
where LDH,, LDH,, and LDH, are the LDH activities measured in sheared, control, and lysed samples, respectively. FL should have a value of 0.0 when no cells were lysed by shearing and a value of 1.0 when all of the cells were lysed. LDH, was determined by measuring LDH activity in a sample in which all of the cells had been disrupted. This was done by sonicating the sample for 5 s (Artek Systems Corp., Farmingdale, NY; sonic dismembrator at power level 60 using a needle tip). Microscope examination of the sample confirmed that this method disrupts the entire cell population. LDH activities were measured within 2 h of shearing or sonication. Metabolite Assays. The concentrations of glucose, lactate, ammonia, and glutamine in culture media samples were determined enzymatically with commercially available assay kits (Boehringer Mannheim Biochemicals, 716251, 139084, 542946, and 139092, respectively). Some glucose and lactate determinations were made on a YSI Model 127 analyzer (Yellow Springs Instrument Co., Inc., Yellow Springs, OH). All of the metabolite assays used in the experiments except for the glutamine assay require measurement of optical absorbances at 340 nm. These measurements were performed in disposable acrylic cuvettes (Sarstedt 67.739) with a Gilford System 2600 spectrophotometer (Corning). Bioreactor Operation. All of the batch and fedbatch experiments were run in the Setric Genie 2C bioreactor (Setric Genie Industriel, Toulouse, France). Cells from T-flask cultures were needed and grown in the bioreactor as discussed earlier (Petersen et al., 1988; Petersen, 1989). The temperature of the reactor was automatically controlled at 37 "C by submersion in an air bath. The dissolved oxygen concentration (DO) was maintained at 160% of air saturation by addition of pure oxygen gas to the reactor headspace. Culture pH was maintained automatically by addition of 0.1 M NaHCO, or HC1 to the reactor. The reactor was nominally operated with a working volume of 1 L. The fed-batch experiments required a 4-fold increase in culture volume during each run. A volume of 750 mL was required to keep the reactor impeller and probes submerged, so this established a minimum working volume. The maximum volume, fixed by the vessel size, is 2000 mL. Hence, a 4-
116
Biorechnol. h g . , 1990, voi. 6, NO. 2
fold volume increase could not be achieved directly if the reactor was operated in a strict fed-batch mode. In these experiments, the reactor was started in batch mode with the initial volume set to 750 mL. Once the cells were growing in mid exponential phase, fresh medium was added at a rate of 68 mL/h to start the fed-batch operation. When the culture volume reached 1500 mL, media and cells were removed to reduce the volume to 750 mL, and the feed rate was set to 34 mL/h. The feed stream was finally shut off when the culture volume reached 1500 mL a second time. This scheme was used to simulate a 4-fold volume increase while remaining within the physical constraints of the bioreactor. When the reactor volume and feed rate were reduced by a factor of 2 a t the same time, the equivalent of a constant feed rate was maintained. The agitation rate was 85 rpm. We consider this level of stirring to be sufficiently low so that damage by agitation is negligible. No evidence of cell damage was seen in any of the cultures prior to shearing, as determined by trypan blue dye exclusion and routine microscopic examination. It has been demonstrated that the shear sensitivity of cells grown under these agitation levels is not significantly different from that of cells grown in the absence of agitation (Petersen, 1989). To ensure that the data from the different cultures were comparable, an indicator of the physiological state of the cultures was needed. We found that the culture growth rate is the best available measure for this comparison (data not shown). In these experiments, all cultures were screened, and only the runs exhibiting “normal” growth rates were compared. Normal growth rates here were taken to be those between 0.043 and 0.049 h-l. For chemostat culture experiments, a stream of fresh feed a t constant flow rate was supplied to the reactor by a peristaltic pump. Consistent volume of the culture was assured by using an overflow medium withdrawal system. In this system, the reactor was made essentially leakproof so that gas and liquid could leave only through the overflow tube in the reactor. The level of the overflow tube was set for all of the chemostat runs to make the culture volume 800 mL. The flow rate of the feed stream was then set to give the required dilution (growth) rate. The medium was made in 2-L aliquots and stored in a reservoir that supplied the feed pump. This supply reservoir was kept in an insulated Styrofoam cooler with ice or cold packs to minimize medium deterioration. Each culture was started in batch mode and converted to chemostat mode when the cells were growing exponentially at a density of 106/mL. Steady states were tentatively assumed after at least 2-2.5 residence times, where a residence time is
t = -v= - 1 ‘ F D
(3)
In this study, residence times ranged from 18 to 125 h for dilution rates from 0.008 to 0.055 h-l. A t this point, the culture was sampled three times over a period of 2 days. Part of the sample was frozen for metabolite assays at a later time, and the remaining sample volume was sheared in the viscometer to measure the shear sensitivity of the culture. Occasionally, the cell density was found to vary during this time, indicating that the culture was not a t steady state. In this case, sampling and shearing of the culture were continued on a daily basis until the culture exhibited steady-state behavior, as indicated by constant cell density over 3 days. Only the data obtained from the cultures at steady state were used to characterize the cells.
