CRISPRCas9:

Engineering
a Revolution
in Gene
Editing

Sponsored by

precision genome editing

Produced by the
Science/AAAS
Custom Publishing
Office

JOIN AAAS

TA B L E O F C O N T E N T S

Get instant access to Science. Support all of the sciences.

Introductions
2 The start of a new genomic era

Tianna Hicklin, Ph.D.

3 CRISPR: The democratization of gene editing

Feng Zhang, Ph.D.

Editorial
4 The power and possibilities of genome engineering
Jeffrey M. Perkel, Ph.D.

CRISPRCas9:
Engineering
a Revolution
in Gene
Editing

Science Articles
7 Genetic screens in human cells using the CRISPRCas9 system

Tim Wang, et al.

Since

1880

11 Genome-scale CRISPR-Cas9 knockout screening in

human cells

Ophir Shalem, et al.

15 The CRISPR craze

Elizabeth Pennisi

18 Multiplex genome engineering using CRISPR/Cas

The American Association for the Advancement of Science (AAAS) is a non-profit community that is open to everyone, from
Nobel laureates to high school students. Ours is a global membership of over 120,000 people who believe in the power of
science to make the world a better place.
From the moment you join, you get immediate access to everything that AAASs award-winning journal Science has to
offer, including:
51 weeks of home delivery of Science;
Instant online retrieval of every Science article ever published, from today, dating back to 1880;
Anytime, anywhere access via the Science mobile site and apps for Android, iPad, and iPhone devices;
Members-only newsletters; and more.
As a member, you are also making a critical contribution to AAASs efforts to provide a public voice for all of science.
With public skepticism about science increasing, and public funding for research more uncertain than ever, this work has
never been more important.
AAAS is hard at work promoting science in government offices, in schools, and in the public commons all around the
worldwith programs like AAAS Senior Scientists and Engineers, which brings volunteer scientists into public school
classrooms, or our sweeping petition drives calling for the preservation of federal R&D funding.

systems

Le Cong, et al.

22 RNA-guided human genome engineering via Cas9

Prashant Mali, et al.

25 A Swiss army knife of immunity

Stan J. J. Brouns

26 CRISPR/Cas, the immune system of bacteria and archaea

Philippe Horvath, et al.

30 A programmable dual-RNAguided DNA endonuclease

in adaptive bacterial immunity (abstract only)

Martin Jinek, et al.

Cover image provided courtesy of Horizon Discovery. This image has been digitally altered from the
original version for publication purposes.
Editors: Tianna Hicklin, Ph.D. and Sean Sanders,
Ph.D.; Designer: Amy Hardcastle; Copyeditor: Yuse
Lajiminmuhip
This booklet was produced by the Science/AAAS
Custom Publishing Office and supported by
Horizon Discovery.
2014 by The American Association for the
Advancement of Science. All rights reserved.
26 September 2014

Visit promo.aaas.org/joinaaas and join today. Together we can make a difference.

SCIENCE sciencemag.org

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

I N T RO D U C T I O N S

H
The start of a
new genomic
era

Science has been

at the forefront of
publishing some of
the groundbreaking
work as scientists
have begun to
unravel the CRISPRCas9 system.

umans are an exceptionally adaptive speciesa characteristic that has

enabled us to flourish all over the planet. We have adapted genetically
and epigenetically to many different climates and habitats, and these
adaptive mutations have been passed down to subsequent generations.
However, these are not the only mechanisms at play; we also have thrived by modifying our surroundings, passing this information on to the next generation so that it
can be built upon, refined, and improved. Now, technological advances in genomic
engineering hold the potential to give us the key to not only modifying our external
environment, but to also engineering genetic adaptations for ourselves as well as
other species.
Starting with the discovery of mysterious palindromic, repeated DNA sequences
in E. Coli in 1987, scientists began investigating the function of this seemingly odd
phenomenon. Out of this natural curiosity grew an entirely new way to modify DNA:
CRISPR-Cas9. The system evolved as a self-defense mechanism for bacteriaessentially, a way to self-vaccinate against invading viruses and plasmids (Science
23 March 2007, p. 1709, scim.ag/1oSz2rE). This adaptive immunity, has enabled
bacteria and archaea to continually arm and rearm themselves against invaders.
However, now that scientists have taken the reins, the CRISPR-Cas9 system has been
retooled into a more globally viable technology, whereby the genetic code of virtually any species can be modified, and for more than simply self-protection.
Named as a runner up for Science Magazines Breakthroughs of the Year in 2013
(Science 20 December 2013, p. 1434; scim.ag/1uykTTC), the CRISPR-Cas9 system
is revolutionizing genomic engineering and equipping scientists with the ability to
precisely modify the DNA of essentially any organism. This gene editing could potentially confer genetic advantages that previously took large amounts of evolutionary time (and perhaps a bit of luck), taxing genetic breeding strategies, or bulkier
and more complex genomic editing tools to acquire. Speculations about whats on
the horizon seem to be limitless at this point.
Just how powerful is this technique? The ability for precision genome engineering
comes with the potential to enhance food production, medicinal discoveries, and
energy solutions, to name a few. Studies over the past several years have shown
promise for altering crop resistance to infection and disease, advancing drug discovery, modifying fuel/energy sources, and even elucidating the multiple genetic
contributions underlying human diseasesfrom heart disease to mental illnesses.
Science has been at the forefront of publishing some of the groundbreaking work
as scientists have begun to unravel the CRISPR-Cas9 system and invent novel ways
to use this tool. Two such papers were the seminal work of Feng Zhang and George
Church, which were simultaneously published and the first studies to show that the
CRISPR-Cas9 system can alter mammalian genomes, including humans (see pages
20, 24). Since that time, a CRISPR Craze has begun (see page 17), whereby scientists have begun exploring more and more ingenious ways to use this technology.
Researchers have already progressed from studies on individual DNA alterations
into studies using CRISPR-Cas9 as an efficient, large-scale, loss-of-function screening method in mammalian cells, in lieu of using RNAi screens (see pages 9, 13).
Of course, it is hoped that these types of studies are just the tip of the iceberg for
this new era of precision genetic engineering. In this booklet, we invite the reader
to explore a selection of articles that highlight both the history of this technique and
how it has grown into one of the most powerful and precise genomic engineering
tools to date.

R
CRISPR: The
democratization
of gene editing
The elegance and
simplicity of Cas9
have sparked the
imagination of
scientists across
many scientific
disciplines.

Tianna Hicklin, Ph.D.

Editor, Science/AAAS Custom Publishing Office

ecent development of genome editing technologies based on the RNAguided CRISPR-associated endonuclease Cas9 has generated enormous
excitement across many fields, including biological research, biotechnology, and medicine. For the first time, researchers have gained the ability to
achieve targeted genomic modifications with efficiency and ease. This is particularly
true when combined with the rapidly increasing amount of information available
from genomic sequencing efforts available as well as innovative nucleic acid synthesis and delivery systems.
Unlike previous generations of genome editing tools based on zinc finger and
transcription activator-like effector proteins, which achieve sequence recognition
via protein-DNA interactions, Cas9 can be targeted to specific genomic loci with a
guide RNA (gRNA). Once the Cas9-gRNA complex finds the DNA target via WatsonCrick base pairing, Cas9 introduces a double-strand DNA break at the target site,
which in turn catalyzes targeted genome editing via non-homologous end joining
or homology directed repair. The ability to use nucleic-acid hybridization rules to
reprogram Cas9 specificity significantly simplifies genome editing applications
particularly given that gRNAs are easily synthesized and introduced into cells to
facilitate targeted genome modifications.
Despite being a nascent technology, Cas9 has been successfully used to
generate an increasing number of cellular and animal models for a variety of basic
research as well as biotechnology applications. For example, Cas9 can facilitate the
generation of isogenic cell lines to identify causal genetic variations. In addition,
Cas9 has already been broadly applied in many species to generate transgenic
models, including mouse, rat, zebrafish, fruit flies, C. elegans, primates, and a
variety of plant species. For each species, gRNAs can be easily designed based
on reference genome sequences to target virtually any locus of choice. Moreover,
direct application of Cas9 in embryos can significantly accelerate transgenic
manipulation of whole organisms, including many previously intractable species.
In addition to facilitating the editing of individual genes, Cas9 can also be used for
high throughput genetic screening applications where many genes are perturbed
in a multiplexed fashion. Large libraries of gRNAs capable of targeting wide
ranges or subsets of genes in the human genome can be easily synthesized using
oligo array synthesis and used to generate genome-scale gene knockout libraries
for functional screening. For example, these Cas9 gRNA libraries have already
been successfully used in a number of studies to identify drug resistance genes.
In the future, Cas9 gRNA libraries may also be used to facilitate high throughput
investigation of non-coding sequences by disrupting regulatory regions in the
genome.
Cas9 can also be converted into a catalytically inactive RNA-guided DNA binding
protein, which can be used to recruit transcription effector domains to specific
genomic loci to modulate transcription state. Alternatively, catalytically inactive Cas9
can also be linked to fluorescent proteins to enable direct visualization of genomic
loci in living cells.
The pace of Cas9 development is only accelerating. Researchers are rapidly
enhancing the functionalities of Cas9, making it more specific, efficient, and
easier to use in a variety of biological contexts. At the same time, the elegance
and simplicity of Cas9 have sparked the imagination of scientists across many
scientific disciplines, with many of them already using the technology to uncover
fundamental biological processes and develop innovative therapeutic strategies
for treating intractable diseases. With all of these developments, the road ahead is
undoubtedly full of exciting possibilities.
Dr. Feng Zhang, Ph.D., is currently serving as the W.M. Keck Career Development
Professor with a joint appointment in the Biological Engineering and Brain and
Cognitive Sciences Departments at the Massachusetts Institute of Technology (MIT),
is one of the 11 core members of the Broad Institute of MIT and Harvard and is an
Investigator at the McGovern Institute for Brain Research.

sciencemag.org SCIENCE

SCIENCE sciencemag.org

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N G E N E E D I T I N G

E D I T O R I A L | T H E P OW E R A N D P O S S I B I L I T I E S O F G E N O M E E N G I N E E R I N G

EDI TOR IAL

The power and

possibilities
of genome
engineering
By Jeffrey M. Perkel

ver since scientists understood that DNA

carried heritable information, they have desired
to bend this code to their will. Manipulating the
fundamental code for life would mean the ability
to correct defects and permanently cure genetic
disorders. But the tools that have been available up
until now, although workable, are crude at bestlike
trying to perform surgery while wearing mittens.
These days, though, the metaphorical mittens are
off. Over the past decade, researchers have devised
a succession of strategies that can handle just
about any genetic rewrites they can imagine. These
genome-editing technologies arent perfect, and
they are typically limited to one or just a few changes
at a time. But they do fill a crucial hole in biologists
toolboxes. In so doing, they are redefining the
genetic frontier, and redefining it for the better.
Genome editing technologies exist in a
handful of basic forms, one of which is based
on homologous recombination. In general,
homologous recombination rates are too low in
mammalian cells to make this approach practical
(a key exception being mouse embryonic stem
cells, which is why mouse transgenic technology
has been so successful). By inserting the repair
template into a single-stranded recombinant adenoassociated viral (rAAV) backbone however, efficiency
improves considerably to the point that the
approach is workable to generate precise genomic
modifications.

Most genome-editing strategies, however, rely on

the concept of a customized DNA-cutting endonuclease. Several variants of this approach have been
developed, but in all these cases, the idea is the
same: To generate a double-stranded DNA break at
a specified location in the genome of a live cell.
In responding to that break, the cell may
successfully stitch the two ends together (i.e., no
change to the sequence occurs); it may inadvertently
disable the gene by non-homologous end joining
(NHEJ)-based frame-shifts; or it may repair the
gene via homology-directed repair (HDR), a process
that researchers can game by supplying their own
templates to repair a point mutation or insert a
missing gene.
That template molecule typically is supplied either
in the form of a short oligonucleotide or doublestranded DNA plasmid. A newer strategy uses a
rAAV vector instead, which combines the enhanced
nuclear uptake of a single-stranded molecule with
the greater donor length of plasmids.
In its original implementation, genome editing
was accomplished using meganucleases, a class of
natural nucleases that targets relatively long DNA
recognition sequences. But these enzymes proved
difficult to tailor, requiring either mutagenesis or sophisticated protein engineering.
Zinc finger nucleases (ZFNs) introduced the
concept of programmability to genome editing.
ZFNs fuse the DNA-binding domains of zinc finger

sciencemag.org SCIENCE

transcription factors to a generic DNA endonuclease

to induce a double-stranded DNA break at a defined
position. Programming is achieved by selecting the
identity and order of fingers used.
In a zinc finger transcription factor, each finger
recognizes a specific three- or four-base sequence.
By simply stringing together an appropriate array of fingers, researchers can in theory target any
sequence they desire, and researchers have been
successful at doing precisely that. Sangamo BioSciences, for instance, has demonstrated it can safely
use ZFN technology in human patients. In a study
published earlier this year in the New England Journal of Medicine, the first ever to document genomeengineering in the clinic, researchers knocked out
the gene for the HIV co-receptor, CCR5, in T cells
from HIV+ individuals, and then safely returned those
cells to the patients, raising T cell counts overall and
CD4 counts in particular (1).
Still, relatively few researchers have hitched their
wagon to ZFN technology, as making high-quality
ZFNs takes considerable skill. One key problem:
Individual fingers dont always function as expected
in the context of an intact ZFN. Optimizationand
therefore additional time and moneyis often
required.
Another class of artificial enzymes, the transcription activator-like (TAL) effector nucleases (TALENs),
deliver similar benefits to ZFNs yet generally are
easier to use. Like ZFNs, TALENs are modular transcription factors, with each so-called TAL module
specifying a single base in the recognition sequence.
As TAL modules function more or less like reusable
building blocks, TALENs are far simpler to design
than ZFNs and can be assembled from kits relatively
quickly. Yet their size and repetitive nature can pose
challenges for the uninitiated.
The newest genome-editing strategy is the socalled CRISPR-Cas9 system. CRISPR-Cas9 is not an
artificial construction; bacteria use it as a primitive
form of adaptive anti-viral immunity. When infected
with a pathogen, bacteria endowed with this system retain a signature of the infecting agent in their
chromosomal DNA. The cell then transcribes those
sequencescalled CRISPRs, or clustered regularly
interspaced short palindromic repeatsand processes them into short RNAs called crRNAs. With
a second transcript called a tracrRNA, the crRNA
guides the Cas9 nuclease to its targetnormally, an
invading viral nucleic acid.
In a seminal 2012 article in Science, Jennifer

SCIENCE sciencemag.org

Doudna and Emmanuelle Charpentier worked out

the mechanism of this process, and showed that it
could be both reprogrammed and simplified into
a two-component system in vitro (2). Subsequent
studies showed that the system could be harnessed
to hit genomic sequences in bacteria, fruit flies,
nematodes, and human cells.
CRISPR-Cas9 offers
CRISPR-Cas9
several key benefits over
competing endonuclease
is also relatively
technologies. First, while
efficient,
meganucleases, ZFNs, and
TALENs can be thought of
editing target
as bespoke single-function
sequences at
machines, Cas9 is basically a
programmable enzyme. All
surprisingly
that is required is a construct
high rates.
expressing the generic
Cas9 nuclease and a set
of instructions in the form of a single-guide RNA
(sgRNA) complementary to the desired target. The
system is simple and inexpensive to implement and
thus more attractive to researchers who might have
been skittish of ZFNs and TALENs.
The second benefit is multiplexing. Since Cas9
is guided by its sgRNA, researchers can program
it with multiple guide RNAs simultaneously. Feng
Zhang and George Church, writing independently in
Science, have both demonstrated the ability to target
two sites simultaneously (3, 4), and Rudolf Jaenisch
has targeted five (5).
CRISPR-Cas9 is also relatively efficient, editing
target sequences at surprisingly high rates. As
a general rule, HDR-mediated genome editing
typically occurs at much lower frequencies than
NHEJ, requiring as it does a second piece of DNA
containing the repair template. Similarly, it is easier
to hit one allele than two. Yet Jaenisch found that
20 of 96 mouse embryonic stem cell clones tested
using three sgRNAs simultaneously contained NHEJinduced mutations at all six alleles of those three
genes, a 20% success rate (5). Church observed
HDR-mediated repair rates in human cells of 3% and
8% using two separate sgRNAs, compared to about
0.5% with a TALEN directed at the same location (4).
Thats not to say CRISPR-Cas9 is perfect. Multiple
studies have documented off-site targeting when
using the system, for instance, at least in its original
incarnationsomething that could significantly limit
potential clinical applications. Researchers have
developed strategies to boost targeting specificity,

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | R E P O RT

RE SE A RCH
including both paired nickase and FokI-fusion
approaches that require two closely spaced sgRNA
binding events for cleavage, but whether they will
be sufficient to make CRISPR-Cas9 clinically useful
remains an open question.
Still, theres little doubt
the system will be useful in
One the research lab. Indeed,
particularly researchers are beginning
to explore its application to
promising more sophisticated genomic
application operations. Echoing work
previously done with ZFNs
of genome and TALENs, for instance,
editing marries they have made surgical
changes in gene expression
its power (as opposed to sequence)
with induced by coupling the catalytically inactive Cas9 enzyme
pluripotent to transcriptional regulatory
stem cell (iPS) domains and illuminated
chromosome structure by
technology. coupling it to fluorescent
proteins. George Church
has experimented with orthologous Cas9 proteins
to expand the technologys reach even further, for
instance by allowing combined genome editing and
transcriptional regulation (6).
Indeed, researchers today are extending the
CRISPR-Cas9 system on multiple fronts. Teams led
independently by Feng Zhang and Eric Lander,
writing in Science, recently demonstrated the
utility of the system for large-scale screening,
much as researchers did a decade ago using RNA
interference (7, 8). Another team used the system to
build a set of biological circuits for synthetic biology
applications (9). And still other researchers are
exploring the clinical applications of the technology,
through startup firms such as Editas Medicine and
CRISPR Therapeutics.
One recent study, led by Daniel Anderson at Massachusetts Institute of Technology (MIT), illustrates
the potential power of CRISPR-Cas9 in the clinic. Andersons team co-delivered an sgRNA, Cas9, and a
199-base single-stranded donor template molecule
into the tail vein of a mouse model of a genetic disease called hereditary tyrosinemia type I, caused by
a single point mutation in the gene for fumarylacetoacetate hydrolase. Though such a strategy cannot
directly be implemented in humans, the treatment

did repair the mutation in one in every 250 liver cells

in the treated animals, minimizing the liver damage
and weight loss typically seen in this model (10).
One particularly promising application of genome
editing marries its power with induced pluripotent
stem cell (iPS) technology. iPS cells, in which researchers turn a somatic cell into an embryonic-like
pluripotent stem cell using four transcription factors
and/or small molecules, are opening the door not
only to new studies of disease etiology, but corrective cellular therapies as well, as evidenced by a new
iPS cell-based clinical study launched recently in
Japan.
That study hopes to differentiate patient-specific
iPS cells into retinal pigment epithelial cells for the
treatment of age-related macular degeneration. But
combined with genome-editing technology, iPS cells
can do much, much morereconstituting a sickle-cell
anemia patients bone marrow with hematopoietic
stem cells in which the mutant hemoglobin gene has
been repaired, for instance. Sangamo BioSciences
demonstrated recently that it could use ZFNs to
drive HDR-based repair of the IL2R-gamma gene in
human hematopoietic stem cells, but that work was
not done in a clinical setting (11).
Where ZFNs, or any genome editing technology
for that matter, will go from here is an open question.
But given the pace of research and the steady stream
of promising findings so far, its a good bet that its
going to be an exciting ride.
References
1. P. Tebas et al., NEJM 370, 901 (2014).
2. M. Jinek et al., Science 337, 816 (2012).
3. L. Cong et al., Science 339, 819 (2013).
4. P. Mali et al., Science 339, 823 (2013).
5. H. Wang et al., Cell 153, 910 (2013).
6. K.M. Esvelt et al., Nat. Methods 10, 1116 (2013).
7. O. Shalem et al., Science 343, 84 (2014).
8. T. Wang et al., Science 343, 80 (2014).
9. S. Kiani et al., Nat. Methods 11, 723 (2014).
10. H. Yin et al., Nat. Biotechnol. 32, 551 (2014).
11. P. Genovese et al., Nature 510, 235 (2014);
10.1038/nature13420.

