Screening to Small-Scale GMP Biomanufacture

Exploring a Simple, Unified Platform Strategy for Handling a Range of Cell-Culture Needs

The cell-line development process starts with in silico procedures combining codon-optimization algorithms, a secretion signal toolbox, flexible expression vector configurations, and high-productivity CHO-K1 cell lines.

Machine learning combined with in vitro screening is used to consider the end product and final process from the onset of the project. Process development can be done in parallel with clone development, thus reducing the number of steps and shortening the overall project time.

Screening is performed using 96-well plates in an automated platform and either transient HEK cells or stable CHO cells.

Multitron Pro incubation shakers configured for 96-well plates are used (3 mm orbit/1,000 rpm mixing). The shakers control all critical running parameters (agitation, temperature, CO2) and have active humidification to minimize evaporative losses. This ensures a consistent and reproducible growth environment from 1-mL volumes in 96-well plates to 5-L flasks.

Each shaker has a capacity of up to 80 plates (7,680 wells). This makes it possible to test large numbers of clones and running conditions in triplicates, and to generate statistically significant data in every screening pass.

Following the initial screening in 96-well plates, the most promising clones are transferred to 24-well deep-well plates and further evaluated prior to scaling up (Figure 1).

Figure 1. Productivity of the top 22 clones derived from stable transfection of a transposon-based IgG1 expression construct in HD-BIOP3 GS null CHO-K1 cells (Horizon Discovery). Cells were grown seven days in a 24-well deep well-plate under non-optimized conditions.


After initial screening and down-selection of stable CHO cell lines, the top candidates are ranked and selected for productivity. The eight best clones are then cultured in 125-mL flasks for 14 days in a Multitron Pro (Infors) with 25-mm orbit and 150-rpm agitation (Figure 2).

Figure 2. Productivity of the eight most- productive clones derived from stable transfection of a transposon-based IgG1 expression construct in HD-BIOP3 GS null CHO-K1 cells (Horizon Discovery). Cells were grown in non-optimized 14-day fed-batch 125-mL shake flask culture. The specific productivity of clonal isolates was >40 pg/cell/day.

Following down-selection, the cell culture can be scaled up further using incubation shakers. One Multitron Pro incubation shaker can hold seven Thomson 5-L flasks, each with a working volume of 2.4 L. This translates to 17 L per shaker and 50–85 g of a typical IgG1. A triple stack of Multitron Pro shakers has the capacity for 50 L of cell culture. In other words, it is possible to produce 150–255 g of antibody in a normal 10–14 day campaign using three incubation shakers. This is sufficient material for both preclinical studies and early clinical work.

Cell Culture Platform

Significant demands are placed on a platform used to develop and scale a cell line from screening in 96-well plates to 5-L flasks. Requirements change as working volumes change as the process moves from screening to development and production. For example, cell culture in 96-well deepwell plates requires far higher agitation and mixing than work in shake flasks or 24-well plates. A system for 96-well plates uses a 3-mm shaking orbit with 1,000-rpm agitation, while work in shake flasks and 24-well plates is done using a 25-mm shaking orbit and 100–150 rpm.

To compare results from different scales and across multiple instruments, growth conditions must be kept uniform, and parameters influencing cell growth and productivity controlled.

96-Well Plate Cell Culture Considerations

Figure 3. Mixing in 96-well deep-well plates using a Multitron Pro 3-mm orbit shaker (500 µm/well). Efficient mixing starts at 800 rpm and peaks at 1,000 rpm.

Mixing: Standard incubation shakers are not configured for work with 96-well plates. For efficient mass transfer and optimal cell growth in small wells, high-speed mixing and small shaking orbits are required (Figure 3). An earlier study showed that a 3-mm shaking orbit and 1,000-rpm mixing gives higher protein expression and HEK titers of as much as 3,000% higher compared with plates in standard 25-mm orbit shakers and lower speeds. The same study also showed that efficient mixing in plates does not start until 800 rpm.

Throughput and Automation:  In ATUM’s laboratory, the switch to 3-mm shakers and microtiter-plate-based screening allowed a 17-fold increase in throughput and enabled automation of protein purification, simultaneously freeing up resources and reducing variability due to human error (Table).

Table. Comparison of transient expression of human IgG1 in HEK-293 cells from 1 mL and 10 mL cultures grown under different agitation conditions. Codon optimization was performed using ATUM’s GeneGPS. Expression vectors were not optimized using  ATUM’s VectorGPS.

Evaporation: Evaporative losses must be considered because of the small volumes used in plates and the extended process times (10–14 days). Active humidification is employed to limit such losses over the course of the experiment.

Uniformity: During selection of clones grown in triplicates in as many as 80 96-well plates stacked and distributed across a large shaking surface, tightly controlled uniform conditions are of critical importance (Figure 4). The accompanying 3D temperature map of the interior of the Multitron Pro illustrates its ability to keep a consistent temperature throughout.

Figure 4. This 3D temperature map of the interior of a Multitron Pro shaker illustrates how even the temperature is all across the tray.

Scaling Up

Qualification: The Multitron Pro shaker can be qualified for validated processes. FAT and SAT documentation is available, as is IQ/OQ support.

To further mitigate risk to the cell culture, several system modifications are available. Those modifications include antimicrobial coating, UV sterilization of the air flow path, and hygienic steam humidification.

Traceability: To allow for interaction with building monitoring systems, the Multitron Pro shaker can be equipped with analog outputs for all process parameters. Alternatively, the shakers can be controlled using eve® bioprocess software. This will provide traceability on par with a bioreactor, and allow both control of complex processes and integration of third-party devices. Also, eve can be qualified for GMP use.


A machine-learning process, combined with protein engineering and efficient screening facilities, is capable of yielding stable high-expressing cell lines, e.g., a 3–5 g/L IgG1 clone, in approximately 12 weeks.

Screening: With up to 7,680 data points (80-microtiter plates) per shaker, it is possible to manipulate multiple factors while performing all experiments in triplicates. This allows for the generation of statistically valid data for design of experiments (DoE), while also considering quality by design (QbD) as stable clones are generated and optimized.

Scaling: Stable systems, expressing hundreds of grams of protein, can be developed in a matter of weeks.

Validation: Systems and process parameters can be qualified and validated for GLP and cGMP work.

Traceability: When the incubation shaker is used in combination with the eve bioprocess software, traceability is comparable to a bioreactor’s. eve also allows full third-party device integration.


Andrew Magno ([email protected]) is manager, technical service, Gary Tompkins serves as chief technology officer, and Travis Scagliarini is manager, QA/QC, at Infors USA Inc. Miles Scotcher, Ph.D., is director of business developent at ATUM.