Chromosomal context and epigenetic mechanisms control the efficacy of genome editing by rare-cutting designer endonucleases.

Abstract

The ability to specifically engineer the genome of living cells at precise locations using rare-cutting designer endonucleases has broad implications for biotechnology and medicine, particularly for functional genomics, transgenics and gene therapy. However, the potential impact of chromosomal context and epigenetics on designer endonuclease-mediated genome editing is poorly understood. To address this question, we conducted a comprehensive analysis on the efficacy of 37 endonucleases derived from the quintessential I-CreI meganuclease that were specifically designed to cleave 39 different genomic targets. The analysis revealed that the efficiency of targeted mutagenesis at a given chromosomal locus is predictive of that of homologous gene targeting. Consequently, a strong genome-wide correlation was apparent between the efficiency of targeted mutagenesis (≤ 0.1% to ≈ 6%) with that of homologous gene targeting (≤ 0.1% to ≈ 15%). In contrast, the efficiency of targeted mutagenesis or homologous gene targeting at a given chromosomal locus does not correlate with the activity of individual endonucleases on transiently transfected substrates. Finally, we demonstrate that chromatin accessibility modulates the efficacy of rare-cutting endonucleases, accounting for strong position effects. Thus, chromosomal context and epigenetic mechanisms may play a major role in the efficiency rare-cutting endonuclease-induced genome engineering.

Full text

Page 1

Chromosomal context and epigenetic mechanisms
control the efficacy of genome editing by
rare-cutting designer endonucleases
Fayza Daboussi1, Mikhail Zaslavskiy1, Laurent Poirot2, Mariana Loperfido3,4,
Agne `s Gouble2, Valerie Guyot1, Sophie Leduc1, Roman Galetto2, Sylvestre Grizot1,
Danusia Oficjalska1, Christophe Perez1, Fabien Delaco ˆte1, Aure ´lie Dupuy2,6,
Isabelle Chion-Sotinel2, Diane Le Clerre2, Ce ´line Lebuhotel2, Olivier Danos5,
Fre ´de ´ric Lemaire1, Kahina Oussedik1, Fre ´de ´ric Ce ´drone1, Jean-Charles Epinat1,
Julianne Smith2, Rafael J. Ya ´n ˜ez-Mun ˜oz7, George Dickson7, Linda Popplewell7,
Taeyoung Koo7, Thierry VandenDriessche3,4, Marinee K. Chuah3,4, Aymeric Duclert1,
Philippe Duchateau1,* and Fre ´de ´ric Pa ˆques1
1CELLECTIS S.A.,2Cellectis Therapeutics, 8 rue de la Croix Jarry, 75 013 Paris, France,3Division of
Gene Therapy & Regenerative Medicine, Free University of Brussels, Laarbeeklaan 103, B-1090 Brussels,
4Department of Molecular and Cellular Medicine, University of Leuven, Herestraat 49, B-3000, Leuven, Belgium,
5Inserm U845, Ho ˆpital Necker Enfants Malades, Universite ´ Paris Descartes, 156, rue de Vaugirard – 75730
Paris Cedex 15,6UMR 8200 CNRS, Institut de cance ´rologie Gustave Roussy, 114, rue Edouard Vaillant,
94805 Villejuif cedex and7School of Biological Sciences, Royal Holloway, University of London, Surrey,
TW20 0EX, UK
Received December 19, 2011; Revised and Accepted March 9, 2012
ABSTRACT
The ability to specifically engineer the genome of
living cells at precise locations using rare-cutting
designer endonucleases has broad implications for
biotechnology and medicine, particularly for func-
tional genomics, transgenics and gene therapy.
However, the potential impact of chromosomal
context and epigenetics on designer endonuclease-
mediated genome editing is poorly understood. To
address this question, we conducted a comprehen-
sive analysis on the efficacy of 37 endonucleases
derived from the quintessential I-CreI meganuclease
that were specifically designed to cleave 39 different
genomic targets. The analysis revealed that the
efficiency of targeted mutagenesis at a given
chromosomal locus is predictive of that of homolo-
gousgene targeting.Consequently,
genome-wide correlation was apparent between
the efficiency of targeted mutagenesis (?0.1% to
?6%) with that of homologous gene targeting
a strong
(?0.1% to ?15%). In contrast, the efficiency of
targeted mutagenesis or homologous gene target-
ing at a given chromosomal locus does not correl-
ate with the activity of individual endonucleases
on transiently transfected substrates. Finally, we
demonstrate that chromatin accessibility modu-
lates the efficacy of rare-cutting endonucleases,
accounting forstrong
chromosomal context and epigenetic mechanisms
may play a major role in the efficiency rare-cutting
endonuclease-induced genome engineering.
positioneffects. Thus,
INTRODUCTION
The development of a comprehensive technology platform
to efficiently engineer the genome at specific sites in living
cells opens new opportunity for functional genomics,
generation of transgenic models, gene therapy, stem cell
engineering and regenerative medicine. Such targeted
genome modification has typically been achieved using
different types of designer rare-cutting endonucleases
*To whom correspondence should be addressed. Tel: +33 1 81 68 16 00; Fax: +33 1 81 68 16 00; Email: duchateau@cellectis.com
Published online 29 March 2012 Nucleic Acids Research, 2012, Vol. 40, No. 136367–6379
doi:10.1093/nar/gks268
? The Author(s) 2012. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/3.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

that induce a DNA double-strand break (DSB) at a
desired specific location in the genome. Different scaffolds
have been used to generate these endonucleases (1,2), and
at least four families of such endonucleases have been
documented so far: zinc finger nucleases (ZFNs) (3–5),
meganucleases (MNs) from the LAGLIDADG family
(6,7),chemicalnucleases
activator-like nucleases (TALENs) (10–12). From a struc-
tural or chemical perspective, these designer rare-cutting
endonucleases differ in several respects, which have been
reviewed elsewhere (1,2). Consequently, one can expect
that each of these platforms will have different properties
in terms of efficacy, specificity, range of cleavable se-
quences, spectrum of induced events (homologous gene
targeting versus targeted mutagenesis, see below), and
therefore, of potential applications.
MNs and ZFNs are currently the most well documented
and widely used reagents. The I-SceI MN has been used
since the 1990s in genome engineering experiments (13,14).
Redesign of the DNA binding interface of the I-CreI
MN by protein engineering has substantially increased
the number of sequences that can be cleaved (15,16).
Similarly, a large number of ZFNs with tailored
specificities have been generated and used by several
laboratories in a wide range of cell types (3–5,17–21).
Both MNs and ZFNs have the ability to deliver targeted
DSBs that subsequently undergo either non-homologous
end-joining(NHEJ) orhomologous
(HR) (1). NHEJ, an error prone DSB repair pathway,
has been used to inactivate the open reading frames of
selected genes by targeted mutagenesis (TM) (22,23). In
contrast, HR is used mainly for targeted gene insertion or
correction,by homologous
Althoughcleavage of promiscuous
sequences have also been observed, MNs and ZFNs
have both been used to induce precise gene modifications
without affecting global genome integrity, even in
organisms with large genomes such as fish, mammals
and plants (1).
Though genome editing based on these designer
rare-cutting endonucleases is determined by the recogni-
tion of a specific nucleotide target sequence, it is likely
that chromosomal context and epigenetic modifications
of target loci could also play an important role.
Consequently, chromatin accessibility and/or the position-
ing of nucleosomes in vivo (24,25) at the target locus may
impact TM and/or HGT. Alternatively, the location of
different target loci in distinct ‘chromosomal territories’
could possibly influence genome editing by modulating
access to limiting factors required for DNA repair (26).
To address these outstanding questions and assess the
potential impact of chromosomal position effects and
epigenetics on designer rare-cutting endonucleases, we
have conducted a comprehensive analysis of the properties
of MNs in human cells. We therefore monitored TM
and HGT efficacies in an unprecedented large number of
chromosomalloci across
determined the effect of chromatin accessibility on
genome editing by MNs. The current study contributes
to a better understanding of the properties of designer
(8,9) and transcription
recombination
genetargeting
or
(HGT).
‘off-target’
thehuman genome and
rare-cutter endonucleases which has important fundamen-
tal biological and translational implications.
MATERIALS AND METHODS
Meganucleases
MNs derived from I-CreI were engineered as described
previously (16,27,28). All MNs were used under a single
chain format, as described in Grizot et al. (16), except for
CAPNS1m and NACAm which were used in 293-H cells
undera previouslydescribed
(16,28).
heterodimeric format
293-H cells culture and transfections
Human 293-H cells (Life Technologies, Carlsbad, CA,
USA) and hamster CHO-KI cells (ATCC) were cultured
at 37?C with 5% CO2in complete medium DMEM and
F12-K, respectively, supplemented with 2mM L-glutam-
ine, penicillin (100IU/ml), streptomycin (100mg/ml),
amphotericin B (Fongizone: 0.25mg/ml, Life Techno-
logies,) and 10% FBS. For extrachromosomal and
survival assays, CHO-KI cells were plated at 2500 cells
per well in 96-well plate. The next day, cells were trans-
fected with 200ng of total DNA using Polyfect transfec-
tion reagent according to the supplier’s (Qiagen) protocol.
In the survival assay, a constant amount of GFP-encoding
plasmid (10ng) was added to various amounts of MN
expression vectors. For mutagenesis and gene targeting
assays, 293-H cells were plated at a density of 1.2?106
cells per 10cm dish. The next day, cells were transfected
with 3 or 5mg of plasmid DNA using Lipofectamine 2000
transfection reagent (Life Technologies) according to the
manufacturer’s protocol.
Extrachromosomal activity assays in CHO-KI and
HEK293 cells
Activity in mammalian cells was measured as previously
reported by Grizot et al. (29) These assays are used to
monitor the activity of the MNs in a non-chromatinized
template, an activity that is a function of MN expression
and specific activity.
Cell survival assay
The CHO-KI cell line was transfected as described above,
with varying amounts of MN expression vector and a
constant amount of GFP-encoding plasmid. GFP levels
were monitored by flow cytometry (Guava EasyCyte,
Guava Technologies) on Days 1 and 6 post-transfection.
Cell survival was calculated as a ratio (MN-transfected
cells expressing GFP on Day 6/control transfected cells
expressing GFP on Day 6) corrected for the transfection
efficiency determined on Day 1.
Area under the curve scores
Area under the curve (AUC) score is used to quantify
(i) the activity of MN in extrachromosomal assays
in CHO-KI and 293-H cells; (ii) cell survival in the
toxicity assay; (iii) chromatin resistance to microccocal
nuclease digestion at a given locus. Examples are given
6368 Nucleic Acids Research, 2012,Vol.40, No. 13

