Comparison of SYTO9 and SYBR Green I for Real-Time Polymerase

Analytical Biochemistry 340 (2005) 24–34 www.elsevier.com/locate/yabio

Comparison of SYTO9 and SYBR Green I for real-time polymerase chain reaction and investigation of the eVect of dye concentration on ampliWcation and DNA melting curve analysis

Paul T. Monis ¤, Steven Giglio, Christopher P. Saint

Microbiology Unit, Australian Water Quality Centre, Private Mail Bag 3, Salisbury, SA 5108, Australia

Received 28 October 2004

Abstract

Following the initial report of the use of SYBR Green I for real-time polymerase chain reaction (PCR) in 1997, little attention has been given to the development of alternative intercalating dyes for this application. This is surprising considering the reported limita- tions of SYBR Green I, which include limited dye stability, dye-dependent PCR inhibition, and selective detection of amplicons during DNA melting curve analysis of multiplex PCRs. We have tested an alternative to SYBR Green I and report the Wrst detailed evaluation of the intercalating dye SYTO9. Our Wndings demonstrate that SYTO9 produces highly reproducible DNA melting curves over a broader range of dye concentrations than does SYBR Green I, is far less inhibitory to PCR than SYBR Green I, and does not appear to selectively detect particular amplicons. The low inhibition and high melting curve reproducibility of SYTO9 means that it can be readily incorporated into a conventional PCR at a broad range of concentrations, allowing closed tube analysis by DNA melting curve analysis. These features simplify the use of intercalating dyes in real-time PCR and the improved reproducibility of DNA melting curve analysis will make SYTO9 useful in a diagnostic context. Crown copyright  2005 Published by Elsevier Inc. All rights reserved.

Keywords: PCR; DNA melting curve analysis; Intercalating dye

Real-time PCR is rapidly becoming a standard of dyes, including YO-PRO-1 [8], SYBR Green I [9,10], method in many diagnostic and research laboratories. BEBO [11], and LC Green [12], have been evaluated for The market for this technology continues to expand and use in real-time PCR applications. While probe-based there have been many probe-based systems developed to detection methods continue to be developed and meet this demand, including Taqman probes [1], molecu- improved, progress for intercalating dyes has been rela- lar beacons [2], FRET1 probes [3], Scorpions [4], and tively slow. iFRET [5]. An alternative to probe-based detection is the SYBR Green I has become the preferred choice for use of Xuorescent double-stranded DNA (dsDNA)-spe- real-time PCR because it is cost eVective compared to ciWc intercalating dyes. Such a dye, ethidium bromide, probe-based systems, allows the generic detection of was used in the Wrst description of real-time PCR by ampliWed DNA, and can be used to diVerentiate DNA Higuchi et al. [6,7]. Since then, a relatively small number by DNA melting curve analysis [9,10]. However, SYBR Green I also has disadvantages that limit its ease of use. V ¤ It cannot generally be used in standard PCR bu ers Corresponding author. Fax: +61 8 8259 0228. without further optimization of conditions and the E-mail address: [email protected] (P.T. Monis). 1 Abbreviations used: FRET, Xuorescence resonance energy transfer; inclusion of additional reagents to improve reaction DMSO, dimethyl sulfoxide. eYciency, such as DMSO [13], bovine serum albumin,

