- Email: [email protected]
Cell density-dependent proliferation in frequently-fed peripheral blood mononuclear cell cultures
Cytotherapy (2000) Vol. 2, No. 4, 267–280
Cell density-dependent proliferation in frequentlyfed peripheral blood mononuclear cell cultures S D Patel1, WM Miller1, JN Winter2 and ET Papoutsakis1 1Department
2Division
of Chemical Engineering, Northwestern University, Evanston, IL, USA of Hematology/Oncology, Northwestern University Medical School and Robert H. Lurie Cancer Center, Chicago, IL, USA
Background
Our goal was to produce granulocyte progenitor (CFU-G) and postprogenitor (CD15 +CD11b +/-) cells for subsequent transplantation. We hypothesized that increasing the feeding frequency and maintaining constant densities may overcome inhibitory growth conditions (i.e. low pH) in high-density cultures. Methods
To study the effect of cell density on total cell expansion, differentiation and lactate production, 50% daily medium exchanges were used in cultures of peripheral blood mononuclear cells (PB MNC) maintained at constant densities (ranging from 5 # 104cells/mL to 2.5 # 106cells/mL). Results
We observed a significant increase in total cell expansion when the density was increased from 5 # 104 cells/mL to 1 # 106 cells/mL, but a further increase to 2.5 # 106cells/mL resulted in a decline in cell expansion. Increasing feeding to 90% daily exchange in cultures with
Introduction It has been hypothesized (and recently shown in clinical trials) that transplantation of large numbers of granulocyte progenitor and post-progenitor cells (+107 CD15+CD11b-/+ granulocytes/kg), as a supplement to a conventional peripheral blood apheresis product, can ameliorate the 7–14 day period of neutropenia in cancer patients following high-dose chemotherapy [1–3]. Most clinical protocols for producing these cells call for the inoculation of culture devices at relatively low cell densities (+ 104 CD34+ cells/mL), and feeding of the cultures once (or not at all) during the period of cell expansion. While this strategy minimizes the handling of cultures
2.5 # 106 cells/mL did not enhance cell expansion; nor did reducing the extent of feeding in cultures with 5 # 104 cells/mL to 10% daily exchange. We did not observe a relationship between cell density and the percentage of granulocyte progenitor and post-progenitor (CD15 +CD11b -/+) cells. While specific lactate production (qlac ) in cultures with 2.5 # 106 cells/mL was approximately 60% of those observed in lower density cultures by Day 13, this difference was largely eliminated by increasing the extent of feeding in cultures with 2.5 # 106 cells/mL. Discussion
Our results suggest that feeding rates must be adjusted according to cell density to maximize culture performance. They also suggest that cellular crowding on the culture surface can limit expansion in suspension (nonadherent) cultures. Keywords
ex vivo expansion, cell density, hematopoietic cell culture, lactate.
(and risk of contamination), this benefit is counterbalanced by the corresponding increase in the overall volume and number of culture devices required (approximately 15 1 L gas-permeable bags in the only clinical trial to successfully abrogate neutropenia [3]). An alternative (and less commonly-used) approach is to reduce the required culture volume by increasing the cell density. The frequency of medium exchange can then be adjusted to achieve the desired cell numbers. In previous studies, using peripheral blood mononuclear cells (PB MNC) and a feeding schedule of once per 10 days, we have shown that, in cultures with high CD34+ content, low seeding densities yield greater total cell expansions than
Correspondence to: Eleftherios T Papoutsakis, Department of Chemical Engineering, Northwestern University; 2145 Sheridan Road, Evanston, IL 60208, USA. © 2000 ISHAGE
267
268
S D Patel et al.
high densities [4]. It was unclear, however, whether these results were influenced by the excessive accumulation of lactic acid in the high-density cultures (+ 20 mM). We have determined that the decrease in pH associated with lactic acid accumulation inhibits cell proliferation [5]. If this is indeed the case, a higher feeding frequency might be expected to reduce the observed difference in cell expansion. Observations in our laboratory have shown that, if the feeding frequency is increased to twice in a 14 day culture period, the difference in total cell expansion between high-density (5 # 105 cells/mL) and low-density (5 # 104 cells/mL) cultures can be decreased (but not eliminated) [6]. Schwartz et al. have also noted that cell expansion in high density cultures (1–5 # 106 cells/mL) is greatly enhanced when the feeding frequency is increased from 1 volume/week to 3.5 volumes/week [7]. Koller et al. have looked at a range of densities (5 # 104 to 5 # 105 cells/mL) in combination with a range of feeding frequencies (0–7 volumes/week) in cultures of bone marrow mononuclear cells (BM MNCs) [8]. They also observed greater total cell expansion in cultures with lower inoculum densities. Similar to the other studies, they varied the cell density only at Day 0 and allowed the density to increase afterwards. In order to maximize culture performance throughout the culture period, it would be desirable to know the effect density plays at later points in the culture. BM MNC cultures possess an adherent stromal layer (which requires a surface for attachment), so the effects of cell density on proliferation might be expected to differ compared with PB MNC cultures. For example, Koller et al. has shown that BM MNC expansion ceases once a critical surface cell density is attained [9]. While available surface area may be important for stromal layer attachment in BM MNC cultures, it is not known if this parameter is also important in suspension cultures of PB MNCs (in which most of the cells are nonadherent). We aimed to produce granulocyte progenitor and post-progenitor cells. We looked again at the issue of density-dependent proliferation and differentiation using cultures of PB MNCs inoculated at four different cell densities: 5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL. Our approach differed from the previous studies in two ways: ■ First, we employed a daily 50% medium exchange protocol to maintain the pH in noninhibitory ranges for all the densities examined.
■ Second, we also readjusted the cell densities daily, to maintain them relatively constant in all four cultures for the duration of the experiment. In this manner, we can gain insight into the effects of cell density for the entire culture period. Proper control of ex vivo culture processes requires an understanding of the culture’s metabolite production, we have therefore also examined the effect of cell density on lactate formation in these studies.
Materials and methods Cell samples and liquid culture initiation Samples of PB MNCs were obtained from cancer patients following chemotherapy-induced mobilization, with or without G-CSF (Response Oncology, Memphis, TN). Samples were collected after written consent under protocols approved by the various Institutional Review Boards. Approximately 48 h after collection, the entire samples were suspended in 60 mL of human long-term medium (HLTM; a serum-containing McCoy’s 5A-based medium) supplemented with 50 ng/mL stem-cell factor (SCF, Amgen, Thousand Oaks, CA), 5 ng/mL IL-3 (Novartis, East Hanover, NJ; R&D Systems, Minneapolis, MN), 10 ng/mL IL-6 (Novartis; R&D Systems), 10 ng/mL G-CSF (Amgen), and 10 ng/mL GM-CSF (Immunex, Seattle, WA). The cell suspension was then inoculated into a T150 flask (Falcon, Lincoln Park, NJ) and incubated for 5 days at 37ºC in a fully humidified atmosphere of air and 5% CO2. The cell densities in these flasks were between 1.2–1.7 # 106 cells/mL at Day 0 and 6 # 105–1.5 # 106 cells/mL at Day 5. At Day 5, the flask contents were withdrawn and spun down at 300 g for 10 min using an IEC CL3R centrifuge (IEC, Needham Heights, MA). Approximately 50 mL of the supernatant was then removed and the cell pellet was gently resuspended in the remaining 10 mL of medium. The total nucleated cell density of the suspension was determined using a Coulter Multisizer IIe (Coulter, Hialeah, FL) after treatment with cetrimide (Sigma, St Louis, MO) to release the nuclei. Aliquots of the cell suspension were then added to 10 mL of fresh HLTM in T25 flasks (Falcon) to achieve the following cell densities: 5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL. Cell densities were confirmed using the Coulter Multisizer and cells were also prepared for colony-forming cell assays and flow cytometry as described below.
Cell density-dependent hematopoietic cell proliferation
Cell and supernatant assays Total nucleated cell counts were performed on each flask daily. Lineage distribution of the cell product was analyzed by flow cytometry using a FACScan [Becton Dickinson (BD), San Jose, CA] as previously described [10]. The following Abs were used for staining the cell samples: FITC-CD15, PE-CD11b (BD), and PE-Cy5 CD14 (Immunotech, Miami, FL). Granulocyte progenitor cells (CFU-G) were counted using previously-described methylcellulose colony assays [10]. Lactate concentrations in the medium supernatant were measured daily using a YSI Model 2700 glucose/lactate analyzer (Yellow Springs Instruments, Yellow Springs, OH). Medium pH in each flask was also measured using a Model 1306 blood gas analyzer (National Instruments, Lexington, MA).
Culture maintenance and feeding protocols In the first set of experiments, flasks were fed daily in a manner that both readjusted the cell density to the Day 5 inoculum density and replaced 50% of the culture volume, while keeping the total volume in the flask constant. For example, if on Day 6, the culture seeded with 5 # 104 cells/mL (10 mL total volume) contained 6.7 # 104 cells/mL, feeding was performed in the following manner. First, 2.5 mL of the cell suspension was withdrawn from the flask and discarded. An additional 2.5 mL was then withdrawn and spun for 10 min at 300 g. The 2.5 mL of spent medium was removed and the cell pellet was gently resuspended in 5 mL of fresh medium and growth factors. The entire suspension was then added back to the original flask, yielding a culture with a new density of 5 # 104 cells/mL and 50% of its volume (still 10 mL) exchanged. Cell counts, pH and lactate concentrations for each flask were measured after feeding to confirm their values. In the second set of experiments, 90% and 10% medium exchanges were utilized in addition to 50% medium exchanges. In cultures with 10% medium exchanges, additional G-CSF was added to obtain an equivalent concentration to that added in cultures with 50% medium exchanges.
