Quantitative Studies of Cell-Bubble Interactions and Cell Damage at Different Pluronic F-68 and Cell Concentrations
Ningning Ma,†,‡,§ Jeffery J. Chalmers,‡ John G. Aunin¸ sˇ,*,† Weichang Zhou,†,| and Liangzhi Xie†,
Fermentation and Cell Culture, Bioprocess R & D, Merck Research Laboratories, Merck & Co., Inc., West Point, PA, 19486, and Department of Chemical Engineering, The Ohio State University, Columbus, Ohio 43210
Pluronic F-68 (PF-68) is routinely used as a shear-protection additive in mammalian cell cultures. However, most previous studies of its shear protection mechanisms have typically been qualitative in nature and have not covered a wide range of PF-68 and cell concentrations. In this study, interactions between air bubbles along with the associated cell damage were investigated using the novel adenovirus-producing cell line PER.C6, a human embryonic retinoblast transfected with the adenovirus type 5 E1 gene. A wide range of PF-68 and cell concentrations (approximately 3 orders of magnitude) were used in these studies. At low PF-68 concentrations (0.001 g/L), cells had a very high affinity for bubbles, indicated by a more than 10-fold increase in cell concentration in the foam layer liquid versus the bulk liquid.
At high PF-68 concentrations (∼3 g/L), however, the cell concentration in the foam layer liquid was only ∼40% of that in the bulk cell suspension. The number of cells associated with each bubble decreased from approximately 1000 cells at 0.001 g/L PF-68 to approximately 120 cells at 3 g/L PF-68. Despite the lower cell affinity for bubbles at a high PF-68 concentration, at high cell concentrations (107 cells/mL and 1 g/L PF-68) significant cell entrapment occurred in the foam layer, on the order of 1000 cells/ bubble. For the cells carried by the bubbles, quantitative cell damage data revealed that the probability of cell death from bubble rupture was independent of bulk cell concentration but was affected by PF-68 concentration. These quantitative studies further indicated that even at a low PF-68 concentration of 0.03 g/L, approximately 30% of the attached cells were killed during the bubble rupture process. At the same time, at low PF-68 concentration (<0.1 g/L), significant cell death occurred prior to bubble rupture. On average, a bubble disrupted more cells in the bulk liquid and/or foam layer than during rupture. For both mechanisms, the number of cells damaged by each bubble increased with decreasing PF-68 concentration and increasing bulk cell concentration. Introduction The current industrial standard practice for economical manufacturing of many biological products is suspension cell culture at scales exceeding 10,000 L. At such large scales, gas sparging remains the most efficient and commonly applied method for oxygenation. However, concerns still exist at this large scale with respect to potential cell damage as a result of gas sparging and foam formation. This is especially true as cell concentration Increases to ∼10 million cells/mL or higher in fed-batch and perfusion cultures. Quantitative studies on the interactions among cells, bubbles, and protective addi- tives are relatively scarce in the literature, especially over a wide range of additive and cell concentrations. The understanding from such studies will greatly assist the design and operation of large-scale bioreactors. Observations of sparging-related cell damage are the subject of numerous published studies and reviews (Handa-Corrigan et al., 1989; Jobses et al., 1991; Pa- poutsakis, 1991; Trinh et al., 1994; Wang et al., 1994; Chalmers, 1994; Wu and Goosen, 1995; Michaels et al., 1996; Meier et al., 1999; Dey and Emery, 1999; Chisti, 2000). Many reports underscore the concept that cell- bubble interactions play a much more important role in physical cell damage than pure agitation (Oh et al., 1989; Kunas and Papoutsakis, 1990; Nienow et al. 1996; Michaels et al., 1996; Ma et al. 2002). In addition, it is generally accepted that cell damage does not occur while the bubble is still in the bulk liquid. This is consistent with observations that the bubble disengagement at the liquid surface is the major cell damage cause (Handa et al., 1987; Tramper et al., 1988; Orton and Wang, 1991; Orton, 1992; Jobses et al., 1991, Michaels et al.1996; Cherry and Hulle, 1992; Trinh et