Targeted gene knockout in mammalian cells by using
engineered zinc-finger nucleases
Yolanda Santiago*, Edmond Chan*, Pei-Qi Liu*, Salvatore Orlando*, Lin Zhang†, Fyodor D. Urnov*, Michael C. Holmes*,
Dmitry Guschin*, Adam Waite*, Jeffrey C. Miller*, Edward J. Rebar*, Philip D. Gregory*‡, Aaron Klug‡§,
and Trevor N. Collingwood*
*Sangamo BioSciences, Inc., 501 Canal Boulevard, Suite A100, Richmond, CA 94804;†Pfizer, Inc., Bioprocess Research and Development, Cell Line
Development, 700 Chesterfield Parkway West, Chesterfield, MO 63017; and§Medical Research Council Laboratory of Molecular Biology, Hills Road,
Cambridge CB2 2QH, United Kingdom
Contributed by Aaron Klug, January 30, 2008 (sent for review November 14, 2007)
Gene knockout is the most powerful tool for determining gene
function or permanently modifying the phenotypic characteristics
of a cell. Existing methods for gene disruption are limited by their
efficiency, time to completion, and/or the potential for confound-
ing off-target effects. Here, we demonstrate a rapid single-step
approach to targeted gene knockout in mammalian cells, using
engineered zinc-finger nucleases (ZFNs). ZFNs can be designed to
target a chosen locus with high specificity. Upon transient expres-
sion of these nucleases the target gene is first cleaved by the ZFNs
and then repaired by a natural—but imperfect—DNA repair pro-
cess, nonhomologous end joining. This often results in the gener-
ation of mutant (null) alleles. As proof of concept for this approach
we designed ZFNs to target the dihydrofolate reductase (DHFR)
gene in a Chinese hamster ovary (CHO) cell line. We observed
biallelic gene disruption at frequencies >1%, thus obviating the
need for selection markers. Three new genetically distinct DHFR?/?
cell lines were generated. Each new line exhibited growth and
functional properties consistent with the specific knockout of the
DHFR gene. Importantly, target gene disruption is complete within
2–3 days of transient ZFN delivery, thus enabling the isolation of
the resultant DHFR?/?cell lines within 1 month. These data
demonstrate further the utility of ZFNs for rapid mammalian cell
line engineering and establish a new method for gene knockout
with application to reverse genetics, functional genomics, drug
discovery, and therapeutic recombinant protein production.
genetic engineering ? zinc-finger proteins
limited by an absence of methods for rapid targeting and
disruption of an investigator-specified gene. Early approaches to
somatic cell gene disruption used genome-wide nontargeted
methods, including ionizing radiation and chemical-induced
mutagenesis (1, 2) whereas more recent methods used targeted
homologous recombination (HR) (3). However, the ?1,000-fold
lower frequency of the targeted HR event relative to random
integration in most mammalian cell lines (beyond mouse ES
cells) can necessitate screening thousands of clones and take
several months to identify a biallelic targeted gene knockout.
Strategies including positive and negative marker selection and
promoter-trap can boost efficiencies considerably, although
these approaches present their own technical challenges and are
not always successful in achieving high efficiency targeting (4, 5).
Although advances with adeno-associated viral delivery strate-
gies continue to improve the efficiency of knockouts (6, 7), the
frequency is still very low and the time required to achieve
biallelic gene knockout remains a barrier to its routine adoption.
Here, we present a general solution for rapid gene knockout in
cells occurs via the distinct mechanisms of homology directed
repair (HDR) or nonhomologous end joining (NHEJ) (8).
he use of gene knockouts in basic research, functional
genomics, and industrial cell line engineering is severely
Although HDR typically uses the sister chromatid of the dam-
aged DNA as a template from which to perform perfect repair
of the genetic lesion, NHEJ is an imperfect repair process that
often results in changes to the DNA sequence at the site of the
DSB. During NHEJ, the cleaved DNA is further resected by
exonuclease activity, and more bases may be added in an
irregular fashion before the two ends of the severed DNA are
rejoined (9). In mammalian systems, such as Chinese hamster
ovary (CHO) cells, the ratio of HDR to NHEJ-based repair has
been found to be 9:13 (10). Studies in Drosophila (11) and later
in both plants and worms (12, 13) showed that DSBs generated
by site-specific zinc-finger nucleases (ZFNs) resulted in targeted
mutagenesis consistent with repair by NHEJ. In this article, we
now extend the ZFN approach to mammalian systems. We make
use of the process of NHEJ to carry out targeted gene knockout
in CHO cells by using transiently expressed site-specific ZFNs to
generate the DSB in the gene that is being targeted (for reviews,
see refs. 14 and 15).
