Enhancing recombinant protein yield and quality using novel CHO GT cells in high density fed batch cultures 6c

Targeting Early Apoptotic Genes in Batch
and Fed-Batch CHO Cell Cultures
Danny Chee Furng Wong,
1,2
Kathy Tin Kam Wong,
1
Peter Morin Nissom,
1
Chew Kiat Heng,
2
Miranda Gek Sim Yap
1,3
1
Bioprocessing Technology Institute, Biomedical Sciences Institutes,
Agency for Science, Technology and Research (A*STAR), 20 Biopolis Way,
#06-01, Centros, Singapore 138668; telephone: þ65-6478-8880; fax: þ65-6478-9561;
e-mail:
2
Department of Paediatrics, National University of Singapore, Singapore
3
Department of Chemical and Biomolecular Engineering,
National University of Singapore, Singapore
Received 2 June 2005; accepted 27 December 2005
Published online 7 August 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20871
Abstract: Based on the transcriptional profiling of CHO
cell culture using microarray, four key early apoptosis
signaling genes, Fadd, Faim, Alg-2, and Requiem, were
identified and CHO GT (Gene Targeted) cell lines were
developed by targeting these four genes. Two were CHO
GT
O

cell lines overexpressing anti-apoptotic genes, Faim
and Fadd DN and two were CHO GT
KD
cell lines involving
knockdown ofAlg-2 andRequiem which are pro-apoptotic
genes using small interfering RNA (siRNA) technology.
Comparisons of these CHO GT cell lines with the parental
cell line in batch culture (BC) and fed-batch culture (FBC)
were performed. Compared to parental cells, the CHO GT
cell lines showed apoptosis resistance as they signifi-
cantly delayed and/or suppressed initiator caspase-8
and -9 and executioner caspase-3 activities during culture.
FBC of CHO GT cell lines reached significantly higher
maximum viable cell densities (up to 9 Â 10
6
cells/mL)
compared with the parental cell line (5 Â 10
6
cells/mL). The
recombinant interferon gamma (IFN-g) yields were
increased by up to 2.5-fold. Furthermore, it was observed
that the IFN-g was more highly sialylated.
ß 2006 Wiley Periodicals, Inc.
Keywords: CHO; apoptosis; fed-batch; FAIM; FADD;
ALG-2; REQUIEM
INTRODUCTION
In many bioprocesses involving animal cells, cell death is a
major barrier to maintaining high cell densities at high
viability and often leads to lower protein yields and quality
(Arden and Betenbaugh, 2004; Fussenegger and Bailey,

1998; Laken and Leonard, 2001; Vives et al., 2003a). The
ability to extend culture life at high viability is important in
processes to maintain recombinant protein quality. For
example, degradative enzymes released during cell death
can detrimentally affect the sialylation of the recombinant
protein resulting in reduc ed circulatory half-life of biother-
apeutics in vivo (Gramer et al., 1994; Varki, 1993).
Cell death in culture has been attributed to apoptosis
(Goswami et al., 1999; Singh et al., 1994). Various strategies
such as nutrient feeding, the addition of anti-apoptotic
chemicals, and genetic manipulation have been used to
extend culture viability (Arden and Betenbaugh, 2004;
Fussenegger and Bailey, 1998; Laken and Leonard, 2001;
Vives et al., 2003a). The most common genetic modification
reported in literature involves overexpression of either Bcl-2
or Bcl-x
L
to prevent mitochondrial release of cytochrome c
(Arden and Betenbaugh, 2004; Laken and Leonard, 2001;
Vives et al., 2003a). Alternatively, the viral homologues of
Bcl-2 such as Bhrf-1 and E1B19K can also be used to confer
apoptosis resistance. Other strategies include approaches that
interfere with caspase activation such as anti-sense RNA
against caspases or overexpression of caspase inhibitors such
as XIAP and CrmA (Kim and Lee, 2002; Sauerwald et al.,
2002, 2003).
In an earlier study (Wong et al., 2006), the transcriptome
analysis of apoptosis signaling pathways using microarray
showed differential expression of FasL, Fadd, Bim, and Bak
in batch culture (BC) and Fadd, Faim, Bim, Bad, Bax, Alg-2

and Requiem in fed-batch culture (FBC). These were
considered to be early (initiator) apoptosis signaling genes
as they were up- or downregulated going from the
exponential to the stationary phase of BC and FBC. As
shown in Figure 1, these genes are involved in death
receptor-, mitochondria-, and endoplasmic reticulum (ER)-
mediated apoptosis signaling.
In the extrinsic death receptor-mediated apoptosis signal-
ing pathway, FasL and Fadd, which are involved in CD95
death receptor apoptosis signaling (Curtin and Cotter, 2003;
Peter et al., 1996), were found to be upregulated during BC
and FBC (Wong et al., 2006). We also observed significant
downregulation of Faim, a pro-survival gene that confers
resistance to CD95 death receptor-mediated apoptosis
signaling (Rothstein et al., 2000; Schneider et al., 1999),
ß 2006 Wiley Periodicals, Inc.
Correspondence to: M.G.S. Yap
suggesting that this particular survival pathway was
compromised during FBC.
In the intrinsic mitochondrial-mediated pathway, we
found significant upregulation of several pro-apoptotic
BCL-2 family members such as Bak, Bax, Bim, and Bad
(Wong et al., 2006). It has been reported that the pro-
apoptotic activities of these four genes can be neutralized by
the protective properties of Bcl-2 and Bcl-x
L
(Cory et al.,
2003; Gross et al., 1999; O’Connor et al., 1998; van Gurp
et al., 2003). Various researchers have shown the effective-
ness of Bcl-2 and Bcl-x

