1. Strategies for selecting host cell populations for enhanced process compatibility
Gary Khoo, Helga Fernandes, Bernie Sweeney, Paul Stephens
UCB - Celltech, Slough, United Kingdom
Introduction Growth data of these cell lines and F(ab)’ and antibody expression after transient transfection are shown in Figure Selection of CHO K1 host cells grown in semisolid media
2B and C. In general, while the antibody expression was significantly higher in the selected hosts, F(ab)’ expression
The development of a therapeutic mammalian cell platform requires the selection of the right host cell line. By was not consistently higher. For cell lines Sp1 and Sp1a, the expression data reflects the viable cell densities. In • Colonies were picked after 10 days of growth in semi-solid media. Only large dense colonies were selected as
selecting a suitable host cell line, process timelines for post-transfection clone development can be reduced and the general, the selected host are able to remain viable for a longer period of time after transient transfection. High shown in Figure 5. After 5 rounds of selection and expansion into shake flasks, 5 pools were evaluated.
probability of obtaining high producing clones during transfection can also be increased. Chinese Hamster Ovary voltage electroporation of these cell lines also revealed that these cell lines were more robust and recovered better
(CHO) cells are by far the most commonly used cell host for the expression of therapeutic proteins and antibodies. than the parental host. • The evaluation of batch growth of the cell lines (5B, 5N,
Various CHO host cell lines have been developed with the aim of improving process capabilities for scale-up. A
120 100
Figure 2: 5K, 5D1, 5D4) revealed that the selected hosts showed
A. Growth curves (solid lines) and cell viabilities similar or slower growth rates when compared to the adapted
Examples of these include the CHO-SV (LONZA Biologics) and the CHO-S (Invitrogen) cell line. Both of these 100 80
(dashed lines) of the cell lines selected for their
cell lines have the ability to grow to high viable cell densities in suspension and protein-free media as well as parental host. Most of the selected hosts showed peak cell
growth rates/anti-apoptotic characteristics.
densities that were similar or lower than that of the parental
Vi able Cell density (x10^5 cells/ml )
expressing recombinant antibodies and proteins at high levels. 80 60
B. F(ab)’ and antibody expression data 6 days post
transfection in the different cell lines. Data was host. However, the slower growth resulted in the cultures
Cell Viability (%)
60 40
normalised against the CHOK1 adapted parental remaining viable for a longer period of time. Figure 5: Images of varying CHOK1 colony growth
Desirable single cell characteristics include those with high growth rates, high transfection efficiency, chromosomal host. in semisolid media. The arrow points to the
stability, ‘humanised’ glycosylation profiles and high protein folding capabilities. At the same time, one has to 40 20
C. Viable cell densities (bars) and viabilities (lines) typical large dense colonies that were picked
consider larger scale biomanufacturing process characteristics and regulatory requirements. These include 6 days post transfection. • Cell lines 5D1 and 5D4 were also found to clump upon
suspension growth in chemically defined protein free media, low nutrient uptake/by-product secretion rates and the
20 0
reaching their peak cell density. This is an undesirable
ability to withstand ‘adverse’ bioreactor conditions ( high shear rate, hypoxia, apoptotic conditions). characteristic in the suspension growth of such cultures.
0 -20
0 50 100 Time (hrs) 150 200 250
Parental CHOK1
Adapted Cell line Sp1 Cell line Sp2 Cell line Sp2a Cell line Sp1a
• With exception of cell line 5N, the other 4 selected host cells showed a relatively good correlation between the
The desire to obtain a ‘super host’ to improve yields and reduce development times has led to two main approaches. B 3 C 25 100
transient production of F(ab)’/ antibody and their respective viable cell densities at day 6 of culture. However, all
F(ab)' Antibody
These include the ‘targeted’ approach where the engineering of certain pathways or proteins within the cell is carried 90
these cultures had lower viabilities than the adapted parental host, which could indicate that cell lines were more
2.5
out. Regulatory requirements pertaining to the characterisation of cell lines used for therapeutic production and the 20 80
susceptible to cell death post transfection.
70
need to ensure a lack of adventitious agents arising from the use of such cell lines are critical factors when
V iabl e Cell density (x10^ 5 cells/ m l)
Normalis ed exp re ssio n data
2 15 60
engineering cell lines. One way of getting around these issues is to harnesses the inherent heterogeneity of CHO cell A 100 100
Cell V iab ility (% )
50 Figure 6:
lines to obtain the desired cell line. In this work, we take the latter approach where cells are either selected or 1.5
10 40
90
Change these to match others
80
adapted to certain predetermined conditions based on the population heterogeneity. 30
80 A. Batch growth (solid lines) and viabilities
1
70
(dashed lines) of the cell lines selected
V i ab le Cel l d en sity (x10^ 5 cel l s/ m l)
5 20
60
from semisolid media
Methods
C el l V iab i li ty (% )
10 60
0.5 B. Expression data of a FAB and antibody
0
Parental CHOK1
adapted sp1 sp1a sp2 sp2a
0 50 40
after 6 days of transient transfection in the
0 Lipofectamine F(ab)' transfection Lipofectamine Antibody transfection 40 different cell lines. Data was normalised
Parental CHOK1
adapted Sp1 Sp1a Sp2 Sp2a Lipofectamine F(ab)' viability Lipofectamine Antibody viability 20 against the CHOK1 adapted parental host.
