Volume 59, Number 3, 2005 APPLIED SPECTROSCOPY 3350003-7028 / 05 / 5903-0335$2.00 / 0q 2005 Society for Applied Spectroscopy

Multiplex Single Strand Conformation PolymorphismAnalysis by Capillary Electrophoresis with On-the-FlyFluorescence Lifetime Detection

TARA MORCONE SNYDER* and LINDA B. MCGOWN*,†

Department of Chemistry, P.M. Gross Chemical Laboratory, Box 90346, Duke University, Durham, North Carolina 27708

This paper describes the use of on-the-fly fluorescence lifetime de-tection (OFLD) for multiplex single strand conformation polymor-phism (SSCP) analysis by capillary electrophoresis (CE). The dyelabels studied for multiplex SSCP-OFLD-CE analyses included RG,NBD, and BODIPY-FL. The dyes were first investigated for a modelsystem of ‘‘Wild Type’’ and ‘‘Mutant’’ 43-base fragments designedto vary by a single A/T substitution. Two dye pairs, BODIPY-FL/RG and BODIPY-FL/NBD, were then used to detect the G20210Amutation in the human prothrombin gene. Mobility correction wasrequired for the BODIPY-FL/RG system. Three ‘‘blind’’ analyseswere performed of three mixtures that combined a control fragment(wild type-BODIPY-FL) with two ‘‘unknown’’ fragments selectedamong four possibilities (wild type or mutant labeled with NBD orRG). In each multiplex analysis, the ‘‘origin’’ of the unknown frag-ments was correctly identified on the basis of fluorescence lifetimeof the dye label and the presence or absence of the mutation wascorrectly determined on the basis of conformation-induced differ-ences in migration time.

Index Headings: Single strand conformation polymorphism; Fluo-rescence lifetime detection; Capillary electrophoresis.

INTRODUCTION

Single strand conformation polymorphism (SSCP)analysis1 is widely considered to be the most commonand versatile technique for the analysis of DNA pointmutations2 due to the simplicity of the underlying theory,the ease of sample preparation, and the 70–90% mutationdetection rates.3 In SSCP analysis, the polymerase chainreaction (PCR) is used to amplify a region of a DNAsequence suspected to contain a mutation. Following am-plification, double-stranded DNA (dsDNA) is thermallydenatured into single-stranded DNA (ssDNA). Uponcooling, the ssDNA assumes a sequence-specific confor-mation due to intramolecular base interactions. Electro-phoresis of the ssDNA fragments is then performed undernon-denaturing conditions. Due to differences in size andshape of the fragments, wild type and mutant DNA mi-grate at different velocities through the gel, allowingthem to be differentiated. Capillary electrophoresis (CE)is the preferred method of electrophoretic separation dueto the speed of analysis and multiplex detection capabil-ities.4

The SSCP technique was originally implemented usingUV-Visible absorbance detection.1,4,5 In order to facilitatethe study of multiple samples in a single electrophoretic

Received 20 July 2004; accepted 29 November 2004.* Current address: Department of Chemistry and Chemical Biology,

Rensselaer Polytechnic Institute, 110 8th St., Troy, NY 12180-3590.† Author to whom correspondence should be sent. E-mail:

lbmcgown@chem.duke.edu.

run, fluorescence color detection was developed as analternative. It was applied in capillary electrophoresis,6–8

in which different fluorescent dyes were used to label thessDNA fragments for multiplex detection. Fluorescentdye attachment has been accomplished both during andafter PCR amplification, via a 59 or 39 labeling process.

