Proc. Nati. Acad. Sci. USA
Vol. 89, pp. 392-396, January 1992
Applied Biological Sciences
Isothermal in vitro amplification of DNA by a restriction
enzyme/DNA polymerase system
(nucleic acid amplification/strand displacement/Mycobacterium tuberculosis)
G. TERRANCE WALKER*, MICHAEL C. LITTLE, JAMES G. NADEAU, AND DARYL D. SHANK
Department of Molecular Biology, Becton Dickinson Research Center, P.O. Box 12016, Research Triangle Park, NC 27709
Communicated by Ignacio Tinoco, Jr., October 14, 1991
ABSTRACT
An isothermal in vitro DNA amplification
method was developed based upon the following sequence of
reaction events. Restriction enzyme cleavage and subsequent
heat denaturation of a DNA sample generates two single-
stranded target DNA fragments (T, and T2). Present in excess
are two DNA amplification primers (P1 and P2). The 3' end of
PI binds to the 3' end ofT,, forming a duplex with 5' overhangs.
Likewise, P2 binds to T2. The 5' overhangs of P1 and P2 contain
a recognition sequence (5'-GTTGAC-3') for the restriction
enzyme HincH. An exonuclease-deficient form of the large
fragment of Escherichia coli DNA polymerase I (exo- Klenow
polymerase) [Derbyshire, V., Freemont, P. S., Sanderson,
M. R., Beese, L., Friedman, J. M., Joyce, C. M. & Steitz,
T. A. (1988) Science 240, 199-201] extends the 3' ends of the
duplexes using dGTP, dCTP, TTP, and deoxyadenosine 5'-[a-
thioltriphosphate, which produces hemiphosphorothioate rec-
ognition sites on PI-T, and P2T2. Hincd
nicks the unprotected
primer strands of the hemiphosphorothioate recognition sites,
leaving intact the modified complementary strands. The exo-
Klenow polymerase extends the 3' end at the nick on PI-T1 and
displaces the downstream strand that is functionally equivalent
to T2. Likewise, extension at the nick on P2-T2 results in
displacement ofa downstream strand functionally equivalent to
Tp. Nicking and polymerization/displacement steps cycle con-
tinuously on PI-T, and P2-T2 because extension at a nick
regenerates a nickable HincH recognition site. Target ampli-
fication is exponential because strands displaced from PI-T,
serve as targets for P2 and strands displaced from P2T2 serve
as targets for P1. A 106-fold amplification of a genomic se-
quence from Mycobacterium tuberculosis or Mycobacterium
bovis was achieved in 4 h at 37C.
A variety of in vitro nucleic acid amplification techniques are
being developed for clinical diagnosis of infectious and ge-
netic diseases. Amplification techniques can be grouped into
those requiring temperature cycling [PCR (1), ligase chain
reaction (2-4), and transcription-based amplification (5)] and
isothermal systems [self-sustained sequence replication (6)
and a Qf3 replicase system (7)] [for a comparative review, see
Kwoh and Kwoh (8)]. Strand displacement amplification
(SDA) is a DNA amplification technique that uses readily
available enzymes and does not require temperature cycling.
The technique is based upon the ability of a restriction
enzyme to nick the unmodified strand of a hemimodified
DNA recognition site, and the ability of a 5'-3' exonuclease-
deficient DNA polymerase to extend the 3' end at the nick
and displace the downstream strand. Exponential target
DNA amplification is achieved by coupling sense and an-
tisense reactions in which strands displaced from a sense
reaction serve as target for an antisense reaction and vice
versa.
Tuberculosis is one of the most common human infectious
diseases, infecting an estimated one billion worldwide with
=.16 million active cases and 3 million deaths per year (9).
Recently, it has received renewed attention due in part to its
high incidence among AIDS patients. The disease is caused
by Mycobacterium tuberculosis and Mycobacterium bovis.
Culture-based diagnosis provides exquisite sensitivity and
specificity but requires 3-6 weeks due to the slow growth rate
of most pathogenic species of mycobacteria. Acid-fast orga-
nisms can be quickly identified in stained smears but with low
sensitivity (>104 organisms per ml). Consequently, there is
an acute need for a rapid and sensitive test.
The PCR meets these diagnostic requirements and has
been applied to mycobacterial infections (10). An IS6110
insertion element, which is specific for M. tuberculosis and
M. bovis, provides an effective target sequence for amplifi-
cation and detection (11). SDA was applied to genomic DNA
samples from M. tuberculosis and M. bovis using a portion of
the IS6110 element as target sequence. We have achieved a
106-fold amplification, thereby establishing SDA as an iso-
thermal alternative for amplifying specific DNA sequences.
