Biotechnol. hog. 1994, 10, 87-96
87
Monoclonal Antibody Process Development Using Medium Concentratesf Theodora A. Bibila,’ Colette S. Ranucci, Konstantin Glazomitsky, Barry C. Buckland, and John G. Aunins Bioprocess R&D Department, Merck Research Laboratories, Rahway, New Jersey 07065
A fed-batch process using concentrated medium was evaluated for its ability to improve cell culture longevity and final monoclonal antibody (MAb) titers for two monoclonal antibody producing cell lines. It was found t o result in up t o 7-fold increases in final antibody titers compared t o batch culture controls. Although the development of cell line specific fed-batch protocols is critical t o the development of cost-efficient largescale production processes, the use of complete medium concentrates provided us with a quick and simple method for producing large quantities of antibodies in the early stages of process development, thus accelerating early work on purification process development, analytical development, biochemical characterizati on, and safety studies. Insights gained from the concentrated medium fed-batch approach were valuable for the development of refined, cell line specific feeding strategies yielding final MAb titers on the order of 1-2 g/L. Process development data on the effects of inhibitory growth byproducts, medium osmolarity, and the mode of nutrient feed addition on culture longevity and MAb production and information on culture metabolic behavior were successfully incorporated in the development of the optimized fed-batch protocols.
1. Introduction Cost-effective large-scale production of monoclonal antibodies (MAbs) is strongly dependent on the development of highly productive, scalable processes. Since final antibody titers in batch and fed-batch cultures are determined by cell density, culture longevity (measured by the integrated number of viable cells over the course of the culture), and specific antibody secretion rate (pg of MAbIcelllday), process development strategies aim at maximizing these two parameters by appropriate bioreactor, culture medium, and feeding protocol design. Significant effort in MAb production is being invested in the development of fed-batch processes as a means of achieving the aforementioned objectives. The strategies for fed-batch process development could, in general, be classified into two categories: the “bottom-up”approach and the “top-down” approach (Bibila, 1993a). The bottom-up approach for fed-batch protocol development consists of supplementation of the culture with medium components that are identified as being quickly consumed or depleted. In the simplest case, the nutrient feeds consist of a single nutrient or a combination of a carbon and an energy source. Glucose and glutaminefeeds have been widely used to improve hybridoma culture performance (Glacken, 1989; De Tremblay, 1992, 1993). Mathematical models of cell growth and MAb production coupled with optimal and feedback control techniques have been used to optimize the timing and mode of feeding in this case (Glacken, 1989; De Tremblay, 1992, 1993). However, multinutrient feeds are usually required for the maximization of culture longevity and final MAb titers. “Nutrient homeostasis”and “rationalmedium design”are examples of bottom-up multinutrient fed-batch protocol
* Author to whom correspondence should be addressed.
+ This work
was presented at the 205th National Meeting of the American Chemical Society, Biochemical Technology Division, Denver, CO, March, 1993.
developmentstrategies. The principle of nutrient homeostasis consists of maintaining nutrient concentrations constant at their initial basal levels during the course of the culture. Using this principle, Robinson et al. (1993a) designed fed-batch protocols yielding MAb titers on the order of 2 g/L. The principle of rational medium design consists of the formulation of nutrient cocktails based on the stoichiometric metabolic and energy requirements for production of cell mass and antibody. This approach has also been used very successfully to achieve increases in final antibody titers on the order of 10-fold compared to batch culture (Xie et al., 1993). In general, the bottomup approach requires the longterm investment of manpower and time. Identification of the important nutrients and determination of the optimal composition of nutrient cocktails to be used for feeding require much analytical work. This is especially true in the case of nutrients other than carbon sources and amino acids (such as vitamins, lipids, trace elements, etc.). The bottom-up approach is also very cell line specific: a high-yield fed-batch protocol developed for one cell line or clone has to be redesigned or reoptimized in order to perform optimally with another cell line or clone. Although this approach is definitely invaluable when it comes to designing the most costeffective production process, the amount of time and effort required to do so is not always available or justifiable in the early phases of process development. On the other hand, the top-down approach for fed-batch protocol development consists of feeding the culture with complete medium in order to quickly achieve culture longevity and antibody titer improvements with no need to identify limiting medium components and formulate customized nutrient cocktails. Complete medium feeding can be easily accomplished by the use of medium concentrates. This technique has been shown to dramatically increase cell densities and final product titers in the case of recombinant CHO cells (Hettwer, 1991; Miller, 1993). The use of complete medium concentrates offers an
8756-7938/94/30 lO-0087$04.50/0 0 1994 American Chemical Soclety and American Instlute of Chemical Engineers
a0
excellent tool for quickly developing an all-purpose, cell line independent, high-yield fed-batch process. This approach can be used to produce, in a simple and timely manner, gram quantities of the MAb needed in the early project phases for initial purification work, analytical/ biochemical/stability testing, and/or preliminary pharmacokinetic/safety studies. Having used the concentrated medium approach for satisfying initial large-quantity MAb project requirements, one can then follow a top-down approach for identifying and removing redundant or excessively accumulating nutrients from the fed-batch protocol and/or using insights gained from this approach in combination with bottom-up techniques in order to develop an optimized fed-batch process. This work presents the evaluation of lox concentrated medium for two MAb-producing recombinant myelomas and demonstrates the general applicability of this approach for successful short-term fed-batch process development (Bibila, 1993a). The effects of the 1OX medium addition on cell growth, antibody production, and nutrient and waste product metabolic profiles will be discussed, along with the development of an optimal fed-batch process based on a combination of the 1OX medium approach with bottom-up techniques. 2. Materials and Methods 2.1. Cell Lines and Culture Medium. Two different
cell lines secreting two therapeutic monoclonal antibodies were used in this work A and B. Both were recombinant NSO murine myeloma cells transfected with a glutamine synthetase (GS) expression vector system (Bebbington, 1992). Glutamine synthetase catalyzes the formation of glutamine from glutamate and ammonia. Addition of this activity allows the selection and culture of recombinants in glutamine-free medium. Amplification of the vector can be achieved by adding the GS inhibitor methionine sulfoximine (MSX). Cell line A was amplified whereas cell line B was not. The specific MAb secretion rates of cell lines A and B, determined during the exponential phase of batch growth, were 30-40 and 15 pg/cell/day, respectively. Cells were grown in a serum-free medium consisting of glutamine-free Iscove's Modified Dulbecco's medium (IMDM, Sigma Chemical Co.) supplemented with nucleosides, 0-mercaptoethanol, ethanolamine, aminoacids,0.1 7% Pluronic F68, bovine low-density lipoprotein (Excyte, Miles, Inc.), bovine serum albumin, insulin, transferrin, and vitamins. The final concentrations of the proteinaceous medium supplements were different for the unamplified and amplified clones. Exact medium composition is proprietary. Methionine sulfoximine was also included in the medium for cell line A for selection pressure. A total of 60 mL of supplements was added per liter of basal IMDM medium. In the optimized fed-batch process, the lipoproteins in the medium were completely replaced by a nonproteinaceous lipid emulsion produced in-house (Seamans, 1993). To minimize osmolarity increases, the 1OX concentrated basal IMDM medium (liquid, Gibco) did not contain sodium chloride, potassium chloride, or sodium bicarbonate but did contain trace elements. Precipitation problems were experienced only after long-term (14 months) storage of the lox basal medium at 4 "C. The exact nature of the precipitate was not determined. Complete 1OX medium was formulated by the addition of concentrated stock solutions of the medium supplements listed above. Sixty milliliters of supplements were added per 100 mL of 1OX concentrated basal medium, thereby making 160 mL of the supplemented concentrated medium, equivalent to 1 L of l x supplemented medium.
