Isothermal reactions for the amplification
of oligonucleotides
Jeffrey Van Ness, Lori K. Van Ness, and David J. Galas*
Keck Graduate Institute of Applied Life Science, 535 Watson Drive, Claremont, CA 91711
Communicated by Eric H. Davidson, California Institute of Technology, Pasadena, CA, February 11, 2003 (received for review November 30, 2002)
We have devised a class of isothermal reactions for amplifying
DNA. These homogeneous reactions rapidly synthesize short oligonucleotides (8 –16 bases) specified by the sequence of an amplification template. Versions of the reactions can proceed in either
a linear or an exponential amplification mode. Both of these
reactions require simple, constant conditions, and the rate of
amplification depends entirely on the molecular parameters governing the interactions of the molecules in the reaction. The
exponential version of the reaction is a molecular chain reaction
that uses the oligonucleotide products of each linear reaction to
create producers of more of the same oligonucleotide. It is a highly
sensitive chain reaction that can be specifically triggered by given
DNA sequences and can achieve amplifications of >106-fold. Several similar reactions in this class are described here. The robustness, speed, and sensitivity of the exponential reaction suggest it
will be useful in rapidly detecting the presence of small amounts
of a specific DNA sequence in a sample, and a range of other
applications, including many currently making use of the PCR.
T
he invention of the PCR changed the practice of molecular
biology. It has become a mainstay of biological research and
diagnostics in providing a method for the rapid detection,
isolation, and measurement of DNA sequences through their
specific amplification. There are currently two widely used
methods for amplifying specific DNA sequences: PCR (1, 2) and
the rolling-circle amplification method (3–5). The PCR method
is the simpler and more flexible of these and has the added
advantage of being geometric rather than linear in character, so
that amplification levels of 106 or more can be achieved. It is by
far the most widely used amplification method in biology. It has
the disadvantage relative to the isothermal rolling-circle amplification method, however, of needing a temperature cycling
protocol to achieve amplification. This imposes instrumentation
constraints on the PCR method that make it more complex and
limit the rate of the amplification to the temperature cycling
schedule. Another limitation of the rate of PCR derives from the
nature of the reaction itself in that a maximum 2-fold amplification can be achieved in each cycle. Advances in speed,
accuracy, and sensitivity, in addition to simplicity, would be most
welcome for applications in biology and medicine.
We report here a class of isothermal reactions for amplifying
DNA that overcomes all the above disadvantages of PCR. This
class includes a linear amplification method, which is fundamental to the others, and several versions of an exponential amplification scheme. These reactions are simple, flexible, and require
no special cycling of conditions. They depend entirely for their
rate of amplification on the molecular parameters governing the
interactions of the molecules in the reaction. Because of the
balance between the thermal properties of the DNA oligonucleotides and the enzymes used, the optimum temperature of the
reaction with these enzymes is 60°C (see Materials and Methods).
The exponential version of the method, designated the exponential amplification reaction (EXPAR), is an isothermal molecular chain reaction in that the products of one reaction
catalyze further reactions that create the same products.
4504 – 4509 兩 PNAS 兩 April 15, 2003 兩 vol. 100 兩 no. 8
Materials and Methods
Oligonucleotides and Enzymes. Sequences used in experiments
described in the text:
Oligonucleotides used in the linear reaction as described in
Figs. 1 and 2: ITAtop, 5⬘-CCGATCTAGTGAGTCGCTC-3⬘;
NBbt12, 5⬘-ACGACTGGA ACTGAGCGACTCACTAGATCGG-3⬘; NBbt16, 5⬘-ACCTACGACTGGAACTGAGCGACTCACTAGATCGG-3⬘; NBbt20, 5⬘-TGAAACCTACGACTGGAACTGAGCGACTCACTAGATCGG-3⬘.
Oligonucleotides used in the exponential reaction described in
Figs. 3 and 4: template oligo, ceap, 5⬘-CCTACGACTGGaacaGACTCACCTACGACTGGA P-3⬘; trigger, seqS, 5⬘-ACCAGTCGTAGG-3⬘ (spacer bases are indicated in lowercase; P
indicates phosphate group; the nicking enzyme site or its complement is indicated by an underline in all above sequences).
