Long-Term Expression of Human Coagulation
Factor VIII in a Tolerant Mouse Model Using
the uC31 Integrase System
Christopher L. Chavez,1Annahita Keravala,1Jacqueline N. Chu,1Alfonso P. Farruggio,1
Vanessa E. Cue ´llar,1Jan Voorberg,2and Michele P. Calos1
We generated a mouse model for hemophilia A that combines a homozygous knockout for murine factor VIII
(FVIII) and a homozygous addition of a mutant human FVIII (hFVIII). The resulting mouse, having no de-
tectable FVIII protein or activity and tolerant to hFVIII, is useful for evaluating FVIII gene-therapy protocols.
This model was used to develop an effective gene-therapy strategy using the uC31 integrase to mediate per-
manent genomic integration of an hFVIII cDNA deleted for the B-domain. Various plasmids encoding uC31
integrase and hFVIII were delivered to the livers of these mice by using hydrodynamic tail-vein injection. Long-
term expression of therapeutic levels of hFVIII was observed over a 6-month time course when an intron was
included in the hFVIII expression cassette and wild-type uC31 integrase was used. A second dose of the hFVIII
and integrase plasmids resulted in higher long-term hFVIII levels, indicating that incremental doses were
beneficial and that a second dose of uC31 integrase was tolerated. We observed a significant decrease in the
bleeding time after a tail-clip challenge in mice treated with plasmids expressing hFVIII and uC31 integrase.
Genomic integration of the hFVIII expression plasmid was demonstrated by junction PCR at a known hotspot
for integration in mouse liver. The uC31 integrase system provided a nonviral method to achieve long-term
FVIII gene therapy in a relevant mouse model of hemophilia A.
within the blood. It is the most common form of hemophilia,
as it affects an estimated one in 5,000 males and accounts for
80–85% of the cases of hemophilia (Cooper and Tuddenham,
1994). Of 1,221 unique mutations identified in the human
FVIII (hFVIII) gene, 47.7% were missense, the largest cate-
11.1% were large deletions, 10.7% were nonsense, 7.8% were
splicing mutants, and 6.6% were insertions (Kemball-Cook
et al., 1998). The levels of hFVIII in circulation determine the
severity of the disease, with plasma levels 5–25% of normal
being mild, 1–5% being moderate, and <1% being severe
(Brettler, 1995). Therefore, only modest levels of circulating
hFVIII are needed to be therapeutic and to provide protection
from spontaneous bleeding episodes. Current treatments for
hemophilia A include infusion of plasma-derived hFVIII,
emophilia A is an X-linked recessive disease caused by
a deficiency of coagulation factor VIII (FVIII) activity
which carries with it the risk of infection, and administration
of recombinant hFVIII, which is expensive (*$100,000/year)
and unavailable in most of the world.
These limitations of current treatments have stimulated
research into alternative therapies for hemophilia A. Several
animal studies using gene therapy have been encouraging.
However, clinical trials have been unable to achieve long-
term, therapeutic levels of hFVIII (Ma ´trai et al., 2010; Petrus
et al., 2010). These failures have been due to factors such as
immune responses to viral vectors (Chuah and Vanden-
Driessche, 2004) and low and/or transient hFVIII gene ex-
pression (Roth et al., 2001; Powell et al., 2003). Additionally,
current FVIII knockout mouse models have unfortunately
been shown reproducibly to generate FVIII inhibitory anti-
bodies after multiple injections of hFVIII (Qian et al., 2000;
Wu et al., 2001). Inhibitory antibodies occur in only 25–30%
of human hemophilia A patients who receive hFVIII infu-
sions (Hoyer and Scandella, 1994). These patients may need
special procedures, such as tolerization to hFVIII, before
1Department of Genetics, Stanford University School of Medicine, Stanford, CA 94305-5120.
2Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, Amsterdam, The Netherlands.
HUMAN GENE THERAPY 23:390–398 (April 2012)
ª Mary Ann Liebert, Inc.
gene therapy can be used or, alternatively, introduction of
activated factor VIIa (Gabrovsky and Calos, 2008).
The uC31 integrase is a site-specific recombinase native to
bacteriophge uC31 of Streptomyces soil bacteria. In nature,
uC31 integrase catalyzes the recombination of the viral ge-
nome with that of the bacterial host through two *30-bp
recognition sequences termed attB and attP (Kuhstoss and
Rao, 1991; Rausch and Lehmann, 1991; Groth et al., 2000).
The integration reaction takes place without a requirement
for any host cofactors (Thorpe and Smith, 1998). Ad-
ditionally, it has been demonstrated that uC31 integrase
autonomously catalyzes its recombination reaction in mam-
malian cells (Groth et al., 2000). Database searches have re-
vealed that the mouse and human genomes do not contain
perfect attP sites. They do, however, contain sites with sim-
ilar sequences, termed pseudo attP sites (Thyagarajan et al.,
2001). Integration of a plasmid bearing a therapeutic gene
can be carried out by placing an attB site on the plasmid,
cotransfecting with a plasmid encoding the uC31 integrase,
and using integration into these native pseudo attP sites.
Because uC31 integrase requires a relatively long and spe-
cific recognition sequence, the number of potential integra-
tion sites may be lower than in systems that integrate with
less sequence specificity (Chalberg et al., 2006).
