Anal. Chem. 2004, 76, 412-417
Evaluation of Surface-Enhanced Resonance Raman Scattering for Quantitative DNA Analysis Karen Faulds, W. Ewen Smith, and Duncan Graham*
Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow, G1 1XL, U.K.
The labeling of biological species using dyes has become common practice to aid in their detection, and immediate positive identification of specific dyes in high dilution is a key requirement. Here the detection by surfaceenhanced resonance Raman scattering (SERRS) of eight commercially available dye labels (ROX, rhodamine 6G, HEX, FAM, TET, Cy3, Cy5, TAMRA) attached to oligonucleotide strands is reported. Each of the eight labels was easily detected by using the SERRS from silver nanoparticles to produce a unique, molecularly specific spectrum. The conditions were optimized to obtain the best signal enhancement, and linear concentration graphs at low oligonucleotide concentrations were obtained. At higher concentrations (above ∼10-8 mol dm-3), curvature was introduced into the concentration graphs with the exception of rhodamine 6G, TET, and FAM, which gave linearity over the entire concentration range studied. Detection limits as low as 0.5 fmol were obtained, with lower possible if a smaller sample was analyzed. Investigation was also carried out into the effect of a Tris-HCl buffer containing the surfactant Tween 20 to aid in the prevention of surface adhesion of the oligonucleotides to the sample vessels at ultralow concentrations. The Tween 20 allowed lower detection limits to be obtained for each of the labels studied. This study shows that the different dyes commonly used with oligonucleotides can give quantitative SERRS at concentration levels not possible when the same dyes are used with fluorescence detection. The labeling of biological components is well established and common practice for detection using a wide range of techniques including microarrays, ELISA, separation science, and real-time PCR. Currently, fluorescent or chemiluminescent labels are the most widely used in biological characterization and diagnostics. These labels generally offer a high degree of sensitivity, down to single-molecule detection levels;1 however, when fluorescence detection is used, the signals are often limited by spectral overlap of the chromophores and by background signals due to other components in the sample. Surface-enhanced resonance Raman scattering (SERRS)2-4 is an alternative technique, which is also highly sensitive with single* To whom correspondence should be addressed. E-mail: Duncan.graham@ strath.ac.uk. (1) Li, H.; Ying L.; Green, J. J.; Balasubramanian, S.; Klenerman, D. Anal. Chem. 2003, 75, 1664-1670.
412 Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