al., 1994). Pluronic F-68 (PF-68) is a nonionic surfactant and one of a family of triblock copolymers consisting of a hydrophobic center, poly(propylene oxide), and two hydrophilic tails composed of poly(ethylene oxide). It is the most commonly used shear protectant in animal and insect cell cultures. The protective effect of PF-68 is concentration- dependent, and a concentration in the range of 0.3-3 g/L, often determined by empirical methods, is typically used (Chisti, 2000; Murhammer and Goochee, 1990). The protective effect and mechanisms of PF-68 have been studied intensively (Murhammer and Goochee, 1988; Goldblum et al., 1990; Ramirez and Mutharasan, 1990; Zhang et al., 1992; Al-Rubeai et al., 1992; Jordan et al., 1994; Tan et al., 1994; Chattopadhyay et al.,1995a; Michales et al., 1995; Wu et al., 1997; Palomares et al., 2000). It is widely accepted that PF-68 protects cells by preventing cells from attaching to bubbles, as shown by microscopic observations and qualitative measurements (Bauer et al., 2000; Bavarian et al., 1991; Chalmers and Bavarian, 1991; Chattopadhyay et al., 1995a; Jordan et al., 1994), but there are few quantitative studies of cell- bubble interactions (Trinh et al., 1994; Michales et al., 1995; Tan et al., 1994; Wen and Tan, 1999; Hellung- Larsen et al., 2000).
In this study, the attachment of a novel human cell line to air bubbles was quantitatively investigated over a wide range of PF-68 and cell concentrations. Using a specially designed bubble column with an exit connected to a collection tube, various cell bubble interaction steps could be isolated and studied, including the isolation of cell-bubble attachment from bubble rupture. Specifical- ly, cell-bubble attachment was investigated quantita- tively as a function of cell and PF-68 concentration. Even more significantly, cell lysis was investigated quantita- tively as a function of location, foam layer versus bubble rupture and as a function of bubble residence time, and PF-68 concentration. Finally, the mechanism of PF-68 protection of cells from sparging damage is discussed in light of this new information.
Materials and Methods
Cells and Maintenance. PER.C6 cells, which were derived from embryonic retinoblast cells transfected with the adenovirus type 5 E1 gene (Fallaux et al., 1998), were used in this study. Having been adapted to serum-free suspension culture, PER.C6 cells were routinely main- tained in a bioreactor (B. Braun Biotech, Allentown, PA) with 2 L working volume in a serum-free medium containing 0.3 g/L PF-68 with approximately 10 mg/L total protein. The cell concentration in the bioreactor was maintained at 0.7-1.5 × 106 cells/mL by replenishing part of the culture with fresh medium daily. Prior to bubble column experiments, PF-68 present in the culture was removed by multiple cycles of low speed centrifuga- tion (250 × g for 10 min) and resuspension of the pelleted cells into a PF-68 free medium. Next, the cell concentra- tion was adjusted to the target value, and just prior to the experimentation, concentrated PF-68 solution was spiked to adjust the final PF-68 concentration. These preparations did not negatively affect cell viability.
Improved Bubble Collector Device. A glass “bubble collector” (Figure 1) was designed and fabricated specif- ically for the cell-bubble interaction studies. It is an improved version of the bubble columns previously developed by others (Michaels et al., 1995; Tan et al., 1994; Wen and Tan, 1999). The device is composed of two parts: a main bubble column (20 mm diameter × 200 mm height) and a bubble exiting tube. The top portion of the bubble column necking down from the main column to the exit tube is conical.
Figure 1. Schematic design of the bubble collector used in this study. The main body is 20 mm (inner diameter) × 200 mm (height), the conical region is 3-20 mm (inner diameter of the upper and lower regions) × 15 mm (height), and the effluent tube is 3 mm (inner diameter) × 20 mm (length).
The improved bubble collector was designed and oper- ated in the following ways so that well-controlled single- variable studies could be quantitatively conducted:
1. Bulk liquid overflow from the column to the collec- tion tube was prevented; consequently, the liquid col- lected in the tube was all from collapsed/ruptured bubble liquid film.
