Toward A Multiplexed Solid-Phase Nucleic Acid Hybridization Assay

Apr 9, 2009 - Toward A Multiplexed Solid-Phase Nucleic Acid Hybridization Assay Using Quantum Dots as Donors in Fluorescence Resonance Energy ...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2009, 81, 4113–4120

Toward A Multiplexed Solid-Phase Nucleic Acid Hybridization Assay Using Quantum Dots as Donors in Fluorescence Resonance Energy Transfer W. Russ Algar and Ulrich J. Krull* Chemical Sensors Group, Department of Chemical and Physical Sciences, University of Toronto Mississauga, 3359 Mississauga Road North, Mississauga, ON L5L 1C6, Canada Solid-phase assays using immobilized quantum dots (QDs) as donors in fluorescence resonance energy transfer (FRET) have been developed for the selective detection of nucleic acids. QDs were immobilized on optical fibers and conjugated with probe oligonucleotides. Hybridization with acceptor labeled target oligonucleotides generated FRET-sensitized acceptor fluorescence that was used as the analytical signal. A sandwich assay was also introduced and avoided the need for target labeling. Green and red emitting CdSe/ZnS QDs were used as donors with Cy3 and Alexa Fluor 647 acceptors, respectively. Quantitative measurements were made via spectrofluorimetry or fluorescence microscopy. Detection limits as low as 1 nM were obtained, and the discrimination of single nucleotide polymorphisms (SNPs) with contrast ratios as high as 31:1 was possible. The assays retained their selectivity and at least 50% of their signal when tested in bovine serum and against a large background of noncomplementary genomic DNA. Mixed films of the two colors of QD and two probe oligonucleotide sequences were prepared for multiplexed solid-phase hybridization assays. It was possible to simultaneously detect two target sequences with retention of selectivity, including SNP discrimination. This research provides an important precedent and framework for the future development of QD-based bioassays and biosensors. A large number of fluorescence-based assays for the detection of metal ions, small molecules, and biomolecules have been developed. These methods can offer advantages in terms of sensitivity, multiplexing, speed, and economy. Quantum dots (QDs) offer a number of potential advantages over traditional molecular fluorophores. For example, QDs exhibit broad absorption spectra and narrow size-tunable photoluminescence (PL) spectra that are ideal for multiplexing. One area of interest has been in developing multiplexed assays with QD-based spectral barcoding.1-3 Another area of interest has been assays combining QDs and fluorescence resonance energy transfer (FRET). Selec* To whom correspondence should be addressed. E-mail: ulrich.krull@ utoronto.ca. (1) Klostranec, J. M.; Xiang, Q.; Farcas, G. A.; Lee, J. A.; Rhee, A.; Lafferty, E. I.; Perrault, S. D.; Kain, K. C.; Chan, W. C. W. Nano Lett. 2007, 7, 2812–2818. 10.1021/ac900421p CCC: $40.75  2009 American Chemical Society Published on Web 04/09/2009

tive binding events at the surface of a QD can modulate energy transfer efficiency, making FRET an ideal transduction mechanism due to its strong distance dependence. The potential for ratiometric measurements when using a fluorescent acceptor is also an advantage. In contrast to absolute measurements, ratiometric measurements are less sensitive to variations between assay preparations and changes in sample conditions. Therefore there is interest in combining the advantages of QDs with the advantages of FRET-based diagnostics. Although recent work has demonstrated the use of QDs as acceptors,4,5 assays have most commonly been developed using QDs as donors in FRET. Recent reviews have highlighted the use of QDs as donors in bioanalyses,6,7 while newer examples have included the detection of proteases, nucleases, and polymerases;8-10 adenosine triphosphate;11 glucose;12 and metal ions.13 Despite advances in the development of assays utilizing QDs and FRET, two important areas of investigation remain underdeveloped: multiplexing and solid-phase assays incorporating immobilized QDs. Only a handful of studies have demonstrated QDFRET multiplexing. These include the multiplexed detection of nucleic acids in solution by our group,14 the multiplexed detection of protease and DNase by Suzuki et al.,10 and the multiplexed detection of proteases by Kim et al.8 Similarly, solid-phase QD(2) Lee, J. A.; Mardyani, S.; Hung, A.; Rhee, A.; Klostranec, J.; Mu, Y.; Li, D.; Chan, W. C. W. Adv. Mater. 2007, 19, 3113–3118. (3) Fournier-Bidoz, S.; Jennings, T. L.; Klostranec, J. M.; Fung, W.; Rhee, A.; Li, D.; Chan, W. C. W. Angew. Chem., Int. Ed. 2008, 47, 5577–5581. (4) Ha¨rma¨, H.; Soukka, T.; Shavel, A.; Gaponik, N.; Weller, H. Anal. Chim. Acta 2007, 604, 177–183. (5) Cissell, K. A.; Campbell, S.; Deo, S. K. Anal. Bioanal. Chem. 2008, 391, 2577–2581. (6) Algar, W. R.; Krull, U. J. Anal. Bioanal. Chem. 2008, 391, 1609–1618. (7) Medintz, I. L.; Mattoussi, H. Phys. Chem. Chem. Phys. 2009, 11, 17–45. (8) Kim, Y. P.; Oh, Y. H.; Oh, E.; Ko, S.; Han, M. K.; Kim, H. S. Anal. Chem. 2008, 80, 4634–4641. (9) Huang, S.; Xiao, Q.; He, Z. K.; Liu, Y.; Tinnefeld, P.; Su, X. R.; Peng, X. N. Chem. Commun. 2008, 5990–5992. (10) Suzuki, M.; Husimi, Y.; Komatsu, H.; Suzuki, K.; Douglas, K. T. J. Am. Chem. Soc. 2008, 130, 5720–5725. (11) Chen, Z.; Li, G.; Zhang, L.; Jiang, J.; Li, Z.; Peng, Z.; Deng, L. Anal. Bioanal. Chem. 2008, 392, 1185–1188. (12) Tang, B.; Cao, L.; Xu, K.; Zhuo, L.; Ge, J.; Li, Q.; Yu, L. Chem.sEur. J. 2008, 14, 3637–3644. (13) Ruedas-Rama, M. J.; Hall, E. A. H. Analyst 2009, 134, 159–169. (14) Algar, W. R.; Krull, U. J. Anal. Chim. Acta 2007, 581, 193–201.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4113

