Creation of Engineered Human Embryonic Stem Cell Lines Using
phiC31 Integrase
BHASKAR THYAGARAJAN,
a
YING LIU,
a
SOOJUNG SHIN,
a
UMA LAKSHMIPATHY,
a
KELLY SCHEYHING,
a
HAIPENG XUE,
a
CATHARINA ELLERSTRO
¨
M,
b
RAIMUND STREHL,
b
JOHAN HYLLNER,
b
MAHENDRA S. RAO,
a
JONATHAN D. CHESNUT
a
a
Invitrogen Corporation, Carlsbad, California, USA;
b
Cellartis AB, Go¨teborg, Sweden
Key Words. Human embryonic stem cells ? phiC31 integrase ? Site-specific integration
ABSTRACT
It has previously been shown that the phage-derived phiC31
integrase can efficiently target native pseudo-attachment
sites in the genome of various species in cultured cells, as
well as in vivo. To demonstrate its utility in human embry-
onic stem cells (hESC), we have created hESC-derived
clones containing expression constructs. Variant human em-
bryonic stem cell lines BG01v and SA002 were used to
derive lines expressing a green fluorescent protein (GFP)
marker under control of either the human Oct4 promoter or
the EF1
promoter. Stable clones were selected by antibiotic
resistance and further characterized. The frequency of in-
tegration suggested candidate hot spots in the genome,
which were mapped using a plasmid rescue strategy. The
pseudo-attP profile in hESC differed from those reported
earlier in differentiated cells. Clones derived using this
method retained the ability to differentiate into all three
germ layers, and fidelity of expression of GFP was verified
in differentiation assays. GFP expression driven by the
Oct4 promoter recapitulated endogenous Oct4 expres-
sion, whereas persistent stable expression of GFP expression
driven by the EF1
promoter was seen. Our results demon-
strate the utility of phiC31 integrase to target pseudo-attP
sites in hESC and show that integrase-mediated site-specific
integration can efficiently create stably expressing engi-
neered human embryonic stem cell clones. STEM CELLS
2008;26:119–126
Disclosure of potential conflicts of interest is found at the end of this article.
INTRODUCTION
Human embryonic stem cells isolated from the inner cell mass
of 5-day blastocysts have the ability to differentiate into cells of
various lineages in culture [1, 2]. The pluripotency of these cells
makes them useful as platforms for basic research and drug
discovery and could eventually allow them to be used as ther-
apeutic material for diseases that are currently difficult or im-
possible to treat. The use of ESC-derived cells in these various
applications would benefit greatly from the development of
systems to facilitate rapid and efficient introduction of exoge-
nous genetic elements into the cell genome.
An example of this is the generation of reporter cell lines,
where specific promoter activity can easily be monitored via
expression of fluorescent protein. The exact culture conditions
and medium additives required to maintain human embryonic
stem cells (hESC) in their pluripotent state are currently being
identified and optimized. In addition, the pathways involved in
directing specific differentiation of hESC are still under inves-
tigation. The development of a hESC reporter line expressing a
fluorescent marker under control of a developmental stage-
specific promoter would aid in the study of stem cell biology by
providing a rapid readout that indicates the state of cells in
culture. Earlier studies have used randomly integrating vectors,
either retroviruses or plasmid DNA, to generate hESC-derived
lines expressing green fluorescent protein (GFP) [3–5]. These
studies describe the development of hESC lines expressing GFP
driven by either a constitutive (EF1
), an inducible (EF1
tied
to tetO element), or a lineage-specific (Oct4) promoter. A recent
study describes the construction of human embryonic stem cells
by transfection of plasmid DNA [6]. Although stable lines were
successfully generated, the silencing of randomly integrated
transgenes in this study underscores the importance of choosing
the appropriate vector.
Use of lentiviral vectors for engineering hESC is popular
because of the high efficiency of gene delivery afforded by viral
infection. Although this system has proven useful, limited pay-
load capacity and the potential for gene disruption from random
integration of DNA could limit its utility in stem cells. Two
recent articles describe the generation of hESC lines with inte-
grated transgenes [7, 8]. Vallier et al. describe the use of a
recombinase to create GFP-expressing hESC lines [7]. This
approach first requires the creation of a recombinase-expressing
line, followed by random integration of the expression con-
struct. Although this is an elegant method to induce expression
of the gene of interest, the efficiency of producing transgenic
lines using this method can be low. Zeng et al. describe a
baculovirus vector coupled with elements of adenoassociated
virus to obtain transduction, integration, and long-term expres-
sion of the transgene [8]. This protocol was efficient at gener-
ating transgenic hESC, and the integrated transgene showed
long-term expression. The large capacity of baculoviruses also
overcomes the disadvantage of retroviruses, which have a lim-
Correspondence: Bhaskar Thyagarajan, B.V.Sc., Ph.D., Invitrogen Corporation, 1600 Faraday Avenue, Carlsbad, California 92008, USA.