2. Low gas flow rate was used to allow accurate counting of the total number of bubbles sparged so that the results could be normalized on a single-bubble basis. Under these conditions, there was no visible bubble rupture in the bubble column. Bubble rupture only occurred at the end of the exiting tube and in the collection tube. This allowed the isolation of foam forma- tion from bubble rupture events; bubbles exited the column close to plug-flow (per visual observation) so that the residence time of each bubble in the foam layer did not deviate significantly.
Experimental Procedures. The bubble column was completely filled with cell suspension (approximately 60 mL) through the open bottom, after which the bottom was plugged with a rubber stopper. An automatic syringe pump (PHD 2000, Harvard Apparatus, Holliston, MA) was used to generate single air bubbles by a syringe needle (16 gauge) inserted through the stopper. The gas flow rate was controlled at 5.0 mL/min by the automatic syringe pump. As a result of the small diameter of the column, the rising force of the bubbles was sufficient to maintain the cells as a homogeneous suspension.
In all experiments (except the residence time experiments), a 1.8 mL foam layer was built up before bubbles coming out the exit tube were collected into the 15 mL centrifuge tube. The small opening and the depth of the 15 mL centrifuge tube minimized the loss of bubble film liquid due to splashes from bubble breakup. A total of 1.5-2.0 mL of the bubble film liquid, representing about 3% of the bulk liquid, was collected before each experi- ment was terminated. As a result, the liquid level in the column was reduced slightly and the headspace was increased to approximately 3.3-3.8 mL. Since the gas sparging flow rate was kept constant at 5.0 mL/min, the mean residence time of the bubbles in the foam layer was approximately 30-34 s before exiting the column. Upon spherical with a uniform liquid film thickness, the volume of the liquid surrounding each bubble can be calculated by subtracting the volume of the gas bubble from the total volume (liquid film and gas): completion of the experiment, the bulk liquid and the bubble film liquid were sampled for analysis.Analytical Methods. The volume of the cell suspension in the bubble column was measured before and after the experiment. The total volume of gas sparged was measured using the syringe pump, the total number of bubbles was counted, and the elapsed time recorded. The total volume of bubble film liquid collected in the 15 mL centrifuge tube was measured with an analytical balance where η is the volume of the liquid film surrounding a bubble, db is the diameter of the bubble, and δ is the theoretical liquid film thickness. One can directly solve for δ by rearranging eq dehydrogenase (LDH) assays (CytoTox 96 assay kit, Promega, Madison, WI), while trypan blue exclusion and supernatant LDH measurements were used for viability and/or cell lysis determination. Typically, two hemocy- tometer measurements were made, and if the measure- ments were off by 15% or more, a third count was taken and the three were averaged. Three independent LDH measurements were made and averaged for both total and supernatant LDH measurement. The standard It should be noted that the liquid film thickness between two neighboring bubbles is assumed to be twice that of a single bubble film thickness.Enrichment Factor. The normalized cell concentra- tion, i.e., the cell concentration in the bubble film relative to the cell concentration in the bulk cell suspension, referred to as the cell enrichment factor, is defined by the following equations: where Xt,bulk,i is the total cell concentration measured with hemocytometer in the bulk liquid in the column at the beginning of the experiment, LDHbulk,w,i and LDHbulk,sup,i are the LDH absorptions of the whole broth (after cell lysis) and the supernatant at the beginning of the experiment, respectively. The cell concentrations in a sample can then be calculated with the conversion factor based on the LDH absorption measured in the whole broth, LDHw: Xt,LDH ) (LDHw – LDHbulk,sup,i)F (2 ) Bubble Liquid Film Thickness. The mean theoreti- cal thickness of the liquid film surrounding a bubble was calculated from the total bubble film liquid collected and the number of bubbles sparged. Assuming the bubble is 1993; Chisti, 2000), and (2) bubble number could be accurately counted and a foam layer, which maintained a constant height, could be generated in the bubble column.