MATERIALS AND METHODS
Materials. The large fragment of Escherichia coli DNA
polymerase
I (Klenow polymerase) was purchased from
Boehringer Mannheim (sequencing grade). Exonuclease-
deficient large fragment of E. coli DNA polymerase I (exo-
Klenow polymerase) (12) was purchased from United States
Biochemical. Rsa I was purchased from Bethesda Research
Laboratories. HincII was purchased from New England
Biolabs at a concentration of 64 units/pkl and from Bethesda
Research Laboratories at 50 units/ptl. Ultra-pure human
placental DNA was purchased from Sigma. 2'-Deoxynucle-
oside 5'-triphosphates (dGTP, dCTP, and TTP) and deoxy-
adenosine 5'-[a-thio]triphosphate (dATP[aS]) were pur-
chased from Pharmacia. Oligodeoxynucleotides were syn-
thesized on an Applied Biosystems model 380B instrument
and purified by denaturing polyacrylamide gel electrophore-
sis. Aerosol-barrier pipet tips (PGC Scientific, Gaithersburg,
MD) were used to minimize aerosol contamination.
Isolation of Genomic DNA from M. tuberculosis and M.
bovis. Stationary-phase cultures of M. tuberculosis H37Rv
(American Type Culture Collection 27294) or M. bovis (kind-
ly supplied by Salmon Siddiqi, Becton Dickinson Diagnostic
Instrument Systems, Sparks, MD) in 60 ml of Middlebrook
7H9 broth were heat-killed at 650C for 4 h. Genomic DNA
was isolated by published procedures (13). Concentrations of
genomic DNA from M. tuberculosis and M. bovis were
calculated based upon respective genome sizes of 2.5 x 109
Abbreviations: SDA, strand displacement amplification; dNTP[aS],
dATP[aS], etc., deoxynucleoside 5'-[a-thio]triphosphate, deoxyad-
enosine 5'-[a-thio]triphosphate, etc.
*To whom reprint requests should be addressed.
392
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 89 (1992)
393
Da and 2.8 x
109 Da (14) and 20 A260 units per mg of
double-stranded DNA.
Recombinant Plasmid Containing the IS6110 Sequence ofM.
tuberculosis and M. bovis. A 1208-base-pair (bp) fragment of
the insertion sequence IS6110 (nucleotide positions 72-1279)
(15), which is specific to M. tuberculosis and M. bovis, was
isolated from genomic M. tuberculosis DNA by a PCR using
primers containing recognition sequences for EcoRI and Xba
I at their 5' ends. A single copy of the IS6110 fragment was
ligated into the EcoRI/Xba I site of pBluescript SK+ (Strat-
agene) and transformed into E. coli DH5aF cells (Bethesda
Research Laboratories). Recombinant plasmid DNA was
purified by standard alkaline lysis procedures and CsCl
centrifugation (16).
SDA Reactions. Amplification reactions were performed on
target DNA isolated from genomic M. tuberculosis DNA,
genomic M. bovis DNA, or the recombinant plasmid con-
taining a single copy of the insertion sequence IS6110 (15).
Rsa I cleavage of these DNA samples liberated a 47-bp
fragment (nucleotide positions 977-1023 of the insertion
sequence IS6110), which served as target for SDA:
5'-ACTCGACCTGAAAGACGTTATCCACCATACGGATAGGGGATCTCAGT
TGAGCTGGACTTTCTGCAATAGGTGGTATGCCTATCCCCTAGAGTCA-5'
Rsa I cleavage was performed using 10 units per gg of DNA
in 100 .ul of 50 mM Tris-HCl, pH 8/10 mM MgCl2 for 1 h at
370C followed by 2 min at 950C. Rsa I-cleaved target samples
were serially diluted into solutions containing Rsa I- cleaved
human placental DNA (10 ,ug/ml) in 50 mM Tris-HCI, pH
7.4/6 mM MgCl2/50 mM NaCI/50 mM KCI, thereby produc-
ing an array of stock target DNA solutions of various
concentrations in the presence of a constant amount of
human carrier DNA. SDA reaction mixtures (100 ,ul) con-
tained various amounts of target DNA and 0.1 ,ug of human
DNA in a solution of 100 units of HinclI, 2.5 units of exo-
Klenow polymerase, 1 mM dGTP, 1 mM dCTP, 1 mM TTP,
and 1 mM dATP[aS], 50 mM Tris HCI (pH 7.4), 6 mM MgCl2,
50 mM NaCl, 50 mM KCI, -1% glycerol (from enzyme stock
solutions), and 1 ,uM SDA primers of the following se-
quences: 5'-d(TTGAATAGTCGGTTACTTGTTGACACTC-
GACCTGAAA) and 5'-d(TTGAAGTAACCGACTATTGT-
TGACACTGAGATCCCCT)
(the HinclI
recognition se-
quence is italicized). Prior to addition of HinclI and exo-
Klenow polymerase, reaction samples were incubated 4 min
at 950C to denature the target fragment followed by 4 min at
37°C to anneal primers. Upon addition of HincIl and exo-
Klenow polymerase, amplification reaction mixtures were
incubated 1-5 h at 37°C.