molecule detection reported.5,6 It is a very attractive technique because it produces molecular-specific spectra which make it feasible to easily identify the components of a mixture without extensive separation procedures.7 Compared to fluorescence, it has a simpler, more extensive, and as yet underdeveloped labeling chemistry. The SERRS-active labels have a chromophore coincident with the excitation frequency and the ability to absorb onto a suitable metal surface.8 Many compounds of interest do not possess these properties, but addition of a SERRS-active label can achieve this, and this approach has been used successfully for a number of targets including DNA.9-12 The use of a metal surface quenches any fluorescence emitted meaning that commonly available fluorescent labels can be used for SERRS. Previously, SERRS detection of oligonucleotides has made use of a covalently attached label followed by a biological event and then detection. Vo-Dinh has used this approach to generate a gene chip for genotyping by SERRS with one label.13 The gene chip was a silver surface in an array format for detection. Using a different methodology with suspensions of silver nanoparticles, we have produced multiplexed genotyping of the mutational status of the cystic fibrosis gene by SERRS and the use of two labels.14,15 Gold nanoparticles have also been used for the detection of DNA by SERS. Mirkin and co-workers have reported the detection of gold nanoparticles with dye-labeled oligonucleotide attached. SERS was used to detect the dye labels for multiplexing purposes, and detection limits of 20 fM were obtained.16,17 (2) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. Chem. Phys. Lett. 1974, 26, 163-166. (3) Stacy, A. M.; Van Duyne, R. P. Chem. Phys. Lett. 1983, 102, 365-370. (4) Hildebrandt, P.; Stockburger, M. J. Phys. Chem. 1984, 88, 5935-5944. (5) Emory, S. R.; Nie, S. Science 1997, 275, 1102-1106. (6) Kneipp, K.; Wang, Y.; Kneipp, H.; Perelman, L. T.; Itzkan, I.; Dasari, R., R.; Feld, M. Phys. Rev. Lett. 1997, 78, 1667-1670. (7) Munro, C. H.; Smith, W. E.; White, P. C. Analyst 1995, 120, 993-1003. (8) Rodger, C.; Smith, W. E.; Dent, G.; Edmonson, M. J. Chem. Soc., Dalton Trans. 1996, 5, 791-799. (9) Isola, N. R.; Stokes, D. L.; Vo-Dinh, T. Anal. Chem. 1998, 70, 13521356. (10) Deckert, V.; Zeisel, D.; Zenobi, R.; Vo-Dinh, T. Anal. Chem. 1998, 70, 2646-2650. (11) Graham, D.; Brown, R.; Smith, W. E. Chem. Commun. 2001, 11, 10021003. (12) Brown, R.; Smith, W. E.; Graham, D. Tetrahedron Lett. 2003, 44 (7), 13391342. (13) Allain, L. R.; Vo-Dinh, T. Anal. Chim. Acta 2002, 469, 149-154. (14) Graham, D.; Mallinder, B. J.; Whitcombe, D.; Smith, W. E. ChemPhysChem. 2001, 2 (12), 746-748. (15) Graham, D.; Smith, W. E.; Linacre, A. M. T.; Munro, C. H.; Watson, N. D.; White, P. C. Anal. Chem. 1997, 69, 4703-4707. (16) Cao, C. C.; Jin, R.; Mirkin, C. A. Science 2002, 297, 1536-1540. 10.1021/ac035060c CCC: $27.50
© 2004 American Chemical Society Published on Web 12/06/2003
Table 1. Dye Labels, Sequence of Each Escherichia coli Oligonucleotide Strand, and Absorbance Maximums of the Dye Label 5′-dye label HEX TAMRA Cy3 ROX Cy5 R6G FAM TET a
oligonucleotide sequencea Int1: VT1: VT1: VT2: VT2: GP1: GP1: EB1:
5′ T*CT*CT*CT*CT*CT*CGGGCGCTCATCATAGTCTTTCTTA 3′ 5′ ATAAATCGCCATTCGTTGACTAC 3′ 5′ T*CT*CT*CT*CT*CT*CATAAATCGCCATTCGTTGACTAC 3′ 5′ GCGTCATCGTATACACAGGAGCAG 3′ 5′ T*CT*CT*CT*CT*CT*CGCGTCATCGTATACACAGGAGCAG 3′ 5′ CCCCACTGCTGCCTCCCGTAG 3′ 5′ T*CT*CT*CT*CT*CT*CCCCCACTGCTGCCTCCCGTAG 3′ 5′ T*CT*CT*CT*CT*CT*CGAAGGTCCCCCTCTTTGGTCTTGC 3′
λmax of dye label 535 565 552 585 643 524 494 521
T* is 5-propargylamine-2′-deoxyuridine.