Jeffrey M. Perkel is a freelance science writer based

in Pocatello, Idaho.

sciencemag.org SCIENCE

Genetic screens in human cells using

the CRISPR-Cas9 system
Tim Wang,1,2,3,4Jenny J. Wei,1,2 David M. Sabatini,1,2,3,4,5* Eric S. Lander1,3,6*
The bacterial clustered regularly interspaced short palindromic repeats (CRISPR)Cas9
system for genome editing has greatly expanded the toolbox for mammalian genetics,
enabling the rapid generation of isogenic cell lines and mice with modified alleles. Here,
we describe a pooled, loss-of-function genetic screening approach suitable for both
positive and negative selection that uses a genome-scale lentiviral single-guide RNA
(sgRNA) library. sgRNA expression cassettes were stably integrated into the genome,
which enabled a complex mutant pool to be tracked by massively parallel sequencing.
We used a library containing 73,000 sgRNAs to generate knockout collections and
performed screens in two human cell lines. A screen for resistance to the nucleotide
analog 6-thioguanine identified all expected members of the DNA mismatch repair
pathway, whereas another for the DNA topoisomerase II (TOP2A) poison etoposide
identified TOP2A, as expected, and also cyclin-dependent kinase 6, CDK6. A negative
selection screen for essential genes identified numerous gene sets corresponding to
fundamental processes. Last, we show that sgRNA efficiency is associated with specific
sequence motifs, enabling the prediction of more effective sgRNAs. Collectively, these
results establish Cas9/sgRNA screens as a powerful tool for systematic genetic analysis
in mammalian cells.

critical need in biology is the ability to efficiently identify the set of

genes underlying a cellular process.
In microorganisms, powerful methods
allow systematic loss-of-function genetic screening (1, 2). In mammalian cells,
however, current screening methods fall
shortprimarily because of the difficulty of
inactivating both copies of a gene in a diploid
mammalian cell. Insertional mutagenesis
screens in cell lines that are near-haploid or
carry Blm mutations, which cause frequent
somatic crossing-over, have proven powerful
but are not applicable to most cell lines and
suffer from integration biases of the insertion vectors (3, 4). The primary solution has
been to target mRNAs with RNA interference (RNAi) (59). However, this approach
is also imperfect because it only partially
suppresses target gene levels and can have
off-target effects on other mRNAs, resulting
in false negative and false positive results
(1012). Thus, there remains an unmet need
for an efficient, large-scale, loss-of-function
Department of Biology, Massachusetts Institute of Technology
(MIT), Cambridge, MA 02139, USA.
2
Whitehead Institute for Biomedical Research, 9 Cambridge
Center, Cambridge, MA 02142, USA.
3
Broad Institute of MIT and Harvard, 7 Cambridge Center,
Cambridge, MA 02142, USA.
4
David H. Koch Institute for Integrative Cancer Research at MIT,
Cambridge, MA 02139, USA.
5
Howard Hughes Medical Institute, Department of Biology, MIT,
Cambridge, MA 02139, USA.
6
Department of Systems Biology, Harvard Medical School,
Boston, MA 02115, USA.
These authors contributed equally to this work.
*Corresponding author. E-mail: sabatini@wi.mit.edu
(D.M.S.); lander@broadinstitute.org (E.S.L.)
1

SCIENCE sciencemag.org

screening method in mammalian cells.

Recently, the clustered regularly interspaced short palindromic repeats (CRISPR)
pathway, which functions as an adaptive
immune system in bacteria (13), has been
co-opted to engineer mammalian genomes
in an efficient manner (1416). In this twocomponent system, a single-guide RNA
(sgRNA) directs the Cas9 nuclease to cause
double-stranded cleavage of matching target
DNA sequences (17). In contrast to previous
genome-editing techniques, such as zincfinger nucleases and transcription activatorlike effector nucleases (TALENs), the target
specificity of CRISPR-Cas9 is dictated by a
20base pair (bp) sequence at the 5 end of
the sgRNA, allowing for much greater ease
of construction of knockout reagents. Mutant
cells lines and mice bearing multiple modified alleles can be generated with this technology (18, 19).
We set out to explore the feasibility of using the CRISPR-Cas9 system to perform largescale, loss-of-function screens in mammalian
cells. The idea was to use a pool of sgRNAexpressing lentivirus to generate a library of
knockout cells that could be screened under
both positive and negative selection. Each
sgRNA would serve as a distinct DNA barcode
that can be used to count the number of cells
carrying it by using high-throughput sequencing (Fig. 1A). Pooled screening requires that
single-copy sgRNA integrants are sufficient
to induce efficient cleavage of both copies of
a targeted locus. This contrasts with the high
expression of sgRNAs achieved through transfection that is typically used to engineer a
Originally published 3 January 2014

specific genomic change by using the CRISPRCas9 system.

We first tested the concept in the near-haploid, human KBM7 CML cell line by creating
a clonal derivative expressing the Cas9 nuclease (with a FLAG-tag at its N terminus) under
a doxycycline-inducible promoter (Fig. 1B).
Transduction of these cells at low multiplicity
of infection (MOI) with a lentivirus expressing
a sgRNA targeting the endogenous AAVS1 locus revealed substantial cleavage at the AAVS1
locus 48 hours after infection (Fig. 1C). Moreover, because the sgRNA was stably expressed,
genomic cleavage continued to increase over
the course of the experiment. Deep sequencing of the locus revealed that repair of Cas9-induced double-strand breaks resulted in small
deletions (<20 bp) in the target sequence, with
tiny insertions or substitutions (<3 bp) occurring at a lower frequency (Fig. 1D). The vast
majority of the lesions, occurring in a proteincoding region, would be predicted to give rise
to a nonfunctional protein product, indicating
that CRISPR-Cas9 is an efficient means of generating loss-of-function alleles.
We also analyzed off-target activity of
CRISPR-Cas9. Although the specificity of
CRISPR-Cas9 has been extensively characterized in transfection-based settings (2022),
we wanted to examine its off-target behavior
in our system, in which Cas9 and a sgRNA targeting AAVS1 (sgAAVS1) were stably expressed
for 2 weeks. We compared the level of cleavage
observed at the target locus (97%) with levels
at 13 potential off-target cleavage sites in the
genome (defined as sites differing by up to 3
bp from sgAAVS1) (fig. S1A). Minimal cleavage (<2.5%) was observed at all sites, with one
exception, which was the only site that had
perfect complementarity in the seed region
(terminal 8 bp) (fig. S1B). On average, sgRNAs
have ~2.2 such sites in the genome, almost always (as in this case) occurring in noncoding
DNA and thus less likely to affect gene function (supplementary text S1).
To test the ability to simultaneously screen
tens of thousands of sgRNAs, we designed a
sgRNA library with 73,151 members, consisting of multiple sgRNAs targeting 7114 genes
and 100 nontargeting controls (Fig. 1E, table
S1, and supplementary materials, materials
and methods). sgRNAs were designed against
constitutive coding exons near the beginning
of each gene and filtered for potential offtarget effects based on sequence similarity
to the rest of the human genome (Fig. 1, F
and G). The library included 10 sgRNAs for
each of 7031 genes and all possible sgRNAs
for each of the 83 genes encoding ribosomal
proteins (Fig. 1H). To assess the effective representation of our microarray synthesized
library, we sequenced sgRNA barcodes from
KBM7 cells 24 hours after infection with the
entire lentiviral pool and were able to detect
the overwhelming majority (>99%) of our
sgRNAs, with high uniformity across constructs (only a sixfold increase in abundance
7

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | R E P O RT

between the 10th and 90th percentiles) (fig.

S2A).
As an initial test of our approach, we
screened the library for genes that function in
DNA mismatch repair (MMR). In the presence
of the nucleotide analog 6-thioguianine (6TG), MMR-proficient cells are unable to repair
6-TGinduced lesions and arrest at the G2-M
cell-cycle checkpoint, whereas MMR-defective
cells do not recognize the lesions and continue
to divide (23). We infected Cas9-KBM7 cells
with the entire sgRNA library, cultured the
cells in a concentration of 6-TG that is lethal
to wild-type (WT) KBM7 cells, and sequenced
the sgRNA barcodes in the final population.
sgRNAs targeting the genes encoding the four
components of the MMR pathway (MSH2,
MSH6, MLH1, and PMS2) (24) were dramatically enriched in the 6-TGtreated cells. At
least four independent sgRNAs for each gene
showed very strong enrichment, and barcodes
corresponding to these genes made up >30%
of all barcodes (Fig. 2, A and B). Each of the
20 most abundant sgRNAs targeted one of
these four genes. The fact that few of the other
73,000 sgRNAs scored highly in this assay sug8

gests a low frequency of off-target effects.

We next addressed the challenge of loss-offunction screening in diploid cells, which require biallelic inactivation of a target gene. We
therefore generated an inducible Cas9 derivative of the HL60 pseudo-diploid human leukemic cell line. In both HL60 and KBM7 cells, we
screened for genes whose loss conferred resistance to etoposide, a chemotherapeutic agent
that poisons DNA topoisomerase IIA (TOP2A).
To identify hit genes, we calculated the difference in abundance between the treated
and untreated populations for each sgRNA,
calculated a score for each gene using a Kolmogorov-Smirnov test to compare the sgRNAs
targeting the gene against the nontargeting
control sgRNAs, and corrected for multiple
hypothesis testing (Fig. 2, C to E, and table S2).
Identical genes were detected in both screens,
with significance levels exceeding all other
genes by more than 100-fold. As expected, loss
of TOP2A itself conferred strong protection to
etoposide (25). The screen also revealed a role
for CDK6, a G1 cyclin-dependent kinase, in mediating etoposide-induced cytotoxicity. Every
one of the 20 sgRNAs in the library targeting

TOP2A or CDK6 was strongly enriched (>90th

percentile) in both screens, indicating that the
effective coverage of our libraries is very high.
We generated isogenic HL60 cell lines with
individual sgRNAs against TOP2A and CDK6
and, consistent with the screen results, these
lines were much more resistant to etoposide
than parental or sgAAVS1-modified HL60 cells
(Fig. 2, F and G). Thus, our Cas9/sgRNA system enables large-scale positive selection lossof-function screens.
To identify genes required for cellular proliferation, we screened for genes whose loss
conferred a selective disadvantage on cells.
Such a screen requires accurate identification
of sgRNAs that are depleted from the final cell
population. A sgRNA will show depletion only
if cleavage of the target gene occurs in the majority of cells carrying the construct.
As an initial test, we screened KBM7 cells
with a small library containing sgRNAs targeting the BCR and ABL1 genes (table S3). The
survival of KBM7 cells depends on the fusion
protein produced by the BCR-ABL translocation (26). As expected, depletion was seen only
for sgRNAs targeting the exons of BCR and
sciencemag.org SCIENCE

Fig. 2. Resistance screens using

CRISPR-Cas9. (A) Raw abundance
(percentage) of sgRNA barcodes after
12 days of selection with 6-TG. (B) MMR
deficiency confers resistance to 6-TG.
Diagram depicts cellular DNA repair
processes. Only sgRNAs targeting components of the DNA MMR pathway were
enriched. The diagram was modified
and adapted from (32). (C) Primary
etoposide screening data. The count
for a sgRNA is defined as the number of
reads that perfectly match the sgRNA
target sequence. (D) sgRNAs from both screens were ranked by their differential abundance between the treated versus untreated populations. For clarity, sgRNAs with no change in abundance are omitted. (E) Gene hit identification by comparing differential abundances of all sgRNAs targeting a gene
with differential abundances of nontargeting sgRNAs in a one-sided Kolmogorov-Smirnov test. P values are corrected for multiple hypothesis testing. (F)
Immunoblot analysis of WT and sgRNA-modified HL60 cells 1 week after infection. S6K1 was used as a loading control. (G) Viability, as measured by cellular ATP concentration, of WT and sgRNA-modified HL60 cells at indicated etoposide concentrations. Error bars denote SD (n = 3 experiments per group).

ABL1 that encode the fusion protein, but not

for those targeting the other exons of BCR and
ABL1 (Fig. 3A).
We then infected Cas9-HL60, Cas9KBM7, and WT KBM7 cells with the entire
73,000-member sgRNA library and used deep
sequencing of the sgRNA barcodes to monitor
the change in abundance of each sgRNA between the initial seeding and a final population obtained after 12 cell doublings (fig. S2,
A and B).
We began by analyzing ribosomal protein
genes, for which the library contained all
possible sgRNAs. We observed strong Cas9dependent depletion of sgRNAs targeting
genes encoding ribosomal proteins, with good
concordance between the sets of ribosomal
protein genes essential for cell proliferation
in the HL60 and KBM7 screens (the median
sgRNA fold-change in abundance was used as
a measure of gene essentiality) (Fig. 3, B and
C). A few ribosomal protein genes were not
found to be essential. These were two genes
encoded on chromosome Y [RPS4Y2, which
is testes-specific (27), and RPS4Y1, which is
expressed at low levels as compared with its
homolog RPS4X on chromosome X (28)] and
ribosome-like proteins, which may be required only in select tissues (27) and generally
are lowly expressed in KBM7 cells (fig. S3A).
We then turned our attention to other genes
SCIENCE sciencemag.org

within our data set, for which 10 sgRNAs were

designed. As for the ribosomal genes, the essentiality scores of these genes were also
strongly correlated between the two cells lines
(fig. S3B and table S4). For the 20 highest scoring genes, we found independent evidence for
essentiality, based primarily on data from
large-scale functional studies in model organisms (table S5).
To evaluate the results at a global level,
we tested 4722 gene sets to see whether they
showed strong signatures of essentiality by using gene set enrichment analysis (29). Gene
sets related to fundamental biological processesincluding DNA replication, gene transcription, and protein degradationshowed
strong depletion, which is consistent with
their essentiality (Fig. 3D and table S6).
Last, we sought to understand the features
underlying sgRNA efficacy. Although the vast
majority of sgRNAs against ribosomal protein
genes showed depletion, detailed comparison
of sgRNAs targeting the same gene revealed
substantial variation in the precise amounts
of depletion. These differences are unlikely to
be caused by local accessibility to the Cas9/
sgRNA complex inasmuch as comparable variability was observed even among sgRNAs targeting neighboring target sites of a given gene
(fig. S4A). Given that our library includes all
possible sgRNAs against each of the 84 ribo-

Fig. 1. A pooled approach

for genetic screening
in mammalian cells by
using a lentiviral CRISPRCas9 system. (A) Outline
of sgRNA library construction and genetic screening
strategy. (B) Immunoblot
analysis of WT KBM7 cells
and KBM7 cells transduced
with a doxycycline-inducible
FLAG-Cas9 construct upon
doxycycline induction.
S6K1 was used as a loading
control. (C) Sufficiency
of single-copy sgRNAs to
induce genomic cleavage.
Cas9-expressing KBM7
cells were transduced with
AAVS1-targeting sgRNA
lentivirus at low MOI. The
SURVEYOR mutation detection assay was performed
on cells at the indicated
days post-infection (dpi).
Briefly, mutations resulting from cleavage of the
AAVS1 locus were detected
through polymerase chain
reaction (PCR) amplification
of a 500-bp amplicon flanking the target sequence,
re-annealing of the PCR
product, and selective
digestion of mismatched
heteroduplex fragments.
(D) Characterization of
mutations induced by
CRISPR-Cas9 as analyzed with high-throughput sequencing. (E) sgRNA library design pipeline. (F) Example of sgRNAs designed for PSMA4. sgRNAs
targeting constitutive exonic coding sequences nearest to the start codon were chosen for construction. (G) Composition of genome-scale sgRNA library.

somal genes, the data allowed us to search

for factors that might explain the differential
efficacy of sgRNAs. Because the majority of
ribosomal protein genes are essential, we reasoned that the level of depletion of a given ribosomal protein-targeting sgRNA could serve
as a proxy for its cleavage efficiency. Applying
this approach, we found several trends related
to sgRNA efficacy: (i) Single-guide sequences
with very high or low GC content were less effective against their targets. (ii) sgRNAs targeting the last coding exon were less effective
than those targeting earlier exons, which is
consistent with the notion that disruption of
the terminal exon would be expected to have
less impact on gene function. (iii) sgRNAs targeting the transcribed strand were less effective than those targeting the nontranscribed
strand (Fig. 3E). Although these trends were
statistically significant, they explained only a
small proportion of differences in sgRNA efficacy (table S7).
We hypothesized that differences in sgRNA
efficacy might also result from sequence features governing interactions with Cas9. To
test this, we developed a method to profile the
sgRNAs directly bound to Cas9 in a highly
parallel manner (supplementary materials,
materials and methods). By comparing the
abundance of sgRNAs bound to Cas9 relative to the abundance of their corresponding
9

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | R E P O RT

genomic integrants, we found that the nucleotide composition near the 3 end of the
spacer sequence was the most important
determinant of Cas9 loading (Fig. 3F). Specifically, Cas9 preferentially bound sgRNAs
containing purines in the last four nucleotides of the spacer sequence, whereas pyrimidines were disfavored. A similar pattern
emerged when we examined depletion of ribosomal protein-targeting sgRNAs [correlation coefficient (r) = 0.81], suggesting that,
in significant part, the cleavage efficiency of
a sgRNA was determined by its affinity for
Cas9 (table S7).
We then sought to build an algorithm to discriminate between strong and weak sgRNAs
(Fig. 3G). We trained a support-vector-machine classifier based on the target sequences
and depletion scores of ribosomal proteintargeting sgRNAs. As an independent test, we
used the classifier to predict the efficacy of
sgRNAs targeting the 400 top scoring (essential) nonribosomal genes. The top two thirds
of our predictions exhibited threefold higher
efficacy than that of the remaining fraction,
confirming the accuracy of the algorithm.
Using this algorithm, we designed a
whole-genome sgRNA library consisting of
sequences predicted to have higher efficacy
(table S8). As with the sgRNA pool used in our
screens, this new collection was also filtered
for potential off-target matches. This reference set of sgRNAs may be useful both for
targeting single genes as well as large-scale
sgRNA screening.
Taken together, these results demonstrate
the utility of CRISPR-Cas9 for conducting
large-scale genetic screens in mammalian
cells. On the basis of our initial experiments,
this system appears to offer several powerful features that together provide substantial
advantages over current functional screening
methods.
First, CRISPR-Cas9 inactivates genes at the
DNA level, making it possible to study phenotypes that require a complete loss of gene
function to be elicited. In addition, the system
should also enable functional interrogation of
nontranscribed elements, which are inaccessible by means of RNAi.

