ARTICLE doi:10.

1038/nature14237

Integrase-mediated spacer acquisition

during CRISPR–Cas adaptive immunity
James K. Nuñez1, Amy S. Y. Lee1,2, Alan Engelman3 & Jennifer A. Doudna1,2,4,5,6

Bacteria and archaea insert spacer sequences acquired from foreign DNAs into CRISPR loci to generate immunological
memory. The Escherichia coli Cas1–Cas2 complex mediates spacer acquisition in vivo, but the molecular mechanism of
this process is unknown. Here we show that the purified Cas1–Cas2 complex integrates oligonucleotide DNA substrates
into acceptor DNA to yield products similar to those generated by retroviral integrases and transposases. Cas1 is the
catalytic subunit and Cas2 substantially increases integration activity. Protospacer DNA with free 39-OH ends and supercoiled
target DNA are required, and integration occurs preferentially at the ends of CRISPR repeats and at sequences adjacent to
cruciform structures abutting AT-rich regions, similar to the CRISPR leader sequence. Our results demonstrate the Cas1–
Cas2 complex to be the minimal machinery that catalyses spacer DNA acquisition and explain the significance of CRISPR
repeats in providing sequence and structural specificity for Cas1–Cas2-mediated adaptive immunity.

Prokaryotic adaptive immunity relies on clustered regularly interspaced EDTA and phenol–chloroform extraction or proteinase K in the pres-
short palindromic repeats (CRISPRs) together with CRISPR associated ence of EDTA and detergent (Extended Data Fig. 1e), indicating that
(Cas) proteins to detect and destroy foreign nucleic acids1,2. CRISPR loci product DNAs are unlikely to be bound to Cas1 and/or Cas2. Consis-
contain an AT-rich leader sequence followed by repetitive sequence ele- tent with product DNA resulting from covalent integration of proto-
ments flanking ,30 base pair (bp) spacer segments that are transcribed spacer DNA into the plasmid, the relaxed and linear forms of pCRISPR
to produce precursor CRISPR RNAs (pre-crRNAs)3–5. Spacers are fre- became radiolabelled in reactions containing 32P-labelled protospacer
quently virus- or plasmid-derived, although ‘self’-derived spacers from DNA (Fig. 1d and Extended Data Fig. 2). Although Cas1 alone catalysed
the host chromosome are present in some CRISPR loci6. After pre-crRNA a low level of protospacer integration in the presence of Mn21, the reac-
processing and assembly with Cas proteins, the resulting surveillance tion was enhanced substantially by the presence of Cas2 (Extended Data
complexes target and cleave foreign nucleic acids bearing sequences Fig. 2b).
complementary to the crRNA spacer sequence7–12. How spacer DNA Bacteria expressing Cas1 active-site mutants, but not Cas2 active-site
sequences, termed protospacers, are acquired into the host CRISPR locus mutants, are incapable of acquiring new spacers in vivo, demonstrating
remains unknown. the catalytic role of Cas1 during spacer acquisition13,14,16. Consistent with
Overexpression of Cas1 and Cas2 nucleases, the only Cas proteins these data, Cas1 active site mutants H208A and D221A were defective
found in all CRISPR–Cas systems, leads to the site-specific acquisition for protospacer integration in vitro, whereas the Cas2(E9Q) active-site
of 33 bp protospacers at the leader end of the CRISPR locus in E. coli13–15. mutant supported integration (Fig. 1c, e and Extended Data Fig. 3). The
Furthermore, Cas1 and Cas2 function as a complex in vivo16, suggest- Cas2 C-terminal (Db6–b7) deletion mutant, which is defective for com-
ing that the Cas1–Cas2 complex might possess DNA recombination plex formation with Cas1 and spacer acquisition in vivo, failed to sup-
activity. We reconstituted CRISPR spacer acquisition using purified Cas1 port Cas1-mediated integrase activity (Fig. 1c, e). We conclude that our
and Cas2 proteins, protospacers and acceptor plasmid DNA, revealing in vitro assay recapitulates the in vivo functions of Cas1 and Cas2 during
an elegant mechanism in which both the sequence and structural ele- spacer acquisition.
ments of the CRISPR repeats specify spacer integration sites.
Integration and disintegration products
Protospacer DNA integration by Cas1–Cas2 We tested whether the reaction products of Cas1–Cas2-mediated DNA
To test whether the Cas1–Cas2 complex is sufficient to catalyse DNA integration resemble those formed by the strand transfer activity of re-
recombination in vitro, assays were conducted using purified Cas1– troviral integrases and cut-and-paste transposases23–26. These enzymes
Cas2 complex, 33 bp protospacer DNA and an acceptor ‘target’ plas- generate two main products in vitro corresponding to half-site and full-
mid consisting of the pUC19 backbone with an inserted CRISPR locus site integration events (Fig. 2a). We observed similar gel mobility of the
(pCRISPR) (Fig. 1a). Co-incubation of these reagents converted the slowly migrating DNA product generated by Cas1–Cas2 and Nb.BbvCI
supercoiled plasmid into three main products: relaxed and linear plas- nickase-digested pCRISPR, consistent with the slow-migrating relaxed
mid species and a fast-migrating species we term band X (Fig. 1b, c DNA species corresponding to half-site products and/or products result-
and Extended Data Fig. 1a). Product formation required Cas1, Cas2 and ing from full-site integration of one protospacer molecule (Extended
the protospacer DNA (Extended Data Fig. 1b–d), and was consistent Data Fig. 1a). Digestion with EcoRI, which cuts pCRISPR once, con-
with previous divalent metal ion-dependent and sequence-nonspecific verted the reaction products to linear DNAs (Fig. 2b, compare lane 4 to
in vitro activity requirements of Cas1 (refs 17–19) and Cas2 (refs 20–22). lane 2, and Fig. 2c). We therefore conclude that both the relaxed and
Product DNA migration was not affected by treatment with EDTA, band X DNA products comprise unit-sized pCRISPR circles.
1
Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, California 94720, USA. 2Center for RNA Systems Biology, University of California, Berkeley, Berkeley, California
94720, USA. 3Department of Cancer Immunology and AIDS, Dana-Farber Cancer Institute and Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA. 4Howard Hughes
Medical Institute, University of California, Berkeley, Berkeley, California 94720, USA. 5Department of Chemistry, University of California, Berkeley, Berkeley, California 94720, USA. 6Physical Biosciences
Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA.

0 0 M O N T H 2 0 1 5 | VO L 0 0 0 | N AT U R E | 1
©2015 Macmillan Publishers Limited. All rights reserved
 RESEARCH ARTICLE

Figure 1 | The Cas1–Cas2 complex integrates

a s 9 s2

)
β7
C (E a
protospacers in vitro. a, Schematic of the in vitro

1+ s2 + C

6–
2( Q)
C + C 8A s2

Δβ
as 2 a
as a )
a b c integration assay (PDB code 4P6I for Cas1–Cas2).

