J. Biochem. Biophys. Methods 63 (2005) 170 – 186
`
`www.elsevier.com/locate/jbbm
`
`Optimization and design of oligonucleotide setup
`for strand displacement amplificationB
`
`Sylvia Ehsesa,*, Jo¨ rg Ackermannb, John S. McCaskillc
`
`aBiomolecular Optical Systems (BioMOS), Fraunhofer-Gesellschaft, Schloss Birlinghoven,
`D-53754 Sankt Augustin, Germany
`bPlasma Analytics Systems, Technologiepark 1, 15236 Frankfurt/O., Germany
`cBiomolecular Information Processing (BioMIP), Ruhr-Universita¨ t Bochum,
`c/o Schloss Birlinghoven, D-53754 Sankt Augustin, Germany
`
`Received 10 March 2005; received in revised form 14 April 2005; accepted 17 April 2005
`
`Abstract
`
`Several advantages of strand displacement amplification (SDA) as an all-purpose DNA
`amplification reaction are due to it
`isothermal mechanism. The major problem of isothermal
`amplification mechanism is the accumulation of non-predictable byproduct especially for longer
`incubation time and low concentrations of initial template DNA. New theoretical strategies to tackle
`the difficulties regarding the specificity of the reaction are experimentally verified. Besides
`improving the reaction conditions, the stringency of primer hybridization can be distinctly improved
`by computer based sequence prediction algorithms based on the thermodynamic stability of DNA
`hybrid a described by the partition function of the hybridization reaction. An alternative SDA
`mechanism, with sequences developed by this means is also investigated.
`D 2005 Elsevier B.V. All rights reserved.
`
`Keywords: Strand displacement amplification; Folding; Partition function; Hybridization
`
`B This work was performed at Biomolecular Information Processing, BioMIP, Fraunhofer Gesellschaft, Schloss
`Birlinghoven, D-53754 Sankt Augustin, Germany.
`* Corresponding author.
`E-mail address: sylvia.ehses@biomos.fraunhofer.de (S. Ehses).
`
`0165-022X/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
`doi:10.1016/j.jbbm.2005.04.005
`
`1
`
`ELIX 1002
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`171
`
`1. Introduction
`
`Strand displacement amplification (SDA) is based on the primer-directed nicking
`activity of a restriction enzyme and an exonuclease-deficient polymerase which is capable
`of initiating synthesis at a nick and displacing the downstream strand. The method exploits
`the ability of several restriction enzymes to create a nick on one DNA strand within a
`hemiphosphothioate form of its recognition site. Because of repeated nicking, strand
`displacement and priming of displaced strands, DNA will be exponentially amplified (see
`Fig. 3b). Since the development of the strand displacement amplification method by
`Walker et. al [1], many enhancements, particularly the use of thermophilic enzymes which
`allows the reaction to take place at incubation temperatures of 50–60 8C [2], have
`decreased non-specific background amplification and thereby turned the method into a
`sensitive and universal tool for amplification of nucleic acids.
`In addition, isothermal amplifications systems like SDA, a general system which allows
`the amplification of a chosen target by the flanking primers, become increasingly
`important with the widespread use of microstructured systems, where thermal convection
`and expansion has to be circumvented. Another potential advantage of SDA is the
`heightened production of single-stranded DNA by asymmetric SDA,
`i.e. using two
`different primer concentrations that is important especially with respect to the use as a
`fluorescent probe for microarrays, single-nucleotide polymorphism, etc.
`Several techniques to analyze the products by end-point detection using sandwich
`based assays [3] or primer extension [2] and real-time detection by fluorescence
`polarization [4] or fluorescence energy transfer [5] are described in the literature. These
`techniques allow the specific detection of a known and distinct DNA subsequence even
`if only few copies are present, and signal can be further amplified by enzymatic
`reaction. This makes SDA valuable in diagnostics [6]. However, byproducts and
`exhaustion of resources due to the synthesis of byproducts were not necessarily detected
`by these detection methods. A variety of rules for primer design in PCR are described in
`detail
`in the literature, see for example Ref. [7], and a wide choice of tools for
`oligonucleotide design in view of optimizing probe-target binding is publicly available.
`The applications range from primer search and analysis for PCR [8–11]
`to the
`calculation of probes for the development of large-scale projects in microarray analysis
`[12–16]. Most of these tools are based on several general criteria for oligonucleotide
`design like melting temperature, oligonucleotide length, product length and GC content,
`although some include secondary structure prediction by energy minimization using
`nearest neighbor energy parameters.
`Since isothermal amplification systems are more susceptible to the occurrence and
`accumulation of non-specific products, none of these tools fulfill the needs of primer
`design satisfactorily for strand displacement amplification. In contrast to polymerase chain
`reaction, the temperature induced steps and thereby synchronization are missing. This non-
`specificity presents a problem, especially in its application to long amplification reactions
`like in vitro evolution [17,18], where the product is unknown and must be analyzed
`afterwards. As a result special care has to be taken in the optimization process of reaction
`conditions and oligonucleotide design. For this reason we searched for a way to improve
`the design strategy using the partition function method to calculate the base pairing
`
`2
`
`