ZFNs employ a heterologous zinc-finger protein (ZFP) DNA
binding domain (which specifically binds to the designated target
sequence) fused to the catalytic domain of the endonuclease
FokI (16). Dimerization of this FokI domain is required for its
DNA binding-dependent endonuclease activity. Thus, two indi-
vidual ZFNs are designed as a pair to bind to the target DNA
stretch with precise sequence specificity, spacing, and orienta-
16). When expressed transiently in cells, the ZFNs generate a
site-specific DSB in the endogenous target gene that subse-
quently can be repaired via NHEJ. The precise nature of the
mutations generated by NHEJ-based repair of the ZFN-induced
DSB cannot be predetermined and, indeed, need not be known.
In this article, we show that the frequency of gene disrupting
mutations generated by this stochastic process is more than
sufficient for utility as a method for gene knockout.
To demonstrate the ZFN approach to gene knockout, we
elected to disrupt the function of the dihydrofolate reductase
(DHFR) gene in a Chinese hamster ovary cell line (CHO-S) that
is diploid for functional DHFR. CHO cells are the dominant
research; Y.S., E.C., P.-Q.L., S.O., A.W., and J.C.M. performed research; P.-Q.L., L.Z., E.J.R.,
P.D.G., and T.N.C. analyzed data; and P.D.G., A.K., and T.N.C. wrote the paper.
Conflict of interest statement: Y.S., E.C., P.-Q.L., S.O., F.D.U., M.C.H, D.G., J.C.M., E.J.R.,
the scientific advisory board for Sangamo BioSciences, Inc. L.Z. is a full-time employee of
Freely available online through the PNAS open access option.
See Commentary on page 5653.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
April 15, 2008 ?
vol. 105 ?
no. 15 ?
system for the production of therapeutic recombinant proteins
(17). DHFR is one of the most widely used and best character-
ized selectable marker genes and is used in conjunction with
CHO cell lines in which endogenous DHFR expression has been
destroyed. This widespread familiarity with DHFR makes it an
ideal choice of target gene with which to validate a ZFN-
mediated approach to gene knockout. Furthermore, existing
DHFR-negative CHO cell lines provide additional benchmarks
against which we can verify the phenotype of our ZFN-generated
DHFR knockout cell lines. Here, we report (i) the successful
generation of ZFNs targeting the DHFR locus in CHO-S cells
devoid of functional DHFR.
Design of Zinc Finger Nucleases Targeting the DHFR Gene. Exon 1 of
the CHO-S DHFR locus was chosen for targeting with pairs
of ZFNs, using the same strategy as in ref. 16. Structural studies
of human DHFR have shown this region to be critical for
substrate and cofactor binding (18). ZFNs that form pairs, which
target the DHFR locus, were designed and screened in vitro for
DNA binding to their target sites (Fig. 1 A and B and data not
shown), using an ELISA-based assay (19). The nuclease function
FokI, which is linked to the DNA-binding zinc-finger proteins.
Two different architectures of the FokI catalytic domain were
tested, either the WT FokI domain or the high-fidelity FokI-KK
ref. 20. These FokI dimerization variants provide an additional
level of specificity to ZFN cleavage by effectively eliminating
unwanted homodimerization events elsewhere in the genome
(20, 21). The ZFN9461/ZFN9684 pair used the ELKK FokI
variants in all cases. The ZFN7843/ZFN7844 pair used the WT
FokI domain in all studies except that shown in Fig. 1C. For this
experiment, all ZFNs used the ELKK variants so that the only
experimental variable was the structure of the DNA binding
domains themselves (Fig. 1B). Plasmids expressing each pair of
ZFNs were transfected into cells at the reported amount. The
frequency of ZFN-mediated disruption at the target site in each
pool of cells was determined by using the CEL-I nuclease assay
(see Materials and Methods for discussion of this assay). We
found that expression of the ZFN9461/ZFN9684 pair gave
enhanced targeted gene cleavage compared with that shown by
the ZFN7843/ZFN7844 pair, as demonstrated by the increased
incidence of allelic mutation (NHEJ frequency) in the ZFN-
treated pools, particularly at lower levels of input ZFN (Fig. 1C).
