ne w s and vie w s
polymethine linkers mediated by hypochlorite short-lived oxidative burst and a more rapid
or peroxynitrite.
Unlike other probes of reactive oxygen
and nitrogen species, the new probe devel-
oped by Rao and colleagues1 combines two
sensors—a chemiluminescence resonance
energy transfer sensor and a fluorescence
resonance energy transfer sensor—in a
single particle called a CF-SPN. The sensors
interact in different ways with a near-infra-
red matrix made of PFODBT (poly(2,7-(9,
9-dioctylfluorene)-alt-4,7-bis(thiophen-
2-yl)benzo-2,1,3-thiadiazole)). PFODBT,
used here for the first time in a nanoparticle,
acts as both a chemiluminescence resonance
energy transfer acceptor and a fluorescence
resonance energy transfer donor (Fig. 1a).
The fluorescence sensor in CF-SPNs is
IR775S, as in the authors’ previous work5.
Because hepatocytes do not produce hypochlo-
rite, IR775S in the liver detects only the reactive
© 2014 Nature America, Inc. All rights reserved.
nitrogen species peroxynitrite. The reaction
abolishes the fluorescence resonance energy
transfer from PFODBT to IR775S that would
predominate in the absence of peroxyni-
trite. Thus, the ratio of PFODBT and IR775S
fluorescence emission signals can be used as
an accurate readout to probe in vivo nitrosative
stress conditions.
The chemiluminescence sensor, which detects
the reactive oxygen species hydrogen peroxide,
is bis-(2,4,5-trichloro-6-(pentyloxycarbonyl)
phenyl) oxalate (CPPO), a hydrophobic
peroxyoxalate molecule. CPPO reacts with
hydrogen peroxide to form the high-energy
intermediate 1,2-dioxetanedione, which gen-
erates light by sensitizing fluorophores in
close proximity, such as PFODBT and IR775S,
triggering chemiluminescence. Indeed, in vitro
npg
calibrations show that hydrogen peroxide
can selectively mediate chemiluminescence
resonance energy transfer from CPPO to both
near-infrared acceptors.
The authors target their probe to the mouse
liver using poly(ethylene glycol) and galactose
moieties that associate with asialoglycoprotein
receptors expressed on the sinusoidal mem-
brane of hepatocytes. They show that produc-
tion of hydrogen peroxide and peroxynitrite
can be readily detected in the liver after a
challenge with high doses of acetaminophen
or isoniazid, two commonly used drugs with
known toxicity.
Notably, the timing and extent of hydrogen
peroxide and peroxynitrite generation varies
between the two drugs, providing insights
into the different pathways and mecha-
nisms of toxicity that are induced by each
(Fig. 1b). In acetaminophen overdose, hydro-
gen peroxide generation precedes that of per-
oxynitrite. In isoniazid overdose, there is a
338 volume 32 number 4 APRIL 2014 nature biotechnology
research applications involving redox biology.
and sustained induction of nitrosative stress.
These subtle yet important differences would
undoubtedly be missed using insensitive
assays that rely on the detection of electro-
philic reactive metabolites through covalent
modification with endogenous nucleophiles
such as glutathione.
In the context of drug design, the reactive
oxygen and nitrogen species profiles detected
with CF-SPNs can act as indicators of drug
metabolism pathways, reporting on whether
phase I or phase II metabolism is involved
(Fig. 1b). In both, drug potency is reduced,
but in phase I metabolism this occurs through
the introduction of polar functionalities by
enzymes such as cytochrome P450 (CYP450),
whereas in phase II metabolism the drug is
conjugated to reactive species such as gluta-
thione, increasing its molecular weight.
Rao and colleagues1 also show clearly that
hydrogen peroxide and peroxynitrite produc-
tion in the liver can be detected before the onset
of histological changes that occur in hepato-
cellular degeneration, presumably through
CYP450 activation reactions, which are the
major, broad-spectrum scavenging pathways
for therapeutics taken into the body.
CF-SPNs offer opportunities for a diverse
array of studies in both clinical and basic
Cas9 in close-up
Erin L Garside & Andrew M MacMillan
Structural and biochemical studies elucidate DNA targeting by the Cas9
endonuclease.
In the past year, a prokaryotic adaptive immune
response—the clustered, regularly interspaced,
short palindromic repeat (CRISPR) system—
has been harnessed to effect genome target-
ing and engineering in many cell types and
organisms1–3. These studies involved RNA-
guided targeting of an endonuclease, Cas9, to
the desired genomic site. Ensuring the speci-
ficity of the Cas9-DNA interaction is critical
to many future applications of the CRISPR
system, particularly clinical applications, but
improving the specificity of the native endo-
nuclease will require a deeper structural and
functional understanding of its activity. Three
recent papers4–6 in Cell, Science and Nature
Erin L. Garside & Andrew M. MacMillan are in
the Department of Biochemistry, University of
Alberta, Edmonton, Canada.
e-mail: andrew.macmillan@ualberta.ca
They are versatile, biocompatible materials,
and because they are engineered entirely from
organic components and lack heavy metal ions,
they should have minimal toxicity. Beyond
drug toxicity testing, CF-SPNs may be helpful
in unraveling the etiology and pathophysiol-
ogy of many health conditions that involve
alterations of oxidative and nitrosative stress,
including chronic inflammatory disease, neu-
rodegenerative disease and cancer. Extension
to other imaging modalities, such as MRI, PET
or photoacoustic imaging, which are more
amenable to deep-tissue imaging, can also be
envisioned. As the work of Rao and colleagues1
suggests, combining new chemical detection
methods and translational research is a prom-
ising avenue for advancing the diagnosis of
human disease.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
1. Shuhendler, A.J., Pu, K., Cui, L., Uetrecht, J.P. & Rao, J.
Nat. Biotechnol. 32, 373–380 (2014).
2. Dickinson, B.C. & Chang, C.J. Nat. Chem. Biol. 7,
504–511 (2011).
3. Winterbourn, C.C. Nat. Chem. Biol. 4, 278–286
(2008).
4. Murphy, M.P. et al. Cell Metab. 13, 361–366
(2011).
5. Pu, K. et al. Nat. Nanotechnol. 9, 233–239 (2014).
now provide an important advance in this
regard through high-resolution crystal struc-
tures of Cas9, with or without bound RNA or
DNA, and biochemical analyses.
In many bacteria and almost all Archaea,
the CRISPR gene-silencing pathway7 incor-
porates short DNA sequences derived from a
foreign source, such as phage, into the genome.
Processing of primary transcripts containing
these sequences produces effector CRISPR
RNAs (crRNAs), which form endonuclease-
RNA complexes that cleave the foreign nucleic
acid, thus conferring resistance to the invader.
Type II CRISPR systems use a single, large
endonuclease, Cas9, which generates double-
strand breaks in target DNA. In eukaryotic
cells, double-strand DNA breaks are repaired
by nonhomologous end joining or homologous
recombination; these pathways can be exploited
to achieve gene knockout or gene editing.
 ne w s and vie w s