Detection of SDA Products. Amplified target DNA frag-
ments and unamplified target standards were detected using
a 5'-32P-labeled probe: 5'-d(CGTTATCCACCATAC). The
detector probe was 32P-labeled using T4 polynucleotide ki-
nase in a 50-pul reaction mixture containing 1 p.M oligodeoxy-
nucleotide probe, 6.9 mM Tricine (pH 7.6), 50 mM Tris-HCl
(pH
8),
10 mM MgCl2,
5 mM dithiothreitol,
2.3 p.M
[y-32P]ATP (3000 Ci/mmol, 10 mCi/ml; 1 Ci = 37 GBq), 50
units of kinase (New England Biolabs). 32P-labeling was
carried out for 1 h at 370C and terminated by heating at 950C
for 2 min.
The SDA reactions shown in Fig. 3 were terminated after
4 h at 37°C by heating 10 min at 65°C and products were
detected as follows. The 32P-labeled probe/kinase mixture (1
,u1) was added to 100 ,ul of SDA reaction sample. The sample
was denatured 4 min at 950C and annealed 2 min at 37°C. The
32P-labeled detector probe was extended to a diagnostic
length upon addition of 2 p.l of the Klenow polymerase (1
unit/,a) and incubation for 15 min at 37°C. The extension
reaction was terminated by heating 10 min at 65°C followed
by addition of 2.6 ,ug of native calf thymus DNA (Boehringer
Mannheim) and 0.1 vol of 3 M sodium acetate. The DNA was
precipitated with 3 vol of ethanol for 10 min at room tem-
perature and recovered by centrifugation for 15 min in a
bench-top microcentrifuge. DNA pellets were dissolved in
50%o (wt/vol) urea/20 mM Na2EDTA/0.5x TBE (16)/0.05%
bromophenol blue/xylene cylanol and heated 2 min at 95°C.
The SDA reaction products shown in Fig. 4 were detected
as follows. Samples (10 ,ul) from a 100-Al SDA reaction
mixture were removed at indicated times and the reaction
was terminated by heating 2 min at 95°C. The 32P-labeled
probe/kinase mixture (2 p.l) was added to each 10-pl sample,
and the mixture was heated 2 min at 95°C and then 2 min at
37°C. The 32P-labeled detector probe was extended to a
diagnostic length upon addition of 2
,ul of the Klenow
polymerase (1 unit/,ul) and incubation for 15 min at 37°C. The
extension reaction was terminated upon addition of 14 p.l of
50%o urea/20 mM Na2EDTA/0.5 x TBE/0.05% bromophenol
blue/xylene cylanol.
Samples were analyzed by denaturing electrophoresis in a
10% (wt/vol) polyacrylamide gel [acrylamide:N,N'-meth-
yleneacrylamide (18:1)] containing 50% urea and 0.5x TBE.
32P-labeled gel electrophoresis bands were excised and ra-
dioactivity was quantified by liquid scintillation counting.
RESULTS AND DISCUSSION
SDA Design. The fundamental reaction scheme of SDA is
depicted in Fig. 1. A single-stranded target DNA fragment
and a DNA amplification primer bind at their complementary
5
primer
--G-T-T-G-A-C-
target
| polymerize using dCTP,
dGTP, TTP and dATPaS.
5
nick with Hinc 11
I-G-T-T G-A-C-
--C-AsAsC-T-G-
4?
polymerize and
G~.-C^
tdisplace strand
--G-T-T -GsA-C-
-C-AsAsC-T-G-
I nick with Hinc 11
T
_-G-T-T GsA-C-
--C-AsAsC-T-G-
4?
polymerize and
Gs,
displace strand
C-_____________
FIG.1.