C 1(H +C
– + + + Protospacer (50 nM)

0
1
b, The presence of Cas1, Cas2 and a protospacer

as s
Cas1–Cas2 complex + + + + pCRISPR (7.5 nM)

C Ca
pCRISPR

1
T
– + – + Cas1 (75 nM) results in the conversion of the supercoiled

W
EcoRI
Cas1 + + + – Complex formation
– – + + Cas2 (75 nM) pCRISPR into relaxed, linear and band X products.
+ – + – In vivo acquisition
c, Neither the Cas1(H208A) active site mutant
Cas2 (kb) nor and the complex formation-defective
10.0 Relaxed (kb) Cas2(Db6–b7) deletion mutant support the
Relaxed
3.0 Linear 10.0 reaction. The Cas2(E9Q) active site mutant is as
2.0 S.C. 3.0 Linear
33 bp protospacer Band X 2.0 S.C. active as the wild type. d, Salt- and metal-
1.5
1.5 Band X dependence of radiolabelled protospacer
1.0 1.0 integration into pCRISPR. e, Same as c except using
CRISPR radiolabelled protospacers. The data presented
0.5 0.5
in b–e are representative of at least three replicates.
AmpR pCRISPR
3240 bp

ori Free Free

protospacer protospacer

as 9 s2

)
β7
C (E Ca

6–
2( Q)
C 1+C 08A 2
1+ s2 +

Δβ
s
as 2 a
as a )
C 1(H +C
Protospacer

d e Protospacer

as s1
C Ca
10 mM MgCl2 10 mM MnCl22
T
W

– 10% DMSO + 10% DMSO + 10% DMSO + + + – Complex formation

KCl + – + – In vivo acquisition
(kb)
10.0 Integration
(kb)
Integration products
10.0 3.0
products
3.0 2.0
2.0 Expected
1.5
1.5 band X
1.0
1.0

0.5 0.5

Free Free
protospacer protospacer

We observed that band X did not become radiolabelled in reactions the presence of Cas2 (Fig. 2h). Disintegration activity was confirmed by
conducted with 32P-labelled protospacer DNA. A time-course analysis radiolabelling the 20 nucleotide (nt) target DNA strand and monitor-
revealed relaxed DNA product formation within the first minute, fol- ing the formation of the joined 40 bp target DNA product (Extended
lowed by accumulation of band X between 10 and 30 min (Fig. 2d). To Data Fig. 5c, d). Thus, Cas1–Cas2 integration and disintegration activ-
determine the properties of band X, the purified product was analysed ities are similar to those of retroviral integrases and transposases.
in two different types of agarose gels—one pre-stained with ethidium
bromide, similar to the gels presented thus far, and the other stained Integration requires 39-OH protospacer ends
with ethidium bromide after electrophoresis (post-stained) (Extended We next investigated the DNA protospacer and target DNA requirements
Data Fig. 4a). Although band X migrated as a single species in the pre- for integration. Single-stranded protospacer DNA failed to support the
stained gel, a ladder of species that migrated faster than the relaxed pro- reaction (Fig. 3a, b). The Cas1–Cas2 complex accommodated various
ducts was observed in the post-stained gel (Fig. 2e, f). These intermediates protospacer lengths in vitro despite the strict 33 bp requirement for spacer
are reminiscent of plasmid topoisomers27,28. The same pre- and post- acquisition in vivo (Extended Data Fig. 6a), suggesting that protospacer
stained agarose gel analysis was performed on the entire integration length is pre-determined before integration in vivo by an unknown
reaction, generating similar results to those observed with purified band mechanism. The Cas1–Cas2 complex integrated DNA substrates with
X (Extended Data Fig. 4b, c). PCR analysis of various segments of blunt-ends or with 39-overhangs up to 5 nt in length (Extended Data
pCRISPR using gel-purified band X as the template yielded amplifica- Fig. 6b). In contrast to retroviral integrases31, substrates with 59-overhangs
tion products indistinguishable from those generated using unreacted were non-viable (Extended Data Fig. 6b).
supercoiled pCRISPR or relaxed integration products, supporting the Retroviral integration and transposition reactions proceed via nu-
conclusion that band X corresponds to pCRISPR topoisomers (Extended cleophilic attack of DNA 39-OH groups at target DNA phosphodie-
Data Fig. 4d). ster bonds31,32. We found that phosphorylation of both 39-ends of the
We wondered whether Band X arose from protospacer excision protospacer ablated integration, whereas phosphorylation of only one
from half-site integration products to regenerate pCRISPR in different 39 end strongly limited integration (Fig. 3a, b). By analogy to known
supercoiled states, analogous to the in vitro reversal of integration activity integrase enzyme mechanisms, DNA integration could proceed by Cas1-
of retroviral integrases and transposases (termed disintegration, Fig. 2g)29,30. catalysed direct nucleophilic attack of the substrate 39-OH on the target
To test this hypothesis, a synthetic Y-structured DNA intermediate that DNA, or by formation of a Cas1–DNA intermediate, as occurs in the
mimics the half-site integration product (Extended Data Fig. 5a, b) was serine and tyrosine families of recombinases33. Four tyrosine residues
radiolabelled such that the liberated 33 bp protospacer DNA could in the vicinity of the Cas1 active site17–19 could be involved in forming
be detected following disintegration activity. Using this substrate, we such a covalent intermediate (Extended Data Fig. 7a, b). Purified Cas1
observed that Cas1 catalysed disintegration activity either by itself or in mutant proteins in which each tyrosine was individually changed to
2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 5
©2015 Macmillan Publishers Limited. All rights reserved
 ARTICLE RESEARCH

a b Post-reaction c d Figure 2 | Half-site, full-site integration and

Supercoiled treatment
Post-reaction pCRISPR topoisomer products. a, Schematic of
pCRISPR – – + + Cas1+ Cas2 treatment
– + – + EcoRI
half-site and full-site integration products.
– EcoRI 0 1 3 5 10 15 20 30 45 60 90 (min)
b, Linearization of the integration products (lane
(kb)
Relaxed 4). Lane 3 is the untreated reaction products.
10.0 (kb)
10.0 Relaxed
+ Cas1–Cas2 3.0 Linear c, Linearization of integration products from
+ protospacers (kb) Relaxed 2.0 3.0
10.0 1.5
2.0 S.C. radiolabelled protospacer reactions. d, The time
1.5 Band X
3.0
Linear
1.0 1.0
course reveals the initial formation of relaxed
2.0 Band X products, followed by band X. The inset reveals
Relaxed/ 0.5
half-site
1.5
0.5 the products detected using 32P-labelled
1.0 protospacers. e, f, Analysis of gel-purified relaxed
Linear/ Relaxed/
full-site full-site
Free
protospacer
and band X on agarose gels pre-stained with
0.5
32P Relaxed ethidium bromide (e) or post-stained after
labelled Linear electrophoresis (f). g, Schematic of the
1 2 3 4 disintegration reaction. h, Native polyacrylamide
gel analysis of the disintegration reaction. The
e Gel purified f Gel purified g h MgCl2 MnCl2
data presented in b–f, h are representative of at
R

R
ed
SP

+ + + + + + + Y DNA (5 nM)
ed

least three replicates.

X

SP

X
ax

d
RI

3′ ′
ax

d
RI

Half-site
n

– + – + + – + Cas1(50 nM)
n
l

5
pC

Re

Ba

l
pC

Re

Ba

(kb) intermediate – – + + – + + Cas2 (50 nM)