Effects of PF-68 Concentration on Cell-Bubble
Interactions. The effects of PF-68 concentration were investigated using cells from an exponentially growing bioreactor culture at a concentration of ∼0.8 × 106 cells/ mL. All other experimental conditions were held con- stant.Figure 2a presents the cell concentration in the col- lected bubble film liquid and the bulk liquid before and after sparging, at PF-68 concentrations between 0.001 and 3 g/L. It is apparent that the PF-68 concentration had a profound effect on the cell concentration in the bubble film. At a low PF-68 concentration (<0.1 g/L), the cell concentration in the bubble film was substantially higher than that in the bulk, indicating high affinity between bubbles and cells, which is consistent with previous studies. For example, at 0.001 g/L PF-68, only a small fraction (∼2%) of the initial bulk volume was taken out of the bubble column by the bubbles; however, under these current conditions, about 27% of the total cells were included in this small portion of liquid (Figure 2b). As a result, the cell concentration in the bulk decreased substantially. Conversely, the cell concentra- tion in the bubble film was actually lower than that in the bulk liquid at PF-68 concentrations higher than 0.1 g/L. Figure 2. (a) Cell concentrations in the bulk solution and in the collected foam liquid before and after sparging. (b) The percentages of cells and liquid carried out of bubble column by bubbles at different PF-68 concentrations. There is a general agreement between the cell concen- trations measured by hemocytometer (solid lines) and LDH assays (dotted lines). However, LDH assays nor- mally gave a higher cell concentration than the measure- ments using the hemocytometer. Consistent with this variation between hemocytometer and LDH assay, the LDH assay indicated more cells were carried out of the bubble column than that based on hemocytometer mea- surements (Figure 2b). Mass balances (cell balances) were conducted with data from both LDH and hemocy- tometer assays. On average, 10% of the cells were missing after sparging based on hemocytometer measurements, while the deviation is only half the value with LDH assay. This discrepancy is most probably the result of cell lysis. Because of the inability of hemocytometer to count the lysed cells, it is concluded that the LDH assay is more accurate in quantifying the number of cells attached to or damaged by each bubble. At PF-68 concentrations between 1 and 3 g/L, even though the cell concentration in the collected bubble liquid was lower than that in the bulk liquid, ap- proximately 1% of the cells were removed from the bulk in about 30 min (Figure 2b). If it is assumed that in a bioreactor all of the cells attached to bubbles and trapped in the foam are eventually lysed (which will be shown later to be a function of the bubble residence time in the foam layer), the cell death rate would be roughly 0.020 h-1. This is significant compared to a typical cell growth rate of 0.029 h-1 (doubling time of 24 h). Figure 3. The effects of PF-68 concentration on (a) the bubble film liquid thickness and the number of cells attached on each bubble and (b) the cell enrichment factor. Figure 3a presents the calculated film thickness at different PF-68 concentrations. At a PF-68 concentration of 0.001-0.01 g/L, the film thickness was approximately 16-20 µm. As PF-68 increased from 0.01 to 0.1 g/L, the liquid film thickness increased significantly until it reached a plateau around 40 µm at 0.1-1 g/L PF-68 concentration. It should be noted that the film thickness between two neighboring bubbles is twice that of a single bubble film thickness. Since the diameter of PER.C6 cells is only 15 µm, the bubble film has sufficient volume to contain cells without attaching to the air-liquid inter- face. Cell attachment (CA), defined as the number of cells associated with each bubble, is also presented in Figure 3a. At 0.001 g/L PF-68, there were approximately 1000 cells associated with each bubble as measured by hemocy- tometer. This number decreased monotonically to ap- proximately 120 cells per bubble at 3 g/L PF-68, an 8-fold reduction. A regression analysis of the number of cells per bubble versus the PF-68 concentration on a log-log scale (Figure 3a), results in the following relationship: log[CA(XB ) 0.8, CPF-68)] ) 2.21 - 0.27 log CPF-68 (7 ) where XB is the bulk cell concentration (106 cells/mL),CPF-68 is the PF-68 concentration (g/L), and CA is the cell attachment (cells per bubble).Figure 3a also indicates that in all but one case, the number of cells attached to bubbles based on LDH measurements was noticeably higher, implying cell lysis. Finally, it should be noted that