Although both ZFN pairs attained a similar maximum level of
in vivo gene cleavage (?15%), the ZFN9461/ZFN9684 pair
achieved this by using ?10-fold less ZFN plasmid. Taken
together, these data demonstrate that the ZFN7843/ZFN7844
and ZFN9461/ZFN9684 pairs successfully cleave, and thereby
initiate mutation of, the target locus within exon 1 of the CHO-S
Generation of DHFR-Deficient Cell Lines. DHFR?/?cell lines were
generated by transfecting CHO-S cells with either of the ZFN
pairs described above and then plating at limiting dilution to
obtain single-cell derived DHFR-deficient cell lines. This ap-
proach was tested in two separate studies. In the first study,
CHO-S cells growing adherently in serum-containing medium
were treated with the ZFN7843/ZFN7844 pair containing the
WT FokI domain. After dilution cloning (one cell per well
average), 68 isolates were analyzed for DHFR gene disruption,
using the CEL-I assay. Five of 68 isolates (7%) showed the
presence of mutated DHFR alleles and were further subcloned
to ensure clonal purity. The exact sequence of the mutant alleles
in each cell line, and thus the genotype, was determined by
PCR-amplifying the target locus and cloning the PCR product,
then sequencing ?90 of the resulting bacterial colonies—each of
which contained the sequence of either allele. Three of the
clones appeared to be heterozygous for DHFR disruption (data
not shown). The remaining two clones, 14/1 and 14/2, showed no
WT sequence, suggesting biallelic disruption. For clone 14/1, 47
of the 89 sequence reads (53%) showed a single base pair
insertion between the ZFN binding sites, whereas the remaining
42 reads (47%) indicated a 2-bp insertion in the same region
(Fig. 2A). The ?1:1 ratio of alleles indicated that clone 14/1 is
a compound heterozygous mutant in which both alleles have
mutations between the individual ZFN binding sites that result
in shifts in reading frame. These give rise to premature termi-
nation products of 46 and 38 codons, respectively, and the mu-
tant mRNA transcripts are also expected to undergo nonsense-
mediated decay. Clone 14/2 is also a compound heterozygous
mutant from which 43 sequence reads contained the same 2-bp
insertion present in clone 14/1, but the remaining 40 contained
a 15-bp deletion (Fig. 2A). Although the deletion preserves
the reading frame for the DHFR protein, the five amino
acid residues (ProTrpProMetLeu) deleted from this peptide
are critical to the integrity of the ligand-binding pocket of
In the second study, CHO-S cells cultured in serum-free
suspension medium were transfected with the other ZFN pair,
ZFN is boxed. ZFN7843 and ZFN9461 bind the same 12-bp site. ZFN7844 binds
the 12-bp site AATGCTCAGGTA, whereas ZFN9684 binds the 15-bp site AAT-
GCTCAGGTACTG. (B) Recognition helix sequences of ZFNs. The sequence of
the recognition helix from position ?1 to ?6 (27) is listed below its target
triplet. Backbone sequences for the ZFPs can be found elsewhere (20). The
of ZFN activity. Plasmids encoding each pair of ZFNs (ZFN7843/ZFN7844 and
ZFN9461/ZFN9684) containing the ELKK FokI variants were delivered in the
amounts shown to CHO-S cells in suspension culture. The frequency of allelic
(gel). Bands migrating at 384, 204, and 180 bp represent the parent amplicon
and the two CEL-I digestion products, respectively. The bands were quanti-
tated by EtBr staining and densitometry to determine the frequency of NHEJ.
The frequency of NHEJ is plotted against ZFN dosage. The lowest band on the
25-bp size ladder is 125 bp.
www.pnas.org?cgi?doi?10.1073?pnas.0800940105Santiago et al.
ZFN9461/ZFN9684, but, in this case, the ELKK FokI variant
domains were used. After dilution cloning, 350 isolates were
analyzed via direct sequencing of PCR products of the genomic
target locus. Eleven of these clones (3%) were found to have at
least one disrupted copy of the DHFR gene (data not shown).