0003-2697/$ - see front matter. Crown copyright  2005 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.01.046 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 25 and Triton X-100 [14]. Depending on the reaction condi- unlike SYBR Green I, SYTO9 readily allows the detec- tions, the dye also appears to be inhibitory to PCR in a tion of multiple amplicons in a multiplex PCR. The util- concentration-dependent manner [9,15] which can be ity of SYTO9 is demonstrated using both prokaryote overcome by increasing the concentration of MgCl2 in and eukaryote systems. the reaction [14,15]. The degradation products of the dye have also been reported to be inhibitory to PCR [14]. DNA melting curve analysis using SYBR Green I is Materials and methods widely applied for the identiWcation of a variety of organisms (e.g., Leptonema [16], Borellia [17,18], Legion- Source of isolates and DNA template preparation ella [19], and Cryptosporidium [20]). However, the melting temperature (Tm) obtained using this dye can be The source of V. cholerae O1 El Tor biotype and L. aVected by the concentration of dye [10] and the concen- pneumophila serotype 1 have been described previously tration of DNA [21]. For example, in the study of Ririe by Giglio et al. [22]. DNA was extracted from cultures et al. [10] a 3 log-fold diVerence in the number of starting using Qiagen DNA-mini spin columns following the copies of target DNA used in real-time PCR resulted in manufacturer’s methods. The origins and characteriza- V a Tm di erence of more than 1 °C for the same amplicon. tion of Giardia duodenalis isolates Ad-1, Ad-19, and Ad- A further limitation of SYBR Green I is that it 23, representative of Assemblages A, B, and F, respec- appears to have limited application for the analysis of tively, and of a Giardia ardeae isolate have been multiplex PCR. In a study by Giglio et al. [22], multiplex described previously [24]. DNA samples from barley PCRs for Vibrio cholerae and Legionella pneumophila (Horedum vulgare) varieties Atlas, Atlas68, Proctor, and were analyzed by DNA melting curve analysis using Shannon were kindly provided by Dr. Brendon King. SYBR Green I. It was found that only one amplicon DNA concentrations for puriWed extracts were deter- could be detected by melting curve analysis but that mined by measuring UV absorbance at 260 nm. both amplicons were being ampliWed as determined by agarose gel electrophoresis. This eVect appeared to be Real-time PCR and melting curve analysis of prokaryotic related to the relative concentrations of the ampliWed DNA DNA and dye because time-based analysis demonstrated that both amplicons could be detected by melting Real-time reactions were conducted in 100-l thin- curve analysis after 20 cycles of PCR but only one walled tubes and monitored using a Rotor Gene 3000 amplicon could be detected after 40 cycles of PCR. (Corbett Research, Sydney, Australia). The multiplex The SYTO family of dyes have many properties that PCRs for V. cholerae and L. pneumophila were per- make them useful for labeling live cells, including good formed as described previously [20], except that the V. membrane permeability, low cytotoxicity, and enhanced cholerae multiplex assay was modiWed to include DMSO Xuorescence in the presence of DNA (US patent (5% v/v Wnal concentration), primer H3F was excluded, 5,436,134 [23]). Although they are ostensibly DNA-bind- and the primer concentrations of the remaining four ing dyes, their low toxicity suggests that they may not primers were reduced to 0.5 M each (Wnal concentra- interfere with cellular processes, such as DNA and RNA tion). The primer set 27f(5Ј-AGAGTTTGATYMTGG synthesis, at the concentrations that are used for staining CTCAG-3Ј) and 1492r (5Ј-TACGGYTACCTTGTT cells. Little information with regard to their speciWcity ACGACT-3Ј), modiWed from Suzuki and Giovannoni for dsDNA versus single-stranded DNA (ssDNA) or [25], was used to amplify a 1.46-kb fragment of the 16S their suitability as dyes for real-time PCR is available. rRNA gene from V. cholerae DNA. Each 20-l reaction Evaluation of the mode of action of the BacLight Live/ mixture contained 200 M each deoxynucleoside tri- Dead kit (Molecular Probes) in our laboratory suggested phosphate, 0.5 M each primer, 2.5 mM MgCl2, 1£ PCR that SYTO9 is a dsDNA-speciWc dye. These factors, buVer II (Applied Biosystems Foster City, CA, USA), together with the similarity in excitation/emission max- 5% v/v DMSO, 1 U of AmpliTaq Gold DNA polymerase ima in SYTO9 and the FAM channel of the Rotor Gene (Applied Biosystems), and 5 l of genomic DNA. Ther- 3000, have led us to select SYTO9 for evaluation in real- mal cycling consisted of an initial denaturation at 95 °C time PCR. for 10 min followed by 50 cycles of denaturation at 94 °C We report herein the novel use of SYTO9 in real-time for 30 s, annealing at 50 °C for 1 min, and extension at PCR. This dye supports PCR ampliWcation over a wide 72 °C for 2 min. range of dye concentrations and produces robust DNA Dye concentrations of SYTO9 (Molecular Probes, melting curves that are not aVected by DNA concentra- OR) or SYBR Green I used in reactions ranged from 10 tion. In addition, we investigated the inhibitory eVect of to 0.5 M. The molar concentration of SYBR Green I dye concentration on PCR and the eVect of dye concen- was calculated using the values determined by Zipper et tration on DNA melting curves. These data extend upon al. [26], who estimated that a 1£ SYBR Green 1 solution the work of Giglio et al. [22] and demonstrate that, represents a dye concentration of 2 M. Fluorescence 26 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 data were acquired on the FAM channel (excitation at SYTO9. DNA melting curve analysis was monitored 470 nm, detection at 510 nm) at the end of each 72 °C from 70 to 95 °C in 1 °C increments using a 30-s hold at extension step of the PCR. The FAM channel closely each step on the FAM channel. AmpliWcation products matches the excitation and emission maxima for SYTO9 were characterized by gel electrophoresis (as described (485 and 498 nm, respectively, according to the manufac- above) to conWrm ampliWcation of the expected sized turer’s speciWcations). Fluorescence was measured over a fragment. range of gain settings to accommodate the detection of A 660-bp fragment of the gene encoding glutamate diVerent dye concentrations within a single experiment. dehydrogenase was ampliWed from Giardia DNA using This was achieved using a feature of the RotorGene primers GDHGENF (5Ј CGYGTNCCNTGGNTGG 3000 that allows the creation of multiple FAM channels, AYGAYGCYGG 3Ј) and GDHGENR (5Ј AGYTT each with diVerent gains ranging from 10 to ¡1. CTCCTCGKTGAASCC 3Ј) and the following condi- Following ampliWcation, melting curves were tions: 95 °C for 5 min and 55 cycles of 94 °C for 30 s, acquired on the FAM channel using 1 °C steps with a 55 °C for 30 s, and 72 °C for 60 s. The reaction mixture hold of 60 s at each step from 70 to 99 °C. The diVerenti- (20 l) contained 2 l of DNA extract, 1£ PCR buVer ated data were analyzed using the Rotor Gene software (20 mM Tris–HCl (pH 8.4), 50 mM KCl), 1.25 mM W (v6.0 build 14) with the digital lter set as “none.” When MgCl2, 200 M each deoxynucleoside, 1 U of Platinum required, samples were analyzed by 1% agarose gel elec- Taq DNA polymerase (Invitrogen Life Technologies, trophoresis with the addition of Gelstar nucleic acid Carlsbad, CA USA), 0.25 M each primer, 5% DMSO, stain (Cambrex Bio Science, Rockland, IN) using stan- and 3.3 M SYTO9. Data were acquired at 72 °C on the dard methods [27]. FAM channel. DNA melting curve analysis (using version 5.0 build 60 of the Rotor-Gene software) was moni- Determination of dye stability and between-run tored from 60 to 99 °C in 1 °C increments using a 10-s reproducibility hold at each step on the FAM channel. AmpliWcation products were characterized by gel electrophoresis (as Master mixes were prepared for the three bacterial described above) to conWrm ampliWcation of the PCR assays described above and 20-l aliquots were dis- expected sized fragment. pensed into 100-l tubes using a CAS-1200 liquid han- dling system (Corbett Research). Each 20-l aliquot Statistical analysis contained 25,000 or 25 pg of DNA and either 2 or 0.5 M dye. Aliquots were stored in the dark at ¡20 °C Statistical analyses (two-sample t test for independent until used. Real-time PCR was conducted on a Rotor- samples) were conducted using Origin version 7 (Origin- Gene 3000 (Corbett Research) as described above. A Lab, MA, USA). threshold value of 0.02 was used to determine the Ct (cycle that crosses the threshold) value for each reaction. For the multiplex reactions, the Ct values represent the Results and discussion combined signal for both amplicons. This information cannot be used for quantitation, but it does provide a EVect of dye concentration on DNA melting curves measure of assay reproducibility because it would be expected that changes in the ampliWcation of either frag- The available product information for the SYTO ment will aVect the total Xuorescence and therefore aVect family of dyes states that they are nucleic acid stains and the Ct. provides absorption and emission spectra for each dye in the presence of DNA and RNA (http://www. Real-time PCR and melting curve analysis of eukaryotic probes.com/media/pis/mp07572.pdf). However, no infor- DNA mation on dye speciWcity with regard to dsDNA or ssDNA is provided. To investigate this for SYTO9, The YLM locus was ampliWed from barley DNA DNA melting curve experiments were conducted using using the YLMF (5Ј CAGGAGCTGGTGAAATAGT genomic DNA ampliWed by the Legionella multiplex GCCT 3Ј) and YLMR (5Ј TTAAAGGGCTCCGT PCR in the absence of any dye. A small volume of GAAGC 3Ј) primers of Paltridge et al. [28]. PCRs were SYTO9 was added post-PCR to provide a concentration conducted on a Rotor-Gene 3000 using the following of 3.3 M, which was selected because it is the working conditions: 95 °C for 10 min and 50 cycles of 94 °C for concentration used to stain live cells. The volume of dye 10 s, 58 °C for10 s, and 72 °C for 15 s. The reaction mix- added was kept small to minimize any change to the V ture (25 l) contained 100 ng of genomic DNA, 1£ PCR bu er or MgCl2 concentrations. For comparison, SYBR V bu er II, 1.5 mM MgCl2, 200 M each deoxynucleoside Green I was added to replicate samples at the same triphosphate, 1 U of AmpliTaq Gold DNA polymerase concentration. The detection of distinct melting peaks (Applied Biosystems), 0.5 M each primer, and 3.3 M (Fig. 1) demonstrates that SYTO9 is speciWc for dsDNA Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 27