Results Batch and Fed-Batch Experiments. The time-dependent volume, cell density, and substrate and product concentrations in an ideal fed-batch reactor can be estimated by integration of standard material balances to give
V ( t )= V(0)+ JtF(Y) dy In [ V ( t )X ( t ) ]= In [ V(0)X ( 0 )J
(4)
+pt
(5)
S’(t) V ( t )= S’(0) V(0)- X ( 0 ) V(O)Y,/, + x ( t ) V(t)ys/,y(6) P’(t) V ( t )= P(0)V(0)- X ( 0 ) V(0)YPiX+ X ( t ) V(tW,/, (7) Here, V is the reactor volume; F is the flow rate of culture media into the reactor; X is the cell density; S and P are substrate and product concentrations, respectively; S’(t) = S, - S ( t ) ;P(t)= Pf - P ( t ) ;and the f subscripts indicate feed concentrations. Constant first-order growth kinetics, yields, and feed concentrations have been assumed. Y,,, and Yplx are the amounts of substrate consumed and the amount of product produced by lo6 cells, respectively. The purpose of the fed-batch experiments was to increase the duration of the exponential growth phase. Equation 5 can be recast as
which is a useful form for estimating the length of the exponential phase. For batch cultures, the culture volume is constant, and this equation reduces to
The value o f t for batch experiments can be estimated as follows: the growth rate, p , for these cells is typically 0.045 h-I. The cells were routinely diluted 1:5 with fresh media when subcultured. Assuming that the cells grow to roughly the same final cell density for each culture, we can estimate that X ( t ) / X ( O )= 5. Substituting these values into eq 9 gives an estimate for t of 36 h for cells in batch culture. Assuming that the media can support the same final cell densities in either batch or fed-batch cultures, then the difference in time of the exponential phases between batch and fed-batch cultures is obtained by subtracting eq 9 from eq 8, which gives the result
From the knowledge that p = 0.045 h-’, a 4-fold increase in volume during the culture will increase the duration of the exponential phase by about 31 h. Hence, use of the fed-batch culture will potentially extend the exponential growth periods of the cultures from approximately 36 to 67 h. The cell density, reactor volumes, and metabolite concentrations for typical batch (BATCH 1)and fed-batch (FEDBATCH 1)cultures are shown in Figures 1 and 2, respectively. Reactor volume was constant in the batch culture and is not shown in the plot. The batch culture exhibited typical behavior: an 18-h lag phase, followed by 34 h of exponential growth, and then a stationary phase. Behavior of the fed-batch culture, shown in Figure 2, it also typical: after the initial lag phase (approximately
117
Biotechnol. Prog., 1990,Vol. 6,No. 2
-
--
0
a
E
L 0
c, U
0 0 0 L
0 U
4
c N
U
m
i L
- 0 2
c
m #-I .-I 0)
0
Q)
2
d
Log phase-70h -,
W 0
OO
10
20
40
30
50
60
70
00
EO
Figure 1. Cell densities and metabolite concentrations as a function of time for a batch culture: cell density (m), glucose concentration (A),and lactate concentration (+). 0
E \
m
m
E-
..
'al UE
t3 04 U O N >
20
40
EO
80
100
120
40
60
80
100
120
140
Figure 3. Example of the determination of the length of exponential growth phase during a fed-batch culture. The data from FEDBATCH 2 are used here. The duration of the exponential phase is determined by measuring the length of time during which the curve is linear. In this example, that time is 70 h. Table I. Growth Rates and Yields Calculated from Batch and Fed-Batch Cultures. yields, mg/lOs cells culture type growth rate, h-' glucose lactate BATCH 1 0.043 2.688 1.40 (0.001) (0.224) (0.94) (Figure 1) BATCH 2 0.045 na na FEDBATCH 1 (Figure 2) FEDBATCH 2
OO
20
Time i n c u l t u r e , h
Time, h
140
Time, h
Figure 2. Cell density (m), culture volume (O), glucose concentration ( A ) , and lactate concentration (+) as a function of time for a fed-batch culture. In the experiment, feed was added until the culture volume reached 1500 mL. At this point, half of the culture volume was removed and the feed rate was cut in half. The reactor volume shown is the effective reactor volume that would have been achieved if the reactor feed rate had been held constant and none of the reactor contents had been removed.