In endogenous type II CRISPR systems, a d 3’

5’
Target DNA
5’
3’
target recognition relies on complementar- RuvC REC1
crRNA
ity between a DNA sequence and the crRNA 180° 5’ 3’

complexed to a second trans-activating RNA 5’

tracrRNA
(tracrRNA). For greater efficiency, engineered
systems often use a fusion of these two RNAs, PI HNH 3’
referred to as single-guide RNA (sgRNA)8,9. REC2
The formation of a heteroduplex between the Arg-rich Cas 9-RNA complex
sgRNA and DNA target displaces the noncom-
plementary DNA strand, forming a structure b
known as an ‘R-loop’. Cas9 then catalyzes cleav- 5’ 3’
age of both DNA strands. Recognition of a short 5’
3’ 5’
sequence in the noncomplementary strand of 5’ 3’

the target DNA, called the protospacer adjacent PAM

motif (PAM), is vital for Cas9 targeting10. The

PAM
three new studies4–6 provide insights into Cas9 recognition
targeting, including the role of the PAM. 90°
Nishimasu et al.4 carried out X-ray crys-
tallographic structure determination of a 5’ 3’
5’
Streptococcus pyogenes Cas9-sgRNA complex 3’
5’
5’
3’
bound to a complementary single-stranded
© 2014 Nature America, Inc. All rights reserved.