Materials and Methods
The uC31 integrase plasmids pVI, pVhP2, and pVmI were
described previously (Keravala et al., 2009). The human
B-domain–deleted FVIII cDNA was amplified from the vec-
tor pCAGEN/HSQ (kind gift of P. Lollar, Emory University)
with the primers HSQ Forward (5¢-ATGCAAATAGAG
CTCTCCACCTGC-3’) and HSQ Reverse (5¢-TCAGTAGAG
GTCCTGTGCCTCGC-3’). PCR conditions were as follows:
an initial denaturation step at 94?C for 2min; then 94?C for
20sec, 55?C for 20sec, and 72?C for 3min, repeated for 30
cycles; and a final hold at 4?C. The resulting 4.3-kb band was
cloned into the Topo TA cloning vector from Invitrogen
(Carlsbad, CA). The human factor IX expression plasmid
pVB9 (Keravala et al., 2009) was cut with NruI/EcoRV, re-
moving the factor IX gene. The hFVIII gene was then cloned
in using the same restriction enzymes, creating pVB8. To
create pVB8ii, the factor IX intron A from pVB9 was intro-
duced into the native FVIII intron-1 location using an
enzyme-free cloning method (Tillett and Neilan, 1999).
Mice were housed in the Research Animal Facility at
Stanford University. huFVIII-R593C mice have been de-
scribed previously (Bril et al., 2006). In brief, an huFVIII-
R593C expression cassette was injected into fertilized oocytes
of FVB mice. Founder mice were then crossed to C57BL/6
mice for five generations. These mice were then crossed to
FVIII knockout mice (B6;129S4-F8tm1Kaz/J) that carry a tar-
geted disruption of exon 16 of the mouse FVIII gene (Bi et al.,
1995), purchased from The Jackson Laboratory (Bar Harbor,
ME). C57BL/6 mice were purchased from Charles River
Laboratories (Wilmington, MA). Animals were fed water
and chow ad libitum. Eight- to 10-week-old mice were used
for the studies. Experimental protocols were approved by the
Administrative Panel on Laboratory Animal Care at Stanford
University. Hydrodynamic tail-vein injection in mice was
carried out as described (Keravala et al., 2009).
Genomic DNA isolation
To obtain mouse-tail DNA, after clipping of 1cm of a
mouse tail, the intact end of the mouse tail was cauterized
with Kwik Stop Styptic Powder (Gimborn Pet Specialties,
Atlanta, GA). DNA was isolated as described previously
(Laird et al., 1991). To obtain mouse liver DNA, mouse livers
were dissected, and different lobes of the liver were minced.
Approximately 25mg of liver tissue was used for genomic
DNA isolation using the DNeasy Blood & Tissue Kit (QIA-
GEN, Valencia, CA).
Mice were screened for the mutant hFVIII transgene by
PCR. DNA (100ng) obtained from mouse tails was used as
template to generate a 505-bp fragment. The primers used
were HF8F (5¢-GAATTCAGGCCTCATTGGAG) and HF8R
(5¢-TCGTAGTTGGGGTTCCTCTG). PCR conditions were as
follows: an initial denaturation step at 94?C for 2min; then
94?C for 30sec, 55?C for 30sec, and 72?C for 30sec, repeated
for 30 cycles; and a final hold at 4?C.
hFVIII activity, expression, and Bethesda assays
Blood was collected by retro-orbital puncture using cap-
illary tubes and transferred to microfuge tubes containing
sodium citrate to a final concentration of 0.38% (vol/vol).
The blood was then centrifuged at 2,000 g for 20min, and the
plasma was removed and stored at -80?C until use. hFVIII
activity in mouse plasma was determined using the Biophen
France), following the manufacturer’s protocol. hFVIII con-
centration was determined using the Matched-Pair Antibody
Set for ELISA of hFVIII antigen, following the manufactur-
er’s protocol (Affinity Biologicals, Ancaster, ON, Canada).
FVIII inhibitor antibodies were determined as described
previously (Jin et al., 2004), with the following modification:
Residual FVIII activity was determined using the Biophen
FVIII:C kit from Hyphen BioMed. Pooled human plasma
from Instrumentation Laboratory (Lexington, MA) was used
to generate standard curves.
We used a method of tail bleeding described previously
(Gui et al., 2009), with a few modifications. In brief, mice
were anesthetized with ketamine xylazine. The tail was
transected at 1cm from the tip and immersed in 14ml of
HBSS at 37?C. The time to cessation of bleeding was re-
corded. After 10min, bleeding was stopped by applying
pressure to the tip of the tail.
Pseudo site PCR
uC31 integrase–mediated integration was confirmed by
PCR at the mpsL1 pseudo attP site (Olivares et al., 2002). The
first round of amplification was performed with primers
attB-F3 (5¢-CGAAGCCGCGGTGCG) and mpsL1-R1 (5¢-
GTAAATGTTATTGCGGCTCT). The second round was
hFVIII EXPRESSION IN MICE WITH uC31 INTEGRASE 391
performed with primers attB-F4 (5¢-CGGTGCGGGTGCCA)
and mpsL1-R2 (5¢-GGTCATGGAGCCCCTTCACAA). Both
rounds used the following PCR conditions: an initial dena-
turation step at 94?C for 3min; then 94?C for 30sec, 66?C for
30sec, and 72?C for 20sec, repeated for 30 cycles; a final
extension at 72?C for 3min; and a final hold at 4?C.