L
overexpression for delaying the
onset of apoptosis (Charbonneau et al., 2003; Chiang
and Sisk, 2005; Chung et al., 1998; Figueroa et al., 2004;
Mastrangelo et al., 2000; Meents et al., 2002; Simpson et al.,
1999; Tey et al., 2000; Vives et al., 2003b).
In the intrinsic ER-mediated pathway, we observed the
upregulation of Apoptosis linked gene 2 (Alg-2) during
culture (Wong et al., 2006). Although it has been shown
that ALG-2 protein forms part of the ER-stress induced
caspase-activating complex (Rao et al., 2004a,b), Alg-2 gene
has also been implicated in CD95 death receptor-mediated
apoptosis (Jung et al., 2001; Krebs and Klemenz, 2000; Vito
et al., 1996).
Requiem, which encodes for a zinc finger protein
transcription factor, has been reported to be required for
apoptosis response following survival factor withdrawal
(Gabig et al., 1994). However, the exact role of Requi em in
apoptosis signaling has not been clearly defined.
Out of the nine early apoptosis signaling genes found to be
differentially expressed in BC and FBC, four of these genes,
Fadd (upregulated in BC and FBC), Faim (downregulated in
FBC), Alg-2 (upregulated in FBC), and Requiem (upr egu-
lated in FBC), were selected for gene targeting to develop
apoptosis-resistant cell lines. FasL was not selected for
targeting because it was upregulated only in BC and not FBC.
Overexpression of Bcl-2 and Bcl-x
L
has been reported to
suppress the pro-apoptotic activities of Bim, Bak, Bax, and

Bad, and as such these genes were not selected. In this study,
four CHO GT (gene targeted) cell lines producing recombi-
nant human interferon gamma (IFN-g) were constructed and
compared with the parental CHO in BC and FBC in terms of
culture viability, IFN-g yields, and quality.
MATERIALS AND METHODS
Construction of Plasmids
FADD Dominant Negative Expression Vector
To overexpress a dominant negative form of FADD from
CHO cells, an artificial Fadd dominant negative (Fadd DN)
fragment with kozak sequence was created by using a 5
0
-PCR
primer, 5
0
-GATAT CGGATCC GCCA CCATG GCCTT
TGACA TTGTA TGCGA CAATG TGGGG-3
0
and a 3
0
-
PCR primer, 5
0
-CCCGG GCTCG AGTGC CTCCC TTCCA
CCAGG TCAG-3
0
. The underlined sequence consisted of a
BamHI and XhoI restriction site, respectively, while the
italicized sequence consisted of an artificial kozak and start
codon to facilitate ‘in frame’ expression of Fadd DN.A

partial Fadd sequence cloned from CHO cells and subcloned
Figure 1. Apoptosis signaling via death receptor-, mitochondria-, and ER-mediated apoptosis signaling pathways during CHO cell culture (shaded box
indicates genes which are targeted in this study).
Wong et al.: Targeting Early Apoptotic Genes 351
Biotechnology and Bioengineering. DOI 10.1002/bit
into pCR
1
-TOPO
1
(Invitrogen, Carlsbad, CA), was used as
the template for PCR amplification. PCR conditions were:
948C for 5 min, followed by 31 cycles of 948C for 1 min, 508C
for 1 min and 728C for 2 min and a final extension at 728C for
10 min. The verified PCR product was then subcloned into
pcDNA3.1(þ) (Invitrogen) and verified again by sequencing.
Verified pcDNA3.1(þ) Fadd DN was then purified using
Maxi Plasmid Purification Kit (Qiagen, Valencia, CA) and its
concentration quantified for transfection into CHO IFN-g
cells to generate CHO GT
O
FADD DN cell line.
FAIM Expression Vector
To overexpress the anti-apoptosis protein FAIM from CHO
cells, we created a FAIM with an artificial kozak sequence
and linker regions using a 5
0
-PCR primer, 5
0
-GAATT
CGCCA CCATG ACAGA TCTTG TAGC-3

0
and a 3
0
-PCR
primer, 5
0
-GAATT CGTGA ACACA TTTAA TTACC A-3
0
.
The underlined sequence consisted of an EcoRI restriction
site while the italicized sequence consisted of an artificial
kozak sequence to facilitate ‘in frame’ expression of
CgFAIM. A Faim cDNA fragment cloned from CHO cells
and subcloned in pCR
1
-TOPO
1
(Invitrogen) was used as
the template. The PCR conditions were: 948C for 5 min,
followed by 60 cycles of 948C for 1 min, 448C for 1 min and
728C for 2min and a final extension at 728C for 10 min. The
verified PCR produc t was cloned into the EcoRI insertion site
of pcDNA3.1(þ) vector (Invitrogen) and verified again by
sequencing. Verified pcDNA3.1(þ) Faim was then purified
using Maxi Plasmid Purification Kit (Qiagen) and its
concentration quantified for transfection into CHO IFN-g
cells to generate CHO GT
O
FAIM cel l line.
ALG-2 Suppression Vector

To achieve the silencing of pro-apoptotic ALG-2 from CHO
cells, a small interfering RNA (siRNA) vector, pSUPER.neo
(OligoEngine, Seattle, WA) was inserted with a pair of
oligonucleotide insert containing unique Alg-2 sequences
shown by underlined nucleotides. The 5
0
oligo insert, 5
0
-
GATCCC
GTGAG CTTCA GCAAG CATTA TTCAA
GAGA
T AATGC TTGCT GAAGC TCATT TTTTG
GAAA-3
0
was annealed to the 3
0
oligo insert, 5
0
-AGCTT
TTCCA AAAAA
TGAGC TTCAG CAAGC ATTAT
CTCTT GAA
TA ATGCT TGCTG AAGCT CACG-3
0
and
ligated into HindIII and BglII digested pSUPER.neo vector
(OligoEngine). Verified pSUPER.neo.ALG-2 siRNA was
then purified using Maxi Plasmid Purification Kit (Qiagen)
and its concentration quantified for transfection into CHO