Note that the trigger oligo above (seqS) is one base longer than
that produced by the primer template. This enables us to
distinguish the initial trigger from the amplified sequence and
does not affect its ability to prime effectively.
Oligonucleotides were synthesized by Midland Certified Reagent Company (Midland, TX), MWG Biotech (High Point,
NC), or Sigma–Genosys (The Woodlands, TX). The oligonucleotides were routinely checked by time-of-flight MS [using LCT
from Micromass (Manchester, U.K.); see below].
All enzymes were purchased from New England Biolabs. The
DNA polymerase used was Vent exo- (6, 7). The nicking enzyme
(N.BstNBI) has a specific activity of ⬇106 units兾mg (H.-M.
Kong, personal communication).
All HPLC components (water and acetonitrile) were purchased from Fisher Scientific. Dimethyl-butylamine was purchased from Sigma–Aldrich, and a salt was made by addition of
acetic acid (Sigma–Aldrich) to pH 7.1. The 2 M stock solution
was filtered by using a 0.2-m nylon filter.
Linear Amplification Reaction. The conditions for the linear reac-
tion were: 85 mM KCl兾25 mM Tris䡠HCl (pH 8.8, 25°C)兾2.0 mM
MgSO4兾5 mM MgCl2兾10 mM (NH4)2SO4兾0.1% (vol兾vol) Triton X-100兾0.5 mM DTT兾0.4 units/l N.BstNBI nicking enzyme兾
0.05 units/l Vent exo⫺ polymerase兾400 M dNTPs (Epicentre,
Madison, WI)兾10 g/ml BSA兾0.05 M template and primer
olignucleotides [NBtop and NB12 (equimolar) in ultrapure
water that is nuclease-free (Ambion, Austin, TX)]. These conditions correspond to 1 part Thermopol buffer and 0.5 parts
N.BstNBI buffer as supplied by New England Biolabs. Reactions
were assembled at 4°C, initiated by transferring to a preheated
thermocycler at 60°C, and stopped by incubation at 4°C. No
further manipulations were performed before placement on the
autoinjector for the HPLC-MS, which is held at 4°C.
Exponential Amplification Reaction. The exponential reactions
were also carried out at 60°C, temperature controlled to within
0.1°C. The exponential reaction conditions were as follows: the
same as described above for the linear reaction except with 0.1
Abbreviation: EXPAR, exponential amplification reaction.
*To whom correspondence should be addressed. E-mail david㛭galas@kgi.edu.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0730811100
in electrospray negative mode, ranging from 800 to 2,000 amu,
1-sec scan time. Analysis of the HPLC-MS data made use of the
software supplied by the manufacturer.
Oligonucleotides are known to exhibit different ionization
efficiencies, which in our measurements would be translated into
sequence-specific differences in measured oligo concentration.
A survey of a range of ⬎80 different 12 mers indicated that the
variation between sequences attributable to this difference is
⬍30%. Almost all relevant quantitative comparisons are with
the same oligo sequence. It is necessary, however, to calibrate for
ionization efficiencies for quantitative comparisons between
different sequences.
Real-Time Fluorescence Measurement. All fluorescence measure-
ments reported here were made on an MJ Opticon instrument
(MJ Research, Waltham, MA) by using software supplied by the
manufacturer. The real-time measurements on this instrument
were made by using an isothermal protocol with a 30-sec interval
read beginning 10 sec after the lid and chamber reached 60°C.
The Exponential Reaction Equations. The simplified mass action
equations use the following variables: a, the annealing rate
between the product oligonucleotide concentration, , and the
amplification template concentration, ; , the concentration of
the transient complex between and ; , the concentration of
the primer template formed by extension of the complex; c, the
rate of conversion of to ; r, the rate of oligonucleotide
production () by each primer template. The equations, using
the simplifying assumptions that annealing is a single-step bimolecular reaction and that the conversion of into can be
represented as a simple effective rate, are then,
d
⫽ r ⫺ a ,
dt
M template oligonucleotide only (unless otherwise noted).
