Quantum Dot-Based Fluorescence Resonance Energy Transfer with

Jun 27, 2006 - Single-molecule FRET between a QD and a dye-labeled DNA in ..... which has a maximum velocity at the center and zero at the walls.26 In...
0 downloads 0 Views 215KB Size
Download PDF
Anal. Chem. 2006, 78, 5532-5537

Quantum Dot-Based Fluorescence Resonance Energy Transfer with Improved FRET Efficiency in Capillary Flows Chun-yang Zhang and Lawrence W. Johnson*

Department of Chemistry, York College and The Graduate Center, The City University of New York, Jamaica, New York 11451

Fluorescence resonance energy transfer (FRET)-based nanosensors with quantum dots (QDs) as donors and organic dyes as acceptors have long been of interest for the detection of biomolecules such as nucleic acids, but their low FRET efficiency in bulk solution has prevented the sensitive detection of nucleic acids due to the large size of the QDs and the long length of nucleic acids. Here we describe a novel approach to improve the detection sensitivity of QD-based nanosensors using single-molecule detection in a capillary flow. In comparison with bulk measurement, single-molecule detection in a capillary flow possesses the unique advantages of improved FRET efficiency, high sensitivity, prevention of photobleaching, and low sample consumption. Greater FRET efficiency was obtained due to the deformation of DNA in the capillary stream. This technique can be easily extended to sensitive bimolecular analysis in microfluidic chips, and it may also offer a promising approach to study the deformation of small nucleic acids in fluid flow.

and drug delivery.5-7 The combination of bright QDs with singlemolecule detection techniques 8,9has greatly improved detection sensitivities and detection limits; QDs have been applied to counting individual biomolecules such as nucleic acids and viruses using single-molecule, two-color coincidence detection.10,11 Recently QD-based nanosensors with QDs as FRET donors and organic dyes as acceptors have been developed for the detection of proteins, DNA, and RNA12-18 and for the RNA-peptide interaction assay;19 furthermore, self-illuminating QD conjugates, with organic dyes as donors and QDs as acceptors, have been developed for in vivo imaging, which employ bioluminescence resonance energy transfer in the absence of external excitation.20 For FRET-based nanosensors, one of the main characteristics is that FRET efficiency falls off with the sixth power of the distance between the donor and acceptor, which results in the detection sensitivity of the sensors being limited by the distance between the donor and the acceptor.15 Due to the large size of QDs and the long length of nucleic acids, the low FRET efficiency in bulk solution prevents the sensitive detection of nucleic acids with the QD-based nanosensor. Single-molecule FRET between a QD and

One of great challenges in biological studies is to obtain information from individual biomolecules, rather than from ensembles of millions of molecules. The progress in singlemolecule detection has made it possible to study the structures and interactions of individual biomolecules using single-pair fluorescence resonance energy transfer (FRET).1,2 Fluorescent dyes are usually used for single-pair FRET investigations, but their functional limitations (e.g., spectral cross-talk and nonuniform fluorophore photobleaching rates) make subsequent analysis complicated. Quantum dots (QDs) provide a viable alternative because of their unique photophysical characteristics, such as sizetunable photoluminescence spectra, high quantum yields, broad absorption, and narrow emission wavelengths.3,4 QDs are now widely used as fluorescent probes in place of traditional fluorescent dyes for genomic analysis, immunoassay, fluorescence imaging,