`

`172
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`probabilities in thermodynamic equilibrium [19] and defining a criterion for elongation
`after binding by the initial primer and target sequences as well as all new build products.
`This allows the design of DNA oligonucleotides which are specific to their respective
`targets.
`
`2. Materials and methods
`
`2.1. Design of an oligonucleotide set
`
`All calculations are based on the algorithms of the Vienna Folding Package (http://
`www.tbi.univie.ac.at/ivo/RNA/[20]).
`For computation of the oligonucleotide design of standard SDA and nicking SDA
`(Table 1) the primer and template sequences were assembled from 6 DNA-words with 16
`bases each [21]. The template is composed of 4 words, 2 words in the middle flanked by
`two primer binding sites. The primers in turn are made up of the recognition sequence of
`the restriction enzyme and 1 word forming the template binding sequence.
`The discrimination of specific and nonspecific hybridization is based on the
`thermodynamic stability of the DNA hybrids formed, whereby the thermodynamic
`stability is described by the free energy of the hybridization reaction. The calculation of
`the probability distribution of alternative DNA/DNA duplex structures was done by
`computation of the partition function [19] and the free energy for the ensemble of
`structures for a sequence, which consists of the two binding partners connected via a
`spacer of 15 artificial nucleotides which are defined to have no binding properties. The
`length of the words was set to 16 nt as mentioned above, the GC content to 50% and the
`melting temperature to 55 8C calculated by nearest neighbor thermodynamics [22,23],
`where Tm ¼ DH 0
`; with the total nucleic acid concentration C.
`DS0þRlnCT
`
`4
`
`System
`
`Description
`
`Standard SDA
`
`Template tSDA
`
`Table 1
`Primer and target sequences (recognition site in upper case and bold types)
`Sequence (5VY 3V)
`cgt tca tct cag tag caa gga cgt acc att ggg cgc aat ttg gta acc
`aca ctg tgc tga tct c
`cga ttc cgc tcc aga ctt CTC GGG cgt tca tct cag tag c
`acc gca tcg aat gca tgt CTC GGG gag atc agc aca gtg t
`gcg caa ttt ggt aac c
`is equivalent to tSDA
`
`Nicking SDA
`
`Minimal SDA
`
`Primer P(SDA)1V
`Primer P(SDA)2R
`Internal probe (R6G-5/6)
`Template
`tSDAnick
`Primer P(SDA)7V
`Primer P(SDA)8R
`Template SDAmin
`Primer SDAminR
`Primer SDAmin5V
`
`cga ttc cgc tcc aga ctt GAG TCa aaa cgt tca tct cag tag c
`acc gca tcg aat gca tgt GAG TCa aaa gag atc agc aca gtg t
`tgc act ctg gaa ttt taa agg gaa cac tgg
`ctc gat cat ctc acc CTC GGG cca gtg ttc cct tta
`agg act gac gca taa CTC GGG tgc act ctg gaa ttt
`
`In standard SDA,the restriction enzyme BsoBI, in the nicking SDA the nicking endonuclease N.BstNBI is used
`(see Materials and methods). Starting from the oligonucleotide design of the standard SDA and nicking SDA, an
`improved design strategy is evolved and tested, resulting in the minimal SDA system (see Fig. 1).
`
`3
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`173
`
`For details on this design strategy we refer to [24]. To improve the design strategy, the
`algorithms used for computation of the minimal SDA system (Table 1) were adapted to the
`SDA reaction and the calculated oligonucleotides were tested by simulation of the
`mechanism. We took into consideration secondary structures as well as the elongation of
`the 3V end because of mis-priming. In this calculation, even less probable intermediate
`products were regarded.
`The test system contains only the essential components for SDA to take place. That
`means, the primer consists of a 5V overlapping end following the recognition site of the
`restriction enzyme and a template-binding region on its 3V end. The template only consists
`of the primer binding target sequence, so there is no sequence inside the two primer
`flanking regions (Fig. 1). The system is also built up of DNA building blocks, each 15 nt
`long with a melting temperature from 50 to 60 8C and a GC content from 40% to 60%.
`To check possible and non-specific binding of a free 3V end of any product, we
`estimated the elongation probability. The elongation probability represents the maximum
`product of the last base pairing probability and the three base pairing probability, which is
`
`Fig. 1. Setup and mechanism of strand displacement amplification. (a) Oligonucleotide setup. All primers (see
`Table 1) consist of a 5V overlapping sequence, the recognition site for the restriction enzyme located 5V to the
`target binding region and a template/target binding region at their 3V end. The two primer bind to opposite strands
`of the target sequence. For the minimal SDA system, they bind without flanking the variable region inside. (b)
`Underlying mechanism of exponential amplification in SDA. Besides amplification by repeated elongation,
`nicking and strand displacement, the displaced strand can also serve as a template for the opposing strand.
`
`4
`
`