Fluorescence lifetime detection in both time9 and fre-quency10 domains has been used as an alternative to fluo-rescence color for multiplex detection in CE. The advan-tages and applications of fluorescence lifetime over colorfor multiplex detection in CE have previously been de-scribed.10–14 We present here the use of frequency-do-main, on-the-fly fluorescence lifetime detection (OFLD)for multiplex SSCP analysis. A model point mutationsystem was selected for evaluation of possible dye com-binations. The dyes should have similar absorbance spec-tra so that they can be excited at the same wavelength,resolvable fluorescence lifetimes, and similar electropho-retic mobilities in order to minimize effects on electrom-igration of the labeled ssDNA. The sequences of the‘‘Wild Type’’ and ‘‘Mutant’’ (single A/T mutation) ss-DNA fragments in the model system are presented inTable I. The Wild Type DNA is a 43-base sequence thatwas chosen because of minimal likelihood of self-asso-ciation (duplex formation).15 The mutation position in thesingle-base Mutant was chosen arbitrarily. An adenine tothymine (A/T) mutation was chosen for its relatively highdetection rate (76%) in previous SSCP studies.16

The results of the model system were used to selectdye combinations for investigation of the G20210A mu-tation in the human prothrombin gene,17 which was pre-viously studied by SSCP analysis with fluorescence colordetection.18 The sequences of the wild type and mutantssDNA fragments are shown in Table I. They correspondto a 40-base region of the prothrombin gene containinga guanine to adenine (G/A) point mutation site at nucle-otide position 20210.

EXPERIMENTAL

Reagents. Oligonucleotides were synthesized and co-valently 59-labeled with fluorescent dyes by Epoch Bio-sciences (San Diego, CA). The dyes include rhodaminegreen (RG), 6-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)hexanoic acid (NBD), and 4,4-difluoro-5,7-di-methyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid(BODIPY-FL). Stock solutions (0.5 mM) of the labeledDNA were prepared in 13 Tris EDTA buffer (pH 7.0).The stock solutions were diluted with tris-borate-EDTA(TBE: 89 mM tris-borate, 2 mM EDTA, pH 8.2) (Sigma,St. Louis, MO) to concentrations of 0.16–1.0 mM for the

336 Volume 59, Number 3, 2005

TABLE I. Sequences of oligonucleotides.

Oligonucleotide Sequence

Model ‘‘WildType’’

59-ATATATATAGCAGCAGCAGCAGCAGCAGACGAC-GACGACGACTCT-39

Model ‘‘Mu-tant’’

59-ATATATATAGCAGCAGCAGCAGCAGCTGACGAC-GACGACGACTCT-39

ProthrombinWild Type

59-CAATAAAAGTGACTCTCAGCGAGCCTCA-ATGCTCCCAGTG-39

ProthrombinMutant(G20210A)

59-CAATAAAAGTGACTCTCAGCAAGCCTCA-ATGCTCCCAGTG-39

SSCP experiments. Prior to analysis, DNA solutions wereheated to 95 8C for 5 min and then snap-cooled on icefor 2 min in order to reduce DNA duplex formation andpromote ssDNA conformer formation.

Instrumentation. Results were obtained using on-the-fly fluorescence lifetime detection (OFLD),10 interfacinga commercial Beckman Coulter P/ACE 5000 CE systemand a SLM Multiharmonic Fourier transform (MHF)phase-modulation spectrophotometer (48 000 MHF, SLMInstruments, Inc., Rochester, NY). The excitation sourcewas the 488 nm line of a water-cooled argon ion laser(Model 307, Coherent, Palo Alto, CA) at a power outputof 100 mW. The excitation beam was passed through a488 nm band pass filter and the emission signal waspassed through a 488 nm holographic notch filter to re-move scattered light and a 515 nm long pass filter forwavelength selection. Scattered light (t 5 0 ns) was usedas the lifetime reference. Lifetime data containing phaseand modulation information was collected at 20 frequen-cies, using a base frequency of 4.1 MHz, a cross corre-lation frequency of 9.375 Hz, and 100 internal averagesper measurement. This yields approximately one signalaveraged measurement per 10 s.