Fundamental reaction scheme of SDA. A single-stranded
target DNA fragment binds to an SDA primer containing a recogni-
tion sequence for HincIl. DNA replication using dCTP, dGTP, TTP,
and dATP[aS] produces a double-stranded hemiphosphorothioate
recognition site. HinclI nicks the unprotected primer strand at its
recognition site, leaving intact the modified complementary strand
(thin strand). DNA polymerase, lacking 5'-3' exonuclease activity,
extends the 3' end at the nick and displaces the downstream
fragment. The polymerization/displacement step regenerates a nick-
able recognition site; the 5'-GsA-3' phosphorothioate linkage does
not inhibit nicking of the thick strand. Nicking and polymerization/
displacement steps cycle continuously producing single-stranded
complementary copies of the target fragment. Oligodeoxynucleotide
lengths are not drawn to scale.
Applied Biological Sciences: Walker et al.

394
Applied Biological Sciences: Walker et al.
3' ends, forming a duplex with a 5' overhang at each end. The
5' overhang of the primer strand contains a recognition
sequence (5'-GTTGAC-3') for the restriction enzyme HincIl.
An exonuclease-deficient form ofthe large fragment ofE. coli
DNA polymerase I (exo- Klenow polymerase) (12) extends
the 3' ends of the duplex using dGTP, dCTP, TTP, and
dATP[aS], which produces a hemiphosphorothioate recog-
nition site. HinclI nicks the unprotected primer strand of the
hemiphosphorothioate recognition site, leaving intact the
modified complementary strand (thin strand in Fig. 1). The
exo- Klenow polymerase extends the 3' end at the nick and
displaces the downstream complement of the target strand.
The polymerization/displacement step regenerates a nick-
able HinclI recognition site; the 5'-GsA-3' phosphorothioate
linkage does not inhibit nicking of the thick strand in Fig. 1.
Nicking and polymerization/displacement steps cycle con-
tinuously producing single-stranded complements of the tar-
get strand.
Fig. 2 depicts SDA using two primers (P1 and P2) with a
double-stranded target DNA fragment (T1.T2), which is gen-
erated by restriction enzyme cleavage of a DNA sample.
After heat denaturation, single-stranded target fragments T1
and T2 bind to respective primers P1 and P2, which are present
in excess. Both primers contain restriction enzyme recogni-
tion sequences located 5' to the target binding sequences.
Each primer target complex (Pl T1 and P2-T2) cycles through
nicking and polymerization/displacement steps in the pres-
ence of a restriction enzyme, DNA polymerase, three
dNTPs, and one dNTP[aS] as illustrated in Fig. 1. SDA with
a single primer produces copies ofa target fragment at a linear
amplification rate (Fig. 1). In contrast, amplification using
two primers proceeds exponentially because strands dis-
placed from the Pj T1 complex serve as target for primer P2
and strands displaced from the P2-T2 complex serve as target
for primer P1 (Fig. 2).
5,
T2
T,
5
P5
denature and
hybridize SDA primers
T2
P2
polymerize using
I dNTPs and dNTPctS
hybridize SDA primers to displaced strands
FIG.
2.
Exponential SDA using two primers with a double-
stranded target DNA fragment. After denaturation of a double-
stranded target fragment, single-strand target fragments T1 and T2
bind to respective primers P1 and P2, which are present in excess.
The primers contain recognition sites for the nicking restriction
enzyme. DNA replication using three dNTPs and one dNTP[aS]
produces double-stranded primertarget complexes with hemiphos-
phorothioate recognition sites, which are subsequently nicked by the
appropriate restriction enzyme. DNA polymerase, lacking 5'-3'
exonuclease activity, extends the 3' end at the nick and displaces the
downstream strand. Nicking and polymerization/displacement steps
cycle continuously due to regeneration of a nickable recognition site.
Exponential amplification results because strands displaced from the
Pj-T1 complex serve as target for primer P2 and strands displaced
from the P2-T2 complex serve as target for primer P1.
Restriction Enzymes and DNA Polymerases Suitable for
SDA. Eckstein and coworkers (17) have identified many
restriction enzymes that nick hemiphosphorothioate recog-
nition sites. Only a small subset of these restriction enzymes
function in SDA because of the special requirement that the
polymerization/displacement step must regenerate a nick-
able recognition site. Phosphorothioate linkages incorpo-
rated into the primer strand must not inhibit subsequent
nicking. Suitable restriction enzymes identified to date sat-
isfy this requirement through nonpalindromic recognition
sites. In addition to HincII used in the present study, HindII,
Ava I, Nci I, and Fnu4HI also function in SDA (G.T.W.,
M.C.L., and J.G.N., unpublished results).