Second, a large proportion of sgRNAs successfully generate mutations at their target

sites. Although this parameter is difficult to
directly assess in pooled screens, we can obtain an estimate by examining the hit rate
at known genes. Applying a z score analysis
of our positive selection screens, we found
that over 75% (46 of 60) of sgRNAs score at a
significance threshold that perfectly separates
true and false positives on a gene level (fig. S5,
A to D). Together, these results show that the
effective coverage of our library is very high
and that the rate of false negatives should be
low, even in a large-scale screen.
Third, off-target effects do not appear to
seriously hamper our screens, according to
several lines of evidence. Direct sequencing
of potential off-target loci detected minimal
cleavage at secondary sites, which typically
reside in noncoding regions and do not affect
gene function. Moreover, in the 6-TG screens
the 20 most abundant sgRNAs all targeted one
of the four members of the MMR pathway. In
total, they represented over 30% of the final
pool, which is a fraction greater the next 400
sgRNAs combined. In the etoposide screen, the
two top genes scored far above background
levels (P values 100-fold smaller than that of
the next best gene), enabling clear discrimination between true and false-positive hits. Last,
new versions of the CRISPR-Cas9 system have
recently been developed that substantially decrease off-target activity (30, 31).
Although we limited our investigation to
proliferation-based phenotypes, our approach
can be applied to a much wider range of biological phenomena. With appropriate sgRNA
libraries, the method should enable genetic
analyses of mammalian cells to be conducted with a degree of rigor and completeness currently possible only in the study of
microorganisms.
R E F E R E N C ES A ND N OTES

Fig. 3. Negative selection screens using CRISPR-Cas9 reveal rules governing sgRNA efficacy. (A) Selective depletion of sgRNAs targeting exons
of BCR and ABL1 present in the fusion protein. Individual sgRNAs are plotted according to their target sequence position along each gene, and the height
of each bar indicates the level of depletion observed. Boxes indicate individual exons. (B) Cas9-dependent depletion of sgRNAs targeting ribosomal proteins. Cumulative distribution function plots of log2 fold changes in sgRNA abundance before and after 12 cell doublings in Cas9-KBM7, Cas9-HL60, and
WT-KBM7 cells. (C) Requirement of similar sets of ribosomal protein genes for proliferation in the HL60 and KBM7 cells. Gene scores are defined as the
median log2 fold change of all sgRNAs targeting a gene. (D) Depleted sgRNAs target genes involved in fundamental biological processes. Gene set enrichment analysis was performed on genes ranked by their combined depletion scores from screens in HL60 and KBM7 cells. Vertical lines underneath the
x axis denote members of the gene set analyzed. (E) Features influencing sgRNA efficacy. Depletion (log2 fold change) of sgRNAs targeting ribosomal
protein genes was used as an indicator of sgRNA efficacy. Correlation between log2 fold changes and spacer %GC content (left), exon position targeted
(middle), and strand targeted (right) are depicted (*P < 0.05). (F) sgRNA target sequence preferences for Cas9 loading and cleavage efficiency. Positionspecific nucleotide preferences for Cas9 loading are determined by counting sgRNAs bound to Cas9 normalized to the number of corresponding genomic
integrations. Heatmaps depict sequence-dependent variation in Cas9 loading (top) and ribosomal protein gene-targeting sgRNA depletion (bottom). The
color scale represents the median value (of Cas9 affinity or log2fold-change) for all sgRNAs with the specified nucleotide at the specified position. (G)
sgRNA efficacy prediction. Ribosomal protein gene-targeting sgRNAs were designated as weak or strong on the basis of their log2 fold change and used
to train a support-vector-machine (SVM) classifier. As an independent test, the SVM was used to predict the efficacy of sgRNAs targeting 400 essential
nonribosomal genes (*P < 0.05).

1. G. Giaever et al., Nature 418, 387391 (2002).

2. M. Costanzo et al., Science 327, 425431 (2010).
3. J. E. Carette et al., Science 326, 12311235 (2009).
4. G. Guo, W. Wang, A. Bradley, Nature 429, 891895
(2004).
5. A. Fire et al., Nature 391, 806811 et al., (1998).
6. T. R. Brummelkamp, R. Bernards, R. Agami, Science

296, 550553 (2002).
7. J. Moffat et al., Cell 124, 12831298 (2006).
8. V. N. Ngo et al., Nature 441, 106110 (2006).
9. H. W. Cheung et al., Proc. Natl. Acad. Sci. U.S.A. 108,

1237212377 (2011).
10. C. J. Echeverri et al., Nat. Methods 3, 777779 (2006).
11. M. Booker et al., BMC Genomics 12, 50 (2011).
12. W. G. Kaelin Jr., Science 337, 421422 (2012).
13. R. Barrangou et al., Science 315, 17091712 (2007).
14. L. Cong et al., Science 339, 819823 (2013).
15. P. Mali et al., Science 339, 823826 (2013).
16. W. Y. Hwang et al., Nat. Biotechnol. 31, 227229 (2013).
17. M. Jinek et al., Science 337, 816821 (2012).
18. T. Horii, D. Tamura, S. Morita, M. Kimura, I. Hatada, Int.

J. Mol. Sci. 14, 1977419781 (2013).
19. H. Wang et al., Cell 153, 910918 (2013).
20. P. D. Hsu et al., Nat. Biotechnol. 31, 827832 (2013).
21. Y. Fu et al., Nat. Biotechnol. 31, 822826 (2013).
22. V. Pattanayak et al., Nat. Biotechnol. 31, 839843
(2013).
23. T. Yan, S. E. Berry, A. B. Desai, T. J. Kinsella, Clin. Cancer
Res. 9, 23272334 (2003).

10

SCIENCE sciencemag.org

sciencemag.org SCIENCE

24. R. D. Kolodner, G. T. Marsischky, Curr. Opin. Genet.

Dev. 9, 8996 (1999).
25. D. J. Burgess et al., Proc. Natl. Acad. Sci. U.S.A. 105,

90539058 (2008).
26. B. Scappini et al., Cancer 100, 14591471 (2004).
27. S. Xue,M. Barna, Nat. Rev. Mol. Cell Biol. 13, 355369
(2012).
28. C. M. Johnston et al., PLOS Genet. 4, e9 (2008).
29. A. Subramanian et al., Proc. Natl. Acad. Sci. U.S.A. 102,

1554515550 (2005).
30. F. A. Ran et al., Cell 154, 13801389 (2013).
31. P. Mali et al., Nat. Biotechnol. 31, 833838 (2013).
32. K. Yoshimoto et al., Front. Oncol. 2, 186 (2012).
AC K N OW L E DG M E NTS
We thank all members of the Sabatini and Lander labs,
especially J. Engreitz, S. Schwartz, A. Shishkin, and Z.
Tsun for protocols, reagents, and advice; T. Mikkelsen for
assistance with oligonucleotide synthesis; and L. Gaffney
for assistance with figures. This work was supported
by the U.S. National Institutes of Health (CA103866)

(D.M.S.), National Human Genome Research Institute

(2U54HG003067-10) (E.S.L.), the Broad Institute of MIT
and Harvard (E.S.L.), and an award from the U.S. National
Science Foundation (T.W.). The composition of the sgRNA
pools and screening data can be found in the supplementary materials. A patent application has been filed by the
Broad Institute relating to aspects of the work
described in this manuscript. Inducible Cas9 and sgRNA
backbone lentiviral vectors and the genome-scale sgRNA
plasmid pool are deposited in Addgene.
SU PPL E M E NTARY MAT ERIALS

www.sciencemag.org/content/343/6166/80/suppl/DC1
Materials and Methods Supplementary Text
Figs. S1 to S5
Tables S1 to S8
References (3343)
8 October 2013; accepted 2 December 2013
Published online 12 December 2013;
10.1126/science.1246981

Genome-scale CRISPR-Cas9
knockout screening in human cells
Ophir Shalem,1,2* Neville E. Sanjana,1,2* Ella Hartenian,1 Xi Shi,1,3 David A. Scott,1,2 Tarjei S.
Mikkelsen,1 Dirk Heckl,4 Benjamin L. Ebert,4 David E. Root,1 John G. Doench,1 Feng Zhang1,2
The simplicity of programming the CRISPR (clustered regularly interspaced short
palindromic repeats)associated nuclease Cas9 to modify specific genomic loci suggests
a new way to interrogate gene function on a genome-wide scale. We show that lentiviral
delivery of a genome-scale CRISPR-Cas9 knockout (GeCKO) library targeting 18,080
genes with 64,751 unique guide sequences enables both negative and positive selection
screening in human cells. First, we used the GeCKO library to identify genes essential
for cell viability in cancer and pluripotent stem cells. Next, in a melanoma model, we
screened for genes whose loss is involved in resistance to vemurafenib, a therapeutic
RAF inhibitor. Our highest-ranking candidates include previously validated genes NF1 and
MED12, as well as novel hits NF2, CUL3, TADA2B, and TADA1. We observe a high level of
consistency between independent guide RNAs targeting the same gene and a high rate of
hit confirmation, demonstrating the promise of genome-scale screening with Cas9.

major goal since the completion of the

Human Genome Project is the functional
characterization of all annotated genetic
elements in normal biological processes
and disease (1). Genome-scale loss-offunction screens have provided a wealth of
information in diverse model systems (25).
In mammalian cells, RNA interference (RNAi)
is the predominant method for genome-wide
loss-of-function screening (2, 3), but its utility
is limited by the inherent incompleteness of
protein depletion by RNAi and confounding
off-target effects (6, 7).
The RNA-guided CRISPR (clustered regularly interspaced short palindrome repeats)
associated nuclease Cas9 provides an effective
means of introducing targeted loss-of-function mutations at specific sites in the genome
(8, 9). Cas9 can be programmed to induce
DNA double-strand breaks (DSBs) at specific
genomic loci (8, 9) through a synthetic singleguide RNA (sgRNA) (10), which when targeted
to coding regions of genes can create frame
shift insertion/deletion (indel) mutations that
result in a loss-of-function allele. Because the
Originally published 3 January 2014

targeting specificity of Cas9 is conferred by

short guide sequences, which can be easily
generated at large scale by array-based oligonucleotide library synthesis (11), we sought
to explore the potential of Cas9 for pooled
genome-scale functional screening.
Lentiviral vectors are commonly used
for delivery of pooled short-hairpin RNAs
(shRNAs) in RNAi because they can be easily
titrated to control transgene copy number
and are stably maintained as genomic
integrants during subsequent cell replication
Broad Institute of MIT and Harvard, 7 Cambridge Center,
Cambridge, MA 02142, USA.
2
McGovern Institute for Brain Research, Department of Brain
and Cognitive Sciences, Department of Biological Engineering,
Massachusetts Institute of Technology, Cambridge, MA 02139,
USA.
3
Stanley Center for Psychiatric Research, Broad Institute of
Harvard and MIT, 7 Cambridge Center, Cambridge, MA 02142,
USA.
4
Division of Hematology, Department of Medicine, Brigham and
Womens Hospital, Harvard Medical School, Boston, MA 02115,
USA.
*These authors contributed equally to this work.
Corresponding author. E-mail: zhang@broadinstitute.org
1

11

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

(2, 12, 13). Therefore, we designed a single

lentiviral vector to deliver Cas9, a sgRNA,
and a puromycin selection marker into target
cells (lentiCRISPR) (Fig. 1A). The ability
to simultaneously deliver Cas9 and sgRNA
through a single vector enables application
to any cell type of interest, without the need
to first generate cell lines that express Cas9.

To determine the efficacy of gene knockout

by lentiCRISPR transduction, we tested six
sgRNAs targeting enhanced green fluorescent
protein (EGFP) in a human embryonic kidney (HEK) 293T cell line containing a single
copy of EGFP (fig. S1). After transduction at
a low multiplicity of infection (MOI = 0.3)
followed by selection with puromycin, lenti-

Fig. 1. Lentiviral
delivery of Cas9
and sgRNA provides
efficient depletion of
target genes. (A)
Lentiviral expression
vector for Cas9 and
sgRNA (lentiCRISPR).
puro, puromycin
selection marker; psi+,
psi packaging signal;
RRE, rev response
element; cPPT, central
polypurine tract; EFS,
elongation factor-1
short promoter;
P2A, 2A self-cleaving
peptide; WPRE,
posttranscriptional
regulatory element.
(B) Distribution of fluorescence from 293T-EGFP cells transduced by EGFP-targeting lentiCRISPR (sgRNAs 1 to 6, outlined
peaks) and Cas9-only (green-shaded peak) vectors, and nonfluorescent 293T cells (gray shaded
peak). (C) Distribution of fluorescence from 293T-EGFP cells transduced by EGFP-targeting shRNA
(shRNAs 1 to 4, outlined peaks) and control shRNA (green-shaded peak) vectors, and nonfluorescent
293T cells (gray shaded peak).

Fig. 2. GeCKO library design and application for genome-scale

negative selection screening. (A) Design of sgRNA library for
genome-scale knockout of coding sequences in human cells (see supplementary text). (B and C) Cumulative frequency of sgRNAs 3 and
14 days after transduction in A375 and human embryonic stem cells,

12

A RT I C L E S | R E P O RT

CRISPRs abolished EGFP fluorescence in 93

8% (mean SD) of cells after 11 days (Fig. 1B).
Deep sequencing of the EGFP locus revealed
a 92 9% indel frequency (n 104 sequencing reads per condition) (fig. S2). In contrast,
transduction of cells with lentiviral vectors
expressing EGFP-targeting shRNA led to incomplete knockdown of EGFP fluorescence
(Fig. 1C).
Given the high efficacy of gene knockout by
lentiCRISPR, we tested the feasibility of conducting genome-scale CRISPR-Cas9 knockout
(GeCKO) screening with a pooled lentiCRISPR
library. We designed a library of sgRNAs targeting 5 constitutive exons (Fig. 2A) of 18,080
genes in the human genome with an average
coverage of 3 to 4 sgRNAs per gene (table S1),
and each target site was selected to minimize
off-target modification (14) (see supplementary text).
To test the efficacy of the full GeCKO library
at achieving knockout of endogenous gene
targets, we conducted a negative selection
screen by profiling the depletion of sgRNAs
targeting essential survival genes (Fig. 2A).
We transduced the human melanoma cell line
A375 and the human stem cell line HUES62
with the GeCKO library at a MOI of 0.3. As
expected, deep sequencing (figs. S3 and S4)
14 days after transduction revealed a significant reduction in the diversity of sgRNAs in
the surviving A375 and HUES62 cells (Fig. 2,
B and C) (Wilcoxon rank sum test, P < 1010 for
both cell types). Gene set enrichment analysis (GSEA) (15) indicated that most of the de-

respectively. Shift in the 14-day curve represents the depletion in a subset of sgRNAs. (D and E) The five most significantly depleted gene sets
in A375 cells [nominal P < 105, false discovery rate (FDR)corrected q <
105] and HUES62 cells (nominal P < 105, FDR-corrected q < 103) identified by GSEA (15).
sciencemag.org SCIENCE

Fig. 3. GeCKO screen in A375 melanoma cells

reveals genes whose loss confers PLX resistance.
(A) Timeline of PLX resistance screen in A375 melanoma cells. (B) Growth
of A375 cells when treated with dimethyl sulfoxide (DMSO) or PLX over 14
days. (C) Box plot showing the distribution of sgRNA frequencies at different time points, with and without PLX treatment (vehicle is DMSO). The
box extends from the first to the third quartile with the whiskers denoting 1.5 times the interquartile range. Enrichment of specific sgRNAs: 7

pleted sgRNAs targeted essential genes such

as ribosomal structural constituents (Fig. 2,
D and E, and tables S2 and S3). The overlap
in highly depleted genes and functional gene
categories between the two cell lines (fig. S5)
indicates that GeCKO can identify essential
genes and that enrichment analysis of depleted sgRNAs can pinpoint key functional
gene categories in negative selection screens.
To test the efficacy of GeCKO for positive
selection, we sought to identify gene knockouts that result in resistance to the BRAF
protein kinase inhibitor vemurafenib (PLX)
in melanoma (16) (Fig. 3A). Exposure to PLX
resulted in growth arrest of transduced A375
cells, which harbor the V600E gain-of-function BRAF mutation (17) (Fig. 3B), therefore
enabling the enrichment of a small group
of cells that were rendered drug-resistant
by Cas9:sgRNA-mediated modification. After
14 days of PLX treatment, the sgRNA distribution was significantly different when
compared with vehicle-treated cells (Fig. 3C)
(Wilcoxon rank-sum test, P < 1010) and clustered separately from all other conditions
(Fig. 3D and fig. S6).
SCIENCE sciencemag.org

days of PLX treatment, 1 sgRNA greater than 10-fold enrichment; 14 days

of PLX treatment, 379 and 49 sgRNAs greater than 10-fold and 100-fold
enrichment, respectively. (D) Rank correlation of normalized sgRNA
read count between biological replicates and treatment conditions. (E)
Scatterplot showing enrichment of specific sgRNAs after PLX treatment.
(F) Identification of top candidate genes using the RIGER P value analysis.

For a subset of genes, we found enrichment of multiple sgRNAs that target each
gene after 14 days of PLX treatment (Fig.
3E), suggesting that loss of these particular genes contributes to PLX resistance. We
used the RNAi Gene Enrichment Ranking
(RIGER) algorithm to rank screening hits
by the consistent enrichment among multiple sgRNAs targeting the same gene (Fig.
3F and table S4) (12). Our highest-ranking
genes included previously reported candidates NF1 and MED12 (18, 19) and also several genes not previously implicated in PLX
resistance, including neurofibromin 2 (NF2),
Cullin 3 E3 ligase (CUL3), and members of
the STAGA histone acetyltransferase complex (TADA1 and TADA2B). These candidates
yield new testable hypotheses regarding PLX
resistance mechanisms (see supplementary
text). For example, NF1 and NF2, although
unrelated in sequence, are each mutated
in similar but distinct forms of neurofibromatosis (20). In addition, epigenetic dysregulation resulting from mutations in the
mechanistically related STAGA and Mediator
complexes (21) may have a role in acquired

drug resistance. All of these hits were also

identified through a second independent library transduction (figs. S7 and S8 and tables
S5 and S6).
A similar screen to identify PLX drug resistance in A375 cells was previously conducted
using a pooled library of 90,000 shRNAs (19).
To compare the efficacy and reliability of
genome-scale shRNA screening with GeCKO,
we used several methods to evaluate the degree of consistency among the sgRNAs or
shRNAs targeting the top candidate genes.
First, we plotted the P values for the top 100
hits using either RIGER (Fig. 4A) or redundant siRNA activity (RSA) (fig. S9) scoring.
Lower P values for the GeCKO versus shRNA
screen indicate better scoring consistency
among sgRNAs. Second, for the top 10 RIGER
hit genes, 78 27% of sgRNAs targeting each
gene ranked among the top 5% of enriched
sgRNAs, whereas 20 12% of shRNAs targeting each gene ranked among the top 5% of
enriched shRNAs (Fig. 4B).
We validated top-ranking genes from
the GeCKO screen individually using 3 to 5
sgRNAs (Fig. 4, C to E, and figs. S10 and S11).
13

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | N E W S F O C U S

The CRISPR craze

Elizabeth Pennisi
A bacterial immune system yields a potentially revolutionary genome-editing technique.

B
Fig. 4. Comparison of GeCKO and shRNA screens and validation of neurofibromin 2 (NF2).
(A) RIGER P values for the top 100 hits from GeCKO and shRNA (19) screens for genes whose loss
results in PLX resistance. Analysis using the RSA algorithm shows a similar trend (fig. S9). (B) For the
top 10 RIGER hits, the percent of unique sgRNAs (top) or shRNAs (bottom) targeting each gene that
are in the top 5% of all enriched sgRNAs or shRNAs. (C) Deep-sequencing analysis of lentiCRISPRmediated indel at the NF2 locus. (D) A375 cells transduced with NF2-targeting lentiCRISPR and
shRNA vectors both show a decrease in NF2 protein levels. (E) Dose-response curves for A375 cells
transduced with individual NF2-targeting lentiCRISPR or shRNA vectors. Controls were EGFP-targeting lentiCRISPR or null-hairpin shRNA vectors. Cells
transduced with NF2-targeting lentiCRISPRs show a significant increase (F1,8 = 30.3, P < 0.001, n = 4 replicates) in the half-maximal effective concentration
(EC50), whereas cells transduced with NF2-targeting shRNA vectors do not (F1,8 = 0.47, P = 0.51, n = 4 replicates).