Southern blots were used to determine whether mice were
singly or doubly transgenic for the mutant hFVIII gene. A
probe was generated to the mutant hFVIII gene by PCR of a
known transgenic mouse. The following primers were used:
HF8F (5¢-GAATTCAGGCCTCATTGGAG) and HF8R (5¢-
TCGTAGTTGGGGTTCCTCTG). The PCR conditions were
as follows: an initial denaturation step at 94?C for 2min; then
94?C for 30sec, 55?C for 30sec, and 72?C for 30sec, repeated
for 40 cycles; and a final hold at 4?C. A 505-bp fragment of
the mutant hFVIII gene was amplified. Eight micrograms of
total liver DNA were digested with BamHI-HF. The samples
were then electrophoresed through a 1% agarose gel and
transferred to a nitrocellulose membrane. The mutant hFVIII
transgene was detected after hybridization with a digox-
igenin-labeled hFVIII probe and chemiluminescence (Roche,
Indianapolis, IN). A 1.9-kb band was detected in transgenic
mice. The band was twice as dark in double-transgenic mice
as in single-transgenic mice.
Data were analyzed using the Microsoft Excel program.
The one-tailed Student’s t test assuming unequal variances
p value of <0.05 was considered to be statistically significant.
Generation of huFVIII-R593C/E-16KO mice
We used a mouse that was a transgenic knockin for the
hFVIII R593C missense mutation (Bril et al., 2006). This
mutation was identified in 48% of patients with mild he-
mophilia A in a past study (Bril et al., 2004). We chose this
mouse model because it had been demonstrated to be tol-
erant to hFVIII under normal conditions and, importantly for
gene-therapy studies, was also shown to have no detectable
FVIII activity or circulating protein by ELISA in mice (Bril
et al., 2006). We crossed this mouse to an FVIII E-16 knockout
mouse (Bi et al., 1995), creating a strain that was murine
FVIII-null and tolerant to hFVIII (huFVIII-R593C/E-16KO).
This mouse was then bred to homozygosity, as confirmed by
PCR and Southern blot (Fig. 1). This strain was used in all of
the following studies.
Hydrodynamic injection of an hFVIII expression
plasmid leads to long-term circulating hFVIII activity
The hFVIII expression plasmidpVB8 carries the B-domain–
deleted hFVIII cDNA under the control of the liver-specific
human a1-antitrypsin promoter and apolipoprotein E en-
hancer (Miao et al., 2000) (Fig. 2). Additionally, pVB8 carries
injected along with a uC31 integrase expression plasmid
(20mg each) into the livers of huFVIII-R593C/E-16KO mice
by hydrodynamic injection. Three different uC31 integrase
expression plasmids were used in this study. pVI expresses
thewild-type uC31integrase thathas beenusedinmostgene-
therapy studies. pVhP2 expresses a mutant uC31 integrase
shown to have increased integration activity in vitro and
in vivo in C57BL/6 mice (Keravala et al., 2009), and was tested
VIII-R593C/E-16KO mice. (A)
Breeding diagram shows how
E-16KO mice were generated
for use in the study. (B) PCR
was conducted on tail DNA
using primers that specifically
amplify the hFVIII gene pres-
ent in transgenic mice. Lanes
1–5, samples from experimen-
tal mice showing that mice 1
and 3–5 are positive for the
hFVIII gene. MW, molecular
-, negative control
mouse DNA. (C) Southern blot
of tail DNA digested with
BamHI and hybridized to a
505-bp hFVIII probe. Lane 1,
negative control (-/-); lanes
2–8, samples from experimen-
tal mice, showing that mice 2
and 6–8 were homozygous
(+/+) and mice 3–5 were
Generation of huF-
392CHAVEZ ET AL.
to determine whether it conferred an advantage. pVmI ex-
presses a catalytically inactive form of the integrase enzyme
(Keravala et al., 2009) and represented a negative control.
Following coinjection of pVB8 with pVI, pVhP2, or pVmI,
plasma was collected from treated mice over a time course,
and hFVIII activity was determined (Fig. 3). In mice that
received pVB8 with pVI, pVmI, or no integrase, the hFVIII
activity levels declined gradually over the first 2 months of
the experiment, then stabilized at relatively low values.
However, animals injected with the pVhP2 mutant form of
uC31 integrase (pVhP2) displayed a rapid drop in hFVIII
activity, with most animals having no detectable hFVIII ac-
tivity by 4 weeks post injection. At the last time point (day
168), there was no significant difference overall between
groups that received pVI and groups that received either
pVmI or no integrase plasmid. The percentages of mice ex-
pressing hFVIII, as well as the hFVIII activity levels, are
presented in Table 1.
In an attempt to increase hFVIII expression, an addi-
tional dose of pVB8+pVI was administered to a separate
group of mice 30 days subsequent to the first injection.