Fig. 1. DNA melting curve analysis of ampliWed mip and 16S rDNA fragments using 3.3 M either SYTO9 or SYBR Green I that has been added post-PCR. and that it produces results comparable with SYBR Green I, although the Tm values determined by each dye diVered from each other by 2–3 °C. To examine the eVect of dye concentration on melting curves, similar experiments using a broad range of concentrations (0.5–33 M) of either SYTO9 or SYBR Green I and either Legionella multiplex PCR or Vibrio multiplex PCR ampliWcation products were conducted. Comparison of amplicon Tm versus dye concentration for the mip, Legionella 16S rDNA, and hlyA amplicons Fig. 2. Comparison of dye concentration versus Tm for mip, Legionella demonstrated that the Tm increases with increasing dye 16S rDNA, and hlyA amplicons. concentration (Fig. 2), suggesting that the dyes have a stabilizing eVect on the dsDNA. This eVect was much dye concentration, as illustrated for the mip and 16S more pronounced for SYBR Green I, where the rate of rDNA amplicons in Fig. 3. In the case of SYBR Green I, increase in the Tm was higher than that for SYTO9 for both amplicons could be detected at high dye concentra- dye concentrations above 2 M. In the case of SYTO9 tions but the peaks were broad and diYcult to interpret. the Tm plateaued for concentrations of dye at or below As the concentration of SYBR Green I decreased, the V 2 M, and the maximum di erence in Tm between low melting peaks became sharper, with best peak resolution (62 M) and high (33 M) dye concentrations was 2 °C obtained for concentrations below 2 M. At lower con- for the three amplicons analyzed. In contrast, the Tm val- centrations of the SYBR Green I, the peak for mip ues determined using SYBR Green I did not plateau became more diYcult to detect or could not be detected until the dye concentration was 0.5 M, and the diVer- (Fig. 3). ence in Tm between the lowest and the highest dye concentrations was 7 °C for hlyA, 10 °C for mip, and >10 °C Comparison of SYTO9 with SYBR Green I for real-time for 16S (the Tm for the latter being >99 °C in the pres- PCR and DNA melting curve analysis ence of 33 M dye). The observation that SYBR Green I has a greater enhancement on Tm compared to SYTO9 The utility of SYTO9 was assessed by using it to mon- under the same conditions suggests either that SYBR itor DNA ampliWcation in real-time for the Legionella Green I binds more tightly to dsDNA or that more multiplex, Vibrio multiplex, and 16S rDNA PCRs. These SYBR Green I molecules can bind per unit DNA. This assays amplify fragments of 114 bp (mip) and 386 bp Wnding is supported by the recent work of Zipper et al. (Legionella 16S rDNA), 385 bp (ctxAB) and 497 bp [26], which suggests that SYBR Green I has a dual mode (hlyA), and approximately 1.5 kb (Vibrio 16S rDNA), of binding to dsDNA which is dependant on the dye/ respectively. The Vibrio 16S rDNA fragment would nor- base pair ratio. The eVect of the dye concentration on mally be considered too large for real-time PCR and rep- X W the Tm was re ected in the shape of the melting peaks resents a challenge for ampli cation if any of the (Fig. 3). For SYTO9, which had relatively little eVect on conditions aVect the reaction eYciency. All reactions V Tm, the shape of the melting peaks was not a ected by were conducted in triplicate, using 250,000–25 pg of 28 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34

Fig. 3. Comparison of DNA melting curves obtained using Legionella DNA ampliWed with mip and 16S rDNA primers and diVerent concentrations of SYTO9 or SYBR Green I.