14

tion, RNAi is limited to transcripts, whereas

Cas9:sgRNAs can target elements across the
entire genome, including promoters, enhancers, introns, and intergenic regions. Furthermore, catalytically inactive mutants of Cas9
can be tethered to different functional domains (2327) to broaden the repertoire of
perturbation modalities, including genomescale gain-of-function screening using Cas9
activators and epigenetic modifiers. In the
GeCKO screens presented here, the efficiency
of complete knockout, the consistency of distinct sgRNAs, and the high validation rate
for top screen hits demonstrate the potential
of Cas9:sgRNA-based technology to transform functional genomics.
REFERENCES AND NOTES

1. E. S. Lander, Nature 470, 187197 (2011).

2. K. Berns et al., Nature 428, 431437 (2004).
3. M. Boutros et al., Heidelberg Fly Array Consortium,
Science 303, 832835 (2004).
4. R. Rad et al., Science 330, 11041107 (2010).
5. J. E. Carette et al., Science 326, 12311235 (2009).
6. A. L. Jackson et al., RNA 12, 11791187 (2006).
7. C. J. Echeverri et al., Nat. Methods 3, 777779 (2006).
8. L. Cong et al., Science 339, 819823 (2013).
9. P. Mali et al., Science 339, 823826 (2013).
10. M. Jinek et al., Science 337, 816821 (2012).
11. A. P. Blanchard, L. Hood, Nat. Biotechnol. 14, 1649
(1996).
12. B. Luo et al., Proc. Natl. Acad. Sci. U.S.A. 105,

2038020385 (2008).
13. P. J. Paddison et al., Nature 428, 427431 (2004).
14. P. D. Hsu et al., Nat. Biotechnol. 31, 827832 (2013).
15. A. Subramanian et al., Proc. Natl. Acad. Sci. U.S.A.

102, 1554515550 (2005).
16. K. T. Flaherty et al., N. Engl. J. Med. 363, 809819
(2010).

17. H. Davies et al., Nature 417, 949954 (2002).

18. S. Huang et al., Cell 151, 937950 (2012).
19. S. R. Whittaker et al., Cancer Discovery 3, 350362
(2013).
20. A. L. Lin, D. H. Gutmann, Nat. Rev. Clin. Oncol. 10,

616624 (2013).
21. X. Liu, M. Vorontchikhina, Y. L. Wang, F. Faiola

E. Martinez, Mol. Cell. Biol. 28, 108121 (2008).
22. F. A. Ran et al., Cell 154, 13801389 (2013).
23. P. Mali et al., Nat. Biotechnol. 31, 833838 (2013).
24. L. A. Gilbert et al., Cell 154, 442451 (2013).
25. S. Konermann et al., Nature 500, 472476 (2013).
26. P. Perez-Pinera et al., Nat. Methods 10, 973976
(2013).
27. M. L. Maeder et al., Nat. Methods 10, 977979 (2013).
ACKNOWLEDGMENTS
We thank G. Cowley, W. Harrington, J. Wright, E. Hodis, S.
Whittaker, J. Merkin, C. Burge, D. Peters, C. Cowan, L. P. Club,
and the entire Zhang laboratory for technical support and
critical discussions. O.S. is a Klarman Family Foundation Fellow,
N.S. is a Simons Center for the Social Brain Postdoctoral Fellow,
D.A.S. is an NSF Fellow, and J.D. is a Merkin Institute Fellow. D.
H. was funded by the German Cancer Foundation. F.Z. is supported by an NIH Directors Pioneer Award (1DP1-MH100706);
a NIH Transformative R01 grant (1R01-DK097768); the Keck,
McKnight, Merkin, Vallee, Damon Runyon, Searle Scholars,
Klingenstein, and Simons Foundations; and Bob Metcalfe and
Jane Pauley. The authors have no conflicting financial interests.
A patent application has been filed relating to this work, and
the authors plan to make the reagents widely available to the
academic community through Addgene and to provide software
tools at the Zhang laboratory Web site (www.genomeengineering.org).
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/343/6166/84/suppl/DC1
Materials and Methods
Supplementary Text
Figs. S1 to S16
Tables S1 to S10
References

9 October 2013; accepted 2 December 2013

Published online 12 December 2013;
10.1126/science.1247005

sciencemag.org SCIENCE

PHOTOS: (TOP TO BOTTOM) REPRINTED BY PERMISSION FROM MACMILLAN PUBLISHERS LTD: NATURE
METHODS (AE FRIEDLAND ET AL., 10, 741743 (2013) DOI:10.1038/NMETH.2532); GONZALES, A.P.W.,
YEH, J.-R. & PETERSON, R.T.; SCOTT GRATZ/UNIVERSITY OF WISCONSIN, MADISON.

For NF2, we found that 4 out of 5 sgRNAs

resulted in >98% allele modification 7 days
after transduction, and all 5 sgRNAs showed
>99% allele modification 14 days after transduction (Fig. 4C). We compared sgRNA and
shRNA-mediated protein depletion and PLX
resistance using Western blot (Fig. 4D) and
cell growth assays (Fig. 4E). Interestingly,
although all five sgRNAs conferred resistance to PLX, only the best shRNA achieved
sufficient knockdown to increase PLX resistance (Fig. 4E), suggesting that even low levels of NF2 are sufficient to retain sensitivity
to PLX. Additionally, sgRNAs targeting NF1,
MED12, CUL3, TADA1, and TADA2B led to a
decrease in protein expression and increased
resistance to PLX (figs. S10 and S11). Deep sequencing confirmed a high rate of mutagenesis at targeted loci (figs. S12 and S13), with
a small subset of off-target sites exhibiting
indels (figs. S14 to S16), which may be alleviated using an offset nicking approach (22, 23)
that was recently shown to reduce off-target
modifications (22).
GeCKO screening provides a mechanistically distinct method from RNAi for systematic perturbation of gene function. Whereas
RNAi reduces protein expression by targeting RNA, GeCKO introduces loss-of-function
mutations into genomic DNA. Although some
indel mutations are expected to maintain the
reading frame, homozygous knockout yields
high screening sensitivity, which is especially important in cases where incomplete
knockdown retains gene function. In addi-

acteria may not elicit much sympathy from us eukaryotes, but they, too,
can get sick. Thats potentially a big
problem for the dairy industry, which
often depends on bacteria such as
Streptococcus thermophilus to make yogurts
and cheeses. S. thermophilus breaks down the
milk sugar lactose into tangy lactic acid. But
certain virusesbacteriophages, or simply
phagescan debilitate the bacterium, wreaking havoc on the quality or quantity of the
food it helps produce.
In 2007, scientists from Danisco, a Copenhagen-based food ingredient company now
owned by DuPont, found a way to boost the
phage defenses of this workhouse microbe.
They exposed the bacterium to a phage and
showed that this essentially vaccinated it
against that virus (Science, 23 March 2007, p.
1650). The trick has enabled DuPont to create
heartier bacterial strains for food production.
It also revealed something fundamental: Bacteria have a kind of adaptive immune system,
which enables them to fight off repeated attacks by specific phages.
That immune system has suddenly become
important for more than food scientists and
microbiologists, because of a valuable feature:
It takes aim at specific DNA sequences.
In January, four research teams reported
harnessing the system, called CRISPR for
peculiar features in the DNA of bacteria that
deploy it, to target the destruction of specific
genes in human cells. And in the following 8
months, various groups have used it to delete,
add, activate, or suppress targeted genes in
human cells, mice, rats, zebrafish, bacteria,
fruit flies, yeast, nematodes, and crops,
demonstrating broad utility for the technique.
Biologists had recently developed several
new ways to precisely manipulate genes, but
CRISPRs efficiency and ease of use trumps
just about anything, says George Church of
Harvard University, whose lab was among the
first to show that the technique worked in
human cells.
With CRISPR, scientists can create mouse
models of human diseases much more quickly
than before, study individual genes much
faster, and easily change multiple genes
in cells at once to study their interactions.
This years CRISPR craze may yet slow
down as limitations of the method emerge,
but Church and other CRISPR pioneers are
already forming companies to harness the
technology for treating genetic diseases. I
dont think theres any example of any field
moving this fast, says Blake Wiedenheft, a
biochemist at Montana State University in
Bozeman.
SCIENCE sciencemag.org

Precise cuts. In just 8 months, CRISPR modifications of DNA resulted in dumpier nematodes
(top, bottom), zebrafish embryos with an excess
of ventral tissue (middle, bottom), and fruit flies
with dark eyes (bottom, right), demonstrating its
broad utility for editing genes in animals.

Humble beginnings
The first inkling of this hot new genetic engineering tool came in 1987, when a research
team observed an oddly repetitive sequence
at one end of a bacterial gene. Few others
took much notice. A decade later, though, biologists deciphering microbial genomes often
Originally published 23 August 2013

found similar puzzling patterns, in which a

sequence of DNA would be followed by nearly
the same sequence in reverse, then 30 or so
seemingly random bases of spacer DNA,
and then a repeat of the same palindromic
sequence, followed by a different spacer DNA.
A single microbe could have several such
stretches, each with different repeat and intervening sequences. This pattern appears in
more than 40% of bacteria and fully 90% of
microbes in a different domain, the archaea,
and gives CRISPR its name. (It stands for clustered regularly interspaced short palindromic
repeats.)
Many researchers assumed that these
odd sequences were junk, but in 2005, three
bioinformatics groups reported that spacer
DNA often matched the sequences of phages,
indicating a possible role for CRISPR in
microbial immunity. That was a very key
clue, says biochemist Jennifer Doudna of
the University of California (UC), Berkeley. It
led Eugene Koonin from the National Center
for Biotechnology Information in Bethesda,
Maryland, and his colleagues to propose that
bacteria and archaea take up phage DNA,
then preserve it as a template for molecules
of RNA that can stop matching foreign DNA
in its tracks, much the way eukaryotic cells
use a system called RNA interference (RNAi)
to destroy RNA.
Enter the Danisco team. In 2007, Rodolphe
Barrangou, Philippe Horvath, and others with
the company showed that they could alter
the resistance of S. thermophilus to phage attack by adding or deleting spacer DNA that
matched the phages. At the time, Barrangou,
who is now at North Carolina State University
in Raleigh, didnt see CRISPRs full potential.
We had no idea that those elements could be
readily exploitable for something as attractive
as genome editing, he says.
Doudna and Emmanuelle Charpentier, currently of the Helmholtz Centre for Infection
Research and Hannover Medical School in
Germany, took the next step. They had independently been teasing out the roles of various CRISPR-associated proteins to learn how
bacteria deploy the DNA spacers in their immune defenses. But the duo soon joined forces
to focus on a CRISPR system that relies on a
protein called Cas9, as it was simpler than
other CRISPR systems.
When CRISPR goes into action in response
to an invading phage, bacteria transcribe the
spacers and the palindromic DNA into a long
RNA molecule that the cell then cuts into
short spacer-derived RNAs called crRNAs. An
additional stretch of RNA, called tracrRNA,
works with Cas9 to produce the crRNA, Charpentiers group reported in Nature in 2011.
The group proposed that together, Cas9,
tracrRNA, and crRNA somehow attack foreign
DNA that matches the crRNA.
The two teams found that the Cas9 protein
is a nuclease, an enzyme specialized for
cutting DNA, with two active cutting sites,
15

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

zinc finger and TALEN technologies both

depend on custom-making new proteins for
each DNA target. The CRISPR systems guide
RNAs are much easier to make than proteins,
Barrangou says. Within a couple weeks you
can generate very tangible results that using
alternative methods would take
months.

Harnessing CRISPR
Speed is not its only advantage.
Churchs group had been pushing
the use of TALENs in human cells,
but when he learned of Doudna
and Charpentiers results, he and
his colleagues made guide RNA
against genes they had already
targeted with TALENs. In three
human cell types, the CRISPR
system was more efficient than
TALENs at cutting the DNA target, and it worked on more genes
than TALENs did (Science, 15
February 2013, p. 823). To demonstrate the ease of the CRISPR system, Churchs team synthesized
a library of tens of thousands of
guide RNA sequences, capable of
targeting 90% of human genes.
You can pepper the genome with
every imaginable CRISPR, he
says.
That makes it possible to alter
virtually any gene with Cas9, exploiting its DNA-cutting ability
to either disable the gene or cut
it apart, allowing substitute DNA
to be inserted. In an independent
paper that appeared at the same
time as Churchs, Feng Zhang, a
synthetic biologist at the Broad
Institute in Cambridge, Massachusetts, and his colleagues
showed that CRISPR can target
and cut two genes at once in huDNA surgeon. With just a guide RNA and a protein called Cas9, researchers first showed that the CRISPR system
man cells (Science, 15 February
can home in on and cut specific DNA, knocking out a gene or enabling part of it to be replaced by substitute DNA.
2013, p. 819). And working with
More recently, Cas9 modifications have made possible the repression (lower left) or activation (lower right) of
developmental biologist Rudolf
specific genes.
Jaenisch at the Whitehead Institute for Biomedical Research in
complex to home in on its target DNA. The
curred. Then, a decade ago, researchers develCambridge, Zhang has since disrupted five
possibility of using a single enzyme by just
oped zinc finger nucleases, synthetic proteins
genes at once in mouse embryonic stem (ES)
changing the RNA seemed very simple,
that have DNA-binding domains that enable
cells.
Doudna recalls.
them to home in and break DNA at specific
Such work lays the foundation for
Before CRISPR could be put to use, howspots. A welcome addition to the genetic engenerating mutant mice, a key tool for
ever, Doudnas and Charpentiers teams had
gineering toolbox, zinc fingers even spawned
biomedical research. One approach would
to show that they could control where Cas9
a company that is testing a zinc finger to
be to add the altered mouse ES cells to a
went to do its cutting. First, Doudnas posttreat people infected with HIV (Science, 23
developing embryo and breed the resulting
doc, Martin Jinek, figured out how to combine
December 2005, p. 1894). More recently, synanimals. But Zhang has demonstrated a
tracrRNA and spacer RNA into a single-guide
thetic nucleases called TALENs have proved
faster option. His team found it could simply
RNA molecule; then, as a proof of principle,
an easier way to target specific DNA and were
inject fertilized mouse eggs, or zygotes, with
the team last year made several guide RNAs,
predicted to surpass zinc fingers (Science, 14
Cas9 messenger RNA and two guide RNAs
mixed them with Cas9, and showed in a test
December 2012, p. 1408).
and, with 80% efficiency, knock out two
tube that the synthetic complexes could find
Now, CRISPR systems have stormed onto
genes. They could also perform more delicate
and cut their DNA targets (Science, 17 August
the scene, promising to even outcompete
genomic surgery on the embryos by shackling
2012, p. 816). That was a milestone paper,
TALENs. Unlike the CRISPR system, which
Cas9, so that it nicks target DNA instead of
Barrangou says.
uses RNA as its DNA-homing mechanism,
cutting it. In this way, they could introduce a
16

sciencemag.org SCIENCE

new part of a gene through a process called

homology-directed repair, they reported in
the 2 May 2013 issue of Cell.
Developing a new mouse model for a disease now entails careful breeding of multiple
generations and can take a year; with Zhangs

The cost of admission is low: Free software exists to design guide RNA to target
any desired gene, and a repository called Addgene, based in Cambridge, offers academics
the DNA to make their own CRISPR system
for $65. Since the beginning of the year, Ad-

CRISPRed rice. Earlier this month, researchers showed CRISPR works in plants, such as rice, where
the knocked-out gene resulted in dwarf albino individuals (right).

PHOTO: GAO CAIXIA LABORATORY

This precision targeting drives the growing

interest in CRISPR. Genetic engineers have
long been able to add and delete genes in a
number of organisms. But they couldnt dictate where those genes would insert into the
genome or control where gene deletions oc-

ILLUSTRATION: K. SUTLIFF/SCIENCE

one site for each strand of the DNAs double

helix. And in a discovery that foreshadowed
CRISPRs broad potential for genome
engineering, the team demonstrated that
they could disable one or both cutting sites
without interfering with the ability of the

A RT I C L E S | NEWS F OCUS

CRISPR technique, a new mouse model could

be ready for testing in a matter of weeks. And
Zhang thinks the approach is not limited to
mice. As long as you can manipulate the embryo and then reimplant it, then you will be
able to do it in larger animals, perhaps even
primates.
Doudnas group and a Korean team reported using CRISPR to cut DNA in human cells 3
weeks after Zhangs and Churchs papers went
online, and, at the same time, another group
revealed they had used CRISPR to make mutant zebrafish. This cascade of papers has had
a synergistic effect, commanding the attention of a broad swath of the biology community. If a single paper comes out, it gets some
attention, but when six papers come out all
together, thats when people say, I have to do
this, says Charles Gersbach, a biomedical engineer at Duke University in Durham, North
Carolina.
Once she saw Doudna and Charpentiers
paper a year ago, Gao Caixia became one of
the early converts. Her group at the Chinese
Academy of Sciences Institute of Genetics and
Developmental Biology in Beijing had been
using zinc finger and TALENs technology
on rice and wheat. Using CRISPR, they have
now disabled four rice genes, suggesting that
the technique could be used to engineer this
crucial food crop. In wheat, they knocked out
a gene that, when disabled, may lead to plants
resistant to powdery mildew. In a measure of
the excitement that CRISPR has generated,
the teams report in the August 2013 issue
of Nature Biotechnology was accompanied
by four other papers describing CRISPR
successes in plants and in rats.
SCIENCE sciencemag.org

dgeneto which 11 teams have contributed

CRISPR-enabling DNA sequenceshas distributed 5000 CRISPR constructs, and in a
single July week the repository received 100
orders for a new construct. They are kind of
crazy hot, says Joanne Kamens, Addgenes executive director.
Fine-tuning gene activity
The initial CRISPR genome-editing papers
all relied on DNA cutting, but other applications quickly appeared. Working with Doudna, Lei S. Qi from UC San Francisco and his
colleagues introduced CRISPRi, which, like
RNAi, turns off genes in a reversible fashion
and should be useful for studies of gene function. They modified Cas9 so it and the associated guide RNA would still home in on a
target but would not cut DNA once there. In
bacteria, the presence of Cas9 alone is enough
to block transcription, but for mammalian applications, Qi and colleagues add to it a section of protein that represses gene activity. Its
guide RNA is designed to home in on regulatory DNA, called promoters, which immediately precede the gene target.
Last month, that team and three other
groups used a Cas9 to ferry a synthetic transcription factora protein fragment that
turns on genesenabling them to activate
specific human genes. Just using one CRISPR
construct had a weak effect, but all four teams
found a way to amplify it. By targeting multiple CRISPR constructs to slightly different
spots on the genes promoter, says Gersbach,
one of the team leaders, we saw a huge synergistic effect.
In the 25 July 2013 issue of Nature Methods,

he reported activating genes tied to human

diseases, including those involved in muscle differentiation, controlling cancer and
inflammation, and producing fetal hemoglobin. Two other teams also targeted biomedically important genes. CRISPR control
of such genes could treat diseases ranging
from sickle cell anemia to arthritis, Gersbach suggests.
CRISPR technology may yet have limitations. Its unclear, for example, how specific
the guide RNAs are for just the genes they
are supposed to target. Our initial data suggest that there can be significant off-target
effects, says J. Keith Joung from the Massachusetts General Hospital in Boston, who
back in January demonstrated that CRISPR
would alter genes in zebrafish embryos and
has used CRISPR to turn on genes. His work
shows that nontarget DNA resembling the
guide RNA can become cut, activated, or
deactivated.
Joungs group showed that a guide RNA
can target DNA that differs from the intended target sequence in up to five of its
bases. Zhang has gotten more reassuring
results but says that the specificity is still
something we have to work on, especially
as more people begin to think about delivering CRISPR systems as treatments for human diseases. To really make the technology safe, we really have to make sure it goes
where we want it to go to and nowhere else.
Researchers must also get the CRISPR
components to the right place. Delivery is
an enormous challenge and will be cell type
and organism specific, Joung notes. With
zebrafish, his team injects guide RNA and
messenger RNA for Cas9 directly into embryos; with mammalian cells, they use DNA
constructs. How CRISPR might be delivered
into adult animals, or to treat disease in people, is just now being considered.
Ultimately, CRISPR may take a place
beside zinc fingers and TALENs, with the
choice of editing tool depending on the particular application. But for now, researchers are dazzled by the ease by which they
can make and test different CRISPR variants and by the technologys unexplored
potential. Charpentier and others are
looking at the versions of Cas9 in other
bacteria that might work better than the
one now being used. Microbiologists have
harnessed the CRISPR system to vaccinate
bacteria against the spread of antibiotic
resistance genes. Church, Doudna, Charpentier, and others are forming CRISPRrelated companies to begin exploring human therapeutic applications, including
gene therapy.
And theres more that can be done, Barrangou says. The only limitation today is
peoples ability to think of creative ways to
harness [CRISPR].
Not bad for a system that started with
sickly bacteria.
17