Plasma was collected over a time course and hFVIII ac-
tivity measured. Mice that received a second dose of
pVB8+pVI displayed significantly higher hFVIII activities
at all time points following the second injection, when
compared with mice receiving a single injection (Fig. 4).
backbone carrying the coding sequence for wild-type uC31
integrase under control of the cytomegalovirus promoter;
carrying an inactive mutant integrase; pVB8, liver-specific
hFVIII plasmid; pVB8ii, liver-specific hFVIII plasmid contain-
ing an intron.The hFVIII plasmids contain the pVax backbone,
uC31 integrase attB site, human a1-antitrypsin (hAAT) pro-
moter, B-domain–deleted hFVIII cDNA, with or without an
intron, and the bovine growth hormone polyA sequence.
uC31 integrase and hFVIII plasmids. pVI, the pVax
16KO mice injected with various plasmids. Mice were hy-
drodynamically injected with pVB8 alone, pVB8+pVmI,
pVB8+pVI, pVB8+pVhP2, or a saline solution only. Plasma
samples were assayed for hFVIII activity at the time point
indicated. Expression of hFVIII was measured by activity
assay. Values are means–SEM.
hFVIII activity in plasma of huFVIII-R593C/E-
Table 1. Frequency of Mice Having hFVIII Activity
After Injection of pVB8 and Various Integrase
Plasmids at the Final Time Point (Day 168)
TreatmentFrequency hFVIII activity (%)
16KO mice injected with a single injection of pVB8+pVI,
versus two injections of pVB8+pVI. Plasma samples were
assayed for hFVIII activity at the time point indicated. Arrow
indicates when the second injection was administered (day
30). Expression of hFVIII was measured by activity assay
through the length of the study. Values are means–SEM.
Asterisks denote values that differ statistically from a single
dose of pVB8+pVI: **p<0.05 (Student’s t test).
hFVIII activity in plasma of huFVIII-R593C/E-
hFVIII EXPRESSION IN MICE WITH uC31 INTEGRASE 393
All mice receiving two injections of pVB8+pVI exhibited
hFVIII activity at the last time point (day 168) and had a
range of expression of 7.5–49.8% (Table 1). By contrast,
only 57% of mice receiving a single injection of pVB8+pVI
expressed hFVIII at the end of the study and had a range
of expression of 5.3–18.6%. Mice that received an injection
of saline displayed no hFVIII activity at any time point
It has been shown that the addition of an intron to a cDNA
can lead to enhanced protein expression (Choi et al., 1991). To
test whether hFVIII expression could be increased in this
fashion, an intron was added to the hFVIII cDNA, creating
the plasmid pVB8ii (ii=internal intron; Fig. 2). This plasmid,
along with pVI, was injected into huFVIII-R593C/E-16KO
mice. Plasma samples were collected over a time course, and
the hFVIII activity was determined (Fig. 5A). Mice receiving
pVB8ii+pVI displayed higher hFVIII activities at all time
points tested, compared with the pVB8ii-only treated cohort.
At day 140, all mice in the pVB8ii+pVI group had hFVIII
activities of 72.7–301.2%. Within the pVB8ii-only group,
83.3% of the mice maintained hFVIII expression of 17.0–
119.4%, whereas 71.4% of the animals in the pVB8ii+pVmI
group expressed hFVIII in the range of 11.7–186.1%. As
pVB8ii resulted in much higher levels of hFVIII expression as
compared with pVB8, pVB8ii was used throughout the re-
mainder of the study.
To eliminate unintegrated pVB8ii and reveal the hFVIII
activity due to integrated pVB8ii only, a single injection of
carbon tetrachloride was administered to all mice on day
154. Carbon tetrachloride is known to induce cycling of
hepatocytes and to cause unintegrated plasmid DNA to be
lost (Liesner et al., 2010). In animals that received pVB8ii
alone or pVB8ii+pVmI, hFVIII activity levels fell below
detection limits by 2 weeks post carbon tetrachloride
treatment. In three of the four animals that received uC31
integrase, hFVIII expression was maintained after exposure
to carbon tetrachloride, albeit at reduced levels. Two ad-
ditional plasma samples were collected from the mice post
carbon tetrachloride treatment, and the hFVIII activity was
measured. Three of the four mice in the pVB8ii+pVI
group continued to express hFVIII over the next 28 days at
levels significantly higher (p<0.05) than mice that received
no functional uC31 integrase (Table 2). None of the mice
without integrase expressed any hFVIII post carbon tetra-
huFVIII-R593C/E-16KO mice injected with various plas-
mids. Mice were hydrodynamically injected with pVB8ii
alone, pVB8ii+pVmI, pVB8ii+pVI, pVB8+pVI, or a saline
solution only. Plasma samples were assayed for hFVIII ac-
tivity by activity assay and for expression by ELISA at the
time points indicated. (A) Percentages are plotted of hFVIII
activity from all groups throughout the experiment. Arrow
denotes time point when a single carbon tetrachloride in-
jection was given. Values are means–SEM. Asterisk denotes
values that differ statistically from the pVB8ii only group:
*p<0.05 (Student’s t test). (B) hFVIII expression was assayed
by ELISA at various time points. Arrow denotes time when a
single carbon tetrachloride injection was given. Negative
control animals were injected with a saline solution. Values
are means–SEM. Asterisk denotes values that differ statis-
tically from the pVB8ii only group: *p<0.05 (Student’s t test).