10.0 Relaxed
(kb) Band X 5′ 5′ 3′ *
10.0 3′ 5′ Y DNA
3.0
3′

3.0
S.C.
5′

2.0 2.0 Disintegration

5′

1.5 5′ OH 3′
1.5 (bp)
3′ 5′
43 33 bp
1.0
1.0 Disintegration 35
31
*
+
3′ 5′
23 40 bp (unlabelled)
5′ 3′
0.5 0.5 5′ 3′
18
Pre-stained with EtBr Post-stained with EtBr 3′ 5′
Native polyacrylamide

alanine supported protospacer integration in vitro at levels comparable observe integration products upon incubation with Cas1 and Cas2 in
to wild-type Cas1–Cas2 (Extended Data Fig. 7c). Thus, the integration the presence of protospacer DNA (Fig. 3c and Extended Data Fig. 6e).
reaction likely proceeds via direct nucleophilic attack of protospacer This finding raised two possibilities: either in vitro spacer integration
39-OH ends onto the target DNA phosphodiester bonds, a mechanism is non-specific with respect to target DNA sequence or structures and/
previously hypothesized to occur in vivo34. or sequence(s) favouring integration are present in the pUC19 plas-
mid. To determine if integration preferentially occurred at the CRISPR
Supercoiled DNA and CRISPR locus requirements locus of pCRISPR, products of radiolabelled reactions were double-
Cas1 and Cas2 overexpression leads to site-selective spacer acquisition digested to separate the CRISPR locus (960 bp) from the pUC19 plas-
proximal to the leader end of the CRISPR locus, a result consistent with mid backbone (,2.27 kb). Suggestive of CRISPR-specific integration, the
32
observations in native populations of CRISPR-containing bacteria13–15,35. P-radiolabel migrated solely with the CRISPR locus fragment (Fig. 3d).
To determine what drives such site-specific integration, we first tested The same result was observed when the experiment was conducted using
various forms of the pCRISPR plasmid to determine target DNA re- a target plasmid containing the CRISPR locus and a different backbone
quirements. Integration requires target DNA supercoiling, as neither sequence (pACYC) (Fig. 3e).
relaxed nor linear pCRISPR, nor the isolated 1 kb CRISPR locus, sup-
ported integration (Fig. 3c and Extended Data Fig. 6c, d). CRISPR repeats provide specificity
As a control, we tested supercoiled pUC19 DNA, the parental plas- To determine the exact sites of protospacer integration in these reac-
mid of pCRISPR that lacks a CRISPR locus, and were surprised to tions, we performed high-throughput sequencing of reaction products

EcoRI
a b c d CRISPR e lacI
EcoNI
CRISPR
(960 bp) (~1 kb)
AmpR pUC19 pACYC
Protospacer

33 bp dsDNA Supercoiled Linear 1 kb Supercoiled CRISPR CRISPR EcoO109I

AflIII
H
5′ H

pCRISPR pCRISPR CRISPR pUC19 3,240 bp p15A ori 4,513 bp

-O
O ′-O

: 3 OH
H

′– H
S1 ′–P 4
S2: 3′ O4

4
: 3 PO

PO
S2 3′- -PO4

9I
1

: 2
S1 H, ′-PO

S1 ′–O

: 3 –O
3′ O4 , -OH

: 3 OH 4
pCRISPR

′-P , 5

10
ssDNA 1
ssDNA 2

S1 A
S2 3′–

S2 3′–

CmR
4,

ori
DN

DN

1+

1+

1+

1+

oO
-O 5

III
-P 5′

: 5′

C 1
C 2

1
C 2

1
C 2

C 1
C 2
:

as
as
as

as
as
as

as
as
as

as
as
as
ss

ss

Afl

Ec
3′ H,

C
C

C
C

C
-O

I+

+
Ec NI

– + Cas1+Cas2
Ec RI

– + – + – + – +
I
3′

oR

Ec e

oN
Ec e

on
on

o
o

(kb)
N

(kb) (kb)
10.0 Integration 10.0 (kb)
(kb) products 10.0 10.0
10.0 Relaxed 3.0 3.0
2.0 2.0 3.0 3.0
3.0 2.0 2.0
S.C. 1.5 1.5 1.5
2.0 1.5
1.5 Band X
1.0 1.0 1.0 1.0
1.0
0.5 0.5 0.5 0.5
0.5

Free
Free protospacer
protospacer

Figure 3 | Integration requires 39-OH protospacer ends and supercoiled integration into different DNA targets. d, e, Restriction enzyme digestion
target DNA. a, b, Integration assays using single-stranded DNAs and of pCRISPR, either in a pUC19 (d) or pACYC backbone (e), after the
either –OH or –PO4 at the 39 or 59 ends of unlabelled (a) or radiolabelled integration assay detects integration into the CRISPR fragment (green arrows).
(b) protospacers. S1 corresponds to one strand of the protospacer and S2 The data presented in a–e are representative of at least three replicates.
corresponds to the complementary strand. c, Comparison of protospacer

0 0 M O N T H 2 0 1 5 | VO L 0 0 0 | N AT U R E | 3
©2015 Macmillan Publishers Limited. All rights reserved
 RESEARCH ARTICLE

a c
pCRISPR target pUC19 target
600 600
Number of integration events

Number of integration events

(+) strand
AmpR
(+) strand

AmpR
CRISPR locus ori AmpR promoter lacZα ori AmpR promoter
300 300

(–) strand
(–) strand

300 300

600 600

Repeat 1 Repeat 2
b Protospacer sequence d 3′-OH nucleotide e C C C C
G A G A
500 preference C G C G
(+) strand

3′ C 3′

Total integration events

Number of integration events

CGGGTTAAATGATGAGCAAGACCACAAAGAGCA G C G C
C G C G
3′ T GCCCAATTTACTACTCGTTCTGGTGTTTCTCGT 3′ 80 * *

into pCRISPR (%)

C G C G
250 C G Spacer-1 C G
C G C G
60 5′ G T G T T A T A A A C C GNNN...............NNN G T G T T A T AAACC 3′
3′ CACA A T A T T T G G CNNN...............NNN C A C A A T A T T T GG 5′
Leader
40 AT- rich G C G C
R S R S R S R S R S R S R S R S R S R S R S R S R S R G C G C
leader G C G C
20 C T C T G C G C
C G C G
(–) strand

G C G C
250 0 C T C T
G G G G
pCRISPR pUC19
empty
500

Figure 4 | Protospacers are specifically integrated into the CRISPR locus. above the plot. c, Integration sites along pUC19. d, Comparison of C 39-OH
a, Integration sites along pCRISPR. b, Magnified view of the integration sites or T 39-OH selection in the total reads from pCRISPR and pUC19 targets
along the ,1 kb CRISPR locus. The cyan peaks represent positions where (n 5 7,866 reads for pCRISPR and n 5 5,524 reads for pUC19, chi-square test,
the 39 T of the protospacer DNA was integrated whereas the black peaks *P , 0.0001). e, Schematic of DNA cruciform formation of the repeat
represent the C 39-OH integration events. The protospacer sequence is depicted sequences. The orange arrows depict the cleavage sites.

that resulted from using either pCRISPR or the parental pUC19 vector When we used protospacer DNAs lacking a 39 C or bearing 39 C on
as the target of integration (Extended Data Fig. 8a). Of the 7,866 pro- both ends, the preference for integration into the minus strand of the
tospacer-pCRISPR junctions retrieved, ,71% mapped to the CRISPR CRISPR locus was significantly decreased (Extended Data Fig. 9). Thus,
locus (Fig. 4a and Extended Data Fig. 8b). Protospacer insertion occurred the Cas1–Cas2 complex plays a critical role in correctly orienting the C
at the borders of each repeat, with the most preferred site at the first 39-OH end of protospacer DNA substrates for incorporation within the
repeat adjacent to the leader (Fig. 4b). The minus strand of each repeat CRISPR locus.
(the bottom strand in Fig. 4a, b that runs 59 to 39 towards the leader
sequence) is also highly preferred, highlighting the role of CRISPR Mechanism of protospacer integration
repeats in providing sequence specificity for the Cas1–Cas2 complex The results presented here explain the mechanistic basis for foreign DNA
(Fig. 4b). Sequence alignment of the integration sites revealed strong acquisition during CRISPR–Cas adaptive immunity (Fig. 5). The Cas1–
preference for sequences resembling the CRISPR repeat on both strands Cas2 complex catalyses integration of protospacers at the leader-end of
of pCRISPR, further supporting the selection of CRISPR repeat bor-
Foreign DNA Mature 33 bp protospacer
ders by the Cas1–Cas2 complex (Extended Data Fig. 8d–f). PAM
G
AA G 3′-OH
The most frequent integration site in the pUC19 control plasmid Protospacer HO-3′
C
TT C
mapped to the amp resistance gene adjacent to the AT-rich promoter Repeat 1 Repeat 2
sequence (,8.8% of 5,524 total retrieved junctions, Fig. 4c and Ex- Leader Spacer 1
First 5′ 3′
tended Data Fig. 8c). An inverted repeat sequence with a propensity to nucleophilic attack 3′ 5′
form a DNA cruciform36 occurs 9 nt adjacent to this integration site C
HO-3′
(plus strand sequence: 59-TTCAATATTATTGAA-39), suggesting that 3′-OH

