Rapid Early Process Development Enabled by Commercial Chemically Defined Media and Microbioreactors

Rapid early bioprocess development is crucial to the timely regulatory filing for biologics. Tight timelines often leave a narrow space for early process development. It is typical to spend a considerable amount of time and resources in late-stage development to achieve a higher titer and improve the manufacturing process. A relatively high titer process in the early stage of a program enables rapid downstream and analytical development and reduces late-stage effort on process improvement and comparability studies.

In mammalian cell culture-based production, protein titer and quality attributes are largely dependent on the cell line and media used. Chemically-defined feeds have been used to replace undefined feed components such as soy hydrolysates and yeastolate in recent years to eliminate potential variability and improve productivity in cell culture processes (1). From an operational perspective, simplicity of preparation and minimum impact of media delivery on bioreactor controls (such as pH) are some key parameters that should be considered. In this paper, the authors describe their efforts in screening a few novel, commercially available, chemically defined feed media in early process development to rapidly achieve desired titer and quality.

All feed media screened are chemically defined and simple in preparation. Within three months, initial screening studies were performed in microbioreactors and top conditions were confirmed in 3-L bioreactors. A desired titer of >4 g/L was obtained with little development effort in a typical 14-day fed-batch process for two cell lines tested. Combining the process with high inoculum density, a titer of approximately 5 g/L was achieved in 14 days. Scale-up to the 200-L scale showed comparable performance to 3-L scale. These results demonstrate the possibility of using commercially available media as a generic option to achieve high titer in early process development under tight timelines.

MATERIALS AND METHODSCell line and inoculum expansion
Two recombinant Chinese hamster ovary (CHO) cell lines (A and B), developed in-house (Merck Research Laboratories) were used in the study. Both were glutamine synthetase (GS) cell lines designed to produce different monoclonal antibodies. The inoculum train started from vial thaw and was expanded in shake flasks at increased volume. The flasks were maintained in 5% CO₂ incubators at 36.5 °C. Wave bioreactors (10 L and 50 L, GE Healthcare) were used in inoculum expansion. 

Media preparation
Chemically defined Chinese hamster ovary (CD–CHO) medium (Life Technologies) was used as basal medium in inoculum expansion and in the production bioreactor. Commercially available feed media included: EfficientFeed A+ AGT, EfficientFeed B+ AGT, EfficientFeed C+ AGT (here abbreviated as A+, B+, and C+), and FunctionMax TiterEnhancer (here abbreviated as Fmax), were evaluated. All media evaluated were sourced from Life Technologies. Feed media were reconstituted from powder to different concentrations (1–3X) and were sterilely filtered through a 0.22 µm filter. 

Bioreactor operations
Ambr15 and ambr250 bioreactors (TAP Biosystems, part of Sartorius Stedim) were used in the initial media screening process. 3-L glass bioreactors (Sartorius Stedim) were used in confirmation and further optimization. Bioreactors were inoculated in the range of 0.5–8 × 10⁶ cells/mL and bolus-fed for 14-day cultivation. High inoculum densities were generated using either ATF-2-based (Repligen) perfusion or centrifugation. 200-L Xcellerex single-use bioreactors (SUB) (XDR200, GE Healthcare) were used in the scale-up evaluation. Dissolved oxygen (DO) was controlled at 30% using pure oxygen. pH was controlled using CO₂ and 1N NaOH (0.5N for ambr15). For the scale-up run with high inoculum density, an ATF-4 system with a 0.22 µm filter was used. 

Cell-culture analysis
Viable cell density (VCD) and viability were measured using Cedex systems (Roche Diagnostics). Offline pH, pO₂, and pCO₂ were measured using an ABL80 blood gas analyzer (Radiometer). Glucose, lactate, glutamine, and glutamate were measured using a YSI Model 2700 analyzer (YSI Life Sciences). Ammonia, electrolytes, and osmolality were measured using a Nova FLEX bioanalyzer (Nova Biomedical). Titer was analyzed using a reverse-phase high-performance liquid chromatography (RP–HPLC) method. 

RESULTS AND DISCUSSIONInitial screening using ambr microbioreactors (cell line A)
Ambr15 microbioreactors have been routinely used in clone selection and media screening activities. The ambr250 was recently introduced to increase the capability of high-throughput development in an environment that better mimics that in stirred tank bioreactors (2). Both ambr15 and ambr250 were used for media screening with cell line A. Peak VCD (PVCD), final viability, peak lactate, and final titer were selected as the main evaluation parameters.