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

Multiplex genome engineering using

CRISPR/Cas systems
Le Cong,1,2* F. Ann Ran,1,4,* David Cox,1, 3 Shuailiang Lin,1, 5 Robert Barretto,6 Naomi
Habib,1 Patrick D. Hsu,1,4 Xuebing Wu,7 Wenyan Jiang,8 Luciano A. Marraffini,8
Feng Zhang1
Functional elucidation of causal genetic variants and elements requires precise genome
editing technologies. The type II prokaryotic CRISPR (clustered regularly interspaced
short palindromic repeats)/Cas adaptive immune system has been shown to facilitate
RNA-guided site-specific DNA cleavage. We engineered two different type II CRISPR/Cas
systems and demonstrate that Cas9 nucleases can be directed by short RNAs to induce
precise cleavage at endogenous genomic loci in human and mouse cells. Cas9 can also
be converted into a nicking enzyme to facilitate homology-directed repair with minimal
mutagenic activity. Lastly, multiple guide sequences can be encoded into a single CRISPR
array to enable simultaneous editing of several sites within the mammalian genome,
demonstrating easy programmability and wide applicability of the RNA-guided nuclease
technology.

recise and efficient genome-targeting

technologies are needed to enable systematic reverse engineering of causal
genetic variations by allowing selective
perturbation of individual genetic elements. Although genome-editing technologies such as designer zinc fingers (ZFs) (14),
transcription activatorlike effectors (TALEs)

(410), and homing meganucleases (11) have

begun to enable targeted genome modifications, there remains a need for new technologies that are scalable, affordable, and easy to
engineer. Here, we report the development of
a class of precision genome-engineering tools
based on the RNA-guided Cas9 nuclease (12
14) from the type II prokaryotic clustered reg-

A RT I C L E S | R E P O RT

ularly interspaced short palindromic repeats

(CRISPR) adaptive immune system (1518).
The Streptococcus pyogenes SF370 type II
CRISPR locus consists of four genes, including the Cas9 nuclease, as well as two noncoding CRISPR RNAs (crRNAs): trans-activating
crRNA (tracrRNA) and a precursor crRNA
(pre-crRNA) array containing nuclease guide
sequences (spacers) interspaced by identical
direct repeats (DRs) (fig. S1) (19). We sought

to harness this prokaryotic RNA-programmable nuclease system to introduce targeted

double-stranded breaks (DSBs) in mammalian
chromosomes through heterologous expression of the key components. It has been previously shown that expression of tracrRNA,
pre-crRNA, host factor ribonuclease (RNase)
III, and Cas9 nuclease is necessary and suf-

ficient for cleavage of DNA in vitro (12, 13)

and in prokaryotic cells (20, 21). We codonoptimized the S. pyogenes Cas9 (SpCas9) andRNase III (SpRNase III) genes and attached
nuclear localization signals (NLSs) to ensure
nuclear compartmentalization in mammalian
cells. Expression of these constructs in human 293FT cells revealed that two NLSs are

most efficient at targeting SpCas9 to the nucleus (Fig. 1A). To reconstitute the noncoding
RNA components of the S. pyogenes type II
CRISPR/Cas system, we expressed an 89-nucleotide (nt) tracrRNA (fig. S2) under the RNA
polymerase III U6 promoter (Fig. 1B). Similarly, we used the U6 promoter to drive the
expression of a pre-crRNA array comprising a

Broad Institute of MIT and Harvard, 7 Cambridge Center,

Cambridge, MA 02142, USA, and McGovern Institute for Brain
Research, Department of Brain and Cognitive Sciences,
Department of Biological Engineering, Massachusetts Institute
of Technology (MIT), Cambridge, MA 02139, USA.
2
Program in Biological and Biomedical Sciences, Harvard
Medical School, Boston, MA 02115, USA.
3
Harvard-MIT Health Sciences and Technology, Harvard
Medical School, Boston, MA 02115, USA.
4
Department of Molecular and Cellular Biology, Harvard
University, Cambridge, MA 02138, USA.
5
School of Life Sciences, Tsinghua University, Beijing
100084, China.6Department of Biochemistry and Molecular
Biophysics, College of Physicians and Surgeons, Columbia
University, New York, NY 10032, USA.
7
Computational and Systems Biology Graduate Program
and Koch Institute for Integrative Cancer Research,
Massachusetts Institute of Technology, Cambridge, MA 02139,
USA.
8
Laboratory of Bacteriology, The Rockefeller University, 1230
York Avenue, New York, NY 10065, USA.
*These authors contributed equally to this work.
Corresponding author. E-mail: zhang@broadinstitute.org
1

Fig. 1. The type II

CRISPR locus from
S. pyogenes SF370
can be reconstituted
in mammalian cells
to facilitate targeted
DSBs of DNA. (A)
Engineering of SpCas9
and SpRNase III with
NLSs enables import
into the mammalian
nucleus. GFP indicates
green fluorescent
protein; scale bars, 10
m. (B) Mammalian
expression of human
codonoptimized
SpCas9 (hSpCas9) and
SpRNase III (hSpRNase
III) genes were driven
by the elongation factor 1 (EF1) promoter,
whereas tracrRNA
and pre-crRNA array
(DR-Spacer-DR) were
driven by the U6 promoter. A protospacer
(blue highlight) from
the human EMX1 locus
with PAM was used as
template for the spacer
in the pre-crRNA array.
(C) Schematic
representation of base pairing between target locus and EMX1-targeting crRNA. Red arrow indicates putative cleavage site. (D) SURVEYOR assay for
SpCas9-mediated indels. (E) An example chromatogram showing a microdeletion, as well as representative sequences of mutated alleles identified from
187 clonal amplicons. Red dashes, deleted bases; red bases, insertions or mutations.

Fig. 2. SpCas9 can be reprogrammed to target multiple genomic loci

in mammalian cells. (A) Schematic of the human EMX1 locus showing the
location of five protospacers indicated by blue lines with corresponding
PAM in magenta. (B) Schematic of the pre-crRNA:tracrRNA complex (top)
showing hybridization between the direct repeat (gray) region of the precrRNA and tracrRNA. Schematic of a chimeric RNA design (12) (bottom).

tracrRNA sequence is shown in red and the 20-bp spacer sequence in blue.
(C) SURVEYOR assay comparing the efficacy of Cas9-mediated cleavage
at five protospacers in the human EMX1 locus. Each protospacer was
targeted by using either processed pre-crRNA:tracrRNA complex (crRNA)
or chimeric RNA (chiRNA). Arrowheads indicate cleavage products for
each protospacer target.

Fig. 3. Evaluation of the SpCas9 specificity and comparison of

efficiency with TALENs. (A) EMX1-targeting chimeric crRNAs with
single point mutations were generated to evaluate the effects of
spacer-protospacer mismatches. (B) SURVEYOR assay comparing

the cleavage efficiency of different mutant chimeric RNAs. (C)

Schematic showing the design of TALENs that target EMX1. (D)
SURVEYOR gel comparing the efficiency of TALEN andSpCas9
(N = 3).

18

SCIENCE sciencemag.org

Originally published 15 February 2013

sciencemag.org SCIENCE

19

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | R E P O RT

Fig. 4. Applications of Cas9 for homologous

recombination and multiplex genome engineering. (A) Mutation of the RuvC I domain
converts Cas9 into a nicking enzyme (SpCas9n).
HNH, histidine-asparagine-histidine endonuclease
domain. (B) Coexpression of EMX1-targeting chimeric RNA with SpCas9 leads to indels, whereas
SpCas9n does not (N = 3). (C) Schematic representation of the recombination strategy. A
homology repair (HR) template is designed to
insert restriction sites into EMX1 locus. Primers
used to amplify the modified region are shown as
red arrows. (D) Restriction fragment length polymorphism gel analysis. Arrows indicate fragments
generated by HindIII digestion. (E) Example chromatogram showing successful recombination. (F)
SpCas9 can facilitate multiplex genome modification by using a crRNA array that contains two
spacers targeting EMX1 and PVALB. Schematic
showing the design of the crRNA array (top). Both
spacers mediate efficient protospacer cleavage
(bottom). (G) SpCas9 can be used to achieve precise genomic deletion. Two spacers targeting EMX1
(top) mediated a 118-bp genomic deletion (bottom).

single guide spacer flanked by DRs (Fig. 1B).

We designed our initial spacer to target a 30
base pair (bp) site (protospacer) in the human
EMX1 locus that precedes an NGG trinucleotide, the requisite protospacer-adjacent motif
(PAM) (Fig. 1C and fig. S1) (22, 23).
To test whether heterologous expression
of the CRISPR system (SpCas9, SpRNase III,
tracrRNA, and pre-crRNA) can achieve targeted cleavage of mammalian chromosomes,
we transfected 293FT cells with different combinations of CRISPR/Cas components. Because DSBs in mammalian DNA are partially
repaired by the indel-forming nonhomologous
end joining (NHEJ) pathway, we used the
SURVEYOR assay (fig. S3) to detect endogenous target cleavage (Fig. 1D and fig. S2B).
Cotransfection of all four required CRISPR
components resulted in efficient cleavage of
the protospacer (Fig. 1D and fig. S2B), which
20

was subsequently verified by Sanger sequencing (Fig. 1E). SpRNase III was not necessary
for cleavage of the protospacer (Fig. 1D), and
the 89-nt tracrRNA is processed in its absence
(fig. S2C). Similarly, maturation of pre-crRNA
does not require RNase III (Fig. 1D and fig.
S4), suggesting that there may be endogenous
mammalian RNases that assist in pre-crRNA
maturation (2426). Removing any of the remaining RNA or Cas9 components abolished
the genome cleavage activity of the CRISPR/
Cas system (Fig. 1D). These results define a
minimal three-component system for efficient
RNA-guided genome modification in mammalian cells.
Next, we explored the generalizability of
RNA-guided genome editing in eukaryotic
cells by targeting additional protospacers
within the EMX1 locus (Fig. 2A). To improve
codelivery, we designed an expression vec-

tor to drive both pre-crRNA and SpCas9

(fig. S5). In parallel, we adapted a chimeric
crRNA-tracrRNA hybrid (Fig. 2B, top) design
recently validated in vitro (12), where a mature crRNA is fused to a partial tracrRNA via
a synthetic stem loop to mimic the natural
crRNA:tracrRNA duplex (Fig. 2B, bottom). We
observed cleavage of all protospacer targets
when SpCas9 is coexpressed with pre-crRNA
(DR-spacer-DR) and tracrRNA. However, not
all chimeric RNA designs could facilitate
cleavage of their genomic targets (Fig. 2C and
table S1). We then tested targeting of additional genomic loci in both human and mouse
cells by designing pre-crRNAs and chimeric
RNAs targeting the human PVALB and the
mouse Th loci (fig. S6). We achieved efficient
modification at all three mouse Th and one
PVALB targets by using the crRNA:tracrRNA
duplex, thus demonstrating the broad applisciencemag.org SCIENCE

cability of the CRISPR/Cas system in modifying different loci across multiple organisms
(table S1). For the same protospacer targets,
cleavage efficiencies of chimeric RNAs were
either lower than those of crRNA:tracrRNA
duplexes or undetectable. This may be due
to differences in the expression and stability
of RNAs, degradation by endogenous RNA
interference machinery, or secondary structures leading to inefficient Cas9 loading or
target recognition.
Effective genome editing requires that
nucleases target specific genomic loci with
both high precision and efficiency. To investigate the specificity of RNA-guided genome
modification, we analyzed single-nucleotide
mismatches between the spacer and its mammalian protospacer target (Fig. 3A). We observed that single-base mismatch up to 11 bp
5 of the PAM completely abolished genomic
cleavage by SpCas9, whereas spacers with
mutations farther upstream retained activity
against the protospacer target (Fig. 3B). This
is consistent with previous bacterial and in
vitro studies of Cas9 specificity (12, 20). Furthermore, SpCas9 is able to mediate genomic
cleavage as efficiently as a pair of TALE nucleases (TALENs) targeting the same EMX1
protospacer (Fig. 3, C and D).
Targeted modification of genomes ideally
avoids mutations arising from the errorprone NHEJ mechanism. The wild-type SpCas9 is able to mediate site-specific DSBs,
which can be repaired through either NHEJ
or homology-directed repair (HDR). We engineered an aspartate-to-alanine substitution (D10A) in the RuvC I domain of SpCas9
to convert the nuclease into a DNA nickase (SpCas9n, Fig. 4A) (12, 13, 20), because
nicked genomic DNA is typically repaired
either seamlessly or through high-fidelity
HDR. SURVEYOR (Fig. 4B) and sequencing
of 327 amplicons did not detect any indels
induced by SpCas9n. However, nicked DNA
can in rare cases be processed via a DSB intermediate and result in a NHEJ event (27).
We then tested Cas9-mediated HDR at the
same EMX1 locus with a homology repair
template to introduce a pair of restriction
sites near the protospacer (Fig. 4C). SpCas9
and SpCas9n catalyzed integration of the
repair template into EMX1 locus at similar
levels (Fig. 4D), which we further verified via
Sanger sequencing (Fig. 4E). These results

SCIENCE sciencemag.org

demonstrate the utility of CRISPR for facilitating targeted genomic insertions. Given
the 14-bp (12 bp from the seed sequence and
2 bp from PAM) target specificity (Fig. 3B)
of the wild-type SpCas9, the use of a nickase
may reduce off-target mutations.
Lastly, the natural architecture of CRISPR
loci with arrayed spacers (fig. S1) suggests
the possibility of multiplexed genome engineering. By using a single CRISPR array
encoding a pair of EMX1- and PVALB-targeting spacers, we detected efficient cleavage at both loci (Fig. 4F). We further tested
targeted deletion of larger genomic regions
through concurrent DSBs by using spacers
against two targets within EMX1 spaced
by 119 bp and observed a 1.6% deletion efficacy (3 out of 182 amplicons, Fig. 4G),
thus demonstrating the CRISPR/Cas system
can mediate multiplexed editing within a
single genome.
The ability to use RNA to program sequence-specific DNA cleavage defines a new
class of genome engineering tools. Here, we
have shown that the S. pyogenes CRISPR
system can be heterologously reconstituted
in mammalian cells to facilitate efficient
genome editing; an accompanying study
has independently confirmed high-efficiency
RNA-guided genome targeting in several human cell lines (28). However, several aspects
of the CRISPR/Cas system can be further
improved to increase its efficiency and versatility. The requirement for an NGG PAM
restricts the target space of SpCas9 to every
8 bp on average in the human genome (fig.
S7), not accounting for potential constraints
posed by crRNA secondary structure or genomic accessibility resulting from chromatin
and DNA methylation states. Some of these
restrictions may be overcome by exploiting
the family of Cas9 enzymes and its differing
PAM requirements (22, 23) across the microbial diversity (17). Indeed, other CRISPR loci
are likely to be transplantable into mammalian cells; for example, the Streptococcus thermophilus LMD-9 CRISPR1 system can also
mediate mammalian genome cleavage (fig.
S8). Lastly, the ability to carry out multiplex
genome editing in mammalian cells enables
powerful applications across basic science,
biotechnology, and medicine (29).
R E F E R E N C ES A ND NOTES
1. M. H. Porteus, D. Baltimore, Science 300, 763 (2003).

2. J. C. Miller et al., Nat. Biotechnol. 25, 778 (2007).

3. J. D. Sander et al., Nat. Methods 8, 67 (2011).
4. A. J. Wood et al., Science 333, 307 (2011).
5. M. Christian et al., Genetics 186, 757 (2010).
6. F. Zhang et al., Nat. Biotechnol. 29, 149 (2011).
7. J. C. Miller et al., Nat. Biotechnol. 29, 143 (2011).
8. D. Reyon et al., Nat. Biotechnol. 30, 460 (2012).
9. J. Boch et al., Science 326, 1509 (2009).
10. M. J. Moscou, A. J. Bogdanove, Science 326, 1501
(2009).
11. B. L. Stoddard, Q. Rev. Biophys. 38, 49 (2005).
12. M. Jinek et al., Science 337, 816 (2012).
13. G. Gasiunas, R. Barrangou, P. Horvath, V. Siksnys,

Proc. Natl. Acad. Sci. U.S.A. 109, E2579 (2012).
14. J. E. Garneau et al., Nature 468, 67 (2010).
15. H. Deveau, J. E. Garneau, S. Moineau, Annu. Rev.
Microbiol. 64, 475 (2010).
16. P. Horvath, R. Barrangou, Science 327, 167 (2010).
17. K. S. Makarova et al., Nat. Rev. Microbiol. 9, 467 (2011).
18. D. Bhaya, M. Davison, R. Barrangou, Annu. Rev. Genet.