IFN-g cells to generate CHO GT
KD
ALG-2 cell line.
Requiem Suppression Vector
To achieve the silencing of pro-apoptotic Requiem from CHO
cells, a siRNA vector, pSUPER.neo (OligoEngine) was
inserted with a pair of oligonucleotide insert containing
unique Requiem sequences shown by underlined nucleotides.
The 5
0
oligo insert, 5
0
-GATCC CGCGG ATCCT TGAAC
CTGAT TTCAA GAGAA TCAGG TTCAG GATCC
GCTTT TTTGG AAA-3
0
was annealed to the 3
0
oligo insert,
5
0
-AGCTT TTCCA AAAAA GCGGA TCCTT GAACC
TGATT CTCTT GAAAT CAGGT TCAAG GATCC GCGG-
3
0
and ligated into HindIII and BglII digested pSUPER.neo
vector. Verified pSUPER.neo.Requiem siRNA plasmid was
then purified using Maxi Plasmid Purification Kit (Qiagen)
and its concentration quantified for transfection into CHO
IFN-g cells to generate CHO GT

KD
REQUIEM cell line.
Real-Time PCR (RT-PCR)
Total RNAwas extracted using Trizol
TM
reagent (Invitrogen)
from samples containing approximately 10 milli on cells and
quantified using GeneQuant
TM
Pro RNA/DNA Calculator
(Amersham Biosciences, Piscataway, NJ). RNA samples
were examined on a 1% denaturing RNA gel to ensure no
RNA degradation after extraction. Total RNA concentration
and purity were determined using a UV spectrometer
(Amersham Biosciences). RT-PCR was used to ascertain
the relative overexpression or suppression of gene of interest
after transfection experiments. RT-PCR was carried out using
the ABI PRISM
1
7000 Sequence Detection System using
SYBR
1
Green PCR Master Mix (Applied Biosystem, Foster
City, CA). Primers used for gene-specific real-time PCR were
as described by Wong et al. (2006). The primers were
designed to resu lt in amplicon sizes ranging from 100 to 200
bp. In order to generate standard curves, quantified pCR
1
-
TOPO

1
(Invitrogen) plasmids containing either Faim, Fadd,
Alg-2orRequiem cloned from CHO cells were serially
diluted and used as standards for RT-PCR. Duplicate runs
were conducted for each sample and normalized against b-
actin (U20114) expression. Fold change in gene expression
was calculated using the delta-delta threshold cycle (DC
T
)
method as described by Livak and Schmittgen (2001).
Cell Line
CHO IFN-g is a suspension Chinese Hamster Ovary cell line
that produces recombinant human IFN-g (Scahill et al.,
1983). CHO IFN-g was maintained in commercially avail-
able serum-free HyQ CHO MPS media (Hyclone, Logan,
UT) supplemented with 4 mM glutamine, 20 mM glucose,
and 0.25 mM methotrexate (Sigma, St. Louis, MO).
Creation of Stable Cell Lines
Transfection was carried out using Lipofectamine reagent
(Invitrogen) according to the manufacturer’s instructions.
Cells were grown overnight in 6-well plates with 0.5 million
cells/well and transfected with approximately 1 mgof
linearized plasmid per well the next day. To generate stable
cell lines, cells were grown for 24 h before changeover to
selection media (1 mg/mL of Geneticin). The cells were
maintained in selection media for 4 weeks where the
untransfected cells in the selection media died within a
352 Biotechnology and Bioengineering, Vol. 95, No. 3, October 20, 2006
DOI 10.1002/bit
week. Stably integrated single cell clones were obtained by

serial dilution of cells into 96-well plates such that there was
only one cell in each well. Wells were checked under a light
microscope and only those containing a single cell were
marked for expansion. Single cell clones were expanded into
24-well plates followed by 6-well plates before culturing in
shake flask.
Batch and Fed-Batch Bioreactor
Control Operations
Batch culture and FBC operations were performed according
to methods previously described by Wong et al. (2005). The
FBC utilize d an online dynamic feeding strategy where at
intervals of 1.5 h, an automated measurement of residual
glutamine concentration was taken. If residual glutamine fell
below setpoint control concentrations of 0.3 mM, feeding
was effected with concentrated media to raise cultur e
glutamine concentration to 0.3 mM.
Measurement of Cell Viability and Apoptosis
Cell viability was determined by trypan blue exclusion assay.
Apoptosis was determined using an Ethidium Bromide/
Acridine Orange assay adapted from Mercille and Massie
(1994) that allowed for the classification of cells into
apoptotic or non-apoptotic populations.
Proteolytic Activities of Initiator
Caspase-8, -9 and Executioner Caspase–3
In addition to morphological analysis, caspase activities were
measured as biochemical indications of apoptosis induction
using BD ApoAlert
TM
Caspase Assay Plates (BD Bios-
ciences Clontech, CA) according to the manufacturer’s

protocol. The assay kit contained substrate specific for
caspases-3, -8, and -9, which upon cleavage by the specific
caspases can be detected using a fluorescence plate reader at
380 nm excitation and 460 nm emission. Normalized caspase
activity was calculated using the following equation:
Normalized caspase activity ¼ Fluorescence
time x
À Fluorescence
time 0
Recombinant Human IFN-g Quantification
Recombinant human IFN-g concentrations in culture were
determined using an enzyme-linked immunosorbent (ELISA)
assay (HyCult Biotechnology, Uden, The Netherlands).
Average specific productivity of IFN-g was then calculated
across the exponential growth phases according to the methods
previously described by Wong et al. (2005).
Sialic Acid Content Assay
Recombinant IFN-g was purified from samples collected at
mid-exponential growth phase and at when the highest IFN-g
concentrations were detected during high viability (>95%)
and during low viability (70–80%). The sialic acid content of
the IFN-g was then determined using a modified thiobarbi-
turic acid assay as described by Wong et al. (2005) .
RESULTS
Creation of Gene Targeted CHO
(CHO GT) Cell Lines
A summary of the anti-apoptosis strategies used to confer
apoptosis resistance on CHO IFN-g cells is shown in Table I.
For ease of reference, these genetically modified cells were
termed ‘CHO GT’ followed by subscript ‘O’ or ‘KD’ to

