Enzymatic assembly of
DNA molecules up to
several hundred kilobases
Daniel G Gibson1, Lei Young1, Ray-Yuan Chuang1,
J Craig Venter1,2, Clyde A Hutchison III2&
Hamilton O Smith2
We describe an isothermal, single-reaction method for
assembling multiple overlapping DNA molecules by the
concerted action of a 5¢ exonuclease, a DNA polymerase
and a DNA ligase. First we recessed DNA fragments, yielding
single-stranded DNA overhangs that specifically annealed, and
then covalently joined them. This assembly method can be used
to seamlessly construct synthetic and natural genes, genetic
pathways and entire genomes, and could be a useful molecular
For nearly 40 years, scientists have had the ability to join DNA
sequences and produce combinations that are not present in
nature. This ‘recombinant DNA technology’ was initiated soon
after the discovery of DNA ligase1and restriction endonucleases2.
Since then, multiple approaches for joining DNA molecules
through the use of restriction enzymes3,4and PCR5–8have been
adapted. Ligation-independent cloning strategies have also been
developed9,10. We had recently described an in vitro recombination
method that we had used to join 101 DNA cassettes into four
quarter molecules of the Mycoplasma genitalium genome, each
136–166 kilobases (kb)11(Supplementary Figs. 1–3 and Supple-
mentary Results online). Because we performed this recombina-
tion methodina thermocycler(towhichwereferas ‘thermocycled’
here), individual reactions were carried out in only two steps. Here
we improved this two-step thermocycled method by using exo-
nuclease III and antibody-bound Taq DNA polymerase, which
allow for one-step thermocycled in vitro recombination (Supple-
mentary Fig. 4 and Supplementary Results online).
We now present an in vitro recombination system that differs
from the ones above by its capacity to assemble and repair over-
lapping DNA molecules in a single isothermal step. This approach
clone joined products in Escherichia coli that are as large as 300 kb.
All reagents and enzymes are commercially available, and all that is
required for DNA assembly is for the reagent-enzyme mix (which
can be stored at ?20 1C until needed) to be combined with
overlapping DNA molecules and then incubated at 50 1C for as few
as 15min(OnlineMethods).Thisapproachdramatically simplifies
the construction of large DNA molecules from constituent parts.
Exonucleases that recess double-stranded DNA from 5¢ ends will
not compete with polymerase activity. Thus, all enzymes required
for DNA assembly can be simultaneously active in a single
isothermal reaction. Furthermore, circular products can be
enriched as they are not processed by any of the three enzymes in
the reaction. We optimized a 50 1C isothermal assembly system
using the activities of the 5¢ T5 exonuclease (Epicentre), Phusion
DNA polymerase (New England Biolabs (NEB)) and Taq DNA
ligase (NEB) (Fig. 1). Taq DNA polymerase (NEB) can be used in
place of Phusion DNA polymerase (data not shown), but the latter
is preferable as it has inherent proofreading activity for removing
noncomplementary sequences (for example, partial restriction
sites) from assembled molecules.
To test this system, we cleaved two restriction fragments that
overlapped by B450 base pairs (bp) from the 6-kb pRS415 vector
and then reassembled them into a circle (Fig. 2a). After 6–8 min at
50 1C, the linear substrate DNA was completely reacted, and the
major product was the 6-kb circle, which migrated just below the
4-kb linear position on an agarose gel. T5 exonuclease actively
are not degraded12. We confirmed the circularity of this assembled
product by treating it with additional T5 exonuclease (Fig. 2a). To
demonstrate that this assembled product was the predicted 6-kb
circle, we digested it with Not I (a single-cutter) and observed the
Chew-back at 50 °C with T5 exonuclease
Anneal at 50 °C
Repair at 50 °C with Phusion polymerase and Taq ligase
Figure 1 | One-step isothermal in vitro recombination. Two adjacent DNA
fragments (magenta and green) sharing terminal sequence overlaps (black)
were joined into a covalently sealed molecule in a one-step isothermal
reaction. T5 exonuclease removed nucleotides from the 5¢ ends of double-
stranded DNA molecules, complementary single-stranded DNA overhangs
annealed, Phusion DNA polymerase filled the gaps and Taq DNA ligase
sealed the nicks. T5 exonuclease is heat-labile and is inactivated during
the 50 1C incubation.