45, 273 (2011).
19. E. Deltcheva et al., Nature 471, 602 (2011).
20. R. Sapranauskas et al., Nucleic Acids Res. 39, 9275
(2011)
21. A. H. Magadn, M. E. Dupuis, M. Villion, S. Moineau,

PLOS ONE 7, e40913 (2012).
22. H. Deveau et al., J. Bacteriol. 190, 1390 (2008).
23. F. J. Mojica, C. Dez-Villaseor, J. Garca-Martnez,

C. Almendros, Microbiology 155, 733 (2009).
24. M. Jinek, J. A. Doudna, Nature 457, 405 (2009).
25. C. D. Malone, G. J. Hannon, Cell 136, 656 (2009).
26. G. Meister, T. Tuschl, Nature 431, 343 (2004).
27. M. T. Certo et al., Nat. Methods 8, 671 (2011).
28. P. Mali et al., Science 339, 823 (2013).
29. P. A. Carr, G. M. Church, Nat. Biotechnol. 27, 1151
(2009).
AC K NOW L E DGMENTS

We thank the entire Zhang lab for their support and advice; P. A.
Sharp for generous help with Northern blot analysis; C. Jennings, R.
Desimone, and M. Kowalczyk for helpful comments; and X. Ye for help
with confocal imaging. L.C. and X.W. are Howard Hughes Medical
Institute International Student Research Fellows. D.C. is supported
by the Medical Scientist Training Program. P.D.H. is a James Mills
Pierce Fellow. X.W. is supported by NIH grants R01-GM34277 and
R01-CA133404 to P. A. Sharp,X.W.s thesis adviser. L.A.M. is supported by Searle Scholars, R. Allen, an Irma T. Hirschl Award, and a
NIH Directors New Innovator Award (DP2AI104556). F.Z. is
supported by a NIH Directors Pioneer Award (DP1MH100706);
the Keck, McKnight, Gates, Damon Runyon, Searle Scholars,
Klingenstein, and Simons foundations; R. Metcalfe; M. Boylan;
and J. Pauley. The authors have no conflicting financial
interests. A patent application has been filed relating to this
work, and the authors plan on making the reagents widely
available to the academic community through Addgene
and to provide software tools via the Zhang lab Web site
(www.genome-engineering.org).
SU PPL E M ENTARY MAT ERIALS

www.sciencemag.org/cgi/content/full/science.1231143/DC1
Materials and Methods
Figs. S1 to S8
Tables S1 and S2
References (3032)
5 October 2012; accepted 12 December 2012
Published online 3 January 2013;
10.1126/science.1231143

21

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

RNA-guided human genome

engineering via Cas9
Prashant Mali,1* Luhan Yang,1,3* Kevin M. Esvelt,2 John Aach,1 Marc Guell,1
James E. DiCarlo,4 Julie E. Norville,1 George M. Church,1,2
Bacteria and archaea have evolved adaptive immune defenses, termed clustered regularly
interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) systems,
that use short RNA to direct degradation of foreign nucleic acids. Here, we engineer the
type II bacterial CRISPR system to function with custom guide RNA (gRNA) in human
cells. For the endogenous AAVS1 locus, we obtained targeting rates of 10 to 25% in
293T cells, 13 to 8% in K562 cells, and 2 to 4% in induced pluripotent stem cells. We
show that this process relies on CRISPR components; is sequence-specific; and, upon
simultaneous introduction of multiple gRNAs, can effect multiplex editing of target loci.
We also compute a genome-wide resource of ~190 K unique gRNAs targeting ~40.5%
of human exons. Our results establish an RNA-guided editing tool for facile, robust, and
multiplexable human genome engineering.

acterial and archaeal clustered regularly

interspaced short palindromic repeats
(CRISPR) systems rely on CRISPR RNAs
(crRNAs) in complex with CRISPR-associated (Cas) proteins to direct degradation of complementary sequences present
within invading viral and plasmid DNA (13).
A recent in vitro reconstitution of the Streptococcus pyogenes type II CRISPR system
demonstrated that crRNA fused to a normally
trans-encoded tracrRNA is sufficient to direct
Cas9 protein to sequence-specifically cleave
target DNA sequences matching the crRNA
(4). The fully defined nature of this twocomponent system suggested that it might
function in the cells of eukaryotic organisms
such as yeast, plants, and even mammals. By
cleaving genomic sequences targeted by RNA
sequences (46), such a system could greatly
enhance the ease of genome engineering.
Here, we engineer the protein and RNA
components of this bacterial type II CRISPR
system in human cells. We began by synthesizing a human codonoptimized version of
the Cas9 protein bearing a C-terminal SV40
nuclear localization signal and cloning it into
a mammalian expression system (Fig. 1A and
fig. S1A). To direct Cas9 to cleave sequences
of interest, we expressed crRNA-tracrRNA
fusion transcripts, hereafter referred to as
guide RNAs (gRNAs), from the human U6
polymerase III promoter. Directly transcribing
gRNAs allowed us to avoid reconstituting the
RNA-processing machinery used by bacterial
CRISPR systems (Fig. 1A and fig. S1B) (4, 79).
Constrained only by U6 transcription initiatDepartment of Genetics, Harvard Medical School, Boston, MA 02115,
USA.
2
Wyss Institute for Biologically Inspired Engineering, Harvard
University, Cambridge, MA 02138, USA.
3
Biological and Biomedical Sciences Program, Harvard Medical
School, Boston, MA 02115, USA.
4
Department of Biomedical Engineering, Boston University, Boston,
MA 02215, USA.
*These authors contributed equally to this work.

Corresponding author. E-mail: gchurch@genetics.med.

harvard.edu
1

22

ing with G and the requirement for the PAM

(protospacer-adjacent motif ) sequence -NGG
following the 20base pair (bp) crRNA target,
our highly versatile approach can, in principle,
target any genomic site of the form GN20GG
(fig. S1C; see supplementary text S1 for a detailed discussion).
To test the functionality of our implementation for genome engineering, we developed
a green fluorescent protein (GFP) reporter assay (Fig. 1B) in human embryonic kidney HEK
293T cells similar to one previously described
(10). Specifically, we established a stable cell
line bearing a genomically integrated GFP
coding sequence disrupted by the insertion
of a stop codon and a 68-bp genomic fragment from the AAVS1 locus that renders the
expressed protein fragment nonfluorescent.
Homologous recombination (HR) using an appropriate repair donor can restore the normal
GFP sequence, which enabled us to quantify
the resulting GFP+ cells by flow-activated cell
sorting (FACS).
To test the efficiency of our system at stimulating HR, we constructed two gRNAs, T1 and
T2, that target the intervening AAVS1 fragment (Fig. 1B) and compared their activity
to that of a previously described TAL effector nuclease heterodimer (TALEN) targeting
the same region (11). We observed successful
HR events using all three targeting reagents,
with gene correction rates using the T1 and T2
gRNAs approaching 3% and 8%, respectively
(Fig. 1C). This RNA-mediated editing process
was notably rapid, with the first detectable
GFP+ cells appearing ~20 hours post transfection compared with ~40 hours for the AAVS1
TALENs. We observed HR only upon simultaneous introduction of the repair donor, Cas9
protein, and gRNA, which confirmed that all
components are required for genome editing (fig. S2). Although we noted no apparent
toxicity associated with Cas9/gRNA expression, work with zinc finger nucleases (ZFNs)
and TALENs has shown that nicking only one
strand further reduces toxicity. Accordingly,
Originally published 15 February 2013

A RT I C L E S | R E P O RT

we also tested a Cas9D10A mutant that is

known to function as a nickase in vitro, which
yielded similar HR but lower nonhomologous
end joining (NHEJ) rates (fig. S3) (4, 5). Consistent with (4), in which a related Cas9 protein
is shown to cut both strands 3 bp upstream
of the PAM, our NHEJ data confirmed that
most deletions or insertions occurred at the 3
end of the target sequence (fig. S3B). We also
confirmed that mutating the target genomic
site prevents the gRNA from effecting HR at
that locus, which demonstrates that CRISPRmediated genome editing is sequence-specific
(fig. S4). Finally, we showed that two gRNAs
targeting sites in the GFP gene, and also three
additional gRNAs targeting fragments from
homologous regions of the DNA methyl transferase 3a (DNMT3a) and DNMT3b genes could
sequence-specifically induce significant HR
in the engineered reporter cell lines (figs. S5
and S6). Together, these results confirm that
RNA-guided genome targeting in human cells
is simple to execute and induces robust HR
across multiple target sites.
Having successfully targeted an integrated
reporter, we next turned to modifying a native
locus. We used the gRNAs described above to
target the AAVS1 locus located in the PPP1R12C
gene on chromosome 19, which is ubiquitously
expressed across most tissues (Fig. 2A). We
targeted 293Ts, human chronic myelogenous
leukemia K562 cells, and PGP1 human induced
pluripotent stem (iPS) cells (12) and analyzed
the results by next-generation sequencing of
the targeted locus. Consistent with our results
for the GFP reporter assay, we observed high
numbers of NHEJ events at the endogenous
locus for all three cell types. The two gRNAs
T1 and T2 achieved NHEJ rates of 10 and 25%
in 293Ts, 13 and 38% in K562s, and 2 and 4%
in PGP1-iPS cells, respectively (Fig. 2B). We
observed no overt toxicity from the Cas9 and
gRNA expression required to induce NHEJ in
any of these cell types. As expected, NHEJmediated deletions for T1 and T2 were centered around the target site positions, which
further validated the sequence-specificity of
this targeting process (figs. S7 to S9). Simultaneous introduction of both T1 and T2 gRNAs
resulted in high-efficiency deletion of the intervening 19-bp fragment (fig. S8), which demonstrated that multiplexed editing of genomic
loci is feasible using this approach.
Last, we attempted to use HR to integrate
either a double-stranded DNA donor construct
(13) or an oligo donor into the native AAVS1
locus (Fig. 2C and fig. S10). We confirmed HRmediated integration, using both approaches,
by polymerase chain reaction (PCR) (Fig. 2D
and fig. S10) and Sanger sequencing (Fig. 2E).
We also readily derived 293T or iPS clones
from the pool of modified cells using puromycin selection over 2 weeks (Fig. 2F and fig.
S10). These results demonstrate that this approach enables efficient integration of foreign
DNA at endogenous loci in human cells.
Our versatile RNA-guided genome-editing
sciencemag.org SCIENCE

Fig. 1. Genome editing in human cells using an engineered type

II CRISPR system. (A) RNA-guided gene targeting in human cells
involves coexpression of the Cas9 protein bearing a C-terminal SV40
nuclear localization signal (NLS) with one or more gRNAs expressed
from the human U6 polymerase III promoter. Cas9 unwinds the DNA
duplex and cleaves both strands upon recognition of a target sequence
by the gRNA, but only if the correct PAM is present at the 3 end. Any
genomic sequence of the form GN20GG can, in principle, be targeted.
CMV, cytomegalovirus promoter; TK, thymidine kinase; pA, polyadenylation signal. (B) A genomically integrated GFP coding sequence is

system can be readily adapted to modify

other genomic sites by simply modifying the
sequence of our gRNA expression vector to
match a compatible sequence in the locus
of interest. To facilitate this process, we bioinformatically generated ~190,000 specific
gRNA-targetable sequences targeting ~40.5%
exons of genes in the human genome (refer to
methods and table S1). We also incorporated
these target sequences into a 200-bp format
compatible with multiplex synthesis on DNA
arrays (14) (fig. S11 and tables S2 and S3). This
resource provides a ready genome-wide reference of potential target sites in the human genome and a methodology for multiplex gRNA
synthesis.
Our results demonstrate the promise of
CRISPR-mediated gene targeting for RNAguided, robust, and multiplexable mammalian
genome engineering. The ease of retargeting our system to modify genomic sequences
greatly exceeds that of comparable ZFNs and
TALENs, while offering similar or greater efficiencies (4). Existing studies of type II CRISPR
SCIENCE sciencemag.org

disrupted by the insertion of a stop codon and a 68-bp genomic fragment from the AAVS1 locus. Restoration of the GFP sequence by HR with
an appropriate donor sequence results in GFP+ cells that can be quantified by FACS. T1 and T2 gRNAs target sequences within the AAVS1
fragment. Binding sites for the two halves of the TALEN are underlined.
(C) Bar graph depicting HR efficiencies induced by T1, T2, and TALENmediated nuclease activity at the target locus, as measured by FACS.
Representative FACS plots and microscopy images of the targeted cells
are depicted below. (Scale bar, 100 m.) Data are shown as means
SEM (N = 3).

specificity (4) suggest that target sites must

perfectly match the PAM sequence NGG and
the 8- to 12-base seed sequence at the 3 end
of the gRNA. The importance of the remaining 8 to 12 bases is less well understood and
may depend on the binding strength of the
matching gRNAs or on the inherent tolerance
of Cas9 itself. Indeed, Cas9 will tolerate single
mismatches at the 5 end in bacteria and in
vitro, which suggests that the 5 G is not required. Moreover, it is likely that the target
locuss underlying chromatin structure and
epigenetic state will also affect the efficiency
of genome editing in eukaryotic cells (13), although we suspect that Cas9s helicase activity
may render it more robust to these factors, but
this remains to be evaluated. Elucidating the
frequency and underlying causes of off-target
nuclease activity (15, 16) induced by CRISPR,
ZFN (17, 18), and TALEN (19, 20) genome-engineering tools will be of utmost importance
for safe genome modification and perhaps for
gene therapy. Potential avenues for improving
CRISPR specificity include evaluating Cas9

homologs identified through bioinformatics

and directed evolution of these nucleases toward higher specificity. Similarly, the range
of CRISPR-targetable sequences could be
expanded through the use of homologs with
different PAM requirements (9) or by directed
evolution. Finally, inactivating one of the Cas9
nuclease domains increases the ratio of HR
to NHEJ and may reduce toxicity (figs. S1A
and fig. S3) (4, 5), whereas inactivating both
domains may enable Cas9 to function as a
retargetable DNA binding protein. As we explore these areas, we note that another parallel study (21) has independently confirmed
the high efficiency of CRISPR-mediated gene
targeting in mammalian cell lines. We expect
that RNA-guided genome targeting will have
broad implications for synthetic biology (22,
23), the direct and multiplexed perturbation
of gene networks (13, 24), and targeted ex vivo
(2527) and in vivo gene therapy (28).
R E F E R E NCES AND NOT ES

1. B. Wiedenheft, S. H. Sternberg, J. A. Doudna, Nature

23

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | P E R S P E C T I V E

A Swiss army knife of immunity

Stan J. J. Brouns

482, 331 (2012).

2. D. Bhaya, M. Davison, R. Barrangou, Annu. Rev. Genet.
45, 273 (2011).
3. M. P. Terns, R. M. Terns, Curr. Opin. Microbiol. 14, 321
(2011).
4. M. Jinek et al., Science 337, 816 (2012).
5. G. Gasiunas, R. Barrangou, P. Horvath, V. Siksnys, Proc.

Natl. Acad. Sci. U.S.A. 109, E2579 (2012).
6. R. Sapranauskas et al., Nucleic Acids Res. 39, 9275
(2011).
7. T. R. Brummelkamp, R. Bernards, R. Agami, Science

296, 550 (2002).
8. M. Miyagishi, K. Taira, Nat. Biotechnol. 20, 497 (2002).
9. E. Deltcheva et al., Nature 471, 602 (2011).
10. J. Zou, P. Mali, X. Huang, S. N. Dowey, L. Cheng, Blood
118, 4599 (2011).
11. N. E. Sanjana et al., Nat. Protoc. 7, 171 (2012).
12. J. H. Lee et al., PLoS Genet. 5, e1000718 (2009).
13. D. Hockemeyer et al., Nat. Biotechnol. 27, 851 (2009).

24

HR at the AAVS1 locus, and the locations of sequencing primers (arrows)

for detecting successful targeted events, are depicted. (D) PCR assay 3 days
after transfection demonstrates that only cells expressing the donor, Cas9 and
T2 gRNA exhibit successful HR events. (E) Successful HR was confirmed by
Sanger sequencing of the PCR amplicon, which showed that the expected DNA
bases at both the genome-donor and donor-insert boundaries are present. (F)
Successfully targeted clones of 293T cells were selected with puromycin for 2
weeks. Microscope images of two representative GFP+ clones is shown. (Scale
bar, 100 m.)

14. S. Kosuri et al., Nat. Biotechnol. 28, 1295 (2010).

15. V. Pattanayak, C. L. Ramirez, J. K. Joung, D. R. Liu,

Nat. Methods 8, 765 (2011).
16. N. M. King, O. Cohen-Haguenauer, Mol. Ther. 16, 432
(2008).
17. Y. G. Kim, J. Cha, S. Chandrasegaran, Proc. Natl. Acad.

Sci. U.S.A. 93, 1156 (1996).
18. E. J. Rebar, C. O. Pabo, Science 263, 671 (1994).
19. J. Boch et al., Science 326, 1509 (2009).
20. M. J. Moscou, A. J. Bogdanove, Science 326, 1501
(2009).
21. L. Cong et al., Science 339, 819 (2013).
22. A. S. Khalil, J. J. Collins, Nat. Rev. Genet. 11, 367
(2010).
23. P. E. Purnick, R. Weiss, Nat. Rev. Mol. Cell Biol.
10, 410 (2009).
24. J. Zou et al., Cell Stem Cell 5, 97 (2009).
25. N. Holt et al., Nat. Biotechnol. 28, 839 (2010).
26. F. D. Urnov et al., Nature 435, 646 (2005).

27. A. Lombardo et al., Nat. Biotechnol. 25, 1298

(2007).
28. H. Li et al., Nature 475, 217 (2011).
ACK N OW LE D G M E N TS
This work was supported by NIH grant
P50 HG005550. We thank S. Kosuri for advice on the
oligonucleotide pool designs and synthesis. G.M.C. and
P.M.have filed a patent based on the findings of this study.
SUP P LE M E N TARY M AT E RIALS
www.sciencemag.org/cgi/content/full/science.1232033/DC1
Materials and Methods
Supplementary Text
Figs. S1 to S11
Tables S1 to S3
References (2946)

26 October 2012; accepted 12 December 2012

Published online 3 January 2013;
10.1126/science.1232033

sciencemag.org SCIENCE

CREDIT: B. STRAUCH/SCIENCE

Fig. 2. RNA-guided genome editing of the native AAVS1 locus in multiple cell types. (A) T1 (red) and T2 (green) gRNAs target sequences in an
intron of the PPP1R12C gene within the chromosome 19 AAVS1 locus. (B)
Total count and location of deletions caused by NHEJ in 293Ts, K562s, and
PGP1 iPS cells after expression of Cas9 and either T1 or T2 gRNAs as quantified by next-generation sequencing. Red and green dashed lines demarcate
the boundaries of the T1 and T2 gRNA targeting sites. NHEJ frequencies for
T1 and T2 gRNAs were 10% and 25% in 293T, 13% and 38% in K562, and
2% and 4% in PGP1 iPS cells, respectively. (C) DNA donor architecture for

elfish genetic elements are more than

a daily nuisance in
the life of prokaryotes.
Whereas viruses can
multiply by reprogramming host cells, or integrate in the host genome
as stowaways, conjugative plasmids (transferrable
extrachromosomal
DNA)
make cells addicted to
plasmid-encoded antitoxin
factors, thus preventing
their disposal. Bacteria and
archaea defend themselves
against these invasive elements using an adaptive
immune system based on
clustered regularly interspaced short palindromic
repeats (CRISPRs). On page
[30 in this booklet], Jinek et
al. (1) show how the CRISPR Fig. 1. All-in-one nuclease. (A) Cas9 requires a crRNA and
effector enzyme Cas9 from tracrRNA to recognize invader DNA sequences by hybridizing
bacteria is directed not by the guide section of the crRNA to one strand of the target DNA to
one, but two small RNAs to form an R-loop. The flanking motif is critical for this process and
may facilitate DNA duplex unwinding and strand invasion by the
cleave invader DNA.
The
CRISPR
system crRNA. Target DNA is then cleaved by both nuclease domains of
integrates
short
DNA Cas9. (B) Cascade-like complexes contain a single crRNA and up
fragments from viruses to five different Cas proteins. Identified invader DNA sequences
and plasmids into a specific are progressively unwound and cleaved by the action of the
repeat locus of the host cell recruited nuclease and helicase Cas3 (11, 12).
genome to function as a
memory of past invasions. This locus of the
Cas systems involve a dedicated nuclease that
cells most wanted is then transcribed into
cleaves the precursor CRISPR RNA in each
RNA (the precursor CRISPR RNA), which
repeat (2), Cas9-based systems also require
is cleaved in each repeat to yield individual
a CRISPR-specific small RNA. This so-called
mature CRISPR RNAs (crRNAs). These
trans-activated crRNA (tracrRNA) base pairs
guide a dedicated set of CRISPR-associated
with each repeat of the CRISPR transcript and
(Cas) proteins to their targets during cellular
provides a substrate for the RNA-specific host
surveillance of the cytoplasm for either
ribonuclease RNase III (6). The cleavage prodforeign DNA or messenger RNA (mRNA)
uct, an RNA hybrid consisting of a 42-nucleoof known invaders. Once identified, foreign
tide crRNA and a 75-nucleotide tracrRNA, was
nucleic acids are permanently damaged
deemed to be the guide for Cas9.
by Cas nucleases, thereby neutralizing the
With this in mind, Jinek et al. could show
invader (2).
that Cas9 from the human pathogenic bacThe CRISPR field was set in motion 5 years
terium Streptococcus pyogenes binds and
ago by the discovery that lactic acid bacteria
cleaves invader DNA within the remembered
become highly resistant to virus infection
region. Although the site specificity was
when they incorporate virus DNA fragments
solely determined by the guiding ability of
in their array of memorized invaders (3).
the crRNA, binding and cleavage of the tarBacterial resistance to the virus is based on
get DNA surprisingly required the tracrRNA.
breaks in the viral DNA within this memoThe tracrRNA thus enables the Cas9-crRNA
rized region, and the bacterial gene cas9 encomplex to locate a DNA sequence complecodes the enzyme responsible (4, 5). However,
mentary to the crRNA in the cellular tangle
the modus operandi of Cas9 has remained
of DNA (see the figure), providing yet another
unknown.
example of the crucial roles that small RNAs
One aspect that had to be resolved first
play in cells (7).
was the unusual way in which Cas9 obtains
Cas9 creates blunt-ended lesions in target
the mature crRNA. Whereas most CRISPRDNA by using two nuclease domains, each