potential DNA cruciform formation adjacent to AT-rich sequences is Half-site 5′ 3′

important for protospacer integration. Sequence analysis of pUC19 intermediate 3′ 5′
O C
target sites revealed the propensity for a G nucleotide to occur at the
3′-OH
22 and 11 positions of the protospacer insertion site, similar to the HO-3
′
preferred pCRISPR sites (Extended Data Fig. 8g, h). These observa-
tions imply that in addition to sequence, pCRISPR repeat selectivity Second 5′
O
3′
stems from the unique structural features of these sites, such as their nucleophilic attack 3′ 5′

ability to form cruciforms (Fig. 4a, b, e).

G
In E. coli, newly acquired spacers harbour a 59 G as the first nucle- Fully integrated
spacer
5′
3′
3′
5′ 3′
3′
5′
otide flanking the leader-proximal end of the repeats, which originates C

from the last nucleotide of the AAG protospacer-adjacent motif

Repeat 1* G Repeat 1* Repeat 2
(PAM) from foreign DNA13–15,37–39. Such positional specificity is crit- DNA repair
5′
3′
3′
5′
C
ical for crRNA-guided interference, as a mutation in this position of the
Repeat duplication and new spacer integration
corresponding crRNA disrupts PAM binding and subsequent target
destruction40–42. We found that ,73% of all integration events into Figure 5 | Model of protospacer integration during CRISPR–Cas adaptive
pCRISPR used the 39 C end instead of the 39 T end of protospacer DNA immunity. The first nucleophilic attack occurs on the minus strand of the first
during integration (see Fig. 4b for protospacer sequence), and there was repeat, distal to the leader, by the C 39-OH end of the protospacer. After
half-site intermediate formation, the second integration event occurs on the
a strong preference for this nucleotide to attack the minus strand of the opposite strand at the leader-repeat border. The resulting single-stranded
repeat sequence (Fig. 4b, d, e). A similar nucleotide bias was observed DNA gaps are repaired by yet uncharacterized mechanisms and the
in the pUC19 target plasmid sequence data (Fig. 4d). This preference protospacer is fully integrated with the G as the first nucleotide at its 59 end.
positions the G at the 59 end of the protospacer substrate as the first The asterisk denotes the duplication of the first repeat, as previously observed
nucleotide of the newly integrated spacer in the CRISPR locus (Fig. 5). in vivo13–15.

4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 5
©2015 Macmillan Publishers Limited. All rights reserved
 ARTICLE RESEARCH