The SSCP-OFLD experiments were performed usingbare fused silica capillaries (Polymicro Technologies)with a total length of 46.5 cm and length to the detectionwindow of 39.5 cm. Capillaries were pre-treated by run-ning 1 mM NaOH through the capillary for 2 min, fol-lowed by a 10 min rinse with deionized water. An entan-gled polymer solution of 1% (w/w) GeneScan polymer(Applied Biosystems, Fullerton, CA) in TBE buffer (89mM tris-borate, 2 mM EDTA, pH 8.2) was used as theseparation medium and as the run buffer. Spectrophoto-metric grade glycerol (Sigma, St. Louis, MO) was addedto the gel and the run buffer solutions at a concentrationof 10% (w/w). The gel solution and run buffer were fil-tered through a 0.45-mm filter prior to use. The gel so-lution was introduced into the capillary via high pressurerinsing for 10 min. DNA samples were injected electro-kinetically under an applied voltage of 193 V/cm (injec-tion times are specified in the figure captions) under re-verse polarity. Separations were performed at 20 8C usingan applied voltage of 300 V/cm.

Data Analysis. The lifetime data were analyzed usingnonlinear least squares (NLLS) analysis software (Glob-als, Unlimited, Urbana, IL). An in-house program wasused to analyze the multifrequency phase and modulationdata acquired at each interval in a single electrophero-gram, yielding simultaneous intensity and lifetime elec-tropherograms.19 Three types of electropherograms arepresented. The steady-state fluorescence intensity electro-

pherogram is simply the intensity recovered at each pointfrom the frequency-domain lifetime data (i.e., the d.c.component of the intensity modulated emission). Life-time electropherograms show the lifetimes recoveredfrom a nonlinear least squares (NLLS) fit of an a priorimodel (one-component or two-component model in thiswork) to the frequency-domain phase-modulation data ateach point along the electropherogram. Lifetime-resolvedelectropherograms are constructed by fitting the data us-ing a two-lifetime-component model in NLLS in whichthe lifetimes are fixed to pre-determined lifetimes of thetwo relevant dyes. The recovered fractional intensities arethen multiplied by the total steady-state fluorescence in-tensity at that point in the electropherogram to obtain thelifetime-resolved electropherogram of each dye.

RESULTS AND DISCUSSION

Model System. BODIPY-FL and RG, which exhibitedlifetimes of approximately 5 ns and 3 ns, respectively,were chosen as the first two-dye pair because of theirsubstantial lifetime difference. As shown in Fig. 1a, how-ever, the dyes cause labeled Wild Type fragments to mi-grate at different rates, with the RG-labeled fragment mi-grating faster than the RG-labeled fragment. In the sep-aration of Wild Type-BODIPY-FL and Mutant-RG (Fig.1b), the migration order is reversed, indicating that ef-fects of conformational differences between the frag-ments more than outweigh the differences in dye mobil-ity. This was verified by comparing the migration of WildType-RG and Mutant-RG (results not shown), supportingthe conclusion that there is a conformation-based sepa-ration mechanism between the two fragments.

Because of their easily resolved lifetimes, we contin-ued to use the BODIPY-FL/RG dye pair despite the dif-ferential effects on fragment mobility. A mobility correc-tion factor was determined from multiple SSCP-OFLDruns of a mixture of Wild Type-RG and Wild Type-BOD-IPY-FL. The difference in migration times (Dtm) betweenthem was determined to be 3.7 (60.6) s. The value ofDtm was then subtracted from the migration time of WildType-BODIPY-FL. Application of the correction factorto the electropherogram in Fig. 1a yielded almost iden-tical migration times for the two fragments (Fig. 1c). Mo-bility correction of the electropherogram in Fig. 1b al-lowed the difference in migration times due to confor-mational differences between dye-labeled Wild Type andMutant fragments to be observed (Fig. 1d).

Prothrombin G20210A Mutation. The prothrombinmutation exists as a single guanine-to-adenine (G/A) mu-tation at base position 20210 on the prothrombin gene.20

This particular mutation occurs in the untranslated 39 re-gion of the prothrombin gene and, as a result, does notdirectly affect the amino acid sequence of prothrombin,but instead affects the rate of production of the protein.17

Prothrombin is a precursor to the production of thrombin,a protein essential for blood clotting by controlling theproduction of fibrin. An increase in prothrombin maylead to thrombosis, in which increased thrombin levelscreate a propensity for venous and arterial clotting.