(5) Alivisatos, A. P. Nat. Biotechnol. 2004, 22, 47-52. (6) Medintz, I. L.; Uyeda, H. T.; Goldman, E.; Mattoussi, H. Nat. Mater. 2005, 4, 437-446. (7) Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538-544. (8) Xu, X. H.; Yeung, E. S. Science 1997, 275, 1106-1109. (9) Anazawa, T.; Matsunaga, H.; Yeung, E. S. Anal. Chem. 2002, 74, 50335038. (10) Agrawal, A.; Zhang, C. Y.; Byassee, T.; Tripp, R. A.; Nie, S. M. Anal. Chem. 2006, 78, 1061-1070. (11) Zhang, C. Y.; Johnson, L. W. Analyst 2006, 131, 484-488. (12) Patolsky, F.; Gill, R.; Weizmann, Y.; Mokari, T.; Banin, U.; Willner, I. J. Am. Chem. Soc. 2003, 125, 13918-13919. (13) Medintz, I. L.; Clapp, A. R.; Mattoussi, H.; Goldman, E. R.; Fisher, B. R.; Mauro, J. M. Nat. Mater. 2003, 2, 630-638. (14) Medintz, I. L.; Konnert, J. H.; Clapp, A. R.; Stanish, I.; Twigg, M. E.; Mattoussi, H.; Mauro, J. M.; Deschamps, J. R. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 9612-9617. (15) Clapp, A. R.; Medintz, I. L.; Mauro, J. M.; Fisher, B. R.; Bawewndi, M. G.; Mattoussi, H. J. Am. Chem. Soc. 2004, 126, 301-310. (16) Zhang, C. Y.; Yeh, H. C.; Kuroki, M. T.; Wang, T. H. Nat. Mater. 2005, 4, 826-831. (17) Bakalova, R.; Zhelev, Z.; Ohba, H.; Baba, Y. J. Am. Chem. Soc. 2005, 127, 11328-11335. (18) Zhou, D.; Piper, J. D.; Abell, C.; Klenerman, D.; Kang, D. J.; Yang, L. Chem. Commun. 2005, 4807-4809. (19) Zhang, C. Y.; Johnson, L. W. J. Am. Chem. Soc. 2006, 128, 5324-5325. (20) So, M. K.; Xu, C. J.; Loening, A. M.; Gambhir, S. S.; Rao, J. H. Nat. Biotechnol. 2006, 24, 339-343.

* To whom correspondence should be addressed. Tel: 718-262-2650. Fax: 718-262-2652. E-mail: [email protected]. (1) Weiss, S. Science 1999, 283, 1676-1683. (2) Myong, S.; Rasnik, I.; Joo, C.; Lohman, T. M.; Ha, T. Nature 2005, 437, 1321-1325. (3) Bruchez, M.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Science 1998, 281, 2013-2016. (4) Chan, W. C. W.; Nie, S. M. Science 1998, 281, 2016-2018.