Biotechnol. Prog., 1994, Vol. 10, No. 1
2.2. Culture Growth and Monitoring. Cells were grown in 2-L bioreactors with a 1.5-L working volume (B. Braun, Allentown, PA). The temperature in the bioreactors was controlled at 37 "C,and dissolved oxygen and pH were controlled at 20% of air saturation and 7.2, respectively, by gas 02/Nz/C02 blending and NaOH addition (NaOH addition was necessary after addition of the 1OX medium shots, since the pH of the 1OX basal medium is around 5.5). Bioreactor cultures were inoculated at 2 X lo5 viable cells/mL. In the case where a total of 3 culture vol equiv of complete concentrated medium was added to the culture, the initial culture volume was 1L. Addition of the 1OX concentrated medium either shotwise or continuouslyresulted in a 50% increase in culture volume up to 1.5 L. During shotwise addition, the equivalent of 1 culture vol of complete concentrated medium was added to the culture with each shot. This corresponded to the addition of 160 mL of complete concentrated medium for the first shot, 185mL for the second shot, and 215 mL for the third shot. During continuous addition, the same total volume (560 mL) of complete concentrated medium was added to the cultures over a period of either 4 (slow feeding) or 8 days (faster feeding). In the case where a total of 3 culture vol equiv of 1OX basal medium was added to the culture, the initial culture volume was 1.15 L and the final volume was 1.5 L, following the addition of three shots of 115, 125, and 130mL, respectively. Likewise, supplementation with 1.5 culture vol equiv of concentrated medium was performed by the addition of a total of 280 mL of concentrated medium to an initial culture volume of 1.22 L (finalvolume 1.5 L). Finally, supplementation with 1.5 culture vol equiv of basal concentrated medium and 3 vol equiv of supplements was performed by the addition of 400 mL of concentrated medium and supplements to an initial culture volume of 1.1 L (final volume 1.5 L). Cell concentration and viability were determined by hemocytometer counts after trypan blue staining. Glucose, lactate, and ammonia assays were performed using a Biolyzer analyzer (Kodak, Rochester, NY). Amino acid analysis was performed by HPLC; precolumn derivatization with o-phthalaldehyde (OPA) and chromatography on a microbore (200 X 2.1 mm) reverse-phase column (Hewlett-Packard Analytical) followed by fluorescent detection were used (Hill, 1979,1982). The OPAtechnique detects 18of the amino acids. Proline, cysteine, and cystine are not detected. Antibody was assayed by HPLC, using a protein A affinity chromatography column (Perseptive Biosystems, Cambridge, MA) (Krips, 1991). Culture medium osmolarity was determined using a freezing point depression osmometer (Model 3M0, Advanced Instruments, Needham Heights, MA). 2.3. Calculation of t h e Integral of Viable Cells and the Specific Antibody Secretion Rate. The mass balance for antibody accumulation in the cell culture supernatant can be expressed as
d[[MAbl U / d t = QmbXvV where [MAbl is the antibody concentration in the cell culture supernatant, t is time, V is culture volume, q m b is the specific antibody secretion rate, and XVis the viable cell concentration. If we assume that the specific antibody secretion rate remains constant during the course of the culture, integration of the above equation under constant-volume conditions suggests that a plot of the MAb titer versus the integral of viable cells should yield a straight line whose slope is equal to the specific antibody secretion rate (Renard, 1988). In the experiments described here, the
89
Bbtechnol. Prog., 1994, Vol. 10, No. 1 10'
-3a
5
L
4
W
2 CI
-aa K
lo6
3
W
.-5
2
>
io5
1
0
Batch
1
2 3 Culture No.
I
I
I
I
I
..
I
4
Figure 1. Effect of lox shot composition on maximum viable cell density (solid bars), final integral of viable cells (diagonally striped bars), specificantibody secretion rate (horizontallystriped bars), and final MAb titers (verticallystriped bars) for cell line A. Culture 1: Fed 3 vol equiv of lox basal medium. Culture 2: Fed 1.5 vol equiv of complete 1OX medium. Culture 3: Fed 1.5 vol equivof lox basalmedium + 3vol equiv worth of Supplements. Culture 4: Fed 3 vol equiv of complete lox medium. reported values for the integral of viable cells have been corrected to account for changes in culture volume as a function of time (where necessary). The integral of viable cells was calculated as a function of culture time using the following formula:
Time (hours) 1000 800
s E
600
v
a
9 B
400
200
0 0
I00
200
300
400
500
Time (hours)
where X V ,=~(XVZ ~ + Xv1)/2, Xvl and XVZare the viable cell concentrations at times tl and tz, respectively, and XV,av is the average viable cell concentration for the time interval, ( t z - tl). Increases in the final integral of viable cells can be mediated by increases in the maximum viable cell density, as well as decreases in cell death rate during the decline phase of the culture.