Upon further cloning and sequencing of these PCR products (as
in the first study), four were found to be compound heterozygous
for mutations in both copies of the DHFR gene, with the
remaining seven clones being heterozygous. All mutations were
centered on the ZFN cleavage site, and all were consistent with
functional DHFR disruption—either by deletion of critical
residues or by introduction of a frameshift. Most mutations we
observed were small deletions (?20 bp). Clone 1.43 (and the
subclone 1.43A5; data not shown) exhibited the largest genetic
loss of all clones analyzed, with a 302-bp deletion in one allele
and a 38-bp deletion in the other allele (Fig. 2A). Both mutations
extended into the following intron, eliminating a splice site, and
thus would be expected to result in the loss of functional DHFR
For each of the three DHFR?/?cell lines produced in the
above two studies, no DHFR protein could be detected by
Western blot analysis using an antibody that recognizes the
carboxyterminal region of DHFR (Fig. 2B). Although the mu-
tant allele that encodes a 5-aa deletion in clone 14/2 might have
been expected to generate a near-full length, albeit nonfunc-
tional, peptide, the Western blot analysis indicates that this
peptide is unstable, perhaps eliminated via the unfolded protein
response (22). To further confirm the loss of functional DHFR,
we used a fluorescence-based assay to detect the binding of
fluoresceine-labeled methotrexate (FMTX), which would indi-
cate the presence of functional endogenous DHFR (23). We
observed that all three DHFR?/?lines did not bind FMTX, in
contrast to the WT parental CHO-S cells (Fig. 2C). This result
further confirms the loss of functional DHFR protein expression
in all three genetically distinct DHFR?/?cell lines generated
from this study.
Functional Analysis of DHFR Knockout Cell Lines. DHFR catalyzes
the reduction of folate during the biosynthesis of purines,
thymidine, and glycine. DHFR negative cells are unable to grow
unless the culture medium is supplemented with essential me-
tabolites, including hypoxanthine and thymidine (HT) (1). To
confirm the loss of functional DHFR expression, each DHFR?/?
cell line was cultured for 4 days in medium either with or without
HT supplement. As predicted, DHFR?/?cell lines 14/1 and 14/2
(Fig. 3A Upper) and 1.43A5 (Fig. 3A Lower) exhibited a strict
requirement for HT. DHFR is frequently used as a selection
marker for the stable expression of recombinant proteins in
DHFR-deficient mammalian cell lines. To further validate the
new cell lines described above, we investigated their capacity to
support DHFR-based selection of transgene expression. Each
mutant cell line and WT CHO-S cells were transfected with
plasmids driving coexpression of a monoclonal antibody and a
DHFR gene as the selection marker. All cells produced similar
levels of antibody two days after transient transfection of the
expression constructs (data not shown). However, after 2 weeks
of culture in the absence of HT supplement, we observed that
stable pools derived from each DHFR?/?clone exhibited sig-
nificantly greater antibody expression levels than did those
need for these DHFR?/?cells to retain coselected exogenous
DHFR expression for survival in the absence of HT, whereas the
WT cells were under no such selection pressure. Conversely, the
DHFR?/?cell lines did not exhibit increased antibody expression
when selection pressure was not applied (i.e., when HT was
present). Furthermore, subsequent incubation of the ‘‘minus-
HT-selected’’ antibody-expressing pool from cell line 14/1 in the
presence of 50 nM or 250 nM the DHFR inhibitor methotrexate
resulted in an increase in antibody expression of 1.7-fold and
2.6-fold, respectively (Fig. 3C). This reflects the selection of cells
that exhibit higher levels of DHFR expression and therefore
elevated levels of coselected antibody. Increased antibody ex-
pression in the pools is likely due to amplification of gene copy
number (24) and/or selection of clones in which the transgenes
have randomly integrated into chromosomal loci that are more
permissive for gene expression. Taken together, the results
obtained in these DHFR selection and amplification studies are
consistent with the characterized behavior of other DHFR-
negative CHO cells (24) and thus further support the phenotype
of the cell lines described here.
It may be relevant to add that, except for the loss of endog-
enous DHFR expression, the cells we have isolated behave in all
other respects like WT CHO-S. For example, clone 1.43A5
exhibits a doubling time of 16.8 h, which is comparable with that
of WT (14.1 h; data not shown), and chromosomal integrity is
also generally preserved [supporting information (SI) Fig. S1].