FRET assays have been limited to the detection of maltose,15,16 the detection of protease,8 a strategy for a solid-phase immunoassay,17 and our earlier work with solid-phase nucleic acid hybridization.18 The limited work that has appeared about the development of novel solid-phase QD-FRET assays may reflect the emphasis on the adaptation of conventional solid-phase assays to the use of QDs as directly excited and detected labels (e.g., microarrays,19 immunoassays20) and also the emphasis on developing QD-based cellular probes.21,22 However, the development of solid-phase assays with immobilized QDs is an essential step toward the development of QD-FRET biosensors. Immobilization of QDs and biorecognition chemistry is advantageous in developing a reusable device, particularly with flow cells or microfluidic platforms. The immobilization of QDs is also a greener approach that reduces the risk of environmental contamination with the toxic components of QDs. Advantages of using multiple FRET pairs with QDs as donors include the detection of multiple targets without the need for discrete sensor elements, spots (e.g., microarrays), or wells, and without sorting technology, single-molecule spectroscopy, or multiple excitation sources. Although microarrays have greater multiplexing capacity, the immobilized QD-FRET strategy is easier to fabricate and may have potential applications within tissues or cells if transferred to microfibers or microparticles. Learning to develop multiplexed assays and sensor platforms using immobilized QDs can significantly advance the role of nanotechnology in biotechnology. In this work, we report a novel multiplexed solid-phase hybridization assay for the selective detection of nucleic acids. Our strategy is shown in Figure 1. Green emitting CdSe/ZnS QDs (gQDs) and red emitting CdSe/ZnS QDs (rQDs) were immobilized on fused silica optical fibers using silane-based multidentate surface ligand exchange.23 A physisorbed layer of Neutravidin provided a bridge between immobilized QDs and biotinylated probe oligonucleotides. Hybridization with acceptor labeled target oligonucleotides provided the proximity required for FRET. Cy3 was used as the acceptor for gQDs (Figure 1a); Alexa Fluor 647 (A647) was used as the acceptor for rQDs (Figure 1c). It was possible to avoid target labeling via a sandwich assay (Figure 1b). Single assays used one color of immobilized QD conjugated with probe oligonucleotides. A multiplexed assay was also possible using mixed films of two different colors of QDs and two different probe oligonucleotides. This new work builds on our earlier report about QD immobilization and proof-of-concept for a solid-phase QD-FRET nucleic acid hybridization assay.18 Advances include increased acceptor PL, lower limits of detection, single nucleotide polymorphism (SNP) discrimination, analysis (15) Medintz, I. L.; Sapsford, K. E.; Clapp, A. R.; Pons, T.; Higashiya, S.; Welch, J. T.; Mattoussi, H. J. Phys. Chem. B 2006, 110, 10683–10690. (16) Sapsford, K. E.; Medintz, I. L.; Golden, J. P.; Deschamps, J. R.; Uyeda, H. T.; Mattoussi, H. Langmuir 2004, 20, 7720–7728. (17) Tran, P. T.; Goldman, E. R.; Anderson, G. P.; Mauro, J. M.; Mattoussi, H. Phys. Status Solidi B 2002, 229, 427–432. (18) Algar, W. R.; Krull, U. J. Langmuir 2009, 25, 633–638. (19) Gerion, D.; Chen, F. Q.; Kannan, B.; Fu, A. H.; Parak, W. J.; Chen, D. J.; Majumdar, A.; Alivisatos, A. P. Anal. Chem. 2003, 75, 4766–4772. (20) Yang, L. J.; Li, Y. B. Analyst 2006, 131, 394–401. (21) 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. (22) Smith, A. M.; Duan, H. W.; Mohs, A. M.; Nie, S. M. Adv. Drug Delivery Rev. 2008, 60, 1226–1240. (23) Algar, W. R.; Krull, U. J. Langmuir 2008, 24, 5514–5520.