3. Results and Discussion 3.1. Effect of 1OX Concentrated Medium Shot Composition. In an effort to minimize the short-term process development time for cell line A and quickly improve culture longevity and final MAb titers following a top-down approach, supplementation of the culture with 1OX concentrated medium was evaluated. Four concentrated medium formulations were evaluated. In all cases, the 1OX medium was added to the culture as three shots starting at lo6 viable cells/mL and following at 48-h intervals. The formulations evaluated were as follows: (1)3 culture vol equiv of lox basal medium: (2) 1.5 culture vol equiv of complete 1OX concentrated medium; (3) 1.5 culture vol equiv of basal 1OX medium supplemented with 3 culture vol equiv worth of supplements; and (4) 3 culture vol equiv of complete 1OX concentrated medium. Figure 1 summarizes the results of these experiments in terms of maximum viable cell density, final integral of viable cells, specific antibody secretion rate, and final antibody titer in batch culture versus the different fedbatch protocols. In all cases, culture performance was improved compared to batch culture. In general, culture performance improved by going from the basal to the supplemented medium feed and by feeding 3 instead of 1.5 vol equiv of the concentrated medium. Addition of the basal concentrated medium had no significant effect on maximum viable cell density or the final integral of viable cells, but it still resulted in a 1.9-fold increase in the final MAb titer compared to batch culture. This increase was mediated by an increase in the specific MAb secretion
Figure 2. (A) Growth of cell line A in batch ( 0 )and lox medium fed fed-batch (W) cultures. Arrows indicate the points of 1 0 ~ medium shot addition. (B)Accumulation profiles of MAb A during batch (0)and 1OX medium fed fed-batch cultures (w). Arrows indicate the points of 1OX medium shot addition. rate. All other fed-batch protocols resulted in substantial increases in the maximum viable cell density (1.7-2-fold), the final integral of viable cells (2.3-3.3-fold1, the specific MAb secretion rate (2-fold), and the final MAb titer (1.97-fold)compared to batch culture. The observed increases in specific MAb secretion rate observed in all cultures fed with concentrated medium can be partially attributed to the osmolarity increases mediated by the addition of the concentrated medium, as discussed below. Addition of 3 culture vol equivof complete medium gave the best results, suggesting that both basal medium components and medium suppIements are necessary to achieve high cell densities and high final antibody titers. Addition of more than 3 culture vol equiv of complete concentrated medium resulted in inferior culture performance due to excessive increases in medium osmolarity as discussed below. In the case of cell line B, addition of five lox complete medium shots resulted in a final titer of 270 mg/L (compared to 365 mg/L for the three shots). 3.2. General Applicability of the 1OX Medium Approach. The cell growth and MAb accumulation profiles for the case of the best concentrated medium protocol (three concentrated complete medium shots) are compared with the corresponding profiles from batch culture for cell line A in Figure 2. The maximum viable cell density and final MAb A titer in batch culture were lo6viable cells/mL and 145mg/L, respectively, in a 175-h cycle. The maximum viable cell density in the concentrated medium fed culture was 2 X lo6 viable cells/mL and the final MAb A titer after 475 h was 1 g/L, a titer comparable with good microbial fermentation titers. The concentrated medium fed-batch protocol that gave optimal results with cell line A was also evaluated with
Biotechnol. Prog.., 1994, Vol. 10, No. 1
6
-1a J al
-al .->m
I
104
0
l
l
-
___.,,
-+---
"""
3
5
2
K
100
1
I
! / ! 200! ! 300; ! ! I 400
4
2 m
CI
P
5
0 Batch
500
1
2
3
Culture No.
Time (hours)
Figure 4. Effect of lox shot addition mode on maximum viable cell density (solid bars), final integral of viable cells (diagonally stripedbars),specific antibodysecretion rate (horizontallystriped bars), and final MAb A titer (verticallystriped bars) for cell line A. In all cases, 3 vol equiv of lox complete medium was added to the culture. Culture 1: shotwise addition of 1OX shots (once every 48 h). Culture 2: Continuousadditionover 4 days. Culture 3: Continuous addition over 8 days. m 0.8 v)
al
a
0
100
200
300
400
500
Time (hours)
Figure 3. (A) Growth of cell line B in batch ( 0 )and lox medium fed fed-batch).( cultures. Arrows indicate the points of lox medium shot addition. (B)Accumulation profiles of MAb B during batch ( 0 )and lox medium fed fed-batch).( cultures. Arows indicate the points of 1OX medium shot addition.
7 > al >
-m
'E
016 0.4
al
0.2 0
1
cell line B in order to test the general applicability of the top-down approach. Figure 3 summarizes the results obtained in a fed-batch culture of cell line B compared to batch culture. The maximum viable cell density and final MAb titer achieved in batch culture during a 325-h cycle were lo6 viable cells/mL and 100 mg/L, respectively. The addition of three lox complete medium shots resulted in a 1.8-fold increase in maximum viable cell density, a 2.9fold increase in the final integral of viable cells, a 3-fold increase in specific MAb secretion rate, and a 3.7-fold increase in final MAb titer, yielding a final MAb titer of 365 mg/L after 400 h. These experiments show that complete concentrated medium can be used in order to quickly achieve dramatic increases in final antibody titers (>250%) and minimize short-term process development time. The approach is very simple and yielded excellent results with two different NSO clones-an unamplified and an amplified clone producing two different MAbs-indicating a more general applicability. For application of the approach with cell lines other than NSO, the lox medium formulation should, of course, reflect the composition of the appropriate basal medium. 3.3. Effect of the Fed-Batch Shot Addition Mode. Having identified a concentrated medium composition and feeding protocol that gave excellent results with multiple cell lines, we then focused our attention on the optimization of the fed-batch shot addition mode. In order to investigate the effect of the mode of 1OX concentrated medium addition on cell culture longevity and final antibody titers, experiments were performed where the 3 culture vol equiv of complete lox concentrated medium were fed to cell line A cultures continuously (as
2
3 4 5 Experiment
6
7
8
No.