These observations further support the notion that ZFN-
mediated gene targeting can be performed without significant
unwanted genetic or phenotypic disruption.
10 10 1010
cells. (A) Each pair of sequences represents the two alleles of the DHFR gene
in the designated cell line. For each mutant allele, inserted bases are boxed,
and deleted bases are represented by dots. (B) Western blot for DHFR protein
in WT CHO-S cells (WT) and the mutant cell lines 14/1, 14/2, and 1.43 and the
commercially available DHFR-null CHO cell line DG44. Note that, for clone
1.43, we analyzed three independent subclones from this line (H7, B11, and
levels as indicated. (C) Fluorescein-labeled methotrexate (FMTX)-based FACS
analysis of DHFR?/?cell lines. WT(-ve), WT CHO-S cells not treated with FMTX;
with FMTX, including DG44 as a DHFR-null CHO cell control. (Left) Cells in
adherent culture. (Right) Cells in suspension culture.
Santiago et al.
April 15, 2008 ?
vol. 105 ?
no. 15 ?
This study demonstrates the feasibility of a ZFN-based approach
to rapid targeted gene knockout in cultured cells. The targeted
gene mutation frequency was sufficient to identify biallelic
knockouts after a single transient ZFN treatment and without
the need for selection methodologies. In the present study, the
frequency of mutated alleles detected upon screening of clones
(2–3%) was lower than expected given the initial level of NHEJ
measured in the treated pool (15%; Fig. 1C). This may be due
to a selective growth disadvantage of clones that are deficient in
endogenous DHFR expression despite the presence of HT in the
medium. Nevertheless, ?1% of all clones analyzed showed
biallelic modification. We have reported that the targeting of
each of the two copies of a gene within a given single cell does
not appear to be an independent event (16). Our observation
here that approximately one-third of all mutant clones have
biallelic modification is consistent with this. Taken together,
these data demonstrate that the reduced screening effort needed
to identify a ZFN-mediated biallelic disruption at a specific
locus—in the absence of selection—offers a significant advan-
tage compared with existing methods.
In previous reports, we have demonstrated that ZFNs are
powerful tools for gene modification by harnessing homologous
have omitted the donor DNA, relying instead simply on NHEJ-
derived modifications, which are effectively random, leading to a
mutated gene with knocked out function. Our method also con-
trasts with the original demonstration of mutating or knocking out
a gene where ‘‘gene targeting’’ was driven only by the homology
and rapid method for gene knockout.
Abrogating the need for a homologous donor thus simplifies
the approach to ZFN-mediated high-efficiency gene knockout
and eliminates the use of selection markers that may hamper
the utility of the cell line for certain industrial applications (see
below). Rather, the simplicity of the process described here is
driven by the naturally random nature of this mode of repair
that frequently results in a gain or loss of genetic information
at the target locus sufficient to disrupt gene expression. As we
have seen, this can occur by causing a shift in reading frame
to generate a premature termination codon (and possibly
destruction of the mRNA by nonsense-mediated decay), as for
cell lines 14/1 and 1.43A5, or by deletion of critical amino acid
residues, as with cell line 14/2. NHEJ-based DNA repair is
inherently less precise in that a specific genetic outcome is not
predetermined by donor DNA design, but this is not relevant
to our purpose here. The high frequencies of gene disruption
strongly support the likelihood of readily achieving the desired
genotype in which each copy of the target gene is functionally
In the absence of viable methods for rapid gene knockout,
several new technologies (e.g., antisense, RNAi, and zinc
finger protein transcription factors) have been developed
recently that enable the knockdown of gene expression, some-
times achieving ?90% reduction in target gene levels for as
long as the repressor molecules are present (26). The ease and
Culture Period (Days)
( s l l e
WT CHO +HT
WT CHO -HT
Clone 14/1 +HT
Clone 14/1 -HT
Clone 14/2 +HT
Clone 14/2 -HT
Culture Period (Days)
x ( l l e
l a t
( y t i l i b
Growth - HT
Viability - HT
14/1 14/2 1.43A5 WT Ad.WT Susp.
