COMMUNIC
Site-specific Recombination
by Phage l Integrase Mutant
S
ras
ve
nct
ou
FL
per
at t
8 (E
nt t
-tra
bi
l li
. R
eth
tors, which could functionally replace the prokaryotic ones normally
required for wild-type Int, are most likely not present in human cells.
*Corresponding author
The recent application of sit
nases has improved significantly
niques for mammalian geno
Notably, Cre and FLP, both m
grase family of conservative si
nases, are used widely in stud
relevance and function of partic
est (for reviews, see Sauer, 1994;
The phage l-encoded integras
type of the integrase family. In
t
m
attB/attP and attL/attR, respectively (for reviews,
sequential pair of strand exchanges during integra-
doi:10.1006/jmbi.2000.3532 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 296, 1175–1181see Landy, 1989, 1993). Each att sequence is com-
posed of two inverted 9 bp core Int binding sites
and a 7 bp overlap region, which is identical in all
tive and excisive recombination (Landy, 1993). The
natural pairs of target sequences for Int, attB and
attP or attL and attR, can be located either on the
same or two different DNA molecules, leading to
intra- or intermolecular recombination, respect-
ively. Here, we consistently use the terms integra-
tive and excisive recombination to describe
intramolecular recombination between inversely
E. Lorbach and N. Christ contributed equally to this
work.
Abbreviations used: Int, phage l-encoded integrase;
XIS, phage-encoded exisionase; IHF, integration hostintegration and excision of the
of the Escherichia coli genome
tion between pairs of attachfactor; FIS, factor for inversion stim
encoded exisionase.
E-mail address of the correspond
p.droege@uni-koeln.de
0022-2836/00/051175–7 $35.00/0# 2000 Academic Press
Keywords: site-specific recombination; l integrase; mutant recombinase;
eukaryotes; genomic engineering
e-specific recombi-
the available tech-
me manipulation.
embers of the inte-
te-specific recombi-
ies elucidating the
ular genes of inter-
Mu
¨
ller, 1999).
e (Int) is the proto-
t executes both the
phage into and out
hrough recombina-
ent sites, termed
wild-type att sites. In addition to core sites where
strand cleavage and religation occurs, each site
except attB contains additional arm Int binding
sites. A varying number of recognition sequences
for the accessory DNA-bending proteins inte-
gration host factor (IHF), factor for inversion
stimulation (FIS), and the phage-encoded excisio-
nase (XIS) protein are also present in the flanking
regions, again with the exception of attB. Int is a
heterobivalent DNA-binding protein and, with
assistance from the accessory proteins and negative
DNA supercoiling, is able to bind simultaneously
to core and arm sites within the same att site.
Int, like Cre and FLP, executes an ordered,Elke Lorbach, Nicole Christ, Micha
Institute of Genetics, University
of Cologne, Weyertal 121
D-50931 Cologne, Germany
Phage lambda Integ
family of conservati
FLP. The natural fu
the phage into and
contrast to Cre and
teins and DNA su
Here, we show th
derivative Int-h/21
factors, are proficie
recombination in co
lar integrative recom
human reporter cel
genomic sequences
detectable by this mo
s
i
ulation; XIS, phage-
ing author:ATION
in Human Cells Catalyzed
s
chwikardi and Peter Dro¨ge*
e (Int) is the prototype of the so-called integrase
site-specific recombinases, which includes Cre and
ion of Int is to execute integration and excision of
t of the Escherichia coli genome, respectively. In
P, however, wild-type Int requires accessory pro-
coiling of target sites to catalyze recombination.
wo mutant Int proteins, Int-h (E174 K) and its
174 K/E218 K), which do not require accessory
o perform intramolecular integrative and excisive
nsfection assays inside human cells. Intramolecu-
nation is also detectable by Southern analysis in
nes harboring target sites attB and attP as stable
ecombination by wild-type Int, however, is not
od. The latter result implies that eukaryotic co-fac-riented attB and attP, or between attL and attR
equences, respectively. Both reactions thus lead to
nversion of the intervening DNA segment.
# 2000 Academic Press

As outlined above, an important difference We co-introduced pGFPattB/attP either with
1176 Genomic Engineering by Phage  Integrase Mutantsbetween the Int and Cre/FLP system is that Int
requires additional protein factors for integrative
and excisive recombination, and negative super-
coiling for integrative recombination. It is this com-
plication that has presumably prevented
application of the l Int system in eukaryotic gene
targeting techniques. As a first step towards the
transfer of the Int system to mammalian cells, we
have therefore explored the use of two mutant Int
proteins, Int-h and its derivative Int-h/218. Both
enzymes promote integrative recombination in the
absence of accessory proteins in E. coli (Miller et al.,
1980; Christ & Dro
¨
ge, 1999) and DNA supercoiling
in vitro (Lange-Gustafson & Nash, 1984). Likewise,
excisive recombination by Int-h (Int-h/218) can
occur in the absence of IHF and XIS (Christ &
Dro¨ge, 1999). The potential to perform recombina-
tion in the absence of accessory factors has been
explained by an enhanced affinity of Int-h for core
recognition sequences (Patsey & Bruist, 1995).