5532 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

10.1021/ac0605389 CCC: $33.50

© 2006 American Chemical Society Published on Web 06/27/2006

a dye-labeled DNA in diffusion solution has previously been reported: in one case with low FRET efficiency (i.e., 1-2%),21 and the other involved a complex and long (more than 4 days) sample preparation.18 Here we describe a novel approach to improve the detection sensitivity of QD-based nanosensors using singlemolecule detection in a capillary flow. Greater FRET efficiency was obtained due to the deformation of DNA in a capillary flow. By using commercially available streptavidin-coated QDs as FRET donors and the simple sample preparation (less than 1h), this method enables the rapid and sensitive detection of DNA and RNA without the separation of unhybridized probes from target-probe hybrids. Moreover, this technique can also be extended to sensitive bimolecular analysis in microfluidic chips. To our knowledge, the improved FRET efficiency of QD-based nanosensor in a capillary flow has not been previously described. EXPERIMENTAL SECTION Sample Preparation. We designed a 5′-Cy5-labeled and 3′biotinylated oligonucleotide (5′-Cy5-AAA GGA CCA GGC GCA ACT AAA TTC A-BioTEG-3′) and its complementary oligonucleotide (5′-TGA ATT TAG TTG CGC CTG GTC CTT T-3′) as models for the study of QD-based FRET. All oligonucleotides had been purified by high-performance liquid chromatography when purchased from Integrated DNA Technology Inc. (Coralville, IA). 5′Cy5-labeled and 3′-biotinylated oligonucleotide was used as singlestranded DNA (ssDNA); double-stranded DNA (dsDNA) was obtained by hybridization of 5′-Cy5-labeled and 3′-biotinylated oligonucleotide with its complementry oligonucleotide. The hybridization experiments were performed in a buffered solution containing 100 mM Tris-HCl, 10 mM (NH4)2SO4, 3 mM MgCl2, pH 8.0. The reactions were performed by mixing 5′-Cy5-labeled and 3′-biotinylated oligonucleotide and its complementry oligonucleotide at 60 °C for 30 min (the molecular ratio was kept at 1:1). After cooling to room temperature, streptavidin-functionalized 605QDs (Invitrogen Corp., Carlsbad, CA) were added to capture the hybrids and formed 605QD/dsDNA/Cy5 complexes. At the same time, streptavidin-functionalized 605QDs were mixed with 5′-Cy5-labeled and 3′-biotinylated oligonucleotides to form 605QD/ ssDNA/Cy5 complexes. Experimental Setup for Single-Molecule Detection. An argon laser was used as the excitation light source for 605QD. The 488-nm beam was collimated, reflected by a dichroic mirror (Z488RDC, Chroma Technology Corp., Rockingham, VT) and then focused by an oil immersion 100×/1.30 NA objective lens (Olympus America, Inc., Melville, NY) on the center of a 50-µmi.d. capillary; the sample was moved through a laser-focused detection volume by the pressure-driven flow from a syringe pump (Harvard Apparatus, Holliston, MA). Photons emitted from 605QD and Cy5 were collected by the same objective, passed through the first dichroic mirror, followed by a 50-µm pinhole (Melles Griot Co., Irvine, CA), and then separated by second dichroic mirror (645 DCLP, Chroma Technology Corp.). After separation, the signal emitted from Cy5 was filtered by a band-pass filter (D680/ 30M, Chroma Technology Corp.) and detected by an avalanche photodiode (model SPCM-AQR-13, EG&G Canada, Vaudreuil, PQ, Canada) in the acceptor channel. At the same time, photons emitted from 605QD were filtered by a band-pass filter (D605/ (21) Hohng, S.; Ha, T. ChemPhysChem, 2005, 6, 956-960.

Figure 1. Principles of QD-based FRET. Cy5-labeled dsDNA/ ssDNA was caught on the 605QD surface to form 605QD/DNA/Cy5 complexes through specific streptavidin-biotin binding. FRET occurred between 605QD and Cy5 upon illumination of 605QD/DNA/ Cy5 with an excitation wavelength of 488 nm.

20M, Chroma Technology Corp.) and detected by an avalanche photodiode in the donor channel. A program written with Labview (National Instruments, Austin, TX) and a digital counter (National Instruments) were used to perform data acquisition and on-line data analysis. Fluorescent signals from both donor and acceptor channels were integrated in a 1-ms interval for a total running time of 100 s for each experiment. In single-molecule detection, a threshold is used to distinguish single-molecule fluorescence signal from random fluctuation in the background. The threshold value is determined by evaluating data from control sample. In this study, a threshold of 15 photon counts‚ms-1 was set for Cy5, and a threshold of 10 photon counts.ms-1 was set for 605QD. A burst was defined as a peak in a filtered data stream that exceeded a preset threshold. Burst counts were defined as the number of bursts detected within certain running time.16 For FRET efficiency analysis, the sum of 605QD bursts was calculated by adding all 605QD bursts one by one. This calculation was accomplished by using the Labview software. Steady-State Fluorescence Measurements. The ensemble measurements were performed in bulk solution with a 700-µL quartz cuvette. All fluorescence spectra were measured at 20 °C on a LS50B luminescence spectrometer (PerkinElmer, Inc., Wellesley, MA). Emission spectra were recorded over the wavelength range of 550-725 nm with an excitation wavelength of 488 nm at a scan rate of 2 nm/s. An excitation and emission slit width of 2.5 nm was used. Absorption Spectra Measurements. The absorption spectra of 605QD/dsDNA/Cy5 and 605QD/ ssDNA/Cy5 complexes were measured on a Lambda 19 UV/VIS/NIR spectrometer (PerkinElmer, Inc.). A spectral range of 550-700 nm was recorded at a scan rate 600 nm/min with a slit width of 2 nm. RESULTS AND DISCUSSION Principle of QD-Based FRET. Figure 1 shows the conceptual scheme of QD-based FRET. The streptavidin-coated QD functioned as both a nanoscaffold and a FRET donor in this nanosensor. Single-stranded Cy5-labeled 25-mer DNA (ssDNA) and double-stranded Cy5-labeled 25-mer DNA (dsDNA) were asAnalytical Chemistry, Vol. 78, No. 15, August 1, 2006