Figure 5. Relationship between the final integral of viable cells (solidbars)and final MAb titer (stripedbars)during lox medium process development with cell line A. opposed to shotwise) over a period of either 4 or 8 days starting at 106 viable cells/mL. In the shotwise addition protocol, the three concentrated medium shots were added over 4 days. Figure 4 summarizes the results of these experiments in terms of maximum viable cell density, the final integral of viable cells, specific antibody secretion rate, and final antibody titers in the batch versus the lox medium fed cultures. There was no difference in performance between the cultures that were fed shotwise or continuously over 4 days. In the case of the slower continuous feed rate (over 8 days), although the final integral of viable cells did increase slightly compared to theshotwise fed culture (by about 13961,thefinalantibody titer decreased (by 30%) due to a simultaneous decrease in the specific antibody secretion rate (by 33%). This phenomenon of improvement in culture longevity at the expense of final antibody titer has been observed repeatedly during process development with these cell lines, using either the concentrated medium or bottom-up fed-batch protocol strategies. Figure 5 summarizes the relationship between the final integral of viable cells and final MAb titer for a series of cell line A lox medium experiments. Increasing experiment numbers represent later process development stages. As shown in this figure, during the course of process development, monotonic increases in the final integral of viable cells were achieved by manipulation of the composition and mode of addition of the concentrated medium shots. However, the increase in final MAb titer is not
e1
B&technol. Prog.., 1994, Vol. 10, No. 1
monotonic. There seems to be an optimum in the final integrd of viable cells versus final MAb titer relationship, below which increases in cell culture longevity translate directly into increases in final MAb titer, since the specific MAb secretion rate under these conditions remains constant. However, above this optimum further increases in the final integral of viable cells are achieved a t the expense of final MAb titer due to parallel decreases in the specific MAb secretion rate, suggesting that more cells do not always translate into more product. In this case, the positive effect on cell growth is not enough to override the negative effect on the specific MAb secretion rate, which results in lower final MAb titers. For long-term process development, identification of the optimum in this relationship is key to operating large-scale cultures under conditions that guarantee maximum MAb yields. The reason for the existence of this optimum is not clear. The decrease in the specific MAb secretion rate as culture longevity improves could not be due to productivity loss or cell line instability over time, since all of the experiments shown in Figure 5 were performed with cells passaged only up to 2 months. After that, fresh cells were thawed for the next series of experiments. These cell lines have been shown to maintain stable productivity for a period of time much longer than 2 months (D. Robinson, personal communication). In addition, the experiments shown in Figure 5 were performed with cell line A, the amplified cell line, and MSX was always included in the culture medium for selective pressure. It might be that, under certain nutritionally enriched conditions, cells divert an increasing percentage of their resources toward maintenance and growth rather than secretory protein synthesis. In the case of the slow continuous feed (over 8 days) versus the shotwise addition, it might also be due to the effect of increased osmolarity on the specific MAb secretion rate. Increased osmolarity has a positive effect on the specific MAb secretion rate, as will be discussed below in detail. In the case of slow continuous addition of the concentrated medium, osmolarity increases at a slower rate than in the case of shotwise additions, although both cultures reach the same final osmolarity level. Therefore, the positive effect of increasing osmolarity on the specific MAb secretion rate is delayed when the concentrated medium is added slowly over 8 days. This would result in comparatively lower specific MAb secretion rates over the same culture period in the cultures fed continuously. Since continuous addition of the concentrated medium did not result in increased final MAb titers, shotwise addition was implemented. 3.4. Analysis of the Optimal Cultures. Although the concentrated medium approach is extremely useful for short-term process development, the cost of the concentrated medium would most likely preclude its use in a production setting. However, a more detailed analysis of the behavior of the optimally performing cultures can provide useful insights for the long-term design of an optimized, cost-effective fed-batch protocol to be used at a large scale. The increases in osmolarity mediated by addition of the concentrated medium, the possible oyeraccumulation of nutrients or their metabolic byproducts, and their effects on cell growth and MAb production are issues of interest when concentrated medium is used for feeding. These are discussed in detail in the following sections for the cell lines A and B fed-batch cultures shown in Figures 2 and 3. 3.4.1. Effect of Osmolarity. Figure 6A shows the medium osmolarity profiles of the fed-batch cultures in Figures 2 and 3. In both cases, despite the fact that the lox medium used did not contain sodium chloride and potassium chloride, addition of the three 1OX medium
500 n
s
-.-*
0 400 E CI
t;r
'0 300
s
0 200 0
100
200 300 400 Time (hours)
500
2m0Ti B v)
-m Q)
1.5
3
>
.-s!
1.o
-m C
0)
a
0.5 0.0
270 300 400 Osmolarity (mOsm)
Figure 6. (A) Osmolarity profiles for lox medium fed-batch cultures of cell lines A (0)and B (m). (B) Effect of osmolarity
on maximum viable cell density (solid bars), final integral of viable cells (diagonallystriped bars), specificantibody secretion rate (horizontally striped bars), and final antibody titers (vertically striped bars) for cell line B. shots results in substantial increases in medium osmolarity (final osmolarity between 400and 450mOsm), which could be expected to affect culture longevity and specific MAb secretion rates (Ozturk, 1991a). The effect of increased medium osmolarity on cell growth and antibody production was studied in separate experiments where medium osmolarity (normally around 270 mOsm) was increased up to 600mOsm by the addition of NaCl in batch cultures. The experiments were performed in T-150 culture flasks using cell line B,and the results are summarized in Figure 6B. These results suggest that, although cell growth is not affected by an osmolarity increase from 270 (normal levels) up to 300 mOsm, there is a substantial negative effect on growth at 400 mOsm, and the cells do not grow at all in media with osmolarities 2500 mOsm. However, increases in osmolarity from 270 up to 300 and 400 mOsm were found to result in 12.5% and 60% increases in the specific antibody secretion rate, respectively. In the case of the modest osmolarity increase to 300 mOsm, the percent improvement in the specific antibody secretion rate translated directly into the same percent increase in the final antibody titer. In the case of the osmolarity increase to 400 mOsm, the positive effect on the specific secretion rate was outweighed by the negative effect on growth, resulting in a 20% lower final MAb titer compared to the control. Similar effects of osmolarity on cell growth and MAb secretion have been reported for other celllines (Ozturk, 1991a). These effects have also been shown to be independent of the chemical source of osmolarity increase (Ozturk, 1991a), suggesting that the results based on NaCl addition could explain culture behavior due to the 1OX medium addition. According to these results, (1)a negative effect on growth
Biotechnol. Prog., 1994, Vol. 10, No. 1
92 1000 800
2
p
600
W
4:
9 I
400
2 0 OetO
le+8
o 2e+8
3e+8
o 4e+8
Integral of viable cells (vc*hr/ml)
h
SEI
300
W
m
200
9I 100
0 Oe+O
let0
2e+8
3e+8
4et8
5etS
Integral of viable cells (vc*hr/ml)
Figure 7. (A) MAb A titer versus the integral of viable cells for lox medium fed-batch culture of cell line A. (B) MAb B titer versus the integral of viable cells for lox medium fed-batch culture of cell line B.
