ABSTRACT
Site-directed DNA integration has been achieved by using a pair of mutant
lox
sites, a right element (RE) mutant
lox
site and a left element (LE) mutant
lox
site [Albert
et al
. (1995)
Plant J
., 7, 649-659], in mouse embryonic stem (ES) cells. We established ES cell lines
carrying a single copy of the wild-type
lox
P
or LE mutant
lox
site as a target and examined the frequency of site-specific integration of a targeting vector carrying a
lox
P
or RE mutant
lox
site induced by Cre transient expression. Since our targeting vector contains a
complete
neo
gene, random integrants can form colonies as in the case of a gene targeting
event through homologous recombination. With our system, the frequency of site-specific integration
via the mutant
lox
sites reached a maximum of 16%. In contrast, the wild-type
lox
P sites yielded very low frequencies (<0.5%) of site-specific integration events. This mutated
lox
system will be useful for `knock-in' integration of DNA in ES cells.
The Cre-
lox
recombination system of bacteriophage P1 is currently the most powerful tool
for genetic manipulation both
in vitro
(
1
-
5
) and
in vivo
(
6
-
9
). Cre recombinase catalyzes reciprocal site-specific recombination between two
lox
P sites. Consequently, Cre mediates both intramolecular (excisive or
inversional) and intermolecular (integrative) recombination. In integrative
recombination, a circular DNA carrying a
lox
P site is inserted into a
lox
P site on a chromosome. However, this integration reaction is quite inefficient,
because the integrated DNA, which has
lox
P sites at both ends, is easily removed again though excisive recombination if
the Cre recombinase is still present (Fig.
1
A). Therefore, a special selection system in which only targeted integrants can
survive is indispensable for targeted integration into
lox
P sites. For example, Sauer and Henderson (
1
) used a promoter fusion strategy. In this system, the targeting vector contains
a
lox
P site followed by a promoterless
tk
gene and the chromosomal target is a
lox
P site located 3' of a promoter. Only upon
lox
P-specific targeted integration does the promoterless
tk
gene on the targeting vector fuse to the chromosomally placed promoter, thereby
making the cells HAT resistant. Fukushige and Sauer (
2
) further improved this strategy using an ATG-less
lox
-
neo
translational fusion system. In this system, the targeting vector contains a
promoter followed by ATG-
lox
and the chromosomal target is a
lox
P site located 5' of the ATG-less
neo
gene. Only upon site-specific recombination dose the promoter-ATG-
lox
element on the targeting vector fuse with the ATG-less
neo
gene, resulting in the generation of a functional
lox
-
neo
fusion gene to confer G418 resistance on the cells.
Recently, Albert
et al
. (
10
) devised a new strategy different from the methods mentioned above. They
identified three sets of mutant
lox
sites that favor integrative recombination over the excisive reaction. The
lox
P site is composed of an asymmetric 8 bp spacer flanked by 13 bp inverted
repeats. They introduced nucleotide changes into the left 13 bp element (LE
mutant
lox
site) or the right 13 bp element (RE mutant
lox
site) (Fig.
1
C). Recombination between the LE mutant
lox
site and the RE mutant
lox
site produces the wild-type
lox
P site and a LE+RE mutant site that is poorly recognized by Cre, resulting in
stable integration (Fig.
1
B and C).
This LE/RE mutant system has many potential uses, because it does not need any
special system for selecting targeted integration. Only a mutant
lox
site is needed as the chromosomal target. However, the efficiency of targeted
integrative recombination in the LE/RE mutant system in mammalian cells is not
known. Albert
et al
. used plant, not mammalian, cells and since they utilized the promoter fusion
strategy as well as the LE/RE mutant system, they could not precisely determine
the efficiency of recombination in the LE/RE mutant system. In this study we
have used embryonic stem (ES) cells, which are extensively used in genome
engineering and adopted a simple selection system to examine the frequency of
targeted insertion over random integration. We show here that the LE/RE mutant
system is more effective than wild-type
lox
P and that the efficiency of targeted integration was in the range 2-16%.