Triggering oligonucleotides were added as described for each
experiment. In the case of fluorescence monitoring, SYBR green
(Molecular Probes) was added to 5⫻ concentration (SYBR
green is supplied by the manufacturer at 10,000⫻).
Chromatography and MS. The chromatography system was an
Agilent (Palo Alto, CA) 1100 Series HPLC composed of a binary
pump, degasser, a column oven, a diode array detector, and
thermostated microwell plate autoinjector. The column is a
Waters Xterra MS C18, incorporating C18 packing with 3.5 M
particle size, with 125-Å pore size, 2.1 mm ⫻ 20 mm. The column
was run at 30°C with a gradient of acetonitrile in 5 mM
dimethyl-butylamine acetate (DMBAA). As a check on the
complete release of the signal oligo during the chromatography
and injection, we ran the column at 50°C after incubating the
sample briefly at 95°C. We saw no increase in the oligo yield over
our standard conditions. Buffer A is 5 mM DMBAA, and buffer
B is 5 mM DMBAA and 50% (V兾V) acetonitrile. The MS was
a Micromass LCT time-of-flight instrument. Samples were run
Van Ness et al.
d
⫽ c,
dt
d
⫽ ⫺a.
dt
can easily be shown from these equations to exhibit exponential behavior. The exponential phase occurs before the template
becomes depleted, but after reaches a steady ratio with . In
this regime, the equation for has the approximate solution, ⬇
oet, where  ⫽ ( (0)ar兾2) 1/2 ⫺ a (0).
A more direct method is simply to solve the above equations
computationally using a direct finite difference method. The
results of a computational solution of the equations in Fig. 2b
show clearly the regime in which the exponential solution
applies.
Results
Linear Amplification. To produce an amplification reaction, we
need to devise a cyclic chain of reactions that will restore the
reactants to their initial state after each synthesis of the molecule
to be amplified. The linear amplification reaction described here
provides such a cycle whose sequence specificity derives from
template-dependent synthesis of the oligonucleotide to be amplified. The reaction synthesizes short oligonucleotides whose
cycle of reactions depends on the idea that, at the reaction
temperature, oligonucleotides above a certain length form stable
duplexes, whereas those below this critical length form unstable
duplexes that dissociate readily. By arranging a specific singlestrand nicking site and nicking enzyme and a compatible DNA
polymerase (6, 7) as described in Fig. 1, a cycle of polymerization
and subsequent oligonucleotide release is created. This cycle
depends on the nicking reaction cleaving a phosphodiester bond
to create an oligonucleotide that is below the threshold of
stability in a duplex and is thereby released from the duplex, thus
regenerating the initial primer template. The synthesized oligoPNAS 兩 April 15, 2003 兩 vol. 100 兩 no. 8 兩 4505
BIOCHEMISTRY
Fig. 1. (a) The cycle of the synthesis and release of the amplified oligonucleotide is shown schematically. On the upper strand is indicated the recognition site for the enzyme N BstNB (5⬘-GAGTC-3⬘) and the specific nicking site
four bases downstream on this strand. The oligonucleotide produced is indicated in blue, the primer in green, and the template in red. The lengths of the
template and amplified oligo are shown (Upper Left). (b) The results of a linear
amplification reaction where the primer template produces a 12 mer as the
full-length product. The primer template was present at 1 M in a 50-l
reaction (see Materials and Methods), and the yield of the reaction products
is shown. The duplex used a top strand (ITAtop) of 16 nucleotides and a
bottom strand (NBbt12) of 28 nucleotides that produced a 12 mer.
d
⫽ a ⫺ c ,
dt
Fig. 2. Exponential amplification reactions. (a) Diagram of the reaction scheme for the exponential amplification of oligonucleotides. The segments in red
represent the sequence complement of the oligonucleotide sequence to be amplified, the signal sequence (shown in blue). The amplification template, t, consists
of two copies of the signal complement flanking the nicking enzyme recognition site, shown as a light blue box, and a spacer sequence, shown as a green segment.