DNA per reaction. For comparison, PCRs were also similar. These results are consistent with the Wndings of conducted using SYBR Green I. Based on the results of Giglio et al. [22] and suggest that in the presence of lower the initial melting curve experiments, SYTO9 concentra- amounts of amplicon (expected for reactions starting tions of 10, 5, 2, and 0.5M and SYBR Green I concen- with less template DNA) both amplicons can be detected trations of 10, 2, 1, and 0.5 M were investigated. These by melting curve analysis, but at higher amounts of values represent concentrations across the range where amplicon the dye becomes limiting and the amplicon V V the dyes had di erent e ects on melting temperature. with the higher Tm is preferentially detected. In contrast, The results, summarized in Table 1, demonstrate that SYTO9 concentrations of 5, 2, and 0.5 M allowed the SYTO9 can be used in PCR over a broader range of con- ampliWcation of both fragments in the Legionella multi- centrations than can SYBR Green I. The Legionella mul- plex assay. All reactions using 10 M SYTO9 and reac- tiplex PCR was inhibited by SYBR Green I at dye tions with 5 and 2 M SYTO9 starting with 250,000 pg concentrations of 10 and 2 M. The smaller mip frag- of template DNA ampliWed only the mip fragment. This ment was ampliWed at a SYBR Green I concentration of suggests that high SYTO9 concentrations or a combina- 1 M, and both fragments were ampliWed using 0.5 M tion of dye concentration and high starting template dye. The ampliWcation of mip and 16S rDNA using causes the preferential ampliWcation of the smaller mip 0.5 M SYBR Green I was aVected by the amount of fragment. starting template in the assay. When starting with The Vibrio multiplex PCR was less aVected by dye 250,000 pg of template DNA only the mip fragment concentration than was the Legionella muliplex PCR. ampliWed (Fig. 4). However, both fragments ampliWed AmpliWcation was inhibited by SYBR Green I at a con- using 25,000 pg of template DNA, suggesting that the centration of 10 M, but the assay worked for the lower mip amplicon preferentially ampliWes at higher template dye concentrations. Only the hlyA amplicon could be concentrations. As the amount of starting template detected by DNA melting curve analysis although both decreased, the height of the 16S rDNA peak increased fragments ampliWed (data not shown). This result is con- and the mip peak height decreased. Using 25 pg of tem- sistent with the Wndings of Giglio et al. [22]. In the case of plate DNA the peak heights for the two amplicons were SYTO9, both hlyA and ctxAB were ampliWed in the Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 29

Table 1 concentration and starting template DNA concentration Summary of results for Legionella multiplex, Vibrio multiplex, and 16S on T . The results for mip and hlyA are presented in Fig. V m rDNA (27f/1492r) PCRs using di erent dye concentrations and 5. The T for mip was not aVected by the amount of amounts of starting DNA template m starting template DNA using 10 M SYTO9, although Dye (M) Amplicon pg DNA/reaction there was an approximate 1 °C decrease in Tm for the 250,000 25,000 2500 250 25 lower SYTO9 concentrations. For 0.5 M SYTO9 there a SYBR Green I 10 16S ¡ ¡¡¡¡ was a small increase (0.1 °C) in mip Tm as the amount of mip ¡¡¡¡¡ starting template decreased. The Tm for mip underwent a 2 16S ¡¡¡¡¡ similar trend using 1 M SYBR Green I, with an mip ¡¡¡¡¡ 1 16S ¡¡¡¡¡ increase by 0.5 °C between reactions with 25 ng com- mip +b ++++ pared with 25 pg DNA template. However, the trend was 0.5 16S ¡ ++++ reversed for mip PCRs using 0.5 M SYBR Green I. The mip +++++ Tm of the hlyA fragment was more consistent when SYTO9 10 16S ¡¡¡¡¡ ampliWed with a range of SYTO9 concentrations. The mip +++++ W 5 16S ¡ ++++ Tm was the same for reactions ampli ed with 10–2 M mip +++++ SYTO9 (Fig. 5). Similarly, reactions ampliWed with 2 M 2 16S + + + + + SYBR Green I had the same Tm irrespective of the mip +++++ amount of starting template. The Tm for hlyA using 0.5 16S + + + + + 0.5 M either SYTO9 or SYBR Green I underwent a mip +++++ SYBR Green I 10 hlyA ¡¡¡¡¡ trend similar to that with mip and 1 M SYBR Green I. ctxAB ¡¡¡¡¡ These results support the initial DNA melting curve 2 hlyA +++++ experiments and show that the relative DNA:dye ratio ctxAB ¡¡¡¡¡ V a ects the Tm. For reactions with high starting DNA 1 hlyA +++++ template and low dye concentration, the dye is limiting ctxAB ¡¡¡¡¡ 0.5 hlyA +++++ and so the Tm is lower than those for reactions that had ctxAB ¡¡¡¡¡ less starting template DNA and the same dye concentra- SYTO9 10 hlyA +++++ tion. An exception to this was the mip Tm values with ctxAB +++++ 0.5 M SYBR Green I. In this case, the Tm decreased 5 hlyA +++++ with decreasing starting template DNA concentration. ctxAB +++++ 2 hlyA +++++ This can be explained by considering the overall results ctxAB +++++ for the Legionella multiplex PCR using 0.5 M SYBR 0.5 hlyA +++++ Green I. At the highest starting template concentration ctxAB +++++ only mip ampliWed, but at the lower concentrations both SYBR Green I 10 27f/1492r ¡¡¡¡¡ mip and 16S rDNA ampliWed. This means that there will 2 27f/1492r ¡¡¡¡¡ W 1 27f/1492r + + + + + be more total ampli ed DNA in reactions starting with 0.5 27f/1492r + + + + + lower amounts of starting template. In addition, prefer- SYTO9 10 27f/1492r + + + + ¡ ential binding of SYBR Green I to the higher-Tm 16S 5 27f/1492r + + + + + rDNA fragment would make less dye available to bind 2 27f/1492r + + + + + to the mip product. These factors would account for the 0.5 27f/1492r + + + + + decrease in T observed for mip. a No speciWc melt peak detected. m b SpeciWc melt peak detected. Determination of dye stability and between-run reproducibility presence of 10–0.5 M dye. The results for the large 16S The stability of SYTO9 and SYBR Green I and the rDNA fragment were similar, with the fragment being reproducibility of ampliWcation and DNA melting ampliWed using all concentrations of SYTO9 tested, curves were examined by testing replicate aliquots although reactions containing 10 M SYTO9 and 25 pg (n D 4) of master mixes stored at ¡20 °C over a 3-week DNA failed. AmpliWcation of the large 16S rDNA frag- period. DiVerent dye concentrations (2 and 0.5 M) and ment was inhibited in the presence of 10 and 2 M starting amounts of DNA template (25,000 and 25 pg) SYBR Green I but not for lower dye concentrations. The were investigated. The resulting Tm and Ct values for the results suggest that the PCR inhibition is dependent not various assays are summarized in Tables 2 and 3, respec- only on the dye concentration but also on other factors tively. Any detrimental changes in the master mixes due such as the primers and amplicons. to dye degradation because of storage should be The DNA melting curve data from these experiments reXected by changes in either of these values. The results allowed the investigation of the combined eVect of dye for the fresh master mixes (week 0) were consistent with 30 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34

