DTD 5
ARTICLE IN PRESS
YMTHE-02698; No. of pages: 8; 4C: 4
ARTICLE
doi:10.1016/j.ymthe.2005.06.094
Custom Zinc-Finger Nucleases for Use in Human Cells
Stephen Alwin,1 Maja B. Gere,1 Eva Guhl,1 Karin Effertz,2 Carlos F. Barbas III,2 David J. Segal,3
Matthew D. Weitzman,4 and Toni Cathomen1,*
2
1
Institute of Virology, Charité Medical School, Campus Benjamin Franklin, Hindenburgdamm 27, 12203 Berlin, Germany
Skaggs Institute for Chemical Biology and Department of Molecular Biology, The Scripps Research Institute, La Jolla, CA 92037, USA
3
Department of Pharmacology and Toxicology, University of Arizona, Tucson, AZ 85721, USA
4
Laboratory of Genetics, The Salk Institute, La Jolla, CA 92037, USA
*To whom correspondence and reprint requests should be addressed. Fax: +49 30 8445 3840. E-mail: toni.cathomen@charite.de.
Genome engineering through homologous recombination (HR) is a powerful instrument for
studying biological pathways or creating treatment options for genetic disorders. In mammalian
cells HR is rare but the creation of targeted DNA double-strand breaks stimulates HR significantly.
Here, we present a method to generate, evaluate, and optimize rationally designed endonucleases
that promote HR. The DNA-binding domains were synthesized by assembling predefined zinc-finger
modules selected by phage display. Attachment of a transcriptional activation domain allowed
assessment of DNA binding in reporter assays, while fusion with an endonuclease domain created
custom nucleases that were tested for their ability to stimulate HR in episomal and chromosomal
gene repair assays. We demonstrate that specificity, expression kinetics, and protein design are
crucial parameters for efficient gene repair and that our two-step assay allows one to go quickly
from design to testing to successful employment of the custom nucleases in human cells.
Key Words: artificial endonucleases, chimeric nucleases, gene correction, site-directed
recombination, zinc-finger
INTRODUCTION
Manipulations to modify the genetic information of an
organism have been applied to answer basic biological
questions or to create treatment options for inherited
disorders. Most of the current methods used in gene
therapy protocols rely on random integration of a transgene. However, since gene expression is typically finetuned among tissues or during development and the cell
cycle, random integration of a transgene under control of
a heterologous promoter is a rather crude technique to
alter the genome. Moreover, random integration can
have genotoxic side effects by disrupting cellular genes or
by misregulating expression of nearby genes, as recently
shown in a gene therapy trial involving randomly
integrating retroviral vectors [1–4].
Genetic manipulations that allow the precise replacement of one sequence with another can be accomplished
through homologous recombination (HR). Unfortunately, HR is a rare event in mammalian cells, while
random integration is much more frequent [5]. The
creation of a targeted DNA double-strand break (DSB),
however, has been shown to stimulate HR several
hundredfold by activating the cellular machinery responsible for homology-directed repair [6–9]. The repair of
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
1525-0016/$30.00
DSBs is crucial to the survival of mammalian cells and
could therefore be a powerful tool to harness for genome
engineering. By intentionally creating a DSB at a specific
site in the genome, most recombination events can be
directed to occur between the target locus and a DNA
template that has been introduced into the cell.
Nucleases that cleave DNA at prechosen sites to stimulate
site-directed recombination (SDR) can be generated
either by recombining existing homing endonucleases
[10,11] or by rational protein design. Chimeric nucleases
that consist of a modified zinc-finger DNA-binding
domain fused to the endonuclease domain of the FokI
restriction enzyme have been shown to cleave DNA
specifically in vitro and in vivo, although some cytotoxicity issues remain unresolved [12–18].
In this paper, we engineer rationally designed zincfinger nucleases (ZFNs) and present a rapid method to
evaluate them. The custom nucleases were generated by
attaching the catalytic domain of FokI to a targetable
DNA-binding domain, which was synthesized by assembling predefined C2H2–zinc-finger modules obtained by
phage display [19,20]. Each module is 28 amino acids in
length and includes an a-helix that contacts 3 bases in
the major groove of DNA. The modules can be assembled
1
ARTICLE IN PRESS
ARTICLE
in the specific order necessary to recognize any given
sequence in a target locus [19–21]. Because the FokI
nuclease domain has to dimerize to become active [16],
two separate custom nucleases were generated. Each
individual ZFN contains three zinc-finger modules recognizing a 9-bp binding site. The complete target
sequence thus consists of two 9-bp binding sites in
opposite orientation separated by a 6-bp spacer [15]. We
have engineered novel custom nucleases and demonstrate that specificity of DNA-binding, expression
kinetics, and protein design are crucial for a low level of
cytotoxicity and a high frequency of SDR.