5533

sembled on the 605QD surface by specific streptavidin-biotin binding. The binding of dsDNA/ssDNA to 605QD resulted in the formation of 605QD/DNA/Cy5 complexes. Upon excitation with a wavelength of 488 nm, FRET occurred between 605QD and Cy5s in the complexes; the fluorescence signals of 605QD and Cy5 were observed simultaneously (Figure 1). The selection of the 605QD/ Cy5 FRET pair produced negligible cross-talk between the donor and acceptor emissions, and the broad absorption band of 605QD allowed sample excitation at 488 nm without direct excitation of Cy5. Moreover, since 605QD has a high quantum yield (∼0.6), Cy5 has a high extinction coefficient (∼250 000 M-1cm-1), and a single 605QD can efficiently couple to multiple Cy5-labeled dsDNA/ssDNA complexes around its surface,13-16 efficient FRET can be achieved even at distances approaching 2R0 for this 605QD/Cy5 FRET pair.16 Bulk Measurement of QD-Based FRET. To determine the influence of the DNA-to-605QD ratio on the FRET efficiency, both 605QD and Cy5 fluorescences were measured as a function of the DNA-to-605QD ratio (Figure 2a and b). Assuming that each QD is conjugated with 12-15 streptavidins and there are 3 available biotin binding sites per streptavidin after conjugation to QDs, in principle, up to ∼36-45 biotinylated dsDNA/ssDNA can be captured by a single 605QD. In this study, the ratio of biotinylated dsDNA/ssDNA-to-605QD (1/1-24/1) was far away from the saturable value, and all the biotinylated dsDNAs/ssDNAs were assumed to be assembled around the 605QD surface. For both the 605QD/ssDNA/Cy5 and the 605QD/dsDNA/Cy5 complexes, the 605QD fluorescence decreased, while the Cy5 fluorescence increased as a function of the increasing DNA-to-605QD ratio. At all ratios examined, both the decrease in 605QD fluorescence and the increase in Cy5 fluorescence were greater in the 605QD/ssDNA/Cy5 complexes (Figure 2b) than in the 605QD/dsDNA/Cy5 complexes (Figure 2a). The FRET efficiencies of 605QD/dsDNA/Cy5 and 605QD/ ssDNA/Cy5 complexes in ensemble measurements were determined according to eq 1,15 where FDA is the 605QD fluorescence

E ) 1 - FDA/FD

(1)

in the presence of Cy5 acceptor and FD is the 605QD fluorescence in absence of Cy5 acceptor. As shown in Figure 2c, the FRET efficiency improved with the increasing DNA-to-605QD ratio in both 605QD/dsDNA/Cy5 and 605QD/ssDNA/Cy5 complexes; this was in agreement with the improved FRET efficiency that was observed when a single 605QD had multiple Cy5-labeled dsDNA/ssDNA assembled around its surface.13-16 Moreover, the FRET efficiency was higher in 605QD/ssDNA/Cy5 complexes than in 605QD/dsDNA/Cy5 complexes at all DNA-to-605QD ratios examined (Figure 2c). The improved FRET efficiency in the 605QD/ssDNA/Cy5 complexes may be explained by the difference in the elasticity between ssDNA and dsDNA. The persistence length, a length scale for polymer stiffness and defined as the distance over which the direction of a polymer segment persists, is often used to characterize the size of DNAs in solutions and the ease with which they can be bent. The persistence length of ssDNA is ∼1.6 nm or three to four bases, whereas the persistence length of dsDNA is ∼50 nm or ∼150 bases.22 The contour length of the 25-mer 5534