0 50 250
cultured in the presence or absence of HT supplement as indicated. (Upper) Growth of DHFR?/?cell lines 14/1 and 14/2 compared with WT CHO cells in adherent
culture conditions. (Lower) Growth and viability of the DHFR?/?cell line 1.43A5 in serum-free suspension culture. (B) IgG expression from WT or DHFR?/?cell
(?HT) of HT supplement for 14 days. IgG expression at ?48 h was measured by ELISA. WT Ad., WT CHO-S cells in adherent medium; WT Susp., WT CHO-S cells
in suspension medium. Cell lines 14/1 and 14/2 were cultured adherently. Cell line 1.43A5 was cultured in suspension. (C) Methotrexate selection/amplification
in DHFR?/?cell line 14/1. Data shown is the level of IgG stably expressed from a clone 14/1 pool after 2 weeks of incubation in the absence of HT then 2 weeks
more in the presence of the noted concentration of methotrexate (see Materials and Methods).
Growth and functional properties of the ZFN-generated DHFR?/?cell lines. (A) WT CHO-S (WT CHO) and ZFN-generated DHFR?/?CHO-S cell lines were
www.pnas.org?cgi?doi?10.1073?pnas.0800940105Santiago et al.
rapidity of use of some of these approaches has made them the
method of choice for applications such as initial target vali-
dation. However, possible toxicity effects on their long term
expression and the incomplete phenotypic penetrance that
may arise from partial knockdown of the target gene may prove
undesirable. Additional metabolic load on cells due to over-
expression of these factors during times of stress (e.g., during
large scale fermentation) may also be a concern. In such
instances, rapid and permanent gene knockout, using ZFNs,
can offer distinct advantages. Moreover, the growing number
of reports, using ZFNs across different species (11–13, 16),
suggest that ZFN-mediated gene disruption may be a robust
and general method for targeted gene knockout.
There are several other critical differences between the
present approach and existing methods that offer further ad-
vantages. ZFN-mediated gene knockout requires only transient
expression of the ZFNs yet results in a permanent genetic mark.
Because this is an alteration to the genome itself, the mutation
is stably transmitted through all subsequent generations of the
cell line, as is the case with conventional gene targeting. Another
unique aspect of the ZFN approach is that no selectable markers
are required to identify or isolate the targeted events. This has
practical value—especially in industrial applications such as
recombinant protein manufacture—where the sustained expres-
sion of exogenous marker genes is undesirable. It allows also for
multiple genes to be targeted without constraints arising from a
limited number of available selectable marker genes. This latter
point, in conjunction with the speed by which the knockout of
any given gene can be achieved, opens up possibilities for
multigene targeting, or trait stacking, in ways that have not
previously been feasible. Future studies will be aimed at testing
Materials and Methods
Cell Culture and Transfection. ZFN-mediated DHFR gene knockout was per-
formed in CHO-S cells (Invitrogen; catalog no. 11619-012), which are
functionally diploid for DHFR. As a DHFR-negative control cell line, we used
the CHO cell derivative DG44 (Invitrogen, catalog no. 12609-012) in which
the DHFR gene had been completely disrupted by chemical mutagenesis
and ionizing radiation (2). For adherent culture, WT CHO-S cells clones and
all derivative mutant cell lines were grown in DMEM (Invitrogen; catalog
no. 11965-092), supplemented with 10% FBS (JRH BioSciences; catalog no.
12117-500M), 10 mM NonEssential Amino Acids (Invitrogen; catalog no.
11140-050), 8 mM L-Glutamine (Invitrogen; catalog no. 25030-081), and 1?
HT supplement (100 ?M sodium hypoxanthine and 16 ?M thymidine;
Invitrogen; catalog no. 11067-030) where specified. For ZFN transfections,
cells were transfected with 100 ng of each ZFN expression plasmid (in pairs)
? 400 ng of pCDNA, using Lipofectamine2000. After 72 h, cells were
replated in 96-well format at limiting dilution (1 cell per well on average).
CHO DG44 cells were cultured as per the manufacturers recommended
Suspension cultured CHO-S cells and all derivative clones were grown and
maintained as 30-ml suspension cultures in EX-CELL CD CHO serum-free
medium (Sigma–Aldrich; catalog no. 14361C) supplemented with 8 mM L-
glutamine in a humidity-controlled shaker incubator (ATR, Inc.) at 125 rpm
with 5% CO2 at 37°C. For studies showing the dependence of the DHFR
knockout cell lines on supplemental HT, we used similar medium that was
deficient in HT (Sigma–Aldrich; catalog no. 14360C) but also contained 8 mM
added glutamine. Cells were transfected with up to 2 ?g of each ZFN plasmid,
using Amaxa Nucleofector II according to manufacturer’s protocol. Viability
was measured by trypan blue staining.