Fig. 4. Comparison of DNA melting curves for the Legionella multiplex PCR incorporating 0.5 M SYBR Green I and diVerent amounts of starting DNA template.

V V Fig. 5. Tm values for mip and hlyA determined from PCRs incorporating di erent dye concentrations and di erent amounts of starting DNA template. the previous results, with 2 M SYBR Green I inhibiting most variable and consistently had larger standard devi- all of the Legionella multiplex PCRs and the 16S rDNA ations than those observed under other reaction condi- PCRs with 25 pg of DNA template. In addition, the tions (Table 2). The results for the 27f/1492r PCR were ctxAB amplicon could not be detected by DNA melting more variable between each week for both dyes but there curve analysis using either concentration of SYBR was marginal variation between replicates ampliWed on W Green I. The Tm values of the amplicons from the the same week. Ampli cation for this assay was variable Legionella multiplex and Vibrio multiplex PCRs were for reactions containing 2 M SYTO9 and 25 pg DNA highly consistent over the 3-week period for reactions after week 0, with some reactions failing to amplify the with either SYTO9 concentrations or 0.5 M SYBR large 16S rDNA fragment. Similar variation was Green I. Although the variation was small (generally observed for reactions containing 2 M SYBR Green I 0.1 °C), the average Tm values for some weeks were sig- and 25,000 pg DNA (with reactions starting with 25 pg niWcantly diVerent from those obtained at week 0. The DNA failing). The reactions that failed had large W Tm values for hlyA using 2 M SYBR Green I were the amounts of nonspeci c product (data not shown). Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 31

Table 2 V Summary of Tm values for amplicons from Legionella multiplex, Vibrio multiplex, and 16S rDNA (27f/1492r) PCRs using di erent concentrations of SYTO9 or SYBR Green I and diVerent storage times Assay Dye conc. pg DNA/ SYTO9 SYBR Green I (M) reaction Week Week 0 1 2.0 3.0 0 1 2.0 3.0 Legionella 16S 2 25,000 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 —a ——— 2 25 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 — — — — 0.5 25,000 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.7 § 0.1 87.5 § 0.0 88.0 § 0.0 — 0.5 25 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.5 § 0.0 87.7 § 0.1 87.6 § 0.1 87.7 § 0.0 87.5 § 0.0 Legionella mip 2 25,000 82.5 § 0.0 82.5 § 0.0 82.4 § 0.1 82.3 § 0.0 — — — — 2 25 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 82.3 § 0.0 — — — — 0.5 25,000 82.3 § 0.0 82.2 § 0.0 82.3 § 0.0 82.1 § 0.1 82.6 § 0.1 82.6 § 0.1 83.0 § 0.2 83.1 § 0.1 0.5 25 82.3 § 0.0 82.2 § 0.0 82.4 § 0.1 82.1 § 0.1 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 82.5 § 0.1 Vibrio hlyA 2 25,000 82.7 § 0.0 82.7 § 0.0 82.7 § 0.0 82.8 § 0.1 84.6 § 0.3 85.3 § 0.1 84.8 § 0.2 84.8 § 0.5 2 25 82.7 § 0.0 82.7 § 0.0 82.7 § 0.0 82.7 § 0.0 85.0 § 0.3 85.4 § 0.2 85.4 § 0.2 85.5 § 0.0 0.5 25,000 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 83.5 § 0.0 83.5 § 0.0 83.5 § 0.0 83.5 § 0.0 0.5 25 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 82.5 § 0.0 83.5 § 0.0 83.5 § 0.0 83.5 § 0.0 83.5 § 0.0 Vibrio ctxAB 2 25,000 76.5 § 0.1 76.5 § 0.0 76.4 § 0.1 76.5 § 0.0 — — — — 2 25 76.3 § 0.1 76.5 § 0.0 76.2 § 0.2 76.5 § 0.1 — — — — 0.5 25,000 76.5 § 0.0 76.5 § 0.1 76.5 § 0.1 76.5 § 0.0 — — — — 0.5 25 76.3 § 0.0 76.5 § 0.0 76.4 § 0.1 76.5 § 0.0 — — — — 27f/1492r 2 25,000 85.4 § 0.1 85.3 § 0.0 85.5 § 0.0 85.5 § 0.0 87.4 § 0.1 86.8 § 0.0 87.3 § 0.2 87.5 § 0.0 16S rDNA 2 25 85.5 § 0.0 85.4 § 0.1 85.5 § 0.0 85.5 § 0.0 — — — — 0.5 25,000 85.1 § 0.1 85.2 § 0.0 85.2 § 0.1 85.3 § 0.0 85.5 § 0.1 85.6 § 0.1 85.5 § 0.0 85.7 § 0.0 0.5 25 85.1 § 0.1 85.1 § 0.3 85.2 § 0.1 85.2 § 0.2 85.5 § 0.0 85.6 § 0.1 85.5 § 0.0 85.5 § 0.0 W V Tm values signi cantly di erent (p < 0.05) from week 0 are shaded. a SpeciWc peak not detected.