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

Figure 2. Bulk measurement of QD-based FRET. (a) Evolution of the fluorescence spectra from 605QD and Cy5 as a function of the increasing dsDNA-to-605QD ratio in 605QD/dsDNA/Cy5 complexes. (b) Evolution of the fluorescence spectra from 605QD and Cy5 as a function of the increasing ssDNA-to-605QD ratio in 605QD/ssDNA/ Cy5 complexes. (c) The variance of FRET efficiency with the DNAto-605QD ratio in 605QD/dsDNA/Cy5 complexes (b) and 605QD/ ssDNA/Cy5 complexes (O). 605QD concentration: 2.8 × 10-8 M; Cy5-labeled ssDNA concentration and Cy5-labeled dsDNA concentration were varied with the DNA-to-605QD ratio as shown. Error bars show the standard deviation of three experiments.

dsDNA used in this study (contour length ∼8.5 nm for a 25 mer) was much smaller than its persistence length, making it relatively

stiff in solution and more difficult to deform without external force. However, the contour length of the 25-base ssDNA was much larger than its persistence length; thus, the ssDNA was more flexible in solution and could form a random coiled conformation. The random coiled conformation of ssDNA brought the Cy5 acceptor spatially closer to the 605QD, which led to improved FRET efficiency in 605QD/ssDNA/Cy5 complexes because of the strong dependence of FRET efficiency on the Cy5-605QD separation distance.15 Single-Molecule Detection of QD-Based FRET in a Capillary Flow. Single-molecule detection techniques have become high-resolution tools for biological studies due to their ability to detect single fluorescent molecules with a high signal-to-noise ratio.1,2,8,9 A hallmark of single-molecule detection, in contrast to ensemble measurement, is the unparalleled ability to obtain the functional form of the distribution of experimental outcomes and not merely their averages.1,2,8,9 In this work, QD-based FRET was examined by single-molecule detection in a capillary flow; photons emitted from 605QD and Cy5 were simultaneously detected by individual donor and acceptor avalanche photodiodes. Panels a and b in Figure 3 show the representative traces of fluorescence bursts from the 605QD/dsDNA/Cy5 complexes; distinct Cy5 bursts were observed (Figure 3a), and each Cy5 burst had a corresponding 605QD burst (Figure 3b), indicating that FRET occurred between 605QD and Cy5. Panels c and d in Figure 3 show the representative traces of fluorescence bursts from the 605QD/ssDNA/Cy5 complexes; distinct Cy5 bursts were also observed (Figure 3c) with each Cy5 burst having a corresponding 605QD burst (Figure 3d). It is worth noticing that there were more Cy5 bursts and 605QD bursts detected from the 605QD/dsDNA/ Cy5 complex (Figure 3a and b) than from the 605QD/ssDNA/ Cy5 complex (Figure 3c and d). In the control experiments with only 605QD, there were only the 605QD bursts from donor channel, but no Cy5 burst from the acceptor channel (data not shown), indicating that there was no cross-talk between the donor and acceptor channels. Control experiments were also performed with 605QD and free Cy5-labeled dsDNA (not biotinylated) at the Cy5-labeled dsDNA-to-605QD ratio of 24/1; no Cy5 burst was observed from the acceptor channel since there was no Cy5 linked to 605QD, further confirming that the FRET between 605QD and Cy5 resulted from Cy5 linking to 605QD. To investigate the FRET efficiency of 605QD/dsDNA/Cy5 and 605QD/ssDNA/Cy5 complexes in single molecule detection in a capillary flow, we propose a new approach to calculate the FRET efficiency of the 605QD/DNA/Cy5 complexes as they move through a capillary. This is based on eq 2, where ∑IDA is the sum

E ) 1 - FDA/FD ) 1 - (

∑I

DA/

∑I

D)

(2)

of 605QD bursts in the presence of Cy5 acceptor, and ∑ID is the sum of 605QD bursts in the absence of Cy5 acceptor. Figure 3e shows the variance of the FRET efficiency as a function of the increasing DNA-to-605QD ratio. As the DNA-to-605QD ratio increased, the FRET efficiency of both 605QD/dsDNA/Cy5 and 605QD/ssDNA/Cy5 complexes significantly improved, which was (22) Sibgh-Zocchi, M.; Dixit, S.; Ivanov, V.; Zocchi, G. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 7605-7610.