In the previous studies, a number of key questions have not been addressed that require investigation before SERRS can be used for routine DNA detection. These include the concentration range that can be used in a quantitative manner, the limit of detection for different labels commonly available, the range of labels available, and the effect of buffer conditions. This study addresses these issues and shows how SERRS can be obtained in a quantitative manner from eight commercially available probes under different conditions. It shows that the development of quantitative methodologies for specific targets using SERRS detection is now practical. EXPERIMENTAL SECTION Labeled Oligonucleotides. The labeled oligonucleotides were purchased from Oswel (Hampshire, U.K.) and were HPLC purified. Silver Nanoparticles Preparation. A colloidal suspension of citrate-reduced silver nanoparticles was prepared using a modified Lee and Meisel18 procedure. Several batches of colloid were prepared and added together to produce 2 L of silver nanoparticles, thus allowing the same batch of colloid to be used throughout the study. Instrumentation. The following Raman instrumentation was used: a Renishaw model 100 probe system with a 514.5-nm argon ion laser, utilizing a 20× objective to focus the laser beam into a 1-cm plastic cuvette containing the sample and a Renishaw microscope system 1000 with a 632.9-nm helium-neon laser utilizing a Ventacon macrosampler to focus the laser beam into a 1-cm plastic cuvette. Sample Preparation. All samples were prepared for SERRS analysis using the following amounts of reagents: 10 µL of dyelabeled oligonucleotide, 10 µL of spermine, 250 µL of water, and 250 µL of citrate-reduced silver nanoparticles. Aggregation Optimization Experiments. Aggregation studies were carried out initially using HEX-labeled oligonucleotide. The samples were prepared as described above using 10 µL of the following concentrations of spermine tetrahydrochloride (Sigma-Aldrich) diluted in sterile water: 0.1, 0.01, 0.001, 0.0001, and 0.000 01 mol dm-3. The GRAMS software was set up to take a spectrum of the dye every 30 s for 30 min to allow the rate of aggregation to be monitored. The ratio of water to colloid and (17) Ginger, D. S.; Cao, Y. C.; Mirkin, C. A. Biophotonics Int. 2003, (July), 48-51. (18) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391-3395.
the order of addition of the analyte, spermine, and water to the colloid were also optimized. The effect of the addition of a buffered solution containing Tween 20, a surfactant to prevent adherence of the oligonucleotide to the walls of the sample vessel, was investigated. The buffer used was a 10 mM tris(hydroxymethyl)aminomethane hydrochloride (Tris-HCl; Aldrich, 99+%), 100 mM NaCl (BDH, 99.5+%), and 0.0 1% Tween 20 (Aldrich), pH 7.4.1 Concentration Studies. Concentration studies were carried out using the dye-labeled oligonucleotides (Table 1). The oligonucleotides were diluted to various concentrations using sterile water or the buffered Tween 20 solution. Samples were prepared by adding 10 µL of labeled oligonucleotide and 10 µL of 0.1 mol dm-3 spermine, followed by 250 µL of distilled water and 250 µL of citrate-reduced silver nanoparticles. The samples were analyzed within 1 min of the addition of the silver colloid, and each oligonucleotide concentration was analyzed five times. The spectra obtained were the result of a 10-s accumulation time with the spectrometer grating centered at 1400 cm-1. The spectra obtained were baseline corrected using the GRAMS/32 software, and the average peak height of the strongest peak in the spectrum was normalized to the silicon standard peak and plotted against the concentration of labeled oligonucleotide. RESULTS AND DISCUSSION SERRS requires a molecule with a chromophore that has an absorption maximum close to the excitation wavelength of the incident laser light. Also, the molecule must come in contact with or be very close to the metal surface used for enhancement. This study uses citrate-reduced silver nanoparticles, which have a net negative charge in aqueous solution due to a layer of citrate that exists on the surface of the silver particles.19 Since DNA is overall negatively charged, due to the phosphate groups present in the DNA backbone, it is unable to absorb efficiently onto the surface of the silver colloid. However, the polyamine spermine hydrochloride has been shown to interact with the DNA backbone and neutralize the charges.20 This has been used previously in aiding oligonucleotides to adsorb onto silver nanoparticles and act as an aggregating agent when in excess.15 The dye labels used here were chosen as they are all commercially available labels used routinely in fluorescence detection of oligonucleotides. Of the eight dye labels chosen, three (19) Munro, C. H.; Smith, W. E.; Garner, M.; Clarkson, J.; White, P. C. Langmuir 1995, 11, 3712-3720. (20) Basu, H. S.; Marton, L. J. Biochem J. 1987, 244, 243-246.
Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
413
Figure 1. Structures of the eight dye labels: Cy3, Cy5, TAMRA, R6G, FAM, TET, HEX, and ROX
Figure 2. SERRS spectra of the eight dye-labeled oligonucleotides, obtained using 514.5-nm laser excitation at ∼1 × 10-8 mol dm-3.
have a net positive charge in aqueous solution (rhodamine 6G (R6G), ROX, and TAMRA, see Figure 1), and no further modification was required to allow oligonucleotides modified with these labels to attach to the negatively charged silver surface. However, five of the labels have a net negative charge in aqueous solution (HEX, Cy3, Cy5, FAM, and TET; see Figure 1); therefore, further modification of the oligonucleotide was required for effective surface adsorption to occur. Propargylamine modification for negatively charged dye labels has previously been reported15 and involves the addition of six modified nucleobases at the 5′-terminus close to the dye label. In aqueous solution, the terminal primary amine groups of the propargylamines are protonated giving a positive charge, allowing the DNA to adsorb onto the negative silver surface. Therefore, the action of spermine combined with either a positively charged dye or a negatively charged dye and the propargylamine-modified bases allows good absorption of DNA and hence successful SERRS to be obtained. SERRS of Labeled Oligonucleotides. SERRS of all eight of the dye-labeled oligonucleotides were obtained using 514.5-nm laser excitation and the spectra shown in Figure 2. It can be seen that there are distinct differences between the spectra of the eight labels, allowing easy identification of the label and hence identification of the DNA sequence present. 414 Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
The SERRS procedure was optimized to obtain maximum surface enhancement. To ensure that the optimum aggregation conditions were obtained, time-dependent aggregation experiments were carried out using the HEX-labeled oligonucleotide and various concentrations of spermine tetrahydrochloride in water. The maximum signal enhancement was obtained when a 0.1 mol dm-3 spermine solution was used as the aggregating agent. Under these conditions, the signal intensity obtained was initially very high but dropped away with time. This is caused by overaggregation of the nanoparticles resulting in a decrease in signal as the silver suspension starts to precipitate due to the formation of large clusters of silver particles. However, the time scale of this process is significantly longer than the spectral acquisition so that reproducible measurements can be taken. Thus, the best signals are obtained when the samples were analyzed immediately upon addition of the spermine. The order of addition of the analyte, the silver nanoparticles, the water, and the aggregating agent was investigated. The labeled oligonucleotides and spermine solution were premixed and left together for 5 min, at room temperature and over ice, before the addition of the silver nanoparticles as this had previously been shown to give the best enhancement in signal.21 A large decrease in the signal intensity occurred when the premixed setup was used. When the samples were analyzed immediately on addition of the spermine to the oligonucleotide, signals double that obtained by premixing were obtained. The largest signal was obtained when the silver nanoparticle suspension was diluted to 50% with water. At higher nanoparticle concentrations, the lower signals were due to an increased selfabsorption of the incident and scattered light. At lower nanoparticle concentrations, there are fewer silver nanoparticles in the suspension resulting in lower signal intensities. Thus, optimum SERRS conditions were obtained when a 0.1 mol dm-3 solution of spermine was used to aggregate the nanoparticles after dilution to 50% with water, and the solution was analyzed immediately. The use of bovine serum albumin (BSA; Sigma) to prevent the adhesion of the labeled oligonucleotides to the surface of the sample containers was investigated. The Eppendorfs used to store the samples were soaked in solutions of various concentrations of BSA, before rinsing thoroughly with distilled water and drying under nitrogen. However, the BSA treatment was found to dramatically reduce the SERRS signal that was obtained from the samples, even when low BSA concentrations were used. The reason for this was unclear; thus, an alternative approach using a surfactant, Tween 20, was investigated. Klenerman and co-workers1 reported the use of a Tris-HCl buffer containing Tween 20 to prevent the surface adhesion of DNA molecules in single-molecule detection using fluorescence and showed improvements in sensitivity. Thus, the procedure used in that study was adapted for the determination of the limits of detection for each of the labeled oligonucleotides. In addition, the effects of Tris-HCl buffer alone were investigated using ROXlabeled oligonucleotide. The following diluents were compared: water, Tris-HCl buffer, Tris-HCl buffer containing 100 mM NaCl, and Tris-HCl buffer containing 100 mM NaCl and 0.01% Tween 20. The results are given in Figure 3. The highest intensity spectra (21) Graham, D.; Mallinder, B. J.; Smith, W. E. Biopolymers (Biospectroscopy) 2000, 57, 85-91.