(C) Bleeding times following tail clip of huFVIII-R593C/E-
16KO mice injected with various plasmids. Bleeding time
was monitored over a 10-min time period. Asterisks denote
values that differ statistically from pVB8ii+pVI: *p<0.05 or
**p<0.005 (Student’s t test).
hFVIII activity, concentration, and bleeding times of
394CHAVEZ ET AL.
uC31 integrase–mediated integration provides
sustained therapeutic levels of hFVIII expression
in huFVIII-R593C/E-16KO mice
To determine the levels of circulating hFVIII, the previ-
ously collected blood samples were also subjected to hFVIII
ELISA (Fig. 5B). On day 2 post injection, all animals dis-
played high levels of circulating hFVIII, ranging from 10.0 to
96.2ng/ml. The initially high hFVIII levels dropped for all
groups, with the groups without active uC31 integrase fall-
ing most rapidly. The hFVIII levels in mice that received
pVB8ii+pVI plateaued after 28 days and remained constant
through day 140; the hFVIII levels were significantly higher
than the levels of other groups during most of the time
course (days 14, 28, 84, 140, 168, 182, and 169; p<0.05).
Animals that received either pVB8ii alone (5.7–3.2ng/ml) or
pVB8ii+pVmI (7.9–2.8ng/ml) exhibited four- and threefold
lower expression of hFVIII, respectively, compared with
those receiving pVB8ii+pVI (22.4–7.1ng/ml) at day 140
To determine the proportion of hFVIII in circulation due to
expression from integrated pVB8ii only, a single injection of
carbon tetrachloride was administered to all mice at day 154,
using the same mice as in Fig. 5A. Mice that received pVI
maintained hFVIII expression, whereas those that did not
receive pVI had no detectable hFVIII 14 days after carbon
tetrachloride injection (day 168; Fig. 5B). Mice receiving pVI
showed a small decrease in hFVIII levels after treatment.
However, these levels stabilized and were within the thera-
peutic range for the remainder of the experiment (196 days).
huFVIII-R593C/E-16KO mice treated with uC31
integrase and an hFVIII plasmid displayed
hemostatic protection in a bleeding assay
To determine if treated mice were hemostatically pro-
tected from a bleeding challenge, a tail-clip assay was per-
formed at 196 days post injection. Protection was determined
by measuring bleeding during a 10-min time period after
having 1cm of the tail clipped. Mice treated with pVB8ii+
pVI bled for a significantly shorter time than mice treated
with pVB8ii alone or pVB8ii+pVmI, or huFVIII-R593C/
E-16KO control mice (p<0.005) (Fig. 5C). Although hemo-
globin was not measured, we observed that total blood loss
correlated well with bleeding time across all groups tested.
C57BL/6 mice were used as a general positive control for
normal bleeding time and had a significantly shorter bleed-
ing time than huFVIII-R593C/E-16KO mice treated with
pVB8ii+pVI (p<0.05). These data indicate that treatment
with pVB8ii+pVI reduced the bleeding time of huFVIII-
R593C/E-16KO mice, although not to fully wild-type
Because they possess a transgenic copy of a mutated
hFVIII gene, the mice used in this study were not expected to
develop inhibitory antibodies when exposed to hFVIII, as
previously reported (Bril et al., 2006). To confirm these re-
sults, we conducted a Bethesda inhibitor assay over the time
course of day 28 through day 168 post treatment (Fig. 6).
Results of this assay revealed no detectable anti-hFVIII in-
hibitor antibodies, because none of the treated samples dis-
played any significant difference in antibody titers from
naive mouse plasma. By contrast, control plasma known to
contain hFVIII inhibitors displayed a significant level of
Table 2. Frequency of Mice Having hFVIII Activity
After Injection of pVB8ii and Various Integrase
Plasmids at Time Points Before (Day 140) and
After (Day 196) Carbon Tetrachloride Injection
4/4 (100%) 26.1–301 3/4 (75%) 52.7–129
Immune response to hFVIII was mea-
sured by Bethesda assay on four huFVIII-
R593C/E-16KO mice injected with 20mg
each of pVI+pVB8ii and an uninjected
naive control. Mice were injected with
plasmid DNA via hydrodynamic tail-vein
injection, and plasma samples were col-
lected at the times indicated. Asterisks
denote values that differ from the naive
group: **p<0.005 (Student’s t test).
Immune response to hFVIII.
hFVIII EXPRESSION IN MICE WITH uC31 INTEGRASE 395
hFVIII inhibition when compared with naive mouse plasma
in the assay (p<0.005; Fig. 6).
Detection of genomic integration within
the livers of treated mice
To determine whether uC31 integrase mediated genomic
integration within hepatocytes, we carried out PCR analysis
at a specific, known integration site commonly used by in-
tegrase in mouse liver. Although uC31 integrase mediates
plasmid integration at many different genomic loci, several
studies have demonstrated that uC31 integrase has a pre-
ferred integration site within the mouse genome that is
preferentially used in liver (Olivares et al., 2002; Held et al.,
2005; Keravala et al., 2009). Therefore, this site represents a
convenient assay for the existence of site-specific integration,
even though it represents only one of many possible inte-
gration sites. This locus is on chromosome 2 and is termed
the mouse pseudo site L1 (mpsL1). Plasmid integration can
be demonstrated at this locus by detection of a specific PCR
band that represents the junction between the chromosomal
mpsL1 site and a portion of the plasmid attB site. At the
termination of the experiment, genomic DNA was isolated
from the livers of mice from different groups, and 100ng was
used to amplify the integration junction. A PCR band indi-
cating integration of pVB8ii was detected in all mice that
received an injection of pVBii+pVI, whereas mice receiving
either pVB8ii alone or pVB8ii with pVmI did not show the
presence of this band (Fig. 7).