could be expected in the 400 mOsm osmolarity range, and (2) osmolarities of around 400 mOsm reached in the 1OX medium fed cultures could result in increases in the specific antibody secretion rate of up to 60%. As shown in Figures 2 and 3, the 1OX medium fed cultures reached higher maximum viable cell densities and higher final integrals of viable cells compared to the batch controls. Thus, it appears that, overall, the positive effects of the enriched nutritional environment override the negative effects of increased osmolarity. However, it is worth noting that the maximum viable cell densities reached in both cases were close to 2-fold higher compared to the batch culture controls, although these cultures received a 4-fold greater medium supply, indicating that the yield of viable cell mass on medium was decreased compared to batch culture. Cultures fed with 3 culture vol equiv of 1 X medium, by repeated medium exchanges with culture circulation through a 0.2-pm hollow fiber filter (once every 48 h), reached a maximum viable cell density of 3.5 X lo6cells/mL (unpublished data), indicating a nearstoichiometric relationship between maximum viable cell mass and total nutrient supply. Although osmolarity is definitely a factor, overaccumulation of medium components or metabolic byproducts rather than osmolarity alone might also be responsible for the decreased viable cell mass yield in the case of the lox medium fed cultures. The expected increases in the specific MAb secretion rate at increased osmolarities were observed in the lox medium fed cultures (Figures 1 and 5). Figure 7 shows the MAb titer versus the integral of viable cells plots used to determine the specific MAb secretion rates for the fedbatch cultures shown in Figures 2 and 3. A plot of MAb A titer versus the integrated number of viable cells (Figure 7A) yields a straight line, indicating that the specific
antibody secretion rate remains constant during the course of the culture and, in this case, equal to 2 times the batch culture specific secretion rate (57 versus 28.5 pg/cell/day). A plot of MAb B titer versus the integrated number of viable cells (Figure 7B) clearly exhibits biphasic behavior, with the specific secretion rate about equal to the corresponding rate in batch culture (12 pg/cell/day) for the first 200 h of culture and increasing by almost 3-fold after that (35pg/cell/day), once the osmolarity has reached its maximum level (450 mOsm). This increase in the specific MAb B secretion rate corresponds with the onset of the cell death phase (Figure 2). Separate experiments with~ cell line B have shown that, in high-yield, long-term fed-batch cultures, the contribution of nonviable cells to the total antibody pool in the supernatant is negligible, suggesting that the increased secretion rate during the second part of the culture is indeed due to active secretion by viable cells (Robinson, 1993b). Increases in the specific MAb secretion rate from viable cells in the decline or death phase of culture have been previously reported (A1Rubeai, 1990). These results suggest that, in the concentrated medium fed-batch cultures, (1)the profile of the specific MAb secretion rate during the course of the culture is strongly dependent on the cell line, and (2) the observed increases in the specific MAb secretion rate (2-3-fold compared to batch culture) can be only partially attributed to osmolarity increases (only a 60% increase expected at 400 m0sm). I t therefore appears that, although increased osmolarity can account for some of the observed effects on growth and the specific MAb secretion rate, its effects are superimposed with those of other factors, prohibiting a straightforward interpretation of the observed phenomena. 3.4.2. Culture Metabolic Profiles. Since the concentrated medium approach supplies the culture with all medium components, irrespective of whether or not they are depleted, some nutrients are bound to overaccumulate in the culture. Overaccumulation of medium components or their metabolic byproducts can be inhibitory or toxic to cell growth and MAb production. It can also indirectly affect culture performance through its contribution to increased osmolarity, as discussed above. Identification of these components can prove useful for the future formulation of well-balanced nutrient shots during the development of an optimized fed-batch protocol. Glucose and Lactate Profiles. Figure 8 shows the glucose and lactate profiles during the course of the 1OX medium fed cultures for cell lines A and B, respectively. Although glucose is being consumed between successive additions of the 1OX medium, overall there is substantial glucose accumulation in both cases, reaching up to 40-50 mM (2-2.5-fold higher than glucose levels in the 1 X culture medium). These results suggest that culture performance might have benefited from the use of either a low-glucose or a glucose-free concentrated medium formulation. In both cases, lactate concentration increases up to a maximum concentration and subsequently decreases. Maximum lactate accumulation is 10 and 30 mM for cell lines A and B, respectively (Figure 8B). Lactate consumption during the later stages of the culture is a phenomenon that we have observed repeatedly in these cultures with a variety of fed-batch protocols. In both cases, the onset of lactate consumption roughly coincides with the onset of the death phase (Figures 2 and 3). In the case of cell line B, this also coincides with the increase in the specific MAb secretion rate during the second part of the culture (Figure 7B). During the later stages of the culture, lactate could be consumed to produce pyruvate (the reverse of the reaction producing lactate) and supply carbon to the TCA cycle.
BbtechnoL Rag, 1994, Vol. 10, No. 1
93
60 50
3 E
40
8
30
8 I
20
Y
Q
10 0
100
300
400
500
200 300 400 Time (hours)
500
200
Time (hours)
n
z
E
Y
”
0
100
Figure 8. Profiles of glucose (A) and lactate (B)accumulation during 1OX medium fed-batchcultures of cell lines A ( 0 )and B (m). Anowe indicate the points of 1OX medium shot addition.
1.o u)
-a (u
0.8
.-L
0.6
2
0.4
.L
z
0.2 0.0
0 5 10 1 5 20 25 30 Initial lactate concentration (mM)
Figure9. Effect of lactate on maximum viable cell density (solid bars),fiial integral of viable cells (diagonallystripedbars),specific MAb secretion rate (horizontallystriped bars), and final MAb titer (verticallystriped bars) for cell line B.