Page 3

in Figure 2b and Supplementary Figure S2c. The classical
approach based on the use of log-normal model (30)
showed a very good fitting quality for both survival
curves and activity curves (after normalization with
respect to the maximal activity level). All the statistics
done in this study are summarized in Supplementary
Table S6.
Western blot
293-H and CHO-KI cells were plated at 1.2?106and
2?105cells per 10cm dish, respectively. These two cell
lines were transfected with the same DNA plasmid-
encoding MN: 3mg for 293-H cells and 1mg for
CHO-K1 cells, according to the usual transfection condi-
tion. Two days post-transfection, total protein extraction
was performed with RIPA (150mM sodium chloride, 1%
NP-40, 0.5% sodium deoxycholate, 01% SDS, 50mM
Tris pH 8.0) buffer. Western blot was performed on
10mg of total protein extract. The MN expression was
revealed using a specific I-CreI antibody. Beta-tubulin
antibody (Cell signaling 2128) was used as loading
control. Quantification of signal was performed on low
exposed blots using ImageJ software. Results were ex-
pressed as a ratio between MN and tubulin signals.
Monitoring of MN-induced TM
293-H cells were transfected with 3mg of MN expressing
vector or empty vector, except for ADCY9-induced mu-
tagenesis where 293-H cells were transfected with 5mg
of MN-encoding plasmid. Two days post-transfection,
genomic DNA was extracted and targets studied were
amplified with specific primers
adaptators needed for HTS sequencing (For: CCATCTC
ATCCCTGCGTGTCTCCGACTCAG-Tag
BiotinTEG/ CCTATCCCCTGTGTGCCTTGGCAGTC
TCAG) on the454 sequencing
Sciences). An average of 10000 sequences per sample
were analysed. Sequence of the primers used to amp-
lify the targeted locus are indicated on Supplementary
Table S3.Inclonal experiments
DMD21m), cells were plated after treatment and indi-
vidual clones were grown. After 21 days, DNA extraction
was performed with the ZR-96 genomic DNA kit (Zymo
research) according to the supplier’s protocol. The region
of the target was amplified by PCR as described above,
and PCR product was sequenced.
flanked by specific
andRev:
system(454 Life
(Rag1m and
Monitoring of MN-induced HGT
293-H cells were co-transfected with 3mg of MN express-
ing vector, and 2mg of DNA matrix and seeded at 10 cells
per well in 96-well plate, and HGT was monitored by PCR
21 days later. Alternatively, cells were plated and individ-
ual colonies were picked 2–3 weeks later in 96-well plates
and analysed (Supplementary Table S1). All DNA
matrices comprise 1kb of homologous sequences located
on both sides of the MN recognition site. The two
homologyarms are separated
fragment of 29bp or 1.7kb. DNA extraction was per-
formed with the ZR-96 genomic DNA kit (Zymo
research) according to the supplier’s protocol. The
bya heterologous
detection of targeted integration is performed by specific
PCR amplification using a primer located within the heter-
ologous insert of the DNA repair matrix, and another one
located on the genomic sequence outside of the homology.
In pool experiments (10 cells/well), HGT frequencies were
normalized to plating efficiencies, e.g. 0.33. This method
was validated by the good correlation between the
frequencies in clonal analysis and the frequencies in
pool analysis (Supplementary Table S1). Sequence of the
primers used to amplify the homologous regions is
available upon request. Sequence of the primers used
to diagnostic targeted integrations are indicated on
Supplementary Table S3. HGT was also confirmed by
SouthernBlot forRag1m,
XPC4m and CLS3690m.
DMD21m,ADCY9m,
Simultaneous monitoring of MN-induced HGT and TM
HEK293-H cells were co-transfected with plasmids ex-
pressing the MN and 2mg of repair matrix. In each experi-
ment, we amplified the region surrounding the target site
by nested PCR using deep sequencing primers, with the
first set of primers being located outside of the repair
matrix(seeSupplementary
sequence). We measured both HGT and TM by deep
sequencing of the amplicons.
Table S3forprimer
CHART-PCR
Cells were resuspended in 1ml of sucrose buffer supple-
mented with 0.5% of NP-40. The pellet nuclei was resus-
pended in ice-cold nuclei buffer supplemented with DTT
and PMSF at a concentration of 2?104nuclei/ml.
Digestion was performed on 1.5?105nuclei at room tem-
perature for 15min with 10–500 U of micrococcal nuclease
(Biolabs). Restriction cleavage was stopped by the
addition of 10mM EDTA. DNA was extracted in the
presence of proteinase K and then quantify with picogreen
(Quant-iTTM
PicoGreen?
Probes). Q-PCR was performed in duplicate with specific
primers described in Supplementary Table S3. The data
are expressed as percentage of Q-PCR signal of undigested
sample.
dsDNA Assay,Molecular
RESULTS
Target selection
We have previously described a combinatorial method
to generate MNs that cleave selected sequences in virtually
any gene or locus (16,28,31). Briefly, this approach
relies on the use of an archive of locally engineered deriva-
tives of the I-CreI MN. These MNs recognize targets
differing from the I-CreI target by a few base pairs only,
but can be used as building blocks to generate combina-
torial libraries of fully redesigned MNs. MNs recognizing
specific chosen targets,unrelated
recognized by the parental I-CreI scaffold, are identified
using a high-throughput yeast cell assay. Typically, a
MN cleavage site is placed between two direct repeats
ina
b-galactosidase (LacZ)
extra-chromosomal target (Supplementary Figure S1).
to thesequence
reporterplasmid, as
Nucleic Acids Research, 2012,Vol.40, No. 13 6369