Figure 3. Single-molecule detection of QD-based FRET in a capillary flow. (a-d) Representative traces of fluorescence bursts from 605QD/dsDNA/Cy5 complexes (a, b) and 605QD/ssDNA/Cy5 complexes (c, d) at the DNA-to-605QD ratio of 18/1. (a) and (c) show the Cy5 bursts in the acceptor channel; (b) and (d) show the 605QD bursts in the donor channel. (e) The variance of FRET efficiency with the DNA-to-605QD ratio in 605QD/dsDNA/Cy5 complexes (b) and 605QD/ssDNA/Cy5 complexes (O). The flow velocity was 1 µL‚min-1. 605QD concentration: 2.0 × 10-11 M; Cy5-labeled ssDNA concentration and Cy5-labeled dsDNA concentration were varied with the DNAto-605QD ratio as shown. Error bars show the standard deviation of three experiments.

consistent with the results from ensemble measurements (Figure 2c). It is worth noting that the FRET efficiency of 605QD/dsDNA/ Cy5 complexes in a capillary flow (Figure 3e) improved substantially in comparison with that in ensemble measurement (Figure 2c). The enhanced FRET efficiency in single-molecule detection may be attributed to the deformation of DNA in the capillary stream. The conformational change of tethered DNA in fluid flow has been studied theoretically and experimentally.23,24 Recently, the conformational changes and tumbling of single DNA molAnalytical Chemistry, Vol. 78, No. 15, August 1, 2006

5535

ecules has been observed even under a slow hydrodynamic flow.25 Previous research has demonstrated the parabolic velocity profile of Poiseuille flow in a capillary flow, which has a maximum velocity at the center and zero at the walls.26 In the Poiseuille flow, the flow velocity of DNA is different on the two sides. This produces a torque that is shear gradient rate dependent. Under this torque, the DNA molecules will be deformed,26 which makes Cy5 and 605QD close to each other in the 605QD/DNA/Cy5 complexes. The deformation of ssDNA and dsDNA results in changes of average QD-Cy5 separation distance (r), which can be calculated according to eq 3,15 where n is the average number of acceptor

(

r ) R0

)

n(1 - E) E

1/6

(3)

molecules interacting with one donor and R0 is the Fo¨rster distance. For 605QD/ssDNA/Cy5 complexes, the average QDCy5 separation distance was 94.3 ( 5.5 Å in ensemble measurements and 91.4 ( 7.2 Å in a capillary flow. No significant difference was observed between these two distance values since they were within experimental error; this was not unexpected considering the shorter persistence length of ssDNA and its flexibility in solution.22 For 605QD/dsDNA/Cy5 complexes, the average QDCy5 separation distance was 120.0 ( 4.3 Å in ensemble measurements and 95.9 ( 8.5 Å in a capillary flow. There was a distinctly shorter average QD-Cy5 separation distance for 605QD/dsDNA/ Cy5 complexes in a capillary flow, which suggested that the deformations of dsDNA in a capillary stream put the Cy5 acceptor spatially closer to the 605QD donor and subsequently resulted in the improved FRET efficiency (Figure 3e). Additionally, since the measurement was performed in a continuous-flow manner inside a microcapillary, each 605QD/DNA/Cy5 complex moved rapidly through the laser illumination region only once; thus, the photobleaching of Cy5 acceptors that occurred in bulk measurement was efficiently prevented in the capillary flow;27this also helped to improve the FRET efficiency of the 605QD/DNA/Cy5 complexes in a capillary flow. To further evaluate the QD-based FRET in single-molecule detection in a capillary flow, the burst counts of 605QD and Cy5 from both the 605QD/dsDNA/Cy5 complexes and the 605QD/ ssDNA/Cy5 complexes were compared, separately. As shown in Figure 4a, the 605QD burst counts were much lower from 605QD/ ssDNA/Cy5 complexes than from 605QD/dsDNA/Cy5 complexes at the DNA-to-605QD ratios of 12/1-24/1, which indicated that more 605QD donors were quenched by Cy5 acceptors. This was consistent with the higher FRET efficiency in the 605QD/ssDNA/ Cy5 complexes (Figure 3e), because higher FRET efficiency would result in efficient quenching of 605QD donors by Cy5 acceptors based on eq 1. Further comparison of Cy5 burst counts revealed that the Cy5 burst counts were surprisingly lower from 605QD/ ssDNA/Cy5 complexes than from 605QD/dsDNA/Cy5 complexes at the DNA-to-605QD ratios of 6/1-24/1 (Figure 4b); this was (23) Smith, D. E.; Babcock, H. P.; Chu, S. Science 1999, 283, 1724-1727. (24) Smith, D. E.; Chu, S. Science 1998, 281, 1335-1340. (25) Kang, S. H.; Shortreed, M. R.; Yeung, E. S. Anal. Chem. 2001, 73, 10911099. (26) Zheng, J. J.; Yeung, E. S. Anal. Chem. 2002, 74, 4536-4547 (27) Wang, T. H.; Peng, Y. H.; Zhang, C. Y.; Wong, P. K.; Ho, C. M. J. Am. Chem. Soc. 2005, 127, 5354-5359.