Figure 2a shows the steady-state and lifetime-resolvedelectropherograms of a mixture of wild type-BODIPY-FL and wild type-RG. The peak resolution due to the

APPLIED SPECTROSCOPY 337

FIG. 1. Intensity (solid line) and lifetime-resolved (dotted line) electropherograms of model system mixtures: (a) 0.25 mM Wild Type-BODIPY-FL (gray dotted line) and 0.25 mM Wild Type-RG (black dotted line); (b) 0.25 mM Wild Type-BODIPY-FL (gray dotted line) and 0.25 mM Mutant-RG (black dotted line). Samples were electrokinetically injected for 80 s. Lifetimes were recovered from NLLS analysis using a two-componentmodel in which the two lifetimes were fixed to pre-determined values. (c, d) Lifetime-resolved electropherograms of (a) and (b), respectively, aftermobility correction.

FIG. 2. Intensity (solid line) and lifetime-resolved (dotted lines) electropherograms of mixtures for prothrombin gene mutation system: (a) 0.25mM wild type-BODIPY-FL (gray dotted line) and 0.25 mM wild type-RG (black dotted line); (b) 0.25 mM wild type-BODIPY-FL (gray dotted line)and 0.25 mM mutant-RG (black dotted line). Samples were electrokinetically injected for 15 s. Lifetimes were recovered from NLLS analysis usinga two-component model in which the two lifetimes were fixed to pre-determined values. (c, d) Lifetime-resolved electropherograms of (a) and (b),respectively, after mobility correction.

338 Volume 59, Number 3, 2005

FIG. 3. Results for unknown Mixture 1. (a) Intensity (solid line) andlifetime (solid circles) electropherograms. Lifetimes were recoveredfrom NLLS analysis using a one-component model; (b) intensity (solidline) and lifetime-resolved (dotted line) electropherograms of Mixture1. Peaks were identified as wild type-BODIPY-FL (light gray), mutant-NBD (‘‘1’’, black), and wild type-RG (‘‘2’’, dark gray); (c) Lifetime-resolved electropherograms after mobility correction. Mixture 1 wasrevealed to contain 0.16 mM wild type-BODIPY-FL, 0.16 mM wildtype-RG, and 1.6 mM mutant-NBD. Samples were electrokineticallyinjected for 10 s. Lifetimes for (b) and (c) were recovered from NLLSanalysis using two-component models in which the two lifetimes werefixed to pre-determined values.

FIG. 4. Intensity (solid line) and lifetime-resolved (dotted line) elec-tropherograms of unknown Mixture 2. Peaks were identified to be wildtype-BODIPY-FL (gray), wild type-NBD (black dotted line of firstpeak), and mutant-NBD (black dotted line of second peak). Mixture 2was revealed to contain 0.16 mM wild type-BODIPY-FL and 1.6 mMeach of wild type-NBD and mutant NBD. Samples were electrokineti-cally injected for 6 s. Lifetimes were recovered from NLLS analysisusing two-component models in which the two lifetimes were fixed topre-determined values.

different dye mobilities is 0.6. Figure 2b shows the elec-tropherogram of a mixture of wild type-BODIPY-FL andmutant-RG, which has a resolution of 0.9. The increasein resolution in Fig. 2b suggests that conformational dif-ferences between wild type and mutant fragments play arole in the separation. A mixture of RG-labeled wild typeand mutant fragments was also resolved (results notshown), confirming that ssDNA conformation contributesto the differential migration of differently labeled wildtype and mutant fragments illustrated in Fig. 2b.

The mobility correction factor for the prothrombinwild type and mutant fragments was determined fromtriplicate separations of a mixture of wild type-BODIPY-FL and wild type-RG to be 10.67 (60.00) s. Figures 2cand 2d show the mobility-corrected electropherogramscorresponding to Figs. 2a and 2b, respectively. The con-formational difference between wild type and mutantfragments results in resolution of the two peaks (Fig. 2d),indicating that SSCP-OFLD is a suitable analytical meth-od for G20210A mutation detection.