Page 4

Cleavage of the target site by the MN in this cell-based
assay induces tandem-repeat recombination, by single-
strand annealing (SSA) between two truncated copies of
the LacZ gene thereby restoring a functional LacZ gene.
This method allowed for the identification of a potential
cleavable target every 250bp, with a success rate of 40%
at the protein engineering step (32). Based on this prin-
ciple, we generated a large collection of 37 different engin-
eered MNs, derived from the parental I-CreI natural
protein, that cleaved 39 different sequences, specifically
designed for genome editing in coding and non-coding
sequences of the human genome (Figure 1 and Table 1).
In this study, m refers to the MN and t its respective target
sequence. For instance, Rag1 designates the chosen locus,
Rag1m the MN, and Rag1t, the 22bp target sequences
cleaved by the Rag1 MN. Our selection was prioritized
towards coding sequences, given their high potential
interest (i.e. gene inactivation and gene correction) and
translational relevance (i.e. gene therapy). Consequently,
this prioritization resulted in a higher proportion of
targets in GC- and CpG-rich regions (Figure 1b).
Characterization of MNs activity based on
extrachromosomal cellular assays
A quantitative cell-based assay in CHO-KI and 293-H
cells was used to characterize the MN activity of this
MN collection was used (Figure 2a and Table 1). This
extrachromosomal assay (29,31) is similar to the cellular
assay in yeast and allows for a quantitative assessment of
Figure 1. Genome-wide distribution of the targets of the MNs used in this study. (a) Schematic representation of the distribution per chromosome.
Thirty-seven MNs cleaving 39 targets were used in this study, including 36 MNs cleaving each single target (represented by filled-in triangles) and 1
MN (CLS3902m), cleaving 3 (represented by open triangles). Target sequences and MN properties are described in Table 1. (b) Distribution per type
of genomic sequence, per GC content and per CpG number. Distributions for target locations (top row) are compared with the distribution found
for the whole genome (bottom row, on assembly GRCh37/hg19). For the type of genomic sequences, two MN target sites are overlapping with an
intron/exon boundary and a special class was made for them. For GC content and CpG number, windows of 1kb are taken around the recognition
site (top row) or successively on the whole genome (bottom row) and the percentage of GC or the number of CpG dinucleotides are computed.
Percentages in each class are displayed. The estimation of the percentage of sequences in exon, intron and intergenic regions on the whole genome is
based on available annotations. The human genome version we used was GRCh37/hg19.
6370Nucleic Acids Research, 2012,Vol.40, No. 13

Page 5

Table 1. Meganucleases and targets
Target characterization
Extra-
chromosomal
test (AUC)
Tox test
Chromatin
accessibility
(AUC)
Targeted
mutagenesis
in 293 cells
(% indels)
Homologous
gene targeting
in 293 cells
(% KI)
MNa
Geneb
Target sequence
Chromosome Seq. type
CHOc
293d
CHO
AUCeact.
dosefToxg293h
CD34hT cellhd7 w/
MNi
d2 w/
MNj
d7 w/o
MNk
w/
MNl
w/o
MNm
CLS3906m
TTTTCCTGTCTTACCAGGTTTTGT
1p34.3
Intergenic
6.63
5.04
6
70
0.10
0.04
CLS3903m
TTAAAACTTCTTAAAGGATCCCAG
1q31.2
Intergenic
8.98
3.14
2.5
60
0.80
0.10
2.00
0.00
CLS3898m
CTAGGCTGGATTACAGCGGCTTGA
2q14.1
Intergenic
8.36
5.37
2.5
80
0.00
0.00
CLS4633m
CCR5
TCGAAATGAGAAGAAGAGGCACAG 3p21
Exon
3.70
1.78
1.2
40
0.04
0.00
XPC4m
XPC
TCGAGATGTCACACAGAGGTACGA
3p25
Exon
10.80
3.52
5.84
2.5
80
188
0.40
0.36
0.00
1.05
0.00
CLS3719m
ATTTCATCACCTAACTAATCTCAC
3p25.3
Intergenic
2.57
0.00
0.02
CLS3191m
Rho
ACTTCCTCACGCTCTACGTCACCG
3q21-q24
Exon
4.82
0.40
5.37
6
95
0.10
0.00
CLS3894m
ACAATCTGAGGTAAGTAATACTGA
5p14.3
Intergenic
8.09
1.45
3.87
6
45
0.50
0.02
0.60
0.00
CLS3902m
TCAAACCATTGTACTCCAGCCTGG
14q24
Intergenic
9.99
3.03
5.36
2.5
90
247
0.19
0.70
0.02
TCAAACCATTGTACTCCAGCCTGG
5q31
Intergenic
337
0.11
0.39
0.02
HECW1 TCAAACCATTGTACTCCAGCCTGG
7p13
Intron
163
0.53
1.17
0.02
CLS2697m
CCAATACAAGGTACAAAGTCCTGA
6p25.1
Intergenic
9.09
4.79
2.5
80
0.70
0.00
CLS2705m
TTAAAACACTGTACACCATTTTGA
7q31.2
Intergenic
6.41
4.49
2.5
75
0.20
0.02
0.90
0.00
CLS2760m
SMC5L1 CAACTCCGTCTTACAGAGGTAAAA
9q21.11
Intron/Exon 8.09
4.96
6
60
0.10
0.00
0.00
0.00
Rag1m
RAG1
TTGTTCTCAGGTACCTCAGCCAGC
11p13
Intron/Exon 9.19
0.82
5.16
2.5
95
106
78
70
1.30
1.28
0.03
8.40
0.00
CLS2607m
HBB
TCAGACTTCTCCTCAGGAGTCAGA
11p15.5
Exon
2.04
5.80
6
95
0.01
0.04
CLS3769m
HBB
AGGGGATAAGTAACAGGGTACAGT 11p15.5
Intron
7.08
4.66
2.5
45
0.06
0.00
CLS4422m
NANOG ACTGAACGCTGTAAAATAGCTTAA
12p13.31
Intron
11.10
5.04
2.5
50
0.10
0.03
CLS3884m
NACA
GGGAACTGCTTCACAGGATCCAGA
12q23-q24.1
Exon
5.88
2.5
0.02
0.06
CLS3896m
ATAAAACAAGTCACGTTATTTTGG
13q34
Intergenic
4.24
5.76
6
85
0.10
0.05
CLS4076m
FUT8
TTGTAATGTCTTACAAGGTTTTAA
14q23.3
Promoteur
10.91
1.29
3.21
2.5
40
55
43
41
0.70
5.09
0.06
0.37
0.00
ADCY9m
ADCY9
CCCAGATGTCGTACAGCAGCTTGG
16p13.3
Exon
8.24
2.87
5.97
2.5
95
410
201
179
0.20
0.17
0.06
0.70
0.00
CLS2276m
p53.7
TCTGGCCCCTCCTCAGCATCTTAA
17p13.1
Exon
9.19
0.10
0.03
0.99
0.00
CAPNS1m
CAPNS1 CAGGGCCGCGGTGCAGTGTCCGAC 19q13.12
50UTR
14.08
3.49
5.15
24.23 21
52
2.40
6.10
0.01
12.00
0.26
scCAPNS1m
13.96
13.96
3.98
1.2
70
ND
6.20
0.01
15.00
0.26
CLS2752m
CTSZ
ATTACCCCCTTCACAGGAGCCAGC
20q13
Intron
9.99
3.89
2.5
40
0.30
0.20
0.00
0.00
CLS3895m
AGAAGCCCAGGTAAAACAGCCTGG 21q21
Intergenic
3.74
4.07
6
50
0.10
0.10
CLS3690m
TTAATACCCCGTACCTAATATTGC
21q21.1
Intergenic
5.71
2.21
4.94
2.5
80
0.50
0.06
1.50
0.00
CLS3899m
GCAAAACATTGTAAGACATGCCAA
22q13.33
Intergenic
10.59
2.73
1.74
2.5
40
0.10
0.01
CLS3702m
WAS
TCCAAACCTCTCACAAAAGTGTAT
Xp11.23-
p11.22
Intron
7.70
2.40
2.5
50
0.10
0.04
CLS3759m
WAS
CCGGGCCCTCGTGCAGGAGAAGAT Xp11.23-
p11.22
Exon
10.52
5.54
2.5
95
286
0.50
0.48
0.00
4.50
0.40
DMD21m
DMD
GAAACCTCAAGTACCAAATGTAAA
Xp21.2
Intron
4.10
0.35
5.62
6
95
61
25
35
1.40
1.20
0.05
8.40
0.00
CLS3326m
DMD
AAATCCTGCCTTAAAGTATCTCAT
Xp21.2
Intron
6.76
5.59
2.5
100
0.10
0.07
0.09
0.00
CLS3402m
DMD
TTTACCTATTTTAAGTCAGATACA
Xp21.2
Intron
2.95
0.36
3.75
2.5
65
0.00
0.03
CLS3631m
DMD
AATGTCTGATGTTCAATGTGTTGA
Xp21.2
Intron
7.03
1.65
3.54
6
50
0.20
0.01
0.60
0.00
CLS4607m
DMD
GAATCCTGTTGTTCATCATCCTAG
Xp21.2
Intron
9.19
0.90
0.08
4.00
0.00
CLS4418m
CCAACATCCTGAACCTCAGCTAC
Xq11
Pseudogene
5.61
3.61
1.2
85
0.30
ND
CLS3368m
IL2RG
CCAGGCTTGAGTGCAGTGGGGCGA Xq13.1
Intron
10.17
4.35
1.2
85
0.02
0.02
CLS3640m
IL2RG
ATTGCACTTGGAACAGCAGCTCTG
Xq13.1
Exon
6.25
1.57
5.13
2,5
80
0.40
0.00
(continued)
Nucleic Acids Research, 2012,Vol.40, No. 13 6371