PAM
DNA target. They find that Cas9 has a bifur-
cated structure in which central nucleic acid
R-loop
recognition (REC) domains form one half c REC1 formation
of a channel binding the sgRNA-target DNA
heteroduplex (Fig. 1a). The other half is 5’ 3’
5’
formed by the RuvC nuclease domain, which 3’ 5’
5’ 3’
cleaves the noncomplementary DNA strand
PAM
(not present in the structure), the HNH nucle-
ase domain, which cleaves the complemen- REC1
tary DNA strand, and the C-terminal PAM Cleavage of both
apo Complex DNA strands
interaction domain. An arginine-rich region
forms a long alpha helix that binds and posi- Figure 1 CRISPR-Cas9 targeting of DNA. (a) Surface representation of X-ray structure of S. pyogenes
tions the sgRNA (Fig. 1a). Cas9-sgRNA bound to target DNA (PDB 4OO8). Front (left) and back (right) views are depicted. The
target DNA strand is colored red and the sgRNA is colored light blue. Cas9 domains are colored blue
This overview of the domain arrangement
(RuvC), purple (also indicated by red arrow) (HNH), orange (REC1), gray (REC2), green (Arg-rich), and
of Cas9 is supported by the crystal structures yellow (PI, PAM interacting motif). (b) Positioning of RuvC and HNH nuclease domains in the complex.
presented by Jinek et al.5, who studied Cas9 (left) Surface representation of X-ray structure with HNH domain removed (top; compare to panel A,
from S. pyogenes and a smaller member of the left) reveals that this domain caps one side of the DNA-sgRNA binding channel also depicted in a 90°
Cas9 family from Actinomyces naeslundii. The
npg

Cas9 structures from these two species, crys-
tallized without nucleic acid (‘apoCas9’), show
that a functional core consisting of the RuvC
and HNH domains is conserved. However,
there are differences in the structures of the
PAM recognition and REC domains, consistent
with their association with distinct crRNAs and
tracrRNAs in the two species.
The work by Nishimasu et al.4 gives detailed
insight into Cas9-RNA contacts. Recognition
of the sgRNA-DNA heteroduplex is largely
mediated by residues in the REC1 domain
and in the arginine-rich helix, and is sequence
independent, as expected, given that Cas9 can
accommodate sgRNAs complementary to dif-
ferent DNA sequences. In contrast, recogni-
tion of the parts of the sgRNA outside of the
heteroduplex region is sequence-specific, con-
sistent with Cas9 associating only with RNAs
from closely related type II CRISPR systems.
The structures provided by Nishimasu
et al.4 and Jinek et al.5 also reveal conformational
rotation (bottom). (right) Disposition of RuvC and HNH active sites with respect to bound DNA-sgRNA
complex. Active sites are highlighted in yellow. The scissile bond of the target DNA is indicated with
a red arrow. (c) Ribbon representations of X-ray structures demonstrating conformational changes
occurring in the assembly of the Cas9-sgRNA-DNA complex PDB 4OO8, highlighting the location of
the nucleic acid binding channel (shaded) from apoCas9 (left; PDB 4CMP). Nucleic acid has been
omitted for clarity. (d) Schematic of target recognition and Cas9 activation. Initial recognition of the
PAM sequence adjacent to the DNA target is followed by directional formation of the DNA-sgRNA
heteroduplex and R-loop structure, including a conformational change in the complex before cleavage of
the target and complementary strands (black vertical arrows).
changes that occur upon binding of Cas9 to and bulk biochemical assays, including com-
RNA and DNA. The HNH nuclease domain can petitive binding and activity experiments, per-
adopt two conformations corresponding to inac- formed by Sternberg et al.6. Notably, they find
tive (Fig. 1b) and, possibly, active Cas9 (ref. 4). that Cas9 recognizes DNA in a stepwise fash-
Global conformational changes in the protein ion, initially sampling for PAM sites in three
form a nucleic acid–binding channel and ori- dimensions, rather than carrying out a one-
ent the active sites of the nuclease domains dimensional search along the DNA as has been
with respect to the target DNA (Fig. 1c). modeled for other DNA-binding proteins. In
These structures provide snapshots of Cas9, the absence of PAM recognition, the residency
but the missing link in understanding how this time of Cas9 on DNA is very short, suggest-
enzyme works is a comprehensive, dynamic ing that Cas9-mediated, off-target cleavage
view of its activity, particularly in relation to will be limited to sequences adjacent to a
PAM recognition. Part of the answer is pro- PAM. Once the PAM is bound, there is direc-
vided by a combination of single-molecule tional and sequential unwinding of the DNA