4114

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

Figure 1. QD-FRET solid-phase hybridization assay design: a mixture of gQDs and rQDs were immobilized on an optical fiber. Probe oligonucleotides were immobilized on the QDs using a Neutravidin (NA) bridge. BSA was used to passivate any remaining adsorption sites. (a) A direct assay was possible using the gQD-Cy3 FRET pair with labeled target. (b) Target labeling was avoided with a sandwich assay. (c) A direct assay was also possible using the rQD-A647 FRET-pair. With the use of mixed films of gQDs, rQDs, and two probe oligonucleotides, it was possible to combine the pairs in parts a and c in a multiplexed assay.

by both spectrofluorimetry and fluorescence microscopy, and multiplexed detection. EXPERIMENTAL SECTION A detailed description of the reagents, procedures, and instrumentation used in these experiments can be found in the Supporting Information. Fiber Preparation. Fused silica optical fibers (0.4 mm diameter × 40 mm) or glass beads (2 mm diameter) were cleaned and modified with immobilized multidentate thiol ligands as described previously.18,23 CdSe/ZnS QDs were obtained from Professor Warren C. W. Chan at the University of Toronto and were prepared using published methods.24-26 The QDs were rendered water-soluble via ligand exchange using 3-mercaptopropionic acid and immobilized on the modified fibers from 0.2-0.8 µM solutions.17,22 When preparing mixed films of QDs for multiplexed assays, the two colors of QDs were mixed in the desired proportion in organic solvent. Ligand exchange with MPA and subsequent immobilization was then done using the mixture of QDs. For an approximately 1:1 gQD-to-rQD PL ratio on the fiber surface, it was found that the gQD-to-rQD concentration ratio should be approximately 6:1. (24) Hines, M. A.; Guyot-Sionnest, P. J. Phys. Chem. 1996, 100, 468–471. (25) Fischer, H. C.; Liu, L.; Pang, K. S.; Chan, W. C. W. Adv. Funct. Mater. 2006, 16, 1299–1305. (26) Jiang, W.; Mardyani, S.; Fischer, H.; Chan, W. C. W. Chem. Mater. 2006, 18, 872–878.

Table 1. Oligonucleotide Sequencesa

a TGT ) target; 1BPM ) single base-pair mismatch; NC ) noncomplementary; REP ) reporter. Additional sequences can be found in the Supporting Information.

Once modified with QDs, the fibers were incubated successively in (1) 0.2 mg/mL solution of Neutravidin in borate buffer at pH 8.5 for 75 min; (2) a 1.0 µM solution of biotinylated probe oligonucleotide(s) in TB buffer at pH 7.4 for 75 min; and (3) 0.5 mg/mL solution of BSA in TB buffer for 30 min. Fibers were rinsed in water between steps. The probe oligonucleotide sequences are given in Table 1. Hybridization Experiments. Once completely modified, three replicate optical fibers were immersed in sample solutions for 45 min. Exceptions were assays in bovine serum or with salmon sperm DNA (1 h) or experiments that measured the limit of detection (4 h). After the incubation period, the fibers were rinsed in TB buffer and then measured. In sandwich assays, the assay fibers were immersed in a 0.20-0.25 µM solution of reporter oligonucleotide (5′Cy3-gREP) for 45 min and rinsed in buffer prior to measurement. In some experiments, additives were used in the sample solutions. Readers should refer to the Results and Discussion section or the Supporting Information for details. Most of the oligonucleotides used in the sample solutions are given in Table 1 (also refer to Table S1, Supporting Information). Cy3 or A647 labeled oligonucleotides were used in direct assays. The Cy3 label was associated with the SMN1 probe/target combination, and the A647 label was associated with the uidA probe/target combination. In sandwich assays, nonlabeled oligonucleotides were used in combination with a Cy3 labeled reporter sequence. The sequences in Table 1 are aligned to illustrate the hybridization between the various probe/target (and reporter) combinations, as well as the location of the Cy3 or A647 label.