cell attachment is a function of multiple factors, including bulk cell concen- tration, PF-68 concentration, bubble size, and bubble residence time in the foam layer. The observation that the thickness of the liquid film was greater at a high PF-68 concentration but signifi- cantly fewer cells were associated with the bubbles implies that PF-68 is effective at either precluding cells from attaching to bubbles or facilitating cell drainage from the foam. This observation is consistent with qualitative observations made by others. Chattopadhyay et al. (1995a) and Garcia-Briones and Chalmers (1992) microscopically observed the top portion of a single bubble emerging from a liquid surface. A large number of insect cells were attached to the bubble when PF-68 or other protective additives were absent, but with the addition of 1 g/L PF-68 the cell attachment was very low. The cell enrichment factors measured by both hemocytometer and LDH are presented in Figure 3b on a log- log scale. As the PF-68 concentration increased, the enrichment factor decreased and reached a value of unity at a PF-68 concentration of approximately 0.08 g/L. This decrease continued until a value of 0.5 was achieved, at which point no further decrease was observed. This is consistent with the thermodynamic argument suggested by Chattopadhyay et al. (1995b). Specifically, Chatto- padhyay et al. (1995b) suggested that once the medium- air interfacial tension is lowered below a critical value, which is a function of the surface tension of the cell-air, cell-medium, and medium-air interfaces, it becomes thermodynamically unfavorable for a cell to attach to a gas-medium interface. Further lowering of the medium- air interfacial tension will not change this unfavorable thermodynamic condition. Effects of Cell Concentration on Cell-Bubble Interaction. Prior studies on the effect of PF-68 on cell-bubble interactions typically tested one cell concentra- tion. However, unpublished anecdotes exist that the effectiveness of PF-68 decreases with increased cell concentration. Consequently, cell-bubble attachment was investigated at two PF-68 concentrations over cell concentrations ranging from 0.1 to 10 × 106 cells/mL. The calculated bubble film thickness versus cell concentration is presented in Figure 4a. At 1 g/L PF-68, the average film thickness was approximately 50 µm, inde- pendent of the bulk cell concentration. Conversely, at a lower PF-68 concentration of 0.03 g/L the film thickness increased with the bulk cell concentration (solid lines in Figure 4a are hand-drawn trends lines). At lower cell concentrations, each bubble carries significantly less liquid, consistent with the data presented in Figure 3a. The calculated cell attachment per bubble is presented in Figure 4b. At both PF-68 concentrations of 0.03 and 1 g/L, the number of cells associated with each bubble increased linearly with the bulk cell concentration on a log-log scale with a similar slope. Linear regressions after log-transforming the data results in eqs 8 and 9 (at PF-68 concentrations of 0.03 and 1 g/L respectively): log[CA(XB, CPF68 ) 0.03 g/L)] ) 2.8 + 0.9 log XB (8) log[CA(XB, CPF-68 ) 1 g/L)] ) 2.3 + 0.9 log XB (9) Although a very large number of cells were attached to a bubble at low PF-68 concentrations and high bulk cell concentrations, cell attachment did not appear to plateau even at the highest bulk cell concentration of 47 × 106 cells/mL. At 47 × 106 cells/mL (and 0.03 g/L PF-68), there were approximately 25,000 cells associated with each bubble. If one assumes that the bubble film liquid, which has a calculated volume of 0.39 µL per bubble, had the same cell concentration as the bulk liquid, i.e., 47 × 106 cells/mL, one would expect 18,000 cells in each bubble’s film liquid. The excess, approximately 7,000 cells, is most likely due to cells that have been adsorbed on the gas-liquid interface. A rough estimate indicates that a monolayer of approximately 40,- 000, cells (assuming a 15 µm diameter) could be packed on the surface of a 1.6 mm bubble. Consequently, significantly more cells could still theoretically adsorb to the bubble surface. Figure 4. The effects of bulk cell concentration on (a) the bubble film thickness, (b) the number of cell carried by each bubble, and (c) the cell enrichment factor. The cell enrichment factor was affected by the bulk cell concentration, in addition to the PF-68 concentration, as presented in Figure 4c. At a PF-68 concentration of 1 g/L, the enrichment factor was about 0.5 and relatively constant at all bulk cell concentrations. However, at 0.03 g/L, the enrichment factor decreased with increasing cell concentration until it approached approximately 1.5. At the same