RESULTS
DNA-Binding Specificity
Affinity and specificity of DNA binding are crucial to the
successful application of artificial nucleases in vivo. As
proof-of-principle, we synthesized four novel DNA-binding domains that recognize four unrelated target sites (Fig.
1A). To assess DNA binding separate from DNA cleavage,
we fused the zinc-finger domains to the transcriptional
activation domain of VP16 and evaluated them for their
ability to activate expression of a reporter gene (Fig. 1B).
We performed assays in 293T cells using reporter constructs containing a single binding site for either GZF1
and GZF2 (1-2/Luc) or GZF3 and GZF4 (3-4/Luc) cloned
upstream of a minimal promoter (Fig. 1C). Although
performed with plasmid DNA, the assay allows assessment
of the specificity of DNA binding in the background of the
human genome. Western blot analysis confirmed the
steady-state levels of all zinc-finger transcription factors to
be similar (Fig. 1D), but only expression of AD-GZF1 and
AD-GZF3 activated expression of luciferase significantly.
doi:10.1016/j.ymthe.2005.06.094
The control zinc-finger protein AD-ZF5C [22] did not
stimulate luciferase expression from either reporter construct. These results imply that GZF1 and GZF3 encode
highly specific DNA-binding domains, while specificity
and/or affinity of GZF2 and GZF4 is not sufficient to
mediate target recognition. This reporter assay thus
provides a convenient way of quickly evaluating zincfinger-mediated binding to a specific DNA sequence.
Design and Expression Kinetics of Custom Nucleases
We then used custom nucleases based on the GZF3 DNAbinding domain to scrutinize design, expression kinetics,
and cytotoxicity issues. Since the length of the amino acid
linker between the zinc-finger domain and the catalytic
FokI domain is critical for efficient cleavage of target DNA
[13,15], we designed GZF3-based nucleases with linker
lengths of either 21 (GZF3n-N) or 5 amino acids (GZF3-N
and nGZF3-N). The presence and position of the nuclear
localization signal are indicated by an bnQ in the name
(Fig. 2A). As confirmed by indirect immunofluorescence,
GZF3-N was distributed evenly throughout the cell,
including the nucleus, while nGZF3-N and GZF3n-N were
predominantly nuclear (data not shown). We determined
expression levels of the nucleases in 293T cells 24 and 48 h
after transfection (Fig. 2B). The CMV promoter induced
very high expression after 1 day, while expression from
the cellular phosphoglycerate kinase (PGK) promoter and
the herpes simplex virus thymidine kinase (HSV-tk)
promoter was weaker. After 2 days, however, the patterns
of nuclease expression were comparable. This suggests
that high expression levels of the custom nucleases are
not well tolerated by the cells, probably leading to their
elimination over time through accumulation of nonspecific DSBs. The variability in the expression levels of
FIG. 1. Analysis of DNA binding. (A) Sequence motifs.
The sequences of the crucial residues of the zinc-finger
modules and the corresponding DNA-binding sites are
shown. Module ‘‘A’’ was designed to recognize DNA
triplet ‘‘a’’ and so on. (B) Schematic. Custom transcription factors (TFs) consist of an HA tag (black box),
the VP16 transcriptional activation domain (AD), the
SV40 NLS (n), and three zinc-finger modules. The
reporter plasmids contain the luciferase gene, a
minimal promoter element, and upstream binding
sites for the custom TFs (xx). Each x represents a 9-bp
recognition motif as defined in (A). (C) Reporter assay.
293T cells were transfected with expression plasmids
for the custom TFs and reporter plasmids containing
binding sites for GZF1 and GZF2 (1-2/Luc) or GZF3
and GZF4 (3-4/Luc). The graph displays luciferase
activity normalized for transfection efficiency and
relative to transfection with empty vector (AD).
**Statistically significant (P b 0.01). (D) Western blot.
Transfected 293T cells were harvested after 40 h and
cell lysates probed with an anti-HA antibody. AD-ZF5C
is a control zinc-finger protein, while ‘‘cto’’ marks a
negative control.
2
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
ARTICLE IN PRESS
doi:10.1016/j.ymthe.2005.06.094
ARTICLE
FIG. 2. Design and expression kinetics. (A)
Design. All ZFNs contain an HA tag (black
box), three zinc-finger modules, and the
nuclease domain of FokI (Fok1-N). GZF3n-N
contains a 21-amino-acid linker, while GZF3N and nGZF3-N contain a short AAARA linker
between the zinc-finger modules and the
Fok1-N domain. ‘‘P’’ stands for promoter
CMV, PGK, or HSV-tk. (B) Western blot.