20 h), the cell density increased exponentially. At t = 50 h, the feed stream into the reactor was started. This caused an apparent decrease in growth rate of the culture as seen by the change in slope of the growth curve. This apparent decrease in growth rate is in fact due to the dilution effect of the feed stream; the cells continue to grow at constant rate. A similar effect was seen on the consumption rates of glucose and glutamine and on the production rates of lactate and ammonia. Growth rates and yields for each culture were estimated by linear regression of eqs 5-7. These results are summarized in Table I. In the table, the standard error is given below each value in parentheses. The raw data for BATCH 2 and FEDBATCH 2 cultures are not shown. The duration of exponential growth can be estimated for each culture by plotting eq 5. A sample plot of this equation is shown in Figure 3 for the FEDBATCH 2 culture. Growth rate is estimated from the slope of the linear portion of the curve. The duration of exponential growth is estimated by measuring the length on the time axis of the linear portion of the curve. The results for all of the cultures are shown in Table 11. The shear sensitivity profiles, as measured by the normalized viability and fraction lysis of the cells after shear-
(0.004) 0.043 (0.002) 0.049 (0.002)
1.416 (0.186) 1.898 (0.126)
1.18 (0.15) 1.76 (0.08)
a These values were obtained by linear regression of kinetic data from the exponential phase of the cultures with eqs 5-7. The numbers in parentheses are the standard errors for each measurement obtained from the linear regression.
Table 11. Fed-Batch and Batch Culture Parameters: Durations of the Exponential Phases of Growth and Durations of the Periods of Reduced Shear Sensitivity in Batch and Fed-Batch Culturesa period of duration of exponential phase, h reduced shear measured sensitivitv. h culture tvDe ea 10 "BATCH 1 36.0 33.5 50.0 BATCH 2 36.0 32.25 50.0 FEDBATCH 1 67.0 75.0 75.0 FEDBATCH 2 67.0 70.0 80.0 a Duration of the exponential growth was extended beyond that of normal batch culture by operation in fed-batch mode. ing, are shown in Figures 4 and 5 for the batch cultures and in Figures 6 and 7 for the fed-batch cultures. To compare the data for these runs, a measure of the length of time that the cells are resistant to shear is needed. We arbitrarily measure the length of time that the cells are insensitive to shear by the width a t half-height (WAHH) of the viability curve peaks. This is evaluated by drawing a line across the normalized viability profile a t a value of 0.5 on the y-axis. The two points where this line intersects the viability curve denote the beginning and end of the period of reduced shear sensitivity (see Figure 7). The difference in time between these two points (measured on the x-axis) is the WAHH. The results of these measurements are summarized in Table 11. In the two batch cultures, the WAHHs were about 50 h. For the fed-batch cultures, the WAHHs were 75 and
118
Biotechnol. Prog., 1990,Vol. 6, No. 2 0
0
d
Time i n culture, h
Figure 4. Shear sensitivity data for BATCH 1 culture. Cells
were shown in batch mode and sheared twice per day. Shearing was for 10 min at 50 dyn.cm-2. Normalized viability (0) and percent lysis (A)were determined after each shear. The width a t half-height of the viability, used as a measure of the length of the period of reduced shear sensitivity, is 50 h.
-0
go/ 0
0
0
,
20
, 40
,
,
60
80
,
,
loL
100 120 1400
Time i n c u l t u r e ,
0
0
P
Time i n c u l t u r e , h Figure 7. Shear sensitivity data for FEDBATCH 2 culture. Other details are the same as in Figure 4. Width at half-height is 70 h.