Calculations Total cell expansions were calculated for day n according to the following formula: J CD N TCEn = (TCEn - 1) KK CD n OO L n - 1P
(1)
269
Where TCE is the total cell expansion and CDn is the pre-feeding cell density at day n and CDn-1 is the postfeeding cell density at day (n - 1). Apparent specific growth rates at day n were calculated using the following time-weighted averages: J CD N J CDn N K ln n + 1 O K ln O CDn O K CDn - 1 O Dt b KK + D t aK Dt a O Dt b O KK OO KK OO L P L P napp, n = Dt a + Dt b
(2)
Where ∆ta and ∆tb are the intervals in hours immediately after day n (tn+1 - tn) and before day n (tn – tn-1), respectively. Specific lactate production rates were calculated using a similar time-weighted average: J N J N DLb DLa O O + Dt K Dt b KK a K (CD (CDLM, a )( Dt a ) O )( Dt b ) O , b LM L P L P q lac, n = (Dt a + Dt b )
(3)
Where ∆L is the change in lactate concentration for the indicated period and CDLM is the log mean cell density for the indicated period.
Statistics/data analysis All results are presented as the mean ! SE of the mean (SEM) of replicate PB MNC samples. Statistically-significant differences are defined as p < 0.05 based on a two-tailed paired Student’s t test.
Results Culture conditions With the goal of producing large numbers of granulocyte progenitor and post-progenitor cells, we sought to determine if there were density-dependent differences in cell expansion, differentiation and metabolism in cultures performed under ‘optimal’ conditions of adequate nutrient supply and removal of lactic acid (to maintain pH in noninhibitory ranges). To this end, four cultures inoculated at different densities (5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL) were fed daily, by exchanging 50% of the medium and withdrawing cells, to maintain the cell density relatively constant for the culture duration. The cell density profiles for all four cultures are shown in Figure 1. As shown, all four densities were controlled within a narrow range of their Day 5 values for the culture duration.
270
S D Patel et al.
1.E+07
+ +
++
+ +
+ +
++
9
10
11
12
13
+
++
+ +
++
++
8
Cell Density (cells/ml)
1.E+06
1.E+05
1.E+04 50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
+ 1.E+03
5
6
2,500,000 cells/ml
7
14
Day
Figure 1. Cell densities for the indicated cultures. The mean ! SEM of seven experiments is shown.
The densities chosen were spaced over wide enough intervals to maintain separation from each other, despite the daily increases in density in each of the cultures. pH and lactate concentration (shown as [lactate]) were also measured to characterize the culture conditions in each flask (Figure 2). As expected, cultures inoculated at 5 # 104 cells/mL had the highest pH (7.25–7.35) and lowest [lactate] (3 mM) of all four cultures. Increasing the density to 2.5 # 105 cells/mL did not significantly change the pH, and only slightly increased [lactate] (3–5 mM), while larger differences in both pH and [lactate] were observed in cultures inoculated at 1 # 106 cells/mL (pH + 7.15–7.3; [lactate] + 6–10 mM). Finally, the highest density cultures (2.5 # 106 cells/mL) were found to have the lowest pH (7.0–7.25) and highest [lactate] (10–16 mM).
Cell proliferation The total cell expansions and apparent specific growth rates for all four cultures are shown in Figure 3. Cultures inoculated at 5 # 104 cells/mL exhibited the least total cell expansion at Day 14 (17-fold). Increasing the density to 2.5 # 105 cells/mL increased the cell expansion to 29fold. While cultures inoculated at 1 # 106 cells/mL proliferated the greatest (51-fold), increasing the density even further to 2.5 # 106 cells/mL resulted in a decline in cell expansion (35-fold). When apparent specific growth rates were measured, as expected, they mirrored the trends observed with cell proliferation. Furthermore, µapp did not significantly change with time in any of the cultures past Day 8, indicating that the observed differences in expansion are not a function of time (and thus, stage of
Cell density-dependent hematopoietic cell proliferation
271
50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
+
7.4
+
7.3
2,500,000 cells/ml
+
+
+
+
+
+
pH
+
+
9
10
7.2
+
7.1
+
7
8
a
+
6
+
5
+
6.9
+
+
+
7
11
12
13
14
Day
50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
18
+
12
+
+
7
+
6
+
+
10
+
[Lactate] (mM)
+
+
+
+
9
+
+
+
8
+
14
+
+
+
16
2,500,000 cells/ml
8
++ 6
4
0 b
5
10
11
12
13
14
Day
Figure 2. (a) pH and (b) lactate concentration for cultures with the indicated densities. The mean ! SEM of seven experiments is shown.
272
S D Patel et al.
50
50,000 cells/ml 250,000 cells/ml
40
1,000,000 cells/ml
+
2,500,000 cells/ml
+
Expansion
+
30
20
+
10
+ +
+
6
+
5
+
0
7
8
9
a
10
11
12
13
14
Day
Figure 3. (a) Total cell expansion and (b) apparent specific growth rates, µapp , for cultures with the indicated densities. The mean ! SEM of seven experiments is shown. For total cell expansion, statistically significant differences (p < 0.05) were observed when cultures with 5 # 10 4cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 12–14, 1 # 10 6cells/mL on 10–13 and 2.5 # 106cells/mL on Days 12–13. Statistically significant differences were observed when cultures with 1 # 10 6 cells/mL were compared with cultures having 2.5 # 105 cells/mL on Days 13–14 and 2.5 # 10 6 cells/mL on Day 9. For apparent specific growth rate, statistically significant differences were observed when cultures with 5 # 10 4 cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 7–10, 1 # 10 6 cells/mL on Days 7–13, and 2.5 # 10 6 cells/mL on Days 9 and 12. Statistically significant differences were observed when cultures with 1 # 10 6cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 10 and 12, and 2.5 # 106cells/mL on Day 8.
differentiation). There were only slight decreases in µapp between Days 8 and 13 in cultures inoculated at 5 # 104 cells/mL (1.50 versus 1.31 # 10-2 h-1, respectively) and 2.5 # 105 cells/mL (1.88 versus 1.57 # 10-2 h-1, respectively). µapp in cultures with 1 # 106 cells/mL was more time invariant, remaining close to 2 # 10-2 h-1 between Days 8–13. The highest density cultures (2.5 # 106 cells/mL) exhibited specific growth rates that fluctuated slightly more (between 1.5 and 1.9 # 10-2 h-1) than the other three densities, but these values were not statistically different from each other. Based on these results, we next sought to understand why there was an observed reduction in cell expansion as the density was increased from 1 # 106 cells/mL to 2.5 #
106 cells/mL, whereas up until that point, increases in density resulted in increases in cell expansion. As seen in Figure 2, pH and [lactate] in cultures with 2.5 # 106 cells/mL were still very close to ranges known to be inhibitory in hematopoietic cell cultures, despite the 50% medium exchange rate [5]. We thus tested the hypothesis that an increase in the extent of feeding (we chose 90% medium exchange/day) may be necessary to create optimal culture conditions. Similarly, because hematopoietic cells (particularly monocytes) are known to produce several growth factors that stimulate proliferation [11,12], it is possible that the lowest density cultures (5 # 104 cells/mL) may be adversely affected by a 50% medium exchange, which may remove these beneficial cytokines.
Cell density-dependent hematopoietic cell proliferation
273
2.5
2
+ +
+
µapp (h-1)
+
+
+
+
1.5
+
1
50,000 cells/ml 250,000 cells/ml 0.5
1,000,000 cells/ml
+
0
5
6
7
8
9
b
We thus tested a second hypothesis that reducing the extent of feeding (we chose 10%/day) is necessary to optimize conditions in these low-density cultures. The results of this set of experiments are shown in Figure 4. Unexpectedly, increasing the extent of feeding from 50% to 90% per day, in cultures with 2.5 # 106 cells/mL, resulted in a decrease in cell expansion from 54-fold to 37-fold at Day 14, despite an increase in pH of approximately 0.08 units and decrease of [lactate] by 3 mM at each day (data not shown). Furthermore, in the cultures with 5 # 104 cells/mL, decreasing the extent of feeding from 50% to 10% day decreased cell expansion from 25fold to 13-fold. We thus conclude that the two hypotheses are not likely to be valid.
Cell differentiation We observed only a small dependence of granulocyte differentiation on cell density. We did not observe significant differences (> 1%) in the percentages of granulocyte progenitor cells (CFU-G) for all four densities (data not shown). Differences in the fraction of granulocyte postprogenitor cells were somewhat more pronounced (Figure 5). The growth factor combination chosen in these experiments promoted the differentiation of granulocytes and monocytes, we therefore chose appropriate
10
11
2,500,000 cells/ml
12
13
14
Day
cellular markers to quantify immature granulocytes (CD15+CD11b- cells), mature monocytes (CD14+ cells), and mature granulocytes/mature monocytes (CD15+ CD11b+ cells). By Day 14, cultures inoculated at 5 # 104 cells/mL had a slightly smaller fraction (5–8%) of CD14+ monocytes, for both 10% and 50% daily medium exchanges. Correspondingly, the fraction of immature granulocytes (CD15+CD11b- cells) in the lowest density cultures (5 # 104 cells/mL) was also consistently 5–10% higher than all other densities, for both 10% and 50% medium exchanges between Days 7 and 14. Finally, apart from a slightly higher fraction (7–8%; p < 0.05) of CD15+CD11b+ mature granulocytes and monocytes when the extent of feeding was reduced to 10%/day in cultures with 5 # 104 cells/mL, no major differences were observed.