Transfected 293T cells were harvested after
24 and 48 h and cell lysates probed with an
anti-HA antibody. ‘‘cto’’ indicates transfection with empty vector. (C) Apoptosis. The
sub-G1 population of transfected 293T cells
was determined by flow cytometry of PIstained cells after 48 h and is shown as the
fraction of transfected cells. **Statistically
significant (P b 0.05).
the different GZF3 nucleases (compare GZF3-N with
nGZF3-N) may be attributed mainly to differences in
protein stability and/or cytotoxicity. To test whether
overexpression of custom nucleases triggers apoptosis,
we determined the number of apoptotic cells by measuring the DNA content of transfected 293T cells (Fig. 2C).
Expression of GZF3-N and nGZF3-N from a CMV promoter induced apoptosis in a significant number of cells,
while expression of I-SceI, an intron-encoded endonuclease present in the mitochondria of yeast, or expression
of the GZF3 nucleases from the weaker PGK and HSV-tk
promoters, did not induce detectable levels of apoptosis.
Site-Directed Recombination in Episomal Gene Repair
Assays
To determine quantitatively the stimulation of SDR by
custom nucleases, we developed a plasmid-based gene
repair assay that consists of three elements: a target
plasmid, a repair plasmid, and the endonuclease (Fig. 3A).
The target plasmid contains a CMV-driven LacZ gene
followed by stop codons and a 5V-truncated EGFP gene
(BGFP). The stop codons and the 33-bp deletion ensure
that no functional EGFP is expressed from the target
plasmid [23]. We placed the 18-bp recognition site for ISceI alone (target plasmid b0-0Q) or in combination with
binding sites for the ZFNs (e.g., two sites for GZF1
nuclease in target plasmid b1-1Q or two sites for GZF3
nuclease in target b3-3Q) between the two open reading
frames. This way, each custom nuclease could be tested
individually for its ability to stimulate SDR through
insertion of a DSB into the target plasmid. The repair
plasmid harbors a BLacZ–EGFP fusion gene and is
designed to rescue EGFP expression through generation
of a LacZ–EGFP fusion protein. To ensure that expression
of LacZ–EGFP strictly depends on SDR, the repair plasmid
does not contain a promoter and the LacZ gene is
truncated at the 5V-end (BLacZ). An expression cassette
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
for the red fluorescent protein DsRed-Express (REx) is
located farther downstream to mark transfected cells.
To assess the effect of nuclease design and expression
kinetics on SDR, we transfected 293T cells with target
plasmid b3-3Q, the repair plasmid, and GZF3 nucleases.
Fig. 3B shows the percentage of green cells normalized for
transfection efficiency. In the absence of a nuclease 4% of
cells turned green, while 36% of cells were green upon ISceI expression. Expression of GZF3-N and nGZF3-N from
the CMV promoter induced cell death very efficiently, as
indicated by the fact that the number of transfected cells
dropped below 1% after 48 h. Due to the low numbers,
we did not determine the frequency of gene repair.
Conversely, expression of the same nucleases from the
weaker PGK and HSV-tk promoters stimulated gene repair
significantly and comparable to I-SceI. These results are in
agreement with the apoptosis assay (Fig. 2C) but point
out that the apoptosis assay highly underestimated
cytotoxicity induced by high levels of ZFNs. Expression
of GZF3n-N from all promoters had no stimulatory effect
on SDR, suggesting that long linkers decrease the
efficiency of DNA cleavage [15].
We validated specificity of the ZFNs by transfecting
293T cells with HSV-tk-driven nuclease expression
vectors along with the repair plasmid and target plasmid
b0-0Q or a target plasmid containing the respective
binding sites (Fig. 3C). Comparison of the repair
frequencies revealed that GZF1-N and GZF3-N stimulated gene repair of targets b1-1Q and b3-3Q to an extent
similar to that of I-SceI but had little effect on control
plasmid b0-0Q. The overall stimulation of SDR by I-SceI,
GZF1-N, and GZF3-N was between 10- and 15-fold.
GZF2-N and GZF4-N did not significantly stimulate
gene repair. These data are in line with the results of
the reporter assay (Fig. 1C), demonstrating that high
affinity and specificity of target site recognition are
crucial for stimulating SDR.
3
ARTICLE IN PRESS
ARTICLE
doi:10.1016/j.ymthe.2005.06.094
FIG. 3. Episomal gene repair assay. (A) Experimental
setup. Repair and target plasmids are described in the
text. The stop codons (bold, italics) and the recognition site for I-SceI (boxed) are highlighted. (B, C)
Episomal gene repair. 293T cells transfected with
repair plasmid (RP), target plasmid, and a nuclease
expression vector were analyzed by flow cytometry
after 2 days. The graph displays the fraction of EGFPpositive cells in relation to transfected cells. The
presence or absence of RP is indicated by + or ,
‘‘cto’’ indicates control. In (B), cells were transfected
with target plasmid ‘‘3-3’’ along with plasmids
encoding different promoters and types of GZF3
nucleases. **Statistically significant (P b 0.01) stimulation of SDR; ‘‘nd’’ stands for ‘‘not determined’’ in
the case in which the number of REx-positive cells was
below 1% due to cell death. In (C), cells were
transfected with ‘‘0-0’’, ‘‘1-1’’, ‘‘2-2’’, ‘‘3-3’’, or ‘‘44’’, along with nuclease expression plasmids. **Statistically significant (P b 0.01) stimulation of HR
comparing gene repair in the absence (‘‘0-0’’) and in
the presence of respective binding sites (e.g., ‘‘1-1’’).