Table 3 V Summary of Ct values for Legionella multiplex, Vibrio multiplex, and 16S rDNA (27f/1492r) PCRs using di erent concentrations of SYTO9 or SYBR Green I and diVerent storage times Assay Dye conc. pg DNA/ SYTO9 SYBR Green I (M) reaction Week Week 01230123 Legionella multiplex 2 25,000 5.0 § 0.3 5.1 § 0.6 4.7 § 0.3 6.1 § 0.1 —a ——— 22512.8§ 0.1 13.1 § 0.2 13.1 § 0.3 13.3 § 0.1 — — — — 0.5 25,000 4.4 § 0.0 5.0 § 0.0 5.0 § 0.0 5.1 § 0.1 6.1 § 0.0 6.1 § 0.0 6.2 § 0.1 6.4 § 0.1 0.5 25 13.7 § 0.1 13.9 § 0.3 14.0 § 0.3 14.0 § 0.2 13.8 § 0.2 13.8 § 0.6 13.7 § 0.5 14.1 § 0.7 Vibrio multiplex 2 25,000 4.4 § 0.0 4.4 § 0.0 4.3 § 0.0 4.4 § 0.0 20.7 § 2.1 23.4 § 1.1 20.8 § 1.8 20.6 § 2.2 22514.5§ 0.2 14.7 § 0.3 14.5 § 0.2 14.7 § 0.2 24.8 § 0.5 26.5 § 0.9 24.4 § 1.0 24.2 § 1.1 0.5 25,000 6.4 § 0.3 6.7 § 0.0 6.5 § 0.3 6.4 § 0.3 4.5 § 0.0 4.5 § 0.0 4.5 § 0.0 4.5 § 0.0 0.5 25 16.5 § 0.2 16.7 § 0.3 16.6 § 0.2 16.8 § 0.1 14.7 § 0.1 14.7 § 0.1 14.6 § 0.0 14.6 § 0.1 27f/1492r 2 25,000 2.5 § 0.0 2.5 § 0.0 2.5 § 0.0 5.3 § 0.4 14.8 § 0.3 12.8 § 0.4 13.3 § 1.0 19.4 § 0.9 16S rDNA 22515.4§ 0.5 15.4 § 0.2 15.3 § 0.6 17.8 § 0.9 24.4 § 0.6 23.0 § 0.3 21.7 § 5.9 26.3 § 2.8 0.5 25,000 3.5 § 0.0 3.4 § 0.2 3.4 § 0.2 6.3 § 0.2 2.9 § 0.0 2.9 § 0.0 2.9 § 0.0 5.5 § 0.5 0.5 25 16.1 § 0.5 16.4 § 0.2 15.9 § 0.4 24.6 § 4.1 15.0 § 0.1 15.1 § 0.4 14.8 § 0.5 21.1 § 0.5 W V Ct values signi cantly di erent from week 0 (p < 0.05) are shaded. a No ampliWcation of speciWc product.

The reaction stability was further investigated by week 0, with the exception of the 0.5 M SYBR reactions W V comparing the average Ct values for the master mixes with 25 pg DNA, which were not signi cantly di erent over time. The Ct values for the Legionella multiplex but had large standard deviations. The Vibrio multiplex assay were consistent over time and did not diVer signiW- reactions with SYTO9 were highly reproducible and the W cantly for the rst 2 weeks for either dye, with the excep- Ct values did not change over the 3 weeks of storage. tion of the master mix with 0.5 M SYTO9 and 25,000 Similar results were obtained for the Vibrio multiplex W pg DNA, which had a signi cantly lower Ct value at using 0.5 M SYBR Green I, but the results were highly week 0 compared to later weeks (Table 3). At week 3 the variable using 2 M SYBR Green I. The Ct values for the W Ct values for reactions with 2 M SYTO9 or 0.5 M ampli cation using the 27f/1492r primers were consis- SYBR Green I were signiWcantly larger than those at tent for the Wrst 2 weeks when using either concentration 32 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34

of SYTO9 or 0.5 M SYBR Green I. In all cases the Ct values were signiWcantly worse at week 3. Collectively, the DNA melting curve and Ct data suggest that master mixes containing SYTO9 or 0.5 M SYBR Green I are stable for 2–3 weeks and that the absolute time may be a function of the PCR assay. Amplicon size may be a fac- tor, considering that reactions amplifying the large 16S rDNA fragment demonstrated the most dramatic change in Ct values at week 3. However, the nature of the amplicons may also play a role considering that the Vib- rio multiplex assay was still functional after 3 weeks of storage, whereas the Legionella multiplex assay was starting to show signs of deterioration, yet the amplicons are of similar size. These results contrast with the report of Karsai et al. [14], who found that SYBR Green I is unstable when stored in TE (pH 8.0) at ¡20 °C. The W reaction buVer used in the experiments described herein Fig. 6. DNA melting curve analysis of the ampli ed Giardia gdh fragment using SYTO9. also had a pH of 8.0. Table 4