Specific lactate production rate The feeding protocols employed maintained glucose at high levels, therefore the amount consumed daily was of the same order as the measurement error. The only metabolic parameter we could accurately determine was the specific lactate production rate. Similarly, because the amount of lactate produced in cultures with 5 # 104 cells/mL was so small, qlac could only be accurately deter-
274
S D Patel et al.
+
50,000 cells/ml 40
50,000 cells/ml - 10%
+
2,500,000 cells/ml 2,500,000 cells/ml - 90%
+
+ +
Expansion
50
30
+ + +
20
++
10
+ +
+
++
5
+
+
0
6
7
8
9
10
11
12
13
14
Day
Figure 4. Total cell expansion for cultures with the indicated densities. Unless otherwise indicated, cultures were fed daily with 50% medium exchanges. The mean ! SEM. of 4 experiments is shown. When cultures with 5 # 104cells/mL fed by 50% medium exchanges were compared to those fed by 10% medium exchanges, statistically significant differences (p < 0.05) were obtained on Days 7–8.
mined for cultures between 2.5 # 105 and 2.5 # 106 cells/mL. As shown in Figure 6, while qlac remains close to 1 # 10-7 µmole/cell h in cultures with 2.5 # 106 cells/mL between Days 6–13, qlac rises from 1.1 to 1.4 # 10-7 µmole/cell h during this interval in cultures inoculated at 2.5 # 105 cells/mL and 1 # 106 cells/mL. Feeding the highest density culture with a 90% daily exchange rate, however, mitigated some of this difference and, at least between Days 6 and 10, increased qlac to values comparable to the other two densities.
Discussion With the aim of producing granulocyte progenitor and post-progenitor cells for subsequent transplantation, we have evaluated whether there are cell density-dependent effects on proliferation, differentiation and metabolism in PB MNC cultures, conducted under conditions of constant cell density and removal of lactic acid. It is important to note that we did not study inoculation densities, but instead focused on density-dependent effects
after a 5 day period of pre-culturing at a high cell density (+ 106 cells/mL). The effect of this pre-culture density on our results is unknown and should be studied separately to accurately assess the importance of inoculation density. Using a daily 50% medium exchange protocol, we have shown that total cell expansion at Day 14 increases as the cell density is increased from 5 # 104 cells/mL to 1 # 106 cells/mL. If the density is increased beyond this point (to 2.5 # 106 cells/mL), cell expansion begins to decline (although it is still significantly higher than that observed in cultures with 5 # 104 cells/mL). At first glance, this data is counter to the results of previous studies, which have shown that lower inoculum densities generally result in greater cell expansions than higher densities. For example, we have previously shown in cultures of PB MNCs that, if the initial CD34+ cell content is high (3.9–5.6%), inoculating cultures with 1 # 105 cells/mL more than doubles the cell expansion observed if the cultures were inoculated
Cell density-dependent hematopoietic cell proliferation
275
30
25
50,000 cells/ml - 10% 50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
% 14⫹ cells
20
2,500,000 cells/ml 2,500,000 cells/ml - 90%
15
10
5
0
7
10
a
12
14
Day
Figure 5 (a–c). Post-progenitor cell content [(a) % CD14+, (b) % CD15+CD11b-, and (c) % CD15+CD11b+ cells] in cultures with the indicated densities. Unless otherwise indicated, cultures were fed daily with 50% medium exchanges. The mean ! SEM of four experiments is shown. Comparisons of statistically-significant differences (p < 0.05) are denoted by * for cultures having 5 # 104 cells/mL; # for cultures having 5 # 104cells/mL(10%); + for cultures having 2.5 # 105cells/mL; ^ for cultures having 1 # 106 cells/mL; ◆ for cultures having 2.5 # 106 cells/mL; and ■ for cultures having 2.5 # 106 cells/mL (90%).
with 4 # 105 cells/mL [4]. These culture were fed only once, however, resulting in concentrations of lactate of approximately 20 mM in the cultures with 4 # 105 cells/mL and < 6 mM in the cultures with 1 # 105cells/mL. We have shown in other studies that, if the concentration of lactic acid approaches 20 mM (corresponding to a pH of + 6.9) in PB MNC cultures, there is significant inhibition of cell proliferation [5]. Thus, it is likely that the high-density cultures were not performed under optimal conditions in the previous studies. While our conclusion was valid for that specific feeding protocol, these experiments suggest that a more frequent feeding protocol may actually favor the higher density cultures.
Koller et al. have also looked at the effect of cell density (5 # 104 cells/mL – 5 # 105 cells/mL) for a variety of feeding frequencies (0 volumes/week – 7 volumes/week) in BM MNC cultures. While they conclude that low density cultures generally yield greater cell expansions than high density cultures, they have shown in other studies [9] that, if the surface cell density in BM MNC cultures exceeds 3 # 106 cells/cm2, cell proliferation ceases. In this regard, their density experiments can be explained by the fact that the high-density cultures had reached this limit in available surface area (thus, inhibiting expansion), whereas the low-density cultures had not. While the requirement of an available surface onto which stromal cells can attach seems reasonable for BM MNCs, there
276
S D Patel et al.
35 50,000 cells/ml - 10%
ⴙ
50,000 cells/ml
ⴙ
250,000 cells/ml
30
1,000,000 cells/ml 2,500,000 cells/ml 2,500,000 cells/ml - 90%
ⴙ
25
% 15⫹11b⫺ cells
Ⲇ
ⴙ
20
Ⲇ 15
ⴙ 10
5
0
7
10
b
12
14
Day
Figure 5. (cont.) have been no studies indicating if it is also necessary for PB MNCs, which are largely nonadherent. Our data show that, at a cell density of 2.5 # 106 cells/mL, a reduction in cell expansion is observed when compared with 1 # 106 cells/mL. A liquid-cell density of 2.5 # 106 cells/mL is equivalent to a surface-cell density of 1 # 106 cells/cm2, which represents close to complete monolayer coverage in T25 flasks, assuming a cell diameter of 10 µm. When the cells are in this close proximity, conditions at the cell surface (pH, lactate concentration, cytokine and nutrient concentrations) are likely to be very different from those in the bulk medium. This was shown by Akatov et al., who measured the local pH in Chinese hamster fibroblast cell cultures inoculated at 1 # 106 cells/cm2 as 6.5, even though the bulk pH was 7.6 [13]. In this regard, maintaining optimal bulk medium conditions will not enhance cell
proliferation as long as this inhibitory microenvironment persists. This hypothesis is consistent with the fact that increasing the medium exchange rate in cultures with 2.5 # 106cells/mL from 50%/day to 90%/day did not increase cell expansion (it actually decreased it in two of the four experiments). Thus, while available surface area may not be required for cell attachment of PB MNCs, our data suggests that it may be necessary to prevent cellular overcrowding, which leads to an inhibitory local environment. It is important to note that, in these instances, culture devices in which all the cells rest on the bottom (i.e. Tflasks, culture bags) might not be optimal for high-density cultures. Instead, a 3D, well-mixed system (such as a spinner flask or other mixed bioreactor) might be more appropriate. There are already published reports of spinner flask BM MNC, PB MNC, and UCB MNC cultures
Cell density-dependent hematopoietic cell proliferation
277
70
60
50,000 cells/ml - 10% 50,000 cells/ml 250,000 cells/ml
50
1,000,000 cells/ml 2,500,000 cells/ml
% 15⫹11b⫹ cells
2,500,000 cells/ml - 90% 40
30
20
10
0
7
10
c
12
14
Day
Figure 5. (cont.) that support cell densities of well over 1 # 107 cells/mL [14,15]. While surface area limitations might explain the decline in cell expansion in cultures with 2.5 # 106 cells/mL, the increase in expansion as density increases from 5 # 104 cells/mL to 1 # 106 cells/mL might be explained by a medium-conditioning effect. It is well known that hematopoietic cells (particularly monocytes) secrete several cytokines conducive to their proliferation [11,12]. Koller et al. have also observed an increase in cell expansion in BM MNC cultures inoculated at 3 # 104 cells/mL when the cells were grown on a pre-formed stromal layer (which secrete a variety of cytokines) [8]. The higher expansion we observed as the density was increased may reflect a greater total production of proliferation-augmenting cytokines in the higher densities. Interestingly, when we reduced the medium exchange rate in cultures with 5 # 104 cells/mL from 50%/day to 10%/day, we actually observed a decrease in average cell expansion. At this low cell density, it is doubtful that key nutrients were consumed to the point of depletion, or
that metabolites had accumulated to high concentrations in the culture fed by 10%/day (pH was still > 7.2 in cultures fed by 10% exchange/day; data not shown). Instead, it is possible that a labile nutrient had degraded over time, and was not sufficiently replenished by the 10% exchange. Although G-CSF has been shown to have a half-life of < 1 day at 37o C [16], our supplementation of extra G-CSF in the lesser-fed cultures should rule out this possibility. Glacken et al. have shown that the oxidation of serum thiols over time results in a decrease in the growth-promoting activity of serum in hybridoma cultures [17]. Only a detailed analysis of nutrient levels in our cultures can test this possibility. We observed only a slight relationship between cell density and differentiation. The tendency to preserve a greater percentage of CD15+CD11b- cells in cultures with 5 # 104 cells/mL (and, to a lesser extent, 2.5 # 105cells/mL) may be a result of higher medium pH in these cultures. Studies in our laboratory have shown that pH values of + 7.3 result in a relatively slower rate of granulopoiesis relative to lower pH cultures [18].