Telephone: 760-268-7460; Fax: 760-602-6691; e-mail: bhaskar.thyagarajan@invitrogen.com Received April 17, 2007; accepted for
publication October 9, 2007; first published online in STEM CELLS EXPRESS October 25, 2007; available online without subscription through
the open access option. AlphaMed Press 1066-5099/2007/$30.00/0 doi: 10.1634/stemcells.2007-0283
TECHNOLOGY DEVELOPMENT
STEMCELLS 2008;26:119–126 www.StemCells.com

ited payload capacity. However, the long-term effects of expres-
sion of the adeno-associated virus rep protein in hESC are still
unknown, and it is possible that they could lead to undesirable
effects [9].
Here, we focus on using a site-specific integration approach
to direct expression constructs to transcriptionally active chro-
mosomal regions using phiC31 integrase. This study outlines a
method for using this site-specific integration system to generate
engineered hESC lines. We describe here the construction of a
pluripotency-specific reporter line using the human Oct4 pro-
moter to drive GFP expression, as well as a constitutively
expressing GFP line.
The integrase from the Streptomyces phage phiC31 has been
shown to target donor plasmids containing a native attB site into
pseudo-attP sites in the human genome [10, 11]. phiC31 inte-
grase has been shown to function in vitro, in cell culture sys-
tems, as well as in vivo. This integrase has been successfully
used in cells derived from a number of species, including
human, mouse, rat, rabbit, Chinese hamsters, Drosophila, and
plants [12–25]. Unlike the better-known recombinases Cre and
Flp, the phiC31 integrase catalyzes recombination between two
nonidentical sites. This feature, along with the apparent lack of
a corresponding excisionase enzyme, makes the recombination
reaction unidirectional, ensuring that constructs integrated into
the genome do not act as substrates for the reverse reaction. The
result is an improvement in integration efficiency compared
with random integration. Another extremely useful feature of
phiC31 integrase is the ability of the enzyme to target pseudo-
attP sites present in the genome. These pseudo-attP sites bear
some resemblance to the native attP site and have been shown
to be present in transcriptionally active areas of the genome
[10]. Many of the sites described have been shown to be in
intronic regions of genes. Since these pseudo-attP sites typically
tend to be in open chromatin [10], our hypothesis was that there
would less interference with expression of the transgene, and
any changes in expression of the transgene would be solely due
to regulation of the promoter used. These features make this
integrase a potentially useful tool for construction of transgenic
lines from unmodified cells, since the targeted sites are already
present in the genome.
In this study, we have used phiC31 integrase to create
variant hESC-derived lines [26, 27] containing the GFP gene
driven by either the human Oct4 promoter or the human EF1

promoter. We also describe a simplified vector construction
design using a targeting vector that is a substrate for Multisite
Gateway (Invitrogen Corporation, Carlsbad, CA, http://www.
invitrogen.com). This greatly reduces the effort involved in
cloning, and allows the creation of multiple constructs in the
same background and with little effort. The combination of
Multisite Gateway technology and site-specific recombinases
provides a powerful tool for the construction of transgenic lines
in human embryonic stem cells, which in turn can be used as
versatile platforms for the study of stem cell biology.
MATERIALS AND METHODS
Plasmid Construction
The plasmids used in this study are shown in Figure 1. The plasmid
pCMV-phiC31 Int has been described earlier [18]. The plasmid
pB2H1-DEST was cloned as follows. The phiC31 attB site was
amplified from the plasmid pBC-PB [18] and cloned into pCR2.1
using the TA Cloning Kit (Invitrogen, Carlsbad, CA, http://www.
invitrogen.com) to generate pCR2.1-phiC31attB. This plasmid was
restricted with EcoRI to release the attB fragment and treated with
Klenow to generate blunt ends. This fragment was ligated with
ZraI-restricted pUC19 vector to generate pUC-phiC31attB2. An
expression cassette containing the Hygromycin phosphatase gene
driven by the HSV-TK promoter was amplified from pTKHyg and
T/A-cloned into pCR2.1. The resulting plasmid was restricted with
SpeI and EcoRV, treated with Klenow to generate blunt ends, and
ligated with pUC-phiC31attB2 that had been restricted with AflIII
and treated with Klenow to generate pB2H1. A fragment containing
the R1-R2 DEST cassette was amplified from pUC-DEST (Invitro-
gen) and T/A-cloned into pCR2.1. The resulting plasmid was re-
stricted with SpeI and EcoRV, treated with Klenow, and cloned into
pB2H1 treated with SalI and Klenow to generate the plasmid
pB2H1-R1R2DEST. This plasmid was used as a recipient for the
expression constructs used in this study.