G20210A ‘‘Blind’’ Analyses. Three ‘‘unknown’’ mix-tures were prepared by a third party. Each mixture con-tained wild type-BODIPY-FL as a control and two otherfragments selected among four possibilities: wild type-RG, mutant-RG, wild type-NBD, and mutant-NBD. Interms of diagnostic applications, the two dyes would cor-

respond to samples from two different sources, i.e., twohuman subjects. The experimental results were submittedto the third party, who then checked them for accuracy.

Figure 3a shows the steady-state intensity and lifetimeelectropherograms of Mixture 1. The lifetimes were re-covered from a single-component (monoexponential)NLLS fit. The recovered lifetimes of the peaks are ap-proximately 5, 2, and 3 ns (in that order), indicating thepresence of BODIPY-FL-, NBD-, and RG-labeled frag-ments. A two-component lifetime model was then usedto analyze both the 750–800 s segment (containing con-tributions from BODIPY-FL-labeled and possibly theNBD-labeled fragment) and the 800–850 s segment (con-taining the NBD-labeled and RG-labeled fragments) ofthe electropherogram. The resulting lifetime-resolvedelectropherograms are shown in Fig. 3b. Since wild type-BODIPY-FL was known to be present in the mixture, itwas only left to determine if the NBD- and RG-labeledfragments (labeled in the figure as 1 and 2, respectively)were wild type or mutant. This was done using the mo-bility correction of 10.67 s determined earlier for BOD-IPY-FL and RG (since there is no mobility shift betweenBODIPY-FL and NBD labeled fragments, there is noneed for a correction factor between those peaks). It isclear from the mobility-corrected electropherogram (Fig.3c) that fragment 2, which overlaps exactly with wildtype-BODIPY, is wild type-RG, and fragment 1 is mu-tant-NBD. These were confirmed by the third party to bethe correct assignments. This particular result illustratesmultiplex analysis of DNA from two sources (subjects),one with normal DNA (RG) and the other with the mu-tation (NBD).

Mixture 2 was analyzed using the same procedure. Itwas determined from the lifetime electropherogram (notshown) that was recovered using a one-component model

APPLIED SPECTROSCOPY 339

FIG. 5. Results for unknown Mixture 3. (a) Lifetime (solid dots) andintensity (solid line) electropherograms. (b) Mobility-corrected lifetimeelectropherogram. Peaks were identified to be wild type-BODIPY-FL(light gray), mutant-NBD (black), and mutant-RG (dark gray). Mixture3 was revealed to contain 0.16 mM wild type-BODIPY-FL, 0.16 mMmutant-RG, and 1.6 mM mutant-NBD. Samples were electrokineticallyinjected for 6 s. Lifetimes were recovered from NLLS analysis using aone-component model.

that the fragments were labeled with only two dyes, NBDand BODIPY-FL. Figure 4 shows the steady-state andlifetime-resolved intensity electropherograms recoveredusing a two-component model. Since the mobility shiftinduced by NBD and BODIPY-FL is negligible, no cor-rection was necessary. The results indicate that the un-known fragments are wild type-NBD and mutant-NBD.This was confirmed to be the correct identification. Thismixture is analogous to a sample obtained from a singleperson who is heterozygous for the prothrombin muta-tion, carrying both wild type and mutant alleles in theirDNA sample that is labeled with a single dye.

Figure 5a shows the intensity and lifetime electrophe-rograms for Mixture 3. The lifetime electropherogramwas obtained using a one-component lifetime model.Three peaks were observed, with lifetimes of approxi-mately 5.5, 1.5, and 3 ns, respectively. Based on the life-times, the first peak was assigned to the wild type-BOD-IPY-FL control, followed by NBD- and RG-labeled frag-ments. After mobility correction (Fig. 5b), the unknownfragments were identified directly from the lifetime elec-tropherogram to be mutant-NBD and mutant-RG. Thisidentification was confirmed by the third party. Like Mix-

ture 1, Mixture 3 corresponds to a situation in whichDNA samples from different subjects, each labeled witha unique dye, are run along with a control (wild type-BODIPY-FL). In the case of Mixture 3, both subjectswould be identified as carriers of the prothrombinG20210A mutation.