To function adequately in SDA, a restriction enzyme must
efficiently nick one strand of a hemiphosphorothioate recog-
nition site and quickly dissociate from the nicked site to allow
the DNA polymerase access. Many restriction enzymes
efficiently nick a nonpalindromic hemiphosphorothioate rec-
ognition site but apparently dissociate too slowly from the
nicked site for efficient operation in SDA (M.C.L., unpub-
lished results).
Clearly, it is essential that the target DNA sequence does
not contain a recognition sequence for the restriction enzyme
used in SDA (HincII in the present investigation).
To function in SDA, a DNA polymerase must (i) initiate
polymerization at a nick, (ii) utilize dNTP[aS]s, and (iii)
displace a downstream strand (and lack 5'-3' exonuclease
activity). In addition to the exo- Klenow polymerase used in
the present study, the Klenow polymerase (sequencing
grade, Boehringer Mannheim) and the large fragment of
Bacillus stearothermophilus (Bio-Rad) also function in SDA
(G.T.W. and M.C.L., unpublished results).
SDA Primer Design. Each SDA primer contains a target
binding region at its 3' end. A recognition site for the nicking
restriction enzyme is located 5' to the target binding region.
The recognition sequence is flanked on the 5' side by a stretch
of DNA typically
-20 nucleotides long, the sequence of
which is not critical. The 5' end of a primer provides flanking
DNA to facilitate binding ofthe nicking restriction enzyme to
the recognition site. The 5' end also must be long enough to
provide a stable polymerase priming site after nicking by the
restriction enzyme. The SDA primers in the current inves-
tigation are partially complementary at their 5' ends. This
was designed to reduce nonspecific binding. However, 5'
complementarity between primers is not essential for SDA.
SDA primers undoubtedly bind nonspecifically at 370C to
nontarget DNA present in the sample (human placental DNA
in the current investigation). However, SDA not only re-
quires that primers bind to DNA fragments with adequate
homology but also requires an appropriately positioned 3'
end on the target strand to form a hemiphosphorothioate
recognition site for the nicking restriction enzyme. Use of a
3'-5' exonuclease-deficient DNA polymerase enhances
primer specificity in SDA because such a DNA polymerase
is less likely to extend nonspecific complexes containing
mismatched 3' ends.
In the PCR, primers can bind nonspecifically to one
another, resulting in amplification of what are commonly
referred to as primer dimers. These background amplification
reactions can effectively attenuate target-specific amplifica-
tion by competing for PCR reagents. Nonspecific binding
between SDA primers can also result in amplification of
primer dimers. Consequently, the terminal 3' nucleotides of
each SDA primer are chosen to minimize formation ofprimer
dimers that can be extended by DNA polymerase. Again, use
of a 3'-5' exonuclease-deficient DNA polymerase reduces
extension of primer dimers with mismatched 3' ends. Exten-
sion of nonspecific complexes can be further reduced by
using SDA primers whose 3'-hydroxyl groups are modified to
prevent extension by DNA polymerase. It is not essential
Proc. Natl. Acad Sci. USA 89 (1992)

Proc. Natl. Acad. Sci. USA 89 (1992)
395
that an SDA primer be extended upon binding to a target
fragment. The hemiphosphorothioate recognition site is still
formed through extension ofthe 3' end ofthe target fragment.
Primers with modified 3'-hydroxyl groups also allow use of
DNA polymerases possessing 3'-5' exonuclease activity
(e.g., the Klenow polymerase) by ensuring excess single-
stranded primers are not degraded during SDA.
SDA of Genomic DNA from M. tuberculosis and M. bovis.
Rsa I cleavage of genomic DNA from either M. tuberculosis
or M. bovis produced a 47-bp target fragment for SDA. A set
of samples containing various amounts of the target DNA
fragment and 0.1 ug of Rsa I-cleaved human placental DNA
was subjected to SDA using the exo- Klenow polymerase
and HincII. Amplified target fragments were detected by
DNA polymerase extension of a 32P-labeled probe that is
complementary to an internal segment ofthe target sequence.
