Results

Creation of the Parental Dock Cell Line

Docks were placed in transcriptionally active sites throughout the genome of CHOZN® GS-/- cells with sequential rounds of transduction using GPEx® retrovector technology (Figure 1). The Dock cell pool contained approximately 75 Dock copies per cell on average. To test the integration efficiency and expression of the pool, we co-transfected with Phi-C31 Recombinase and Boat construct expressing an Fc fusion protein (Figure 2). After transfection, pools were allowed to recover for 2-4 days prior to selection by removal of glutamine from the media.
The system utilizes multiple methods to reduce GS expression resulting in survival of only high copy number clones regardless of the initial copy number of the pool (see methods). For example, a pool containing 31 inserted copies per cell on average prior to selection via glutamine withdrawal resulted in 63 copies after selection while another pool that contained 64 copies per cell average before selection reached 94 copies after selection. Even though the second pool already contained 64 copies per cell on average, selection still resulted in a substantial increase in copy number.
After pools recovered from selection, QPCR using the attR primers, which only detect Boat that has integrated into the Dock site, revealed that these pools contained approximately 30 copies of recombined Boat per cell. In generic fed-batch production, these pools produce up to 5 g/L indicating the system was working as designed.
To isolate a high Dock copy number clone that had good growth characteristics, the Dock pool was subjected to single cell cloning using the Berkeley Lights Beacon® instrument. Multiple clones with the highest number of Dock copies were co-transfected with plasmid encoding the Phi-C31 recombinase and the same test Boat construct expressing an Fc fusion protein. After recovering from selection, the cell pool generated from the top clone contained about 40 integrated Boat copies on average. This clone was found to have suitable reproducibility of insertion, averaging 40 to 130 copies per pool. Robust growth characteristics and high protein expression were observed; therefore, this clone was used as the parental Dock cell line for all further process optimization studies and projects.

Creation of Cell Pools Expressing Proteins of Interest

To test the insertion efficiency, expression, and consistency of the technology, we also utilized a Boat expression construct that contains two cloning sites for the independent expression of both heavy chain and light chain sequences in the same Boat as opposed to the single chain construct described above. The GPEx® Lightning Parental Dock line was co-transfected with Phi-C31 recombinase plasmid and each Boat plasmid construct in triplicate. To select high expressing cells, glutamine was removed from the media and viability was monitored during selection (Figure 3A). Each triplicate transfection showed very consistent recovery from selection, and selection dynamics were very similar between molecules. Pools consistently achieved about 50 integrated Boat copies per cell on average (Figure 3B).

Fed Batch Production from Cell Pools

Pools were subjected to a fed-batch production using a generic, unoptimized feeding strategy (GFB1) for each of the three replicate pools. All three replicate pools of mAb1 reached approximately 20 x 106 cells per mL, maintained high viability for the full 20 days of production, and produced very consistent titers of about 6.5 g/L (Figure 4A). The same feed strategy was used for production of the Fc fusion protein (test molecule) and three other mAb molecules. Average viable cell densities for the three replicates of each molecule reached 18-26 million cells per mL and viability remained high throughout production (Figure 4B). Average final titers ranged from 3.5 to 5.5 g/L across these molecules. These data indicate that the technology consistently produces robust and high titer producing pools.
Upstream process development was performed on mAb1 pool population #3. Media/feed scouting experiments identified the GFB2 strategy as superior to GFB1 (Figure 5). The Ambr® 250 bioreactor system was used for further control of the GFB2 process. Using the optimized strategy, the pool reached 28 million cells per mL viable cell density, maintained high viability throughout production, and attained a final titer of approximately 12 g/L.

Stability of Cell Pools

To determine the stability of pools expressing the Fc fusion protein, three unique pools made at three different generations from the GPEx® Lightning Dock Pool (not clone) were subjected to stability analysis. For this study, two different media and feed strategies were employed. GFB1 typically provides robust titer and culture longevity for most products. GFB2 is more variable but often results in higher cell density and significantly higher titers than GFB1. Final titers and average number of integrated transgenes for all generations in each of the two medias remained stable for more than 40 generations (Figure 6A).
The genetic and production stability of pools expressing mAb1 and mAb2 made using the Parental Dock clone were also assessed using GFB1 (Figure 6B). These pools also showed high production and genetic stability across more than 40 generations. Together, these data indicate robust stability of transgene copy number and protein expression. Stability and consistency of expression are critical requirements for the potential use of these stable cell pools in drug substance manufacturing.

Expression and Stability of Clones

To determine the expression and stability of clones, the highest titer pool expressing the Fc Fusion molecules (pool population 4, Figure 6) was cloned using the Berkeley Lights Beacon® instrument. Clones were ranked and exported based on relative expression using the SpotlightTM Assay. 26 clones were expanded and their attR copy numbers ranged from 37 to approximately 200. At high copy numbers such as this, intrinsic variability in QPCR assays make it difficult to accurately measure precise copy number (D’haene, 2010). Since QPCR is better able to quantify lower copy numbers, we used primers that amplify attP, which is only present in empty dock sites. Highlighting the efficiency of the process, the attP amplicon could not be detected in one of the clones indicating that this clone had all dock sites filled.
Next, we performed a small-scale 16-day production run on these 26 clones using the GFB1 strategy. Titers ranged from 2.5 to 5 g/L in this production run. We found that Boat copy number correlated well with final titer (Figure 7A). To determine the stability of the three highest titer clones, a study was performed, this time for over 60 generations. As with cell pools, the clones showed the same robust stability of transgene copy number (not shown) and product expression (Figure 7B).

Protein Quality Characterization of Pools and Clones

To examine the consistency of the protein product quality, a generational stability study was conducted using two pools made at different times, but expressing the same Fc-fusion protein, which contains five potential N-glycan attachment sites (Figure 8A and Table 1). While there are minor differences between the quantity of each particular N-glycan between the two pools, different generations within each pool are essentially identical, demonstrating consistent quality over the generations studied.
Comparisons of product quality between pools and their clones were performed for 3 different antibody products. To create sample diversity for the evaluation, we chose clones that produced significantly more product than the pool (mAb05), less product than the pool (mAb06), and about the same amount of product as the pool (mAb07). No matter how the clones produced relative to the pool, no significant differences in glycosylation pattern or levels were observed (Figure 8B and Table 2). In each case the pools and their corresponding clone(s) are superimposable. Table 2 shows more detail and concordance for each species between the pool and the clone(s). Of particular interest is the similarity in high mannose and total fucosylated species.