A gene-therapy approach for hemophilia A holds con-
siderable promise, because only modest levels of hFVIII are
needed to provide a therapeutic benefit to patients. In the
present study, we generated an FVIII knockout mouse tol-
erant to hFVIII and tested the efficacy of uC31 integrase to
mediate sustained and therapeutically relevant levels of
hFVIII expression. This mouse model allowed us to evaluate,
for the first time, long-term hFVIII expression in a mouse
model of hemophilia A without the need for immunosup-
pression or complex tolerance regimes.
In our initial experiments, we injected mice with the
hFVIII expression plasmid pVB8 and one of three different
integrase plasmids. pVI expressed wild-type uC31 integrase,
whereas pVmI encoded an inactive, negative control in-
tegrase. pVhP2 expressed a mutant integrase with higher
activity, but possessing a 33-amino acid amino-terminal ex-
tension, as well as several amino acid substitutions (Keravala
et al., 2009). When pVB8 was injected along with pVhP2, a
rapid decline in hFVIII activity was observed (Fig. 3). The
kinetics of this decline in hFVIII activity correlated well with
that of a host immune response, possibly due to novel fea-
tures of the P2 integrase sequence or structure that were
immunogenic in this strain background. Due to this obser-
vation, further use of pVhP2 was discontinued. Injection of
pVB8 alone or with pVI or pVmI resulted in similar hFVIII
expression kinetics (Fig. 3). These data suggested that pVB8
by itself provided low, sustained levels of hFVIII expression
and that a single dose of integrase provided no large benefit
during this time frame (Table 1).
In an attempt to increase the amount of circulating hFVIII,
as well as to test the consequence of administering an addi-
tional dose of uC31 integrase, a second dose of pVB8+pVI
was administered by hydrodynamic injection. It was possible
that a second injection of pVI would result in lower hFVIII
activity asa resultofimmunologicalmechanisms,because the
animals had previously been exposed to the enzyme. How-
ever, the second injection of pVB8+pVI resulted in signifi-
cantly higher and sustained hFVIII expression, as compared
with a single injection (Fig. 4). Circulating hFVIII activity
levels stabilized at *20% of normal, well within the thera-
peutic range for hemophilia A. Therefore, readministration of
pVI had a beneficial effect on the levels of hFVIII activity.
To achieve higher expression levels, a redesigned hFVIII
expression plasmid was constructed. A previous report
demonstrated that introduction of an intron into the hFVIII
cDNA resulted in increased hFVIII expression in stably
transfected cells (Plantier et al., 2001). We constructed an
hFVIII expression plasmid, pVB8ii, with a truncated version
of the human factor IX intron-1 placed at the native intron-1
location within the hFVIII cDNA (Fig. 2). When injected,
pVB8ii gave high day 2 hFVIII activities (250–330% of nor-
mal) across all groups tested. hFVIII activities this high
would not be desirable clinically, due to the elevated risk of
venous thrombosis and development of inhibitory anti-
bodies. However, such levels would not be expected clini-
cally, due the poor efficiency of hydrodynamic liver injection
in large animals to date, compared with the levels seen in
mice (Fabre et al., 2008). Other nonviral liver delivery
methods are similarly inefficient in large animals, so the
problem of excessive hFVIII levels is not one that needs to be
addressed at this time. The wide therapeutic window for
hFVIII levels is also helpful in this regard.
When pVB8ii was injected along with pVI, hFVIII activity
stabilized by day 14 at *100% of normal and was main-
tained at this level through day 140 (Fig. 5A). When injected
alone or with an inactive integrase, the initially high hFVIII
activities fell rapidly and stabilized between 25% and 50% of
normal through day 140. These data showed that although
injection of pVB8ii plasmid alone provided therapeutic levels
of hFVIII in this time frame, genomic integration of pVB8ii
resulted in statistically higher levels of hFVIII expression.
Extended expression of FVIII from unintegrated plasmid
DNA has been reported previously (Ye et al., 2004).
Liver cells are known to be quite stable, turning over at a
relatively slow rate. To induce cell cycling and to model long-
term liver-cell turnover, a single injection of carbon tetra-
chloride was administered to the mice. Carbon tetrachloride
has been shown to cause liver necrosis (Weber et al., 2003),
pseudo attP site in mouse liver. PCR demonstrates genomic
integration of pVB8ii at a preferred pseudo attP site (mpsL1).
Lanes 1–4, livers that were injected with pVB8ii+pVI; lane 5,
liver that was injected with pVB8ii only; lane 6, liver that was
injected with pVB8ii+pVmI. PCR band of 300bp indicates
plasmid integration at the mpsL1 pseudo att site. MW, DNA
molecular weight marker;+, positive control from an animal
known to have integration at mpsL1; Saline, liver injected
with saline only.
Genomic integration of pVB8ii into a genomic
396CHAVEZ ET AL.
leading to liver regeneration and loss of episomal DNA. As
can be seen in Fig. 5A, hFVIII activity levels dropped after
carbon tetrachloride administration in all groups. However,
in the pVB8ii+pVI group, hFVIII expression was maintained,
whereas in all other groups no hFVIII activity could be de-
tected. A similar expression profile was shown by ELISA (Fig.
5B). These data suggested that for hFVIII expression over the
longer term, integration may be required.
To determine if the circulating levels of hFVIII present in
treated mice were sufficient to provide protection from a
bleeding challenge, a tail-clip assay was performed. Mice
treated with pVB8ii+pVI displayed reduced bleeding times
when compared with the saline-injected control mice (Fig.