RECEIVED 5 JANUARY; ACCEPTED 16 MARCH; PUBLISHED ONLINE 12 APRIL 2009; DOI:10.1038/NMETH.1318
1The J. Craig Venter Institute, Synthetic Biology Group, Rockville, Maryland, USA, and2San Diego, California, USA. Correspondence should be addressed to
NATURE METHODS | VOL.6 NO.5 | MAY 2009 | 343
© 2009 Nature America, Inc. All rights reserved.
6-kb linear fragment (Fig. 2b). We concluded that DNA molecules
can be assembled and repaired in a single isothermal step using
We next determined whether DNA molecules with overlaps of
only40bp couldbe joined.We accomplished this whenwe reduced
the concentration of T5 exonuclease (Fig. 2c). Three 5-kb DNA
fragments, F1–F3, were efficiently assembled into an 8-kb bacterial
artificial chromosome (BAC). Furthermore, when we transformed
these assembled DNA molecules into E. coli, we obtained 4,500
colonies, and nine out of ten colonies tested had the predicted
15-kb insert (Fig. 2d).
During the construction of the synthetic M. genitalium genome,
E. coli11. To determine whether the isothermal assembly method
could be used to join and clone DNA fragments of larger size,
we reacted two synthetic M. genitalium quarter DNA molecules,
C25–49 (144 kb) and C50–77 (166 kb), with BAC25–77 (8 kb), a
cloning vector specific for the assembly of these two DNA mole-
cules. The 318-kb Mgen25–77 product was efficiently produced, so
we conclude that DNA fragments this size can be joined by this
method (Fig.2e).Todeterminewhether this method couldbe used
to clone assembled DNA fragments this size, we transformed a
fraction of this assembly reaction into E. coli. We obtained several
hundred clones, and 5 out of 10 colonies screened had the correct
used to join and clone DNA molecules up to several hundred
kilobases in length in E. coli, the approximate upper limit for
transformation into this bacterium13. In a direct comparison of all
our assembly methods, we found that only the one-step in vitro
recombination methods could be used to clone assembled DNA
fragments this size (Supplementary Fig. 5 online).
During in vitro recombination, errors may be introduced in
the assembled DNA. However, sequencing of 30 cloned DNA
molecules (210 repaired junctions) after two-step thermocycled
assembly revealed only 4 errors (Supplementary Table 1 online).
This equates to only about 1 error per 50 DNA molecules joined.
Therefore, if our hierarchical scheme to assemble the M. genitalium
genome was used11without sequence verification at intermediate
of mutations would be even lower with the isothermal assembly
system because gaps are filled in by Phusion DNA polymerase,
which has higher fidelity than Taq polymerase.
Our isothermal method can be used to assemble DNA molecules
of unprecedented sizes, and we used it to assemble the complete
synthetic 583-kb M. genitalium genome (Fig. 2g). The size limit for
have been observed (Supplementary Fig. 6 online). Of the three
in vitro recombination methods, we prefer the one-step-isothermal
system because of its simplicity. This approach could be very useful
for cloning multiple inserts into a vector without relying on the
availability of restriction sites and for rapidly constructing large
DNA molecules. For example, regions of DNA too large to be
amplified by PCR can be divided into multiple overlapping PCR
amplicons and then assembled into one piece. The one-step
thermocycled method could be used to generate linear assemblies
as the exonuclease is inactivated during the reaction (Supplemen-
tary Figs. 4 and 5).
Synthetic biologists are engineering genetic pathways for the
production of biofuels, pharmaceuticals and industrial com-
pounds14,15. Here we provide efficient methods for constructing
these pathways, from natural or synthetic DNA.
Methods and any associated references are available in the online
version of the paper at http://www.nature.com/naturemethods/.
Note: Supplementary information is available on the Nature Methods website.
Figure 2 | Examples of the one-step isothermal
assembly method. (a) Two DNA molecules
(4,024 bp, green (i) and 2,901 bp, magenta (ii)),
overlapping by B450 bp at the termini (black,
overlaps labeled A and B), were reacted for
0–16 min to form a 6-kb circle, pRS415 (iv). Linear
assembly products (iii) were then removed by
incubation with additional tenfold excess T5
exonuclease after the 16-min incubation (T5).