CEL-I Nuclease Mismatch Assay. The frequency of targeted gene mutation in
ZFN-treated pools of cells was determined by using the CEL-1 nuclease assay
from WT as a result of NHEJ-mediated imperfect repair of ZFN-induced DNA
ZFN-treated cells generates a mixture of WT and mutant amplicons. Melting
and reannealing of this mixture results in mismatches forming between
heteroduplexes of the WT and mutant alleles. A DNA ‘‘bubble’’ formed at the
site of mismatch is cleaved by the surveyor nuclease CEL-I, and the cleavage
etry. The relative intensity of the cleavage products compared with the
parental band is a measure of the level of CEL-I cleavage of the heteroduplex.
This, in turn, reflects the frequency of ZFN-mediated cleavage of the endog-
The sequence of PCR primers used for amplification of the DHFR target locus
is: Forward primer (129F), 5?-TAGGATGCTAGGCTTGTTGAGG; Reverse primer
from WT genomic sequence.
Western Blot Analysis. The primary antibody used to detect DHFR was from
Santa Cruz Biotechnology (catalog no. sc-14780); for detection of TFIIB, the
antibody was from Santa Cruz Biotechnology (catalog no. scsc-225); and, for
glutamine synthetase, the antibody was from BD Bioscience (catalog no.
Fluorescent Methotrexate (FMTX) Assay. HT-supplemented growth medium
was used throughout. The medium on 1 ? 106cells was replaced with 500 ?l
of fresh medium containing 10 ?M FMTX (Invitrogen; catalog no. M1198MP)
and incubated at 37°C for 2 h. The medium was then replaced with 1 ml of
once with PBS and resuspended in PBS ? 1% FBS at a density of 1 ? 105cells
at 516 nm.
Antibody Expression and ELISA. WT and DHFR?/?cells were grown initially in
medium supplemented with HT. Cells (1 ? 106) were cotransfected with 2 ?g
of each of two monoclonal antibody expression plasmids in which one con-
tained an expression cassette for the antibody light chain, whereas the other
contained a cassette for the antibody heavy chain and a cassette for DHFR
expression. The following day, the growth medium was replaced with or
without HT supplement, and the cells were left to grow for up to 14 days,
inoculated into 1.0 ml of fresh HT-supplemented medium. After 48 h, the
medium was collected and analyzed for antibody expression by ELISA accord-
ing to manufacturers protocol (Bethyl Laboratories; Human IgG ELISA quan-
titation kit, catalog no. E80-104).
Methotrexate Selection. The DHFR?/?cell line 14/1 was transfected with the
antibody constructs described above and grown adherently in the absence of
HT supplement for 2 weeks. Methotrexate was then added to the medium at
the noted concentration followed by a further 2 weeks of incubation. The
growth medium was then removed, and 1.0 ml of fresh HT-supplemented
medium was added. After 48 h, the medium from each sample was collected
and analyzed for antibody expression by ELISA as above.
ACKNOWLEDGMENTS. We thank Yingying Wu, George Katibah, Anna Vin-
cent, and Sarah Hinkley for generation of the ZFN reagents; Jingwei Yu for
performing the cytogenetic analyses; and Andreas Reik for helpful discussion
and review of the manuscript. Cytogenetic studies were performed in the
laboratory of Dr. Jingwei Yu (Cytogenetics Laboratory, University of Califor-
nia, San Francisco, CA). Early studies leading to this work were supported by
Advanced Technology Program grants from the National Institutes of Stan-
dards and Technology Grant 70NANB4H3006.
1. Urlaub G, Chasin LA (1980) Isolation of Chinese hamster cell mutants deficient in
dihydrofolate reductase activity. Proc Natl Acad Sci USA 77:4216–4220.
2. Urlaub G, Kas E, Carothers AM, Chasin LA (1983) Deletion of the diploid dihydrofolate
reductase locus from cultured mammalian cells. Cell 33:405–412.
3. Mansour SL, Thomas KR, Capecchi MR (1988) Disruption of the proto-oncogene int-2
in mouse embryo-derived stem cells: A general strategy for targeting mutations to
non-selectable genes. Nature 336:348–352.