The signal oligonucleotide (labeled ) is produced in the linear amplification cycle for each amplification template created. The labels on each structure in the
figure correspond to the symbols used for their concentrations in the equations. (b) MS measurement results for the reaction. The oligonucleotide concentration
(M) of the oligonucleotide ( in the equations) was measured as described in Materials and Methods. The initial point is not measurable in the mass spectrometer
and is the initial concentration introduced into the reaction. The template oligos and trigger oligos are shown in Materials and Methods. Solution of the
differential equations in the text describing the mass-action kinetics of the reaction scheme shown in a. The kinetic parameters used for the solution (see Materials
and Methods) were: r ⫽ 0.4 sec⫺1; a ⫽ 2 ⫻ 10⫺5 M⫺1䡠sec⫺1; c ⫽ 2 sec⫺1. The theoretical curves are shown as heavy lines. Parameter c was chosen to give a reasonable
fit to the data, although the curve is not very sensitive to this parameter. The other parameters are determined as described. The initial (trigger) concentrations
were chosen to match the curves in b. The curve for the higher concentration of trigger (10⫺11) is indicated by the blue line. The curve of for the lower
concentration corresponds to the lower curve (green). (c) ‘‘Real-time’’ fluorescence monitoring of the EXPAR reaction. The reaction was carried out under the
conditions of Fig. 3. The trigger oligonucleotide, , was present at 10⫺5 M at time 0. The fluorescence of SYBR green was monitored every 30 sec in six
independent identical reactions. The error bars indicate the standard deviations of these reactions at each time point.
nucleotide is fully stable at 60°C when it is covalently joined to
the rest of the upper stand, as it is immediately after its synthesis,
but is only transiently stable as a 12 mer after the nicking
reaction. Therefore, when the bond is cleaved at the nicking site,
the oligonucleotide dissociates recreating a primer template,
ready for elongation. This cycle thus generates oligonucleotides
that are complementary to the template beyond the nicking site
(shown in blue in Fig. 1a).
When the nicking enzyme is present with a compatible
polymerase, the reaction proceeds around the cycle shown in Fig.
1a, and amplification of the product oligonucleotide occurs. In
Fig. 1b, we show the results of one of these reactions. The
experiment was devised to produce a 12 mer as its amplified
product. The products of the reaction were analyzed on the
LC-MS system after the indicated incubation times at 60°C (see
Materials and Methods). Because the exact masses of all of the
relevant molecules are known, the relative concentrations of all
of the components, including the amplified oligonucleotide, can
be directly measured. The yield of oligonucleotide is perhaps
best characterized in this case as the number of molecules
produced per primer template per second. For the experiment
shown in Fig. 1, this initial rate is about one molecule (12 mer)
per primer template every 2.5 sec, or ⬇0.4 molecules per primer
template 䡠 sec. Note that the reaction slows down noticeably after
10 min or so. This is consistent with the reaction rate declining
exponentially, as if an essential component of the reaction is
being inactivated. We expect the nicking enzyme is responsible,
as the optimum temperature (⬇55°) of the enzyme is lower than
the 60°C of the reaction, and preliminary experiments show a
clear difference in the rate decline between different starting
nicking enzyme concentrations: more enzyme makes the reaction stay linear longer. Further experiments, however, are
needed to verify this hypothesis. An extensive set of experiments
(data not shown) show that the absolute initial rate of the
reaction is proportional to the primer-template concentration, as
expected, over a wide range of concentrations. The balance
between the nicking enzyme and the DNA polymerase is more
complex.
4506 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0730811100
To investigate this relationship, we examined the dependence
of the reaction yield (12-mer product) on the amounts of the two
enzymes. It is clear that the reaction is completely dependent on
the presence of both enzymes, the template, and the primer
oligonucleotide (data not shown), but the yield is a complex
function of the amounts of both enzymes. What we find is that
for small amounts of NE, there is a broad range of low reaction
yields. At higher NE concentrations, there is a sharp maximum
as a function of polymerase concentration. In addition, it is clear
from the data (published as supporting information on the
PNAS web site, www.pnas.org) that we can modulate the yield of
partial products by changing the ratio of the enzymes. Although
we do not know precisely how the enzymes interact, cooperate,
or compete with one another, it is clear that there are optimal
concentration ranges of both enzymes. In amplifying an oligonucleotide, we see that the extension leads to some partial
products (see supporting information on the PNAS web site)
identified by their masses to be the result of incomplete elongation of the primer to the full length of the template. The
reaction favors 12 mers as partial products for reasons not well
understood but may have to do with the structural details of the
DNA–polymerase complex for this distributive polymerase (7).