`

`174
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`allowed. The three base pairing probabilities means the sum of the base pairing
`probabilities of any base of a DNA sequence at position N to the last base and at position
`N + 1 to the last but one base and at position N + 2 to the last but two base at the 3V end.
`These probabilities are calculated either for one or two sequences linked by a spacer of
`15 artificial nucleotides with no binding properties, to take inter- as well as intramolecular
`binding into account. Every combination is tested: that means every primer, template and
`intermediate product with itself and with any other sequence from the reaction (see Fig. 2).
`As long as this maximum product is greater than the elongation probability one
`randomly selected base will be exchanged by another base chosen randomly as well (point
`mutation). Fixing the elongation probability is therefore a criterion for the specificity of the
`computed oligonucleotides.
`
`2.1.1. Materials
`BsoBI (10 U/Al), N.BstNBI (10 U/Al), Bst DNA Polymerase Large Fragment (exo-
`Bst, 8 U/Al)were purchased from New England Biolabs. Denaturing ladder was a 10
`
`Fig. 2. Partition function as a tool for the search of non-specific hybridization reaction.
`
`5
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`175
`
`base pair DNA ladder (Invitrogen). Ultra-pure human placental DNA was purchased
`from Sigma and diluted in sterile water. Oligonucleotides (Table 1) were synthesized
`and purified by IBA and MWG. Deoxynuleoside 5V-triphosphates (dGTP, dTTP, dATP,
`dCTP) were purchased from QBiogene, 2V-deoxycytosine 5V-O-(1-thiotriphosphates)
`(dCTPaS) from IBA and TriLinkBiotech. TO-PRO-1, SYBR Green II and SYBR Gold
`were from Molecular Probes, denaturing stop/loading dye from Epicentre.
`
`2.1.2. Standard SDA
`SDA was performed in a 30 Al volume with final concentrations of 35 mM KiPO4 (pH
`7.6), 8 mM magnesium acetate, 0.4 mM each dATP, dGTP, dTTP, 0.8 mM dCTPaS, 0.1
`mg/ml BSA,0.5 AM of each primer, either 1 AM TO-PRO-1 or 1 : 5 SYBR Gold, 0.24 U/
`Al exo-Bst ,1.7 U/Al BsoBI), 10% (v/v) glycerin. After addition of template DNA into final
`volume of 24 Al and before addition of any enzymes, the reaction sample was incubated
`for 3 min at 95 8C, followed by 1 min at 55 8C. Upon addition of the enzymes, the
`amplification mixture was incubated 15–60 min in an ICycler (BioRAD) and the increase
`in fluorescence intensity was monitored. The reaction was stopped by addition of
`denaturing Stop/Loading dye and products were denatured at 95 8C for 10 min. Aliquots
`of 10 Al from each sample were added to 5 Al of the loading dye and analyzed by 12%
`denaturing polyacrylamide gel electrophoresis, with subsequent staining via SYBR Gold
`or SYBR Green II and visualization under UV illumination.
`The partition function calculates the probability distribution of alternative DNA/DNA
`duplex structure. Above the diagonal, the dot matrix represents base pairing probabilities
`calculated by the partition function, below, the minimal free energy by the hybridization
`of sequence P(SDA)8R to itself. The loop consisting of 15 artificial nucleotide (n) is
`defined to have no binding properties and do not contribute to the entropy term.
`P(SDA)8Rs corresponds to the target-binding part of the primer P(SDA)8R (see Fig. 1).
`The rectangle contains the part of the sequence used to calculate the probability that the
`3V end of sequence P(SDA)8Rs can be elongated. The product of the binding probability
`of the last base and the sum of the last three bases (diagonal) to any other base is called
`the 3V end,
`the elongation probability. By analysis of the binding probabilities at
`structures can be found, which provide an indication of problems in primer design. A
`possible hybridization reaction resulting in the production of non-specific amplification
`products (see Fig. 4) is shown in the lower part of the plot, although calculation of the
`minimal free energy shows no secondary structures. The size of the squares corresponds
`to the binding probabilities. The Vienna Folding package (http://www.tbi.univie.ac.at/
`ivo/RNA/) is used to calculate the partition function and the minimal free energy.
`
`2.1.3. Nicking SDA
`Nicking SDA reactions were performed in a 30 Al volume with final concentrations of
`100 mM KCl, 35 mM Tris–HCl (pH 7.5), 10 mM MgCl2, 0.4 mM each dATP, dGTP,
`dTTP and dCTP, 0.1 mg/ml BSA, 0.5 AM of each primer, either 1 AM TO-PRO-1 or 1 : 5
`SYBR Gold, 0.24 U/Al exo-Bst 1.7 U/Al N.BstNBI). After addition of template DNA into
`final volume of 24 Al and before addition of any enzymes, the reaction sample was
`incubated for 3 min at 95 8C, followed by 1 min at 55 8C. Upon addition of the enzymes,
`the amplification mixture was incubated 15–60 min in an ICycler (BioRAD) and the
`
`6
`
`