278
S D Patel et al.
2
1.8
1.4
1.2
+
1
+
+ +
+
+
+
q lac ⴛ 107 (µmole/cell hr)
1.6
+
0.8 250,000 cells/ml 1,000,000 cells/ml
0.4
+
0.6
5
6
2,500,000 cells/ml
7
8
9
a
10
11
12
13
14
Day
Figure 6 (a and b). Specific lactate production rates, qlac, in cultures with the indicated densities. Unless otherwise stated, cultures were fed daily with 50% medium exchanges. The mean ! SEM of (a) seven experiments; and (b) three experiments is shown. In (a), statistically significant differences (p < 0.05) were observed when cultures having 2.5 # 106 cells/mL were compared with cultures having 2.5 # 105 cells/mL on Days 8–13 and 1 # 106cells/mL on Days 8–9 and 11–13. In (b), statistically significant differences were observed when cultures with 2.5 # 106cells/mL fed by 50% exchanges were compared with 90% exchanges on Day 12. Although immature granulocytes are thought to be important for mediating the later stages of short-term engraftment (approximately Days 3–10 following the onset of neutropenia [1]), it is unlikely that the slightly higher percentage of CD15+CD11b- cells in the lowdensity cultures is enough to significantly affect the clinical outcome of patients relative to the high density cultures. Finally, we observed qlac to be lower in cultures with 2.5 # 106 cells/mL than that in cultures with 2.5 # 105 and 1 # 106 cells/mL. We have shown in other studies that qlac decreases as the medium pH decreases from 7.2 to 6.7 [5]. In these studies, the pH in cultures with 2.5 # 106cells/mL was 0.15–0.25 units lower than for the other two densities, possibly explaining the lower qlac. The fact that increasing the extent of feeding in cultures with 2.5 # 106 cells/mL from 50%/day to 90%/day increased
both pH (by approximately 0.08 units) and q lac to values comparable to those found in cultures with 2.5 # 105 cells/mL (from Days 6–11) and 1 # 106 cells/mL (from Days 6–10) is consistent with this hypothesis. Moreover, because the cells in the cultures with 2.5 # 106 cells/mL fed with 90% daily exchanges are producing more lactic acid, the local pH at the cell surface is likely to be lower than that in the same density cultures fed by 50% exchanges. This lower pH may in turn be responsible for the relatively lower cell expansion as the extent of feeding is increased. Finally, the rise in qlac we observed in cultures with 2.5 # 105 cells/mL and 1 # 106 cells/mL between Days 6 and 11 has been previously correlated with increasing percentages of mature granulocyte and monocyte post-progenitor cells, which are believed to increase their production lactic acid on a per cell basis as they mature [19].
Cell density-dependent hematopoietic cell proliferation
279
1.8 250,000 cells/ml 1,000,000 cells/ml
+ +
1.6
2,500,000 cells/ml 2,500,000 cells/ml - 90%
+
8
9
+
+
+
1.2
+
+
q lac ⴛ 107 (µmole/cell hr)
1.4
+
+
1
+ +
+
++
+
+
0.8
+
0.6
5
6
7
b
In conclusion, we have shown that by utilizing a 50% daily medium exchange rate, significant cell expansion can occur in cultures with densities as high as 2.5 # 106 cells/mL. Current perfusion systems, or fed-batch cultures, utilize uniform feeding rates for the duration of the culture, regardless of the cell density. The clinical significance of the results of this study is that, in order to maximize culture performance, feeding rates must be modified according to the cell density. Our data suggests factors such as pH (due to lactic acid accumulation), surface limitations and medium conditioning should all be taken into account, along with the cell density, to create a more appropriate feeding strategy. For example, slow perfusion rates or less extensive medium exchanges may be optimal for low-density cultures. In these studies, we looked only at the extreme case of daily medium exchanges to test this hypothesis. It is possible that a lower feeding frequency (i.e. 2–3 50%
10
11
12
13
14
Day
exchanges/week) may also yield similar cell expansions as would daily exchanges, thereby reducing the labor involved. Gradually, increasing the feeding rate is likely to be required as the cell density (and lactic acid concentration) increases. However, at some point, when the surface cell density is very high, there may be no benefit to a further increase in the feeding rate. Realizing this would prevent the wasting of expensive cytokines.
Acknowledgement We would like to thank Amgen for donation of Stem Cell Factor and Novartis for donation of IL-3 and IL-6. We are grateful to Response Oncology (especially Chet Cudak, Cathy Allen and Dr Bonnie Hazelton) for providing apheresis products. This work was supported by National Science Foundation Grants BES-9809730 and BES-9410751 and the State of Illinois Excellence in Academic Medicine Act.
280
S D Patel et al.
References 1
Nielsen LK, Bender JG, Miller WM, Papoutsakis ET. Population balance model of in vivo neutrophil formation following bone marrow rescue therapy. Cytotechnology 1998;28:157–62. 2 Scheding S, Franke H, Diehl V et al. How many myeloid post-progenitor cells have to be transplanted to completely abrogate neutropenia after peripheral blood progenitor cell transplantation? Results of a computer simulation. Exp Hematol 1999;27:956–64. 3 Reiffers J, Calliot C, Dazey B et al. Abrogation of postmyeloablative chemotherapy neutropenia by ex-vivo expanded autologous CD34+ cells. Lancet 1999;354:1092–3. 4 Guo M, Miller WM, Papoutsakis, ET et al. Ex vivo expansion of CFU-GM and BFU-E in unselected peripheral blood mononuclear cell cultures with Flt3L is enhanced by autologous plasma. Hematotherapy 1999;1:183–94. 5 Patel S, Papoutsakis ET, Winter JN, Miller WM. The lactic acid issue revisited: novel feeding protocols to examine inhibition of cell proliferation and glucose metabolism in hematopoietic cell cultures. Biotech Progr. In press. 6 Patel S. Improved substrates and feeding protocols for hematopoietic cultures [MS Thesis]. Northwestern University, 1996. 7 Schwartz RM, Palsson BO, Emerson SG. Rapid medium perfusion rate significantly increases the productivity and longevity of human bone marrow cultures. Proc Natl Acad Sci USA 1991;88:6760–4. 8 Koller MR, Manchel I, Palsson MA et al. Different measures of ex vivo human hematopoiesis culture performance are optimized under vastly different conditions. Biotech Bioeng 1996;50:505–13. 9 Oh DJ, Koller MR, Palsson BO. Frequent harvesting from perfused bone marrow cultures results in increased overall cell and progenitor expansion. Biotech Bioeng 1994;44:609–16. 10 Collins PC, Patel SD, Miller WM, Papoutsakis ET. Initiation, maintenance, and quantification of human hematopoietic cell cultures. In: Morgan JR, Yarmush ML,
11
12
13
14
15
16
17
18
19
editors. Methods in molecular medicine. Vol.18: Tissue engineering methods and protocols. Totowa: Humana Press Inc., 1998:271–92. Migliaccio A, Migliaccio G, Johnson G et al. Comparative analysis of hematopoietic growth factors released by stromal cells from normal donors or transplanted patients. Blood 1990; 75:305–12. Koller MR, Bender JG, Miller WM, Papoutsakis ET. Expansion of primitive human hematopoietic progenitors in a perfusion bioreactor system with IL-3, IL-6, and stem cell factor. Bio/Technology 1993;11:358–63. Akatov VS, Lezhnev E I, Vexler AM, Kublik LN. Low pH value of pericellular medium as a factor limiting cell proliferation in dense cultures. Exp Cell Res 1985;160:412–18. Zandstra PW, Eaves CJ, Piret JM. Expansion of hematopoietic progenitor cell populations in stirred suspension bioreactors of normal human bone marrow cells. Bio/Technology 1994;12:909–14. Collins PC, MillerWM, Papoutsakis ET. Stirred culture of peripheral and cord-blood hematopoietic cells offers advantages over traditional static systems for clinically relevant applications. Biotech Bioeng 1998;59:534–43. Martinson JA, Unverzagt K, Schaeffer A et al. Neutrophil precursor generation: Effects of culture conditions. J Hematother 1998;7:463–71. Glacken MW, Adema E, Sinskey AJ. Mathematical descriptions of hybridoma culture kinetics: II. the relationship between thiol chemistry and the degradation of serum activity. Biotechnol Bioeng 1989;33:440–50. Hevehan D, Papoutsakis ET, Miller WM. Physiologically significant effects of pH and oxygen tension on granulopoiesis. Exp Hematol 2000;28:267–75. Collins PC, Nielsen LK, Patel S D et al. Characterization of hematopoietic cell expansion, oxygen uptake, and glycolysis in a controlled, stirred-tank bioreactor system. Biotech Progr 1998;14:466–72.