• CHO K1 parental cells were adapted to chemically defined animal component free media (CDCHO, Invitrogen) 30
C. Viable cell densities (bars) and viabilities
supplemented with GlutamaxTM (Invitrogen). Cells were cultivated in shake flasks (25ml to 200ml working Selection of CHO K1 host cells to withstand oxidative stress 20
0 (lines) of transiently transfected cell lines
volumes) and grown at 370C/ 8% CO2. Pseudo-perfusion cultivation of the CHO K1 parent host was carried out 10 after 6 days of culture.
under the conditions stated below. Cell counts and viability determination were carried out using Cedex • Oxidative stress has been observed in high antibody producing clones. Oxidative stress results from the rapid 0
0 50 100 Time (hrs) 150 200 250
-20
(Innovartis) machines. folding of proteins and has been linked to the generation of reactive oxygen species (ROS) (see figure 3). ROS 5b 5n 5k 5d1 5d4 ParentalCHOK1
Adapted
generation, if allowed to accumulate for a period of time will lead to cell death. B C
• Selection of improved growth and anti-apoptotic host cells was carried out by growth in 50% spent media. Spent 2.5
F(ab)' Antibody
20 80
media was collected from a prior culture when cell viabilities were below 50%. This same batch of spent media •After separate cell pools were recovered from methyl viologen application, growth and F(ab)’/antibody expression 18
70
16
was used throughout. To select for cells able to cope with oxidative stress, 2µΜ methyl viologen (Sigma Aldrich) testing was carried out. Growth rates of these cell lines (MV1a, MV1b, MV2a, MV2b) were found to be similar to 2
V iab le C ell den sity (x10^ 5 cells/m l)
60
14
was added to cultures. These conditions were applied to cultures separately and the parental cells were allowed to that of the parental host. However, two cell lines (MV1a and MV2b) were found to be able to reach higher peak
C ell V iability (% )
50
12
adapt and the surviving cells were expanded. Once recovered, another round of selection was carried out. This cell densities than the parental host with MV2b showing a 50% increase in peak cell density over the parental line.
N o rma lis e d e xp res s io n d a ta
1.5
10 40
cyclic selection process was carried for 70-140 days.
8
• Transient expression data did not show consistent trends with cell growth and viabilities. MV2a was found to be 1
30
6
• Our mammalian expression platform utilises the Clonepix FL system (Genetix) to select for high producing a poor expresser of both the F(ab)’ and antibody. Unlike the data from the spent media experiments there was no 20
4
clones after transfection. The Clonepix system uses a methyl cellulose semisolid media for the growth of such significant difference in expression of the F(ab)’ vs antibody. 0.5 10
2
clones. To enable the host cells lines to adapt in semisolid media growth, cells were plated into 6 well plates with
0 0
semisolid media and repeatedly selected for large dense colonies. 5 rounds of selection in clonemedia was carried Parental CHOK1
adapted 5b 5n 5k 5d1 5d4
Figure 3: Oxidative stress resulting from 0
out as shown in figure 1. Fast growing colonies were picked and pooled before they were expanded into shake peptide disulphide formation. Proteins Parental CHOK1
adapted 5N 5B 5K 5D1 5D4
Lipofectamine F(ab)' transfection
Lipofectamine F(ab)' viability
Lipofectamine Antibody transfection
Lipofectamine Antibody viability
flasks. Pools that clumped after such expansion were discarded while the remaining pools were replated. in the Endoplasmic Reticulum facilitate
this process. Figure is adapted from Tu
•Transient transfections were carried out on the selected host cells lines using a lipid based transfection agent and Weissman, JCB, 164:341-346, 2004. Summary of Results
(Lipofectamine 2000, Invitrogen). Host cell lines were transfected with DGVs vectors encoding the heavy chain
(F’ab or Full length) and kappa chain of an in-house antibody. • We have investigated various methods for improving host cell lines for transfection. Process considerations such
Large dense colonies
Oxidative stress as cell growth/ protein assembly and compatibility with our present expression platform were determinants for the
Cells seeded at 1000 cells/ml of
Semisolid media picked and recovered into selection process.
Figure 1: Selection of liquid media
host cells by repeated • Amongst all the conditions, cell lines generated from spent media selection resulted in most robust cell growth
seeding and colony and increased antibody titres (greater than 1.5 fold increase). However, the expression of F(ab)’ was not
picking in semi-solid A 120 100
Figure 4:
media A. Growth curves (solid lines) and
consistently better in these same cell lines. While adaptation to semisolid growth resulted in larger colonies, their
Incubate for robustness in terms of growth in suspension liquid media was reduced. However, despite this fact, their ability to
Cells which 100 80
viabilities (dashed lines) of the cell
10 days at
clump during lines selected from the oxidative express antibody was also consistently higher than that of the parental host. While none of the selection methods
37oC
V iab le Cel l de nsity (x1 0^5 c ells/ m l )
adaptation to 80 60
stress experiments. increased expression of the F(ab)’ it must be noted that expression of recombinant proteins is very dependant on