indicate overexpression or knockdown of the candidate
genes. CHO GT
O
FAIM cells were genetically modified to
constitutively overexpress recombinant FAIM protein to
enhance the endogenous Faim pro-survival pathway, which
was observed to be compromised during FBC (Wong et al.,
2006). CHO GT
O
FADD DN cells were genetically
Table I. Fold increase/decrease in gene expression in CHO GT and control cell lines.
Cell line
Candidate
apoptosis gene
Anti-apoptosis
strategy
Relative increase/decrease in expression (fold change)
Faim Fadd DN Alg-2 Requiem
CHO GT
O
Blank pool None None þ0.2 Æ 0.3 À0.3 Æ 0.4 — —
CHO GT
O
FAIM pool Faim Overexpression þ3.3 Æ 0.2 — — —
CHO GT
O
FAIM clone 5 Faim Overexpression þ3.9 Æ 0.3 — — —
CHO GT
O
FADD DN pool Fadd Dominant negative — þ4.3 Æ 0.3 — —

CHO GT
O
FADD DN clone 7 Fadd Dominant negative — þ4.5 Æ 0.5 — —
CHO GT
KD
Blank pool None None — — þ0.1 Æ 0.2 À0.2 Æ 0.3
CHO GT
KD
ALG-2 pool Alg-2 siRNA knockdown — — À2.3 Æ 0.1 —
CHO GT
KD
ALG-2 clone 1 Alg-2 siRNA knockdown — — À3.3 Æ 0.3 —
CHO GT
KD
REQUIEM pool Requiem siRNA knockdown — — — À3.4 Æ 0.4
CHO GT
KD
REQUIEM clone 4 Requiem siRNA knockdown — — — À4.2 Æ 0.2
Relative increase or decrease in expression in various cell lines compared to the parental CHO IFN-g cell line as determined by quantitative real-time PCR
using primers specific for the genes examined.
Wong et al.: Targeting Early Apoptotic Genes 353
Biotechnology and Bioengineering. DOI 10.1002/bit
engineered to overexpress a dominant negative form of
FADD protein. It has been shown that overexpression of the
dominant negative form, which contains the death domain
(DD) but not the death effector domain (DED) of FADD
allowed for effective suppression of apoptosis signaling
via FADD and CD95 death receptors (Chinnaiyan et al.,
1996).
Unfortunately, as the functional domains of the other pro-

apoptotic genes, Alg-2 and Requiem, are not well defined, it is
difficult to design a dominant negative approach. Thus, a
vector that allowed for stable expression of siRNA was used
to knockdown the expression of these two pro-apoptotic
genes. The siRNA can achieve gene-specific knockdown by
guiding an RNA-induced silencing complex to cleave
specific mRNAs based on sequence complement (Hammond
et al., 2001; Moss and Taylor, 2003; Samuel, 2004). The
gene-specific siRNA was designed according to criteria
described by Brummelkamp et al. (2002). To further ensure
specific gene targeting, oligo insert sequences were com-
pared against Genebank database using BLAST to ensure the
sequence specificity to the desired target gene.
In order to validate the effectiveness of the above-
mentioned strategies to overexpress or suppress specifically
targeted genes, we used real-time PCR to quantify specific
gene expression of either Faim, Fadd DN, Alg-2orRequiem
in CHO GT cells. Real-time PCR analysis can be used
reliably for detection of mRNA abundance and has a wide
dynamic range of quantification (Klein, 2002). The quanti-
tative real-time PCR results showed that the CHO GT cells
indeed have overexpression/suppression of the intended gene
targets while the negative controls consisting of CHO GT
O
blank cells transfected with pcDNA3.1(þ) blank vector and
CHO GT
KD
blank cells transfected with pSUPER.neo blank
vector, did not show significant changes in gene expression
when compared to the parental CHO IFN-g cells (Table I).

Batch Cultures of CHO GT Transfected Pools
Figures 2 and 3 show the profiles of viable cell density, %
viability and % apoptotic cells for the four different CHO GT
cell lines during BC using transfected pools. The use of
transfected pools in this initial characterization ensured that
any observed apoptosis-resistant phenotype is not due to
clonal variation caused by single cell selection. CHO GT
O
FADD DN and CHO GT
O
FAIM cells were able to achieve
similar maximum viable cell densities comparable to the
control cells (Fig. 2A). The loss in culture viability for CHO
GT
O
FADD DN and CHO GT
O
FAIM cells were delayed for
Figure 2. CHO GT
O
cells in batch cultures. The viable cell density (A), viability (B) and percentage of apoptotic cells (C) of CHO GT
O
FADD DN pool ( )
and CHO GT
O
FAIM pool ( ) compared to CHO GT
O
Blank pool (&) in batch cultures.
354 Biotechnology and Bioengineering, Vol. 95, No. 3, October 20, 2006
DOI 10.1002/bit

24 h (Fig. 2B) corresponding to a 24 h delay in the onset of
apoptosis (Fig. 2C) compared to control CHO GT
O
Blank
cells. In the case of CHO GT
KD
ALG-2 and CHO GT
KD
REQUIEM cells, maximum viable cell densities achieved
were higher than the control CHO GT
KD
Blank cells (Fig. 3A).
Both CHO GT
KD
ALG-2 and CHO GT
KD
REQUIEM cells
also showed a 24 h delay in the loss of culture viability
(Fig. 3B) and the onset of apoptosis (Fig. 3C) compared to
CHO GT
KD
Blank cells. These results indicate clearly that all
four gene-targeting strategies were able to delay the onset of
apoptosis resulting in prolonged BC.
As shown in Figure 4, the extended culture viabilities (from
72 to 96 h of culture time) for CHO GT
O
and CHO GT
KD
cell

lines translated into greatly enhanced recombinant IFN-g
yields. The CHO GT
O
and CHO GT
KD
cells achieved two- to
threefold higher recombinant IFN-g yields (4–6 mg/L)
compared to the average 2 mg/L yield in CHO IFN-g parental
and CHO GT Blank cells in BC. Specific productivities of the
CHO GT were similar to the parental CHO IFN-g cell line as
shown in Figure 5, indicating that the improvement in
recombinant IFN-g yields achieved was due to prolonged
culture viability rather than to improvement in specific
productivities.
Proteolytic Activities of Caspases in
Batch Cultures
Figures 6 and 7 show the impact of early apoptosis gene
targeting in CHO GT
O
and CHO GT
KD
cells on initiator
caspase-8 and -9 and executioner caspase-3 activation during
BC. Executioner caspase-3 is involved in the execution of
apoptosis and can be activated either by initiator caspase-8,
which is primarily involved in death receptor-mediated
apoptosis signaling or by initiator caspase-9, which is
primarily involved in mitochondria-mediated apoptosis
signaling (Riedl and Shi, 2004).
For both CHO GT