The
lox
P,
lox
66 and
lox
71 sequences (
10
) were synthesized and cloned into pBluescript (pBS)-SK-
(Stratagene). The pCAGloxPbsr and pCAGlox71bsr plasmids were constructed by
first introducing a
bsr
Hin
dIII fragment from pSV2bsr (Kaken-Seiyaku, Tokyo) (
11
) into
lox
P- or
lox
71-containing plasmids. Then, the
lox
P-
bsr
or
lox
71-
bsr
fragment was ligated into pBS-CAGG, which was constructed by inserting a
Sal
I-
Pst
I fragment from pCAGGS (
12
) into pBS-SK-
.
Plasmid ploxPNZneo and plox66NZneo were assembled from components of pBSnlslacZ,
pMC1NeopolyA (Stratagene) and
lox
sequences. Plasmid pBSnlslacZ contains the
lacZ
gene fused with the nuclear localization signal (NLS) derived from the SV40
large T gene and an intron and poly(A) signal derived from SV40 (
13
). The MC1neopolyA fragment was inserted into the 3'-end of NLS-
lacZ
and the
lox
P or
lox
66 fragment was inserted into the 5'-end of NLS-
lacZ
.
The Cre expression vector, pCAGGS-Cre, was constructed by inserting the
cre
fragment from pBS185 (Life Technologies) into pCAGGS (
12
).
The ES cell line TT2 (
14
) was grown as described (
15
) except for the use of G418-resistant primary mouse embryo fibroblasts as feeder layers.
In the case of electroporation with the pCAGloxPbsr or pCAGlox71bsr plasmids, 10
[mu]g
Spe
I-digested DNA and 1 * 107
cells were used. The cells were electroporated using a BioRad Gene Pulser and
after 48 h they were fed with medium supplemented with 4 [mu]g/ml Blasticidin S (Kaken-Seiyaku, Tokyo). Selection was maintained for 5 days and then colonies
were picked into 24-well plates and expanded for freezing. The clones were analyzed by
Southern blotting to select cell lines showing a single copy integration
pattern.
For the electroporation experiments designed to detect targeted integration into
the
lox
site, ploxPNZneo, plox66NZneo and pCAGGS-Cre were used in their circular forms. The electroporated cells were
selected with G418 at 200 [mu]g/ml for 1 week. Colonies were stained with X-gal as described (
16
) or picked and expanded for DNA analysis.
Six micrograms of genomic DNA were digested with appropriate restriction
enzymes, electrophoresed on a 0.9% agarose gel and then blotted onto a nylon
membrane (Boehringer Mannheim). Hybridization was performed using a DIG DNA
Labeling and Detection Kit (Boehringer Mannheim). For PCR analysis, DNAs (0.1-0.5 [mu]g) were subjected to 28 cycles of amplification (each cycle
consisting of 1 min at 94oC, 2 min at 55oC and 2 min at 72oC) with a thermal cycler. The 5'-primers and 3'-primers were AG2 (5'-CTGCTAACCATGTTCATGCC-3') and LZUS3
(5'-GCGCATCGTAACCGTGCAT-3') for amplification of the 5'-junction fragment and T7 (5'-AATACGACTCAGTATAG-3') and BSR1 (5'-CTTCTCTGTCGCTACTTCTAC-3') for amplification of the 3'-junction fragment. One half of the PCR reaction product was loaded and analyzed on an agarose gel.
The nucleotide sequences of the regions surrounding the
lox
sites derived from PCR from the targeted clones were determined with a thermo
sequenase fluorescent labeled primer cycle sequencing kit (US Biochemicals).
The experimental design outlined in Figure
2
was used to assess the efficiency of targeted recombination between the mutant
lox sequences,
lox
71 and
lox
66 (Fig.