Continuous Culture (Chemostat)Experiments. Our previous observation (Petersen, 1989) was that cells in batch culture were less sensitive to shear during exponential growth. This was confirmed by the fed-batch experiments. One interpretation of this finding is that slowly growing cells are more sensitive to shear than rapidly growing cells. This suggests that the shear sensitivity of cells should be investigated at different growth rates. To achieve this, the cells were grown at different dilution rates ( D ) ,and thus different growth rates, in a chemostat, since, at steady state
The metabolic yields can also be determined from material balances:
h
Figure 5. Shear sensitivity data for BATCH 2 culture. Other details are the same as in Figure 4. Width at half-height is 50 h.
0
Time i n c u l t u r e ,
--
h
Figure 6. Shear sensitivity data for FEDBATCH 1 culture.
Other details are the same as in Figure 4. Width at half-height is 75 h. 80 h. The difference between the batch and fed-batch cultures is then around 25-30 h. Exponential growth in the fed-batch cultures was designed to last 31 h longer than in comparable batch cultures. The actual measured difference in these cultures was about 40 h (Table 11). This increase in duration of the exponential phase compares well with the increase in WAHH observed for fed-batch cultures. This supports the hypothesis that extending the period of exponential cell growth will also extend the period of reduced shear sensitivity.
P,-P YPJX = X The metabolic data from selected steady states were used to calculate yields from eqs 12 and 13. The metabolite concentrations and yields are shown in Table 111. A t low dilution rates, the glutamine concentration was too small to be detected by our assay technique (less than 0.2 mM), and this indicates that the glutamine sample is a good candidate for the growth-limiting nutrient. The substrate yields tended to be lower at both the lowest and highest dilution rates. The normalized viability and lysis data for the cultures are shown in Figure 8. Each point is the average of three shearing experiments, and the error bars mark the 90% confidence levels. No obvious trend in shear sensitivity was observed with changes in growth rate or the variable metabolic yields of cells in continuous culture. Though the shear sensitivity of the cells was higher at D = 0.01 and 0.045 h-', the difference was not greater than the experimental error. The culture was converted to batch mode after the chemostat steady states were established to see if the change of culture modes affected shear sensitivity. The culture was converted from continuous mode (D = 0.040 h-l) to batch by diluting the culture 1 5 with fresh media while bringing the total culture volume to 1000 mL. This change was effected by shutting off continuous feed to the reactor, removing 600 mL of cell suspension, and replacing it with 800 mL of fresh media. Figure 9 shows the normalized viability for the cells grown in batch mode.
119
Biotechnol. Prog., 1990,Vol. 6, No. 2
[bl
A
0 .
I
0.00 0.01
,
,
0.02
I
I
0.03 0.04 0.05 0
00 0 . 0 1
1
0.02
I
I
I
0.03 0.04 0.05 0 . 0 6
Dilution rate, h - I D i l u t i o n r a t e , h-' Figure 8. Normalized viability (a) and fractional lysis (b) of hybridoma cells from chemostat culture after shearing for 10 min at 50 dyn-cm-'. Each point is the average normalized viability obtained after shearing three samples of the culture. The error bars indicate the 90% confidence intervals.
Discussion
"
I
0
20 40 6'0 Time i n batch
loo
60
120
culture, h Figure 9. Shear sensitivity of cells when culture is converted from chemostat mode to batch mode. The data are the normalized viabilities after shearing for 10 min at 50 dyn-cm-'. Table 111. Metabolite Concentrations, Growth Rates, and Yields (Y,,, or YpIx)Calculated from the Continuous Culture Experiments.
cell density, D or p, h-'
10s/mL
0.0100 0.0190 0.0275 0.0363 0.0450
1.321 1.979 1.618 1.851 1.100
concentrations mg/mL glucose lactate glutamine NH, 0.460 1.475 1.696 1.794 1.675
2.096 1.617 1.706 1.589 1.750
0.000 0.000 0.026 0.086 0.131
0.048 0.044 0.042 0.033 0.032
Y,,, or YPlx, mg/W cells D or p , h-'
glucose
lactate
glutamine
NH,
0.0100 0.0190 0.0275 0.0363 0.0450
3.058 1.529 1.733 1.462 2.571
1.587 0.817 1.054 0.858 1.591
0.442 0.295 0.345 0.269 0.412
0.036 0.022 0.026 0.018 0.030
a
The growth rates and yields were calculated from eqs 11-13.
The "down-up-down" trend in shear sensitivity was observed as for other batch cultures (Figures 4 and 5 and Petersen (1989)). The sudden increase in shear sensitivity upon dilution is thought to be caused by some as yet uncharacterized "shock" to the culture and has been previously reported (Petersen et al., 1988).