O
FADD DN and CHO GT
O
FAIM cells,
the onset of initiator caspase-8 activities (Fig. 6A) was
delayed by $72 h, while initiator caspase-9 activities
(Fig. 6B) were suppressed compared to the control CHO
GT
O
Blank cells. There was a delay of 24 h in the downstream
activation of executioner caspase-3 (Fig. 6C) for CHO GT
O
FADD DN and CHO GT
O
FAIM cells compared with control
CHO GT
O
Blank.
In CHO GT
KD
ALG-2 cells, low activities of caspase-8
(Fig. 7A) and caspase-9 (Fig. 7B) were observed, indicating
minimal activation of these initiator caspases compared to
Figure 3. CHO GT
KD
cells in batch cultures. The viable cell density (A), viability (B) and percentage of apoptotic cells (C) of CHO GT
KD
ALG-2 pool (~)
and CHO GT
KD

REQUIEM pool (*) compared to CHO GT
KD
Blank pool (&) in batch cultures.
Wong et al.: Targeting Early Apoptotic Genes 355
Biotechnology and Bioengineering. DOI 10.1002/bit
CHO GT
KD
Blank cells. However, as shown in Figure 7C, the
executioner caspase-3 was observed to be activated at 144 h,
albeit at a suppressed level with a 24 h delay compared with the
activation of executioner caspase-3 in CHO GT
KD
Blank cells.
In the case of CHO GT
KD
REQUIEM cells, caspase
activation profiles for caspase-8 (Fig. 7A) and caspase-9
(Fig. 7B) were similar to that for CHO GT
O
FADD DN and
CHO GT
O
FAIM. There was also a correspo nding 24 h delay
in downstream activation of executioner caspase-3 (Fig. 7C)
compared to the control CHO GT
KD
Blank cells.
The above results showing that the CHO GT cells exhibit
significant delays and/or suppression of caspase activities
compared to CHO GT Blank cells are evidences that clearly

demonstrate that all four gene-targeting strategies were
effective in delaying apoptosi s in BC.
Fed-Batch Cultures of CHO GT Cell Lines
The fed-batch studies of CHO GT cells were carried out using
single cell clones selected to have comparable targeted gene
overexpression or suppression with the transfected pool
based on real-time PCR as shown in Table I. Using single cell
clones as opposed to transfected pools for the fed-batch
studies is advantageous as a homogeneous population can be
maintained over the extended culture time. This would
ensure that the effects of the targeted genes (namely Fadd,
Faim, Alg-2, and Requiem) on apoptosis in culture would be
clearly exhibited.
As shown in Figure 8, all four CHO GT cells could delay
loss in culture viability by 24–48 h compared to the parental
CHO IFN-g cells during FBC. In addition, three of the CHO
GT cells, namely CHO GT
O
FAIM, CHO GT
KD
ALG-2, and
CHO GT
KD
REQUIEM achieved much higher viable cell
densities of up to 9.0 Â 10
6
cells/mL compared to the parental
cell line ($5.0 Â 10
6
cells/mL) in FBC.

Most importantly, Figure 9 shows that the extension in
culture viability and improvement in culture cell density for
CHO GT cell lines in FBC translated into improvement in
recombinant protein yield. Compa red to the 20 mg/L of
recombinant human IFN-g yield in CHO parental FBC, the
best performing cell line, CHO GT
KD
REQUIEM showed a
2.5-fold increase in yield (up to 49 mg/L) of IFN-g during
FBC.
Sialylation of Recombinant Human
IFN-g in CHO GT Cell Lines
Figure 10 shows the sialylation of recombinant human IFN-g
harvested at three time points during FBC, namely at the mid-
exponential (>95% viability), stationary (>95% viability),
and death phase (70–80% viability) for the CHO GT and
parental CHO IFN-g cell lines. For the latter, the sialic acid
content of recombinant human IFN-g decreased as the culture
progressed from mid-exponential (2.9 mol of SA/mol of
IFN-g) to stationary (2.3 mol of SA/mol of IFN-g) to death
phase (2.1 mol of SA/mol of IFN-g). In contrast, the sialic
acid content of IFN-g harvested at the three time points for the
four CHO GT cell lines was maintained, and even showed
increase in sialylation, ranging from 2.7 to 3.5 mol of SA/mol
of IFN-g. These results seem to suggest that another potential
benefit of apoptosis-resistant CHO cells is the maintenance/
Figure 4. Recombinant human IFN-g yields in batch cultures. Concentra-
tions of recombinant human IFN-g during batch cultures of CHO GT
O
FADD