1
), which we chose from the three sets of mutants reported by Albert
et al
. (
10
). We established six ES cell lines carrying a single copy of
lox
P (loxP-5 and loxP-6) or
lox
71 (lox71-2, lox71-19, lox71-20 and lox71-22). The
lox
site was placed between the CAG promoter and the
bsr
gene as a target. Since the transformants were selected with Blasticidin S
(i.e.
bsr
gene expression), the CAG promoter was active in the cell lines. After
introducing the targeting plasmids, ploxPNZneo and plox66NZneo, we could detect
lox
site-mediated integration by monitoring expression of the inserted gene, i.e.
the
lacZ
gene fused with the NLS derived from the SV40 large T gene (NLS-
lacZ
) in this study. The targeting plasmids also contained the complete
neo
gene with a promoter and poly(A) signal (see Fig.
2
). In this system, although both random and targeted integrants become G418
resistant, only the targeted integrants are stained blue by X-gal and the percentage of blue colonies indicates the frequency of
targeted integration. It is expected that in the case of site-specific recombination between wild-type
lox
P sites, the inserted vector will be excised by the remaining Cre recombinase
activity. On the other hand, targeted integration via
lox
66 and
lox
71 will be stable because of the poor affinity of the LE+RE mutant site for the
Cre recombinase.
The cell lines carrying a
lox
P or
lox
71 site were co-electroporated with a constant amount (20 [mu]g) of ploxPNZneo or plox66NZneo and various amounts of the Cre
expression vector, pCAGGS-Cre, as indicated in Table
1
. After G418 selection for 1 week, the colonies were stained with X-gal and then scored. All of the positive colonies turned blue within 2 h,
reflecting the strong activity of the CAG promoter. The blue colonies derived
from the same parental cell line showed almost the same intensity of staining,
suggesting that no `gene-trap'-type integration had occurred. However, the staining intensities
varied among different parental lines, probably due to the different promoter
activity in each parental cell line. The results are summarized in Table
1
. The targeting frequencies for normal
lox
P sites were very low (<0.5%) irrespective of the amount of pCAGGS-Cre. We could find two blue colonies at most. On the other hand, the
frequencies of targeted integration between
lox
66 and
lox
71 were in the range 2-16% with 20 [mu]g each of pCAGGS-Cre and plox66NZneo. These results indicate that the LE/RE
mutant system is more efficient as to Cre-mediated site-specific integration. When we introduced plox66NZneo into the
lox
P-carrying cell lines or ploxPNZneo into the
lox
71-carrying cell lines, the frequencies were low (0.3-1%), although they were slightly better than in the case of
lox
P-
lox
P recombination. This demonstrates that the combination of the RE and LE mutants
is important for achieving targeted integration.
Table 1
We found differences in the targeting frequency among the four cell lines
carrying
lox
71. Cell line lox71-19 showed 8-fold higher efficiency than cell line lox71-2. These results indicate that the targeting efficiency may
depend on the chromosomal position of the integration site. Interestingly, the
positive colonies derived from cell line lox71-19 showed the strongest X-gal staining among the four lines (data not shown), suggesting the
possibility that the
lox
71 target site is located at an `active' chromosomal position. Concerning the
positive colonies derived from the
lox
P-carrying cell line, their staining intensities were similar to those
derived from cell lines lox71-20 and lox71-22. Therefore, the low targeting efficiency of
lox
P sites is not due to a position effect.
To confirm site-directed integration of the vector into chromosomal
lox
targets, we picked 48 colonies from among the transformants of cell line lox71-19 electroporated with 20 [mu]g each of pCAGGS-Cre and plox66NZneo. Five out of the 48 clones showed positive X-gal staining. We prepared genomic DNAs from these five
clones and also from seven negative clones and analyzed them by Southern
blotting using pBS as a probe. As shown in Figure
3
A, targeted integration should give a 4.3 kb band, whereas bands of various
sizes are expected in the case of random integration. In agreement with this
expectation, all the five X-gal-positive clones gave a 4.3 kb band (Fig.