Episomal and Chromosomal Gene Repair by
Heterodimeric Custom Nucleases
We evaluated the efficiency of heterodimeric nucleases
to promote SDR initially in the episomal gene repair
assay (Fig. 4A). Combined expression of GZF1-N and
GZF3-N stimulated SDR to an extent similar to that of
4
I-SceI, with almost 40% of transfected cells showing
gene repair. Although some gene repair was measurable when GZF1-N or GZF3-N was expressed alone,
stimulation of SDR was specific for target plasmid b31Q, as only background levels were detected with target
b0-0Q.
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
doi:10.1016/j.ymthe.2005.06.094
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
ARTICLE IN PRESS
FIG. 4. SDR stimulated by heterodimeric nucleases. (A) Episomal gene repair. 293T cells were transfected with ‘‘3-1’’ and plasmids coding for I-SceI, GZF1-N, and GZF3-N as indicated by +. **Statistically
significant stimulation of gene repair (P b 0.001). (B) Chromosomal gene repair. 293/3-1 cells, which contain a single copy of target locus ‘‘3-1’’, were transfected with repair plasmid (RP) and
endonucleases as indicated by + (0.25 Ag) and z (0.75 Ag). The columns designate the fraction of EGFP-positive cells 7 days posttransfection normalized for transfection efficiency. The presence or
absence of RP is indicated by + or . **Statistically significant stimulation (P b 0.01) of gene repair above basal HR. (C) Cytotoxicity. 293/3-1 cells were transfected with the same amounts of
endonuclease expression vectors as for (B). The percentage of dead 293/3-1 cells was determined after 48 h by analyzing PI incorporation. The number of PI-positive cells is shown as a fraction of
transfected cells.
ARTICLE
5
ARTICLE IN PRESS
ARTICLE
To assess whether the custom nucleases were able to
promote gene repair in a chromosomal context, we
generated a 293-based cell line. As opposed to the
episomal gene repair assay, in which each cell probably
contains about 1000 copies of the target plasmid, the
293/3-1 cell line contains a single integrated copy of
target locus b3-1Q. Accordingly, we determined the
number of EGFP-positive 293/3-1 cells 7 days after
transfection as expression of LacZ–EGFP originated
from a single bcorrectedQ gene (Fig. 4B). In the absence
of a nuclease, about 0.1% of cells underwent HR between the target locus and the repair plasmid. Expression of I-SceI stimulated SDR in a dose-dependent
manner by a factor of 116 and 189, respectively, which
is in good agreement with previously published results
[7–9,24,25]. About 1% of transfected cells turned green
upon expression of PGK-driven GZF1-N and GZF3-N.
Transfections with higher amounts did not increase
SDR and expression of a single custom nuclease had no
stimulatory effect. Cytotoxicity induced by nuclease
expression during chromosomal gene repair was estimated by analyzing the incorporation of propidium
iodide (PI) of transfected 293/3-1 cells (Fig. 4C). Compared to control or I-SceI, coexpression of GZF1N/GZF3N from the PGK promoter did not affect cell survival,
which is in good agreement with the apoptosis assay
(Fig. 2C).
Together, these data establish that custom nucleases
can stimulate gene repair in episomal and chromosomal
contexts. Moreover, the custom nucleases promoted
replacement of up to 54 bp, indicating that this
technology can be used to correct more than just point
mutations.
DISCUSSION
Targeted manipulation of the genome through HR can be
used to induce precise changes at specific locations in the
genetic program of a living cell. Changing the genotype of
a cell or organism will allow studies of gene function as
well as treatments for patients suffering from inherited
disorders. This report demonstrates that custom nucleases
are a valuable tool for stimulating targeted manipulations
of the genome. It also points out the ease and the
difficulties of creating custom endonucleases. The two
assays we present here allow a rapid evaluation of the
DNA-binding specificity and a quick optimization of
design and expression kinetics of the custom nucleases.
Thus, using our simple two-step procedure, custom
nucleases can be prescreened and optimized prior to their
utilization in target cells.
Design of Custom Nucleases
Custom nucleases to stimulate SDR at a defined location
can be chosen as follows. (i) Select a suitable target site.
Thus far, zinc-finger modules recognizing GNN and ANN
6
doi:10.1016/j.ymthe.2005.06.094
triplets have been identified by phage display [19,20,26].