The SMN1 sequence is diagnostic of spinal muscular atrophy; the uidA sequence is diagnostic of Escherichia coli.27 Abbreviations are used through the text to refer to sample oligonucleotides. These included noncomplementary (5′Cy3-gNC/ 5′A647-rNC), single base pair mismatched (3′Cy3-g1BPM/ g1BPM), and fully complementary (3′Cy3-gTGT/gTGT/3′A647rTGT) sequences. The lowercase prefix refers to the FRET pair and probe sequence, g ) gQD + SMN1 probe + Cy3; r ) rQD + uidA probe + A647. Data Analysis. Raw PL spectra obtained from the assay of optical fibers were background corrected, and a FRET ratio was calculated to obtain quantitative data. Details on the calculations can be found in the Supporting Information and varied slightly between single versus multiplexed assays and spectrofluorimetry versus microscopy. In each case, the ratio was calculated for a FRET-pair as a measure of the acceptor PL (numerator) over the donor PL (denominator). Estimates of Fo¨rster distances were calculated as described previously.14 RESULTS AND DISCUSSION The FRET Pairs. The PL spectra for each donor (gQD, rQD) and acceptor (Cy3, A647) are shown in Figure 2a (refer to Supporting Information, Figure S3, for absorption spectra). The spectral overlap integrals for the gQD-Cy3 and rQD-A647 pairs were 5.6 × 10-10 and 1.9 × 10-9 cm6, respectively. The solution phase quantum yields for these MPA-capped QDs were (27) Watterson, J. H.; Raha, S.; Kotoris, C. C.; Wust, C. C.; Gharabaghi, F.; Jantzi, S. C.; Haynes, N. K.; Gendron, N. H.; Krull, U. J.; Mackenzie, A. E.; Piunno, P. A. E. Nucleic Acids Res. 2004, 32, e18.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4115

Figure 2. (a) Normalized PL spectra for (i) gQD, (ii) Cy3, (iii) rQD, and (iv) A647. (b) gQD-Cy3 direct labeling assay for the SMN1 target. The solid black line is the initial PL spectrum. The dashed black line is the response to 150 nM 5′Cy3-gNC. The response to 3′Cy3-gTGT is also shown: (i) 10, (ii) 25, (iii) 50, (iv) 75, (v) 100, (vi) 200, and (vii) 300 nM. No SDS was added to the sample solutions. (c) gQD-Cy3 sandwich assay for the SMN1 target. The solid black line is the initial PL spectrum. The dashed black line is the response to 150 nM gNC. The response to gTGT is also shown: (i) 10, (ii) 50, (iii) 75, (iv) 100, and (v) 200 nM. No SDS was added to the sample or reporter solutions. (d) rQD-A647 direct labeling assay for the uidA target. The solid black line is the initial PL spectrum. The dashed black line is the response to 100 nM 5′A647rNC. The response to 3′A647-rTGT is also shown: (i) 25, (ii) 50, (iii) 75, and (iv) 100 nM. 0.1% SDS was added.

∼0.2-1.0%. Assuming a refractive index of n ) 1.33 and an orientation factor of κ2) 2/3, the expected Fo ¨rster distances for the gQD-Cy3 and rQD-A647 pairs were in the range of 2.3-3.0 and 2.8-3.6 nm, respectively. These values are comparable to the dimensions of a Neutravidin bridge between the QDs and probe oilgonucleotides. This suggested, a priori, that the FRET pairs and surface chemistry were suitable for developing the FRET-based binding assay. The gQD-Cy3 FRET Pair and the SMN1 Sequence. Three assay formats were initially investigated using the gQD-Cy3 FRET pair with the SMN1 probe sequence. In a direct assay (Figure 1a), the sample oligonucleotides were Cy3-labeled. Figure 2b shows the response of the gQD-SMN1 probe modified fibers to different amounts of 3′Cy3-gTGT. A linear increase in FRETsensitized emission was observed as the 3′Cy3-gTGT concentration was increased. Exposure to 5′Cy3-gNC yielded only a very small amount of FRET-sensitized acceptor emission. A contrast of 47:1 was obtained for 100 nM 3′Cy3-gTGT versus 150 nM 5′Cy3gNC. These experiments were diffusion limited, so the signal magnitude depended on both the target concentration and the incubation period. When the incubation period was 45 min, the upper limit of the dynamic range was approximately 100-150 nM. Binding curves for 200 nM 3′Cy3-gTGT, 100 nM 3′Cy3-gTGT, and 200 nM 5′Cy3-A20 are shown in Figure 3a. Full signal was achieved in approximately 75 min. Note that the signal for 200 nM 3′Cy3-gTGT was not twice as large as that for 100 nM 3′Cy3gTGT. This was consistent with fiber saturation above ∼100-150 nM of target. The limit of detection (LOD) was experimentally measured as the concentration of 3′Cy3-gTGT that gave a FRET ratio 3 standard deviations above the baseline FRET ratio. An incubation 4116