Page 6

MN activity based on LacZ reporter gene expression
(Figure 2b). Importantly, the activity measured in this
extrachromosomal in vivo cellular assay depends not
only on the intrinsic specific activity of the MN, but also
on the MN expression level in the target cell. Whereas
specific activity can vary widely even between closely
related I-CreI derivatives (28,33), MN expression levels
can also be strongly affected by the few amino acid sub-
stitutions that differ from one engineered MN to another
(Figure 2c). Nevertheless, there is a strong correlation in
relative MN expression levels in different cell types, at
least based on 293-H and CHO-KI cells (Figure 2d).
Most importantly, when comparing MN activity in
CHO-KI and 293-H cells, we observed very consistent
results (Figure 2e), with linear (Pearson’s r) and non-linear
(Spearman’s r) correlation coefficients ranging between
0.71 and 0.75, and P?0.002. We also compared MN
activity measured in our CHO-KI cell assay with that
obtained with aforementioned yeast assay, that also
measures MN-induced tandem repeat recombination on
a plasmid-borne reporter system (10,16,21,27). Here
again, we observed a very good consistency (r=0.8; and
r=0.79withP<2.2?10?16
Supplementary Figure S1b), suggesting that our extra-
chromosomal assay in CHO cells is predictive for a wide
range of different cell types. This MN activity assay in
CHO cells therefore serves as an appropriate and repre-
sentative assay for MN activity.
inboth cases;
MN-induced TM and HGT efficiencies are strongly
correlated
MNs were then tested for their efficacy, defined in this
study as their ability to induce targeted gene modification
at their cognate chromosomal locus in human cells. Results
are summarized in Table 1. TM was monitored for all MNs
by pyrosequencing of the targeted locus 7 days after trans-
fection of 3mg of MN-expressing vector (Figure 3a). The
locus was amplified by PCR, and an average of 10000 mol-
ecules was sequenced per amplicon. The MN-induced rate
varied between background level (up to 0.06% due to PCR
and sequencing errors) and 2.4% of the analyzed chroma-
tids (Table 1). HGT was monitored for 18 MNs by PCR in
pools of 10 cells grown in 96-well plates, 21 days after
transfection, as described previously (16) (Figure 3b).
HGT was also monitored in clonal experiments for
4 MNs, with the results confirming the results from pools
in all cases (Supplementary Table S1). For the 18 MNs that
were characterized for both TM and HGT, a strong linear
correlation (r=0.93 with P=2?10?8) was observed
between the frequency of TM ‘per chromatid’ and the fre-
quency of HGT ‘per cell’ throughout the target genome
(Figure3c). Most importantly,
analysis revealed that the efficiency of TM at a given
chromosomal locus is predictive of that of HGT and that
the HGT/TM ratio is relatively conserved throughout the
genome.
In Figure 3c, we compared HGT events per cell and TM
events ‘per chromatid’ in 293-H cells. However, these
immortalized cells typically do not have a normal
diploid genome. Therefore, we refined our analysis and
thiscomprehensive
Table 1. Continued
Target characterization
Extra-
chromosomal
test (AUC)
Tox test
Chromatin
accessibility
(AUC)
Targeted
mutagenesis
in 293 cells
(% indels)
Homologous
gene targeting
in 293 cells
(% KI)
MNa
Geneb
Target sequence
Chromosome Seq. type
CHOc
293d
CHO
AUCeact.
dosefToxg293h
CD34hT cellhd7 w/
MNi
d2 w/
MNj
d7 w/o
MNk
w/
MNl
w/o
MNm
CLS3676m
IL2RG
CCTAAACTTGGTGCCCCATCTCTC
Xq13.1
Intron
11.53
2.09
3.98
6
50
0.20
0.04
1.08
0.00
I-CreI
NA
TAGGGATAACAGGGTAAT
NA
NA
12.39
2.20
1
95
NA
NA
NA
NA
NA
NA
NA
NA
I-SceI
NA
TCAAAACGTCGTGAGACAGTTTGG
NA
NA
5.56
0.61
5.98
6
95
NA
NA
NA
NA
NA
NA
NA
NA
aFor statistics of Figures 3 and 4, results for CAPNS1 target are from CAPNS1m.
bWhen in a gene, or in the promoter.
cMeasured as shown in Figure 2A and B, in CHO-KI cells.
dMeasured as shown in Figure 2A and B, in 293-H cells.
eMeasured as shown in Figure 2G and 2H, in CHO-KI cells.
fMeasured in nanograms, amount of transfected DNA corresponding to MN maximal activity.
gMeasured in percentage of survived cells at the active dose.
hMeasured as shown in Figure 5, see also ‘Materials and Methods’ section.
iMeasured in percentage of chromatids, as shown in Figure 3A, 7 days after transfection with MN expressing vector in 293-H cells.
jMeasured in percentage of chromatids, as shown in Figure 3A, 2 days after transfection with MN expressing vector in 293-H cells.
kMeasured in percentage of chromatids, as shown in Figure 3A, 7 days after mock transfection in 293-H cells.
lMeasured in percentage of cells, as shown in Figure 3B, 21 days after transfection with MN expressing vector vector in 293-H cells. HGT frequencies were normalized to plating efficiencies, e.g. 0.33.
mMeasured in percentage of cells, as shown in Figure 3B, 21 days after mock transfection in 293-H cells HGT frequencies were normalized to plating efficiencies, e.g. 0.33.
6372 Nucleic Acids Research, 2012,Vol.40, No. 13