The 32P-labeled detector probe {[5'-32P]d(CGTTATCCAC-
CATAC)} is extended to a length ofeither 35 or 56 nucleotides
when hybridized to one of the following two strands pro-
duced during SDA: 5'-d(GACACTGAGATCCCCTATCCG-
TATGGTGGATAACGTCTTTCAGGTCGAGTGTC) and
5'-d(TTGAAGTAACCGACTATTG1TGACACTGAGATC-
CCCTATCCGTATGGTGGATAACGTCTTTCAGGTC-
GAGTGTC). These two strands are respectively analogous
to the displaced strand and the unnicked strand (containing
the sequence 5'-GTTGsAC-3') in Fig. 1.
Results from amplification of genomic DNA from M.
tuberculosis and M. bovis are shown in Fig. 3. The number
of genome copies (copies of the single Mycobacterium chro-
mosome) initially present in each SDA sample is indicated
above each lane in Fig. 3. There are 10-12 copies ofthe 47-bp
target DNA fragment in the genome of M. tuberculosis. In
comparison, there are 1-3 copies in the genome of M. bovis
(ref. 18; D.D.S., unpublished observations). Consequently,
band intensities from each ofthe M. tuberculosis samples are
-10-fold higher than the corresponding M. bovis samples due
to this difference in the number of target fragments per
genome. Intensities of the SDA product bands decrease as a
function of decreasing initial target over the array of M.
tuberculosis or M. bovis samples. For the experiments shown
in Fig. 3, a maximum signal is exhibited for amplification of
>103 genome copies of M. tuberculosis and >l04 genome
M. tuberculosis
NI' bovis
copies of M. bovis due to the limiting amount of the 32p_
labeled detector probe. Reactions performed without addi-
tion of target DNA exhibit faint SDA-specific 32P-labeled
bands in Fig. 3 due to accidental contamination of reagents
with '-10 target molecules (=1 M. tuberculosis or -10 M.
bovis genome copies). It is extremely difficult to maintain
reagents that are free of target DNA contaminants on the
scale of 10 molecules. Contamination control is a common
problem with the PCR as well (19).
A recombinant plasmid containing a single copy of the
IS6110 sequence served as a standard for estimating the
degree of amplification in SDA reactions. (The plasmid was
used as a standard because it was available in sufficient
quantity.) Comparison of amplified samples with standards
allows us to estimate an amplification factor of 106-fold. A
constant amplification factor is observed for each sample,
regardless of the initial amount of target.
The current method of detection using a 32P-labeled probe
allows detection of as few as -108 DNA molecules. This
sensitivity in conjunction with 106-fold amplification allows
clear detection of as few as 10 genome copies of M. tuber-
culosis and 100 genome copies of M. bovis using the 47-bp
Rsa I fragment from the insertion element IS6110 as target.
We have not yet examined the replication fidelity of the
exo
Klenow polymerase during SDA. Detection using a
target-sequence-specific 32P-labeled probe indicates fidelity
is sufficient for diagnostic applications. However, fidelity
may be a concern in other applications. Only a few pertinent
investigations are published. The exo- Klenow polymerase is
4- to 7-fold less accurate than the Klenow polymerase (20). E.
coli DNA polymerase I exhibits a 20-fold decrease in fidelity
when dCTP is replaced with dCTP[aS] (21). This decrease in
fidelity was attributed to inhibition of the 3'-5' exonuclease
proofreading activity of DNA polymerase I by dCTP[aS].
The fidelity of avian myeloblastosis virus DNA polymerase
and DNA polymerase (3, which lack 3'-5' proofreading
activity, was not altered by dCTP[aS] substitution. Finally,
Taq DNA polymerase accurately incorporates dNTP[aS]
using a template containing phosphorothioate nucleotides as
judged by sequencing PCR products (22).
SDA Reaction Kinetics. Reaction kinetics were studied
using the recombinant plasmid that contains a single copy of
the IS6110 target fragment. A time-course profile for the
initial stage of an SDA reaction is shown in Fig. 4A Inset.
()4
1()
10)2
()
0)
FIG. 3.
Denaturing gel electrophoresis analysis ofSDA products.
SDA reaction mixtures contained 0.1 Ag of Rsa I-cleaved human
placental DNA and various amounts of target genomic DNA from
either M. tuberculosis or M. bovis. SDA reactions proceeded for 4
h. Amplified target fragments were detected by extension of a
32P-labeled probe to the indicated lengths. The number of copies of
the single Mycobacterium chromosome initially present in each SDA
reaction is shown above each lane.