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

Figure 3. Changes in FRET-ratio as a function of time for various samples: (a) gQD-Cy3 direct assay with (i) 200 nM 5′Cy3-A20, (ii) 100 nM 3′Cy3-gTGT, and (iii) 200 nM 3′Cy3-gTGT; (b) rQD-A647 direct assay with (i) 150 nM 5′A647-rNC, (ii) 50 nM 3′A647-rTGT, and (iii) 100 nM 3′A647-rTGT. No SDS was added.

period of 4 h was used to allow sufficient binding at low target concentrations. The slow binding was a diffusion limitation due to the small amount of target. For experiments with three replicate fibers in a 1.25 mL sample volume, the LOD was 3 nM. The number of moles of target was limiting in these experiments, and lower detection limits were possible using a single fiber in the same volume. Experiments showed a 2.7-fold increase in signal when only a single fiber was immersed in 1.25 mL of 3 nM target solution. Experiments also confirmed that for a single fiber in a 1.25 mL sample volume, the limit of detection was 1 nM. This is

comparable to many molecular beacons28-30 and is an improvement on the 5 nM LOD reported in our previous work.18 In practical application, it would be advantageous to avoid labeling target oligonucleotides. One possibility is a competitive binding assay. Proof-of-concept for this type of assay was demonstrated and is shown in the Supporting Information (Figure S4). However, this approach was not very sensitive, requiring high concentrations of target in the sample for comparatively small changes in FRET-sensitized acceptor emission. Another approach that avoided direct labeling was a sandwich assay (Figure 1b). Figure 2c shows the results of the sandwich assay, where FRETsensitized acceptor PL increased with increasing gTGT concentration. In contrast to directly labeled 3′Cy3-gTGT oligonucleotides, the 5′Cy3-gREP showed a significant amount of nonspecific adsorption. The result was a relatively large amount of FRETsensitized acceptor PL in the absence of gTGT. Although the gTGT signals were easily resolved above this background, an attempt was made to suppress the nonspecific adsorption of reporter by the addition of 0.1 mg/mL BSA, 0.1% w/v SDS, or 0.1% v/v Triton X-100 to the reporter solution. Only 0.1% SDS was found to be advantageous. Adsorption was reduced by 83 ± 10% on average and increased the gTGT-to-gNC contrast almost 3-fold. With the addition of 0.1% SDS to the reporter solution, and a prior 4 h incubation period with the sample, the LOD of the sandwich assay with three fibers in a 1.25 mL volume was 3 nM. The average baseline was determined from fibers incubated with gNC for 4 h, followed by incubation with 5′Cy3-gREP. It should also be noted that the addition of 0.1% SDS also reduced nonspecific adsorption in the gQD-Cy3 direct assay to levels which were undetectable within the precision of the experiment. The rQD-A647 FRET Pair and the uidA Sequence. As a step toward multiplexing, we also investigated the direct assay format using the rQD-A647 FRET pair (Figure 1c). Although the amount of nonspecific adsorption was small in the gQD-direct assay, we found that it was not small in direct assays done using fibers modified with rQD-uidA probe conjugates. Typical contrast between 100 nM 3′A647-rTGT and 150 nM 5′A647-rNC was in the range of 3:1-6:1 (cf. 47:1 for the gQD-Cy3 pair). This can be seen in Figure 3b, which shows binding curves for the rQD system in the absence of SDS. There was a large baseline from the nonspecific adsorption of 5′A647-rNC, but the 3′A647-rTGT signals clearly scaled in proportion above this baseline. Full signal was reached within 45 min. The addition of 0.1% SDS to sample solutions reduced adsorption by a minimum of 95% and often to levels that were undetectable within the precision of the experiment. The near elimination of nonspecific adsorption can be seen in Figure 2d, along with the increase in FRET-sensitized A647 PL with increasing 3′A647-rTGT concentration. With 45 min incubation periods, the upper limit of the dynamic range was approximately 150-200 nM of 3′A647-rTGT. With a 4 h incubation period, the LOD for three fibers in the 1.25 mL sample of 3′A647rTGT was 3 nM. (28) Sunkara, V.; Hong, B. J.; Park, J. W. Biosens. Bioelectron. 2007, 22, 1532– 1537. (29) Zuo, X.; Yang, X.; Wang, K.; Tan, W.; Li, H.; Zhou, L.; Wen, J.; Zhang, H. Anal. Chim. Acta 2006, 567, 173–178. (30) Sassolas, A.; Leca-Bouvier, B. D.; Blum, L. J. Chem. Rev. 2008, 108, 109– 139.