From our initial studies (Petersen, 1989), there seemed to be a relation between exponential growth and shear sensitivity of the cells. To study this relation, the shear sensitivities of cultures with different lengths of exponential phases were studied. Fed-batch culture was used as a convenient technique for changing the duration of the exponential phase of the culture. In an effort to select cultures with similar initial physiological states, the cultures in the fed-batch experiments were screened so that all of the cultures that were analyzed exhibited similar growth characteristics. This was done to increase the likelihood that any observed differences between the batch and fed-batch cultures were the result of the change in growth mode and not interculture variation. Even though these cultures had similar growth rates, considerable variation was observed in the metabolite yields (1.416-2.688 mg/106 cells for glucose, 1.18-1.76 mg/106 cells for lactate), so the attempts to analyze a uniform set of cultures were not entirely successful. Papoutsakis and Kunas (1989) reported glucose yields of 1.6-3.3 mg/106 cells for the same cell line under similar conditions, which are in agreement with the values calculated here. Despite the variation in metabolic yields, the batch and fed-batch cultures were still consistent in the durations of their exponential growth periods and width at half-height of the normalized viability curve (WAHH). One may also interpret this finding in that as long as the cells are actively growing, the metabolic state of the cells, as reflected by the variable metabolic yields, does not affect their shear sensitivity. The continuous culture experiments confirm this observation. Also, the increase in duration of the exponential phase of the fed-batch cultures agreed nicely with that predicted in the design of the experiments. The difference in WAHH between fed-batch and batch cultures seemed to correlate with the extension of the exponential phase due to operation of the culture in fedbatch mode. This observation supports the hypothesis of a relation between exponential growth and ability to tolerate shear. It is also interesting to note that, for the cells from batch cultures, the WAHH was considerably longer than their exponential phase, while, in fed-batch cultures, the WAHH was nearly equal to the duration of the exponential phase.
Biotechnol. Prog., 1990, Vol. 6, No. 2
120
Another interpretation of our initial observation (Petersen, 1989) that shear sensitivity is related to growth is that cells become more robust at higher growth rates. This was investigated by growing cultures in a chemostat, where culture growth rate can be readily controlled. Shear sensitivity of the cells was fairly constant over a wide range of dilution rates and metabolic yields, suggesting that shear sensitivity does not depend directly on specific growth rate or the metabolic state of the cells as expressed by variable metabolic yields, as long as the cells are actively growing. Thus, of the mechanisms suggested in the Introduction to explain the variable shear sensitivity of cells in batch culture, a dependence of shear sensitivity on growth rate seems unlikely. A dependence of shear sensitivity on either culture synchronization or nutrient limitations is currently under investigation. A practical conclusion of this work is that nonactively growing cultures (in the lag and stationary phases) should be agitated or mixed mildly due to their increased sensitivity to fluid-mechanical forces. This is consistent with the reduced oxygen-transfer requirements of the cultures under such conditions. On the other hand, agitation or mixing of the nongrowing cultures during a desirably prolonged product-expression phase may become problematic. This is an especially acute problem in microcarrier cultures since the product-expression phase is typically carried out in a serum-free medium under nongrowth conditions. In the absence of serum, cells can be damaged even more easily compared to serum-containing media (Kunas and Papoutsakis, 1989; 1990). One strategy that we found useful to resolve this difficulty in our lab is to use a serum-free medium that allows just enough cell growth to compensate for cell death due to agitation during the product-expression phase. Another strategy is to alternate between long product-expression phases and short growth phases to maintain a high and actively expressing cell population.
Acknowledgment This research was supported in part by the National Science Foundation (USA) under Grant ECE-8896100 and matching grants from the Monsanto Corporation and the Eastman Kodak Company, by the National Institutes of Health (USA) under Grants HL-17437 and HL-18672, and by the Robert A. Welch Foundation under Grant C938.