(b) Not I digest of the assembled circles (pRS415)
shown in a. (c) Fragments F1–F3 were assembled
into an 8-kb PCR-amplified pCC1BAC (Epicentre)
containing 40-bp overlaps to F1 and F3, using
the indicated amounts of T5 exonuclease. The
B23-kb circular assembly products are indicated
by the arrow. (d) Not I digestion of BACs purified
from ten E. coli clones after electroporation of
the 4 U ml–1T5 exonuclease reaction shown in c.
*, correct 15-kb insert. (e) Assembly of quarter
M. genitalium genomes C25-49 and C50-77 with
BAC25-77 (ref. 11) to produce Mgen25-77.
(f) Not I digestion of BACs purified from ten
E. coli clones after electroporation of the assembly
reaction shown in e. The correct insert size (310 kb) is indicated by the arrow. (g) Assembly of quarter M. genitalium genomes C1–24, C25–49, C50–77 and
C78–101 (ref. 11) to produce a complete M. genitalium genome. DNA products were analyzed by conventional gel electrophoresis (a,b) and by field-inversion gel
electrophoresis (c–g). M, 1-kb DNA extension markers; l, lambda markers.
Time (min) at 50 °C
468 10 12 14 16 T5
T5 exonuclease (U ml–1)
344 | VOL.6 NO.5 | MAY 2009 | NATURE METHODS
© 2009 Nature America, Inc. All rights reserved.
This work was supported by the Office of Science (Biological and Environmental
Research) United States Department of Energy grant number DE-FG02-02ER63453,
and Synthetic Genomics, Inc.
D.G.G., L.Y., R.-Y.C., J.C.V., C.A.H. and H.O.S. designed research; D.G.G., L.Y.,
R.-Y.C., C.A.H. and H.O.S. performed research; D.G.G., L.Y., R.-Y.C., J.C.V., C.A.H.
and H.O.S. analyzed data; and D.G.G., C.A.H. and H.O.S. wrote the paper.
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at http://www.nature.com/naturemethods/.
Published online at http://www.nature.com/naturemethods/
Reprints and permissions information is available online at
1. Gellert, M. Proc. Natl. Acad. Sci. USA 57, 148–155 (1967).
2. Smith, H.O. & Wilcox, K.W. J. Mol. Biol. 51, 379–391 (1970).
3. Shetty, R.P., Endy, D. & Knight, T.F. Jr. J. Biol. Eng. 2, 5 (2008).
4. Yount, B., Denison, M.R., Weiss, S.R. & Baric, R.S. J. Virol. 76, 11065–11078
5. Horton, R.M., Cai, Z.L., Ho, S.N. & Pease, L.R. Biotechniques 8, 528–535
6. Horton, R.M. Mol. Biotechnol. 3, 93–99 (1995).
7. Bang, D. & Church, G.M. Nat. Methods 5, 37–39 (2008).
8. Geu-Flores, F., Nour-Eldin, H.H., Nielsen, M.T. & Halkier, B.A. Nucleic Acids Res.
35, e55 (2007).
9. Aslanidis, C. & de Jong, P.J. Nucleic Acids Res. 18, 6069–6074 (1990).
10. Li, M.Z. & Elledge, S.J. Nat. Methods 4, 251–256 (2007).
11. Gibson, D.G. et al. Science 319, 1215–1220 (2008).
12. Sayers, J.R. & Eckstein, F. J. Biol. Chem. 265, 18311–18317 (1990).
13. Sheng, Y., Mancino, V. & Birren, B. Nucleic Acids Res. 23, 1990–1996
14. Endy, D. Nature 438, 449–453 (2005).
15. Drubin, D.A., Way, J.C. & Silver, P.A. Genes Dev. 21, 242–254 (2007).
NATURE METHODS | VOL.6 NO.5 | MAY 2009 | 345
© 2009 Nature America, Inc. All rights reserved.
Preparation of DNA molecules for in vitro recombination. The
DNA molecules used in the assembly analyses were derived from
several sources including (i) the assembly intermediates of the
synthetic M. genitalium genome11, (ii) PCR products derived from
plasmids (F6 and F8), Clostridium cellulolyticum genomic DNA
(F1–F4) and Mycoplasma gallisepticum genomic DNA (F5 and F7)
and (iii) pRS415 restriction fragments. E. coli strains carrying each
of M. genitalium cassettes 66–69 (contained in pENTR223), each
of M. genitalium cassettes 78–85 (contained in pBR322), C1–24,
C25–49, C50–77 and C78–101 (each contained in pCC1BAC) or
pRS415 were propagated in LB medium containing the appro-
priate antibiotic and incubated at 30 1C or 37 1C for 16 h. The
cultures were collected and the DNA molecules were purified
using Qiagen’s HiSpeed Plasmid Maxi Kit according to the
manufacturer’s instructions, with the exception of C1–24, C25–
49, C50–77 and C78–101, which were not column-purified.