Real-time PCR and melting curve analysis of eukaryote Summary of Tm values obtained by DNA melting curve analysis of the systems gdh amplicon using SYTO9 and GC% of the amplicon for each Giardia isolate Having established the utility of SYTO9, the dye was Sample Peak 1°CPeak 2°CGC% used to convert a conventional PCR assay for barley yel- Giardia duodenalis A 86.7 89.5 60.0 low dwarf virus resistance to a real-time format. Initial Giardia duodenalis B 87.0 58.3 experiments found that the expected sized fragments, Giardia duodenalis F 88.3 90.4 62.6 which were 90bp for the barley yellow dwarf virus resis- Giardia ardeae 86.4 88.8 62.8 tance-associated allele and 101bp for the susceptible-associated allele, were ampliWed for SYTO9 concentrations of products were ampliWed (data not shown) and the Wrst 16.5M and lower but that 33 M SYTO9 was inhibitory derivative proWles of the DNA melting curves were diag- (data not shown). The GC compositions of these ampli- nostic for each assemblage or species of Giardia (Fig. 6). cons were 42.0 and 44.5%, respectively, and DNA melting The GC composition for these samples was relatively curve analysis identiWed a single peak for each allele with high (58.3–62.8%, Table 4), demonstrating that SYTO9 W Tms of 79.8§0.1 °C (nD16) for the resistance-associated is also useful for the ampli cation and analysis of GC- allele and 81.4 §0.2°C (nD16) for the susceptible-associ- rich DNA. With the exception of the Assemblage B sam- ated allele. Based on a comparative quantitation analysis ple, all amplicons produced two peaks. The average Tm of the ampliWcation data, the reaction eYciencies for sam- values from the two experiments for each sample are ples with the susceptible-associated allele (0.57§0.02, summarized in Table 4. The data suggest that the overall nD16) were signiWcantly poorer (p< 0.0001) than those GC composition of a DNA fragment is not necessarily a for samples with the resistance-associated allele good predictor of the Tm. The G. ardeae and Assemblage (0.72§0.06, nD16). The results suggest that DNA melting F amplicons possess similar GC compositions (62.8% V V curve analysis can be readily applied to the di erentiation versus 62.6%), yet the Tm values are 2 °C di erent for of these allelic variants and should be simpler and easier to each peak. In contrast, the GC composition of G. ardeae V interpret than electrophoretic analysis of small amplicons. di ers from Assemblage A by 2.8% yet the Tm values are The utility of SYTO9 was further evaluated by apply- similar (although the relative heights of the two peaks ing it to a conventional PCR assay for the parasitic pro- for each sample are diVerent—see Fig. 6). These data tozoan Giardia. Sequence data from the gdh locus suggest that DNA melting analysis can detect Wner-scale ampliWed by this assay have previously been used to diVerences in sequences, such as localized domains of determine the evolutionary relationships of the major diVerent GC composition. However, additional experi- assemblages of G. duodenalis [24], and it was predicted mental data are required to support this hypothesis. that the level of diversity may allow diVerentiation of some of these groups by DNA melting curve analysis. Isolates representing three of these assemblages and Conclusions another species of Giardia, G. ardeae, were analyzed by real-time PCR and DNA melting curve analysis in two The data clearly show that SYTO9 can be used for replicate experiments. In each case, the expected sized real-time PCR and DNA melting curve analysis of both Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34 33 prokaryote and eukaryote systems and that the results are [5] W.M. Howell, M. Jobs, A.J. Brookes, iFRET: an improved Xuo- highly reproducible. Comparison of SYTO9 with SYBR rescence system for DNA-melting analysis, Genome Res. 12 Green I demonstrated that both dyes aVect amplicon T (2002) 1401–1407. m [6] R. Higuchi, G. Dollinger, P.S. Walsh, R. GriYth, Simultaneous values in a concentration-dependent manner, but the ampliWcation and detection of speciWc DNA sequences, Biotech- eVect of SYBR Green I is much more extreme, suggesting nology (NY) 10 (1992) 413–417. either that SYBR Green I binds dsDNA more tightly or [7] R. Higuchi, C. Fockler, G. Dollinger, R. Watson, Kinetic PCR that more dye molecules can bind per unit DNA com- analysis: real-time monitoring of DNA ampliWcation reactions, pared with SYTO9. The key performance advantages of Biotechnology (NY) 11 (1993) 1026–1030. [8] T. Ishiguro, J. Saitoh, H. Yawata, H. Yamagishi, S. Iwasaki, Y. SYTO9 compared with SYBR Green I are that SYTO9 Mitoma, Homogeneous quantitative assay of hepatitis C virus can be used in real-time PCR assays over a broader range RNA by polymerase chain reaction in the presence of a Xuores- of dye concentrations without causing PCR inhibition, cent intercalater, Anal. Biochem. 229 (1995) 207–213. that it produces more robust and consistently shaped [9] C.T. Wittwer, M.G. Herrmann, A.A. Moss, R.P. Rasmussen, Con- X W DNA melting curves that are less aVected by dye concen- tinuous uorescence monitoring of rapid cycle DNA ampli cation, Biotechniques 22 (1997) 130–131 134–138. tration, and that it can be more readily used for the detec- [10] K.M. Ririe, R.P. Rasmussen, C.T. Wittwer, Product diVerentiation tion of multiple amplicons by DNA melting curve by analysis of DNA melting curves during the polymerase chain analysis without any apparent bias. Both dyes were dem- reaction, Anal. Biochem. 