Figure 4. SNP discrimination using 25% formamide in buffered samples at room temperature. (a) gQD-Cy3 direct assay with 150 nM 5′Cy3-gNC, 3′Cy3-gTGT, and 3′Cy3-g1BPM. (b) gQD-Cy3 sandwich assay with 150 nM concentrations of gNC, gTGT, and g1BPM. No SDS was added.

FRET Efficiency. These experiments did not represent traditional FRET experiments. Instead of discrete donor-acceptor pairs, acceptor binding sites were gradually occupied on a twodimensional film with multiple donors. Therefore, only an apparent FRET efficiency could be measured since not every QD was necessarily acting as a donor. It was found that intensity measurements were unreliable and measuring the change in QD lifetime following hybridization with labeled target was required. PL decay curves are given in the Supporting Information (Figure S5) and show multiexponential decays, as observed with many QD systems.31-33 For the purpose of providing an estimate of apparent FRET efficiency, the decay curves were fit with a monoexponential model. The resulting apparent FRET efficiencies were 18% and 16% for the gQD-Cy3 and rQD-A647 systems, respectively. We expect that the actual efficiency in exciting an acceptor bound at the surface of the QD film was actually much higher, with the lower apparent efficiency being due to an “excess” of donors. Selectivity and Complex Sample Matrixes. In many applications of hybridization assays, it is necessary to resolve a SNP. To explore the possibility of SNP discrimination using the gQDCy3 FRET assay, we first investigated discrimination between fully matched and three base pair (3BPM) mismatched sequences. Two methods were effective: increasing the sample temperature to 45 °C and adding 20% v/v of formamide to the sample. Detailed data is available in the Supporting Information. Lower nonspecific adsorption was observed when using formamide to destabilize binding of the 3BPM, and so this method was used to pursue SNP discrimination. To resolve SNPs, it was necessary to increase the formamide concentration to 25% v/v. The results are shown in Figure 4a for a direct assay with gQD-Cy3 and in Figure 4b for a sandwich assay with gQD-Cy3. In the direct assay, the 25% formamide reduced the signal for 3′Cy3-1BPM to 3% of the signal for 3′Cy3-gTGT. Similar results were obtained with the sandwich (31) Labeau, O.; Tamarat, P.; Lounis, B. Phys. Rev. Lett. 2003, 90, 257404. (32) Jones, M.; Nedeljkovic, J.; Ellingson, R. J.; Nozik, A. J.; Rumbles, G. J. Phys. Chem. B 2003, 107, 11346–11352. (33) Morello, G.; Anni, M.; Cozzoli, P. D.; Manna, L.; Cingolani, R.; De Giorgi, M. J. Phys. Chem. C 2007, 111, 10541–10545.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4117

assay, although with a higher background signal from 5′Cy3-gREP adsorption (no SDS added). Taking the 5′Cy3-gNC response as the baseline, the direct and sandwich assays provided contrast ratios of 31:1 and 12:1, respectively, for the target over the same concentration of the SNP. Another practical aspect of assays is the ability to detect target in complex sample matrixes. We tested the assay fibers in bovine serum (75% v/v with buffer and 0.1% SDS w/v) and against a large background of sheared salmon sperm DNA (587 and 831 bp fragments). The salmon sperm DNA solutions contained 1.1 mg/ mL of salmon sperm DNA (cf. 0.6 µg/mL 3′Cy3-gTGT) and 0.1% SDS. The direct gQD-Cy3, direct rQD-A647, and sandwich gQDCy3 assays all functioned in the presence of the salmon sperm, giving respective signals of 72 ± 12%, 89 ± 11%, and 131 ± 11% compared to buffer. The direct and sandwich gQD-Cy3 assays also functioned in 75% bovine serum, with respective signals of 89 ± 11% and 63 ± 10% of the value obtained in only buffer. Even in 100% serum (no additives, except the 3′Cy3-gTGT spike), the direct gQD-Cy3 assay retained 50% of the target signal compared to buffer and no problematic quenching or band gap emission was observed. However, the rQD-A647 system showed significant quenching of the rQD PL and strong band gap emission after exposure to 75% bovine serum. However, an A647 fluorescence profile became clearly visible in the PL spectrum only when 5′A647-rNC was added to the serum. DNA extraction or protein precipitation steps may circumvent the problem of rQD quenching in serum. All of the assays functioned when run with a background of noncomplementary nucleic acid present, as is most likely to be encountered in real samples. Glass Bead Substrates and Microscopy Experiments. Assays using fluorescence microscopy have become very common. To test if these assays were suitable for analysis by fluorescence microscopy, we moved the chemistry onto glass beads instead of optical fibers. The glass beads had approximately the same surface area as our fibers (2 mm diameter beads), and no change in experimental methods was required. Potentially, the assay could also be reduced to micrometer scale particles for compatibility with techniques such as flow cytometry and suspension arrays. In microscopy experiments, gQD PL and Cy3 PL were separated into “channels” using band-pass filters. An image was serially acquired in each channel, and quantitative comparisons were made via the intensity ratio between the Cy3 and gQD channels. Figure 5 illustrates qualitative comparisons using falsecolor composite images. In Figure 5a, the response of beads to different concentrations of 3′Cy3-gTGT in a direct assay (no SDS) is shown. The relative Cy3 PL intensity increased as the target concentration increased. The contrast between 100 nM 3′Cy3gTGT and 100 nM 5′Cy3-gNC was 9.7:1. Saturation occurred on the beads at target concentrations above ∼100-150 nM (see Supporting Information, Figure S8). This was consistent with fiber experiments using the same incubation time. In a sandwich assay with 0.1% SDS, a gTGT-to-gNC contrast ratio of 16:1 was obtained. Figure 5b shows that SNP discrimination was also possible using the glass beads and microscopy analysis. The addition of 25% formamide provided a 3′Cy3-gTGT-to-3′Cy3-g1BPM contrast of 22:1 and a 3′Cy3-gTGT-to-5′Cy3-gNC contrast of 40:1. SNP discrimination was also possible with 25% formamide and 0.1% 4118