The impact of individual feeds and combinations of feeds was evaluated using the ambr15. For initial screening, 1X A+, 1X B+, and 1X C+ were fed either individually or in combination (at equal volume). Total feed volume was 48% of initial bioreactor volume for all conditions. All feed volumes in this paper are referring to the percentage of the initial bioreactor volume. Figure 1 shows PVCD, final viability, peak lactate, and final titer results for different media conditions. For single-feed conditions, 1X B+ was the best in promoting cell growth and sustaining viability. As a result, it supported the highest titer. Cultures with 1X A+ or 1X C+ had lower PVCD, higher lactate production, and lower final viability and titer. Combining 1X A+ or 1X C+ with 1X B+ (24% each) led to improved PVCD and final viability, as well as lower peak lactate compared with the corresponding individual feeds. Both 1X A+/1X B+ and 1X C+/1X B+ produced titers close to 2 g/L on day 14.

Concentrated feeds have been used to improve productivity and overall yield (3, 4). Another experiment on the ambr15 was performed to evaluate the effect of B+ and C+ concentration (02X) using a central composite design, Design Expert 9 (Stat-Ease, Inc.). Figure 2 shows the contour plots of PVCD, final viability, peak lactate, and titer using B+ and C+ at 02X concentrations. Regardless of concentration, feeding B+ resulted in a more desired culture performance than C+. Concentrated B+ (2X) improved overall culture performance. Based on the ambr15 results, it was decided to further evaluate B+ and the B+/C+ combination on the ambr250. The A+/B+ combination was not pursued further since the peak lactate level in the A+/B+ combination was higher than that in other conditions, as shown in Figure 1, which could potentially pose scale-up issues, especially in fed-batch processes.

The following conditions were evaluated in the ambr250: feeding schedules, 13X B+, and B+/C+ combinations. Three feeding schedules were evaluated using 3X B+: 1) day 0, 1, 4, 7, 10; 2) day 1, 4, 6, 8, 11; 3) day 1, 3, 5, 7, 11. Sensitivity of culture performance to feeding schedules, particularly glucose/lactate metabolism and viability, was observed. Table I lists three feeding schedules used with 3X B+, with the same feed volume used. Feeding schedule 2 for 3X B+ led to the highest titer among the three tested. The other schedules generated higher lactate, lower viability, and lower titer. It is possible that overfeeding of 3X B+ early in the culture could have contributed to the higher lactate production and lower viability because 3X B+ is rich in glucose. Overfeeding could also explain the lower titer observed (Figure 3) when the combination of 30% 3X B+/30% 2X C+ was used. 

As shown in Figure 3, comparable bioreactor productivity (combining titer and bioreactor volume) was observed with the increase of B+ concentration from 1X to 3X when the same amount of material was used (i.e., 30% 1X B+ vs. 10% 3X B+). Thus, concentrated feeds were chosen to reduce the feed volume required in the process in future studies. There were two conditions that reached a day 14 titer of >3 g/L: 30% 3X B+ and 12% 3X B+/12% 2X C+ combination. These were evaluated in the next stage of 3-L confirmation and optimization runs. 

Culture performance in 3-L bioreactors (cell line A)
The top conditions from the ambr250 evaluation were confirmed in 3-L bioreactors. Additionally, supplementation of Fmax (6−15% of initial volume) was applied to further boost titer. Fmax has been reported to improve titer considerably (5).

Figure 4 shows the culture performance comparison between the ambr250 and 3-L bioreactors, as well as the further titer improvement from Fmax addition. The 3X B+ and 3X B+/2X C+ conditions from the ambr250 and 3-L bioreactors showed comparable results. A 3050% titer increase was reached by using Fmax supplementation in both 3X B+ and 3X B+/2X C+ conditions. The two conditions showed comparable performance. To reduce the types of feeds needed, the authors decided to focus on 3X B+ and Fmax for further development and optimization.

The impact of inoculum density on culture performance was evaluated using 3X B+ and Fmax. High inoculum density culture has been reported to shorten culture duration of a typical fed-batch process (6, 7). Figure 5 shows the impact of inoculum density (0.5, 4, and 8 × 10⁶ cells/mL) on culture performance. Although similar PVCDs of 20−25 × 10⁶ cell/mL were achieved for all conditions, a faster decline in viability was observed for the high inoculum density cultures (4 and 8 × 10⁶ cell/mL). Peak lactate levels were similar for all three inoculum densities, and lactate consumption started earlier in the 4 and 8 × 10⁶ cell/mL conditions, as expected. Titers of approximately 5 g/L were reached in 14 days for the 4 and 8 × 10⁶ cell/mL conditions, compared with an average titer of 4.3 g/L for the 0.5 × 10⁶ cell/mL condition. However, since the 0.5 × 10⁶ cell/mL condition still had high viability on day 14, it would have been expected to reach >5 g/L titer if the culture was extended for a few more days. As shown in Figure 5, higher inoculation density has the merit of shorter cultivation time for a target titer because antibody production is typically non-growth associated. 