DN pool ( ), CHO GT
O
FAIM pool ( ), CHO GT
KD
ALG-2 pool (~), and
CHO GT
KD
REQUIEM pool (*) compared to CHO GT blank pool (&).
Figure 5. Specific producti vities of recombinant human IFN-g in CHO G T
cells. The four CHO GT cells, CHO GT
O
FADD DN pool ( ), CHO GT
O
FAIM
pool ( ), CHO GT
KD
ALG-2 pool (~), and C HO GT
KD
REQUIEM pool (*)
showed similar productivities to the parental CHO IFN- g cell line ( &).
356 Biotechnology and Bioengineering, Vol. 95, No. 3, October 20, 2006
DOI 10.1002/bit
enhancement of protein glycosylation quality over extended
culture time, regardless of loss in culture viability (70–80%).
DISCUSSION
Targeting of Fadd, Faim, Alg-2, and Requiem to
Prolong BC and FBC
Overcoming apoptosis, the major mode of cell death in many
bioprocesses is desirable to enhance product yield and
quality. We have previously demonstrated that DNA micro-

arrays could be used to profile apoptosis signaling pathways
in culture and identified early apoptosis signaling genes
(Wong et al., 2006). In this study, we selected Fadd, Faim,
Alg-2, and Requiem to target prolonged culture viability. And
as hypothesized, the results in this study showed that the four
CHO GT cell lines constructed by targeting the early
apoptosis signaling genes prolonged culture viability by
delaying the onset of apoptosis in BC and FBC. Recombinant
IFN-g yields of between 1.2- to 2.5-fold higher were achieved
compared to parental CHO IFN-g cells. These enhanced yields
are comparable, or in some cases better than those reported
involving either Bcl-2orBcl-x
L
overexpression (Fussenegger
et al., 1998; Chiang and Sisk, 2005; Figueroa et al., 2004;
Jung et al., 2002; Meents et al., 2002; Tey et al., 2000).
For all four CHO GT cell lines, the enhancement in IFN-g
yields was accompanied by the notable finding that the sialic
acid content of the IFN-g produced in FBC was maintained/
enhanced and did not decrease over the extended culture
time, regardless of loss in culture viability (70–80%). This is
a distinct advantage for cell lines used for manufacturing
biotherapeutics as a lower degree of sialylation can decrease
the in vivo half-life of protein-based drugs (Gramer et al.,
1995; Varki, 1993). It is hypothesized that apoptosis-related
signaling can detrimentally affect protein glycosylation. As
such, in CHO GT cells where the onset of apoptosis is
delayed, the glycosylation machinery continues to be active,
resulting in more highly sialylated glycoproteins. Decrease
in sialidase released by CHO GT cells may account for the

observed maintenance of sialylation of IFN-g over extended
culture time and loss in culture viability.
Strategies to Enhance Apoptosis Resistance
Most reported anti-apoptosis strategies block apoptosis via
targeting the late execution pathways such as caspase-3
activation or cytochrome c release (Arden and Betenbaugh,
2004; Laken and Leonard, 2001; Vives et al., 2003a). The
advantage is that apoptosis signaling pathways converging
on a few executioner genes can be blocked simultaneously.
However, targeting of late executioner genes may not be as
effective, as the activation of the early apoptotic signals
may cause detrimental effects on cell metabolism and
productivity.
Figure 6. Caspase-8, -9, and -3 activities during batch culture of CHO GT
O
cell lines. The activities of initiator caspase-8 (A), initiator caspase-9 (B), and
executioner caspase-3 (C) of CHO GT
O
FADD DN pool ( ) and CHO GT
O
FAIM pool ( ) compared to CHO GT
O
Blank pool (&) during batch cultures
(caspase activities are normalized to caspase activity at time 0 h).
Wong et al.: Targeting Early Apoptotic Genes 357
Biotechnology and Bioengineering. DOI 10.1002/bit
Figure 7. Caspase-8, -9, and -3 activities during batch culture of CHO GT
KD
cell lines. The activities of initiator caspase-8 (A), initiator caspase-9 (B), and
executioner caspase-3 (C) of CHO GT

KD
ALG-2 pool (~) and CHO GT
KD
REQUIEM pool (*) compared to CHO GT
KD
Blank pool (&) during batch cultures
(caspase activities are normalized to caspase activity at time 0 h).
Figure 8. Viable cell densities of CHO GT cell lines in fed-batch cultures. Viable cell densities of CHO GT
O
FADD DN cell line ( ), CHO GT
O
FAIM cell
line ( ), CHO GT
KD
ALG-2 cell line (~) and CHO GT
KD
REQUIEM cell line ( *) compared to the parental CHO IFN-g cell line (&) during fed-batch culture.
358 Biotechnology and Bioengineering, Vol. 95, No. 3, October 20, 2006
DOI 10.1002/bit
In contrast, the anti-apoptosis strategies described in this
study involve targeting early apoptosis signaling genes,
acting prior to activation of caspases (Chin naiyan et al., 1996;
Gabig et al., 1994; Jung et al., 2001; Rothstein et al., 2000;
Schneider et al., 1999). The results here showed that
intervention at the early initiation stages could delay the
onset of apoptosis, leading to enhanced culture viability and
recombinant protein yields. However, the possible drawback
of targeting early initiation signals is that the strategy may be
cell type- and/or stimulus-specific and the activation of other
initiator apoptosis signaling pathways may compromise

apoptosis resistance.
A multi-gene approach which targets both early apoptosis
signaling genes such as Fadd, Faim, Alg-2, and Requiem
together with late executioner genes such as Caspase-3 may
offer further improvement in extension of culture viability
Figure 9. Enhanced recombinant human IFN-g yields in CHO GT cell lines during fed-batch cultures. Recombinant IFN-g yields for CHO GT
O
FADD DN
cell line (
), CHO GT
O
FAIM cell line ( ), CHO GT
KD
ALG-2 cell line (~), and CHO GT
KD
REQUIEM cell line (*) during fed-batch culture compared to
the parental CHO IFN-g cell line during batch (&) and fed-batch (&) culture.
Figure 10. Sialylation of recombinant IFN-g in CHO GT cell lines during mid-exponential, stationary, and death phase of FBC. Sialylation of IFN-g during
fed-batch culture of CHO GT
O
FADD DN cell line ( ), CHO GT
O
FAIM cell line ( ), CHO GT
KD
ALG-2 cell line (~) and CHO GT
KD
REQUIEM cell line
(*) compared to the parental CHO IFN-g cell line (&).
Wong et al.: Targeting Early Apoptotic Genes 359
Biotechnology and Bioengineering. DOI 10.1002/bit