Taking the 6-bp spacer between the two half-sites into
consideration [15], a full target site follows the sequence
5V-NNYVNNYVNNYVNNNVNNNVRNNVRNNVRNN (R =
purine, Y = pyrimidine), which is statistically found every
64 bp in the human genome. The target site should be
located as close as possible to the sequence to be replaced
because the frequency of SDR drops steeply with increasing distance from the DSB [27]. To prevent continuous
cleavage of the breplaced locus,Q at least one of the two
ZFN binding sites should not remain after SDR. (ii)
Assemble the custom DNA-binding domains. Our results
show that not all DNA-binding domains assembled in
this manner mediate specific binding to the target site.
Since the neighbors influence the binding activity of an
individual zinc-finger module, the use of alternative
modules might improve specificity and affinity [21,26].
Alternatively, the whole DNA-binding domain can be
optimized through in vivo selection, as recently reported
using a bacterial two-hybrid system [28]. (iii) Generate the
reporter constructs and the custom transcription factors.
The reporter assay allows one to preselect DNA-binding
domains of high affinity and specificity. (iv) Generate the
custom nucleases and the target plasmids and evaluate
the custom nucleases in episomal gene repair assays
individually or in combination as heterodimers. The
choice of promoter is critical because high nuclease levels
provoke the death of the target cells. The best combinations can then be employed in the final target cells.
As shown here, preselected custom nucleases are
expected to perform well in a chromosomal context.
Further Improvements and Outlook
The frequency of chromosomal SDR achieved here is in
good agreement with results obtained in Drosophila [14]
and Arabidopsis [29], confirming that custom nucleases
can perform well in complex genomes. Nonetheless,
while stimulation of SDR by the custom nucleases in
the episomal assay equaled the gene repair frequency
measured with I-SceI, it was 10-fold lower for chromosomal gene repair. We assume that the lower frequency was
at least partially due to cytotoxic side effects upon longterm expression of the nucleases, as previously suggested
[17]. It will therefore be crucial to achieve a bburst-likeQ
expression pattern, with high initial concentrations of
the ZFNs that quickly drop after creation of the DSB. This
could be achieved by mRNA transfer [30] or by use of
inducible promoters [14].
The full target sequence consists of 2 9 bp. Every
sequence encompassing a total of 18 bp defines a unique
location in the human genome. However, the DNAbinding domains of the ZFNs recognize only the 9-bp
half-site of the full target sequence. Statistically, a 9-bp
sequence is found more than 10,000 times in the human
genome. To compensate for this dilution effect more
protein has to be expressed in the cell, which increases
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
ARTICLE IN PRESS
ARTICLE
doi:10.1016/j.ymthe.2005.06.094
the probability of nuclease dimers being formed in the
absence of DNA binding and hence the risk of unregulated cleavage of DNA. It would be interesting to see in a
direct comparison whether four-finger proteins, which
recognize a 12-bp half-site [18], will ameliorate the
problem by reducing the number of potential binding
sites and increasing affinity and specificity.
The results shown here illustrate that custom
nucleases facilitate the replacement of sequence stretches
up to 54 bp. It remains to be determined whether even
larger stretches of DNA can be substituted. This would
offer treatment options for genetic diseases caused not
only by point mutations but also by deletions and
insertions.
Using ZFNs in combination with a repair template
allows the correction of mutated genes in situ, thus
presenting a novel and potentially safer form of gene
therapy. Importantly, this technology can be easily
combined with viral gene transfer systems. Vectors based
on adeno-associated virus have proven to be particularly
well suited to serve as templates for gene repair
[24,25,31,32]. Gene repair of human stem cells will be
particularly intriguing [33], as corrected cells could be
expanded ex vivo and then reintroduced into patients.
Our assay will provide a valuable tool for further
optimizing the nucleases with regard to cytotoxicity,
specificity, and design, so they can develop into routinely
used genetic scissors.
MATERIAL
AND
METHODS
Plasmids and cell lines. DNA-binding proteins containing three zincfinger domains were assembled by overlap PCR and cloned into
appropriate expression vectors using methods and domains described
previously by Barbas and co-workers [19–21,34,35]. Briefly, overlapping
PCR primers were designed to encode zinc-finger domains that had been
obtained by phage display to bind unique 3-bp sites. Typically, zincfinger domains were chosen from those listed in Ref. [21], although
some variants were chosen from Ref. [34]. All plasmids were assembled
by PCR and oligonucleotide-directed cloning. Relevant sequences are
given in Figs. 1 and 3. Complete sequence files can be obtained upon
request. Briefly, the repair plasmids contain a 5V-deleted LacZ gene fused
in-frame with a downstream EGFP open reading frame (Fig. 3A, top). In
the case of pUC.Zgfp/REx, an expression cassette for DsRed-Express
(Clontech) to identify transfected cells is located farther downstream.