`

`176
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`increase in fluorescence intensity was monitored. The reaction was stopped by addition of
`denaturing stop/loading dye and products were denatured at 95 8C for 10 min. Analysis
`was performed as described above.
`
`2.1.4. Elongation
`Elongation was performed as with SDA or nicking SDA, depending on the
`oligonucleotides. Only the restriction enzyme was omitted. The oligonucleotides were
`used in a concentration of 1 AM.
`
`2.1.5. Cloning and sequencing
`Cloning was performed using TOPO TA Cloning from Invitrogen. For sequencing,
`SequiTherm EXCEL II (Epicentre) was used. Products were inserted into a pCR2.1
`vector and subsequently transformed into E. coli Top10 chemically competent cells.
`Direct sequencing without previous cloning was performed by elongation of single-
`stranded DNA with an overlapping primer including the reverse complement sequence for
`T7 promoter sequence, a T20 spacer and a template binding region:
`TAATACGACTCACTATAGGG T20 GAGATCAGCACAGTGT.
`Cycle sequencing was performed using M13 forward ( 20)primer for sequencing of
`plasmids or T7 promoter sequencing primer for direct sequencing of reaction products
`labeled with IRD800 (MWG Biotech). Subsequent gel electrophoresis and automated
`detection and analysis were carried out on a LICOR DNA Sequencer 4000 L.
`
`3. Results
`
`3.1. Performance of standard SDA and nicking SDA
`
`In this work, an optimization process for the SDA reaction setup is described to
`enable selective amplification processes. The assembly differs
`from other SDA
`reaction setups by the use of short synthetic DNA oligonucleotides as starting
`template for the amplification reaction, which means that an initial target generation
`process that makes copies of the target sequence flanked by nickable restriction sites
`is missing [25]. At the outset, reaction conditions like temperature, salt and enzyme
`concentration were adapted to the theoretically optimized oligonucleotide setup to
`prevent non-specific binding by hybridization of
`the DNA strands in complex
`structures [24,21].
`The estimation of optimized amplification reaction conditions involved comparison of
`product yield and specificity after denaturing gel electrophoresis. Product analysis is
`accomplished by 5V Rhodamine labeled primers as well as by fluorescence dyes
`intercalating to single- or double stranded nucleic acids. The use of intercalating dyes
`permits a sensitive detection of all constructed products. In fact, in contrast to labeled
`primers we detected non-specific products only by using of these dyes.
`When performing conventional
`thermophilic SDA on templates we detected a
`successful amplification at a temperature range from 50 to 65 8C, with higher yields
`and higher tendency for side-reactions at increased temperature and a BsoBI / exo-Bst ratio
`
`7
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`177
`
`of 1.7 to 0.24 U/Al. Optimization of the salt concentration up to 10 mM shows no
`amplification products when performing SDA with less than 5 mM magnesium acetate and
`an optimum at 8 mM.
`In the alternative mechanism of nicking SDA we introduced a nicking endonuclease.
`This enzyme N.BstNBI is a site specific endonuclease that cleaves only one strand of
`DNA on a double-stranded DNA substrate. The need for a-thiophosphate nucleotides can
`be circumvented by the use of N.BstNBI. This is in contrast to standard thermophilic SDA,
`which uses BsoBI as restriction enzyme and the modified nucleotides dCTPaS to achieve
`nicking of the hemiphosphothioate double stranded DNA. This reaction shows poor yields
`below 6 mM magnesium chloride. Between 6 and 14 mM magnesium ions, no change in
`product yields could be observed. Between 50 and 65 8C the highest yield was detected at
`55 to 60 8C and a N.BstNBI / exo-Bst ratio of 1.7 to 0.24 U/Al. Estimation of the optimized
`enzyme concentration is more difficult to perform because of the interdependency of the
`two enzymes to each other (see Fig. 4 (b)).
`Since the optimization of reaction conditions like temperature, salt and enzyme
`concentration did not yield satisfactory results regarding the suppression of non-specific
`byproducts, we realized the need for further improvements to ensure optimal primer-
`target binding with concomitant exclusion of other hybridization events. For these
`reasons, we looked for stringent constraints for computation of primer and template
`sequence.
`
`3.2. Analysis and reaction mechanism for byproducts
`
`When performing the standard reaction as described above, the expected products were
`achieved with amplification starting with template concentrations of 10 and 100 pM in 20
`min for SDA and nicking SDA respectively, with product analysis on a SYBR Green II or
`SYBR Gold stained denaturing polyacrylamide gel (Fig. 3a and b). When starting with
`less template, omitting the initial denaturation step or increasing the reaction time, the
`standard SDA system as well as the nicking system shows the tendency to side-reactions.
`Assuming that this is no matter of contamination with foreign DNA molecules, at least two
`side-reactions can be observed (Fig. 3c).
`On one hand, an unspecific primer reaction takes place, which we did not expect under
`the reaction conditions employed and after optimization of the sequences to reduce non-
`specific binding. On the other hand, a DNA ladder is built because of partitial primer
`concatemerization.
`The incubation of the different primer pairs of the nicking SDA system, which is
`equivalent to the standard system except of the recognition site for the endonuclease, with
`the polymerase,
`indeed showed elongation of primers, or single building blocks of
`primers, without the corresponding binding partner in the form of a template molecule
`(Fig. 4a). We observed, that the forward primer P7V as well as the target-binding region of
`the reverse primer P8R (in Fig. 4a called P8Rs), in contrast to the bwholeQ sequence of the
`primer, can prime themselves. In the SDA, this reaction is favored by using less template
`concentrations, probably because the reaction of the primer with the template is
`energetically preferred. Furthermore,
`the initial denaturation step before addition of
`enzymes can delay the formation of the dimer. Obviously, non-specific hybridization of
`
`8
`
`