Cell density-dependent proliferation in frequentlyfed peripheral blood mononuclear cell cultures S D Patel1, WM Miller1, JN Winter2 and ET Papoutsakis1 1Department
2Division
of Chemical Engineering, Northwestern University, Evanston, IL, USA of Hematology/Oncology, Northwestern University Medical School and Robert H. Lurie Cancer Center, Chicago, IL, USA
Background
Our goal was to produce granulocyte progenitor (CFU-G) and postprogenitor (CD15 +CD11b +/-) cells for subsequent transplantation. We hypothesized that increasing the feeding frequency and maintaining constant densities may overcome inhibitory growth conditions (i.e. low pH) in high-density cultures. Methods
To study the effect of cell density on total cell expansion, differentiation and lactate production, 50% daily medium exchanges were used in cultures of peripheral blood mononuclear cells (PB MNC) maintained at constant densities (ranging from 5 # 104cells/mL to 2.5 # 106cells/mL). Results
We observed a significant increase in total cell expansion when the density was increased from 5 # 104 cells/mL to 1 # 106 cells/mL, but a further increase to 2.5 # 106cells/mL resulted in a decline in cell expansion. Increasing feeding to 90% daily exchange in cultures with
Introduction It has been hypothesized (and recently shown in clinical trials) that transplantation of large numbers of granulocyte progenitor and post-progenitor cells (+107 CD15+CD11b-/+ granulocytes/kg), as a supplement to a conventional peripheral blood apheresis product, can ameliorate the 7–14 day period of neutropenia in cancer patients following high-dose chemotherapy [1–3]. Most clinical protocols for producing these cells call for the inoculation of culture devices at relatively low cell densities (+ 104 CD34+ cells/mL), and feeding of the cultures once (or not at all) during the period of cell expansion. While this strategy minimizes the handling of cultures
2.5 # 106 cells/mL did not enhance cell expansion; nor did reducing the extent of feeding in cultures with 5 # 104 cells/mL to 10% daily exchange. We did not observe a relationship between cell density and the percentage of granulocyte progenitor and post-progenitor (CD15 +CD11b -/+) cells. While specific lactate production (qlac ) in cultures with 2.5 # 106 cells/mL was approximately 60% of those observed in lower density cultures by Day 13, this difference was largely eliminated by increasing the extent of feeding in cultures with 2.5 # 106 cells/mL. Discussion
Our results suggest that feeding rates must be adjusted according to cell density to maximize culture performance. They also suggest that cellular crowding on the culture surface can limit expansion in suspension (nonadherent) cultures. Keywords
ex vivo expansion, cell density, hematopoietic cell culture, lactate.
(and risk of contamination), this benefit is counterbalanced by the corresponding increase in the overall volume and number of culture devices required (approximately 15 1 L gas-permeable bags in the only clinical trial to successfully abrogate neutropenia [3]). An alternative (and less commonly-used) approach is to reduce the required culture volume by increasing the cell density. The frequency of medium exchange can then be adjusted to achieve the desired cell numbers. In previous studies, using peripheral blood mononuclear cells (PB MNC) and a feeding schedule of once per 10 days, we have shown that, in cultures with high CD34+ content, low seeding densities yield greater total cell expansions than
Correspondence to: Eleftherios T Papoutsakis, Department of Chemical Engineering, Northwestern University; 2145 Sheridan Road, Evanston, IL 60208, USA. © 2000 ISHAGE
267
268
S D Patel et al.
high densities [4]. It was unclear, however, whether these results were influenced by the excessive accumulation of lactic acid in the high-density cultures (+ 20 mM). We have determined that the decrease in pH associated with lactic acid accumulation inhibits cell proliferation [5]. If this is indeed the case, a higher feeding frequency might be expected to reduce the observed difference in cell expansion. Observations in our laboratory have shown that, if the feeding frequency is increased to twice in a 14 day culture period, the difference in total cell expansion between high-density (5 # 105 cells/mL) and low-density (5 # 104 cells/mL) cultures can be decreased (but not eliminated) [6]. Schwartz et al. have also noted that cell expansion in high density cultures (1–5 # 106 cells/mL) is greatly enhanced when the feeding frequency is increased from 1 volume/week to 3.5 volumes/week [7]. Koller et al. have looked at a range of densities (5 # 104 to 5 # 105 cells/mL) in combination with a range of feeding frequencies (0–7 volumes/week) in cultures of bone marrow mononuclear cells (BM MNCs) [8]. They also observed greater total cell expansion in cultures with lower inoculum densities. Similar to the other studies, they varied the cell density only at Day 0 and allowed the density to increase afterwards. In order to maximize culture performance throughout the culture period, it would be desirable to know the effect density plays at later points in the culture. BM MNC cultures possess an adherent stromal layer (which requires a surface for attachment), so the effects of cell density on proliferation might be expected to differ compared with PB MNC cultures. For example, Koller et al. has shown that BM MNC expansion ceases once a critical surface cell density is attained [9]. While available surface area may be important for stromal layer attachment in BM MNC cultures, it is not known if this parameter is also important in suspension cultures of PB MNCs (in which most of the cells are nonadherent). We aimed to produce granulocyte progenitor and post-progenitor cells. We looked again at the issue of density-dependent proliferation and differentiation using cultures of PB MNCs inoculated at four different cell densities: 5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL. Our approach differed from the previous studies in two ways: ■ First, we employed a daily 50% medium exchange protocol to maintain the pH in noninhibitory ranges for all the densities examined.
■ Second, we also readjusted the cell densities daily, to maintain them relatively constant in all four cultures for the duration of the experiment. In this manner, we can gain insight into the effects of cell density for the entire culture period. Proper control of ex vivo culture processes requires an understanding of the culture’s metabolite production, we have therefore also examined the effect of cell density on lactate formation in these studies.
Materials and methods Cell samples and liquid culture initiation Samples of PB MNCs were obtained from cancer patients following chemotherapy-induced mobilization, with or without G-CSF (Response Oncology, Memphis, TN). Samples were collected after written consent under protocols approved by the various Institutional Review Boards. Approximately 48 h after collection, the entire samples were suspended in 60 mL of human long-term medium (HLTM; a serum-containing McCoy’s 5A-based medium) supplemented with 50 ng/mL stem-cell factor (SCF, Amgen, Thousand Oaks, CA), 5 ng/mL IL-3 (Novartis, East Hanover, NJ; R&D Systems, Minneapolis, MN), 10 ng/mL IL-6 (Novartis; R&D Systems), 10 ng/mL G-CSF (Amgen), and 10 ng/mL GM-CSF (Immunex, Seattle, WA). The cell suspension was then inoculated into a T150 flask (Falcon, Lincoln Park, NJ) and incubated for 5 days at 37ºC in a fully humidified atmosphere of air and 5% CO2. The cell densities in these flasks were between 1.2–1.7 # 106 cells/mL at Day 0 and 6 # 105–1.5 # 106 cells/mL at Day 5. At Day 5, the flask contents were withdrawn and spun down at 300 g for 10 min using an IEC CL3R centrifuge (IEC, Needham Heights, MA). Approximately 50 mL of the supernatant was then removed and the cell pellet was gently resuspended in the remaining 10 mL of medium. The total nucleated cell density of the suspension was determined using a Coulter Multisizer IIe (Coulter, Hialeah, FL) after treatment with cetrimide (Sigma, St Louis, MO) to release the nuclei. Aliquots of the cell suspension were then added to 10 mL of fresh HLTM in T25 flasks (Falcon) to achieve the following cell densities: 5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL. Cell densities were confirmed using the Coulter Multisizer and cells were also prepared for colony-forming cell assays and flow cytometry as described below.
Cell density-dependent hematopoietic cell proliferation
Cell and supernatant assays Total nucleated cell counts were performed on each flask daily. Lineage distribution of the cell product was analyzed by flow cytometry using a FACScan [Becton Dickinson (BD), San Jose, CA] as previously described [10]. The following Abs were used for staining the cell samples: FITC-CD15, PE-CD11b (BD), and PE-Cy5 CD14 (Immunotech, Miami, FL). Granulocyte progenitor cells (CFU-G) were counted using previously-described methylcellulose colony assays [10]. Lactate concentrations in the medium supernatant were measured daily using a YSI Model 2700 glucose/lactate analyzer (Yellow Springs Instruments, Yellow Springs, OH). Medium pH in each flask was also measured using a Model 1306 blood gas analyzer (National Instruments, Lexington, MA).
Culture maintenance and feeding protocols In the first set of experiments, flasks were fed daily in a manner that both readjusted the cell density to the Day 5 inoculum density and replaced 50% of the culture volume, while keeping the total volume in the flask constant. For example, if on Day 6, the culture seeded with 5 # 104 cells/mL (10 mL total volume) contained 6.7 # 104 cells/mL, feeding was performed in the following manner. First, 2.5 mL of the cell suspension was withdrawn from the flask and discarded. An additional 2.5 mL was then withdrawn and spun for 10 min at 300 g. The 2.5 mL of spent medium was removed and the cell pellet was gently resuspended in 5 mL of fresh medium and growth factors. The entire suspension was then added back to the original flask, yielding a culture with a new density of 5 # 104 cells/mL and 50% of its volume (still 10 mL) exchanged. Cell counts, pH and lactate concentrations for each flask were measured after feeding to confirm their values. In the second set of experiments, 90% and 10% medium exchanges were utilized in addition to 50% medium exchanges. In cultures with 10% medium exchanges, additional G-CSF was added to obtain an equivalent concentration to that added in cultures with 50% medium exchanges.