Expression of a LacZ–EGFP fusion protein from the repair plasmids is
prevented by an 1197-bp deletion at the 5Vend of LacZ and by omitting
a promoter. LacZ–EGFP expression from target plasmids is prevented by
a stop codon that terminates translation of LacZ and a 33-bp truncation
at the 5Vend of the EGFP open reading frame [23] (Fig. 3A, bottom). A
binding site for I-SceI follows the stop codon in all target loci, while b11Q, b2-2Q, b3-3Q, b4-4Q, and b3-1Q contain additional bindings sites for the
respective ZFNs, which have been introduced into the PacI site of b0-0Q
(Fig. 3A, bottom). All endonuclease expression plasmids are based on
plasmid pRK5 [36], in which the CMV promoter was substituted for the
PGK or the HSV-tk promoter where indicated. The custom transcription
factors were generated by cloning the zinc-finger domains into vector
pRK5.AD [22].
293 and 293T cells were grown in DMEM supplemented with 10%
fetal calf serum. Cell line 293/3-1 was generated by transfecting 293 cells
with target plasmid b3-1Q followed by selection in Geneticin (Gibco-
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy
Invitrogen) at 400 Ag/ml. After clonal expansion, the presence of a singlecopy insertion was confirmed by Southern blot analysis (data not shown).
Reporter assays and immunoblotting. 293T cells in 35-mm wells were
transfected in duplicate by calcium phosphate precipitation as previously
described [22]. Transfection cocktails included 0.5 Ag of plasmids encoding
the zinc-finger domains fused to the transcriptional activation domain of
VP16, 1 Ag of the reporter plasmid, 0.1 Ag of pEGFP-N1 (Clontech), and 0.5
Ag of plasmid pCMVh (Clontech) expressing h-galactosidase to normalize for transfection efficiency. The amount of DNA was kept constant by
adding pUC118 (ATCC 37462) to 4 Ag. Cells were harvested 40 h after
transfection in PBS. Luciferase and h-galactosidase activities were
measured in a luminometer (Berthold) using BrightGlo and BetaGlo
substrates (Promega). Statistical significance was determined using the
Student t test.
Immunoblotting was basically performed as previously described [36].
Briefly, 293T cells in 35-mm wells were transfected by calcium phosphate
precipitation. Typically 2 Ag of the specific expression plasmid was used
and pUC118 was added to 5 Ag. In the case of HSV-tk-containing plasmids
5 Ag was used. Cells were harvested 24 to 48 h after transfection. After
resuspension in lysis buffer (20 mM Tris–HCl, pH 8.0, 100 mM NaCl, 0.2%
NP-40, 0.2% Triton X-100, 0.2% deoxycholate), equal amounts of
proteins (50 Ag) were separated by SDS–polyacrylamide gel electrophoresis
and transferred to Immobilon-P membrane (Millipore). Specific proteins
were detected with a rabbit anti-HA tag antibody (Novus Biologicals,
NB600-363) and visualized with Western Lightning chemiluminescence
reagent (Perkin–Elmer).
Apoptosis and cell survival assay. To assess apoptosis, 293T cells in 12well plates were transfected with 1.5 Ag of pEGFP-tubulin (obtained from
Tony Hunter) and 0.5 Ag of nuclease expression vectors by calcium
phosphate precipitation. Cells were harvested after 48 h in PBS, fixed for 1
h at 48C in 75% EtOH, washed with PBS, incubated in DNA extraction
buffer (200 mM Na2HPO4, 100 mM citric acid, pH 7.8) for 5 min at room
temperature, pelleted, resuspended in PI staining solution (3.8 mM Na
citrate, 25 Ag/ml propidium iodide, 0.2 mg/ml RNase A, PBS) for 30 min at
room temperature, and then analyzed by flow cytometry (FACSCalibur;
BD Bioscience) to determine the sub-G1 population in EGFP-positive cells.
Statistical significance was determined using the Student t test.
To assess cell survival, 293/3-1 cells in six-well plates were transfected
with 0.5 Ag of pEGFP-C1 (Clontech) and 0.5 or 1.5 Ag of nuclease
expression vector by calcium phosphate precipitation. The amount of
DNA was kept constant by adding pUC118 to 4.5 Ag. Cells were harvested
after 48 h in PBS/10% FCS and stained with PI, which is excluded by
viable cells but penetrates cell membranes of dying or dead cells. After 30
min incubation at room temperature (final concentration 20 Ag PI/ml),
105 cells were analyzed by flow cytometry (FACSCalibur; BD Bioscience)
to determine the percentage of EGFP- and PI-positive cells. Statistical
significance was analyzed using the Student t test.