The effect of the observed maximum lactate levels on cell growth and MAb production was assessed by performing separate experiments. Experiments were performed in T-flask batch cultures with cell line B. Initial lactate concentrations in the range of 0-30 mM were examined. The initial pH was adjusted to 7.2 in all cultures. Figure 9 summarizes the results of these experiments. Lactate was not inhibitory to cell growth or antibody production at concentrations up to about 10mM. At higher lactate concentrations, a negative effect on growth and the final MAb titer was observed. At 30 mM lactate, a 30% decrease in the fiial MAb titer was observed, mediated by a 20 % decrease in the final integral of viable cells and a 16% decrease in the specific MAb secretion rate. Maximum viable cell density in this case decreased
by 25%. These results suggest that, at least in the case of the 1 0 medium ~ fed culture of cell line B, lactate accumulation most likely had a negative effect on growth and MAb production. Lactate accumulation can be controlled by the use of low-glucose medium and/or slow glucose feeding. Amino Acid Profiles. Figure lOA-D summarizes the amino acid profiles during the course of the 1OX medium fed culture for the case of cell line A, the amplified clone. Although not shown here, the results for cell line B-the unamplified clone-were analyzed in the same way (Bibila, 1993b). Most of the amino acids transiently accumulate to 2-3-fold their initial levels following the concentrated medium shots. The effect of this transient exposure to high amino acid concentrations on NSO cell health is not known. Specific amino acid uptake rates remained constant for the majority of the amino acids after the addition of the first and second concentrated medium shots and decreased by 50% or more following the third shot. This decrease coincides with the onset of the death phase. The phenomenon of increased specific amino acid uptake rates in cells that are grown in concentrated or nutritionally enriched media reported for recombinant CHO cells (Miller, 1993)was not observed here. The specific rate of alanine production increases by 10-foldfollowingthe third concentrated medium shot. This dramatic increase in the alanine production rate coincides with the onset of lactate uptake and is in agreement with the hypothesis that at this stage of the culture lactate is converted to pyruvate, which then is converted toalanine. Glutamate, asparagine, and glutamine are of special interest since these are the amino acids directly involved in the glutamine synthetase reaction. In this reaction, glutamate is combined with ammonia produced mainly by asparagine metabolism in order to produce glutamine. The specific uptake rate of glutamate decreases by 65% following the third concentrated medium shot, but at the same time the specific uptake rate of asparagine increases by 70%. In fact, asparagine is the only amino acid whose specific uptake rate increases during this last phase of the culture. Since the glutamine synthetase reaction seems to be slowing down (due to decreased glutamate uptake rate) at the same time that the asparagine uptake is increasing, substantial ammonia accumulation would be expected during this last stage of the culture. This is indeed the case as is discussed below. The glutamine concentration in the medium is basically zero for most of the culture, suggesting that the glutamine derived from the glutamine synthetase reaction is completely consumed by the cells (1OX feeds are glutamine-free). There is, however, a transient accumulation of glutamine in the medium between 250 and 350 h. During this period, glutamine is most likely being produced by the glutamine synthetase reaction faster than it is being consumed by the cells, suggesting a decrease in the glutamine utilization rate. These results indicate that nutrient feeds should be wellbalanced in an optimized fed-batch protocol in order to avoid starvation by or overaccumulation of amino acids. In this case, the feeding of the amino acids supplying the glutamine synthetase reaction as well as differences between the amplified and unamplified clones’ metabolic requirements should be considered carefully (Gould, 1992; Bibila, 1993b). Ammonia Profiles. Since these cells scavengetheir own ammonia to make glutamine, there is very little ammonia accumulation during the course of a typical batch culture. Figure 11A shows the ammonia accumulation profiles for the 1OX medium fed cell line A and B cultures. The patterns of ammonia accumulation are distinctly different for the two cell lines. For cell line A-the amplified
94
Biotechnol. Prog., 1994, Vol. 10, No. 1
2.51A1
6
I
I
I
I
I
I
I
I
2.5
2.0
E
Y
E
.-0
1.5
sE
1.o
(3
-
L
E
L
c Q
I
z
0
c 0
0
3.0
I
0.5
2.0 1.5
1.0
0.5
0.0 100
0
300 400 Time (hours) 200
0.0
500
0
2.0
v)
-sm
Y
c
E
1.o
> aJ .-c> 5 aJ
c C
0 0
z 0
0.5
100
300 400 Time (hours)
200
500
2.0
E
v
E
1.5
c
1.o
E E
Q
0 C
s
0.5
0.0
100
0
2.0ID
I
~
200 300 400 Time (hours)
1
I I1
--e-0-
500
Trp Phe
h
I
E
Y
0.5
0.0
0 0 . 5 1 2 3 4 Initial NH3 concentration (mM)
Figure 11. (A) Ammonia accumulation profiles during 1OX medium fed-batchcultures of cell lines A ( 0 )and B (m), Arrows indicate the points of 1OX medium shot addition. (B) Effect of ammonia on maximum viable cell density (solid bars), final integral of viable cells (diagonally striped bars), specific MAb secretion rate (horizontally striped bars), and final MAb liter (verticallystriped bars) for cell line B.
2.5
.-c0
1.0
K
0.0 0
f
500
B
I 1.5 E L
200 300 400 Time (hours)
1.5
h
.-0
100
l,
0.0
100
200 300 400 Time (hours)
500
Figure 10. Profiles of amino acid concentrations during 1OX medium fed-batch culture of cell line A (A-D). Arrows indicate the points of l o x medium shot addition. clone-ammonia accumulates up to 3 mM; ammonia levels are very low (less than 0.25 mM) during the growth and stationary phases of the culture (as in a typical batch culture), and rapid ammonia accumulation starts at around
200 h. The onset of rapid ammoniaaccumulation coincides with the onset of the death phase and the onset of lactate consumption and parallels the fast asparagine consumption during the late stages of the culture (as discussed above). For cell line B-the unamplified clone-ammonia accumulates up to 1.2 mM; in this case, ammonia accumulates continuously during the growth and stationary phases of the culture up to 1.2 mM and then plateaus at that level during the death phase. These results indicate that there are basic differences in the control of asparagine metabolism, as well as in the control of the glutamine synthetase reaction, in the amplified versus the unamplified cell line. A more detailed analysis of these differences has been presented elsewhere (Bibila, 199313). Separate experiments investigating the effects of increasing ammonia concentrations on cell growth and antibody production were performed in batch T-flask cultures with cell line B. Different initial ammonia concentrations in these cultures were achieved by the addition of appropriate amounts of ammonium chloride. The initial pH was adjusted to 7.2 in all cultures. These experiments showed that ammonia has no (or even a slightly positive) effect on cell growth and antibody production within the 0.5-4 mM concentration range examined (Bibila, 1993b). Figure 11B summarizes the results of these experiments. According to these results, the ammonia accumulation in the 1OX medium fed cultures, 1.2 and 3 mM for cell lines B and A, respectively, is well within the nontoxic concentration regime. 3.5. Development of an Optimized Fed-Batch Process. The knowledge gained from the 1OX medium feeding experiments was invaluable in the long-term design of an
Bbtechnol. Rog., 1994, Vol. 10, No. 1 1000
4
800
T 600
2
3
v
m
400 p
2
200
1
2
0
5
10 15 Time (days)
20
0 25
Figure 12. Cellgrowth ( 0 )and MAbB (m) accumulation profides for cell line B using the optimized fed-batch protocol (Robinson et al., 1993a).