Page 7

conducted additional studies to simultaneous monitor
HGT and TM ‘per chromatid’ by deep sequencing.
Events were monitored at Day 2, to alleviate toxicity
effect over time. Sixteen measurements were made with
seven different MNs (Supplementary Table S2). Again, a
very stronglinearcorrelation
P=3?10?6) was observed between the frequency of
HGT ‘per chromatid’ and frequency of TM ‘per chroma-
tid’ (Figure 3d, Supplementary Table S2), and in this case,
(r=?0.90with
a global HGT/TM ratio of 1.7 could be inferred from the
slope of the regression line.
MN efficacies are poorly correlated with their activities
based on extrachromosomal substrates
The differences in MN efficacy on chromosomal target loci
may reflect different parameters, particularly the activity
of a given MN, as measured using extrachromosomal
(a)(b)
(c)
(e)
(d)
I
Transfection
CHO-KI or 293-H cells
ONPG test (day 2)
Activity (AUC)
CLS4076m
I-SceI
Rag1m
MN
Amount of transfected DNA (ng)
β-galactosidase activity (AU)
4
3
2
1
0
4
3
2
1
0
02
46
028
46
AUC
SSA in 293-H cells (AUC)
4 6 8 10 12 14
SSA in CHO cells (AUC)
N=16
r=0.75; p=9x10-4
rho=0.71; p=2x10-3
1.0
1.5
2.0
2.5
3.0
3.5
0.5
CLS4076m
CAPNS1m
ADCY9m
XPC4m
DMD21m
CLS3897m
Rag1m
WASm
CLS3899m
Tubulin
MN
293H
CHO
Relative expression in CHO
Relative expression in 293-H
N=9
r=0.74; p=2.3X10-2
rho=0.80; p=1.4X10-2
Figure 2. Characterization of MNs properties using cell-based assays. (a) An extrachromosomal assay for the characterization of MN activity in
mammalian cells. Briefly, MN expressing vectors and reporter vector are co-transfected into CHO-KI or 293-H cells. Upon cleavage of the target
site, tandem repeat recombination by SSA between two truncated copies of the LacZ gene restores a functional b-galactosidase gene, which can be
monitored by standard assays. For each MN, a dose response is performed and AUC is used as a quantitative measure of MN activity. (b) Example
of read out of the extrachromosomal assay described in panel. The activity of the CLS4076m, I-SceIm and Rag1m MNs is featured as an example.
AUC is used as a quantitative measure of MN activity in this type of extrachromosomal assay. For CLS4076, AUC is the area in grey. (c)
Monitoring of MN expression by western blotting. 293-H cells and CHO-K1 were transfected with a plasmid-encoding MN. The MN expression
level was monitored by western blot. (d) Quantification of MN expression level. The MN expression level from western blot (c) was quantified and
normalized to b-tubulin signal. These ratios were plotted for both cell lines and statistical analysis was performed. r, Pearson coefficient (linear
correlation coefficient); r, Spearman coefficient (non-linear correlation coefficient); N, sample size; P, probability of finding a given correlation when
the underlying variables are not correlated. (e) Comparison between the extra-chromosomal SSA assays in CHO-KI and HEK293 cells. Sixteen MNs
were tested in both assays (Table 1). Dotted grey curves represent 95% confidence interval for the regression line. r, Pearson coefficient (linear
correlation coefficient); r, Spearman coefficient (non-linear correlation coefficient); N, sample size; P, probability of finding a given correlation when
the underlying variables are not correlated.
Nucleic Acids Research, 2012,Vol.40, No. 136373

Page 8

substrates,
However, other parameters may also contribute to these
differences in MN efficacy, such as bona fide target site-
dependent chromosomal position effects. Consequently,
MN efficacy (i.e. TM, HGT) may result from a com-
bination of ‘intrinsic’ factors (i.e. MN activity, MN
toxicity) and ‘extrinsic’ factors (i.e. MN target locus).
To assess the contribution of each of these distinct
parameters on the overall MN efficacy, we performed a
comprehensive quantitative analysis. We first compared
and/orMN-associatedcellular toxicities.
MN efficacy based on TM and HGT assays and MN
activity based on the extra-chromosomal assay, as
measured in CHO-KI and 293-H cells (Figure 4a).
Overall, MN efficacies based on the frequency of TM
per chromatid correlated poorly with their respective in-
trinsic activities on extrachromosomal substrates (Figure
4a and b), consistent with the low correlation coefficients
and P-values. A somewhat higher correlation was
observed only when comparing MN efficacies (based on
the frequency of TM per chromatid) with activity in
CHO-KI cells (r=0.40 with P=10?2, r=0.41 with
P=8?10?3), and was statistically significant only when
a large sample (n=39) was characterized (compare with
n=18 on Figure 4a), though the correlation coefficient
remained low (r=0.40). Notably, some of the highest
frequencies of indels (as a measure of the frequency of
TM per chromatid) were achieved with DMD21m,
which has relatively moderate activities in the extrachro-
mosomal assay (Figure 4a and Table 1). Similarly, HGT
frequencies were poorly correlated with MN activity in
CHO-KI and 293-H cells (Figure 4c and d). These
modest correlations contrast with the high consistency
observed between the extrachromosomal tests (Figure 2e
and Supplementary Figure S1b) and between the two
types of efficacy tests (Figure 3c and d).
Hence, the intrinsic activity of a given MN on extra-
chromosomal substrates is not predictive for its efficacy on
chromosomal targets. Therefore, we decided to test the
impact of other factors.
MN efficacies are not correlated with their
cellular toxicity
It has been shown before that in a population of cells
treated with a rare cutting endonuclease, the proportion
of cells modified by the nuclease could vary over time,
probably due to nuclease toxicity (34). We actually
observed a similar phenomenon, whereas treatment with
the non-toxic Rag1m molecule results in a stable rate of
?1% of TM, most of the events generated by CLS4076m
at Day 2 (?5% of TM) have disappeared by Day 13
(Supplementary Figure S2a). Thus, cell death resulting
from MN toxicity could impact with apparent efficacy
of MN, especially when measured after day several days.
WeassessedthetoxicityofallMNsfeaturedinthisstudy,
using a cell survival assay that was previously validated
and relies on decreased green fluorescent protein (GFP)
expression as a read-out for cellular toxicity (16,20,35;
Supplementary Figure S2b and S2c). Toxicity was MN
dose-dependent and typically ranged between that of
I-SceI (the most specific reported rare-cutting endonucle-
ase) and I-CreI, with the exception of two MNs (CLS3899
andCLS4633,seeTable1).Importantly,toxicitywasfound
not to be related to the level of activity (r=?0. 15 with
P=0.52; and r=?0.14 with P=0.43).
However, no correlation could be established between
toxicity and efficacy. The estimated values of linear
(Pearson’s r) and non-linear (Spearman, r) correlation
coefficients between TM and toxicity are 0.04 and 0.03,
respectively, with corresponding P-values of 0.81 and 0.86,
respectively. Therefore, we looked for more complex
Transfection
293-H cells
gDNA extraction (day 7)
PCR &
Deep sequencing
frequency of TM per
chromatids (%)
I
I I I I I I
Transfection
293-H cells
Distribution into 96w plates
gDNA extraction (day 21)
PCR
Frequency of HGT per cell (%)
(a)(b)
(c)(d)
% HGT/cell
0.0 0.5 1.0 1.5 2.0 2.5 3.0
N=18
r=0.93; p=2x10-8
rho=0.72; p=6x10-4
N=16
r=0.90; p=3x10-6
rho=0.83; p=7x10-5
% HGT/chromatid
% TM/chromatid
0
5
10
15
20
1
2
3
4
5
6
0
-1
0.0 0.5 1.0 1.5 2.0 2.5 3.0
% TM/chromatid
MN MN
Figure 3. Characterization of MNs efficacy on their chromosomal
cognate target. (a) A chromosomal assay for the characterization of
MN-induced TM. MN expressing vectors are transfected into 293-H
cells; the targeted locus is amplified by PCR and 10000 molecules (on
average) are sequenced to detect indels. (b) A chromosomal test for the
detectionofMN-inducedHGT.
co-transfected into 293-H cells together with a repair matrix; cells are
grown in 96-well plates, and targeted alleles are detected by PCR.
(c) Comparison between the TM and HGT assays in 293-H cells.
Eighteen MNs were characterized in both assays (Table 1). Dotted
grey curves represent 95% confidence interval for the regression line.
Linear regression fits the formula y=5.46x?0.66. r, Pearson coeffi-
cient (linear correlation coefficient); r, Spearman coefficient (non-linear
correlation coefficient); N, sample size; P: probability of finding a given
correlationwhentheunderlying
(d) Comparison between the TM and HGT frequencies measured sim-
ultaneously by deep sequencing. Sixteen independent experiments were
conducted, using different MNs (Rag1m, ADCY9m, CLS3676m,
CLS3759m,CLS3894m,CLS4076m
(SV40or CMV)andvector
Supplementary Table S2. Dotted grey curves represent 95% confidence
interval for the regression line. Linear regression fits the formula
y=1.71x+0.20. r, r, P and N have the same meaning as on the
previous panel.
MN-expressingvectorsare
variablesarenotcorrelated.
andCLS4607m),promoters
describedconcentrations,allin
6374Nucleic Acids Research, 2012,Vol.40, No. 13