and recombinant protein production. Alternatively, a dual
gene-targeting strategy, which simultaneously blocks the
death receptor- and mitochondria-mediated pathways, may
also be effective. An example drawn from this study would be
to combine the overexpression of Fadd DN or Faim (which
targets the death receptor-mediated apoptosis signaling
pathway) with the overexpression of Bcl-2orBcl-x
L
(mediated apoptosis signaling pathway). Indeed, Figue roa
et al. (2004) has shown that the overexpression of Bcl-x
L
together with Aven, which inhibits caspase-9, was more
effective for apoptosis suppression than the individual
overexpression of these genes (Figueroa et al., 2004).
CONCLUSION
This study showed that the targeting of early apoptosis
signaling genes, Fadd, Faim, Alg-2, and Requiem, identified
via transcriptome analysis, are promising strategies, which
can effectively suppress apoptosis in culture. Most impor-
tantly, the prolonged culture viability can be translated into
improved recombinant protein yield and quality.
NOMENCLATURE
Alg-2 Apoptosis linked gene 2
Fadd Fas associated death domain gene
Faim Fas apoptosis inhibitory molecule
The authors acknowledge Ong Peh Fern, Jennifer Lo, Tan Kher Shing,
and Breana Cham for their technical assistance in molecular biology.
References
Arden N, Betenbaugh MJ. 2004. Life and death in mammalian cell culture:
Strategies for apoptosis inhibition. Trends Biotechnol 22:174–180.

Brummelkamp TR, Bernards R, Agami R. 2002. A system for stable expression
of short interfering RNAs in mammalian cells. Science 296:550–553.
Charbonneau JR, Furtak T, Lefebvre J, Gauthier ER. 2003. Bcl-xL
expression interferes with the effects of L-glutamine supplementation
on hybridoma cultures. Biotechnol Bioeng 81:279–290.
Chiang GG, Sisk WP. 2005. Bcl-x
L
mediated increased production of
humanized monoclonal antibodies in Chinese Hamster Ovary cells.
Biotechnol Bioeng 91:779–792.
Chinnaiyan AM, Tepper CG, Seldin MF, O’Rourke K, Kischkel FC,
Hellbardt S, Krammer PH, Peter ME, Dixit VM. 1996. FADD/MORT1
is a common mediator of CD95(Fas/APO-1 and TNF receptor-induced
apoptosis. J Biol Chem 271:4961–4965.
Chung JD, Sinskey AJ, Stephanopoulos G. 1998. Growth factor and bcl-2
mediated survival during abortive proliferation of hybridoma cell line.
Biotechnol Bioeng 57:164–171.
Cory S, Huang DCS, Adams JM. 2003. The Bcl-2 family: Roles in cell
survival and oncogenesis. Oncogene 22:8590–8607.
Curtin JF, Cotter TG. 2003. Live and let die: Regulatory mechanisms in Fas-
mediated apoptosis. Cell Signal 15:983–992.
Figueroa BJ, Chen S, Oyler GA, Hardwick JM, Betenbaugh MJ. 2004. Aven
and Bcl-x
L
enhance protection against apoptosis for mammalian cells
exposed to various culture conditions. Biotechnol Bioeng 85:589–600.
Fussenegger M, Schlatter S, Datwyler D, Mazur X, Bailey JE. 1998.
Controlled proliferation by multigene metabolic engineering enhances
the productivity of Chinese Hamster ovary cells. Nat Biotechnol 16:
468–472.

Fussenegger M, Bailey JE. 1998. Molecular regulation of cell-cycle
progression and apoptosis in mammalian cells: Implications for
Biotechnology. Biotechnol Prog 14:807–833.
Gabig TG, Mantel PL, Rosli R, Crean CD. 1994. Requiem: A novel zinc
finger gene essential for apoptosis in myeloid cells. J Biol Chem 269:
29515–29519.
Goswami J, Sinskey AJ, Steller H, Stephanopoulos GN, Wang DI. 1999.
Apoptosis in batch cultures of Chinese hamster ovary cells. Biotechnol
Bioeng 62:632–640.
Gramer MJ, Goochee CF. 1994. Glycosidase activities of the 293 and NS0
cell lines and of an antibody-producing hybridoma cell line. Biotechnol
Bioeng 43:423–428.
Gross A, McDonnell JM, Korsmeyer SJ. 1999. Bcl-2 family members and the
mitochondria in apoptosis. Genes Dev 13:1899–1911.
Hammond SM, Caudy AA, Hannon GJ. 2001. Post-transcriptional
gene silencing by double-stranded RNA. Nat Rev Genet 2:110–111.
Jung YS, Kim KS, Kim KD, Lim JS, Kim JW, Kim E. 2001. Apoptosis-linked
gene 2 binds to the death domain of Fas and dissociates from Fas during
Fas-mediated apoptosis in Jurkat cells. Biochem Biophys Res Commun
288:420–426.
Jung D, Cotes S, Drouin M, Simard C, Lemieux R. 2002. Inducible
expression of Bcl-XL restricts apoptosis resistance to the antibody
secretion phase in hybridoma cultures. Biotechnol Bioeng 79:
180–187.
Kim NS, Lee GM, 2002. Inhibition of sodium butyrate-induced apoptosis in
recombinant Chinese hamster ovary cells by constitutively expressed
ansense RNA of caspase-3. Biotechnol Bioeng 78:217–228.
Klein D. 2002. Quantification using real-time PCR technology: Application
and limitations. Trends Mol Med 8:257–260.
Krebs J, Klemenz R. 2000. The ALG-2/AIP-complex, a modulator at the