`

`178
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`the primer provokes elongation of the 3V end. This may arise e.g. by primer dimerization or
`hairpin formation, although the design of the sequences is intended to prevent such
`behavior.
`An indication of the amplification of non-specific products is also given by a two step
`kinetic profile. When using an intercalating fluorescence dye TOPRO-1 in real-time
`detection, after about 20 min the fluorescence intensity signal shows a steep increase
`(results not shown). Besides primer dimerization a regular ladder of larger byproducts was
`built. This behavior can also be observed in the elongation reaction without a restriction
`enzyme (see Fig. 4), when both oligonucleotides consisting of the target-binding region of
`the forward resp. reverse primer and template are present. This ladder building is
`intensified when performing longer SDA reactions,
`i.e.
`longer than 45 min. Direct
`sequencing of these products gave a surprising result (Fig. 5). The products were generated
`by successive addition of distinctive sequences, and could be enhanced by increasing
`concentration of the polymerase (see Fig. 4).
`Likewise, it is possible that, besides the polymerase, the restriction enzyme and the
`modified nucleotides in standard SDA could enhance this effect. It is known that the
`employed restriction enzyme BsoBI shows star activity at elevated temperature. So we
`tested the enzyme AvaI in the SDA reaction, which has the same recognition sequence as
`BsoBI but shows no significant star activity at elevated temperature. Side-products could
`not be suppressed. Furthermore the rate of yield was lower because of the reaction
`temperature that was farther away from the optimum at 37 8C, so activity and stability are
`reduced. In addition strand termination by a-substituted nucleotides and E. coli DNA
`Polymerase I Large Fragment has been described [26] and this may be true for Bst DNA
`Polymerase Large Fragment too. However, because nicking SDA shows the same behavior
`regarding the side-products and this system needs no modified nucleotides, strand
`termination should be a subsidiary reason at most. Although the effect relies on the
`concatemerization of the monomers building multimers by self-ligation,
`the exact
`mechanism is not clear although the polymerase or polymerase activity clearly plays a
`leading role. Moreover, a comparable reaction for amplification of nucleic acids has been
`patented [27]. Amplification and multimerization circular and linear padlock probes and
`thereby the building of multimers by Bst polymerase is known and used, i.a. in the so-
`called linear target isothermal multimerization and amplification (LIMA)and cascade
`rolling-circle amplification (CRCA) [28], though it gives no account for the observation
`made here.
`
`Fig. 3. Analysis of strand displacement amplification (SDA) reaction of variable initial concentration of template
`tSDA after optimization of reaction conditions. In (a) and (b) SDA was performed for 20 min at 55 8C. Initial
`template concentration for each sample are indicated at the bottom. The two product bands correspond to nicked and
`unnicked SDA products. (a) Standard SDA. The strands are nicked by the endonuclease BsoBI and dCTPaS. The
`product length is 93 nt for the full length product resp. 74 nt for the nicked product. (b) Nicking SDA. The strands are
`nicked by the nicking endonuclease N. BstNBI. The product length is 91 nt for the full length product resp. 64 nt for
`the nicked product. (c) Comparative product analysis after 90 min reaction time at 55 8C with 10 pM initial target
`concentration (+ T) and without target ( T) performing standard SDA and nicking SDA. Because of the increased
`side-reactions, the desired product cannot be detected by subsequent staining with SYBR Gold. A 10 bp ladder (bp)
`was included for reference. Denaturing gel electrophoresis (12% acrylamide) (see Material and methods).
`
`9
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`179
`
`To solve the problems regarding unintended byproducts, a new oligonucleotide set of
`primers and template with improved features regarding the mechanism of the SDA
`reaction was calculated.
`
`10
`
`