the CRISPR locus and also selects the terminal C 39-OH as the attacking 4. Bolotin, A., Quinquis, B., Sorokin, A. & Ehrlich, S. D. Clustered regularly interspaced
short palindrome repeats (CRISPRs) have spacers of extrachromosomal origin.
nucleophile, resulting in the 59 G on the opposite strand of the proto- Microbiology 151, 2551–2561 (2005).
spacer becoming the first nucleotide of the newly integrated spacer. This 5. Pourcel, C., Salvignol, G. & Vergnaud, G. CRISPR elements in Yersinia pestis acquire
orientation bias, previously observed in vivo39, is a key step during new repeats by preferential uptake of bacteriophage DNA, and provide additional
tools for evolutionary studies. Microbiology 151, 653–663 (2005).
immunity for productive downstream foreign DNA targeting by the 6. Stern, A., Keren, L., Wurtzel, O., Amitai, G. & Sorek, R. Self-targeting by CRISPR: gene
Cascade complex and Cas3 effector nuclease (Extended Data Fig. 10). regulation or autoimmunity? Trends in Genet. 26, 335–340 (2010).
Interestingly, the presence of the complete AAG PAM in the proto- 7. Carte, J., Wang, R., Li, H., Terns, R. M. & Terns, M. P. Cas6 is an endoribonuclease
that generates guide RNAs for invader defense in prokaryotes. Genes Dev. 22,
spacer is not required for in vitro integration, suggesting that a highly 3489–3496 (2008).
specific selection or processing step occurs in vivo to exclude the AA 8. Haurwitz, R. E., Jinek, M., Wiedenheft, B., Zhou, K. & Doudna, J. A. Sequence- and
nucleotides from the mature protospacer before integration. structure-specific RNA processing by a CRISPR endonuclease. Science 329,
1355–1358 (2010).
We propose a two-step integration mechanism in which the C 39-
9. Deltcheva, E. et al. CRISPR RNA maturation by trans-encoded small RNA and host
OH first attacks the minus strand of the CRISPR repeat to produce a factor RNase III. Nature 471, 602–607 (2011).
half-site intermediate (Fig. 5). The 39-OH on the opposite strand of the 10. Brouns, S. J. et al. Small CRISPR RNAs guide antiviral defense in prokaryotes.
integrating DNA then attacks the target DNA 28 bp away on the oppo- Science 321, 960–964 (2008).
11. Garneau, J. E. et al. The CRISPR/Cas bacterial immune system cleaves
site side of the repeat on the plus strand, leading to full integration of bacteriophage and plasmid DNA. Nature 468, 67–71 (2010).
the protospacer (Fig. 5). Our in vitro system predominantly traps the 12. Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive
first step of this two-step integration mechanism, suggesting that the bacterial immunity. Science 337, 816–821 (2012).
13. Yosef, I., Goren, M. G. & Qimron, U. Proteins and DNA elements essential for the
second nucleophilic attack is greatly accelerated in vivo in the presence CRISPR adaptation process in Escherichia coli. Nucleic Acids Res. 40, 5569–5576
of cellular factors. This model is consistent with spacer integration (2012).
intermediates that are observed in vivo, in which protospacers are inte- 14. Datsenko, K. A. et al. Molecular memory of prior infections activates the CRISPR/
Cas adaptive bacterial immunity system. Nature Commun. 3, 945 (2012).
grated such that staggered cleavage at each end of the repeat generates
15. Swarts, D. C., Mosterd, C., van Passel, M. W. & Brouns, S. J. CRISPR interference
single-stranded gaps that ensure repeat duplication34. The in vivo con- directs strand specific spacer acquisition. PLoS ONE 7, e35888 (2012).
ditions could also promote the high specificity of integration to occur 16. Nuñez, J. K. et al. Cas1–Cas2 complex formation mediates spacer acquisition
solely downstream of the first repeat of the CRISPR locus in E. coli, during CRISPR–Cas adaptive immunity. Nature Struct. Mol. Biol. 21, 528–534
(2014).
instead of at every repeat, as observed in our in vitro assay. 17. Wiedenheft, B. et al. Structural basis for DNase activity of a conserved
CRISPR spacer integration shares mechanistic similarities with ret- protein implicated in CRISPR-mediated genome defense. Structure 17, 904–912
roviral integration and DNA transposition, where the integrase/trans- (2009).
18. Babu, M. et al. A dual function of the CRISPR–Cas system in bacterial antivirus
posase enzyme uses donor DNA 39-OH ends to make a staggered cut immunity and DNA repair. Mol. Microbiol. 79, 484–502 (2011).
at the DNA target site, which concurrently joins the donor DNA to tar- 19. Kim, T. Y., Shin, M., Huynh Thi Yen, L. & Kim, J. S. Crystal structure of Cas1 from
get DNA 59-phosphates31,32. Completion of the integration reaction Archaeoglobus fulgidus and characterization of its nucleolytic activity. Biochem.
Biophys. Res. Commun. 441, 720–725 (2013).
requires a DNA polymerase to fill in sequence gaps and a DNA ligase 20. Beloglazova, N. et al. A novel family of sequence-specific endoribonucleases
to seal the phosphodiester backbone43. Similar polymerase and ligase associated with the clustered regularly interspaced short palindromic repeats.
functions are required to complete CRISPR spacer acquisition in vivo, J. Biol. Chem. 283, 20361–20371 (2008).
although the specific enzymes involved have not yet been identified. 21. Samai, P., Smith, P. & Shuman, S. Structure of a CRISPR-associated protein Cas2
from Desulfovibrio vulgaris. Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 66,
Despite these similarities, we note that the Cas1 active site does not 1552–1556 (2010).
harbour the RNase H fold that defines the retroviral integrase enzyme 22. Nam, K. H. et al. Double-stranded endonuclease activity in Bacillus halodurans
superfamily44. This structural difference could explain the unexpected clustered regularly interspaced short palindromic repeats (CRISPR)-associated
Cas2 protein. J. Biol. Chem. 287, 35943–35952 (2012).
production of different topoisomers of pCRISPR (band X) in vitro, although 23. Li, M. & Craigie, R. Processing of viral DNA ends channels the HIV-1 integration
the physiological significance of band X production remains unclear. reaction to concerted integration. J. Biol. Chem. 280, 29334–29339 (2005).
Our results highlight the fundamental role of repeat sequences at 24. Cherepanov, P. LEDGF/p75 interacts with divergent lentiviral integrases
and modulates their enzymatic activity in vitro. Nucleic Acids Res. 35, 113–124
multiple stages of CRISPR–Cas adaptive immunity. In addition to cre- (2007).
ating structures within nascent CRISPR transcripts that ensure correct 25. Hare, S. et al. A novel co-crystal structure affords the design of gain-of-function
RNA processing during crRNA maturation45, the repeats operate at the lentiviral integrase mutants in the presence of modified PSIP1/LEDGF/p75. PLoS
Pathog. 5, e1000259 (2009).
DNA level to recruit the Cas1–Cas2 complex for sequence- and structure- 26. Yang, J. Y., Jayaram, M. & Harshey, R. M. Positional information within the Mu
specific protospacer integration. We envision that this recruitment transposase tetramer: catalytic contributions of individual monomers. Cell 85,
involves transient DNA cruciform formation within the CRISPR inverted 447–455 (1996).
27. Dinardo, S., Voelkel, K. A., Sternglanz, R., Reynolds, A. E. & Wright, A. Escherichia coli
repeats that occurs as a function of target DNA supercoiling46. The ob- DNA topoisomerase I mutants have compensatory mutations in DNA gyrase
servation that a preferred non-CRISPR site of Cas1–Cas2-mediated genes. Cell 31, 43–51 (1982).
DNA integration is proximal to an inverted repeat adjacent to an AT- 28. Pruss, G. J., Manes, S. H. & Drlica, K. Escherichia coli DNA topoisomerase I mutants:
rich sequence suggests the fascinating possibility that CRISPR loci arise increased supercoiling is corrected by mutations near gyrase genes. Cell 31,
35–42 (1982).
in naive genomes through integration events that become self-propagating 29. Chow, S. A., Vincent, K. A., Ellison, V. & Brown, P. O. Reversal of integration and DNA
through creation of repetitive sequences with properties that ensure con- splicing mediated by integrase of human immunodeficiency virus. Science 255,
tinual recognition and activity by the Cas1–Cas2 integration machinery. 723–726 (1992).
30. Au, T. K., Pathania, S. & Harshey, R. M. True reversal of Mu integration. EMBO J. 23,
Online Content Methods, along with any additional Extended Data display items 3408–3420 (2004).
and Source Data, are available in the online version of the paper; references unique 31. Engelman, A., Mizuuchi, K. & Craigie, R. HIV-1 DNA integration: mechanism of viral
to these sections appear only in the online paper. DNA cleavage and DNA strand transfer. Cell 67, 1211–1221 (1991).
32. Mizuuchi, K. & Adzuma, K. Inversion of the phosphate chirality at the target site of
Mu DNA strand transfer: evidence for a one-step transesterification mechanism.
Received 6 November 2014; accepted 15 January 2015.
Cell 66, 129–140 (1991).
Published online 18 February 2015. 33. Curcio, M. J. & Derbyshire, K. M. The outs and ins of transposition: from mu to
kangaroo. Nature Rev. Mol. Cell Biol. 4, 865–877 (2003).
1. Barrangou, R. et al. CRISPR provides acquired resistance against viruses in 34. Arslan, Z., Hermanns, V., Wurm, R., Wagner, R. & Pul, U. Detection and
prokaryotes. Science 315, 1709–1712 (2007). characterization of spacer integration intermediates in type I-E CRISPR-Cas
2. van der Oost, J., Westra, E. R., Jackson, R. N. & Wiedenheft, B. Unravelling the system. Nucleic Acids Res. 42, 7884–7893 (2014).
structural and mechanistic basis of CRISPR–Cas systems. Nature Rev. Microbiol. 35. Tyson, G. W. & Banfield, J. F. Rapidly evolving CRISPRs implicated in acquired
12, 479–492 (2014). resistance of microorganisms to viruses. Environ. Microbiol. 10, 200–207 (2008).
3. Mojica, F. J., Diez-Villasenor, C., Garcia-Martinez, J. & Soria, E. Intervening 36. Sheflin, L. G. & Kowalski, D. Altered DNA conformations detected by mung bean
sequences of regularly spaced prokaryotic repeats derive from foreign genetic nuclease occur in promoter and terminator regions of supercoiled pBR322 DNA.
elements. J. Mol. Evol. 60, 174–182 (2005). Nucleic Acids Res. 13, 6137–6154 (1985).

0 0 M O N T H 2 0 1 5 | VO L 0 0 0 | N AT U R E | 5
©2015 Macmillan Publishers Limited. All rights reserved
 RESEARCH ARTICLE