Gene repair assay. For episomal gene repair, 293T cells in 6-well plates
were transfected by calcium phosphate precipitation with 50 ng of target
plasmid, 2 Ag of repair plasmid pUC.Zgfp/REx or repair control pUC.REx,
and 1 Ag of a nuclease expression vector, I-SceI (pRK5.LHA-Sce1) or
control vector pCMV.Luc. The amount of DNA was kept constant by
adding pUC118 to 4 Ag. After 2 days 105 cells were analyzed by flow
cytometry to determine the percentage of EGFP- and REx-positive cells.
For chromosomal gene repair, 293/3-1 cells in 12-well plates were
transfected by calcium phosphate precipitation with 1.5 Ag of repair
plasmid pUC.Zgfp or repair control pUC118, 0.75 Ag of a nuclease
expression vector or control vector pCMV.Luc, and 5 ng of pDsRedExpress-N1 (Clontech). After 3 and 7 days 105 cells were analyzed by
flow cytometry. The number of EGFP-positive cells at day 7 was
normalized for transfection efficiency measured at day 3. Statistical
significance of all assays was determined using the Student t test.
ACKNOWLEDGMENTS
We thank Katharina Gellhaus, Stefan Weger, and Regine Heilbronn for critical
discussions and Tony Hunter for plasmid pEGFP-tubulin. This work was
7
ARTICLE IN PRESS
ARTICLE
supported by Grant CA311/1 from the German Research Foundation (T.C.),
NIH Grant RO1GM065059 (C.F.B.) and R21CA103651 (D.J.S.), and an
equipment grant from the Sonnenfeld-Stiftung (T.C.).
RECEIVED FOR PUBLICATION APRIL 15, 2005; REVISED JUNE 3, 2005;
ACCEPTED JUNE 6, 2005.
REFERENCES
1. Hacein-Bey-Abina, S., et al. (2002). Sustained correction of X-linked severe combined
immunodeficiency by ex vivo gene therapy. N. Engl. J. Med. 346: 1185 – 1193.
2. Hacein-Bey-Abina, S., et al. (2003). A serious adverse event after successful gene
therapy for X-linked severe combined immunodeficiency. N. Engl. J. Med. 348:
255 – 256.
3. Bushman, F. D. (2003). Targeting survival: integration site selection by retroviruses and
LTR–retrotransposons. Cell 115: 135 – 138.
4. Baum, C., et al. (2004). Chance or necessity? Insertional mutagenesis in gene therapy
and its consequences. Mol. Ther. 9: 5 – 13.
5. Vasquez, K. M., Marburger, K., Intody, Z., and Wilson, J. H. (2001). Manipulating the
mammalian genome by homologous recombination. Proc. Natl. Acad. Sci. USA 98:
8403 – 8410.
6. Rouet, P., Smih, F., and Jasin, M. (1994). Introduction of double-strand breaks into the
genome of mouse cells by expression of a rare-cutting endonuclease. Mol. Cell. Biol. 14:
8096 – 8106.
7. Choulika, A., Perrin, A., Dujon, B., and Nicolas, J. F. (1995). Induction of homologous
recombination in mammalian chromosomes by using the I-SceI system of Saccharomyces cerevisiae. Mol. Cell. Biol. 15: 1968 – 1973.
8. Smih, F., Rouet, P., Romanienko, P. J., and Jasin, M. (1995). Double-strand breaks at
the target locus stimulate gene targeting in embryonic stem cells. Nucleic Acids Res. 23:
5012 – 5019.
9. Sargent, R. G., Brenneman, M. A., and Wilson, J. H. (1997). Repair of site-specific
double-strand breaks in a mammalian chromosome by homologous and illegitimate
recombination. Mol. Cell. Biol. 17: 267 – 277.
10. Chevalier, B. S., et al. (2002). Design, activity, and structure of a highly specific artificial
endonuclease. Mol. Cell 10: 895 – 905.
11. Epinat, J. C., et al. (2003). A novel engineered meganuclease induces homologous
recombination in yeast and mammalian cells. Nucleic Acids Res. 31: 2952 – 2962.
12. Kim, Y. G., Cha, J., and Chandrasegaran, S. (1996). Hybrid restriction enzymes:
zinc finger fusions to Fok I cleavage domain. Proc. Natl. Acad. Sci. USA 93:
1156 – 1160.
13. Porteus, M. H., and Baltimore, D. (2003). Chimeric nucleases stimulate gene targeting
in human cells. Science 300: 763.
14. Bibikova, M., Beumer, K., Trautman, J. K., and Carroll, D. (2003). Enhancing gene
targeting with designed zinc finger nucleases. Science 300: 764.
15. Bibikova, M., et al. (2001). Stimulation of homologous recombination through
targeted cleavage by chimeric nucleases. Mol. Cell. Biol. 21: 289 – 297.
16. Smith, J., et al. (2000). Requirements for double-strand cleavage by chimeric
restriction enzymes with zinc finger DNA-recognition domains. Nucleic Acids Res. 28:
3361 – 3369.