Page 9

correlations factors. Supplementary Figure S2d features
an attempt of correlation between toxicity, on the one
hand, and the ratio of the efficacy of TM in 293-H cells
versus the activity in the extrachromosomal assay in
CHO-KI cell (TM/SSA ratio), on the other. Correlation
coefficients are very low, with very high P-values (r=0.16
with P=0.42; and r=0.03 with P=0.86), indicating a
lack of correlation, in spite of the relatively large sample
size (n=39). Fitting of TM as a linear function of toxicity
and SSA (TM=a*SSA+b*TOX+c), lead to similar
results witha=2.38
(P=0.55) and c=7.9385 (P=2?10?4). This outcome
indicates that when measuring the contribution of activity
in CHO-KI cells (SSA) and toxicity (TOX) to efficacy, the
contribution of activity in CHO-KI cell is significant
(P=9?10?3), consistent with the results shown in
Figure 4a for the same sample, but the contribution of
toxicity is not (P=0.55), consistent with the results of
Supplementary Figure S2d.
While toxicity has no direct correlation with the efficacy
of TM, it seems that it might play an important role in the
(P=9?10?3),b=?0.24
dynamic of TM over time. There is a significant cor-
relation between toxicity and the ratio of TM in 293-H
at Day 7 versus Day 2 (r=0.86 P=6?10?3; r=0.9,
P=5?10?3), with more toxic MNs having a bigger
drop in the efficacy of TM from Day 2 to Day 7.
In conclusion, whereas toxicity can impact the apparent
efficacy of MNs, it cannot account for the global
discrepancies observed between the on extrachromosomal
and chromosomal targets, and MN efficacy is not solely
the consequence of its intrinsic properties (activity and
toxicity). Therefore, we decided to test the impact of ex-
trinsic factors, such as chromatin-based position effect.
MN efficacy is strongly correlated with chromosomal
position effects
Chromatin-based position effects could impact on MN
efficacy by modulating access of the MN, the repair
matrix and/or cellular repair factors to the target DNA.
Eukaryotic genomes are packaged in chromatin, and
nuclease sensitivity remains one of the main methods
used to monitor chromatin accessibility and/or the
(d)(c)
N=18
r=0.24; p=0.34
rho= 0.18; p=0.47
N=10
r=-0.07; p=0.84
rho= 0.06; p=0.85
SSA in CHO-KI cells (AUC)
SSA in 293-H cells (AUC)
% HGT/cell
% HGT/cell
0
2
4
6
8
10
12
14
024
0
2
4
6
8
10
12
14
05 1015
% TM/chromatid
SSA in CHO-KI cells (AUC)
(a)
2 4 6 8 10 12 14
N=18
r=0.33; p=0.18
rho= 0.21; p=0.39
N=39
r=0.40; p=10-2
rho= 0.41; p=8x10-3
(b)
0
0.5
1.0
1.5
2.0
2.5
3.0
024
0.5
1.0
1.5
2.0
0.0
% TM/chromatid
SSA in 293-H cells (AUC)
N=15
r=0.09; p=0.73
rho= -0.02; p=0.94
Figure 4. The efficacy of MNs on their chromosomal target is poorly correlated with the activity of the nucleases in extrachromosomal assays.
Throughout the figure, r, Pearson coefficient (linear correlation coefficient); r, Spearman coefficient (non-linear correlation coefficient); N, sample
size; P, probability of finding a given correlation when the underlying variables are not correlated. (a) Comparison between the extrachromosomal
SSA assay in CHO-KI and the TM assay in HEK293 cells. Thirty-seven MNs cleaving 39 targets were tested in both assays (Table 1). Correlation
was also made with a smaller sample of 18 MNs (black circles), which are the same as the 18 MNs characterized in the HGT assay (Figure 3b), and
displayed on Figure 3c. Vertical grey lines represent the activity levels of I-SceIm (left line) and Rag1m (right line). (b) Correlation between the
extrachromosomal SSA assays in 293-H cells and the TM assay in 293-H cells. (c) Correlation between the extrachromosomal SSA assays in CHO
cells and the HGT assay in 293-H cells. (d) Correlation between the extrachromosomal SSA assays in 293-H cells and the HGT assay in 293-H cells.
Nucleic Acids Research, 2012,Vol.40, No. 13 6375

Page 10

(a)
(b)
(c)
(d)
(e)(f)
Figure 5. Impact of position effect on MNs efficacy. (a) The Rag1m MNs as well as Iso1Rag1m and Iso2Rag1m, two isoschizomers were tested for
activity in our extrachromosomal assay in CHO-KI cells, for toxicity, and for their ability to induce targeted insertion at the endogenous Rag1 locus.
Activity, efficacy and toxicity assays were conducted as described in Figures 2 and 3. (b) Similar studies were conducted with the DMD21m MN and
Iso1DMDm and Iso2DMDm cleaving the DMD21t target. (c) The CLS3902m MN cleaves a sequence found in four copies in the human.
CLS3902m was tested for activity, toxicity and efficacy of targeted mutagenesis at three of its cognate targets, CLS3902t_5, CLS3902t_7 and
CLS3902t_14. In A, B and C, each value is the average of two independent experiments but for Rag1m (n=7) and DMD21 (n=3).
(d) CHART-PCR assay for micrococcal nuclease accessibility of the Rag1t, DMD21t, CLS3902t_5, CLS3902t_7, CLS3902t_14, CLS4076t,
CAPNS1t, ADCY9t, CLS3759t and XPC4t loci. The GAPDH and PAX7 loci are used as controls for accessible and non accessible chromatin,
respectively (37). Nuclei were isolated and treated with various amounts of micrococcal nuclease, and for each locus, the amount of non digested
DNA was monitored by Q-PCR as described in materials and methods. Data are normalized to the value obtained in absence of nuclease, and each
6376Nucleic Acids Research, 2012,Vol.40, No. 13
(continued)