37. Goren, M. G., Yosef, I., Auster, O. & Qimron, U. Experimental definition of a clustered 46. Paleček, E. Local supercoil-stabilized DNA structures. Crit. Rev. Biochem. Mol. Biol.
regularly interspaced short palindromic duplicon in Escherichia coli. J. Mol. Biol. 26, 151–226 (1991).
423, 14–16 (2012).
38. Savitskaya, E., Semenova, E., Dedkov, V., Metlitskaya, A. & Severinov, K. High- Acknowledgements We are grateful to M. Chung, P. J. Kranzusch and A.V. Wright for
throughput analysis of type I-E CRISPR/Cas spacer acquisition in E. coli. RNA Biol. technical assistance and members of the Doudna laboratory and J. Cate for
10, 716–725 (2013). discussions. This project was funded by US National Science Foundation grant no.
39. Shmakov, S. et al. Pervasive generation of oppositely oriented spacers during 1244557 to J.A.D. and by NIH grant AI070042 to A.E. This work used the Vincent
CRISPR adaptation. Nucleic Acids Res. 42, 5907–5916 (2014). J. Coates Genomics Sequencing Laboratory at UC Berkeley, supported by NIH S10
40. Deveau, H. et al. Phage response to CRISPR-encoded resistance in Streptococcus Instrumentation Grants S10RR029668 and S10RR027303. J.K.N. is supported by a
thermophilus. J. Bacteriol. 190, 1390–1400 (2008). US National Science Foundation Graduate Research Fellowship and a UC Berkeley
41. Semenova, E. et al. Interference by clustered regularly interspaced short Chancellor’s Graduate Fellowship. A.S.Y.L. is supported as an American Cancer Society
palindromic repeat (CRISPR) RNA is governed by a seed sequence. Proc. Natl Acad. Postdoctoral Fellow (PF-14-108-01-RMC). J.A.D. is an Investigator of the Howard
Sci. USA 108, 10098–10103 (2011). Hughes Medical Institute and a member of the Center for RNA Systems Biology.
42. Westra, E. R. et al. Type I-E CRISPR-cas systems discriminate target from non- Author Contributions J.K.N. performed the biochemical experiments. A.S.Y.L.
target DNA through base pairing-independent PAM recognition. PLoS Genet. 9, processed and analysed the high-throughput sequencing data. J.K.N., A.S.Y.L., A.E. and
e1003742 (2013). J.A.D. designed the study, analysed the data and wrote the manuscript.
43. Craigie, R. & Bushman, F. D. HIV DNA integration. Cold Spring Harbor Perspect. Med.
2, a006890 (2012). Author Information Sequencing data are deposited in Gene Expression Omnibus
44. Nowotny, M. Retroviral integrase superfamily: the structural perspective. EMBO under accession number GSE64552. Reprints and permissions information is
Rep. 10, 144–151 (2009). available at www.nature.com/reprints. The authors declare competing financial
45. Hochstrasser, M. L. & Doudna, J. A. Cutting it close: CRISPR-associated interests: details are available in the online version of the paper. Readers are welcome
endoribonuclease structure and function. Trends Biochem. Sci. 40, 58–66 to comment on the online version of the paper. Correspondence and requests for
(2015). materials should be addressed to J.A.D. ([email protected]).

6 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 5
©2015 Macmillan Publishers Limited. All rights reserved
 ARTICLE RESEARCH

METHODS Disintegration assays. The four single-stranded DNA substrates were annealed
Cas1, Cas2 and DNA preparation. The cas1 and cas2 genes from E. coli K12 to form the Y DNA in a stepwise manner: 95 uC for 3 min, 65 uC for 20 min, 50 uC
(MG1655) were cloned into expression vectors and the proteins were separately for 20 min, and gradual cooling to room temperature. The annealing reactions
purified as previously described16. The proteins were stored in 100 mM KCl, 20 mM were analysed on a 15% native polyacrylamide gel to confirm the formation of the
HEPES-NaOH, 5% glycerol and 1 mM TCEP at 280 uC before use. Single-stranded Y DNA (Extended Data Fig. 5b). The disintegration assay was performed in the
DNAs were synthesized (Integrated DNA Technologies). Double-stranded DNA integration reaction buffer with 50 nM protein and 5 nM Y DNA at 37 uC for 1 h.
protospacers were annealed in 20 mM HEPES-NaOH, pH 7.5, 25 mM KCl, 10 mM For native polyacrylamide gel analysis, the reaction was quenched with DNA load-
MgCl2 or MnCl2, 1 mM DTT, 10% DMSO by heating at 95 uC for 3 min and slow ing buffer with 50 mM EDTA and analysed on 15% polyacrylamide gels. For dena-
cooling to room temperature. The sequence of the 33 bp protospacer used in this turing polyacrylamide gel analysis, the reactions were quenched with formamide
study was shown to be the most acquired in vivo in E. coli K12 after M13 bacterio- buffer and heated at 95 uC before loading on 15% 8M urea-polyacrylamide gels.
phage infection14: strand 1 (59-GCCCAATTTACTACTCGTTCTGGTGTTTCT The sequences of the four strands are as follows: A (59-GGCCCCAGTGCTGCA
CGT-39) and strand 2 (59-ACGAGAAACACCAGAACGAGTAGTAAATTGG ATGAT-39); B (59-GTGAGCGTGGGTCTCGCGGTATCATTGCAGCACTGG
GC-39). The pCRISPR target plasmid was constructed by PCR amplifying the GGCC-39); C (59-GCCCAATTTACTACTCGTTCTGGTGTTTCTCGTACCGC
E. coli BL21-AI genomic CRISPR locus and cloning the fragment into pUC19 using GAGACCCACGCTCAC-39); and D (59-ACGAGAAACACCAGAACGAGTAG
the following primers with the underlines indicating the respective restriction TAAATTGGGC-39).
sites used: forward/EcoRI: 59-ACGTCGAATTCTACCTTTTTAATCAATGG-39 High-throughput sequencing. The integration reaction was performed with 75 nM
and reverse/AflIII: 59-ACGTCACATGTGGTTATATGGTGGTTTATCC-39. The Cas1–Cas2, 200 nM protospacer and 7.5 nM pCRISPR or pUC19 in 20 mM HEPES,
pACYC CRISPR plasmid was constructed by cloning the CRISPR fragment into a pH 7.5, 25 mM KCl, 10 mM MgCl2, 10% DMSO and 1 mM DTT. The DNAs were
pACYCDuet-1 vector using the EcoNI and AvrII restriction sites. isolated by phenol–chloroform extraction and ethanol precipitation. The excess
In vitro integration assays. The integration reactions were performed in 20 mM protospacers were removed using 100K MWCO Amicon Ultra-0.5 ml centrifugal
HEPES-NaOH, pH 7.5, 25 mM KCl, 10 mM MgCl2 or MnCl2, 1 mM DTT and 10% filters. The resulting integration products were digested into smaller DNA frag-
DMSO. There was little difference when DMSO was omitted from the reaction ments using dsDNA Fragmentase (New England Biolabs) for 75 min at 37 uC and
(Fig. 1d), in contrast to its in vitro integration enhancement with HIV-1 integrase47. quenched at 65 uC for 15 min. Fragments were end repaired using T4 DNA Poly-
All of the reactions were conducted with MgCl2 unless otherwise noted. For reac- merase (NEB), Klenow (NEB) and T4 PNK (NEB) and A-tailed with Klenow exo
tions with the Cas1–Cas2 complex, separately purified Cas1 and Cas2 were pre- (39 to 59 exo minus) (NEB). Adapters were ligated onto fragments using T4 DNA
incubated for 20–30 min at 4 uC to allow complex formation. The protospacer DNAs ligase (NEB) and cDNA libraries were amplified by PCR using Phusion (NEB).
were incubated with the protein(s) for 10–15 min at 4 uC, followed by the addition Libraries were sequenced on an Illumina HiSeq2500 on rapid run mode. The oli-
of the target pCRISPR or pUC19 plasmid DNA. The reactions were conducted at gonucleotides used are: universal adaptor: 59-AATGATACGGCGACCACCGA
37 uC for 1 h and quenched with DNA loading buffer containing a final concentra- GATCTACACTCTTTCCCTACACGA CGCTCTTCCGATC*T-39 (*phosphor-
tion of 50 mM EDTA. The products were analysed on 1.5% agarose gels pre-stained othioate bond); indexed adaptor: 59-/5Phos/GATCGGAAGAGCACACGTCTG
with ethidium bromide. All of the reactions, except those shown in Fig. 1 and Ex- AACTCCAGTCAC-index-ATCTCGTATGCCGTC TTCTGCTTG-39); PCR pri-
tended Data Fig. 1a, c–e, were conducted with 75 nM protein, 200 nM protospa- mers: 59-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACAC
cers and 7.5 nM pCRISPR to clearly visualize band X from pCRISPR. Reactions in GA-39, 59-CAAGCAGAAGACGGCATACGAGAT-39.
Fig. 1 and Extended Data Fig. 1a, c, e were performed with 50 nM protospacers. Computational analysis. For preprocessing, 39 adapters were removed from raw
Each integration and disintegration assay was performed a minimum of three times. Illumina reads using Cutadapt (http://code.google.com/p/cutadapt/), discarding
Radiolabelled protospacer integration assays. Pre-annealed double-stranded pro- reads shorter than 15 nt. Reads containing integrated protospacer were selected
tospacer DNA substrates were 59-radiolabelled using [c-32P]-ATP (PerkinElmer) using Cutadapt, requiring the presence of at least 10 nt of protospacer sequence
and T4 polynucleotide kinase (New England Biolabs). Protospacers with 39-PO4 with no errors. After creating Bowtie48 indexes from fasta files of the pUC19 empty
ends were 59-radiolabelled using T4 polynucleotide kinase with 39 phosphatase and pCRISPR plasmid sequences, these reads were mapped to the respective plas-
minus activity (New England Biolabs). The reactions were carried out in the same mids using Bowtie, allowing up to 2 mismatches and requiring unique mapping.
buffer as above. Unless otherwise noted, 200 nM of Cas1–Cas2 was first incubated Sequence motif analysis depicted in Extended Data Fig. 8 were generated using
with 20 nM protospacers at 4 uC for 10–15 min, followed by the addition of 200 ng WebLogo, using integration sites that are represented at least ten times in the
(,5 nM) of pCRISPR. The reactions were conducted at 37 uC for 1 h and quenched sequencing data49.
with 25 mM EDTA and 0.4% SDS. The DNA samples were deproteinized with Sample size. No statistical methods were used to predetermine sample size.
30 mg of proteinase K for 1 h at 37 uC and ethanol precipitated. The reactions were
analysed on 1.5% agarose gels. After electrophoresis, the gels were dried onto pos- 47. Engelman, A. & Craigie, R. Efficient magnesium-dependent human
itively charged nylon transfer membrane (GE Healthcare) and imaged using Phos- immunodeficiency virus type 1 integrase activity. J. Virol. 69, 5908–5911 (1995).
48. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory-
phor Screens (GE Healthcare). The restriction enzyme digest experiments were efficient alignment of short DNA sequences to the human genome. Genome Biol.
performed by first conducting the integration reaction, followed by addition of the 10, R25 (2009).
respective enzymes (New England Biolabs), which were allowed to digest for an 49. Crooks, G. E., Hon, G., Chandonia, J. M. & Brenner, S. E. WebLogo: a sequence logo
additional 1 h at 37 uC. generator. Genome Res. 14, 1188–1190 (2004).