CONCLUSION

The results support the use of OFLD as an alternativeto fluorescence color detection for the multiplex identi-fication of point mutations by SSCP. BODIPY-FL, RG,and NBD were found to be suitable dyes for fluorescencelifetime resolution of multi-component DNA fragmentmixtures, within which point mutations could be detectedin all cases. The use of lifetime rather than color enablesresolution of the individual peaks of different fragments,even if they are completely overlapping, using multi-ex-ponential decay models that are based on simple, first-order kinetics.

ACKNOWLEDGMENTS

This work was supported by the National Institutes of Health (GrantR01HG01161). T.M.S. would also like to acknowledge graduate studentRebecca L. Owen for preparing the samples used in the ‘‘blind’’ anal-yses and verifying the experimental results.

1. M. Orita, H. Iwahana, H. Kanazawa, K. Hayashi, and T. Sekiya,Proc. Natl. Acad. Sci., U.S.A. 86, 2766 (1989).

2. I. V. Kourkine, C. N. Hestekin, and A. E. Barron, Electrophoresis23, 1375 (2002).

3. V. C. Sheffield, J. S. Beck, A. E. Kwitek, D. W. Sandstrom, and E.M. Stone, Genomics 16, 325 (1993).

4. A. W. H. M. Kuypers, P. M. W. Willems, M. J. van der Schans, P.C. M. Linssen, H. M. C. Wessels, C. H. M. M. de Bruijn, F. M.Everaerts, and E. J. B. M. Mensink, J. Chromatogr. 621, 149 (1993).

5. C. Cheng, T. Kasuga, N. D. Watson, and K. R. Mitchelson, J. Cap.Electrophor. 2, 24 (1995).

6. K. Hebenbrock, P. M. Williams, and B. L. Karger, Electrophoresis16, 1429 (1995).

7. J. Ren, A. Ulvik, P. M. Ueland, and H. Refsum, Anal. Biochem.245, 79 (1997).

8. D. H. Atha, H. M. Wenz, H. Morehead, J. Tian, and C. D.O’Connell, Electrophoresis 19, 172 (1998).

9. S. A. Soper, B. L. Legendre, and D. C. Williams, Anal. Chem. 67,4358 (1995).

10. L.-C. Li and L. B. McGown, Anal. Chem. 68, 2737 (1996).11. B. K. Nunnally, H. He, L.-C. Li, S. A. Tucker, and L. B. McGown,

Anal. Chem. 69, 2392 (1997).12. H. He and L. B. McGown, Anal. Chem. 72, 5865 (2000).13. E. Waddell, S. Lassiter, and S. A. Soper, J. Liq. Chrom. Rel. Tech-

nol. 23, 1139 (2000).14. S. Lassiter and S. A. Soper, Electrophoresis 23, 1480 (2002).15. S. Winter and G. L. Kirschstein, Nucleosides Nucleotides 16, 531

(1997).16. V. C. Sheffield, J. S. Beck, A. E. Kwitek, D. W. Sandstrom, and E.

M. Stone, Genomics 16, 325 (1993).17. S. R. Poort, F. R. Rosendaal, P. H. Reitsma, and R. M. Bertina,

Blood 88, 3698 (1996).18. T. Walz, J. Geisel, M. Bodis, J.-P. Knapp, and W. Herrmann, Elec-

trophoresis 21, 375 (2000).19. M. B. Smalley, J. M. Shaver, and L. B. McGown, Anal. Chem. 65,

3466 (1993).20. S. J. Freizner Degen and E. W. Davie, Biochemistry 26, 6165

(1987).