Page 11

positioning of nucleosomes in vivo (24,25). Nuclease acces-
sibility is known to vary largely across the genome and at
least in the promoter region, to correlates with a transcrip-
tionally active state (24,36). In order to assess the possible
role of chromatin position effect on genome editing using
MN, we first compared (i) the efficacy of different MNs
(isoschizomers) at the same locus (ii) the efficacy of the
same MN at different loci.
The Rag1m MN cleaves a sequence from the human
Rag1 gene (16). DMD21m cleaves a sequence from the
38th intron of the gene (Table 1). Several isoschizomers
were obtained when we engineered these proteins, and
we compared Rag1m and DMD21m with two additional
isoschizomers. At each individual locus, efficacy was
found to correlate with activity in CHO-KI cells (Figs.
5a and 5b), with the most active cognate MN giving the
highest rates of HGT. However, this correlation was not
found when comparing the results achieved at the two loci.
For example, the iso2DMDm MN is significantly less
active than the iso2RAG1m MN, but has a similar
efficacy. Thus, efficacy is correlated with activity in
CHO-KI cells only when assessed at the ‘same target
locus’. We also characterized CLS3902m, a MN whose
cognate target is found four times in the genome.
CLS3902m has the same activity as Rag1m, and a
slightly higher toxicity (Figure 5c). Examination of TM
at three target loci (the fourth one could not be amplified
by PCR) gave clearly different frequencies of events,
ranging from 0.39% to 1.17% at Day 2 and from 0.13
to 0.75 at Day 7 (Figure 5c), demonstrating locus
dependence.
This type of position effect is likely the consequence of
chromatin-based differences of target accessibility. We
used CHART-PCR (36,37) to test micrococcal nuclease
accessibility at 10 loci targeted by eight engineered MNs
(Figure 5d), and a nuclease resistance index (AUC) was
calculated for each locus, as indicated in figure legend and
‘Materials and Method’ section. CLS4076t and CAPNS1t,
two targets for which we observed very high rates of TM,
were indeedlocatedin
Interestingly, CAPNS1t is in a non-methylated CpG
island in the 50UTR of the CAPNS1 gene, whereas
CLS4076t is close to a CpG island, about 4kb upstream
of the region coding for the FUT8 protein. More gener-
ally, a very strong ‘negative’ correlation was observed
between nuclease resistance and percentage of TM per
chromatid at Day 2 (Figure 5e) and at Day 7
(Figure 5f), fitting with strong linear correlation between
log (TM) andaccessibility
P=7?10?5and r=?0.87 with P=2?10?3, respect-
ively). Thus, target accessibility is a major determinant
of efficacy of rare-cutting designer endonucleases.
veryaccessibleregions.
index (r=?0.93with
DISCUSSION
We describe here the first genome-wide study of a rare-
cutting designer endonuclease platform. We showed that a
wide range of efficacies of TM and HGT (between ?0.1%
and 15%) can be achieved in human cells using MNs. This
provides an effective benchmark for genome editing. In
addition, these comprehensive studies allow us to accur-
ately address several long-standing questions. In particu-
lar, since both TM and HGT depend on cellular DNA
repair mechanisms, it is tempting to hypothesize that
HGT and TM would be correlated at the genome level.
Moreover, the potential impact of chromosomal context
andepigeneticsondesigner
genome editing remains largely unexplored. Our current
study is the first of its kind that specifically addresses the
role of chromosomal context and epigenetics in genome
editing.
Our comprehensive analysis revealed a robust genome-
wide correlationbetween
MN-induced TM (r=0.93; P=2?10?8). Thus, the effi-
ciency of repair of a designer endonuclease-induced
DSB by either NHEJ or homologous recombination es-
sentially depends on extrinsic factors at the cellular level
(although one cannot exclude subtle locus-specific or
MN-specific differences). One of the direct consequences
is that the TM rate, which is relatively easy to monitor, is
predictive of the HGT rate, which monitoring requires the
construction of a repair matrix and a more sophisticated
protocol. Therefore, TM can be used as a surrogate assay
to investigate the ability of a given locus to be edited by
MN-induced HGT. We could actually infer a global rela-
tively constant HGT/TM ratio throughout the genome
close to 1.7 in 293-H cells. This ratio is likely to be
cell-type dependent, particularly since HGT is thought
to be largely dependent on the rate of cell division (38).
It would be interesting to determine whether, the average
HGT/TM ratio is similar or substantially different with
alternative platforms, such as ZFNs or TALENs in the
same cell type.
The availability our comprehensive platform of MN
allowed us to assess the potential impact of chromosomal
context and epigenetics on rare-cutting endonuclease-
mediated genome editing. In particular, it has long been
assumed that the efficacy of these nucleases could be
subject to position effects. We showed here that ‘extrinsic’
factors (i.e. chromosome locus) have a strong impact that
can outweighs that of ‘intrinsic’ factors (i.e. MN activity,
expression, toxicity) on the net MN efficacy. Most import-
antly, molecular analysis shows a very robust correlation
between chromatin accessibility and MN ‘efficacy’, which
appears as a key predictive factor in genome engineering
endonuclease-mediated
MN-inducedHGTand
Figure 5. Continued
point is the average of four measurements from two independent experiments. (e) Comparison between the micrococcal nuclease accessibility assay
and the TM assay at Day 2. The values of panel d were used to construct a curve, and a nuclease resistance index (AUC) was calculated as the AUC,
based on the same principle as activity and cell survival index calculations (Figure 2b and h). Efficacy of TM at Day 2 was plotted against this
resistance index, using a semi-logarithmic scale. Here, r represents the linear correlation coefficient between log(TM) and AUC. r, Pearson coefficient
(linear correlation coefficient); r, Spearman coefficient (non-linear correlation coefficient); N, sample size; P, probability of finding a given correlation
when the underlying variables are not correlated. (f) Comparison between the micrococcal nuclease accessibility assay and the TM assay at Day 7.
Here, r represents the linear correlation coefficient between log(TM) and AUC.
Nucleic Acids Research, 2012,Vol.40, No. 136377

Page 12

experiments. These results should not hide the fact that at
a given individual locus, the most active isoschizomers will
have the highest efficacies, as shown on Figure 5a and b.
However, these results underscore the potential import-
ance of chromosomal context and epigenetic factors in
genome editing using designer rare-cutting endonucleases.
Our results also argue that relatively simple techniques
such as CHART-PCR, or the simple data mining of
existing databases could be used to better predict the
loci allowing for the best efficiencies. The promoters or
50UTR regions of housekeeping genes could be ideal
targets for applications requiring high efficacies, as
illustrated by the high efficacies observed with the
CAPNS1t and CLS4076t targets. Consequently, our
results set the stage for a more rational design of
genome engineering strategies using designer endonucle-
ases and contribute to a better understanding of their
properties at the genome-wide level. It is likely that
chromosomal position effects and epigenetics, particularly
target site accessibility, will have broad implications for
other genome engineering platforms such as those based
on ZFNs and TALENs.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online:
Supplementary Tables 1–6, Supplementary Figures 1–2.
ACKOWLEDGEMENTS
We thank E. Samara and Dr E. Belay for technical
assistance.
FUNDING
Cellectis; an FWO fellowship (Belgium; to M.L.); Free
UniversityofBrusselsand
research Project Funding [FWO N?G.0632.07, in part];
AssociationNationale de
Technologie, contrat [Cifre 535/2008 to A.D.]. Funding
for open access charge: Cellectis.
University ofLeuven;
la rechercheet dela
Conflict of interest statement. All authors but M.L.,
O.D., M.C., T.V., G.D., L.P., and T.K. are employees
of Celectis. F.P. has shares and stock options of Cellectis.
REFERENCES
1. Silva,G., Poirot,L., Galetto,R., Smith,J., Montoya,G.,
Duchateau,P. and Paques,F. (2011) Meganucleases and other
tools for targeted genome engineering: perspectives and challenges
for gene therapy. Curr. Gene Ther., 11, 11–27.
2. Pingoud,A. and Silva,G.H. (2007) Precision genome surgery. Nat.
Biotechnol., 25, 743–744.
3. Pattanayak,V., Ramirez,C.L., Joung,J.K. and Liu,D.R. (2011)
Revealing off-target cleavage specificities of zinc-finger nucleases
by in vitro selection. Nat. Methods, 8, 765–770.
4. Urnov,F.D., Rebar,E.J., Holmes,M.C., Zhang,H.S. and
Gregory,P.D. (2010) Genome editing with engineered zinc finger
nucleases. Nat. Rev. Genet., 11, 636–646.
5. Lombardo,A., Cesana,D., Genovese,P., Di Stefano,B., Provasi,E.,
Colombo,D.F., Neri,M., Magnani,Z., Cantore,A., Lo Riso,P.
et al. (2011) Site-specific integration and tailoring of cassette
design for sustainable gene transfer. Nat. Methods, 8, 861–869.
6. Pa ˆ ques,F. and Duchateau,P. (2007) Meganucleases and DNA
double-strand break-induced recombination: perspectives for gene
therapy. Curr. Gene Ther., 7, 49–66.
7. Stoddard,B.L. (2011) Homing endonucleases: from microbial
genetic invaders to reagents for targeted DNA modification.
Structure, 19, 7–15.
8. McNeer,N.A., Chin,J.Y., Schleifman,E.B., Fields,R.J.,
Glazer,P.M. and Saltzman,W.M. (2010) Nanoparticles deliver
triplex-forming PNAs for site-specific genomic recombination in
CD34+ human hematopoietic progenitors. Mol. Ther., 19,
172–180.
9. Eisenschmidt,K., Lanio,T., Simoncsits,A., Jeltsch,A., Pingoud,V.,
Wende,W. and Pingoud,A. (2005) Developing a programmed
restriction endonuclease for highly specific DNA cleavage. Nucleic
Acids Res., 33, 7039–7047.
10. Christian,M., Cermak,T., Doyle,E.L., Schmidt,C., Zhang,F.,
Hummel,A., Bogdanove,A.J. and Voytas,D.F. (2010) Targeting
DNA double-strand breaks with TAL effector nucleases. Genetics,
186, 757–761.
11. Miller,J.C., Tan,S., Qiao,G., Barlow,K.A., Wang,J., Xia,D.F.,
Meng,X., Paschon,D.E., Leung,E., Hinkley,S.J. et al. (2011) A
TALE nuclease architecture for efficient genome editing. Nat.
Biotechnol., 29, 143–148.
12. Mussolino,C., Morbitzer,R., Lu ¨ tge,F., Dannemann,N., Lahaye,T.
and Cathomen,T. (2011) A novel TALE nuclease scaffold enables
high genome editing activity in combination with low toxicity.
Nucleic Acids Res., 39, 9283–9293.
13. 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.
14. 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.
15. Smith,J., Grizot,S., Arnould,S., Duclert,A., Epinat,J.C., Prieto,J.,
Redondo,P., Blanco,F.J., Bravo,J., Montoya,G. et al. (2006) A
combinatorial approach to create artificial homing endonucleases
cleaving chosen sequences. Nucleic Acids Res., 34, e149.
16. Grizot,S., Smith,J., Daboussi,F., Prieto,J., Redondo,P.,
Merino,N., Villate,M., Thomas,S., Lemaire,L., Montoya,G. et al.
(2009) Efficient targeting of a SCID gene by an engineered
single-chain homing endonuclease. Nucleic Acids Res., 37,
5405–5419.
17. Zou,J., Maeder,M.L., Mali,P., Pruett-Miller,S.M., Thibodeau-
Beganny,S., Chou,B.K., Chen,G., Ye,Z., Park,I.H., Daley,G.Q.
et al. (2009) Gene targeting of a disease-related gene in human
induced pluripotent stem and embryonic stem cells. Cell Stem
Cell, 5, 97–110.
18. Lombardo,A., Genovese,P., Beausejour,C.M., Colleoni,S.,
Lee,Y.L., Kim,K.A., Ando,D., Urnov,F.D., Galli,C.,
Gregory,P.D. et al. (2007) Gene editing in human stem cells using
zinc finger nucleases and integrase-defective lentiviral vector
delivery. Nat. Biotechnol., 25, 1298–1306.
19. Shukla,V.K., Doyon,Y., Miller,J.C., DeKelver,R.C., Moehle,E.A.,
Worden,S.E., Mitchell,J.C., Arnold,N.L., Gopalan,S., Meng,X.
et al. (2009) Precise genome modification in the crop species Zea
mays using zinc-finger nucleases. Nature, 459, 437–441.
20. Maeder,M.L., Thibodeau-Beganny,S., Osiak,A., Wright,D.A.,
Anthony,R.M., Eichtinger,M., Jiang,T., Foley,J.E., Winfrey,R.J.,
Townsend,J.A. et al. (2008) Rapid ‘‘open-source’’ engineering of
customized zinc-finger nucleases for highly efficient gene
modification. Mol. Cell, 31, 294–301.
21. Townsend,J.A., Wright,D.A., Winfrey,R.J., Fu,F., Maeder,M.L.,
Joung,J.K. and Voytas,D.F. (2009) High-frequency modification
of plant genes using engineered zinc-finger nucleases. Nature, 459,
442–445.
22. Liang,F., Han,M., Romanienko,P.J. and Jasin,M. (1998)
Homology-directed repair is a major double-strand break repair
pathway in mammalian cells. Proc. Natl Acad. Sci. USA, 95,
5172–5177.
6378 Nucleic Acids Research, 2012,Vol.40, No. 13

