Characterization of the Stability of
Recombinant Protein Production in the
GS-NS0 Expression System
Louise M. Barnes,1 Catherine M. Bentley,2 Alan J. Dickson1
12.205 School of Biological Sciences, University of Manchester, Stopford
Building, Oxford Road, Manchester, M13 9PT, UK; telephone: 0161 275
5101; fax: 0161 275 5082; e-mail: l.m.barnes@fs1.scg.man.ac.uk
2GlaxoWellcome Research and Development, South Eden Park Road,
Beckenham, Kent, BR3 3BS, UK
Received 20 July 2000; accepted 24 November 2000
Abstract: The GS-NS0 system is an important mamma-
lian expression system used largely within industry for
the high-level expression of recombinant proteins for
therapeutic use. It is essential that the productivity of this
system remains stable throughout culture expansion for
the successful long-term production of recombinant pro-
teins. Here we present a study of the stability of recom-
binant protein production from unamplified GS-NS0 cell
lines over extended period of continuous culture. The
cell lines used in this study were generated by the trans-
fection of NS0 cells with DNA encoding for a secreted
recombinant protein and by two subsequent rounds of
limiting dilution cloning prior to analysis of stability. The
stability of recombinant protein production was as-
sessed at intervals over a period of 134 days using re-
peated batch culture in shake flasks. Heterogeneous sta-
bility was identified. The productivity of some clones re-
mained consistent throughout 134 days of continuous
culture. Others exhibit rapid and progressive loss of pro-
ductivity. Analysis of the causal relationships underlying
stability indicates that the initial transfectant determines
the susceptibility to loss or retention of productivity. Se-
lection of production clones on the basis of growth and
productivity alone will not predict stability during long-
term culture. Our research indicates that stable high-
producing clones can readily be obtained from use of the
GS-NS0 system in the absence of amplification but there
may be molecular features of the original transfectants
that could serve as very important predictive indicators
of the stability of recombinant protein production. © 2001
John Wiley & Sons, Inc. Biotechnol Bioeng 73: 261–270, 2001.
Keywords: clonal heterogeneity; GS-NS0; instability; my-
eloma; productivity; recombinant protein
INTRODUCTION
Mammalian cell expression systems are now an important
part of the biotechnology industry and are used for the high-
level production of many therapeutic and diagnostic pro-
teins. One of the major reasons for this is their ability to
carry out posttranslational modifications of the correct fi-
delity (Werner et al., 1998). Two expression systems are of
particular importance for the production of therapeutic pro-
teins, the dihydrofolate reductase (DHFR) system (Kauf-
man et al., 1985; Page and Sydenham, 1991) and the glu-
tamine synthetase (GS) system (Bebbington et al., 1992). In
comparison to the DHFR system, the GS system, especially
when used with nonsecreting (NS0) myeloma cells, offers a
large time advantage during development and requires
fewer copies of the recombinant gene per cell (Brown et al.,
1992). The GS system has been used in both Chinese ham-
ster ovary (CHO) (Cockett et al., 1990) and NS0 cells (Beb-
bington et al., 1992). However NS0 cells, unlike CHO cells,
are phenotypically GS-deficient and hence this allows easier
selection of successful transfectants. Therefore, the NS0 cell
line, which originated from mouse tumour cells in the
1960s, is commonly the cell line of choice used in combi-
nation with the GS system, particularly within industry
(Barnes et al., 2000).
The GS technology, which was developed by scientists at
Celltech, also has the added advantage that it can be used in
combination with methionine sulphoximine (MSX), a toxic
drug which is an inhibitor of GS activity (Brown et al.,
1992). This allows selection of cells lines producing high
levels of transfected recombinant protein and also enables
the possibility of gene amplification using increased levels
of MSX. Through optimisation of culture conditions, pro-
duction from the GS-NS0 system reaching levels of 2.7 g of
recombinant protein/litre in fed batch cultures have been
obtained (Zhou et al., 1997). The success and importance of
the GS system is highlighted by the fact that it has now been
used to produce two human therapeutic products that have
reached the marketplace, Zenapax® and Synagis (Lonza,
1998).
The production of recombinant protein at the large scale
required by industry necessitates significant time to scale up
cultures. During this time it is important that the level of
recombinant protein produced from the cell lines, selected
for high productivity, remains stable (Brown et al., 1992),
ideally in the absence of selection using toxic drugs, as these
are costly and complicate downstream purification of the
Correspondence to: Louise M. Barnes
Contract grant sponsors: the Biotechnology and Biological Sciences Re-
search Council (BBSRC) of the UK and GlaxoWellcome
© 2001 John Wiley & Sons, Inc.