Calculations Total cell expansions were calculated for day n according to the following formula: J CD N TCEn = (TCEn - 1) KK CD n OO L n - 1P
(1)
269
Where TCE is the total cell expansion and CDn is the pre-feeding cell density at day n and CDn-1 is the postfeeding cell density at day (n - 1). Apparent specific growth rates at day n were calculated using the following time-weighted averages: J CD N J CDn N K ln n + 1 O K ln O CDn O K CDn - 1 O Dt b KK + D t aK Dt a O Dt b O KK OO KK OO L P L P napp, n = Dt a + Dt b
(2)
Where ∆ta and ∆tb are the intervals in hours immediately after day n (tn+1 - tn) and before day n (tn – tn-1), respectively. Specific lactate production rates were calculated using a similar time-weighted average: J N J N DLb DLa O O + Dt K Dt b KK a K (CD (CDLM, a )( Dt a ) O )( Dt b ) O , b LM L P L P q lac, n = (Dt a + Dt b )
(3)
Where ∆L is the change in lactate concentration for the indicated period and CDLM is the log mean cell density for the indicated period.
Statistics/data analysis All results are presented as the mean ! SE of the mean (SEM) of replicate PB MNC samples. Statistically-significant differences are defined as p < 0.05 based on a two-tailed paired Student’s t test.
Results Culture conditions With the goal of producing large numbers of granulocyte progenitor and post-progenitor cells, we sought to determine if there were density-dependent differences in cell expansion, differentiation and metabolism in cultures performed under ‘optimal’ conditions of adequate nutrient supply and removal of lactic acid (to maintain pH in noninhibitory ranges). To this end, four cultures inoculated at different densities (5 # 104 cells/mL, 2.5 # 105 cells/mL, 1 # 106 cells/mL and 2.5 # 106 cells/mL) were fed daily, by exchanging 50% of the medium and withdrawing cells, to maintain the cell density relatively constant for the culture duration. The cell density profiles for all four cultures are shown in Figure 1. As shown, all four densities were controlled within a narrow range of their Day 5 values for the culture duration.
270
S D Patel et al.
1.E+07
+ +
++
+ +
+ +
++
9
10
11
12
13
+
++
+ +
++
++
8
Cell Density (cells/ml)
1.E+06
1.E+05
1.E+04 50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
+ 1.E+03
5
6
2,500,000 cells/ml
7
14
Day
Figure 1. Cell densities for the indicated cultures. The mean ! SEM of seven experiments is shown.
The densities chosen were spaced over wide enough intervals to maintain separation from each other, despite the daily increases in density in each of the cultures. pH and lactate concentration (shown as [lactate]) were also measured to characterize the culture conditions in each flask (Figure 2). As expected, cultures inoculated at 5 # 104 cells/mL had the highest pH (7.25–7.35) and lowest [lactate] (3 mM) of all four cultures. Increasing the density to 2.5 # 105 cells/mL did not significantly change the pH, and only slightly increased [lactate] (3–5 mM), while larger differences in both pH and [lactate] were observed in cultures inoculated at 1 # 106 cells/mL (pH + 7.15–7.3; [lactate] + 6–10 mM). Finally, the highest density cultures (2.5 # 106 cells/mL) were found to have the lowest pH (7.0–7.25) and highest [lactate] (10–16 mM).
Cell proliferation The total cell expansions and apparent specific growth rates for all four cultures are shown in Figure 3. Cultures inoculated at 5 # 104 cells/mL exhibited the least total cell expansion at Day 14 (17-fold). Increasing the density to 2.5 # 105 cells/mL increased the cell expansion to 29fold. While cultures inoculated at 1 # 106 cells/mL proliferated the greatest (51-fold), increasing the density even further to 2.5 # 106 cells/mL resulted in a decline in cell expansion (35-fold). When apparent specific growth rates were measured, as expected, they mirrored the trends observed with cell proliferation. Furthermore, µapp did not significantly change with time in any of the cultures past Day 8, indicating that the observed differences in expansion are not a function of time (and thus, stage of
Cell density-dependent hematopoietic cell proliferation
271
50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
+
7.4
+
7.3
2,500,000 cells/ml
+
+
+
+
+
+
pH
+
+
9
10
7.2
+
7.1
+
7
8
a
+
6
+
5
+
6.9
+
+
+
7
11
12
13
14
Day
50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
18
+
12
+
+
7
+
6
+
+
10
+
[Lactate] (mM)
+
+
+
+
9
+
+
+
8
+
14
+
+
+
16
2,500,000 cells/ml
8
++ 6
4
0 b
5
10
11
12
13
14
Day
Figure 2. (a) pH and (b) lactate concentration for cultures with the indicated densities. The mean ! SEM of seven experiments is shown.
272
S D Patel et al.
50
50,000 cells/ml 250,000 cells/ml
40
1,000,000 cells/ml
+
2,500,000 cells/ml
+
Expansion
+
30
20
+
10
+ +
+
6
+
5
+
0
7
8
9
a
10
11
12
13
14
Day
Figure 3. (a) Total cell expansion and (b) apparent specific growth rates, µapp , for cultures with the indicated densities. The mean ! SEM of seven experiments is shown. For total cell expansion, statistically significant differences (p < 0.05) were observed when cultures with 5 # 10 4cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 12–14, 1 # 10 6cells/mL on 10–13 and 2.5 # 106cells/mL on Days 12–13. Statistically significant differences were observed when cultures with 1 # 10 6 cells/mL were compared with cultures having 2.5 # 105 cells/mL on Days 13–14 and 2.5 # 10 6 cells/mL on Day 9. For apparent specific growth rate, statistically significant differences were observed when cultures with 5 # 10 4 cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 7–10, 1 # 10 6 cells/mL on Days 7–13, and 2.5 # 10 6 cells/mL on Days 9 and 12. Statistically significant differences were observed when cultures with 1 # 10 6cells/mL were compared with cultures having 2.5 # 10 5 cells/mL on Days 10 and 12, and 2.5 # 106cells/mL on Day 8.
differentiation). There were only slight decreases in µapp between Days 8 and 13 in cultures inoculated at 5 # 104 cells/mL (1.50 versus 1.31 # 10-2 h-1, respectively) and 2.5 # 105 cells/mL (1.88 versus 1.57 # 10-2 h-1, respectively). µapp in cultures with 1 # 106 cells/mL was more time invariant, remaining close to 2 # 10-2 h-1 between Days 8–13. The highest density cultures (2.5 # 106 cells/mL) exhibited specific growth rates that fluctuated slightly more (between 1.5 and 1.9 # 10-2 h-1) than the other three densities, but these values were not statistically different from each other. Based on these results, we next sought to understand why there was an observed reduction in cell expansion as the density was increased from 1 # 106 cells/mL to 2.5 #
106 cells/mL, whereas up until that point, increases in density resulted in increases in cell expansion. As seen in Figure 2, pH and [lactate] in cultures with 2.5 # 106 cells/mL were still very close to ranges known to be inhibitory in hematopoietic cell cultures, despite the 50% medium exchange rate [5]. We thus tested the hypothesis that an increase in the extent of feeding (we chose 90% medium exchange/day) may be necessary to create optimal culture conditions. Similarly, because hematopoietic cells (particularly monocytes) are known to produce several growth factors that stimulate proliferation [11,12], it is possible that the lowest density cultures (5 # 104 cells/mL) may be adversely affected by a 50% medium exchange, which may remove these beneficial cytokines.
Cell density-dependent hematopoietic cell proliferation
273
2.5
2
+ +
+
µapp (h-1)
+
+
+
+
1.5
+
1
50,000 cells/ml 250,000 cells/ml 0.5
1,000,000 cells/ml
+
0
5
6
7
8
9
b
We thus tested a second hypothesis that reducing the extent of feeding (we chose 10%/day) is necessary to optimize conditions in these low-density cultures. The results of this set of experiments are shown in Figure 4. Unexpectedly, increasing the extent of feeding from 50% to 90% per day, in cultures with 2.5 # 106 cells/mL, resulted in a decrease in cell expansion from 54-fold to 37-fold at Day 14, despite an increase in pH of approximately 0.08 units and decrease of [lactate] by 3 mM at each day (data not shown). Furthermore, in the cultures with 5 # 104 cells/mL, decreasing the extent of feeding from 50% to 10% day decreased cell expansion from 25fold to 13-fold. We thus conclude that the two hypotheses are not likely to be valid.
Cell differentiation We observed only a small dependence of granulocyte differentiation on cell density. We did not observe significant differences (> 1%) in the percentages of granulocyte progenitor cells (CFU-G) for all four densities (data not shown). Differences in the fraction of granulocyte postprogenitor cells were somewhat more pronounced (Figure 5). The growth factor combination chosen in these experiments promoted the differentiation of granulocytes and monocytes, we therefore chose appropriate
10
11
2,500,000 cells/ml
12
13
14
Day
cellular markers to quantify immature granulocytes (CD15+CD11b- cells), mature monocytes (CD14+ cells), and mature granulocytes/mature monocytes (CD15+ CD11b+ cells). By Day 14, cultures inoculated at 5 # 104 cells/mL had a slightly smaller fraction (5–8%) of CD14+ monocytes, for both 10% and 50% daily medium exchanges. Correspondingly, the fraction of immature granulocytes (CD15+CD11b- cells) in the lowest density cultures (5 # 104 cells/mL) was also consistently 5–10% higher than all other densities, for both 10% and 50% medium exchanges between Days 7 and 14. Finally, apart from a slightly higher fraction (7–8%; p < 0.05) of CD15+CD11b+ mature granulocytes and monocytes when the extent of feeding was reduced to 10%/day in cultures with 5 # 104 cells/mL, no major differences were observed.