109
--
u'108
00i0
4)~~~~~4
0
Time, mi
1
3
5
1
3
5
Time, h
FIG. 4.
Time-course profiles of SDA reactions. The SDA target
was a recombinant plasmid containing a single copy of the IS6110
fragment. (A Inset) Observed exponential amplification for an SDA
reaction with 1 x 108 initial target molecules per ,ul at time 0. The
observed doubling time is 5.3 min. (A) Semilogarithmic plot of
theoretical time courses for SDA reactions with various amounts of
initial target DNA. Calculated curves (Eq. 1) are based upon a
doubling timie of 5.3 min and an SDA reagent at a limiting concen-
tration of 1x 1010 molecules per ,ul. (B) Semilogarithmic plot of
observed time courses for experimental SDA reactions with various
amounts of initial target DNA.
(;L)nic Copspis
104
1()3
1(-
10
Applied Biological Sciences: Walker et al.

396
Applied Biological Sciences: Walker et al.
Exponential amplification is observed with a doubling time
(cycle time) X of 5.3 min (P, = P02'/, where Pt is the number
of product molecules at time t and PO is the number of target
molecules at time 0). The following sequential steps occur
within each 5.3-min cycle as illustrated in Figs. 1 and 2: (i)
nicking of the hemiphosphorothioate recognition
site by
HincII, (ii) dissociation of HincII from the nicked site, (iii)
polymerization and strand displacement by the exo- Klenow
polymerase, (iv) hybridization of the displaced strand to the
other primer, and (v) extension of the displaced strand and
primer forming another hemiphosphorothioate recognition
site.
The net result ofeach cycle is a doubling ofproduct number
(P) and consumption of reagents (R):
P + R
--
2P.
If reagents are not replenished during the course of the
reaction, the level of product at time t is given by:
Pt = (ROQek')/(j + Qek)
[1]
where Q = Po/(Ro - PO), PO is the initial product (target)
concentration, and Ro is initial concentration of limiting
reagent. The apparent rate constant k is related to the cycling
time r by k = (In 2)/r.
An SDA reaction should proceed exponentially until the
concentration of some reagent R becomes limiting. Fig. 4A is
a theoretical time course profile for SDA reactions with
various amounts of initial target DNA. The curves were
calculated from Eq. 1 based upon a cycle time Tof 5.3 min and
a limiting reagent concentration RO of 1010 molecules per 1ul.
Product concentration is expected to increase exponentially
until it reaches a level -5-fold less than RO. The exponential
phase is then followed by a much less efficient amplification
phase as the product concentration approaches RO. The final
product concentration should be independent of the initial
target concentration. t
Time courses of experimental SDA reactions are shown in
Fig. 4B. Each experimental SDA reaction exhibits exponen-
tial amplification over the first 60 min. By the 2-h mark, each
reaction has switched from exponential amplification to a
much less efficient amplification phase, during which product
concentration asymptotically approaches a level that de-
pends on the initial amount of target. In each reaction, the
number of target molecules is amplified by a constant amount
(_106-fold after 5 h), regardless of the number of initial
targets. This results in a dose-response relationship between
initial and final target levels, in sharp contrast to the results
predicted by the theoretical curves in Fig. 4A. A similar
dose-response relationship was reported for another ampli-
fication technique referred to as self-sustained sequence
replication (6).
In SDA reactions, addition of fresh enzymes at the 1-h
mark does not prolong the exponential phase, indicating
enzyme degradation is not responsible for the reduction in
amplification efficiency after that point. The observed dose-
tFig. 4A is intended to be a qualitative illustration of the reaction
kinetics expected for SDA. Eq.
1 is quantitatively correct only if
reagent R is consumed in the rate-determining reaction step. (We
have yet to establish the contribution of individual steps to the
overall reaction rate.) If R does not participate in the rate-
determining step, we still expect the same limiting behavior defined
by Eq. 1, namely, (i) exponential growth during the early phase of
the reaction with a doubling time of 5.3 min and (ii) a final product
concentration that is independent of initial target levels and ap-
proaches a value ofRo. Furthermore, the growth ofPt should begin
to deviate from an exponential increase at the same level of Pt,
regardless of initial target concentration.