nature biotechnology volume 32 number 4 APRIL 2014 339

 ne w s and vie w s
double helix, proposed by Sternberg et al.6 to
be a Brownian ratchet motion that forms the
R-loop structure present in the active form of
the Cas9 complex (Fig. 1d).
Competition assays using DNAs of vary-
ing sequence and length show that the PAM is
required not only for stable Cas9-DNA asso-
ciation but also for efficient DNA cleavage. A
single-stranded DNA target is cleaved much
more slowly than a double-stranded DNA
substrate. In a hybrid substrate containing a
short double-stranded sequence adjacent to the
single-stranded target, the presence of a PAM
sequence in the noncomplementary strand
activates the target for cleavage. This is an
important observation as it suggests that there
might be a second PAM-mediated checkpoint
in target specificity. It also highlights the need
for further work to achieve a high-resolution
structural analysis of the whole Cas9-DNA-
RNA complex, including the PAM-containing
© 2014 Nature America, Inc. All rights reserved.
strand, to visualize whether a PAM-dependent
rearrangement is required for activation.
Although some tolerance for target mis-
match is no doubt desirable in the context of
phage resistance, the opposite is true for gene
editing or gene modification applications.
The three new studies provide the functional
and structural foundation for analyzing and
optimizing the specificity of Cas9 targeting.
Given the requirement of the PAM for both
targeting and activation, swapping individual
PAM interacting motifs among different Cas9
npg
340 volume 32 number 4 APRIL 2014 nature biotechnology
proteins is one possible approach. Another is 1. Cong, L. et al. Science 339, 819–823 (2013).
2. Mali, P. et al. Science 339, 823–826 (2013).
fusing Cas9 to other DNA binding domains 3. Wang, H. et al. Cell 153, 910–918 (2013).
as a means of enhancing specificity, similar 4. Nishimasu, H. et al. Cell 156, 935–949 (2014).
to what has been done for homing endonu- 5. Jinek, M. et al. Science doi:10.1126/science.1247997
(6 February 2014).
cleases11. As these examples indicate, there are 6. Sternberg, S.H. et al. Nature 507, 62–67 (2014).
many exciting directions for engineering Cas9 7. Wiedenheft, B., Sternberg, S.H. & Doudna, J.A. Nature
482, 331–338 (2012).
to modulate the specificity of DNA recognition
8. Deltcheva, E. et al. Nature 471, 602–607 (2011).
and cleavage. 9. Jinek, M. et al. Science 337, 816–821 (2012).
10. Sapranauskas, R. et al. Nucleic Acids Res. 39,
9275–9282 (2011).
COMPETING FINANCIAL INTERESTS 11. Boissel, S. et al. Nucleic Acids Res. 42, 2591–2601
The authors declare no competing financial interests. (2014).
Research Highlights
Papers from the literature selected by the Nature Biotechnology editors. (Follow us on
Twitter, @NatureBiotech #nbtHighlight)
Necroptosis drives motor neuron death in models of both sporadic and familial ALS
Re, D.B. et al. Neuron 81, 1001–1008 (2014)
Detection of circulating tumor DNA in early- and late-stage human malignancies
Bettegowda, C. et al. Sci. Transl. Med. 19, 224ra24 (2014)
Structure-based programming of lymph-node targeting in molecular vaccines
Liu, H. et al. Nature doi:10.1038/nature12978 (16 February 2014)
Highly multiplexed subcellular RNA sequencing in situ
Lee, J.H. et al. Science doi:10.1126/science.1250212 (27 February 2014)
Universal count correction for high-throughput sequencing
Hashimoto, T.B. et al. PLoS: Comput. Biol. doi:10.1371/journal.pcbi.1003494
(6 March 2014)
Expanding ester biosynthesis in Escherichia coli
Rodriguez, G.M. et al. Nat. Chem. Biol. doi:10.1038/nchembio.1476 (9 March 2014)