3
B, lanes 1-5), in contrast to the `white' clones (lanes 6-12). Since there were no extra bands other than the 4.3 kb one, a
single copy of the vector was inserted in each case. PCR analyses were also
performed for confirmation. The junction of the integration can be amplified
using the primer pairs AG2 and LZUS3 for the 5'-junction and T7 and BSR1 for the 3'-junction (see Fig.
3
A). As shown in Figure
3
C, only the X-gal-positive clones gave a band of the expected size. Furthermore, we
determined the nucleotide sequence surrounding the
lox
sites of the PCR products and confirmed generation of the wild-type
lox
P site and the double mutant
lox
site (data not shown). These results demonstrate that the clones stained with X-gal are targeted integrants and that the recombination occurred between
lox
sites.
We show here that the mutant
lox
sites promoted site-specific integration events much more frequently than those obtained with
wild-type
lox
sites in ES cells. The reason for this clear-cut result is probably the use of a simple selection system. It allows us
to score the efficiency of targeted integration over random integration by
counting blue colonies. We believe this is the first report showing successful
Cre-mediated site-specific targeting without selection of targeting events. The
highest frequency of site-specific integration was 16% with cell line lox71-19. This frequency is comparable to that of `knock-out' gene targeting, in which negative selection or promoter
trapping is not involved. Therefore, this LE/RE mutant system is a practical
method for genetic manipulation in ES cells.
The efficiency of Cre-mediated, targeted integration depends on the amount of Cre expression
vector and thus the level of Cre recombinase activity. In our experiment, the
optimal amount was 20 [mu]g and a higher or lower amount of the Cre expression plasmid reduced the
targeting efficiency. Baubonis and Sauer (
3
) reported a similar observation for purified Cre protein in the
lox
-
neo
fusion system. Since the CAG promoter has strong activity, 3- to 5-fold higher than that of the phosphoglycerate kinase-1 (PGK) promoter (unpublished data), we originally expected
that <20 [mu]g would be optimal for targeted recombination. Our results suggest that it
is necessary to use a strong promoter for effective insertion. On the other
hand, an excess amount of the Cre expression plasmid causes excisive
recombination, even when a double mutant
lox
site is used. If a more effective pair of mutant
lox
sequences is found, the targeting frequency might increase.
We observed an 8.5-fold difference in targeting efficiency among the cell lines carrying
lox
71. Since all the cell lines used carried a single copy of the target
lox
site, the chromosomal position effect is considered to be the cause of the
difference. Baubonis and Sauer (
3
) also observed a 50-fold difference in targeting efficiency per target
lox
P site in human osteosarcoma cell lines using purified Cre protein. As they used
cell lines carrying different numbers of
lox
P sites, the difference in targeting efficiency could be larger. Our finding
that the highest
lacZ
-expressing cell line gave the highest targeting efficiency suggests that
targeting efficiency might be proportional to the level of gene expression,
although the number of examined cell lines was limited. Increasing the number
of parental cell lines could make the cause(s) of the position effects on
recombination and gene expression more apparent.
When we introduced plox66NZneo into the
lox
P-carrying cell lines or ploxPNZneo into the
lox
71-carrying cell lines, the frequencies were low, as in the case of
lox
P-
lox
P recombination. This indicates that excisive recombination between the
lox
71 and
lox
P sites occurs at a normal rate.
The advantage of this mutant
lox
system is that only a mutant
lox
site is needed as the chromosomal target. We believe that the method described
here will be useful for genetic manipulation in ES cells, including conditional
gene targeting and gene trapping, as this system allows site-specific integration of any DNA sequence into a defined
lox
site.
We wish to thank Y.Kiyonaga for technical assistance and Dr K.Abe for critical
reading of the manuscript. This work was supported by grants from the Ministry
of Education, Science and Culture, by a grant from the Yamanouchi Foundation
for Research on Metabolic Disorders, by a grant from the Osaka Foundation for
Promotion of Clinical Immunology and a grant from the Science and Technology
Agency.
*To whom correspondence should be addressed. Tel: +81 96 373 5316; Fax: +81 96
373 5321; Email: yamamura@gpo.kumamoto-u.ac.jp
REFERENCES
Return