5C). Mice that received no functional integrase, and therefore
had only unintegrated pVB8ii, displayed bleeding times
similar to those of the saline-injected control mice (Fig. 5C).
This result was not unexpected, because none of these mice
had detectable hFVIII activity (Fig. 5A). Although the bleed-
ing times of pVB8ii+pVI–treated mice were longer than those
of the control C57BL/6 mice, they were significantly shorter
than when no active uC31 integrase was present.
The FVIII knockout mouse is known to develop high-titer
inhibitory antibodies to hFVIII rapidly and vigorously when
plasmid DNA is delivered by hydrodynamic injection (Ye
et al., 2004). Bethesda units of over 100 have been reported at
day 40 post plasmid injection (Ye et al., 2004). To determine if
our mouse model of hemophilia A developed inhibitor an-
tibodies to hFVIII, we conducted a Bethesda assay. As shown
in Fig. 6, we were unable to detect inhibitory antibodies to
hFVIII. This result was expected, because a previous report
had shown this mouse to be tolerant to hFVIII (Bril et al.,
2006). This result was also consistent with the long-term
presence of active hFVIII that was observed in many of the
plasma samples from treated mice.
When PCR was conducted on liver sections from the mice
at the conclusion of the study, integration of pVB8ii at the
known genomic integration hotspot mpsL1 was detected
(Fig. 7). Integration at other sites was not determined, as we
wished only to demonstrate that genomic integration had
occurred. However, it would be beneficial to the field to
examine the integration specificity of uC31 integrase more
completely by using contemporary deep-sequencing tech-
nologies. These data, taken along with the presence of cir-
culating hFVIII protein and activity, suggested that mice
treated with pVB8ii+pVI had genomically integrated pVB8ii
and were actively expressing hFVIII. uC31 integrase has
been shown to integrate plasmids more than 85% of the time
as single or double integrants per cell (Sivalingam et al.,
2010). Therefore, expression of hFVIII from one or two in-
tegrated plasmids per cell resulted in expression levels that
provided a therapeutic effect. Future studies seeking to in-
crease hFVIII expression further could include administra-
tion of a second dose of pVB8ii+pVI, and possibly the
introduction of a second intron into the pVB8ii plasmid.
Translation of these results to the clinic currently awaits the
development of efficient delivery methods for plasmid DNA
to the liver that are applicable to large mammals.
We thank W.E. Jung for help with mouse crosses and
genomic DNA analysis and Gabriel L. Ramos for assistance
with the preparation of mouse liver DNA. C.L.C. was sup-
ported in part by a Dean’s Fellowship from the Stanford
University School of Medicine. This work was supported by
NIH grant HL068112 to M.P.C.
Author Disclosure Statement
M.P.C is an inventor on Stanford-owned patents covering
Bi, L., Lawler, A.M., Antonarakis, S.E., et al. (1995). Targeted
disruption of the mouse factor VIII gene produces a model of
haemophilia A. Nat. Genet. 10, 119–121.
Brettler, D.B., Kraus, E.M., and Levine, P.H. (1995). Clinical
Aspects of and Therapy for Hemophilia A. (Churchill Livingstone,
New York, NY) pp. 1648–1663.
Bril, W.S., MacLean, P.E., Kaijen, P.H., et al. (2004). HLA class II
genotype and factor VIII inhibitors in mild hemophilia A pa-
tients with Arg593to Cys mutation. Haemophilia 5, 509–514.
Bril, W.S., van Helden, P.M., Hausl, C., et al. (2006). Tolerance to
factor VIII in a transgenic mouse expressing human factor VIII
cDNA carrying an Arg(593) to Cys substitution. Thromb.
Haemost. 95, 341–347.
Chalberg, T.C., Portlock, J.L., Olivares, E.C., et al. (2006). In-
tegration specificity of phage phiC31 integrase in the human
genome. J. Mol. Biol. 357, 28–48.
Choi, T., Huang, M., Gorman, C., and Jaenisch, R. (1991). A
generic intron increases gene expression in transgenic mice.
Mol. Cell. Biol. 11, 3070–3074.
Chuah, M.K., and VandenDriessche, T. (2004). Clinical gene
transfer studies for hemophilia A. Semin. Thromb. Hemost.
Cooper, D.N., and Tuddenham, E.G. (1994). Molecular genetics
of familial venous thrombosis. Br. Med. Bull. 50, 833–850.
Fabre, J.W., Grehan, A., Whitehorne, M., et al. (2008). Hydro-
dynamic gene delivery to the pig liver via an isolated segment
of the inferior vena cava. Gene Ther. 15, 452–462.
Gabrovsky, V., and Calos, M.P. (2008). Factoring nonviral gene
therapy into a cure for hemophilia A. Curr. Opin. Mol. Ther.
Groth, A.C., Olivares, E.C., Thyagarajan, B., and Calos, M.P.
(2000). A phage integrase directs efficient site-specific integra-
tion in human cells. Proc. Natl. Acad. Sci. U.S.A. 97, 5995–6000.
Gui, T., Reheman, A., Ni, H., et al. (2009). Abnormal hemostasis in
a knock-in mouse carrying a variant of factor IX with impaired
binding to collagen type IV. J. Thromb. Haemost. 7, 1843–1851.