Culture performance in 3-L bioreactors (cell line B)
To evaluate whether the conditions optimized using cell line A are applicable to other cell lines, the same feeds were tested using cell line B. The two cell lines have different characteristics for cell growth and product formation, from the authors’ previous experience. Figure 6 shows the culture performance of cell line B using 3X B+ and Fmax in 3-L bioreactors. A target inoculation density of 0.5 × 10⁶ cells/mL was used in the process. As shown in Figure 6, >4 g/L average titer was reached on day 14. Peak lactate was comparable to that in cell line A. Culture viability was still >90% on day 14, which makes primary recovery relatively easier. As mentioned for the cell line A experiments previously, high viability leaves room for culture duration extension if required for titer improvement at a later time.

Initial results from cell line B showed the possibility of using commercial feeds for other programs with no or little development work required. The generic application of the feeds implies that relatively high titer could be achieved under tight timelines. As a result of the improved productivity, smaller size bioreactors could be sufficient to provide materials for downstream and analytical development and safety assessment. For both cell lines, Protein A-purified bioreactor samples showed very low level of aggregates and desirable acidic/basic species distribution (data not shown), demonstrating appropriate quality from using the feeds screened.

Scalability
To evaluate the scalability of using 3X B+ and Fmax, scale-up of the cell line A process to a 200-L SUB was performed. A high inoculum density process including N-1 perfusion was scaled-up to evaluate N-1 perfusion at the same time. N-1 perfusion was performed using an ATF-4 on a 200-L SUB. The production bioreactor was inoculated with a target inoculation density of 4 × 10⁶ cells/mL.

Figure 7 shows the cell culture results from the 200-L SUB. Cell growth and viability trends from the 3-L bioreactors were reproduced well in the 200-L SUB. Osmolality trends also matched well between the two scales. The 200-L SUB pCO2 trend was maintained at <100 mmHg, although slightly higher than that in the 3-L bioreactors due to the scale effect. Comparable ammonia levels were seen in both scales. Importantly, glucose/lactate metabolism in the 200-L SUB was similar to that in the 3-L bioreactors, which demonstrated a good scalability of cell metabolism. The desired titer (4−5 g/L) was reached in 14 days in the 200-L SUB.

CONCLUSION
In this paper, the feasibility of rapid screening and application of commercial feeds in early process development programs is demonstrated. Top conditions were quickly identified from two rounds of ambr microbioreactors and further confirmed/optimized in 3-L bioreactors. The desired titer for a typical 14-day fed-batch process for early stage programs (>4 g/L) was obtained within three months. Increase of inoculum density (4 × 10⁶ cells/mL) showed the potential of manufacturing cadence improvement, and increased titer to around 5 g/L in 14 days. Comparable titer was achieved in a 200-L SUB. The feeds developed for cell line A were successfully applied to cell line B with little modification needed. The performance in cell lines A and B demonstrated the potential of using the commercial feeds identified in this study as generic feeds for other cell lines in early programs with no or little further development.

ACKNOWLEDGEMENTS
The authors would like to thank the bioprocess technical operations group for media preparation/bioreactor operations and the purification in-process support group for titer assay. The authors would also like to thank Mr. Shawn Barrett and Dr. Robin Ng from Thermo Fisher Scientific for kindly providing technical support and part of the raw materials for the initial evaluation.

References
1. N. Ma et al., Biotechnol. Prog. 25 (5) 1353-1363 (2009).
2. R. Bareither et al., Biotechnol. Bioeng. 110 (12) 3126-3138 (2013).
3. P. Hossler et al., Biotechnol. Prog.  29 (4) 1023-1033 (2013).
4. P. Xu et al., BioPharm Intl. 27 (6) 24-32 (2014).
5. M. Liu et al., BioProcess Intl. 12 (8) 52-61 (2014).
6. I. Padawer et al., Biotechnol. Prog. 29 (3) 829-832 (2013).
7. W.C. Yang et al., Biotechnol. Prog. 30 (3) 616-625 (2014).

ALL FIGURES ARE COURTESY OF THE AUTHORS 

About the Authors:
Sen Xu, PhD, is associate principal scientist, Balrina Gupta is associate principal scientist, Linda Hoshan is senior scientist, and Hao Chen, PhD, is director, all at Process Development & Engineering, Merck Research Laboratories.

Article Details

BioPharm International
Vol. 28, No. 2
Pages: 28-33
Citation: When referring to this please cite it as S. Xu at al., "Rapid Early Process Development Enabled by Commercial Chemically Defined Media and Microbioreactors," BioPharm International28 (2) 2015.