Instead, after neutralization of the lysed cells, these DNA mole-
cules were centrifuged then precipitated with isopropanol. DNA
pellets were dissolved in Tris-EDTA (TE) buffer (pH 8.0) then
RNase treated, phenol-chloroform extracted and ethanol precipi-
tated. DNA pellets were dissolved in TE buffer. Cassettes 66–69
and 78–85 were excised from the vectors by restriction digestion
with either Fau I or Bsm BI and C1–24, C25–49, C50–77 and C78–
101 were excised by digestion with Not I. To generate the 4,024-bp
and 2,901-bp overlapping fragments of pRS415, DNAwas digested
with Pvu II and Sca I or Psi I, respectively. Restriction digestions
were terminated by phenol-chloroform extraction and ethanol
precipitation. DNA was dissolved in TE buffer, then quantified by
gel electrophoresis with known DNA standards. Fragments F1–F8
were generated by PCR using the Phusion Hot Start High-Fidelity
DNA polymerase with HF buffer (NEB) according to the manu-
facturer’s instructions. PCR products were extracted from agarose
gels after electrophoresis and purified using the QIAquick Gel
Extraction kit (Qiagen) according to the manufacturer’s instruc-
tions, except DNAwas eluted from the columns with TE buffer pH
8.0. Fragments F1–F4 were amplified from Clostridium cellulolyti-
cum genomic DNA using primers F1-For and F1-Rev, F2-For and
F2-Rev, F3-For and F3-Rev, and F4-For and F4-Rev, respectively.
F5 and F7 were amplified from Mycoplasma gallisepticum genomic
DNA using primers F5-For and F5-Rev, and F7-For and F7-Rev,
respectively. F6 and F8 were amplified from pRST2 (ref. 16) using
primers F6-For and F6-Rev, and F8-For and F8-Rev, respectively.
Primer sequences are listed in Supplementary Table 2 online.
Two-step thermocycled DNA assembly. A 4? chew-back and
anneal (CBA) reaction buffer (20% PEG-8000, 800 mM Tris-HCl
pH 7.5, 40 mM MgCl2, 4 mM DTT) was used for thermocycled
DNA assembly. DNA molecules were assembled in 20-ml reactions
consisting of 5 ml 4? CBA buffer, 0.2 ml of 10 mg ml–1BSA (NEB)
and 0.4 ml of 3 U ml–1T4 polymerase (NEB). T7 polymerase can be
substituted for T4 polymerase (data not shown). Approximately
10–100 ng of each B6 kb DNA segment was added in equimolar
amounts. For larger DNA segments, proportional amounts of
DNA were added (for example, 250 ng of each 150 kb DNA
segment). Assembly reactions were prepared in 0.2 ml PCR tubes
and cycled as follows: 37 1C from 0 to 18 min as indicated in the
text, 75 1C for 20 min, 0.1 1C s–1to 60 1C, held at 60 1C for
30 min, then cooled to 4 1C at a rate of 0.1 1C s–1. In general, a
chew-back time of 5 min was used for overlaps less than 80 bp and
15 min for overlaps greater than 80 bp. Ten microliters of the CBA
reactions were then added to 25.75 ml of Taq repair buffer (TRB),
which consisted of 5.83% PEG-8000, 11.7 mM MgCl2, 15.1 mM
DTT, 311 mM each of the four dNTPs and 1.55 mM NAD. Four
microliters of 40 U ml–1Taq DNA ligase and 0.25 ml of 5 U ml–1Taq
polymerase were added and the reactions were incubated at 45 1C
for 15 min. For the T4 polymerase fill-in assembly method, 10 ml
of the CBA reaction was mixed with 0.2 ml of 10 mM dNTPs and
0.2 ml of 3 U ml–1T4 polymerase. This reaction was carried out at
37 1C for 30 min.