optimized cell line specific fed-batch process. Information on the nutrient and metabolic byproduct profiles, osmolarity effects, and growth versus MAb production relationship, as presented above, was successfully applied to the development of well-balanced nutrient feeds in the optimized final process. For the development of the optimized process, an amino acid feeding protocol developed on the basis of the strategy of nutrient homeostasis, i.e., designed to maintain constant amino acid concentrations at their initial basal levelsduring the course of the culture (bottom-up approach; Silberklang, 1992),was combined with glucose and protein/lipid feeds as indicated by the lox medium experiments in section 3.1 (topdown approach). Refinements of this combination finally yielded-with the investment of significant process development time and effort-an optimized fed-batch culture process shown in Figure 12 (Robinson, 1993a) for cell line B. The maximum viable cell density was increased to 3.5 X lo6 viable cells/mL, and a final antibody titer of 850 mg/L was achieved a t the end of a 22-day process. This process was successfully scaled up to the 200-L scale (Bibila, 1993a). Similar optimized fed-batch protocols used with cell line A and an amplified clone of cell line B yielded final titers in excess of 1 g/L (1.8 and 1.2 g/L, respectively) (Robinson, 1993a). Daily feeding with amino acids, glucose, lipids, vitamins, and proteins was found to be essential for the optimal culture performance shown in Figure 12, as indicated by the 1OX medium experiments on shot composition (Figure 1). Initial implementation of the amino acid homeostasis feeding protocol alone for cell line B had only yielded a final MAb titer or 140 mg/L, i.e., only a 40% improvement over the batch culture titer (Robinson, 1993a). In the optimal process, daily glucose feeding (instead of upfront glucose supplementation) was implemented in order to minimize lactate accumulation, on the basis of the results of the 1OX medium experiments. In addition, the optimized fed-batch protocol was performed in a manner similar to the 1OX medium protocol with respect to the effect of shot mode addition on culture performance and the relationship between cell culture longevity and final MAb titers. These results are summarized in Figure 13. Again, there was basically no difference in cell growth or MAb production between shotwise and continuously fed cultures (Figure 13A); shotwise feeding was therefore implemented. The optimum in the final integral of viable cells versus final MAb titer relationship was also shown to exist in a series of process development experiments using the optimized fedbatch process (Figure 13B). The final osmolarity levels in the optimized fed-batch cultures were also similar to those in the 1OX medium fed cultures, and the observed
0
Batch
1 .o
] B
1 Culture No.
2
M
0.8
0.6 0.4
0.2
n 1 2 3 4 5 7 8 91011 Experiment No.
Figure 13. (A) Effect of the shot addition mode on maximum viable cell density (solid bars), final integral of viable cells (diagonallystripedbars),specificMAb secretionrate (horizontally striped bars), and final MAb titer (vertically striped bars) for cell line A using the optimized fed-batch protocol. Culture 1: Shotwise addition. Culture 2: Continuous addition. (B) Relationship between final integral of viable cells (solid bars) and final MAb titer (striped bars) during process development with cell line A using the optimized fed-batch culture protocol.
effects on specific MAb secretion rates were identical to those observed in the 1OX medium fed cultures. Concentrated medium can therefore be used not only as a tool for rapid production of large MAb quantities in early project phases but also as a useful guide for the longterm development of cost-efficient fed-batch protocols for large-scaleproduction. The bottom-up and top-down fedbatch protocol development strategies are not mutually exclusive, but can be successfully combined for optimal results. 4. Conclusions 1OX concentrated medium was evaluated here for ita ability to improve cell culture longevity and final monoclonal antibody titers in two different antibody-producing cell lines. Increases in final antibody titers on the order of 265-590 % were achieved, showing that complete medium concentrates are a very useful tool for achieving high antibody titers in early stages of process development. Increases in final titers were mediated by the prolongation of cell viability combined with increases in the specific antibody secretion rate. Increases in the specific antibody secretion rate could be partially attributed to increases in osmolarity mediated by the addition of the concentrates. Continuous instead of shotwise addition of the concentrated medium shots was found to improve culture longevity but not final MAb titers due to the existence of an optimum in the final integral of viable cells/final MAb titer relationship. Analysis of the nutrient and metabolic
96
Biotechnol. Prog., 1994,' Vol. 10, No. 1
byproduct profiles of the optimal 1OX medium fed cultures provided information valuable for the development of an optimized, cell line specific fed-batch process. A combination of the 1OX medium approach and a nutrient homeostasis approach resulted in optimized fed-batch protocols yielding titers on the order of 1-2 g/L. In conclusion,the use of complete medium concentrates can minimize the short-term process development effort that otherwise must be invested in the development of a cell line specific, customized fed-batch protocol. I t also can prove invaluable for the rapid production of antibody for initial purification process development, pharmacokinetidsafety studies, and analytical/ biochemical characterization work. Cost and problems such as excessive osmolarity increases and nutrient/byproduct overaccumulation would most likely prevent the application of this approach in a large-scale production environment, and thereby explain the need for long-term cell line specific fed-batch process development, but this does not undermine the technique's usefulness during early process development phases. The knowledge gained from this method is valuable for the long-term development of optimized fed-batch protocols for large-scale MAb production.
Acknowledgment The authors acknowledge Ms. Ellen Dahlgren for performing the amino acid analysis, Dr. Melvin Silberklang, Dr. David Robinson, and Mr. Craig Seamans for useful discussions on medium and process development issues and for providing the optimized fed-batch culture results in Figure 12, and Dr. Cynthia Oliver for helping with the POROS MAb assay.