5536 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

Figure 4. Variance of 605QD burst counts (a) and Cy5 burst counts (b) with the DNA-to-605QD ratio from 605QD/dsDNA/Cy5 complexes (b) and 605QD/ssDNA/Cy5 complexes (O). 605QD concentration: 2.0 × 10-11 M; Cy5-labeled ssDNA concentration and Cy5-labeled dsDNA concentration were varied with the DNA-to-605QD ratio as shown. Error bars show the standard deviation of three experiments.

contrary to the results of Figure 4a and Figure 3e, since in principle higher FRET efficiency (Figure 3e) should lead to lower 605QD burst counts (Figure 4a) and higher Cy5 burst counts from 605QD/ssDNA/Cy5 complexes. The discrepancy in the Cy5 burst counts between 605QD/dsDNA/Cy5 and 605QD/ssDNA/Cy5 complexes might be attributed to the formation of nonfluorescent Cy5 dimers.28,29 Evidence of Cy5 dimer formations was obtained by spectroscopic analysis (Figure 5). The absorption spectra of 605QD/dsDNA/Cy5 and 605QD/ssDNA/Cy5 complexes at DNAto-605QD ratios of 18/1 and 24/1 were compared (Figure 5a). For both ratios examined, the 605QD/ssDNA/Cy5 complexes showed a reduction in the absorption peak around 645 nm in comparison with 605QD/dsDNA/Cy5 complexes, indicating the formation of nonfluorescent ground-state dimers in the 605QD/ ssDNA/Cy5 complexes. Moreover, the absorption spectra of 605QD/ssDNA/Cy5 complexes underwent a red shift in comparison with 605QD/dsDNA/Cy5 complexes (Figure 5a), indicating a change of Cy5 local environment, which could have been caused by their being wrapped in the coiled ssDNA. The fluorescence spectra of 605QD/ssDNA/Cy5 complexes also exhibited a red shift in comparison with 605QD/dsDNA/Cy5 complexes at the DNA-to-605QD ratios of 18/1 and 24/1 (inset in Figure 5b), further confirming the change of Cy5 local environment due to multiple Cy5 molecules being wrapped in the coiled ssDNA. The presence of the inner-filter effect at high local (28) Anderson, G. P.; Nerurkar, N. L. J. Immunol. Methods 2002, 271, 17-24. (29) Cox, W. G.; Beaudet, M. P.; Agnew, J. Y.; Ruth, J. L. Anal. Biochem. 2004, 331, 243-254.