One-step thermocycled DNA assembly. A 4? chew-back, anneal
and repair (CBAR) reaction buffer (20% PEG-8000, 600 mM Tris-
HCl pH 7.5, 40 mM MgCl2, 40 mM DTT, 800 mM each of the four
dNTPs and 4 mM NAD) was used for one-step thermocycled
DNA assembly. DNA molecules (added in amounts described
above for CBA reactions) were assembled in 40 ml reactions
consisting of 10 ml 4? CBAR buffer, 0.35 ml of 4 U ml–1ExoIII
(NEB), 4 ml of 40 U ml–1Taq DNA ligase and 0.25 ml of 5 U ml–1
Ab-Taq polymerase (Applied Biosystems). ExoIII was diluted 1:25
from 100 U ml–1in its stored buffer (50% glycerol, 5 mM KPO4,
200 mM KCl, 5 mM 2-mercaptoethanol, 0.05 mM EDTA and
200 mg ml–1BSA, pH 6.5). DNA assembly reactions are prepared
in 0.2 ml PCR tubes and cycled using the following conditions:
37 1C for 5 or 15 min as indicated in the text, 75 1C for 20 min,
0.1 1C s–1to 60 1C, then held at 60 1C for 1 h. In general, a chew-
back time of 5 min was used for overlaps less than 80 bp and
15 min for overlaps greater than 80 bp. ExoIII is less active on 3¢
protruding termini17, which can result from digestion with certain
restriction enzymes. This can be overcome by removing the
overhangs to form blunt ends with the addition of T4 polymerase
and dNTPs, as described above, before assembly (data not shown).
One-step isothermal DNA assembly. A 5? isothermal reaction
buffer (25% PEG-8000, 500 mM Tris-HCl pH 7.5, 50 mM MgCl2,
50 mM DTT, 1 mM each of the four dNTPs and 5 mM NAD) was
used for one-step DNA assembly at 50 1C. DNA molecules (added
in amounts described above for CBA reactions) were assembled in
40 ml reactions consisting of 8 ml 5? isothermal buffer, 0.8 ml of
0.2 U ml–1or 1.0 U ml–1T5 exonuclease, 4 ml of 40 U ml–1Taq DNA
ligase and 0.5 ml of 2 U ml–1Phusion DNA polymerase. T5
exonuclease was diluted 1:50 or 1:10 from 10 U ml–1in its stored
buffer (50% glycerol, 50 mM Tris-HCl pH 7.5, 0.1 mM EDTA,
1 mM DTT, 0.1 M NaCl and 0.1% Triton X-100) depending on
the overlap size. For overlaps shorter than 150 bp, 0.2 U ml–1T5
exonuclease is used. For overlaps larger than 150 bp, 1.0 U ml–1T5
exonuclease was used. All isothermal assembly components can be
stored at –20 1C in a single mixture at 1.33? concentration for
more than one year. The enzymes are still active after more than
ten freeze-thaw cycles. To constitute a reaction, 5 ml DNA was
added to 15 ml of this mixture. Incubations were carried out at
50 1C for 15 to 60 min, with 60 min being optimal.
One-step isothermal DNA assembly protocol. Six milliliters of
5? isothermal reaction buffer were prepared by combining 3 ml of
1 M Tris-HCl pH 7.5, 150 ml of 2 M MgCl2, 60 ml of
100 mM dGTP, 60 ml of 100 mM dATP, 60 ml of 100 mM dTTP,
60 ml of 100 mM dCTP, 300 ml of 1 M DTT, 1.5 g PEG-8000 and
© 2009 Nature America, Inc. All rights reserved.
300 ml of 100 mM NAD. This buffer can be aliquoted and stored at
–20 1C. An assembly master mixture was prepared by combining
320 ml 5? isothermal reaction buffer, 0.64 ml of 10 U ml–1T5
exonuclease, 20 ml of 2 U ml–1Phusion DNA polymerase, 160 ml of
40 U ml–1Taq DNA ligase and water up to a final volume of 1.2 ml.
Fifteen microliters of this reagent-enzyme mix were aliquoted and
stored at –20 1C. This mixture can tolerate numerous freeze-thaw
cycles and remains stable even after one year. The exonuclease
amount is ideal for the assembly of DNA molecules with 20–
150 bp overlaps. For DNA molecules overlapping by greater than
150 bp, 3.2 ml of 10 U ml–1T5 exonuclease was used to prepare the
assembly master mixture above. Frozen 15 ml assembly mixture
aliquots were thawed and then kept on ice until ready to be used.