SCIENCE sciencemag.org

Originally published 17 August 2012

cleaving one DNA strand of the target doublestranded DNA R-loop. Whereas the HNHnuclease domain cleaves the DNA strand that
base pairs with the crRNA, the RuvC-nuclease
domain cleaves the displaced strand of the
DNA. Jinek et al. show that cleavage was robust and occurred with multiple turnovers in
both relaxed and supercoiled DNA targets, implying that Cas9 is functionally recycled after
cleavage to destroy more invader DNA copies
that may be present in the host cell.
Despite the seeming efficiency of this cleavage and recycling process, the Achilles heel
of Cas9 was also uncovered. Viruses escape
immunity by making point mutations in either the memorized regions of their genomes
(8), or just outside this region in a conserved
nucleotide motif. When testing these mutant
DNA molecules, Jinek et al. found that binding and cleavage by Cas9 was severely compromised, suggesting that these mutated virus
DNA molecules adopt a stealth mode inside
the cell and require a new cycle of memory
formation before they are subject to interference once again. Cycles like these contribute
to the ongoing coevolution between invaders
and their hosts.
Jinek et al. realized that a highly specific,
customizable RNA-directed DNA nuclease
could be useful to edit whole genomes. Based
on the 20-nucleotide guide section of the
crRNA, the enzyme could theoretically introduce breaks at unique sites in any eukaryotic
genome. As a proof of concept, the authors
programmed Cas9 to cleave a plasmid carrying the gene encoding green fluorescent
protein at predetermined loci using a single
chimeric crRNA containing just the critical
segment of the tracrRNA. DNA breaks induce
cellular DNA repair pathways (9) and this can
be harnessed to disrupt, insert, or repair specific genes of cells. Introducing DNA breaks
at desired loci using just Cas9 and a chimeric
crRNA would be a substantial improvement
over existing gene-targeting technologies,
such as zinc finger nucleases and transcription activatorlike effector nucleases, as these
require protein engineering for every new target locus (10). Efficient gene repair strategies
in cells from patients, and the reintroduction
of repaired cells, could become increasingly
important for treating many genetic disorders.
Cas9 is thus a remarkably compact
and multifunctional enzyme compared to
CRISPR effector complexes from other bacteria or archaea. These are typically 350- to
450-kD crRNA-protein complexes and contain up to 11 protein subunits encoded by
four to seven different cas genes (see the
figure) (2, 11). Yet, nuclease activities are
not always part of the complex and need to
be recruited when the target DNA is identified (12). With all its activities at hand, Cas9
Laboratory of Microbiology, Wageningen University, 6703 HB
Wageningen, Netherlands.
E-mail: stan.brouns@wur.nl

25

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

is truly the Swiss army knife of CRISPR

immunity.
REF ERENC ES AND NOTES

1. M. Jinek et al., Science 337, 816 (2012); 10.1126/

science.1225829.
2. D. Bhaya, M. Davison, R. Barrangou, Annu. Rev.
Genet. 45, 273 (2011).
3. R. Barrangou et al., Science 315, 1709 (2007).
4. J. E. Garneau et al., Nature 468, 67 (2010).
5. R. Sapranauskas et al., Nucleic Acids Res. 39, 9275

(2011).
6. E. Deltcheva et al., Nature 471, 602 (2011).
7. G. Storz, J. Vogel, K. M. Wassarman, Mol. Cell
43, 880 011).
8. H. Deveau et al., J. Bacteriol. 190, 1390 (2008).
9. E. C. Friedberg, Nature 421, 436 (2003).
10. M. Baker, Nat. Methods 9, 23 (2012).
11. B. Wiedenheft, S. H. Sternberg, J. A. Doudna,
Nature 482, 331 (2012).
12. E. R. Westra et al., Mol. Cell 46, 595 (2012).
10. 1126/science.1227253

CRISPR/Cas, the immune system of

bacteria and archaea
Philippe Horvath1* and Rodolphe Barrangou2*
Microbes rely on diverse defense mechanisms that allow them to withstand viral predation and exposure to invading nucleic acid. In many bacteria and most archaea, clustered
regularly interspaced short palindromic repeats (CRISPR) form peculiar genetic loci,
which provide acquired immunity against viruses and plasmids by targeting nucleic acid
in a sequence-specific manner. These hypervariable loci take up genetic material from
invasive elements and build up inheritable DNA-encoded immunity over time. Conversely,
viruses have devised mutational escape strategies that allow them to circumvent the
CRISPR/Cas system, albeit at a cost. CRISPR features may be exploited for typing purposes, epidemiological studies, host-virus ecological surveys, building specific immunity
against undesirable genetic elements, and enhancing viral resistance in domesticated
microbes.

icrobes have devised various strategies that allow them to survive exposure to foreign genetic elements.
Although outpopulated and preyed
upon by abundant and ubiquitous viruses, microbes routinely survive, persist, and
occasionally thrive in hostile and competitive
environments. The constant exposure to exogenous DNA via transduction, conjugation, and
transformation have forced microbes to establish an array of defense mechanisms that allow
the cell to recognize and distinguish incoming
foreign DNA, from self DNA and to survive
exposure to invasive elements. These systems
maintain genetic integrity, yet occasionally allow exogenous DNA uptake and conservation
of genetic material advantageous for adaptation to the environment. Certain strategies,
such as prevention of adsorption, blocking of
injection, and abortive infection, are effective
against viruses; other defense systems specifically target invading nucleic acid, such as the
restriction-modification system (R-M) and the
use of sugar-nonspecific nucleases. Recently,
Danisco France SAS, BP10, F-86220 Dang-Saint-Romain,
France.
2
Danisco USA, Inc., 3329 Agriculture Drive, Madison,
WI 53716, USA.
*Corresponding author. E-mail: philippe.horvath@danisco.
com (P.H.), rodolphe.barrangou@danisco.com (R.B.)
1

26

an adaptive microbial immune system, clustered regularly interspaced short palindromic

repeats (CRISPR) has been identified that
provides acquired immunity against viruses
and plasmids.
CRISPR represents a family of DNA repeats found in most archaeal (~90%) and
bacterial (~40%) genomes (13). Although
the initial discovery of a CRISPR structure
was made fortuitously in Escherichia coli in
1987, the acronym was coined in 2002, after
similar structures were observed in genomes
of various bacteria and archaea (1). CRISPR
loci typically consist of several noncontiguous
direct repeats separated by stretches of variable sequences called spacers (which mostly
correspond to segments of captured viral and
plasmid sequences) and are often adjacent
to cas genes (CRISPR-associated) (Fig. 1). cas
genes encode a large and heterogeneous family of proteins that carry functional domains
typical of nucleases, helicases, polymerases,
and polynucleotide-binding proteins (4).
CRISPR, in combination with Cas proteins,
forms the CRISPR/Cas systems. Six core
cas genes have been identified, including the
universal markers of CRISPR/Cas systems
cas1 (COG1518) and cas2 (COG1343, COG3512,
occasionally in a fused form with other cas
genes). Besides the cas1 to cas6 core genes,
Originally published 8 January 2010

A RT I C L E S | R E V I E W

subtype-specific genes and genes encoding repeat-associated mysterious proteins

(RAMP) have been identified and grouped
into subtypes functionally paired with particular CRISPR repeat sequences (48). The
size of CRISPR repeats and spacers varies between 23 to 47 base pairs (bp) and 21 to 72
bp, respectively. Generally, CRISPR repeat sequences are highly conserved within a given
CRISPR locus, but a large assortment of repeat sequences has been shown across microbial species (1, 9). Most repeat sequences are
partially palindromic, having the potential
to form stable, highly conserved secondary
structures (7). The number of repeat-spacer
units is documented to reach 375 (Chloroflexus sp. Y-400-fl), but most loci commonly
contain fewer than 50 units, as exemplified
in lactic acid bacteria genomes (8). Microbes
may contain more than one CRISPR locus; up
to 18 such loci have been identified in Methanocaldococcus jannaschii, totaling more than
1% of the genome (10). CRISPRs are typically
located on the chromosome, although some
have been identified on plasmids (1113).
The CRISPR loci have highly diverse and
hypervariable spacer sequences, even between
closely related strains (1416), which were initially exploited for typing purposes. A variety
of putative roles for CRISPR sequences was
originally suggested, including chromosomal
rearrangement, modulation of expression of
neighboring genes, target for DNA binding
proteins, replicon partitioning, and DNA repair (5). In 2005, three independent in silico
studies reported homology between spacer
sequences and extrachromosomal elements,
such as viruses and plasmids (11, 14, 15). This
led to the hypothesis that CRISPR may provide adaptive immunity against foreign genetic elements (6).
A vast spectrum of immunity
In 2007, it was shown in Streptococcus thermophilus that during natural generation of
phage-resistant variants, bacteria commonly
alter their CRISPR loci by polarized (i.e., at
the leader end) incorporation of CRISPR repeat-spacer units (Fig. 1) (17, 18), consistent
with observed spacer hypervariability at the
leader end of CRISPR loci in various strains
(14, 16). The integrated sequences were identical to those of the phages used in the challenge, which suggested that they originate
from viral nucleic acid. To determine whether
CRISPR impacts phage resistance, spacer
content was altered via genetic engineering,
which showed that spacer addition can provide novel phage resistance, whereas spacer
deletion could result in loss of phage resistance (17). These findings were confirmed in
Streptococcus mutans, where phage-resistant
mutants acquired novel CRISPR spacers with
sequences matching the phage genome, in vitro and in vivo (19). Although the ubiquitous
and predatory nature of phages may explain
the overwhelming representation of phage sesciencemag.org SCIENCE

Fig. 1. Overview of the four CRISPR/cas systems present in Streptococcus thermophilus

DGCC7710. For each system, gene organization is depicted on the top, with cas genes in gray, and
the repeat-spacer array in black. Below the gene scheme, the repeat and spacer (captured phage or
plasmid nucleic acid) content is detailed as black diamonds (T, terminal repeat) and white rectangles,
respectively. Bottom line, consensus repeat sequence. L1 to L4, leader sequences. The predicted
secondary structure of the CRISPR3 repeat is shown on the right. S. thermophilus CRISPR2, CRISPR3,
and CRISPR4 systems are homologous to the CRISPR systems of Staphylococcus epidermidis (20),
Streptococcus mutans (19), and E. coli (28), respectively.

quences in CRISPR loci, CRISPR spacers can

also interfere with both plasmid conjugation
and transformation, as shown in Staphylococcus epidermidis (20). Furthermore, several
metagenomic studies investigating host-virus
populations dynamics showed that CRISPR
loci evolve in response to viral predation and
that CRISPR spacer content and sequential
order provide insights both historically and
geographically (2124).
The ability to provide defense against invading genetic elements seems to render
CRISPR/Cas systems particularly desirable
in hostile environments and may explain
their propensity to be transferred horizontally between sometimes distant organisms
(12). There is extensive evidence that defense
systems such as CRISPR have undergone
horizontal transfer between genomes, notably differences observed in codon bias, GC
content variability, their presence on mobile
genetic elements, the presence of neighboring
insertion sequence elements, and their variable presence and location in closely related
SCIENCE sciencemag.org

genomes. This is in agreement with the lack

of congruence between the phylogenetic relation of various CRISPR elements and that of
the organisms in which they are found (8, 12).
This horizontal gene transfer may be mediated by plasmids, megaplasmids, and even
prophages, all of which are documented to
carry CRISPR loci (2).
Given the variety of defense systems in
microbes and their role in controlling the
presence of plasmids, prophages, transposons, and, perhaps, chromosomal sequences,
studies should investigate whether CRISPR/
Cas systems preferentially target certain elements and could determine whether they are
symbiotic or mutually exclusive with other
defense systems.
Idiosyncrasies of the CRISPR/Cas
mechanism of action
The mechanism by which CRISPR provides
resistance against foreign genetic elements is
not fully characterized (Fig. 2). Even so, the
functional link between Cas and CRISPR re-

peats has been inferred from the congruence

observed between their sequence patterns.
cas genes provide CRISPR-encoded immunity,
because inactivating the CRISPR1-associated
cas7 gene (Fig. 1) impairs the ability of the
host to integrate novel CRISPR spacers after
phage exposure (17), which suggests that it is
necessary for recognizing foreign nucleic acid
and/or integrating the novel repeat-spacer
unit. Cas1 appears to be a double-stranded
DNA (dsDNA) endonuclease involved in the
immunization process (25). It has also been
proposed that Cas2 may act as a sequencespecific endoribonuclease that cleaves uracilrich single-stranded RNAs (ssRNAs) (26). The
mechanistic steps involved in invasive element recognition, novel repeat manufacturing, and spacer selection and integration into
the CRISPR locus remain uncharacterized.
Although some Cas proteins are involved
in the acquisition of novel spacers, others
provide CRISPR-encoded phage resistance
and interfere with invasive genetic elements.
Mechanistically, although defense is spacerencoded, the information that lies within
the CRISPR repeat-spacer array becomes
available to the Cas machinery through
transcription. The CRISPR leader, defined
as a low-complexity, A/T-rich, noncoding sequence, located immediately upstream of the
first repeat, likely acts as a promoter for the
transcription of the repeat-spacer array into
a CRISPR transcript, the pre-crRNA (13, 27).
The full-length pre-crRNA is subsequently
processed into specific small RNA molecules
that correspond to a spacer flanked by two
partial repeats (2729). In E. coli, processing
is achieved by a multimeric complex of Cas
proteins named Cascade (CRISPR-associated
complex for antiviral defense), which specifically cleaves the pre-crRNA transcript within
the repeat sequence to generate small CRISPR
RNAs, crRNAs (28). Similarly, in Pyrococcus,
Cas6 is an endoribonuclease that cleaves the
pre-crRNA transcript into crRNA units that
include a partial [8-nucleotide (nt)] repeat sequence at the 5 end, as part of the Cas-crRNA
complex (27, 29, 30). The crRNAs seem to specifically guide the Cas interference machinery
toward foreign nucleic acid molecules that
match its sequence, which leads ultimately to
degradation of the invading element (30). The
involvement of cas genes in CRISPR defense
was originally demonstrated when inactivating the CRISPR1-associated csn1-like gene
(Fig. 1) resulted in loss of phage resistance
despite the presence of matching spacers (17).
The observation that CRISPR spacers
match both sense and antisense viral DNA
led to the hypothesis that some CRISPR/
Cas systems may target dsDNA, and this
was confirmed by disruption of target
DNA with an intron (the excision of which
restores the native mRNA) on a plasmid that
allows conjugation despite the presence of a
matching CRISPR spacer (20). Conversely,
the Pyrococcus CRISPR effector complex, a
27

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

A RT I C L E S | R E V I E W

ribonucleoprotein complex that consists of

crRNA and Cas proteins, targets invader RNA
by complementary-dependent cleavage, in
vitro (30). Given the large diversity of CRISPR/
Cas systems in bacteria and archaea (4, 6), it is
likely that both DNA and RNA may be targets.
More information is needed to establish and
understand what the functional differences
are among distinct CRISPR/Cas systems.
The initial hypothesis that CRISPR may
mediate microbial immunity via RNA interference (RNAi) (6) is misguided. RNAi allows
eukaryotic organisms exposed to foreign genetic material to silence the invading nucleic
acid sequence before or after it integrates into
the host chromosome, and/or to subvert cellular processes through a small interfering
RNA guide (31). A key difference between
RNAi and CRISPR-encoded immunity lies in
the enzymatic machinery involved. Although
both are mediated by a guide RNA in an inhibitory ribonucleoprotein complex, only
Dicer, Slicer, and the RNA-induced silencing
complex (RISC) may have analogous counterparts (6, 30). Mechanistically, although the
short RNA duplexes at the core of RNAi are
typically 21 to 28 nt in length (32), crRNAs
are larger, because they contain a CRISPR
spacer (23 to 47 nt) flanked by partial repeats.
Also, RNA-dependent transcription generating dsRNA and using the cleaved target RNA
seen in RNAi have not been characterized in
the CRISPR/Cas systems. In other ways, the
sequence-specific and adaptive CRISPR/Cas
systems share similarities with the vertebrate
adaptive immune system, although CRISPR
spacers are DNA-encoded and can be inherited by the progeny.
Circumventing CRISPR-based immunity
Even though CRISPR can provide high
levels of phage resistance, a relatively small
proportion of viruses retain the ability to
infect the immunized host. These viral
particles have specifically mutated the protospacer (sequence within the invading nucleic
acid that matches a CRISPR spacer), with a
single point mutation that allows the viruses
to overcome immunity, which indicates that
the selective pressure imposed by CRISPR can
rapidly drive mutation patterns in viruses (17,
18, 23). Analysis of phage sequences adjacent
to proto-spacers revealed the presence of conserved sequences, called CRISPR motifs (13, 16,
18, 19, 33, 34), or proto-spacer adjacent motifs
(PAMs) (35). Phages may also circumvent the
CRISPR/Cas system by mutating the CRISPR
motif (18), which indicates that it is involved
in CRISPR-encoded immunity. Additionally,
CRISPR motif mutation can result in loss of
phage resistance despite the presence of a
matching CRISPR spacer (34). The absence of
this motif in the CRISPR locus likely allows
the system to act on the invading target DNA
specifically and precludes an autoimmune
response on the host chromosome (Fig. 2).
Such a motif may not be necessary in CRISPR/
28

Fig. 2. Overview of the CRISPR/Cas mechanism of action. (A) Immunization process: After insertion of exogenous DNA from viruses or plasmids, a Cas complex recognizes foreign DNA and integrates a novel repeat-spacer unit at the leader end of the CRISPR locus. (B) Immunity process: The
CRISPR repeat-spacer array is transcribed into a pre-crRNA that is processed into mature crRNAs,
which are subsequently used as a guide by a Cas complex to interfere with the corresponding invading nucleic acid. Repeats are represented as diamonds, spacers as rectangles, and the CRISPR leader
is labeled L.