©2015 Macmillan Publishers Limited. All rights reserved

 RESEARCH ARTICLE

a b

Ec SPR

PR
C vC

2
1+
Bb
N I

IS
oR

pC I
Cas1 only Cas2 only Cas1+Cas2

1
2
I

oR
R

as
as
as

R
b.
pC

Ec
KCl

C
C
kb
Relaxed kb Relaxed
10.0
10.0
3.0 Linear Linear
3.0
S.C.
2.0 2.0 S.C.
Band X
1.5 1.5 Band X

1.0 1.0

0.5 0.5

Free
Free protospacer
protospacer

c d e

+
K
m

e
n as
ip pr for
Reaction

tio ein
ec + oro
Protospacer post-treatment

ita ot
l
H D ch
10 25 50 75 100 200 200 nM protospacer

O S l-
pCRISPR

Et A + eno
pCRISPR

pCRISPR

Nb.BbvCI

pr S
Cas1+2 pCRISPR (7.5 nM)

h
ED + p
EcoRI

EcoRI
Cas1 (75 nM)

ED A
TA
g 2+

n 2+
TA

T
Additive (10 mM)

ED
ED

Cas2 (75 nM)

kb kb kb
Relaxed 10.0 Relaxed
Relaxed 10.0
10.0
3.0 3.0 Linear 3.0 Linear
Linear
2.0 S.C. 2.0 S.C.
2.0 S.C.
Band X 1.5 Band X
1.5 Band X 1.5
1.0
1.0 1.0

0.5
0.5 0.5

Free Free Free

protospacer protospacer protospacer

Extended Data Figure 1 | The integration reaction is dependent on the carried in from the reaction reagents. c, Integration assays in the presence of
presence of protospacers, low salt and divalent metal ions. a, In vitro 10 mM EDTA, Mg21, Mn21 or no additive. d, Integration assays with
integration assay alongside EcoRI- and Nb.BbvCI nickase-treated pCRISPR. increasing protospacer concentrations. e, A comparison of post-reaction
b, Salt-dependence assay using Cas1 or Cas2 only and Cas11Cas2. The treatments as indicated. The data presented in a–e are representative of at least
titration corresponds to 0, 25, 50, 100 and 200 nM KCl, in addition to the salt three replicates.

©2015 Macmillan Publishers Limited. All rights reserved

 ARTICLE RESEARCH

a b c

Protospacer
10 mM MnCl2

Protospacer
10 mM MgCl2
Cas1–Cas2 complex
Cas1 Cas2 Cas1+2 Cas1 Cas2 Cas1+2
Cas1

kb
kb
10.0 Integration 10.0 Integration
Cas2 products products
3.0 3.0
2.0 2.0
*33-bp protospacer* 1.5 1.5
1.0 1.0

0.5 0.5
CRISPR

AmpR pCRISPR
3240 bp
Free Free
protospacer protospacer
ori

Extended Data Figure 2 | Cas1 requires Cas2 for robust protospacer 0, 50, 100 and 200 nM protein. c, Same as b except in the presence of
integration. a, Schematic of the integration assays using 32P-labelled 10 mM MgCl2. The data presented in b and c are representative of at least
protospacers (PDB code 4P6I for Cas1–Cas2). b, Integration assays in the three replicates.
presence of increasing protein and 10 mM MnCl2. The titration corresponds to

©2015 Macmillan Publishers Limited. All rights reserved

 RESEARCH ARTICLE

a b c
Cas1 mutants + Cas1 mutants +
Cas2 WT Cas2 WT

Protospcer
pCRISPR

8A

1A

D A
K2 A
A

A
8
1
24

24
20

22

20
22
T

T
K2
W

W
H

H
kb
α7 Integration
kb 10.0
products
3.0
D218 10.0 Relaxed 2.0
H208 Linear
1.5
3.0
D221 S.C. 1.0
K224 2.0
Band X
1.5

1.0 0.5
α4 E141

0.5

α5

Free Free
Cas1 active site protospacer
protospacer

Extended Data Figure 3 | The catalytic activity of Cas1 is required for purified Cas1 active site mutants complexed with wild-type Cas2. c, The same
integration. a, Close-up view of the Cas1 active site with the conserved as b except using radiolabelled protospacers. The data presented in b and c are
residues shown in stick configurations (PDB 4P6I). b, Integration assays of representative of at least three replicates.

©2015 Macmillan Publishers Limited. All rights reserved

 ARTICLE RESEARCH

a b c
Gel excised

n
products

io

n
tio
at
r

a
ts g

ts gr
uc i n t e

uc t e
PR

PR

od i n
PR
ed

od d
X

IS

IS

pr i f i e

pr f i e d
ax

IS
nd

r
el

R
pC

pC
Ba

Pu

ri
R

pC

Pu
kb
kb Relaxed 10.0 Relaxed
10.0
Band X
kb Linear
10.0 3.0 3.0
S.C. S.C.
2.0
3.0 Band X 2.0
2.0 1.5
1.5
1.5
1.0
1.0 1.0

0.5
0.5
0.5

Pre-stained with EtBr Post-stained with EtBr

d
CRISPR
locus AmpR AmpR + ori Primer target
100bp ladder
1kb ladder

PR

PR

PR
ed

ed

ed
pC X

pC X

pC X
IS

IS

IS

Template
ax

ax

ax
nd

nd

nd
R

R
el

el

el
Ba

Ba

Ba
R

kb CRISPR locus
10.0 (1 kb)
R
Amp
3.0
(1 kb) pCRISPR
2.0
1.5 ~1.7 kb 3240 bp
1.0 ~1.0 kb

0.5 ori
(0.6 kb)
0.3

0.1

Extended Data Figure 4 | Band X corresponds to topoisomers of pCRISPR. products of various segments of pCRISPR using the relaxed, band X or
a, Agarose gel of purified relaxed and band X integration products. b, Analysis pCRISPR template shown in a. The laddering effect of minor products using
of the total reaction products, after phenol chloroform extraction and CRISPR locus primers likely reflects the propensity of CRISPR repeats to
ethanol precipitation, on a pre-stained agarose gel. c, Same as b except ethidium form DNA hairpins. The data presented in a–d are representative of at least
bromide staining was performed after electrophoresis. d, PCR amplification three replicates.