Specific lactate production rate The feeding protocols employed maintained glucose at high levels, therefore the amount consumed daily was of the same order as the measurement error. The only metabolic parameter we could accurately determine was the specific lactate production rate. Similarly, because the amount of lactate produced in cultures with 5 # 104 cells/mL was so small, qlac could only be accurately deter-
274
S D Patel et al.
+
50,000 cells/ml 40
50,000 cells/ml - 10%
+
2,500,000 cells/ml 2,500,000 cells/ml - 90%
+
+ +
Expansion
50
30
+ + +
20
++
10
+ +
+
++
5
+
+
0
6
7
8
9
10
11
12
13
14
Day
Figure 4. Total cell expansion for cultures with the indicated densities. Unless otherwise indicated, cultures were fed daily with 50% medium exchanges. The mean ! SEM. of 4 experiments is shown. When cultures with 5 # 104cells/mL fed by 50% medium exchanges were compared to those fed by 10% medium exchanges, statistically significant differences (p < 0.05) were obtained on Days 7–8.
mined for cultures between 2.5 # 105 and 2.5 # 106 cells/mL. As shown in Figure 6, while qlac remains close to 1 # 10-7 µmole/cell h in cultures with 2.5 # 106 cells/mL between Days 6–13, qlac rises from 1.1 to 1.4 # 10-7 µmole/cell h during this interval in cultures inoculated at 2.5 # 105 cells/mL and 1 # 106 cells/mL. Feeding the highest density culture with a 90% daily exchange rate, however, mitigated some of this difference and, at least between Days 6 and 10, increased qlac to values comparable to the other two densities.
Discussion With the aim of producing granulocyte progenitor and post-progenitor cells for subsequent transplantation, we have evaluated whether there are cell density-dependent effects on proliferation, differentiation and metabolism in PB MNC cultures, conducted under conditions of constant cell density and removal of lactic acid. It is important to note that we did not study inoculation densities, but instead focused on density-dependent effects
after a 5 day period of pre-culturing at a high cell density (+ 106 cells/mL). The effect of this pre-culture density on our results is unknown and should be studied separately to accurately assess the importance of inoculation density. Using a daily 50% medium exchange protocol, we have shown that total cell expansion at Day 14 increases as the cell density is increased from 5 # 104 cells/mL to 1 # 106 cells/mL. If the density is increased beyond this point (to 2.5 # 106 cells/mL), cell expansion begins to decline (although it is still significantly higher than that observed in cultures with 5 # 104 cells/mL). At first glance, this data is counter to the results of previous studies, which have shown that lower inoculum densities generally result in greater cell expansions than higher densities. For example, we have previously shown in cultures of PB MNCs that, if the initial CD34+ cell content is high (3.9–5.6%), inoculating cultures with 1 # 105 cells/mL more than doubles the cell expansion observed if the cultures were inoculated
Cell density-dependent hematopoietic cell proliferation
275
30
25
50,000 cells/ml - 10% 50,000 cells/ml 250,000 cells/ml 1,000,000 cells/ml
% 14⫹ cells
20
2,500,000 cells/ml 2,500,000 cells/ml - 90%
15
10
5
0
7
10
a
12
14
Day
Figure 5 (a–c). Post-progenitor cell content [(a) % CD14+, (b) % CD15+CD11b-, and (c) % CD15+CD11b+ cells] in cultures with the indicated densities. Unless otherwise indicated, cultures were fed daily with 50% medium exchanges. The mean ! SEM of four experiments is shown. Comparisons of statistically-significant differences (p < 0.05) are denoted by * for cultures having 5 # 104 cells/mL; # for cultures having 5 # 104cells/mL(10%); + for cultures having 2.5 # 105cells/mL; ^ for cultures having 1 # 106 cells/mL; ◆ for cultures having 2.5 # 106 cells/mL; and ■ for cultures having 2.5 # 106 cells/mL (90%).
with 4 # 105 cells/mL [4]. These culture were fed only once, however, resulting in concentrations of lactate of approximately 20 mM in the cultures with 4 # 105 cells/mL and < 6 mM in the cultures with 1 # 105cells/mL. We have shown in other studies that, if the concentration of lactic acid approaches 20 mM (corresponding to a pH of + 6.9) in PB MNC cultures, there is significant inhibition of cell proliferation [5]. Thus, it is likely that the high-density cultures were not performed under optimal conditions in the previous studies. While our conclusion was valid for that specific feeding protocol, these experiments suggest that a more frequent feeding protocol may actually favor the higher density cultures.
Koller et al. have also looked at the effect of cell density (5 # 104 cells/mL – 5 # 105 cells/mL) for a variety of feeding frequencies (0 volumes/week – 7 volumes/week) in BM MNC cultures. While they conclude that low density cultures generally yield greater cell expansions than high density cultures, they have shown in other studies [9] that, if the surface cell density in BM MNC cultures exceeds 3 # 106 cells/cm2, cell proliferation ceases. In this regard, their density experiments can be explained by the fact that the high-density cultures had reached this limit in available surface area (thus, inhibiting expansion), whereas the low-density cultures had not. While the requirement of an available surface onto which stromal cells can attach seems reasonable for BM MNCs, there
276
S D Patel et al.
35 50,000 cells/ml - 10%
ⴙ
50,000 cells/ml
ⴙ
250,000 cells/ml
30
1,000,000 cells/ml 2,500,000 cells/ml 2,500,000 cells/ml - 90%
ⴙ
25
% 15⫹11b⫺ cells
Ⲇ
ⴙ
20
Ⲇ 15
ⴙ 10
5
0
7
10
b
12
14
Day
Figure 5. (cont.) have been no studies indicating if it is also necessary for PB MNCs, which are largely nonadherent. Our data show that, at a cell density of 2.5 # 106 cells/mL, a reduction in cell expansion is observed when compared with 1 # 106 cells/mL. A liquid-cell density of 2.5 # 106 cells/mL is equivalent to a surface-cell density of 1 # 106 cells/cm2, which represents close to complete monolayer coverage in T25 flasks, assuming a cell diameter of 10 µm. When the cells are in this close proximity, conditions at the cell surface (pH, lactate concentration, cytokine and nutrient concentrations) are likely to be very different from those in the bulk medium. This was shown by Akatov et al., who measured the local pH in Chinese hamster fibroblast cell cultures inoculated at 1 # 106 cells/cm2 as 6.5, even though the bulk pH was 7.6 [13]. In this regard, maintaining optimal bulk medium conditions will not enhance cell
proliferation as long as this inhibitory microenvironment persists. This hypothesis is consistent with the fact that increasing the medium exchange rate in cultures with 2.5 # 106cells/mL from 50%/day to 90%/day did not increase cell expansion (it actually decreased it in two of the four experiments). Thus, while available surface area may not be required for cell attachment of PB MNCs, our data suggests that it may be necessary to prevent cellular overcrowding, which leads to an inhibitory local environment. It is important to note that, in these instances, culture devices in which all the cells rest on the bottom (i.e. Tflasks, culture bags) might not be optimal for high-density cultures. Instead, a 3D, well-mixed system (such as a spinner flask or other mixed bioreactor) might be more appropriate. There are already published reports of spinner flask BM MNC, PB MNC, and UCB MNC cultures
Cell density-dependent hematopoietic cell proliferation
277
70
60
50,000 cells/ml - 10% 50,000 cells/ml 250,000 cells/ml
50
1,000,000 cells/ml 2,500,000 cells/ml
% 15⫹11b⫹ cells
2,500,000 cells/ml - 90% 40
30
20
10
0
7
10
c
12
14
Day
Figure 5. (cont.) that support cell densities of well over 1 # 107 cells/mL [14,15]. While surface area limitations might explain the decline in cell expansion in cultures with 2.5 # 106 cells/mL, the increase in expansion as density increases from 5 # 104 cells/mL to 1 # 106 cells/mL might be explained by a medium-conditioning effect. It is well known that hematopoietic cells (particularly monocytes) secrete several cytokines conducive to their proliferation [11,12]. Koller et al. have also observed an increase in cell expansion in BM MNC cultures inoculated at 3 # 104 cells/mL when the cells were grown on a pre-formed stromal layer (which secrete a variety of cytokines) [8]. The higher expansion we observed as the density was increased may reflect a greater total production of proliferation-augmenting cytokines in the higher densities. Interestingly, when we reduced the medium exchange rate in cultures with 5 # 104 cells/mL from 50%/day to 10%/day, we actually observed a decrease in average cell expansion. At this low cell density, it is doubtful that key nutrients were consumed to the point of depletion, or
that metabolites had accumulated to high concentrations in the culture fed by 10%/day (pH was still > 7.2 in cultures fed by 10% exchange/day; data not shown). Instead, it is possible that a labile nutrient had degraded over time, and was not sufficiently replenished by the 10% exchange. Although G-CSF has been shown to have a half-life of < 1 day at 37o C [16], our supplementation of extra G-CSF in the lesser-fed cultures should rule out this possibility. Glacken et al. have shown that the oxidation of serum thiols over time results in a decrease in the growth-promoting activity of serum in hybridoma cultures [17]. Only a detailed analysis of nutrient levels in our cultures can test this possibility. We observed only a slight relationship between cell density and differentiation. The tendency to preserve a greater percentage of CD15+CD11b- cells in cultures with 5 # 104 cells/mL (and, to a lesser extent, 2.5 # 105cells/mL) may be a result of higher medium pH in these cultures. Studies in our laboratory have shown that pH values of + 7.3 result in a relatively slower rate of granulopoiesis relative to lower pH cultures [18].