17. Wilson, J. H. (2003). Pointing fingers at the limiting step in gene targeting. Nat.
Biotechnol. 21: 759 – 760.
8
doi:10.1016/j.ymthe.2005.06.094
18. Urnov, F. D., et al. (2005). Highly efficient endogenous human gene correction using
designed zinc-finger nucleases. Nature 435: 646 – 651.
19. Segal, D. J., Dreier, B., Beerli, R. R., and Barbas, C. F., III (1999). Toward controlling
gene expression at will: selection and design of zinc finger domains recognizing each of
the 5V-GNN-3VDNA target sequences. Proc. Natl. Acad. Sci. USA 96: 2758 – 2763.
20. Dreier, B., Beerli, R. R., Segal, D. J., Flippin, J. D., and Barbas, C. F., III (2001).
Development of zinc finger domains for recognition of the 5V-ANN-3Vfamily of DNA
sequences and their use in the construction of artificial transcription factors. J. Biol.
Chem. 276: 29466 – 29478.
21. Segal, D. J. (2002). The use of zinc finger peptides to study the role of specific factor
binding sites in the chromatin environment. Methods 26: 76 – 83.
22. Cathomen, T., Stracker, T. H., Gilbert, L. B., and Weitzman, M. D. (2001). A genetic
screen identifies a cellular regulator of adeno-associated virus. Proc. Natl. Acad. Sci. USA
98: 14991 – 14996.
23. Li, X., et al. (1997). Deletions of the Aequorea victoria green fluorescent protein
define the minimal domain required for fluorescence. J. Biol. Chem. 272:
28545 – 28549.
24. Porteus, M. H., Cathomen, T., Weitzman, M. D., and Baltimore, D. (2003). Efficient
gene targeting mediated by adeno-associated virus and DNA double-strand breaks.
Mol. Cell. Biol. 23: 3558 – 3565.
25. Miller, D. G., Petek, L. M., and Russell, D. W. (2003). Human gene targeting by adenoassociated virus vectors is enhanced by DNA double-strand breaks. Mol. Cell. Biol. 23:
3550 – 3557.
26. Liu, Q., Xia, Z., Zhong, X., and Case, C. C. (2002). Validated zinc finger protein designs
for all 16 GNN DNA triplet targets. J. Biol. Chem. 277: 3850 – 3856.
27. Stark, J. M., Pierce, A. J., Oh, J., Pastink, A., and Jasin, M. (2004). Genetic steps of
mammalian homologous repair with distinct mutagenic consequences. Mol. Cell. Biol.
24: 9305 – 9316.
28. Hurt, J. A., Thibodeau, S. A., Hirsh, A. S., Pabo, C. O., and Joung, J. K. (2003). Highly
specific zinc finger proteins obtained by directed domain shuffling and cell-based
selection. Proc. Natl. Acad. Sci. USA 100: 12271 – 12276.
29. Lloyd, A., Plaisier, C. L., Carroll, D., and Drews, G. N. (2005). Targeted
mutagenesis using zinc-finger nucleases in Arabidopsis. Proc. Natl. Acad. Sci. USA
102: 2232 – 2237.
30. Galla, M., Will, E., Kraunus, J., Chen, L., and Baum, C. (2004). Retroviral pseudotransduction for targeted cell manipulation. Mol. Cell 16: 309 – 315.
31. Cathomen, T. (2004). AAV vectors for gene correction. Curr. Opin. Mol. Ther. 6:
360 – 366.
32. Liu, X., et al. (2004). Targeted correction of single-base-pair mutations with adenoassociated virus vectors under nonselective conditions. J. Virol. 78: 4165 – 4175.
33. Chamberlain, J. R., et al. (2004). Gene targeting in stem cells from individuals with
osteogenesis imperfecta. Science 303: 1198 – 1201.
34. Dreier, B., Segal, D. J., and Barbas, C. F., III (2000). Insights into the molecular
recognition of the 5V-GNN-3 family of DNA sequences by zinc finger domains. J. Mol.
Biol. 303: 489 – 502.
35. Beerli, R. R., Segal, D. J., Dreier, B., and Barbas, C. F., III (1998). Toward controlling
gene expression at will: specific regulation of the erbB-2/HER-2 promoter by using
polydactyl zinc finger proteins constructed from modular building blocks. Proc. Natl.
Acad. Sci. USA 95: 14628 – 14633.
36. Cathomen, T., Collete, D., and Weitzman, M. D. (2000). A chimeric protein containing
the N terminus of the adeno-associated virus Rep protein recognizes its target site in an
in vivo assay. J. Virol. 74: 2372 – 2382.
MOLECULAR THERAPY Vol. xx, No. xx, Month Year
Copyright C The American Society of Gene Therapy