Literature Cited Al-Rubeai, M.; Emery, A. N. Mechanisms and kinetics of monoclonal antibody synthesis and secretion in synchronous and asynchronous hybridomacellcultures. J .Biotechnol. 1990, 16,67-86. Bebbington, C. R.; Renner, G.; Thompson, S.; King, D.; Abrams, D.; Yarranton, G. T. High level expression of a recombinant antibody from myeloma cells using a glutamine synthetase gene as an amplifiable selectable marker. Biotechnology 1992, 10, 169-175. Bibila, T.; Glazomitsky, K.; Ranucci, C.; Robinson, D.; Silberklang, M.; Buckland, B.; Aunins, J. Approach to optimization of fed-batch culture conditions for monoclonal antibody production. Presented a t the the 205th National Meeting of the American Chemical Society, Biochemical Technology Division, Denver, CO, 1993a. Bibila, T.; Ranucci, C.; Glazomitsky, K.; Buckland, B.; Aunins, J. Investigation of NSO cell metabolic behavior in MAb producing clones. Presented at the Biochemical Engineering VI11 Engineering Foundation Meeting, Princeton, NJ, 1993b. De Tremblay, M.; Perrier, M.; Chavarie, C.; Archambault, J. Optimization of fed-batch culture of hybridoma cells using dynamic programming: single and multi feed cases. Bioprocess Eng. 1992, 7, 229-234. De Tremblay, M.; Perrier, M.; Chavarie, C.; Archambault, J. Fedbatch culture of hybridoma cells: comparison of optimal control and closed loop strategies. Bioprocess Eng. 1993, 9, 13-21. Glacken, M. W.; Huang, C., Sinskey, A. J. Mathematical description of hybridoma culture kinetics. I11 Simulation of fed-batch bioreactors. J . Biotechnol. 1989, 10, 39-66. Gould, S.; DiStefano, D.; Cuca, G.; Robinson, D.; Silberklang, M. Major metabolic changes accompany transfection and selection for high level expression of recombinant genes. In Vitro Cell. Deu. Biol. 1992, 3 (part 11), 162A.
Hettwer, D. J.; Escobar, E.; Fieschko, J. Development of a serumfree suspension process for recombinant CHO cells. Presented a t the American Institute of Chemical Engineers Annual Meeting, Los Angeles, CA, 1991. Hill, D.; Walters, F. H.; Wilson, T. D.; Stuart, J. D. High performance liquid chromatography determination of amino acids in the picomole range. Anal. Chem. 1979, 51, 13381341. Hill, D.; Burnworth, L.; Skea, W.;Pfeifer, R. Quantitative HPLC analysis of plasma amino acids as orthophthaldialdehyde/ ethanethiol derivatives. J . Liquid Chromatogr. 1982,5,23692393. Krips, D. M.; Sitrin, R. D.; Oliver, C. N. A very rapid 2 min protein A HPLC assay for monoclonal antibodies. FASEB J . 1991,5, A465. Miller, W. M.; Blanch, H. W.; Wilke, C. R. The effects of dissolved oxygen concentration on hybridoma growth and metabolism in continuous culture. J . Cell Physiol. 1984, 132, 524-530. Miller,D.A.; Drapeau,D.;Luan,Y.-T.; Whiteford, J.C.;Adamson, S. R. Elevated nutrient levels in CHO cell culture: effect on cellular productivity and amino acid uptake rates. Presented at the 205th National Meeting of the American Chemical Society, Biochemical Technology Division, Denver, CO, 1993. Ozturk, S. S.; Palsson, B. 0. Effect of medium osmolarity on hybridoma growth, metabolism and antibody production. Biotechnol. Bioeng. 1991a, 37, 989-993. Ozturk, S. S.; Palsson, B. 0. Growth, metabolic and antibody production kinetics of hybridoma cell culture: effects of serum concentration, dissolved oxygen concentration and medium pH in a batch reactor. Biotechnol. h o g . 1991b, 7,481-494. Ozturk, S. S.; Riley, M. R.; Palsson, B. 0. Effects of ammonia and lactate on hybridoma growth, metabolism, and antibody production. Biotechnol. Bioeng. 1992, 39, 418-431. Renard, J. M.; Spagnoli, R.; Mazier, C.; Salles, M. F.; Mandine, E. Evidence that monoclonal antibody production kinetics is related to the integral of viable cells in batch systems. Biotechnol. Lett. 1988, 10 (2), 91-96. Robinson, D.; Bibila, T.; Chan, C.; Yu, Ip, C.; Tung, J. S.; Lenny, A.; Seamans, C.; Gould, S.; Glazomitsky, K.; Ranucci, C.; DiStefano, D.; Lee, D.; Cuca, G.; Buckland, B.; Mark, G.; Aunins, J.; Silberklang M. In pursuit of the optimal fed-batch process for monoclonal antibody production. Presented at the Biochemical Engineering VI11 Engineering Foundation Metting, Princeton, NJ, 1993a. Robinson, D.; Chan, C.; Lee, D.; Lenny, A.; Seamans, C.; Tung, J. S.; Tsai, P. K.; Yu Ip, C.; Mark, G.; Silberklang, M. Product consistency during long term fed-batch culture. Presented at the 12th Meeting of the European Society for Animal Cell Technology (ESACT), Wiuzburg, Germany, 1993b. Seamans, T. C.; Gould, S. L.; DiStefano, D. J.; Silberklang, M.; Mark, G. E.; Robinson, D. K. Use of lipid emulsions as nutritional supplements in mammalian cell culture. Presented at the Biochemical Engineering VI11Engineering Foundation Meeting, Princeton, NJ, 1993. Silberklang, M.; Jain, D.; Gould, S.; DiStefano, D.; Cuca, G.; Benincasa, D.; Ramasubramanyan, K.; Lenny, A.; Mark, G. E. A nutrient homeostasis fed-batch process design for recombinant antibody production. Presented at the 111Engineering Foundation Conference on Cell Culture Engineering, Palm Coast, FL, February, 1992. Xie, L.; Wang, D. I. C. Rational design and control of animal cell metabolism. Presented at the 205th National Meeting of the American Chemical Society, Denver, CO, 1993. Accepted November 4, 1993.' @
Abstract published in Advance ACS Abstracts, December 15,
1993.