Held, P.K., Olivares, E.C., Aguilar, C.P., et al. (2005). In vivo
correction of murine hereditary tyrosinemia type I by uC31
integrase-mediated gene delivery. Mol. Ther. 11, 399–408.
Hoyer, L.W., and Scandella, D. (1994). Factor VIII inhibitors:
structure and function in autoantibody and hemophilia A
patients. Semin. Hematol. 31, 1–5.
Jin, D., Zhang, T., Gui, T., et al. (2004). Creation of a mouse
expressing defective human factor IX. Blood 104, 1733–1739.
Kemball-Cook, G., Tuddenham, E.G.D., and Wacey, A.I. (1998).
The factor VIII structure and mutation resource site: HAM-
STeRS version 4. Nucleic Acids Res. 26, 216–219.
Keravala, A., Lee, S., Thyagarajan, B., et al. (2009). Mutational
derivatives of phiC31 integrase with enhanced efficiency and
specificity. Mol. Ther. 17, 112–120.
Kuhstoss, S., and Rao, R.N. (1991). Analysis of the integration
function of the Streptomycete bacteriophage FC31. J. Mol. Biol.
hFVIII EXPRESSION IN MICE WITH uC31 INTEGRASE 397
Laird, P.W., Zijderveld, A., Linders, K., et al. (1991). Simplified Download full-text
mammalian DNA isolation procedure. Nucleic Acids Res. 19,
Liesner, R., Zhang, W., Noske, N., and Ehrhardt, A. (2010).
Critical amino acid residues within the phiC31 integrase
DNA-binding domain affect recombination activities in
mammalian cells. Hum. Gene Ther. 21, 1104–1118.
Ma ´trai, J., Chuah, M.K., and VandenDriessche, T. (2010). Pre-
clinical and clinical progress in hemophilia gene therapy.
Curr. Opin. Hematol. 17, 387–392.
Miao, C.H., Ohashi, K., Patijn, G.A., et al. (2000). Inclusion of the
hepatic locus control region, an intron, and untranslated re-
gion increases and stabilizes hepatic factor IX gene expression
in vivo but not in vitro. Mol. Ther. 1, 522–532.
Olivares, E.C., Hollis, R.P., Chalberg, T.W., et al. (2002). Site-
specific genomic integration produces therapeutic factor IX
levels in mice. Nat. Biotechnol. 20, 1124–1128.
Petrus, I., Chuah, M., and VandenDriessche, T. (2010). Gene
therapy strategies for hemophilia: benefits versus risks. J.
Gene Med. 12, 797–809.
Plantier, J.L., Rodriguez, M.H., Enjolras, N., et al. (2001). A factor
VIII minigene comprising the truncated intron I of factor IX
highly improves the in vitro production of factor VIII.
Thromb. Haemost. 86, 596–603.
Powell, J.S., Ragni, M.V., White, G.C., 2nd, et al. (2003). Phase 1
trial of FVIII gene transfer for severe hemophilia A using a
retroviral construct administered by peripheral intravenous
infusion. Blood 102, 2038–2045.
Qian, J., Collins, M., Sharp, A.H., and Hoyer, L.W. (2000). Pre-
vention and treatment of factor VIII inhibitors in murine he-
mophilia A. Blood 95, 1324–1329.
Rausch, H., and Lehmann, M. (1991). Structural analysis of the
actinophage FC31 attachment site. Nucleic Acids Res. 19,
Roth, D.A., Tawa, N.E., Jr., O’Brien, J.M., et al. (2001). Nonviral
transfer of the gene encoding coagulation factor VIII in patients
with severe hemophilia A. N. Engl. J. Med. 344, 1735–1742.
Sivalingam, J., Krishnan, S., Ng, W.H., et al. (2010). Biosafety
assessment of site-directed transgene integration in human
umbilical cord-lining cells. Mol. Ther. 18, 1346–1356.
Thorpe, H.M., and Smith, M.C.M. (1998). In vitro site-specific
integration of bacteriophage DNA catalyzed by a recombinase
of the resolvase/invertase family. Proc. Natl. Acad. Sci. U.S.A.
Thyagarajan, B., Olivares, E.C., Hollis, R.P., et al. (2001). Site-
specific genomic integration in mammalian cells mediated by
phage uC31 integrase. Mol. Cell. Biol. 21, 3926–3934.
Tillett, D., and Neilan, B.A. (1999). Enzyme-free cloning: a rapid
method to clone PCR products independent of vector restric-
tion enzyme sites. Nucleic Acids Res. 27, e26.
Weber, L.W., Boll, M., and Stampfl, A. (2003). Hepatotoxicity
and mechanism of action of haloalkanes: carbon tetrachloride
as a toxicological model. Crit. Rev. Toxicol. 33, 105–136.
Wu, H., Reding, M., Qian, J., et al. (2001). Mechanism of the
immune response to human factor VIII in murine hemophilia
A. Thromb. Haemost. 85, 125–133.
Ye, P., Thompson, A.R., Sarkar, R., et al. (2004). Naked DNA
transfer of factor VIII induced transgene-specific, species-
independent immune response in hemophilia A mice. Mol.
Ther. 10, 117–126.
Address correspondence to:
Dr. Michele P. Calos
Department of Genetics
Stanford University School of Medicine
Stanford, CA 94305-5120
Received for publication June 29, 2011;
accepted after revision November 11, 2011.
Published online: November 11, 2011.
398 CHAVEZ ET AL.