Five microliters of the DNA to be assembled were added to the
master mixture in equimolar amounts. Between 10 and 100 ng of
each B6 kb DNA fragment was added. For larger DNA segments,
proportional amounts of DNAwere added (for example, 250 ng of
each 150 kb DNA segment). Incubations were performed at 50 1C
for 15 to 60 min (60 min was optimal).
Rolling-circle amplification (RCA) of assembled products. RCA
was carried out as previously described18. One microliter of the
repaired or unrepaired reaction was mixed with 1 ml of 100 mM
NaOH and incubated at room temperature (18–22 1C) for 5 min
to denature the double-stranded DNA. One microliter of this
alkaline-treated mixture was then added to 19 ml of RCA compo-
nents in a 0.2 ml PCR tube. The final reaction concentrations for
RCA are as follows: 37 mM Tris-HCl pH 7.5, 50 mM KCl, 10 mM
MgCl2, 5 mM (NH4)2SO4, 100 mg ml–1BSA, 1 mM DTT,
3.25 mM random hexamers (Fidelity System), 1 U ml–1yeast
pyrophosphatase (United States Biochemical) and 250 U ml–1
phi29 DNA polymerase (NEB). The reaction was incubated at
30 1C for 20 h then terminated by incubation at 65 1C for 10 min.
Cloning the DNA assembly products. To clone assembled pro-
ducts, reactions were carried out in the presence of PCR-amplified
BACs containing 40 bp of overlapping sequence to the ends of the
assembled product. Not I restriction sites were also included to
allow release of the vector11. To produce BAC-F1/F3, primers
BACF1 For and BACF3 Rev were used in PCR. To produce
BAC66-69, primers BAC66 For and BAC69 Rev were used in
PCR. To produce BAC25-77, primers BAC25 For and BAC77 Rev
were used in PCR. Primer sequences are listed in Supplementary
Table 2. In general, pCC1BAC was used as DNA template.
However, for cloning Mgen25-77, a version of pCC1BAC, named
KanBAC, was constructed that contains the kanamycin resistance
gene in place of the chloramphenicol resistance gene. Samples (up
to 1 ml) of the assembly reactions were transformed into 30 ml
TransforMax EPI300 (Epicentre) electrocompetent E. coli cells in a
1-mm cuvette (BioRad) at 1,200 V, 25 mF and 200 O using a Gene
Pulser Xcell Electroporation system (BioRad). Cells were allowed
to recover at 30 1C or 37 1C for 2 h in 1 ml SOC medium then
plated onto LB medium containing 12.5 mg ml–1chloramphenicol
or LB medium containing 25 mg ml–1kanamycin. After incubation
at 30 1C or 37 1C for 24–48 h, individual colonies were selected
and grown in 3 ml LB medium with 12.5 mg ml–1chloramphenicol
or 25 mg ml–1kanamycin overnight at 30 1C or 37 1C. DNA was
prepared from these cells by alkaline lysis using the P1, P2 and P3
buffers (Qiagen) followed by isopropanol precipitation. DNA
pellets were dissolved in TE buffer containing RNase and then
digested with Not I to release the insert from the BAC.
Agarose gel analyses of assembled DNA molecules and cloned
products. U-5 field-inversion gel electrophoresis analysis was
performed on 0.8% E-gels (Invitrogen) and the parameters were
forward 72 V, initial switch 0.1 s, final switch 0.6 s, with linear
ramp and reverse 48 V, initial switch 0.1 s, final switch 0.6 s, with
linear ramp. U-2 field-inversion gel electrophoresis analysis was
performed on 1% agarose gels (BioRad) in 1? TAE buffer with
0.5 mg ml–1ethidium bromide without circulation, and the para-
meters were forward 90 V, initial switch 5.0 s, final switch 30 s, with
linear ramp, and reverse 60 V, initial switch 5.0 s, final switch 30 s,
with linear ramp. DNA bands were visualized with a BioRad Gel
Doc or an Amersham Typhoon 9410 Fluorescence Imager.
16. Lartigue, C., Duret, S., Garnier, M. & Renaudin, J. Plasmid 48, 149–159 (2002).
17. Henikoff, S. Gene 28, 351–359 (1984).
18. Hutchison, C.A., III, Smith, H.O., Pfannkoch, C. & Venter, J.C. Proc. Natl. Acad.
Sci. USA 102, 17332–17336 (2005).
© 2009 Nature America, Inc. All rights reserved.