`

`180
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`3.3. Optimized template-primer design with minimal byproducts
`
`The optimization process is based on the same algorithms [20]used for the first
`design of standard SDA and nicking SDA. The calculated primers and template present
`a minimal system because there is no sequence between the two flanking primers (Fig.
`
`11
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`181
`
`Fig. 5. Sequenced ladder building product (see Fig. 4 (b)). Direct sequencing by T7 promoter primer (see
`Materials and methods) of products of different length (different steps) after nicking SDA (see Fig. 4 (b)). The
`products are isolated by gel elution after denaturing polyacrylamide gel electrophoresis. The primer binding part
`of the template emerge in multiple copies (bold type) and thereby cause the observed ladder building manner of
`elongation and SDA reaction. The underlying mechanism remains unclear (see text).
`
`1). The calculation takes into account the successive steps and hence generated products
`like restriction after elongation. This means that all
`intermediate products and their
`reaction with other fragments were considered. To decide if a sequence could act as
`template for another sequence, or itself, a threshold value called elongation probability
`was introduced. Two sequences were considered to prime each other when the product
`of the last base and the sum of the last three nucleotides at 3V end of a DNA fragment of
`the probability matrix calculated by the partition function [19] is greater than this
`elongation probability (for details see Material and methods). If this results in a non-
`desired product, one sequence is changed by a randomly substituted base (point mutation).
`The substitutions are repeated as long as there is no more binding of the 3V end possible. In
`a last step, the sequences were checked again by simulation of the reaction.
`In the simulation procedure the calculated oligonucleotide sequences were checked if
`they are able to prime themselves or each other at the chosen elongation probability. If they
`hybridize to any 3V end, new sequences will be calculated by elongation of the
`oligonucleotide. The old as well as the new sequences are cut if they carry a restriction site
`which is not protected (by means of the modified oligonucleotides), and new sequences
`may then be produced. The computation is continued as long as new sequences can be
`found. No mis-priming should occur at the chosen elongation probability for the optimized
`sequences.
`The next step is the evaluation of the developed algorithm. When the computed
`sequences showed no tendency for side-reactions theoretically,
`the corresponding
`oligonucleotides were used in the experiment and products were analyzed by gel
`electrophoresis. No tendency for side-reactions of the primers in the simulation is achieved
`when the elongation probability is fixed to 0.001 and the maximum free energy of the
`primers is set to 2.5 kcal/mol at 37 8C. When testing the calculated DNA oligonucleotide
`design in the SDA reaction experimentally, we actually also found a stable system,
`
`Fig. 4. Analysis of mechanism of forming byproducts. (a) Elongation. Incubation of different primer resp. primer
`pairs on SDA condition with exo-Bst polymerase but no restriction enzyme. Reaction stopped after 1 h. Each pair
`was incubated without (first