Tuning the reaction conditions via the enzyme concentrations
thus appears to be important for maximizing the yield of any
particular product.
Exponential Amplification. We have devised a simple way to use
the above-described linear amplification to create an exponential
amplification reaction. It has several variants that can be adapted
for different uses. The key idea is to arrange it so that the
oligonucleotide product of the linear reaction serves to create a
new primer that in turn anneals to a target template and creates
a new primer template, which in turn produces more of the same
oligonucleotide product, creating a chain reaction. Our simplest
scheme for doing this is depicted in Fig. 2a. The scheme depends
on our observation that even though the product oligonucleotide
is unstable as a duplex, it will form a transient duplex molecule
with its complement, and this transient duplex can act as a primer
Van Ness et al.
Van Ness et al.
Fig. 3. Triggering mechanisms for the EXPAR chain reaction. Schematic
representation of a mechanism for producing the initial oligonucleotides
from naturally occurring nicking sites in targeted DNA. The trigger template
(green) is made up of sequences matching the target DNA shown in yellow (s0,
s1, the nicking site and the 4-base spacer). A tilde over a sequence symbol
indicates its complement.
as described above with the addition of a ‘‘double-strandspecific’’ dye, SYBR-green, to the reaction (see Materials and
Methods), and the reaction was carried out on a temperaturecontrolled fluorescence reader with which measurements of
fluorescence were made at regular time intervals. The fluorescence in this case is generated during the amplification reaction,
not by the presence of the amplified oligonucleotide itself, but
rather from the double-stranded primer templates produced
during the reaction ( in Fig. 2a). The results of this experiment
are shown in Fig. 2c. MS measurements show that the amplification in this case was ⬇106- to 107-fold.
Triggering Mechanisms. To initiate the exponential reaction, we
need to produce from the sample the first few oligonucleotide
molecules to form the first primer templates that then will start
generating the amplified signal oligonucleotide. The initial oligonucleotide products must be accurate representations of the
sequence to be amplified. The mechanism by which these first
few oligonucleotides are produced is called the triggering mechanism for EXPAR. There are several ways to do this, each of
which requires that we provide a 3⬘OH group-terminated strand
of DNA that can anneal with a complementary template to form
a primer template with the proper configuration and sequence.
One of these triggering schemes, and probably the simplest,
relies on the natural occurrence of nicking enzyme recognition
sites (5⬘-GAGTC-3⬘) in the DNA of interest. For example, as is
shown in Fig. 3, a linear amplifier of a genomic sequence can be
created by providing an oligonucleotide (trigger template, shown
as the green line) that is complementary to the genomic DNA
flanking a specific nicking site. When the trigger template is
annealed to the genomic DNA, this creates a structure that the
nicking enzyme can convert to a primer-template structure,
similar to the linear amplification structure shown in Fig. 1. This
structure will then produce oligonucleotide corresponding to the
sequence to the right of the nicking site. This oligonucleotide is
then used as the trigger for a subsequent exponential amplification reaction. Because the nicking enzyme recognition sequence occurs naturally in both bacterial and human DNA at the
expected frequency for a five-base sequence (about 1 in 1,000),
potential trigger sites abound in DNA and provide a wide range
of target sites for triggering reactions. This simple scheme is the
one demonstrated here. Triggering does not depend on occurrence of these sites, but for simplicity they are used in this
demonstration of the triggering reaction.