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

Figure 5. False-color composite images of glass bead-based gQDCy3 direct assays: (a) after exposure to different 3′Cy3-gTGT concentrations and 5′Cy3-gNC. The contrasts observed were (clockwise from top): 1:2.7:5.8:10.5:1.9. (b) SNP discrimination using 25% formamide. No SDS was added in parts a or b. Green ) gQD PL; red ) Cy3 PL; yellow/orange ) gQD PL + Cy3 PL.

SDS when using the sandwich assay format on beads. Contrasts of 9.1:1 and 7.4:1 were obtained for gTGT versus gNC and g1BPM, respectively (see Supporting Information, Figure S8). An advantage of using beads over optical fibers is that a sample solution with beads is easily agitated, overcoming diffusion limitations. For a 50 nM 3′Cy3-gTGT solution and 45 min incubation, it was found that the agitation increased the observed signal 3-fold on average compared to a bead in a solution that was not agitated (see Supporting Information, Figure S9). Therefore, with sufficient mixing, it is anticipated that 1-3 nM detection limits could be obtained in less than 2 h. Mixing is therefore an important consideration for the further development of this assay in moderate-throughput screening or real-time biosensing applications. Simultaneous Assay with Two FRET Pairs. We have previously demonstrated the multiplexing capability of combining two FRET-pairs with QD donors in solution phase experiments.14 Here, we demonstrate that this multiplexing capability can be transferred to a solid-phase assay while retaining the simplicity of the fiber preparation. In contrast to solution phase experiments where two separate preparations of the gQD-SMN1 probe and rQD-uidA probe conjugates would be required, our interfacial approach allowed the single-step preparation of fibers using mixed films of QDs and biotinylated probes. Although there was no discrete pairing of the gQD-SMN1 probe and rQD-uidA probe, immobilization at the interface retained the proximity required for both FRET pairs. Figure 1 illustrates the mixed film of QDs and oligonucleotide probes, where the multiplexed assays were a combination of parts a and c of Figure 1. Figure 6a shows the PL spectra of fibers prepared with mixed films after exposure to a variety of sample solutions containing 3′Cy3-gTGT and 3′A647-rTGT (and 0.1% SDS). The initial spectrum

Figure 6. Multiplexed assays for SMN1 and uidA target sequences. (a) PL spectra obtained with different concentrations of 3′Cy3-gTGT/ 3′A647-rTGT: (i) 25 nM/120 nM, (ii) 50 nM/80 nM, (iii) 75 nM/50 nM, and (iv) 100 nM/20 nM. The solid black line is the initial PL spectrum. The dashed black line shows the response to 100 nM 5′Cy3-gNC and 100 nM 5′A647-rNC. (b) FRET ratios for the spectra in part a. (c) PL spectra demonstrating SNP discrimination in the multiplexed assay. The samples contained 100 nM 3′A647-rTGT and 0.1% SDS, in addition to (i) 100 nM 3′Cy3-gTGT, (ii) 100 nM 3′Cy3-g1BPM, (iii) 100 nM 3′Cy3-gTGT + 25% formamide, and (iv) 100 nM 3′Cy3-g1BPM + 25% formamide. The solid black line is the initial PL spectrum. The dashed black line shows the response to 100 nM 5′Cy3-gNC. (d) FRET ratios for the spectra in part b. The arrow indicates the SNP.