Cas systems targeting RNA. Although protospacers seem to be randomly located on phage
genomes, a given CRISPR spacer may be acquired independently by different lineages.
It is thus tempting to speculate that CRISPR
motifs also play a key role in the selection
of spacers.
These mutations may have an impact on
the amino acid sequence, as either nonsynonymous mutations or premature stop codons
that truncate the viral protein (18). In addition to mutations, phages may also circumvent CRISPR-encoded immunity via deletion
of the target sequence (18, 21). This perhaps

indicates a strong cost associated with circumventing the CRISPR/Cas systems. Alternative strategies that allow viruses to escape
CRISPR, such as suppressors that could interfere with crRNAs biogenesis or Cas machinery
remain uncovered. Defense tactics employed
by viruses to circumvent the CRISPR/Cas
systems are yet another critical difference between RNAi and CRISPR: Eukaryotic viruses
may express inhibitors such as dsRNA-binding
proteins that interfere with the RNA silencing
machinery (32), which are yet to be identified
in response to CRISPR, whereas microbial
viruses specifically mutate or recombine (21)
sciencemag.org SCIENCE

the sequence corresponding

to the CRISPR spacer or that
of the PAM.
The impact of CRISPR on
phage genomes is illustrated
by extensive genome recombination events observed in
environmental phage populations in response to CRISPR
(21). This contrasts with the
fact that acquisition of novel
CRISPR spacers does not seem
to have a fitness cost for the
host, apart from maintaining
the CRISPR/Cas system as
active.
Although it seems intuitive
that CRISPR loci should not
be able to expand indefinitely
(21, 36), little is known about
the parameters that define
the optimal and maximum
size of a CRISPR locus. Also,
Fig. 3. CRISPR interference. The CRISPR/Cas systems may
the fitness cost of CRISPR
target either DNA or RNA to interfere with viruses, plasmids,
expansion in the host should
prophages, or other chromosomally encoded sequences.
be compared with that of
CRISPR evasion in the virus
populations, so as to determine whether prey
increase the selective pressure on invading
or predators incur the higher evolutionary
elements and, consequently, could increase
cost of this genetic warfare.
the chances of host survival by using multiple
Although CRISPR loci primarily evolve
hurdles.
via polarized addition of novel spacers at the
Because CRISPR spacers correspond to
leader end of the locus after phage exposure,
prior episodes of phage and plasmid expointernal spacer deletions have also been resure, they provide a historical and geographiported, likely occurring via homologous recalalthough limitedperspective as to
combination between CRISPR repeats (1, 16,
the origin and paths of a particular strain,
18). Perhaps this allows the host to limit the
which may be used for ecological and epiexpansion of the CRISPR locus so that the
demiological studies. Many intrinsic aspects
relative size of the locus does not increase to
of CRISPR-based immunity have provided
a detrimental level. The propensity of spacavenues for industrial applications, includers located at the trailer end (opposite to the
ing exploiting hypervariability for typing
leader end) to be deleted preferentially would
purposes, driving viral evolution, predicting
mitigate the loss of fitness associated with the
and modulating virus resistance in domesdeletion, because ancestral spacers would articated microbes, and performing natural
guably provide resistance against viruses that
genetic tagging of proprietary strains. The
were historically, but are not currently, presinheritable nature of the CRISPR spacer conent in the environment. The combination of
tent provides potential for perennial use of
locus expansion via spacer acquisition and
industrial microbes. Alternatively, the abilcontraction via spacer loss, in the context of
ity of CRISPR/Cas systems to impede the
rapid evolution in space and time because of
transfer of particular nucleic acid sequences
viral predation, which generate a high level of
(such as phage or plasmid DNA) into a host
spacer polymorphism, suggests that CRISPR
might be exploited via genetic engineering
loci undergo dynamic and rapid turnover on
to specifically preclude the dissemination of
evolutionary time scales (16, 21, 36). Indeed, in
undesirable genetic elements, such as antibimicrobes with an active system, CRISPR loci
otic-resistance markers and genes harmful to
have been shown to be the most hypervariable
humans and other living organisms. It may
genomic regions (21).
also be designed to limit the intracellular
spread of mobile genetic elements such as
Applications and future directions
insertion sequences and transposons. In adA priori, the concurrent presence of disdition to providing immunity, CRISPR/Cas
tinct defense systems against foreign genetic
systems that target RNA have the potential to
elements in bacteria and archaea seems ineffiaffect the transcript stability of chromosomal
cient and redundant, although it might reflect
elements (Fig. 3).
functional preferences and increased fitness.
Although significant progress has been
Because all defense mechanisms have their
made in the last few years, many mechanisadvantages and caveats, the accumulation
tic aspects remain uncovered, notably vis-and combination of different systems would
vis the immunization process (key elements
SCIENCE sciencemag.org

involved in spacer selection and integration

between repeats and/or possible involvement of degenerate infectious particles in
building immunity) and the interference
mechanism (other cellular components involved). Also, more knowledge is desirable
regarding the elements necessary to have
functional CRISPR/Cas systems and the
basis for the absence of CRISPR in 60% of
bacteria.

R E F E R ENCES AND NOT ES

1. R. Jansen, J. D. Embden, W. Gaastra, L. M. Schouls,

Mol. Microbiol. 43, 1565 (2002).
2. R. Sorek, V. Kunin, P. Hugenholtz, Nat. Rev. Microbiol.

6, 181 (2008).
3. J. van der Oost, M. M. Jore, E. R. Westra, M. Lundgren,

S. J. Brouns, Trends Biochem. Sci. 34, 401 (2009).
4. D. H. Haft, J. Selengut, E. F. Mongodin, K. E. Nelson,

PLOS Comput. Biol. 1, e60 (2005).
5. K. S. Makarova, L. Aravind, N. V. Grishin, I. B. Rogozin,

E. V. Koonin, Nucleic Acids Res. 30, 482 (2002).
6. K. S. Makarova, N. V. Grishin, S. A. Shabalina, Y. I.

Wolf, E. V. Koonin, Biol. Direct 1, 7 (2006).
7. V. Kunin, R. Sorek, P. Hugenholtz, Genome Biol. 8, R61
(2007).
8. P. Horvath et al., Int. J. Food Microbiol. 131, 62 (2009).
9. I. Grissa, G. Vergnaud, C. Pourcel, BMC Bioinformatics

8, 172 (2007).
10. R. K. Lillestl, P. Redder, R. A. Garrett, K. Brugger,

Archaea 2, 59 (2006).
11. F. J. Mojica, C. Dez-Villaseor, J. Garca-Martnez,

E. Soria, J. Mol. Evol. 60, 174 (2005).
12. J. S. Godde, A. Bickerton, J. Mol. Evol. 62, 718 (2006).
13. R. K. Lillestl et al., Mol. Microbiol. 72, 259 (2009).
14. C. Pourcel, G. Salvignol, G. Vergnaud, Microbiology

151, 653 (2005).
15. A. Bolotin, B. Quinquis, A. Sorokin, S. D. Ehrlich,

Microbiology 151, 2551 (2005).
16. P. Horvath et al., J. Bacteriol. 190, 1401 (2008).
17. R. Barrangou et al., Science 315, 1709 (2007).
18. H. Deveau et al., J. Bacteriol. 190, 1390 (2008).
19. J. R. van der Ploeg, Microbiology 155, 966 (2009).
20. L. A. Marraffini, E. J. Sontheimer, Science 322,

1843 (2008).
21. A. F. Andersson, J. F. Banfield, Science 320, 1047
(2008).
22. V. Kunin et al., Genome Res. 18, 293 (2008).
23. J. F. Heidelberg, W. C. Nelson, T. Schoenfeld, D.

Bhaya, N. Ahmed, PLOS ONE 4, e4169 (2009).
24. N. L. Held, R. J. Whitaker, Environ. Microbiol. 11, 457
(2009).
25. B. Wiedenheft et al., Structure 17, 904 (2009).
26. N. Beloglazova et al., J. Biol. Chem. 283, 20361
(2008).
27. C. Hale, K. Kleppe, R. M. Terns, M. P. Terns, RNA 14,

2572 (2008).
28. S. J. Brouns et al., Science 321, 960 (2008).
29. J. Carte, R. Wang, H. Li, R. M. Terns, M. P. Terns, Genes
Dev. 22, 3489 (2008).
30. C. R. Hale et al., Cell 139, 945 (2009).
31. C. C. Mello, D. Conte Jr., Nature 431, 338 (2004).
32. G. Meister, T. Tuschl, Nature 431, 343 (2004).
33. S. A. Shah, N. R. Hansen, R. A. Garrett, Biochem. Soc.
Trans. 37, 23 (2009).
34. E. Semenova, M. Nagornykh, M. Pyatnitskiy, I. I.

Artamonova, K. Severinov, FEMS Microbiol. Lett. 296,

110 (2009).
35. F. J. Mojica, C. Dez-Villaseor, J. Garca-Martnez,

C. Almendros, Microbiology 155, 733 (2009).
36. G. W. Tyson, J. F. Banfield, Environ. Microbiol. 10, 200
(2008).
37. We thank our colleagues P. Boyaval, C. Fremaux,

D. Romero, and E. Bech Hansen for their support and

scientific contributions, and S. Moineau, V. Siksnys,

and J. Banfield for their insights and expertise. This

work was supported by Danisco A/S. P.H. and R.B.

have submitted patent applications relating to various

uses of CRISPR.

10.1126/science.1179555

29

T E C H N I C A L N OT E

C RIS PR-C A S 9: ENG IN E E R I N G A R E V O LU TI O N I N GE N E E D I T I N G

Real-World Gene Editing Issues: Improving

Knock-In Efficiency

A B ST R AC T ON LY

Authors: Tom Henley, Ramu Mangena, Rohan Sivapalan, Eric Rhodes

A programmable dual-RNAguided
DNA endonuclease in adaptive
bacterial immunity
Martin Jinek, * Krzysztof Chylinski, * Ines Fonfara, Michael Hauer,
Jennifer A. Doudna,1,2,5,6 Emmanuelle Charpentier4
1,2

3,4

Clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas)

systems provide bacteria and archaea with adaptive immunity against viruses and plasmids by using
CRISPR RNAs (crRNAs) to guide the silencing of invading nucleic acids. We show here that in a subset
of these systems, the mature crRNA that is base-paired to trans-activating crRNA (tracrRNA) forms
a two-RNA structure that directs the CRISPR-associated protein Cas9 to introduce double-stranded
(ds) breaks in target DNA. At sites complementary to the crRNA-guide sequence, the Cas9 HNH
nuclease domain cleaves the complementary strand, whereas the Cas9 RuvC-like domain cleaves the
noncomplementary strand. The dual-tracrRNA:crRNA, when engineered as a single RNA chimera, also
directs sequence-specific Cas9 dsDNA cleavage. Our study reveals a family of endonucleases that use
dual-RNAs for site-specific DNA cleavage and highlights the potential to exploit the system for RNAprogrammable genome editing.

Howard Hughes Medical Institute (HHMI), University of

California, Berkeley, CA 94720, USA.
2
Department of Molecular and Cell Biology, University of
California, Berkeley, CA 94720, USA.
3
Max F. Perutz Laboratories (MFPL), University of Vienna,
A-1030 Vienna, Austria.
4
The Laboratory for Molecular Infection Medicine Sweden,
Ume Centre for Microbial Research, Department of
Molecular Biology, Ume University, S-90187 Ume,
Sweden.
5
Department of Chemistry, University of California,
Berkeley, CA 94720, USA.
1

Physical Biosciences Division, Lawrence Berkeley

National Laboratory, Berkeley, CA 94720, USA.
*These authors contributed equally to this work.
Present address: Friedrich Miescher Institute for
Biomedical Research, 4058 Basel, Switzerland.
Corresponding author.
E-mail: doudna@berkeley.edu (J.A.D.);
emmanuelle.charpentier@mims.umu.se (E.C.)
6

Read full article: scim.ag/1piiXv7

Genome editing has gotten a real boost from the recent

interest in CRISPR technology. Unfortunately while
many experimenters are having great success using
CRISPR to knock-out genes, the ability to use CRISPR for
knock-ins has proven somewhat more challenging.
The difficulties associated with efficiently generating knock-ins
can be attributed to several issues, but two are key: the need
to position the cut site close to the region of the intended
change and the need for effective delivery of a donor DNA
template. These constraints generally dont exist when doing
targeted knock-outs but present several real-world problems
when attempting to generate knock-ins.
It has been demonstrated that the efficiency of incorporation
of a targeted alteration is highly dependent on the distance
from the nuclease-induced cut site to the site of the desired
genomic alteration1. The closer these two are to each other,
the higher the efficiency, so it would make sense to use a guide
RNA (gRNA) that lies as close to the site of the desired alteration
as possible. However, the best gRNA to drive incorporation
may not be the most specific, leading to off-target effects, and
so a difficult choice needs to be made as to the right balance
to strike between these two factors.
It has also been demonstrated that not all gRNAs are equal in
activity. Therefore, the lack of sufficient activity by a given gRNA
might also require one to use a gRNA which lies somewhat
further away from the target site. It is even possible, in some
cases, that there simply is no suitable gRNA available that lies
close enough (within 40bp) to the target site.
To address the donor side of the equation, researchers have
mostly been using either single-stranded oligos or doublestranded plasmids to provide a donor template. Because of
Horizons long history of using rAAV (recombinant adeno-

Figure 1: An inactive GFP gene, split into exons,

driven by the CMV promoter is integrated as a
single copy into the genome. gRNA target sites
are shown flanking the mutation which renders
the GFP non-functional, with numbers representing the distance of each PAM site from the
mutation.

associated virus) to edit and create hundreds of isogenic lines,

we were interested to see whether the combination of rAAV
and CRISPR might offer advantages not necessarily realized
when using an oligo or plasmid alone.
It has been known for some time that rAAV is able to drive
homologous recombination levels that are up to 1,000fold higher than those seen when using a simple plasmid.
Despite this vast improvement over plasmids, the levels of
recombination seen are still relatively modest, making rAAV
optimal for single allele alterations, but less so when looking to
affect multiple alleles simultaneously. The efficiency of using
rAAV is boosted significantly however, when a double-strand
break is introduced in the vicinity of the homology region2.
In addition, rAAV has an extremely wide tropism making it
very effective for delivery of a single-stranded piece of long
DNA (approx. 4.8kb) directly to the nucleus of many cell types,
particularly those which may be difficult to transfect.
To test the potential for a combination approach, we set up an
experiment to look at more real-world situations where the
gRNA cut wasnt necessarily located directly over the desired
change and used a common cell line (HCT116) that is not
particularly easy to transfect. We used CRISPR to introduce
double strands breaks at various points in the genome of a line
carrying a disabled copy of the GFP gene on one chromosome.
The figure below shows the organization of the artificial GFP
gene split into three exons, with the third exon carrying a
mutation which prematurely terminates the GFP protein
rendering it non-functional. Conversion of the adenosine
(A) residue to a cysteine (C) restores GFP functionality. The
location and relative distance of five different gRNA targets
sites from the point mutation are shown below the expanded
Exon 3 region.

CMV Promoter

Originally published 17 August 2012

Exon 2

Exon 3

PolyA

A>C (GFP Switch)

Exon 3

Mouse Hbb-b2 intron 2-3

gRNA 2
-60bp

30

Exon 1

gRNA 1
26bp

gRNA 3
97bp

SV40 PolyA

gRNA 4
162bp

gRNA 5
454bp

T E C H NIC A L NOT E

3.00

Targeting Frequency (% of cells with intact GFP)

Figure 2: Plasmid and rAAV donors were compared

for their ability to knock-in the activating mutation
and restore GFP function through homology-directed repair. For each gRNA tested, rAAV donors provide
higher frequencies of correct gene targeting than
equivalent plasmids.

2.85

2.50

2.00
1.54
1.50

Plasmid

1.31

TM

rAAV
1.05

1.00

0.87
0.63

0.50

0.00

0.03

0.00

Plasmid only rAAV only

0.05

0.05

gRNA 1

gRNA 2

26bp

-60bp

0.11

0.07

gRNA 3

gRNA 4

gRNA 5

97bp

162bp

454bp

GENASSIST | CRISPR & rAAV Genome Editing Tools

A range of gene-editing kits and reagents to enable easier, robust implementation of CRISPR and rAAV gene-editing
experiments. The GENASSIST product and service solution includes:

Distance from GFP Mutation

Custom guide RNA manufacturing and Validation

We used both a plasmid and rAAV derived from that same

plasmid as a donor and measured the ability of these donors
to stimulate a genetic modification at the range of distances
from the induced cut site.

References

Remarkably the rAAV donor performs markedly better than the

corresponding plasmid with an average 10-fold improvement
at most distances. We are currently working to extend this
research and recent results comparing rAAV with oligo and
plasmid donors at different concentrations have supported
this trend.

Contact us to learn more about Horizons research:

L. Yang et al., Nucleic Acids Res. 41, 9049 (2013).

M. H. Porteus et al., Molec. Cell. Biol. 23, 3558 (2003).

Website:
Email:
Phone:

www.horizondiscovery.com
info@horizondiscovery.com
UK +44 (0)1223 655 580

Custom Donor Design and manufacturing

Both CRISPR and rAAV technologies for the right solution to your challenge
Adenovirus and rAAV packaging for hard to transfect cells
Consultation with our Gene-Editing Expert Scientists
Over 550 genetically defined Isogenic Cell Lines available for further genomic modification
And an extensive plasmid library!

We acknowledge that the data is representative of only one

cell line which does not transfect particularly well, but we feel
it does fairly represent a real-world situation which often arises
as experiments are designed in an ever widening variety of
cellular backgrounds. We and other groups continue to look
for technology combinations and other approaches that can
be used to boost gene targeting efficiencies and lower the
barriers to make even complex gene editing more routine.

Generate the cell line you want in your lab:

Point Mutations

Gene Knockouts

Deletions

Insertions

Translocations

Amplifications

Or choose any of our over 550 off the shelf X-MAN Isogenic
Cell Line pairs to start driving your research right away!

www.horizondiscovery.com/GENASSIST to learn more

Visit
about GENASSIST and Horizons many other products and services
precision genome editing

ZFN

AV
rA
TM

C RIS P R

Translating Genomes | Personalizing Medicine

Horizon Discovery combines long scientific heritage in translational research with GENESIS, a precision gene-editing
platform incorporating rAAV, CRISPR and ZFN technologies. Horizon supplies genetically-defined cell lines, gene-editing
tools and services, custom cell line generation, molecular reference standards, and contract research services to almost
1,000 academic, clinical and biopharmaceutical organizations.

Horizons precision gene-editing platform

As the only translational genomics company to have access to all three gene-editing technologies, Horizon is in a unique
position to be able to generate virtually any genomic modification upon customer request.

Horizons Capabilities:
GENASSIST
CRISPR & rAAV
gene-editing tools

Bioproduction

Custom Cell Line

Development
Validated customer-defined
cell lines

X-MAN Isogenic Cell Lines

Cell line development for

biomanufacturing

Off-the-shelf cell line pairs

Discovery Research Services

X-MAN Reporter Kits

Assay development, disease

modeling & compound profiling

Endogenous pathway reporters

CombinatoRx
Combination drug and other
biological molecule screening

Custom Screening
Services

HDx Reference
Standards
DNA and FFPE reference
standards

Cas9/sgRNA & siRNA screens to

ID targets in oncology and
other indications

For more information or to learn more about Horizons products & services:

Visit www.horizondiscovery.com