complexes than from the 605QD/dsDNA/Cy5 complexes, and this is contrary to our experimental results (Figures 3e and 4a). Therefore, we can exclude the possibility that the elongation of ssDNA in a capillary flow caused the lower Cy5 burst counts from the 605QD/ssDNA/Cy5 complexes. Additionally, it has been previously reported that there is the nucleobase-specific fluorescence quenching of fluorescent dyes by neighboring DNA bases as a result of the photoinduced electron-transfer reaction from guanosine ground state to the excited singlet state of the dye;31 however, since Cy5 exhibits a lower electron-accepting tendency, the quenching of Cy5 by a nucleobase such as guanosine residue can be ruled out.32 Consequently, we concluded that both Cy5 dimer formations and the inner-filter effect contributed to quenching of Cy5 in the 605QD/ssDNA/Cy5 complexes. Nucleic acids are usually detected by the formation of dsDNA through sandwich hybridization of capture probes, reporter probes, and target oligonucleotides.10,11,16 The linear increase of Cy5 burst counts with the increasing DNA-to-605QD ratio from the 605QD/dsDNA/Cy5 complexes (Figure 4b) demonstrates the high sensitivity of this QD-based nanosensor for nucleic acid detection. In principle it can distinguish even one-copy differences of target oligonucleotides linked to 605QD by using singlemolecule detection in a capillary flow. Also, it is worth mentioning that single-molecule detection in a capillary flow possesses another advantage of low sample consumption in comparison with bulk measurement; the amount of 605QD used in this single-molecule detection was only 33 amol (calculated according to the flow rate of 1 µL‚min-1 for a measurement time of 100 s), which was ∼5 orders of magnitude less than that used in bulk measurement. Figure 5. Absorption spectra (a) and steady-state fluorescence measurement (b) of 605QD/ssDNA/Cy5 and 605QD/dsDNA/Cy5 complexes. Inset b shows the Cy5 fluorescence spectra. 605QD concentration: 4.0 × 10-8M; Cy5-labeled ssDNA concentration and Cy5-labeled dsDNA concentration were varied with the DNA-to605QD ratio as shown.

concentration of Cy5 might also make contributions to the decrease of Cy5 fluorescence in 605QD/ssDNA/Cy5 complexes. The inner-filter effect could lead to the decrease of fluorsecence emission and the distortion of band shape due to the reabsorption of emitted radiation.30 The change in the Cy5 emission maximum (inset in Figure 5b) was also indicative of the inner-filter effect. It should be noted that the elongation of ssDNA in a capillary flow might also cause the lower Cy5 burst counts in the 605QD/ ssDNA/Cy5 complexes, since the elongation of ssDNA might put Cy5 acceptors out of the range of 2R0. However, if this happened, then there should be correspondingly a lower FRET efficiency and higher 605QD burst counts from the 605QD/ssDNA/Cy5 (30) Porta, P. A.; Summers, H. D. J. Biomed. Opt. 2005, 10, 034001. (31) Seidel, C. A. M.; Schulz, A.; Sauer, M. J. Phys. Chem. 1996, 100, 55415553. (32) Lieberwirth, U.; Arden-Jacob, J.; Drexhage, K. H.; Herten, D. P.; Mu ¨ ller, R.; Neumann, M.; Schulz, A.; Siebert, S.; Sagner, G.; Klingel, S.; Sauer, M.; Wolfrum, J. Anal. Chem. 1998, 70, 4771-4779.

CONCLUSIONS We have used single-molecule detection to study the FRET between a QD donor and a dye acceptor attached to DNA in a capillary flow, and improved FRET efficiency was obtained due to the deformation of DNA in the capillary stream. In comparison with bulk measurement, single-molecule detection in a capillary flow possesses the unique advantages of improved FRET efficiency, high sensitivity, prevention of photobleaching, and low sample consumption. This technique can be easily extended to sensitive bimolecular analysis in microfluidic chips and has important significance in the FRET-based flow cytometry for DNA and RNA detection. Since the deformation of DNA in fluid flow is usually observed with large DNA (such as λ-DNA) due to the limitation of observation techniques,23,24 this study may also offer a promising approach to study the deformation of small nucleic acids in fluid flow. ACKNOWLEDGMENT The authors thank Dr. R. Desamero for helpful discussions. This work was supported by NIH (GM08153). Received for review March 24, 2006. Accepted May 22, 2006. AC0605389

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5537