©2015 Macmillan Publishers Limited. All rights reserved

 RESEARCH ARTICLE

a b

Annealed, A labeled

Annealed, C labeled
C strand only
A strand only
Ladder
Y DNA substrate

3’ ’ G
5
C CC
G C
G A
G A
Y DNA

TT TT
AA TA
AT CT
G AC
Strand D

AT T
G C
bp

AG G
(33 nt)
Strand C

C TC
AA T
43
T
(53 nt) 53 nt ssDNA

G GG
AC T
35

C GT
31

AC T
AA TC
AG TC
AG GT
Strand A (20 nt) 23

5’
C
A
5’ GGCCCCAGTGCTGCAATGAT ACCGCGAGACCCACGCTCAC 3’
3’ CCGGGGTCACGACGTTACTA TGGCGCTCTGGGTGCGAGTG 5’ 18
Strand B (40 nt)
20 nt ssDNA
Disintegration

33 bp 3’ CGGGTTAAATGATGAGCAAGACCACAAAGAGCA 5’
5’ GCCCAATTTACTACTCGTTCTGGTGTTTCTCGT 3’

5’ GGCCCCAGTGCTGCAATGATACCGCGAGACCCACGCTCAC 3’
40 bp 3’ CCGGGGTCACGACGTTACTATGGCGCTCTGGGTGCGAGTG 5’

Native polyacrylamide

c d

2
Strand A labeled

as
C
+
MgCl2

20 2
MnCl2

as + 8A

8A
H s
as W l y

1 Ca
C s1 20
C s1 o n
C 1 T
a H
Y DNA (5 nM) C NA
D

Cas1 (50 nM)

a
Y-

Cas2 (50 nM)

*
Y DNA

nts
53 Strand C (53 nt)
bp
43
Disintegration 40
35 product Disintegration
31 33 product (33 nt)
*

23

18 20
Native polyacrylamide
Denaturing

Extended Data Figure 5 | Cas1 catalyses the disintegration of half-site radiolabelled. c, Native polyacrylamide gel analysis of disintegration assay
integrated protospacers. a, Schematic of the four strands constituting the products using Y DNA substrates with strand A labelled. d, Denaturing gel
Y DNA substrate used in the disintegration assays. b, Native polyacrylamide analysis of the disintegration assay products with strand A labelled.
gel analysis of the annealing products with either strand A or strand C

©2015 Macmillan Publishers Limited. All rights reserved

 ARTICLE RESEARCH

Extended Data Figure 6 | Cas1–Cas2 can integrate various lengths of pCRISPR target. e, Integration assay using different target plasmids with or
double-stranded DNA with blunt- or 39-overhang ends into a supercoiled without a CRISPR locus. The green arrows correspond to the relaxed product
target plasmid. a, Integration assays using the indicated lengths of protospacer of each target and the cyan arrows correspond to the band X product.
DNA. b, Integration assays using varying 59 or 39 overhang lengths. c, d, A The data presented in a–e are representative of at least three replicates.
comparison of integration assays using pCRISPR or Nb.BbvCI-nicked

©2015 Macmillan Publishers Limited. All rights reserved

 RESEARCH ARTICLE

c
a
Cas1 mutants +

Protospacer
Cas2

Y2 A
A

A
Y1 A
88

17

65
49
T

Y1
Y1
W
kb
10.0 Integration
Y217
Y188 products
3.0
2.0
1.5
D218 Y165
D221 1.0

H208
Y149
0.5

E141

Free
protospacer

b α5

49

65
Y1

Y1
E. coli 180
T. maritima 186
P. horikoshii 186
P. aeolicus 178
P. aeruginosa 226

α6 α7
88

17
Y1

Y2

E. coli 240
T. maritima 261
P. horikoshii 261
P. aeolicus 253
P. aeruginosa 287

Extended Data Figure 7 | Cas1 tyrosine mutants support integration c, Radiolabelled protospacer integration assay of Cas1 tyrosine mutants
activity in vitro. a, A close-up of the Cas1 active site with the tyrosine residues complexed with wild-type Cas2. The gel presented in c is representative of at
labelled in blue. b, Structure-based sequence alignment of Cas1 proteins, least three replicates.
highlighting the tyrosine residues mutated to alanine in this study.

©2015 Macmillan Publishers Limited. All rights reserved

 ARTICLE RESEARCH

Extended Data Figure 8 | High-throughput sequencing of integration pUC19 target. d, Sequence of the leader-end of the CRISPR locus in E. coli.
products reveals sequence-specific integration. a, Schematic of the workflow e, f, WebLogo analysis from the 25 to 15 positions surrounding the
for high-throughput sequencing analysis of the integration sites. b, Raw map protospacer integration sites on the plus (e) and minus (f) of pCRISPR.
of the total reads along pCRISPR before collapsing into single peaks of The arrow points to the nucleotide that is covalently joined to the protospacer.
protospacer–pCRISPR junctions depicted in Fig. 4. c, Same as b, except for the g, h, Same as e, f, except for the pUC19 target.

©2015 Macmillan Publishers Limited. All rights reserved

 RESEARCH ARTICLE

Extended Data Figure 9 | Cas1–Cas2 correctly orients the protospacer DNA CRISPR locus strand biases in a, c, e are plotted in b, d and f, respectively, as
during integration. a–f, Mapped integration sites along the CRISPR locus percentages of integration events within the CRISPR locus. The black and
of pCRISPR when using protospacer DNA with nucleotide ends ‘wild-type’ 39 clear bars represent the (2) and (1) strands of the CRISPR locus, respectively.
C and 39 T (a), 39 A and 39 T (c), and 39 C and 39 C (e). The red arrow in c and NS corresponds to not significant and *P , 0.0001 by chi-square test.
e points to the nucleotide change in the protospacer DNA compared to The n values for b, d and f are 5,623, 5,685 and 12,453 reads along the CRISPR
the ‘wild-type’ sequence in a. The protospacer DNA 39 nucleotide and the locus, respectively.

©2015 Macmillan Publishers Limited. All rights reserved

 ARTICLE RESEARCH

Extended Data Figure 10 | Model of the CRISPR–Cas adaptive immunity strand-specific recognition of the 39-TTC-59 PAM sequence in the target strand
pathway in E. coli. Mature double-stranded protospacers bearing a 39 C-OH by the crRNA-guided Cascade complex. Incorrect protospacer integration
are site-specifically integrated into the leader-end of the CRISPR locus. Correct (right) cannot initiate foreign DNA destruction due to the inability for the
protospacer integration (left) results in the 59G/39C as the first nucleotide of crRNA to recognize the strand with the 39-TTC-59 PAM. Thus, foreign DNA
the spacer, proximal to the leader. After transcription of the CRISPR locus interference during CRISPR–Cas adaptive immunity relies on the Cas1–Cas2
and subsequent crRNA processing, foreign DNA destruction is initiated by complex for correctly orienting the protospacer during integration.

©2015 Macmillan Publishers Limited. All rights reserved