PNAS 兩 April 15, 2003 兩 vol. 100 兩 no. 8 兩 4507
BIOCHEMISTRY
for extension by the DNA polymerase. Once extension of the
oligonucleotide has occurred, the duplex is stabilized by the
additional complementary duplex section and will not readily
dissociate. Extending the primer thus creates a stable primer
template that will produce oligonucleotide products in a linear
fashion (Fig. 1). To create these new duplexes, we need only
provide a ready supply of complementary oligonucleotides we
call amplification templates. The key feature of these singlestranded oligonucleotides is that they contain two copies in
tandem of the complement of the oligonucleotide product to be
amplified, separated by the complement of the nicking enzyme
recognition site (3⬘-CTCAG-5⬘) and a four-base spacer (on the
5⬘ side). When the transient duplex (formed with the first of
these copies, by hybridization with the complementary oligonucleotide) is extended, a stable new primer template is created.
This primed template will then continue to produce oligonucleotide product via the linear amplification cycle as described
above (nicking after the four-base spacer, dissociating the oligonucleotide, and reelongating the primer) as long as the
enzymes remain active and dNTPs are available. One might ask
what happens when a transient duplex is formed with the second
copy of the complementary sequence (Fig. 1 Right). The thermodynamics of the situation are essentially the same as after the
extended product has been nicked. The key difference is that
there can be no extension to stabilize the duplex by elongating
it, because it provides no primer template structure for the
polymerase, and it rapidly dissociates. If the amplification templates are present at a high concentration (experiments reported
here use 0.01–0.1 M amplification template oligos), we can
rapidly create primer-template structures that will produce
product oligonucleotide at an accelerating rate. In our reactions,
we take the important precaution of blocking the 3⬘ ends of the
template oligonucleotides (with 3⬘ PO4 groups, for all of the
experiments reported here, or tethered amines) to prevent
spurious self-priming by pairs of template molecules. We have
seen no such spontaneous priming in any of our experiments to
date (data not shown). Finally, it is clear that when all of the
template has been converted into primer template, the exponential reaction kinetics must shift to a linear amplification
mode.
To examine the kinetics of amplification, we carried out the
full exponential reaction in the presence of differing initial
amounts of amplifying oligonucleotide ( in Fig. 2a) and measured the amounts of at a number of time points with the mass
spectrometer. We find that the oligonucleotide amplifies approximately exponentially for the first 2 min or so, as shown by
the data points in Fig. 2b. Note that this amplification proceeds
approximately exponentially until the concentration of approaches the concentration of the template pool. After this
point, it proceeds in an approximately linear fashion, as expected. The total amplification of ⬇106 to 107 in shown in Fig.
2. In end-point measurements from the same reaction, for a
range of starting concentrations, we find that the amplification
levels are all in the range of 106.
Although it may be intuitive that there will be an approximately exponential increase in the product oligonucleotide in a
chain reaction that proceeds as described, it is informative to
look carefully at the mass action reaction equations. If we write
out these equations making the simplest assumptions, we can
show that indeed the kinetics of product generation are predicted to be exponential in character, while the template lasts
(see Materials and Methods). Solutions of the mass action
equations using parameters estimated from our experimental
results are shown as heavy lines in Fig. 2b.
Because the EXPAR reaction is rapid and simple, it is
potentially appealing as a ‘‘real-time’’ reaction in which the
amplification is monitored in the reaction volume during the
reaction itself. To test this possibility, we carried out the reaction
To demonstrate triggering from a naturally occurring nick site,
we specifically amplified certain oligonucleotides contained in
cDNA. These results show that the reaction is triggered only
when the trigger oligonucleotide cognate to the specific cDNA
is present (see supporting information on the PNAS web site).
When no trigger oligonucleotide is present, no reaction occurs,
demonstrating the power of the chain reaction to detect the
presence of small amounts of the cDNAs, and the strict dependence on the triggering reaction.
Discussion
The amplification scheme described here appears to have several
major advantages for many research and diagnostic applications.