desired product. Cell line production stability is generally
required throughout a period of at least 60 generations be-
yond the production of a Manufacturers Working Cell Bank
(Birch et al., 1993, Brown et al., 1992). Varying levels of
stability of recombinant protein production have been re-
ported from industrially important mammalian expression
systems. The DHFR-CHO system has been reported to
show some signs of instability in terms of recombinant pro-
tein production in the absence of toxic drug selective pres-
sure (i.e., methotrexate [MTX]) (Kim et al., 1998a,b; Page
and Sydenham, 1991; Weidle et al., 1988). The extent of the
decrease of productivity varies but is often attributed to loss
of recombinant gene copies during long-term culture (Kim
et al., 1998a). Similarly, the GS-CHO system has been re-
ported to lose the ability to accumulate product after a rea-
sonably short period of culture in the absence of selection
with MSX (Brown et al., 1992; Pu et al., 1998). As CHO
cells contain an endogenous GS activity, it is likely that
such a phenotype decreases the certainty of use of exog-
enous GS as a selectable marker and hence promotes insta-
bility (Brown et al., 1992). Due to the usage of the GS-NS0
system within the industrial environment, information on
stability is limited. The overall consensus in the literature
states that the system is relatively stable over the desired
time period (Bebbington et al., 1992; Brown et al., 1992),
although it is possible that unstable clones can be selected
under specific conditions (Bird et al., 1998).
It is important to ensure that stability of recombinant
protein expression can always be reproducibly obtained,
particularly with industrially important cell lines. To this
end it would be valuable in the long-term to identify mo-
lecular markers that could act as predictive indicators of cell
line instability. Currently, after identification of potential
clones on the basis of growth and production, promising
candidates for stable production are subjected to prolonged
culture and results are extrapolated to the full-scale produc-
tion environment. It would be of great value in terms of both
time and cost to remove the need for this prolonged culture
when confirming long-term stability of clones. This article
examines the stability of the GS-NS0 system producing a
secreted recombinant protein over an extended culture pe-
riod of 134 days. This time period of culture covers the time
that would be required for production scale expression. Us-
ing procedures that are typically used for clone generation
within industrial environments, we have produced the most
comprehensive analysis of stability of recombinant protein
production in the GS-NS0 system to date. The data pre-
sented here suggest that the generation of unstable cell lines
arises from molecular features of initial transfectants.
MATERIALS AND METHODS
Media and Reagents
NS0 cells (ECACC No. 85110503) and the DNA en-
coding for the recombinant protein were obtained from
GlaxoWellcome at Beckenham, Kent, UK, and Stevenage,
Herts, UK, respectively. Four media preparations were used
during cell line development and culture. The base medium,
Medium A, is a protein-free, serum-free medium formulated
at GlaxoWellcome (exact composition confidential). Me-
dium B consists of Medium A supplemented with 2 mM
L-glutamine (Gibco BRL, Renfrewshire, UK). Medium C is
a nonselective medium consisting of Iscove’s Modified
Dulbecco’s Medium (IMDM; Sigma, Dorset, UK) supple-
mented with 10% (v/v) fetal bovine serum (FBS; Gibco
BRL) and 2 mM L-glutamine. Medium D is a selective
medium consisting of IMDM supplemented with 10% (v/v)
dialysed (against 0.15 M NaCl) FBS (Selborne Biological
Services, New Zealand), 1% (v/v) G&A solution (0.041
mM glutamic acid, 0.045 mM asparagine) and 2% (v/v)
nucleoside solution (1.44 mM cytidine, 1.43 mM uridine,
1.31 mM adenosine, 1.24 mM guanosine, 0.5 mM thymi-
dine).
Restriction enzymes were purchased from Boehringer
Mannheim (Sussex, UK) and all other reagents were ob-
tained from BDH (Liverpool, UK) or Sigma, unless other-
wise stated.
Cell Culture and Growth Curve Analysis
Transfected cell lines were routinely cultured at 37°C with
5% CO2 in T-75 static flasks and cloned cell lines and
untransfected cells were routinely cultured at 37°C with
shaking (100 rpm) in 125 mL Erlenmeyer flasks. All cells
were diluted to a density of 0.2–0.3 × 106 viable cells/mL on
a 2,2,3-day subbing regime using Medium B for untrans-
fected cells, Medium D for transfected cell lines, and Me-
dium A for cloned cell lines. Long-term culture of cloned
cell lines was performed for a total of 134 days.
Batch culture growth curves were performed in duplicate
on transfected and cloned cell lines. Initial seeding densities
were between 0.25–0.4 × 106 viable cells/mL and cells were
counted at 24 h intervals by light microscopy in the pres-
ence of 0.5% Trypan blue in PBS on an improved Neubauer
haemocytometer. Growth curves were continued until cells
reached less than 50% viability.
Generation of Cell Lines
Transfectants were generated by electroporation of cells
with a plasmid containing genes for a secreted recombinant
protein and a glutamine synthetase selectable marker. 1 ×
107 viable cells in exponential growth with >90% viability
were transfected with 40 mg of linearised (Sal I) plasmid
dissolved in a total of 50 mL of sterile water. Cells were
washed in 1× PBS prior to the addition of DNA. Two con-
secutive pulses of 1,500 volts, 3 mFd were delivered to the
mixture in an electroporation cuvette via a Gene Pulsery
machine (Bio-Rad Labs, Herts, UK). Cells were placed on
ice for 5 min, diluted in Medium C to give a range of
dilutions (3.3 × 105 viable cells/mL, 8.25 × 104 viable cells/
mL and 1.65 × 104 viable cells/mL) and then plated out at
262 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 73, NO. 4, MAY 20, 2001