In order to test whether Int-h (Int-h/218) is com-
petent to perform integrative and/or excisive
recombination inside human cells, it was first
necessary to demonstrate that Int-h protein is syn-
thesized. We constructed eukaryotic expression
vectors, termed pKEXInt-h and pPGKInt-h, which
contain the Int-h coding region under control of
the CMV and the PGK promoter, respectively
(Figure 1(a)). After transient transfection of pKEX-
Int-h into two BL60 reporter cell lines, termed B2
and B3 (see below), we could detect both unpro-
cessed and correctly spliced Int-h mRNAs by RT-
PCR (data not shown). Cell lysates were analyzed
72 hours after transfection of pKEXInt-h by Wes-
tern blotting using mouse polyclonal antibodies
raised against Int. The parental vector pKEX, lack-
ing the Int-h gene, was used as a control. The
results show that a protein with the expected mol-
ecular mass of Int-h is detectable in both cell lines
when pKEXInt-h was introduced (Figure 2(a); com-
pare lanes 2 and 4 with M). This protein is absent
from cells containing the control vector (lanes 1
and 3). The Int-h protein can also be detected in
HeLa cells after transfection of pPGKInt-h and its
derivative pPGKInt-hNLS; the latter carries a
nuclear localization signal fused to the C terminus
of the Int-h gene (Figure 2(b), lanes 1 and 2,
respectively). The protein is absent when the con-
trol vector pPGK is used (lane 3). We conclude that
presumably full-length Int-h protein is produced
by two different human cell lines.
In order to test whether the Int-h protein pro-
duced in human cells can perform integrative
recombination, we constructed substrate vector
pGFPattB/attP (Figure 1(b)). The two recombina-
tion sites attB and attP are in inverted orientation
with respect to each other and flank the gene for
green fluorescence protein (GFP). The GFP gene
itself is in inverted orientation with respect to the
CMV promoter. Hence, recombination between
attB and attP should lead to its inversion and sub-
sequent expression.pPGKInt-h or pKEXInt-h to HeLa or BL60 cells,
respectively. DNA was isolated 72 hours after
transfection and recombination was monitored by
PCR employing primer pair p3/p4. These primers
should yield a 0.99 kb product if inversion of the
GFP gene has occured (compare Figure 1(b)). The
results show that the expected product is detect-
able in both cell lines only when the respective
expression vector is present (Figure 3(a), lanes 1
and 4). Control PCRs furthermore confirmed that
the substrate pGFPattB/attP is present in all trans-
fected cell lines (data not shown).
In order to test for excisive recombination, we
constructed pGFPattL/attR. This vector is identical
with pGFPattB/attP, except that attL and attR
replace attB and attP, respectively (Figure 1(c)).
Recombination assays were performed as
described for pGFPattB/attP and the results show
that the expected PCR product (1.09 kb) is detect-
able only when the respective expression vector
was introduced together with pGFPattL/attR
(Figure 3(b), lanes 1 and 3). The presence of the
substrate vector in cell lines was again confirmed
by PCR. DNA sequencing of isolated PCR pro-
ducts that result either from integrative or excisive
recombination confirmed that the strand transfer
reactions occured as expected for wild-type Int
(data not shown). We conclude that Int-h is profi-
cient at performing integrative and excisive recom-
bination on episomal DNA substrates in two
different human cell lines.
In order to test whether Int-h can also catalyze
integrative recombination when the target sites are
stably integrated into the host genome, we gener-
ated three BL60 reporter cell lines (B1-B3).
Southern analysis of their genomic DNA confirmed
that B1 and B3 carry multiple copies of pGFPattB/
attP integrated as tandem repeats, while B2 con-
tains only a single copy (data not shown). Vectors
pKEXInt-h and pKEX were transfected separately
into these reporter cell lines. Cells were harvested
72 hours after electroporation, genomic DNAs pur-
ified, and analyzed by PCR for recombination
events. The results revealed that the expected pro-
ducts generated with primer pairs p3/p4 and p1/
p2 (compare Figure 1(b)) are detectable only when
pKEXInt-h is introduced into the cells (data not
shown). DNA sequencing of isolated PCR products
confirmed that both attL and attR are present, and
that the GFP gene has been inverted. Furthermore,
RT-PCR analysis confirmed that the GFP gene is
expressed due to recombination (data not shown).
We performed these assays three times and could,
in each case, detect recombination between attB
and attP by genomic PCR and RT-PCR in all three
cell lines expressing Int-h.
In order to test whether wild-type Int and a
second Int mutant, Int-h/218 (Christ & Dro
¨
ge,
1999), is proficient for integrative recombination
in a HeLa reporter cell line, we introduced
pCMVSSInt and pCMVSSInt-h/218, as well as
pCMVSSInt-h and pCMV as positive and negative