245 (1997) 154–160. onstrated to be stable in master mixes for 2–3 weeks, but [11] M. Bengtsson, H.J. Karlsson, G. Westman, M. Kubista, A new the absolute reaction stability and dye concentration that minor groove binding asymmetric cyanine reporter dye for real- V time PCR, Nucleic Acids Res. 31 (2003) e45. causes PCR inhibition appear to di er between assays. [12] C.T. Wittwer, G.H. Reed, C.N. Gundry, J.G. Vandersteen, R.J. The performance characteristics of SYTO9 make this dye Pryor, High-resolution genotyping by amplicon melting analysis easier to use in real-time PCR applications than SYBR using LCGreen, Clin. Chem. 49 (2003) 853–860. Green I, particularly for adapting conventional assays to [13] M. Jung, J.M. Muche, A. Lukowsky, K. Jung, S.A. Loening, a real-time format and for DNA melting curve analysis. Dimethyl sulfoxide as additive in ready-to-use reaction mixtures for real-time polymerase chain reaction analysis with SYBR These characteristics will extend the scope of using Green I dye, Anal. Biochem. 289 (2001) 292–295. dsDNA-speciWc dyes in diagnostic applications and make [14] A. Karsai, S. Muller, S. Platz, M.T. Hauser, Evaluation of a home- SYTO9 an exciting alternative to SYBR Green I. Further made SYBR green I reaction mixture for real-time PCR quantiWca- studies using a broad range of PCR assays are required to tion of gene expression, Biotechniques 32 (2002) 790–792 794–796. V determine the performance characteristics of SYTO9 for [15] K. Nath, J.W. Sarosy, J. Hahn, C.J. Di Como, E ects of ethidium bromide and SYBR Green I on diVerent polymerase chain reac- quantitative PCR, particularly in relation to how SYTO9 tion systems, J. Biochem. Biophys. Methods 42 (2000) 15–29. compares with SYBR Green I for reaction eYciency. [16] T.H. Woo, B.K. Patel, L.D. Smythe, M.L. Symonds, M.A. Norris, Given the demonstrated utility of SYTO9 for real-time R.S. Weyant, M.F. Dohnt, IdentiWcation of Leptospira inadai by PCR, it will be interesting to investigate other SYTO dyes continuous monitoring of Xuorescence during rapid cycle PCR, to determine whether the properties of SYTO9 are com- Syst. Appl. Microbiol. 21 (1998) 89–96. [17] J. Pietila, Q. He, J. Oksi, M.K. Viljanen, Rapid diVerentiation of mon to other members of this family of dyes. Borrelia garinii from Borrelia afzelii and Borrelia burgdorferi sensu stricto by LightCycler Xuorescence melting curve analysis of a PCR product of the recA gene, J. Clin. Microbiol. 38 (2000) 2756–2759. Acknowledgments [18] S. Mommert, R. Gutzmer, A. Kapp, T. Werfel, Sensitive detection of Borrelia burgdorferi sensu lato DNA and diVerentiation of Bor- relia species by LightCycler PCR, J. Clin. Microbiol. 39 (2001) We thank the Australian Water Quality Centre and 2663–2667. the South Australian Water Corporation for funding [19] K. Rantakokko-Jalava, J. Jalava, Development of conventional this work. We also thank the referees for their valuable and real-time PCR assays for detection of Legionella DNA in feedback. respiratory specimens, J. Clin. Microbiol. 39 (2001) 2904–2910. [20] S. Tanriverdi, A. Tanyeli, F. Baslamisli, F. Koksal, Y. Kilinc, X. Feng, G. Batzer, S. Tzipori, G. Widmer, Detection and genotyping of oocysts of Cryptosporidium parvum by real-time PCR and melt- References ing curve analysis, J. Clin. Microbiol. 40 (2002) 3237–3244. [21] H.X. Xu, Y. Kawamura, N. Li, L. Zhao, T.M. Li, Z.Y. Li, S. Shu, T. [1] C.A. Heid, J. Stevens, K.J. Livak, P.M. Williams, Real time quanti- Ezaki, A rapid method for determining the G + C content of bac- tative PCR, Genome Res. 6 (1996) 986–994. terial chromosomes by monitoring Xuorescence intensity during [2] A.S. Piatek, S. Tyagi, A.C. Pol, A. Telenti, L.P. Miller, F.R. DNA denaturation in a capillary tube, Int. J. Syst. Evol. Micro- Kramer, D. Alland, Molecular beacon sequence analysis for biol. 50 (Pt 4) (2000) 1463–1469. detecting drug resistance in Mycobacterium tuberculosis, Nat. Bio- [22] S. Giglio, P.T. Monis, C.P. Saint, Demonstration of preferential technol. 16 (1998) 359–363. binding of SYBR Green I to speciWc DNA fragments in real-time [3] X. Chen, P.Y. Kwok, Homogeneous genotyping assays for single multiplex PCR, Nucleic Acids Res. 31 (2003) e136. nucleotide polymorphisms with Xuorescence resonance energy [23] R.P. Haugland, S.T. Yue, P.J. Millard, B.L. Roth, 1995. USA Pat- transfer detection, Genet. Anal. 14 (1999) 157–163. ent No. 5,436,134. [4] A. Solinas, L.J. Brown, C. McKeen, J.M. Mellor, J. Nicol, N. Thel- [24] P.T. Monis, R.H. Andrews, G. Mayrhofer, P.L. Ey, Molecular sys- well, T. Brown, Duplex Scorpion primers in SNP analysis and tematics of the parasitic protozoan Giardia intestinalis, Mol. Biol. FRET applications, Nucleic Acids Res. 29 (2001) E96. Evol. 16 (1999) 1135–1144. 34 Real-time PCR and melting curve analysis using SYTO9 / P.T. Monis et al. / Anal. Biochem. 340 (2005) 24–34

[25] M.T. Suzuki, S.J. Giovannoni, Bias caused by template annealing [27] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning A Lab- in the ampliWcation of mixtures of 16S rRNA genes by PCR, oratory Manual, second ed., Cold Spring Harbor Laboratory Appl. Environ. Microbiol. 2 (1996) 625–630. Press, NY, 1989. [26] H. Zipper, H. Brunner, J. Bernhagen, F. Vitzthum, Investigations [28] N.G. Paltridge, N.C. Collins, A. Bendahmane, R.H. Symons, on DNA intercalation and surface binding by SYBR Green I, its Development of YLM, a codominant PCR marker closely linked

structure determination and methodological implications, Nucleic to the Yd2 gene for resistance to barley yellow dwarf disease, Acids Res. 32 (2004) e103 Theor. Appl. Genet. 96 (1998) 1170–1177.