response relationship probably results from amplification of
nontarget DNA (human placental DNA) that is present in
large excess over specific target sequences. Although the
intrinsic efficiency of this background amplification is ex-
pected to be very low, the high level of nontarget DNA may
lead to significant accumulation ofundesired products. These
background reactions should not interfere with target ampli-
fication early in the reaction when reagents are present in
excess. However, the exponential amplification phase (dur-
ing which the number of target sequences doubles every 5.3
min) can only be sustained while the concentration of limiting
reagent R remains at least 5- to 10-fold higher than amplified
product levels (see Eq. 1 and Fig. 4A). Competing back-
ground reactions will prematurely deplete reagent levels and
as a consequence will shorten the period during which
exponential amplification of target sequences can occur. In
addition, competing reactions will limit the final quantity of
amplified target sequences that can be produced because
reagents converted to background products will be unavail-
able for the production of target sequences. Consequently, a
dose-response relationship will result because the degree to
which target amplification can compete against background
reactions will depend on the ratio of target to nontarget
sequences present at the beginning of the reaction.
1.
Saiki, R. K., Scharf, S., Faloona, R., Mullis, K. B., Horn,
G. T., Erlich, H. A. & Arnheim, N. (1985) Science 230, 1350-
1354.
2.
Wu, D. Y. & Wallace, R. B. (1989) Genomics 4, 560-569.
3.
Barringer, K., Orgel, L., Wahl, G. & Gingeras, T. R. (1990)
Gene 89, 117-122.
4.
Barany, F. (1991) Proc. Natl. Acad. Sci. USA 88, 189-193.
5.
Kwoh, D. Y., Davis, G. R., Whitfield, K. M., Chappelle,
H. L., DiMichele, L. J. & Gingeras, T. R. (1989) Proc. Natl.
Acad. Sci. USA 86, 1173-1177.
6.
Guatelli, J. C., Whitfield, K. M., Kwoh, D. Y., Barringer,
K. J., Richman, D. D. & Gingeras, T. R. (1990) Proc. Natl.
Acad. Sci. USA 87, 1874-1878.
7.
Lizardi, P. M., Guerra, C. E., Lomeli, H., Tussie-Luna, I. &
Kramer, F. R. (1988) BioTechnology 6, 1197-1202.
8.
Kwoh, D. Y. & Kwoh, T. J. (1990) Am. Biotechnol. Lab. 8,
14-25.
9.
Snider, D. E. (1989) Rev. Infect. Dis. 11, S336-S338.
10.
Pierre, C., Lecossier, D., Boussougnat, Y., Bocart, D., Joly,
V., Yeni, P. & Hance, A. J. (1991) J. Clin. Microbiol. 29,
712-717.
11.
Eisenach, K. D., Cave, M. D., Bates, J. H. & Crawford, J. T.
(1990) J. Infect. Dis. 161, 977-981.
12.
Derbyshire, V., Freemont, P. S., Sanderson, M. R., Beese, L.,
Friedman, J. M., Joyce, C. M. & Steitz, T. A. (1988) Science
240, 199-201.
13.
Visuvanathan, S., Moss, M. T., Stanford,
J. L., Hermon-
Taylor, J. & McFadden, J. J. (1989) J. Microbiol. Methods 10,
59-64.
14.
Bradley, S. G. (1973) J. Bacteriol. 113, 645-651.
15.
Thierry, D., Cave, M. D., Eisenach, K. D., Crawford, J. T.,
Bates, J. H., Gicquel, B. & Guesdon, J. L. (1990) Nucleic
Acids Res. 18, 188.
16.
Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular
Cloning:A Laboratory Manual (Cold Spring Harbor Lab., Cold
Spring Harbor, NY).
17.
Olsen, D. B., Kotzorek, G. & Eckstein, F. (1990) Biochemistry
29, 9546-9551.
18.
Cave, M. D., Eisenach, K. D., McDermott, P. F., Bates, J. H.
& Crawford, J. T. (1991) Mol. Cell. Probes 5, 73-80.
19.
Erlich, H. A., Gelfland, D. & Sninsky, J. J. (1991) Science 252,
1643-1651.
20.
Bebenek, K., Joyce, C. M., Fitzgerald, M. P. & Kunkel, T. A.
(1990) J. Biol. Chem. 265, 13878-13887.
21.
Kunkel, T. A., Eckstein, F., Mildvan, A. S., Koplitz, R. M. &
Loeb, L. A. (1981) Proc. Natl. Acad. Sci. USA 78, 6734-6738.
22.
Kakamaye, K. L., Gish, G., Eckstein, F. & Vosberg, H.-P.
(1988) Nucleic Acids Res. 16, 9947-9959.
Proc. Natl. Acad. Sci. USA 89 (1992)