These include the isothermal conditions required, the relative
speed of the reaction, and the flexibility with which it can be
triggered and elaborated into multiple coupled reactions. We
have shown clearly that the linear amplification reaction can be
turned into a rather simple exponential amplification scheme
(EXPAR). The linear reaction itself is quite distinct from the
strand displacement amplification scheme (8). It depends fundamentally on the transformation of duplex thermal stability
into instability by the cleavage of a phosphodiester bond in the
nicking reaction (Fig. 1). We have also demonstrated this
essential distinction by experiments physically separating the
polymerization step from the nicking and release steps by using
the enzymes separately (data not shown).
It is difficult to be sure which step in the reaction as described
limits the speed of the overall cycle of amplification and therefore the overall rate of amplification. These issues are being
investigated further to optimize the reaction in speed, sensitivity,
and accuracy, and adapting the reaction for a wider range of
applications. It seems likely that the diffusion and annealing of
the product oligonucleotides to the amplification template is a
slow step, judging from the diffusion constants of oligonucleotides in this molecular weight range. At low trigger concentrations, this is expected to have a significant effect on the reaction rate, because we expect that the reaction in this case is
heterogeneous.
In addition to the triggering reaction demonstrated here, we
can also easily construct variant forms of the above triggering
reaction using any technique that creates a discrete 3⬘ end in the
target DNA extendable by the polymerase, by using a restriction
site, for example. This fragment is then annealed to a trigger
template, just as shown in Fig. 5a, except that it contains the
nicking site in the single-stranded region of the oligonucleotide.
Polymerization of this primed template can then create a duplex
nicking site and complete the amplifying structure. The key to
creating a trigger to the exponential reaction is simply to make
a structure strictly dependent on the target DNA that will
linearly amplify a target oligonucleotide, which can be done in
many different ways.
Because the reaction is a true molecular chain reaction, once
the reaction is triggered, it will proceed without change of the
conditions or further stimulus. There is a concern, therefore, that
the reaction may spontaneously or spuriously trigger. We take
the precaution of blocking the 3⬘ ends of all templates present to
prevent them acting as primers through mispairing with each
other and do not see any spurious priming at the concentrations
of template used in these reactions (up to 0.1 M). The full range
of possible amplification levels of triggering DNA sequences is
not yet fully known. We routinely get 106- to 107-fold amplification within a few minutes but have observed amplification
levels as high as 108-fold. The experiments here use MS and
real-time fluorescence measurements to analyze and monitor
the reactions and their products, but it is clear that they are
amenable in practice to simple end-point fluorescence measurements as well.
4508 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0730811100
Fig. 4. Two alternative EXPAR schemes that can be used for different
applications. ‘‘Direct EXPAR’’ is the scheme described in Fig. 3 in shorthand
form, in which the trigger sequence (blue) is exponentially amplified using the
template (red). The ‘‘copy EXPAR’’ scheme consists of two parts. The upper
bracket represents a template with a nicking site in the reverse orientation,
relative to those in Fig. 3 and Left. This template amplifies the complement of
the triggering sequence (including the 5⬘ overhang). The lower bracket represents the exponential amplification of that complement, now containing a
copy of the 5⬘ overhang on its 3⬘ end (described in the text). The bases
represented by the yellow and purple circles in the copy EXPAR section
indicate complementary bases. The base represented by the green circle in
template b indicates another base variant.
There are several variant forms of the exponential amplification scheme shown in Fig. 3. One in particular has been devised
to provide an accurate copy of a polymorphic site that can
subsequently be amplified. This latter scheme, illustrated in Fig.
4 in a shorthand form, is contrasted with the ‘‘direct’’ EXPAR
scheme described above and is called ‘‘copy EXPAR.’’ The
‘‘copy EXPAR’’ scheme (Fig. 4 Right) is slightly more complex
than ‘‘direct EXPAR’’ (Fig. 4 Left) in that there is a second
template whose amplification reaction is driven by the products
from the first template. The first reaction (upper bracket in Fig.
4 Right) is essentially a linear amplification of the oligonucleotide
trigger with the polymorphic base on its 5⬘ end. Because the
nicking sequence is reversed relative to the orientation shown in
Fig. 3, and the template does not include this 5⬘ base, the
replication component of the reaction (see Fig. 1) creates a
primer template with a 3⬘ terminal base in the template that
matches the polymorphic base. The amplification reaction then
produces the complement of the initiating trigger with an
accurate copy of the polymorphism at its 3⬘ end. The second
bracket indicates the exponential amplification of the product of
the first reaction, shown in the same shorthand as for Fig. 4 Left.