Page 13

23. Perez,E.E., Wang,J., Miller,J.C., Jouvenot,Y., Kim,K.A., Liu,O.,
Wang,N., Lee,G., Bartsevich,V.V., Lee,Y.L. et al. (2008)
Establishment of HIV-1 resistance in CD4+ T cells by genome
editing using zinc-finger nucleases. Nat. Biotechnol., 26, 808–816.
24. Rigaud,G., Roux,J., Pictet,R. and Grange,T. (1991) In vivo
footprinting of rat TAT gene: dynamic interplay between the
glucocorticoid receptor and a liver-specific factor. Cell, 67, 977–986.
25. Carey,M.F., Peterson,C.L. and Smale,S.T. (2009) In vivo DNase
I, MNase, and restriction enzyme footprinting via
ligation-mediated polymerase chain reaction (LM-PCR). Cold
Spring Harb. Protoc., 2009, doi:10.1101/pdb.prot5277.
26. Misteli,T. and Soutoglou,E. (2009) The emerging role of nuclear
architecture in DNA repair and genome maintenance. Nat. Rev.
Mol. Cell. Biol., 10, 243–254.
27. Smith,J., Grizot,S., Arnould,S., Duclert,A., Epinat,J.C., Chames,P.,
Prieto,J., Redondo,P., Blanco,F.J., Bravo,J. et al. (2006) A
combinatorial approach to create artificial homing endonucleases
cleaving chosen sequences. Nucleic Acids Res., 34, e149.
28. Arnould,S., Perez,C., Cabaniols,J.P., Smith,J., Gouble,A.,
Grizot,S., Epinat,J.C., Duclert,A., Duchateau,P. and Paques,F.
(2007) Engineered I-CreI derivatives cleaving sequences from the
human XPC gene can induce highly efficient gene correction in
mammalian cells. J. Mol. Biol., 371, 49–65.
29. Grizot,S., Duclert,A., Thomas,S., Duchateau,P. and Paques,F.
(2011) Context dependence between subdomains in the DNA
binding interface of the I-CreI homing endonuclease. Nucleic
Acids Res., 39, 6124–6136.
30. Sen,A.K. and Srivastava,M.S. (1990) Regression Analysis: Theory,
Methods and Applications. Springer, New-York.
31. Arnould,S., Chames,P., Perez,C., Lacroix,E., Duclert,A.,
Epinat,J.C., Stricher,F., Petit,A.S., Patin,A., Guillier,S. et al.
(2006) Engineering of large numbers of highly specific homing
endonucleases that induce recombination on novel DNA targets.
J. Mol. Biol., 355, 443–458.
32. Galetto,R., Duchateau,P. and Paques,F. (2009) Targeted
approaches for gene therapy and the emergence of engineered
meganucleases. Expert Opin. Biol. Ther., 9, 1289–1303.
33. Redondo,P., Prieto,J., Munoz,I.G., Alibes,A., Stricher,F.,
Serrano,L., Cabaniols,J.P., Daboussi,F., Arnould,S., Perez,C.
et al. (2008) Molecular basis of xeroderma pigmentosum group C
DNA recognition by engineered meganucleases. Nature, 456,
107–111.
34. Porteus,M.H. and Baltimore,D. (2003) Chimeric nucleases
stimulate gene targeting in human cells. Science, 300, 763.
35. Cornu,T.I., Thibodeau-Beganny,S., Guhl,E., Alwin,S.,
Eichtinger,M., Joung,J.K. and Cathomen,T. (2008) DNA-binding
specificity is a major determinant of the activity and toxicity of
zinc-finger nucleases. Mol. Ther., 16, 352–358.
36. Rao,S., Procko,E. and Shannon,M.F.J. (2001) Chromatin
remodeling, measured by a novel real-time polymerase chain
reaction assay, across the proximal promoter region of the IL-2
gene. Immunol., 167, 4494–4503.
37. Cruickshank,M., Fenwick,E., Abraham,L.J. and Ulgiati,D. (2008)
Quantitative differences in chromatin accessibility across
regulatory regions can be directly compared in distinct cell-types.
Biochem. Biophys. Res. Commun., 367, 349–355.
38. Delacote,F. and Lopez,B.S. (2008) Importance of the cell cycle
phase for the choice of the appropriate DSB repair pathway, for
genome stability maintenance: the trans-S double-strand break
repair model. Cell Cycle, 7, 33–38.
Nucleic Acids Research, 2012,Vol.40, No. 136379