interface between cell proliferation and cell death? A hypothesis.
Biochim Biophys Acta 1498:153–161.
Laken HA, Leonard MW. 2001. Understanding and modulating apoptosis in
industrial cell culture. Curr Opin Biotechnol 12:175–179.
Livak KJ, Schmittgen TD. 2001. Analysis of relative gene expression data
using real-time quantitative PCR and the 2^[-delta delta Ct] method.
Methods 25:402–408.
Mastrangelo AJ, Hardwick JM, Bex F, Betenbaugh MJ. 2000. Part I. Bcl-2
and Bcl-x(L) limit apoptosis upon infection with alphavirus vectors.
Biotechnol Bioeng 67:555–564.
Meents H, Enenkel B, Eppenberger HM, Werner RG, Fussenegger M. 2002.
Impact of coexpression and coamplification of sICAM and antiapoptosis
determinants bcl-2/bcl-x
L
on productivity, cell survival and mitochon-
dria number in CHO-DG44 grown in suspension and serum-free media.
Biotechnol Bioeng 80:706–716.
Mercille S, Massie B. 1994. Induction of apoptosis in nutrient-deprived
cultures of hybridoma and myeloma cells. Biotechnol Bioeng 44:1140–
1154.
Moss EG, Taylor JM. 2003. Small-interfering RNAs in the radar of the
interferon system. Nat Cell Biol 5:771–772.
O’Connor L, Strasser A, O’Reilly LA, Hausmann G, Adams JM, Cory S,
Huang DCS. 1998. Bim: A novel member of the Bcl-2 family that
promotes apoptosis. EMBO J 17:384–395.
Peter ME, Kischkel FC, Hellbardt S, Chinnaiyan AM, Krammer PH, Dixit
VM. 1996. CD95(APO-1/Fas)-associating signaling proteins. Cell
Death Differ 3:161–170.
Rao R, Poksay K, Castro-Obregon S, Schilling B, Row RH, Rio G, Gibson
BW, Ellerby Hm, Bredesen DE. 2004a. Molecular components of a cell

death pathway activated by endoplasmic reticulum stress. J Biol Chem
279:117–187.
Rao RV, Ellerby HM, Bredesen DE. 2004b. Coupling endoplasmic reticulum
stress to the cell death program. Cell Death Differ 11:372–380.
Riedl SJ, Shi Y. 2004. Molecular mechanisms of caspase regulation during
apoptosis. Nat Rev Mol Cell Biol 5:897–907.
Rothstein TL, Zhong X, Schram BR, Negm RS, Donohoe TJ, Cabral DS,
Foote LC, Schneider TJ. 2000. Receptor-specific regulation of B-cell
360 Biotechnology and Bioengineering, Vol. 95, No. 3, October 20, 2006
DOI 10.1002/bit
susceptibility to Fas-mediated apoptosis and a novel Fas apoptosis
inhibitory molecule. Immunol Rev 176:116– 133.
Sauerwald TM, Betenbaugh MJ, Oyler GA. 2002. Inhibiting apoptosis in
mammalian cell culture using caspase inhibitor XIAP and deletion
mutants. Biotechnol Bioeng 77:704–716.
Sauerwald TM, Oyler GA, Betenbaugh MJ. 2003. Studying caspase
inhibitors for limiting death in mammalian cell culture. Biotechnol
Bioeng 81:329–340.
Samuel CE. 2004. Knockdown by RNAi—Proceed with caution. Nat
Biotechnol 22:280–282.
Scahill SJ, Devos R, Der Heyden VJ, Fiers W. 1983. Expression and
characterization of the product of a human immune interferon cDNA
gene in Chinese hamster ovary cells. Proc Natl Acad Sci USA 80:4654–
4658.
Schneider TJ, Fischer GM, Donohoe TJ, Colarusso TP, Rothstein TL. 1999.
A novel gene coding for a fas apoptosis inhibitory molecule (FAIM)
isolated from inducibly fas-resistant B lymphocytes. J Exp Med 189:
949–956.
Simpson NH, Singh RP, Emery AN, Al-Rubeai M. 1999. Bcl-2 over-
expression reduced growth rate and prolongs G1 phase in continuous

chemostat cultures of hybridoma cells. Biotechnol Bioeng 64:
174–186.
Singh RP, Al-Rubeai M, Gregory CD, Emery AN. 1994. Cell death in
bioreactors: A role for apoptosis. Biotechnol Bioeng 44:720–726.
Tey BT, Singh RP, Piredda L, Piacentini M, Al-Rubeai M. 2000. Influence of
Bcl-2 on cell death during the cultivation of a Chinese hamster ovary cell
line expressing a chimeric antibody. Biotechnol Bioeng 68:31–43.
van Gurp M, Festjens N, van Loo G, Saelens X, Vandenabelle P. 2003.
Mitochondrial intermembrane proteins in cell death. Biochem Biophys
Res Commun 304:487–497.
Varki A. 1993. Biological roles of oligosaccharides: All of the theories are
correct. Glycobiology 3:97–130.
Vito P, Lacana E, D’Adamio L. 1996. Interfering with apoptosis: Ca2þ
binding protein ALG-2 and Alzheimer’s disease gene ALG-3. Science
271:521–525.
Vives J, Juanola S, Cairo JJ, Godia F. 2003a. Metabolic engineering of
apoptosis in cultured animal cells: Implications for the biotechnology
industry. Metab Eng 5:124–132.
VivesJ, Juanola S, Cairo JJ, Prats E, Cornudella L, Godia F. 2003b. Protective
effect of viral homologues of Bcl-2 on hybridoma cells under apoptosis-
inducing conditions. Biotechnol Prog 19:84–89.
Wong CFD, Wong TKK, Goh LT, Heng CK, Yap GSM. 2005. Impact of
dynamic online fed-batch strategies on metabolism, productivity and
N-glycosylation quality in CHO cell cultures. Biotechnol Bioeng
89:164–177.
Wong CFD, Wong TKK, Lee YY, Nissom PM, Heng CK, Yap MGS. 2006.
Transcriptional profiling of apoptotic pathways in batch and fed-batch
CHO cell cultures. Biotechnol Bioeng 94:373–382.
Wong et al.: Targeting Early Apoptotic Genes 361
Biotechnology and Bioengineering. DOI 10.1002/bit