The effect of the two reactions as shown is to amplify the
complement of the triggering oligonucleotide (shown in blue).
The scheme enables the creation of a template that carries an
extra base, which can be interrogated for polymorphic variation
by the mass of the resulting amplified oligonucleotide. This is
shown as the yellow (or purple) disk in Fig. 4. There are two (or
more) second templates available, each cognate to a different
extra base. Fig. 4 shows the amplification of a sequence with the
purple variant (triggered by its complement). Thus, the scheme
can be used to detect and measure polymorphisms in the target
DNA.
Variant forms of the amplification reaction attest to the
flexibility of the method. One of the variants, just described, can
be used to amplify and characterize polymorphic sites in genomic
DNA. The ‘‘copy EXPAR’’ reaction in Fig. 4 depends for its
specificity (that is, giving only the appropriate product) on a
Van Ness et al.
currently exploring the tethering of the amplification templates
to solid supports, including slides and microbeads, so that in situ
amplification can be triggered. This flexibility might open up a
large number of new possibilities.
This work was supported in part by the W. M. Keck Foundation, the
Norris Foundation, and Defense Advanced Research Projects Agency
(contract no. MDA972-02-C-0047). We are grateful to Bill Jack of New
England Biolabs for useful discussions about DNA polymerases and for
comments on the manuscript, and to Huimin Kong for unpublished
information on the nicking enzyme N. BstNBI. We are grateful for
insightful comments on the manuscript by Eric Davidson and an
anonymous referee. The technology described in this paper has been
licensed to Ionian Technologies Incorporated, a company in which the
authors have a financial interest.
1. Mullis, K., Faloona, F., Scharf, S., Saiki, R., Horn, G. & Erlich, H. (1986) Cold
Spring Harb. Symp. Quant. Biol. 51 Pt 1, 263–273.
2. Saiki, R. K., Scharf, S. J., Faloona, F., Mullis, K. B., Horn, G. T., Erlich, H. A.
& Arnheim N. (1985) Science 230, 1250–1354.
3. Fire, A. & Xu, S. Q. (1995) Proc. Natl. Acad. Sci. USA 92, 4641–4645.
4. Liu, D., Daubendiek, S. L., Zillman, M. A., Ryan, K. & Kool, E. T. (1996) J. Am.
Chem. Soc. 118, 1587–1594.
5. Lizardi, P. M., Huang, X., Zhu, Z., Bray-Ward, P., Thomas, D. C. & Ward, D. C.
(1998) Nat. Genet. 19, 225–232.
6. Morgan, R. D., Calvet, C., Demeter, M., Agra, R. & Kong, H. (2000) Biol.
Chem. 381, 1123–1125.
7. Kong, H., Kucera, R. B. & Jack, W. E. (1993) J. Biol. Chem. 268, 1965–1967.
8. Walker, G. T., Little, M. C., Nadeau, J. G. & Shank, D. D. (1992) Proc. Natl.
Acad. Sci. USA 89, 392–396.
BIOCHEMISTRY
phenomenon that appears to be specific to the transient annealing and priming process that creates the primer templates. That
process, in which an oligonucleotide transiently anneals to the
template and is extended by the polymerase, is sharply inhibited
by mispairing near the 3⬘ end, much more so than inhibition by
mispairing of a more stable priming duplex (data not shown). It
is sufficiently inhibited that we see no detectable amplification
when we attempt to prime with such a mispaired oligonucleotide
(data not shown).
Many potential variations of the coupled reactions are described here, including the use of one amplifying oligonucleotide
to trigger another amplification reaction. There are a number of
ways in which this coupling can be used. We are currently
investigating several of these variations. In addition, we are
Van Ness et al.
PNAS 兩 April 15, 2003 兩 vol. 100 兩 no. 8 兩 4509