conditions of 0.8 × 106 cell/mL and 0.03 or 1 g/L PF-68, Figure 4c agrees with Figure 3b. It should be noted that this expression was obtained for a constant bubble size of 1.6 mm and bubble residence time around 30-34 s in the foam. Figure 5 shows eq 11 and the experimental data used to determine the cor- relation.Effect of Residence Time on Cell-Bubble Attach- ment. Equation 11 was obtained from experiments in which the bubble residence time was approximately 30- 34 s. To investigate the potential effects of the residence time of bubbles in a foam layer on cell-bubble attach- ment, the volume of the foam headspace was varied while maintaining a constant sparge rate of 5.0 mL/min (Figure 6). Compared to a foam layer in most bioreactors, the bubbles collected at shorter residence time resemble those close to the bottom of a foam layer, while the bubbles with longer residence time resemble those close to the top of a foam layer. Figure 7a indicates that the bubble film thins as residence time is increased. The thinning rate is almost identical at PF-68 concentrations between 0.1 and 1 g/L but is much more rapid at the lower PF-68 concentration of 0.01 g/L. This implies that the foam layer is stabilized more at higher PF-68 concentrations. The right end of each line in Figure 7a represents the point at which bubbles would burst before exiting the headspace; hence the lines end at a specific calculated bubble film thickness (as opposed to dropping to zero). Figure 7b presents the number of cells associated with each bubble for the same set of experiments presented in Figure 7a. A number of notable observations can be made with respect to this Figure. First, arrow 1 indicates that an increase of PF-68 concentration from 0.01 to 1 g/L decreases the number of cells attached to each bubble from approximately 2000 to less than 800. Second, as the residence times increases, the number of cells associated with each bubble drops, most significantly at the highest PF-68 concentrations, as indicated by arrow 2. This indicates that allowing cells to drain from the foam layer also plays a significant role in cell protection. LDH Release Measurement as an Indicator of Cell Damage. Figure 8 presents the effect of sparging at different PF-68 concentrations on supernatant LDH concentration. In this figure, all of the data were obtained in experiments conducted at a constant bulk cell concentration (0.8 × 106 cells/mL). Before sparging, the super- natant LDH in the bulk liquid was low, which was consistent with the high viability of the cells used in the experiments. After sparging, elevated supernatant LDH was observed in the collected foam liquid at all PF-68 concentrations; however, the difference was significantly elevated at PF-68 concentrations below 0.03 g/L. Super-more cells were lysed in the bubble column (where only foam was observed) than in the collecting tube (where all bubbles eventually ruptured). At first glance, when compared to Figure 8, one might think that an error exists since the supernatant LDH concentration in the bubble film liquid was much higher than that in the bulk liquid. However, the volume of the bubble film liquid at the end of an experiment was only about 3% of that in the bubble column. When the cell damage was examined over a wide range of bulk cell concentrations (0.12-47 × 106 cells/mL), the general trend was that higher bulk cell concentration resulted in more cell damage. Figure 5. A three-dimensional plot of the number of cells associated with each bubble as a function of cell concentration (cells/mL) and PF-68 concentration. The dots indicate experi- mental data, and the surface represents the estimation given by eq 11. Figure 6. The utilization of bubble collector to study the liquid and cell content in different layer of the foam in bioreactors. ficially, this observation is expected; PF-68 is known to protect cells from sparging. However, the supernatant LDH in the bulk also increased at the lower PF-68 concentration. In these experiments, the “bubble film liquid” was obtained from the foam, and these experi- ments were operated in such a manner that no bubbles ruptured in this foam. These results indicate significant cell lysis in the foam layer itself. Similar results were obtained in the experiments conducted at different bulk cell concentrations and two PF-68 concentrations: 0.03 and 1 g/L (data not shown). Estimation of the Number of Cells Lysed per Bubble. The number of cells lysed per bubble from sparging was estimated from the LDH release data by using the conversion factor (F) defined previously. At a cell concentration of 0.8 × 106 cells/mL and high PF-68 concentrations (g0.3 g/L), on average less than 20 cells were