278
S D Patel et al.
2
1.8
1.4
1.2
+
1
+
+ +
+
+
+
q lac ⴛ 107 (µmole/cell hr)
1.6
+
0.8 250,000 cells/ml 1,000,000 cells/ml
0.4
+
0.6
5
6
2,500,000 cells/ml
7
8
9
a
10
11
12
13
14
Day
Figure 6 (a and b). Specific lactate production rates, qlac, in cultures with the indicated densities. Unless otherwise stated, cultures were fed daily with 50% medium exchanges. The mean ! SEM of (a) seven experiments; and (b) three experiments is shown. In (a), statistically significant differences (p < 0.05) were observed when cultures having 2.5 # 106 cells/mL were compared with cultures having 2.5 # 105 cells/mL on Days 8–13 and 1 # 106cells/mL on Days 8–9 and 11–13. In (b), statistically significant differences were observed when cultures with 2.5 # 106cells/mL fed by 50% exchanges were compared with 90% exchanges on Day 12. Although immature granulocytes are thought to be important for mediating the later stages of short-term engraftment (approximately Days 3–10 following the onset of neutropenia [1]), it is unlikely that the slightly higher percentage of CD15+CD11b- cells in the lowdensity cultures is enough to significantly affect the clinical outcome of patients relative to the high density cultures. Finally, we observed qlac to be lower in cultures with 2.5 # 106 cells/mL than that in cultures with 2.5 # 105 and 1 # 106 cells/mL. We have shown in other studies that qlac decreases as the medium pH decreases from 7.2 to 6.7 [5]. In these studies, the pH in cultures with 2.5 # 106cells/mL was 0.15–0.25 units lower than for the other two densities, possibly explaining the lower qlac. The fact that increasing the extent of feeding in cultures with 2.5 # 106 cells/mL from 50%/day to 90%/day increased
both pH (by approximately 0.08 units) and q lac to values comparable to those found in cultures with 2.5 # 105 cells/mL (from Days 6–11) and 1 # 106 cells/mL (from Days 6–10) is consistent with this hypothesis. Moreover, because the cells in the cultures with 2.5 # 106 cells/mL fed with 90% daily exchanges are producing more lactic acid, the local pH at the cell surface is likely to be lower than that in the same density cultures fed by 50% exchanges. This lower pH may in turn be responsible for the relatively lower cell expansion as the extent of feeding is increased. Finally, the rise in qlac we observed in cultures with 2.5 # 105 cells/mL and 1 # 106 cells/mL between Days 6 and 11 has been previously correlated with increasing percentages of mature granulocyte and monocyte post-progenitor cells, which are believed to increase their production lactic acid on a per cell basis as they mature [19].
Cell density-dependent hematopoietic cell proliferation
279
1.8 250,000 cells/ml 1,000,000 cells/ml
+ +
1.6
2,500,000 cells/ml 2,500,000 cells/ml - 90%
+
8
9
+
+
+
1.2
+
+
q lac ⴛ 107 (µmole/cell hr)
1.4
+
+
1
+ +
+
++
+
+
0.8
+
0.6
5
6
7
b
In conclusion, we have shown that by utilizing a 50% daily medium exchange rate, significant cell expansion can occur in cultures with densities as high as 2.5 # 106 cells/mL. Current perfusion systems, or fed-batch cultures, utilize uniform feeding rates for the duration of the culture, regardless of the cell density. The clinical significance of the results of this study is that, in order to maximize culture performance, feeding rates must be modified according to the cell density. Our data suggests factors such as pH (due to lactic acid accumulation), surface limitations and medium conditioning should all be taken into account, along with the cell density, to create a more appropriate feeding strategy. For example, slow perfusion rates or less extensive medium exchanges may be optimal for low-density cultures. In these studies, we looked only at the extreme case of daily medium exchanges to test this hypothesis. It is possible that a lower feeding frequency (i.e. 2–3 50%
10
11
12
13
14
Day
exchanges/week) may also yield similar cell expansions as would daily exchanges, thereby reducing the labor involved. Gradually, increasing the feeding rate is likely to be required as the cell density (and lactic acid concentration) increases. However, at some point, when the surface cell density is very high, there may be no benefit to a further increase in the feeding rate. Realizing this would prevent the wasting of expensive cytokines.
Acknowledgement We would like to thank Amgen for donation of Stem Cell Factor and Novartis for donation of IL-3 and IL-6. We are grateful to Response Oncology (especially Chet Cudak, Cathy Allen and Dr Bonnie Hazelton) for providing apheresis products. This work was supported by National Science Foundation Grants BES-9809730 and BES-9410751 and the State of Illinois Excellence in Academic Medicine Act.
280
S D Patel et al.
References 1
Nielsen LK, Bender JG, Miller WM, Papoutsakis ET. Population balance model of in vivo neutrophil formation following bone marrow rescue therapy. Cytotechnology 1998;28:157–62. 2 Scheding S, Franke H, Diehl V et al. How many myeloid post-progenitor cells have to be transplanted to completely abrogate neutropenia after peripheral blood progenitor cell transplantation? Results of a computer simulation. Exp Hematol 1999;27:956–64. 3 Reiffers J, Calliot C, Dazey B et al. Abrogation of postmyeloablative chemotherapy neutropenia by ex-vivo expanded autologous CD34+ cells. Lancet 1999;354:1092–3. 4 Guo M, Miller WM, Papoutsakis, ET et al. Ex vivo expansion of CFU-GM and BFU-E in unselected peripheral blood mononuclear cell cultures with Flt3L is enhanced by autologous plasma. Hematotherapy 1999;1:183–94. 5 Patel S, Papoutsakis ET, Winter JN, Miller WM. The lactic acid issue revisited: novel feeding protocols to examine inhibition of cell proliferation and glucose metabolism in hematopoietic cell cultures. Biotech Progr. In press. 6 Patel S. Improved substrates and feeding protocols for hematopoietic cultures [MS Thesis]. Northwestern University, 1996. 7 Schwartz RM, Palsson BO, Emerson SG. Rapid medium perfusion rate significantly increases the productivity and longevity of human bone marrow cultures. Proc Natl Acad Sci USA 1991;88:6760–4. 8 Koller MR, Manchel I, Palsson MA et al. Different measures of ex vivo human hematopoiesis culture performance are optimized under vastly different conditions. Biotech Bioeng 1996;50:505–13. 9 Oh DJ, Koller MR, Palsson BO. Frequent harvesting from perfused bone marrow cultures results in increased overall cell and progenitor expansion. Biotech Bioeng 1994;44:609–16. 10 Collins PC, Patel SD, Miller WM, Papoutsakis ET. Initiation, maintenance, and quantification of human hematopoietic cell cultures. In: Morgan JR, Yarmush ML,
11
12
13
14
15
16
17
18
19
editors. Methods in molecular medicine. Vol.18: Tissue engineering methods and protocols. Totowa: Humana Press Inc., 1998:271–92. Migliaccio A, Migliaccio G, Johnson G et al. Comparative analysis of hematopoietic growth factors released by stromal cells from normal donors or transplanted patients. Blood 1990; 75:305–12. Koller MR, Bender JG, Miller WM, Papoutsakis ET. Expansion of primitive human hematopoietic progenitors in a perfusion bioreactor system with IL-3, IL-6, and stem cell factor. Bio/Technology 1993;11:358–63. Akatov VS, Lezhnev E I, Vexler AM, Kublik LN. Low pH value of pericellular medium as a factor limiting cell proliferation in dense cultures. Exp Cell Res 1985;160:412–18. Zandstra PW, Eaves CJ, Piret JM. Expansion of hematopoietic progenitor cell populations in stirred suspension bioreactors of normal human bone marrow cells. Bio/Technology 1994;12:909–14. Collins PC, MillerWM, Papoutsakis ET. Stirred culture of peripheral and cord-blood hematopoietic cells offers advantages over traditional static systems for clinically relevant applications. Biotech Bioeng 1998;59:534–43. Martinson JA, Unverzagt K, Schaeffer A et al. Neutrophil precursor generation: Effects of culture conditions. J Hematother 1998;7:463–71. Glacken MW, Adema E, Sinskey AJ. Mathematical descriptions of hybridoma culture kinetics: II. the relationship between thiol chemistry and the degradation of serum activity. Biotechnol Bioeng 1989;33:440–50. Hevehan D, Papoutsakis ET, Miller WM. Physiologically significant effects of pH and oxygen tension on granulopoiesis. Exp Hematol 2000;28:267–75. Collins PC, Nielsen LK, Patel S D et al. Characterization of hematopoietic cell expansion, oxygen uptake, and glycolysis in a controlled, stirred-tank bioreactor system. Biotech Progr 1998;14:466–72.