A 3.2-kilobase fragment containing the human Oct4 promoter
[28, 29] was amplified from human genomic DNA using the prim-
ers hO-For (5
-GGAGAGGTGGGCCTCACC-3
) and hO-Rev (5
-
GGGGAAGGAAGGCGCCCC-3
). The resulting fragment was
T/A-cloned into pCR2.1 to generate pCR2.1-phOct4. Assembly of
the final phOct4-GFP and pEF1a-GFP expression constructs was
accomplished by using protocols recommended for Multisite Gate-
way.
Cell Culture and Transfection
BG01v cells (49, XXY, 12, 17,) were obtained from BresaGen,
Inc., (Athens, GA). SA002 cells (47, 13, XY) were obtained from
Cellartis AB (Goteborg, Sweden, http://www.cellartis.se). All re-
agents were obtained from Invitrogen unless indicated otherwise.
The cells were maintained either on a mouse embryonic fibroblast
(MEF) feeder layer in Dulbecco’s modified Eagle’s medium
(DMEM)/Ham’s F-12 medium (F12) medium supplemented with
20% knock-out serum replacement (KSR), 4 ng/ml basic fibroblast
growth factor, 1 ml of nonessential amino acids, and 100 M
-mercaptoethanol or on Matrigel (BD Biosciences, Franklin
Lakes, NJ, http://www.bdbiosciences.com) in the same medium
conditioned on MEF feeder layer. Fresh medium was provided to
the cells every day, and the cells were passaged every 4–5 days.
One day prior to transfection with Lipofectamine 2000 (Invitro-
gen), cells were treated with Accutase (Sigma-Aldrich, St. Louis,
http://www.sigmaaldrich.com) and plated on Matrigel in condi-
tioned medium. Lipofectamine 2000-mediated transfection was car-
ried out according to the manufacturer’s protocol. We typically used
4 g of the expression vector and 4 g of the phiC31 integrase
expression vector to transfect 2 million cells. Control transfections
omitted the phiC31 integrase plasmid or the GFP expression vector.
After transfection, cells were allowed to recover for 1 day, and
selection was started with medium containing Hygromycin at a
concentration of 10 g/ml. After 14–21 days of selection, individ-
ual colonies were manually picked and expanded for further anal-
ysis.
Electroporation was carried out with the BTX ECM630 elec-
troporator (Harvard Bioscience, Holliston, MA, http://www.
btxonline.com). Six to 8 million cells were harvested using Ac-
cutase and resuspended in 800 l of OptiPro SFM (Invitrogen).
These cells were placed in an electroporation cuvette with a gap of
0.4 cm. Cells were electroporated with a pulse of 500 V at 250 F.
Electroporated cells were plated on MEF feeders and allowed to
recover for 48–72 hours before selection was started with Hygro-
mycin (10 g/ml; Invitrogen). As with lipid-mediated transfection,
individual drug-resistant clones were manually picked and ex-
panded for further analysis.
Plasmid Rescue and Sequence Analysis
Genomic DNA isolated from individual clones was restricted with
the restriction enzymes NheI, SpeI, and XbaI. The enzymes were
heat-inactivated, and the DNA was self-ligated at low DNA and T4
DNA ligase concentrations. After overnight incubation at 16°C, the
DNA was extracted with phenol:chloroform, ethanol-precipitated,
and resuspended in water. Electrocompetent DH10B Escherichia
coli were then electroporated with the ligated DNA using the Gene
Pulser II (Bio-Rad, Hercules, CA, http://www.bio-rad.com) using
the recommended conditions. The resulting transformation was
plated on Luria-Bertani (LB)-agar plates containing ampicillin.
Plasmid DNA isolated from the resulting colonies was sequenced
using the primer ChoSeqR (5
-TCCCGTGCTCACCGTGAC-
120 Site-Specific Integration in hESC