4. Vasquez KM, Marburger K, Intody Z, Wilson JH (2001) Manipulating the mammalian
genome by homologous recombination. Proc Natl Acad Sci USA 98:8403–8410.
5. Yamane-Ohnuki N, et al. (2004) Establishment of FUT8 knockout Chinese hamster
ovary cells: An ideal host cell line for producing completely defucosylated antibodies
with enhanced antibody-dependent cellular cytotoxicity. Biotechnol Bioeng 87:
6. Kohli M, Rago C, Lengauer C, Kinzler KW, Vogelstein B (2004) Facile methods for
generating human somatic cell gene knockouts using recombinant adeno-associated
viruses. Nucleic Acids Res 32:e3.
7. Hirata R, Chamberlain J, Dong R, Russell DW (2002) Targeted transgene insertion into
human chromosomes by adeno-associated virus vectors. Nat Biotechnol 20:735–738.
Santiago et al.
April 15, 2008 ?
vol. 105 ?
no. 15 ?
8. Sonoda E, Hochegger H, Saberi A, Taniguchi Y, Takeda S (2006) Differential usage of Download full-text
non-homologous end-joining and homologous recombination in double strand break
repair. DNA Repair (Amst) 5:1021–1029.
9. Weterings E, van Gent DC (2004) The mechanism of non-homologous end-joining: A
synopsis of synapsis. DNA Repair (Amst) 3:1425–1435.
10. Liang F, Han M, Romanienko PJ, Jasin M (1998) Homology-directed repair is a major
double-strand break repair pathway in mammalian cells. Proc Natl Acad Sci USA
11. Bibikova M, Golic M, Golic KG, Carroll D (2002) Targeted chromosomal cleavage and
mutagenesis in Drosophila using zinc-finger nucleases. Genetics 161:1169–1175.
nucleases in Arabidopsis. Proc Natl Acad Sci USA 102:2232–2237.
nuclease-targeted double-strand breaks in Caenorhabditis elegans somatic cells. Proc
Natl Acad Sci USA 103:16370–16375.
14. Porteus MH, Carroll, D (2005) Gene targeting using zinc finger nucleases. Nat Biotech-
15. Durai S, et al. (2005) Zinc finger nucleases: Custom-designed molecular scissors for
genome engineering of plant and mammalian cells. Nucleic Acids Res 33:5978–5990.
16. Urnov FD, et al. (2005) Highly efficient endogenous human gene correction using
designed zinc-finger nucleases. Nature 435:646–651.
17. Wurm FM (2004) Production of recombinant protein therapeutics in cultivated mam-
malian cells. Nat Biotechnol 22:1393–1398.
18. Klon AE, et al. (2002) Atomic structures of human dihydrofolate reductase complexed
with NADPH and two lipophilic antifolates at 1.09 Å and 1.05 Å resolution. J Mol Biol
19. Bartsevich VV, Miller JC, Case CC, Pabo CO (2003) Engineered zinc finger proteins for
controlling stem cell fate. Stem Cell 21:632–637.
genome editing. Nat Biotechnol 25:778–785.
21. Szczepek M, et al. (2007) Structure-based redesign of the dimerization interface
reduces the toxicity of zinc-finger nucleases. Nat Biotechnol 25:786–793.
22. Zhang K, Kaufman RJ (2006) Protein folding in the endoplasmic reticulum and the
unfolded protein response. Handb Exp Pharmacol 172:69–91.
23. Gaudray P, Trotter J, Wahl GM (1986) Fluorescent methotrexate labeling and flow
cytometric analysis of cells containing low levels of dihydrofolate reductase. J Biol
24. Omasa T (2002) Gene amplification and its application in cell and tissue engineering.
J Biosci Bioeng 94:600–605.
genome using designed zinc-finger nucleases. Proc Natl Acad Sci USA 104:3055–3060.
26. Tan S, et al. (2003) Zinc-finger protein-targeted gene regulation: Genomewide single-
gene specificity. Proc Natl Acad Sci USA 100:11997–12002.
27. Pabo CO, Peisach E, Grant RA (2001) Design and selection of novel Cys2his2 zinc finger
proteins. Annu Rev Biochem 70:313–340.
www.pnas.org?cgi?doi?10.1073?pnas.0800940105 Santiago et al.