Figure 3. Comparison between ROX-labeled oligonucleotide (at 9.77 × 10-10 mol dm-3) diluted in (A) water, (B) Tris-HCl buffer (pH 7.4), (C)Tris-HCl buffer and NaCl (pH 7.4), and (D) Tris-HCl buffer containing NaCl and Tween 20 (pH 7.4). The average of five replicates is displayed.
were obtained when the buffered solution contained Tween 20 and NaCl was used, as previously reported by Klenerman and co-workers. This was the desired result whereby the signal intensity was increased by the addition of the surfactant, which was expected to stop the surface adhesion of the DNA to the walls of the sample vessel. The Tris-HCl buffer itself actually gave a
slight reduction in the signal intensity, and when NaCl was added, the signal intensity was very similar to that obtained when water was used as the diluent. It was still necessary to use spermine with all oligonucleotide samples to give good SERRS. All of the labeled oligonucleotides were analyzed using 632.9nm laser excitation, as well as 514.5 nm. However, although effective SERS could be obtained for all of the labels, the intensity of the spectra obtained was much lower. This resulted in higher detection limits being obtained. This was due to the fact that, with the exception of Cy5, which has an absorption maximum of 643 nm, 632.9 nm is further from the λmax of the dyes than 514.5-nm laser excitation, resulting in less enhancement in signal as the excitation wavelength is no longer close to the excitation maximum of the dye. Concentration Studies. Concentration studies were carried out with each of the eight dye-labeled oligonucleotides, both in water and in the buffered Tween 20 solution. In the case of rhodamine 6G, FAM, and TET (Figure 4), the concentration graphs were linear over the entire range of concentrations studied (∼10-11-10-7 mol dm-3). The other five dye labels showed a good degree of linearity at low concentrations (below ∼10-8 mol dm-3); however, a degree of curvature was introduced at high concentrations (above ∼10-8 mol dm-3). The contrast between R6G and
Figure 4. Concentration dependence of the signal of a selected peak for each of the labeled oligonucleotides. All spectra were obtained using 514.5-nm laser excitation with the exception of Cy5, which was carried out using 632.9-nm laser excitation. Each point is the average of five spectra, and the error bars shown are one standard deviation.
Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
415
Table 2. Detection Limits Obtained Using SERRS with Silver Nanoparticles for Eight Different Dye-Labeled Oligonucleotides, Diluted in Water and Tris-HCl Buffer Containing the Surfactant Tween 20 dye label HEX TAMRA Cy3 ROX Cy5 R6G FAM TET
Figure 5. SERRS intensity versus concentration for R6G and TAMRA over the entire concentration range studied. The R6Glabeled oligonucleotide was linear over the full range, whereas the TAMRA-labeled oligonucleotide was only linear at the lower concentrations.