showed only the QD PL profiles, and exposure to 5′Cy3-gNC and 5′A647-rNC yielded almost no detectable nonspecific adsorption. The samples contained increasing concentrations of 3′Cy3-gTGT and decreasing concentrations of 3′A647-rTGT. Figure 6b shows the FRET ratios for the spectra in Figure 6, illustrating the concentration dependence in the two-color experiment. Similar results were obtained when the 3′Cy3-gTGT and 3′A647-rTGT were increased simultaneously (see Supporting Information, Figure S10). Therefore, the two probe-target combinations behaved independently of one another. Figure 6c shows that SNP discrimination was also possible in multiplexed assays using 25% formamide and 0.1% SDS. The FRET ratios are shown in Figure 6d. The 25% formamide reduced the 3′Cy3-g1BPM signal to near baseline levels, with a 3′Cy3-TGT-to-3′Cy3-1BPM contrast of at least 15:1. Good 3′Cy3-gTGT and 3′A647-rTGT signal levels were retained. The FRET signals in the multiplexed assay were lower than in the single assays. This was an expected result of the mixed film preparation. The observed FRET signals depended on the number of QD donors, the number of bound acceptor labeled targets, and their average separation. In a mixed film, and compared to a single assay using only gQDs and the SMN1 probe, the gQDs were diluted by the added rQDs, and the SMN1 probes were diluted by the added uidA probes. This dilution, and the resulting larger mean separation between the immobilized gQDs and bound Cy3, yielded lower FRET ratios. However, it was possible to exercise some control over this dilution effect by changing the relative numbers of SMN1 and uidA probes. Figure 7 shows the effect of varying the SMN1 probe (gQD/Cy3 pair) to uidA probe (rQD/A647 pair) ratio on the FRET ratios for 200

Figure 7. The effect of varying the SMN1-to-uidA probe ratio in the preparation of the multiplexed assay fibers. The response to 200 nM 3′Cy3-gTGT and 200 nM 3′A647-rTGT is shown, and 0.1% SDS was added. The response to 100 nM 5′Cy3-gNC and 5′A647-rNC is also shown.

nM concentrations of 3′Cy3-gTGT and 3′A647-rTGT. As the relative amount of SMN1 probe was increased, the Cy3/gQD FRET ratio increased. Similarly, the A647/rQD FRET ratio decreased as the relative amount of uidA probe decreased. Thus the sensitivity of one FRET-pair can be enhanced, albeit at the expense of the FRET-pair, by increasing the relative amount of the corresponding probe oligonucleotide. CONCLUSIONS We have developed a multiplexed solid-phase nucleic acid hybridization assay using QDs as donors in FRET. Immobilized Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4119

QD-probe oligonucleotide conjugates were prepared using multidentate surface ligand exchange and a Neutravidin bridge. Signal transduction was based on FRET-sensitized acceptor PL when probe-target hybridization provided the proximity required for FRET. Assays were demonstrated using a single FRET pair (gQDCy3 or rQD-A647) and spectrofluorimetry on fused silica optical fiber substrates and using fluorescence microscopy on glass bead substrates. SNP discrimination was possible in the presence of 25% formamide. In addition, 0.1% SDS proved to be very effective at reducing nonspecific adsorption. The assays were possible using labeled target or a sandwich format in which a labeled reporter oligonucleotide avoided the need for a labeled target. LODs were in the range of 1-3 nM. When the assays were applied to complex matrixes such as bovine serum or a solution containing a large background of noncomplementary nucleic acid, they remained functional. The observed signal losses were less than 50%. For multiplexing, gQDs, rQDs, and two probe oligonucleotides were coimmobilized in mixed films to allow a simultaneous assay of two different target sequences on a single optical fiber. SNP discrimination was also possible in the multiplexed assay. The signal generated by each FRET-pair in the multiplexed assay was also tunable by changing the relative amounts of the two probe oligonucleotides that were coimmobilized. This work provides an

4120

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

important precedent and framework for the future development of QD-based bioassays and biosensors. ACKNOWLEDGMENT We gratefully acknowledge financial support of this research by the Natural Sciences and Engineering Research Council of Canada (NSERC). WRA is grateful to NSERC for a graduate fellowship. We thank Mr. Dawei Li in the Chan Laboratory (Professor Warren C.W. Chan) at the Institute of Biomaterials and Biomedical Engineering (IBBME), University of Toronto, for synthesizing the QDs. SUPPORTING INFORMATION AVAILABLE Detailed experimental methods, absorption spectra, competitive binding assay data, PL decay curves, direct excitation of acceptors, data for three base pair mismatches, additional data with complex matrices, results and discussion about regeneration of optical fibers, additional data with glass bead substrates, and additional results and discussion of multiplexed assays. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review February 25, 2009. Accepted March 31, 2009. AC900421P