Literature Cited Absher, M. In Tissue Culture: Methods and Applications; Kruse, P. F., Jr., Patterson, M. K., Jr., Eds.; Academic Press: New York, 1973; pp 395-397. Ashihara, T.; Baserga, R. In Methods in Enzymology; Jakoby, W. B., Pastan, I. H., Eds.; Academic Press London, 1979; Vol. 58, pp 248-262. Diamond. S. L.: Eskin. S. G.: McIntire. L. V. Fluid Flow Stimulates Tissue Plasminogen Activator Secretion by Cultured Human Endothelial Cells. Science 1989, 243, 1483-1485. Frangos, J. A.; Eskin, S. G.; McIntire, L. V.; Ives, C. L. Flow Effects on Prostacyclin Production by Cultured Human Endot-
helial Cells. Science 1985,227, 1477-1479. Frangos, J. A.; McIntire, L. V.; Eskin, S. G. Shear Stress Induced Stimulation of Mammalian Cell Metabolism. Biotechnol. Bioeng. 1988,32, 1053-1060. Handa, A,; Emery, A. N.; Spier, R. E. On the Evaluation of Gas-Liquid Interfacial Effects on Hybridoma Viability in Bubble Column Reactors. Deu. Biol. Stand. 1987, 66, 241-253. Handa-Corrigan, A,; Emergy, A. N.; Spier, R. E. Effect of GasLiquid Interfaces on the Growth of Suspended Mammalian Cells: Mechanisms of Cell Damage by Bubbles. Enzyme Microb. Technol. 1989, 11, 230-235. Kunas, K. T.; Papoutsakis, E. T. Increasing Serum Concentrations Decrease Cell Death and Allow Growth of Hybridoma Cells at Higher Agitation Rates. Biotechnol. Lett. 1989, I 1 (8), 525-530. Kunas, K. T.; Papoutsakis, E. T. Protective Effect of Serum Against Hydrodynamic Damage of Hybridoma Cells in Agitated and Surface Aerated Bioreactors. J . Biotechnol. 1990, in press. Lee, G. M.; Huard, T. K.; Kaminski, M. S.; Palsson, B. 0. Effect of Mechanical Agitation on Hybridoma Cell Growth. Biotechnol. Lett. 1988, 10, 625-628. McQueen, A.; Meilhoc, E.; Bailey, J. E. Flow Effects on the Viability and Lysis of Suspended Mammalian Cells. Biotechnol. Lett. 1987, 9, 831-836. Nelson, K. L. Industrial-Scale Mammalian Cell Culture, Part 11: Design and Scale-up. BioPharm 1988, I , 34-41, 50. Papoutsakis, E. T.; Kunas, K. T. Presented at Proceedings of the 9th Meeting of the European Societyfor Animal Cell Technology, Knokke, Belgium, Sept 26-30, 1988. Also in: Advances in Animal Cell Biology and Technology for Bioprocesses; Spier, R. E., Griffiths, J. B., Stephenne, J., Crooy,
P. J., Eds.; Butterworths: Kent, England, 1989; pp 203-208. Petersen, J. F. Ph.D. Thesis, Rice University, Houston, TX, 1989. Petersen, J. F.; McIntire, L. V.; Papoutsakis, E. T. Shear Sensitivity of Cultured Hybridoma Cells (CRL-8018) Depends on Mode of Growth, Culture Age and Metabolite Concentration. J . Biotechnol. 1988, 7, 229-246. Phillips, H. J. In Tissue Culture: Methods and Applications; Kruse, P. F., Jr., Patterson, M. K., Jr., Eds.; Academic Press: New York, 1973; pp 406-408. Schurch, U.; Kramer, H.; Einsele, A,; Widmer, F.; Eppenberger, H. M. Experimental Evaluation of Laminar Shear Stress on the Behavior of Hybridoma Mass Cell Cultures, Producing Monoclonal Antibodies Against Mitochondrial Creatine Kinase. J . Biotechnol. 1988, 7, 179-184. Stathopoulos, N. A,; Hellums, J. D. Shear Stress Effects on Human Embryonic Kidney Cells In Vitro. Biotechnol. Bioeng. 1985,27, 1021-1026. Tolbert, W. R.; Schoenfeld, R. A.; Lewis, C.; Feder, J. Design and Use of an Economical Batch Suspension System. Biotechnol. Bioeng. 1982,24, 1672-1679. Tramper, J.; Vlak, J. M. In Advances Biotechnology Processes; Mizrahi, A., Ed.; Alan R. Liss: New York, 1988; Vol. 7, pp 199-208. Tramper, J.; Williams, J. B.; Joustra, D.; Vlak, J. M. Shear Senstivity of Insect Cells in Suspension. Enzyme Microb. Technol. 1986, 8 , 33-36. Tramper, J.; Smit, D.; Straatman, J.; Vlak, J. M. Bubble Column Design for Growth of Fragile Insect Cells. Bioprocess Eng. 1988, 3, 37-41. Accepted February 12, 1990.