damaged by each bubble in either the bubble column or the collecting tube (Figure 9). However, the total number of cells lysed increased by over an order of magnitude as the PF-68 concentration was reduced. Figure 9 presents a somewhat surprising result in that Figure 7. The effect of residence time on (a) the bubble film thickness and (b) the number of cell carried by each bubble. Figure 8. LDH release into the culture medium as a result of sparging. Percentage of Cells Lysed from Bubble Rupture. Figures 8 and 9 indicate a surprising result: at PF-68 concentrations below 0.1 g/L, significantly more LDH was released as a result of cells only being trapped in the foam layer (where it is assumed that no rupture occurred) than from the bubble rupture as the bubble emerges from the exit tube and into the centrifuge tube. Another interest- ing result, as shown in Figure 10a, is that most cells being carried out of the bubble column will survive the eventual bubble burst event. Specifically, at a cell concentration of 0.8 × 106 cells/mL the total number of cells attached to a bubble decreased with increasing PF- 68 concentration. The percentage of these attached cells that were lysed as a result of the bubble exiting the exit tube was initially low (right y-axis), around 5%, and then increased to 21% at a PF-68 concentration of 0.03 g/L and then significantly decreased to only a couple of percent at high (g1 g/L) PF-68 concentrations. At this point it is not known what caused the peak around 0.03 g/L PF-68. Figure 9. The number of cells damaged by each bubble in the bubble column (due to cell-bubble interactions) and in the collecting tube (due to bubble rupture). Figure 10. The percentage of cells attached to bubbles being killed during bubble rupture at various conditions: (a) at different PF-68 concentrations and (b) at different bulk cell concentrations. Figure 10b presents the percentage of the carried out cells damaged by bubble rupture as a function of cell concentration at two different PF-68 concentrations, 0.03 and 1 g/L. Consistent with current understanding, the percentage of cells damaged as a result of bubble rupture remained low at a PF-68 concentration of 1 g/L but varied between 20% and 40% for the experiments conducted at a PF-68 concentration of 0.03 g/L. Discussion Over the years, numerous studies have been conducted to obtain a better understanding of the damage mecha- nism of mammalian cells from gas sparging. In general, it is generally understood that bubble rupture is a major factor in cell damage (Cherry and Hulle, 1992; Bavarian et al., 1991; Chalmers and Bavarian, 1991; Jobses et al., 1991; Michaels et al., 1996; Wu and Goosen, 1995). This study confirms that bubble rupture results in lethal cell damage but also suggests that cell death also occurs prior to bubble rupture. In fact, the data presented in Figure 9 indicate that a majority of the cell lysis occurs when cells are trapped in the foam layer where no observable bubble rupture was observed. Two potential sources of this damage are the shear force in draining liquid films (or lamellae) in foams (Handa-Corrigan et al., 1989) and forces, possibly resulting from surface tension, affecting the cells on the air-medium interfaces (Jordan et al., 1994). A second, somewhat surprising result is that while it is well-known that PF-68 inhibits cell-bubble attach- ment, at high cell concentrations (on the order of 107 cells/ mL) significant attachment or entrapment still occurs in the foam layer (Figure 4b). This will not be surprising if one compares the size of bubble film thickness with cell size. At PF-68 concentration of 1 g/L, the film liquid between two medium-air interfaces can be around 100 µm wide. This is significantly larger than the average diameter of PER.C6 cells, 15 µm. Therefore, even though PF-68 is able to preventing cells attaching to the medium- air interface, it cannot prevent cells from staying in the film liquid without attaching to either interface. In addition, there will be more cells in the film liquid at higher bulk cell density. A third important observation is the importance of the residence time of bubbles in the foam layer, especially at high PF-68 concentrations, with respect to the number of cells associated with a bubble (Figure 7b). Earlier studies with PF-68 noted the presence of a “stable foam layer” and the potential importance of that layer with respect to cell protection. These current results seem to confirm the importance of a foam layer, which allows cells to drain back into the bulk cell suspension. Acknowledgment The authors wish to thank Mr. Martin King, Mr. Joseph Peltier, and Mr. Christian Metallo for their assistance in experimentation.