`trace) and with (second trace) template DNA. P(SDA)7Vs and P(SDA)8Rs
`correspond to the 3V
`target binding part of primer P(SDA)7V resp. P(SDA)8R (see Fig. 1). Elongated
`oligonucleotides are marked by an arrow. P(SDA)7V is elongated unexpectedly, as well as P(SDA)8Rs. A
`hypothetical explanation for elongation of P(SDA)8Rs is described in Fig. 2. 10 bp ladder (bp) is indicated.
`Denaturing gel electrophoresis (12% acrylamide), staining with SYBR Gold. (b) SDA reaction. Occurrence of
`byproduct in nicking SDA subject to enzyme activity. Incubation of reaction mixtures in nicking SDA with
`different amount of nicking restriction enzyme and polymerase. Reaction topped after 30 min. With increasing
`enzyme concentration the formation of a regular ladder of byproducts become more pronounced. Non-denaturing
`gel electrophoresis is (12% acrylamide) (see Material and methods).
`
`12
`
`

`

`182
`
`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`Fig. 6. Minimal SDA. Increasing the specificity of strand displacement amplification of target tSDAmin by
`calculation of an optimized template-primer design. SDA was performed for 1 h at 55 8C and the product are
`visualized by subsequent staining with SYBR Gold. Initial template concentration for each sample are indicated at
`the bottom. The two product bands correspond to nicked and unnicked SDA product. A 10 bp ladder (bp) is
`included for reference. Denaturing gel electrophoresis (12% acrylamide) (see Material and methods).
`
`showing no sideproducts after gel electrophoresis (Fig. 6) even after incubation longer
`than 1 h. So the algorithm works satisfactory and this stringent test confirms that we
`developed a well-suited tool for SDA oligonucleotide design.
`
`4. Discussion
`
`There have been many efforts that were made to develop algorithms for improved
`oligonucleotide design, e.g. primer design in PCR-based detection methods [10,11,9]and
`development of probes in microarray analysis [16,15,14]. Most of the offered tools are
`based on simple criteria or rules [7]as well as thermodynamic calculations derived from
`secondary structure prediction [20,29]. However, hitherto it has remained difficult to
`choose good primers for the use in isothermal amplification reactions.
`
`13
`
`

`

`S. Ehses et al. / J. Biochem. Biophys. Methods 63 (2005) 170–186
`
`183
`
`We found that for strand displacement amplification it is not satisfactory to calculate
`the minimum free energy secondary structure. Self-homology and non-specific binding
`the 3V end are underestimated in the minimum free energy picture.
`especially at
`Moreover, including products like the nicked and displaced strand after nicking turned
`out to be essential for the primer design of the isothermal SDA. So we looked for a
`criterion to calculate oligonucleotide sequences which allows
`for
`stable strand
`displacement amplification. In contrast to PCR, the possibilities in optimization with
`regard to the hybridization temperature, i.e. the annealing temperature of the primers,
`are restricted to a relative small temperature range, in which the enzymes work with
`high activity. In addition,
`the high magnesium salt concentration, used in SDA to
`ensure an optimal product yield, makes the primer calculation more difficult. High
`Mg2+ concentrations can promote incorrect annealing of the primer to its intended
`binding site. We found that below 5 mM Mg2+ no product is detectable but the high