TAMRA is given in Figure 5. Dye-labeled oligonucleotides that readily adsorb onto the silver surface give a linear response at submonolayer concentrations; however, once monolayer is reached, competition between the formation of multilayers and adsorption onto the walls of the vessel becomes a factor. A low concentration of Tween 20 appears to prevent appreciable absorption onto the walls of the vessel. Therefore, the concentration studies were carried out within the range of concentrations that was as close to linearity as possible for each of the labels (Figure 4). The greatest deviations from linearity occurred with HEX and TAMRA. In all cases, the addition of the buffer containing Tween 20 gave an increase in the SERRS signal obtained, even though very slight in the cases of FAM and R6G. The addition of Tween 20 also generally increased the slope of the graph obtained and in most cases allowed a lower concentration of analyte to be detected suggesting that the Tween 20 helps prevent the DNA from adhering to the surface of the sample container and allows more of the sample to be detected. The concentration graphs for the three positively charged dyelabels, R6G, ROX, and TAMRA, are shown in Figure 4a-c. In the case of R6G, Figure 4a, the buffered Tween 20 solution has little effect on the intensity of the signal obtained. However, a lower concentration of the labeled oligonucleotide could be observed, presumably due to the reduced surface adhesion when highly diluted solutions were studied. The calibration graph for the R6G label was linear over the entire concentration range studied. The ROX-labeled oligonucleotide diluted with the Tween 20 buffered solution gave the largest increase in intensity of all the dye labels (Figure 4b). This may be due to the fact that the ROX dye is more hydrophobic than the other labels and more likely to stick to the surface of the containers, resulting in a higher signal when the surfactant was used. The concentration graph obtained was only 416
Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
detection limit (mol dm-3) in water in buffer/Tween 20 7.76 × 10-12 3.47 × 10-12 2.57 × 10-10 8.10 × 10-11 8.34 × 10-11 1.17 × 10-12 2.73 × 10-12 1.61 × 10-11
6.21 × 10-12 3.12 × 10-12 2.38 × 10-10 4.05 × 10-11 5.96 × 10-11 1.05 × 10-12 2.18 × 10-12 1.34 × 10-11
linear at low concentrations of oligonucleotide. In the case of TAMRA (Figure 4c), less deviation from linearity was observed when the Tween 20 was present and lower concentrations of oligonucleotide could be detected. Of the three positively charged dyes, the lowest detection limit was obtained with rhodamine 6G, where it was possible to detect the labeled oligonucleotide down to a concentration of 1.05 × 10-12 mol dm-3, which equates to 0.5 fmol of the labeled oligonucleotide in the sampling volume. The concentration graphs obtained for the five negatively charged dye labels incorporated into the oligonucleotide strand with propagylamine modification are given in Figure 4d-h. With Cy3 (Figure 4d) there is a slight curvature in the graph when water is used as the diluent, whereas when the Tween 20 buffered solution is used as the diluent, the graph becomes linear. For Cy5, 632.9-nm excitation was used and higher intensity spectra were obtained with Tween 20 (Figure 4e). The SERRS of FAM- and TET-labeled oligonucleotides (Figure 4f and g, respectively) gave a linear concentration range over all concentrations investigated. The effect of the buffered Tween 20 solution was very minor with very little signal increase, and no increase in the degree of dilution that could be analyzed. In the case of HEX-labeled oligonucleotide, a slight improvement in the signal intensity was obtained when the buffered Tween 20 solution was used (Figure 4h). The detection limits obtained for each of the eight dye-labeled oligonucleotides are given in Table 2. It can be seen that, in each case, the addition of Tween 20 decreased the detection limit obtained slightly, although not by a huge amount in the majority of cases. The most dramatic decrease was seen with the ROX label, where the detection limit was halved. Thus, we can state that the optimal conditions for quantitative detection of labeled oligonucleotides are to use spermine and a Tris-HCl, NaCl buffer with Tween 20, followed by immediate accumulation of the spectra. At 632.9 nm, Cy5 gave the lowest detection limits, and at 514.5 nm, R6G is the best, although FAM, HEX, and TAMRA are not far behind. CONCLUSIONS Eight differently labeled oligonucleotides have been designed and all gave excellent SERRS with silver nanoparticles. The characteristics of the dye label affect the degree of SERRS produced, and it was found that the optimal conditions were of spermine with a Tris-HCl, sodium chloride buffer containing the
surfactant Tween 20. This allowed linear concentration-dependent graphs of SERRS intensity to be obtained over a range of biologically relevant concentrations. The quantitative nature of the response using the optimized conditions gave accurate limits of detection for SERRS detection of labeled oligonucleotides for the first time. This allowed an assessment of the labels for use in SERRS and provided detailed information on the optimal label, modification, and conditions. The information from this study will aid progress in the growing field of using SERRS for labeled oligonucleotide detection.
ACKNOWLEDGMENT The authors thank the DTI Measurements for Biotechnology program through their funding of the project. D.G. thanks the RSC for the award of their Analytical Grand Prix fellowship.
Received for review September 11, 2003. Accepted November 4, 2003. AC035060C
Analytical Chemistry, Vol. 76, No. 2, January 15, 2004
417
