The effects of temperature, salt concentration, and formamide concentration on the emission characteristics of commonly used fluorescent labels were evaluated on DNA microchips. The emission intensities of different fluorophores without hybridization were observed to vary, each to a different extent, to mainly temperature changes. Rhodamine red, TAMRA [tetramethylrhodamine], and dyes from the carbocyanide group exhibited the largest variations, and Texas Red and Oregon Green exhibited the smallest variations. This temperature dependency was shown to affect results obtained during melting curve analysis in DNA microarray studies. To minimize the bias associated with the temperature-dependent emission of different fluorescent labels, a normalization step was proposed.
Microarray technology using rRNA-based oligonucleotide probes has increasingly been used in microbial community analysis and identification [1, 7, 16]. One promising approach [referred to as melting curve analysis hereafter] to accurately discriminate perfect-match [PM] target-probe duplexes from mismatch [MM] non-target-probe duplexes is to compare nonequilibrium dissociation curves and the differences in dissociation temperatures [Tds], the temperature at which 50% of the starting duplex remained intact, between PM and MM duplexes [8, 9, 13, 14]. In brief, rRNA or rRNA gene targets are first fragmented and labeled with a fluorophore and then hybridized to planar or gel-based microchips containing oligonucleotide probes [∼15 to 25 nucleotides [nt]] with complementary sequences to targets. After hybridization and a brief washing, the microchip is covered with a transparent chamber containing the proper washing solution and placed on the top of a heating device underneath a microscopy-based imaging system [9, 14]. The microchip is subsequently subjected to thermal heating from low to high [i.e., from 10 to 80°C at a 1 to 2°C increment per min] to gradually denature the target-probe duplexes. By quantifying the respective fluorescent intensities of all spots on the array at every temperature, dissociation curves of all duplexes can be plotted based on the measured fluorescent intensities, and Tds of all duplexes can be calculated in a single experiment. This melting approach has shown promising results to correctly identify targets based on differences in Td and minimize false-positive signals in DNA microarray studies but will require further optimization and development before being extensively applied in microbial-community analysis.
Recent reports [4, 6, 12] suggested that different fluorescent dyes can exhibit different stabilities and intensities with respect to the changes of external parameters, such as solution temperature. This temperature dependency can further affect the hybridization intensity, dissociation curves, and Td measurement during melting curve analysis in microarray studies [5] but apparently has not been taken into consideration in past studies [2, 3, 9, 14]. To address this issue, the fluorescence characteristics of eight commonly used fluorophores [Table 1] were systematically investigated with respect to changes in temperature, salt concentration, and formamide concentration using gel-based microchips in this study.
Fluorophore excitation/emission characteristics and filters used in this study
| Oregon Green | 490 | 514 | 470/40 | 525/50 |
| Cy2 | 489 | 506 | 470/40 | 525/50 |
| FAM | 492 | 517 | 470/40 | 525/50 |
| FITC | 494 | 518 | 470/40 | 525/50 |
| Cy3 | 550 | 565 | 555/50 | 610/75 |
| TAMRA | 542 | 568 | 555/50 | 610/75 |
| Rhodamine red | 570 | 590 | 555/50 | 610/75 |
| Texas Red | 596 | 615 | 555/50 | 610/75 |
To perform DNA microarray studies, we selected a target DNA sequence [5′-AGTCGTAACAAGGTAGCCGT-3′, Escherichia coli positions 1492 to 1511] from a conserved region of the 16S rRNA gene for the domain Bacteria. The target [single-stranded DNA] was synthesized and labeled at the 5′ end with different fluorophores [Table 1]. In addition, four different oligonucleotide probes were designed specifically for this target sequence. They were PM probe BC [5′-[T]15-ACGGCTACCTTGTTACGACT-3′], probe BC_9G · G with a single internal MM [5′-[T]15-ACGGCTACGTTGTTACGACT-3′; the mismatch nucleotide is underlined], probe BC_9A · G with a single internal MM [5′-[T]15-ACGGCTACATTGTTACGACT-3′], and probe BC_9G · G10A · A with two internal MMs [5′-[T]15-ACGGCTACGATGTTACGACT-3′]. These four probes, which also contained a C-6 NH linker modification at the 5′ end for coupling onto microchips and a poly[T] spacer before the target sequence to increase on-chip accessibility for target DNA, were synthesized from MWG Biotech-AG [Ebersberg, Germany]. In addition, control probes [GGGG] with modifications of a fluorophore label at the 5′ end and a C-6 NH linker at the 3′ end [5′-linker-[T]15-GGGG-fluorophore-3′] were included as references on the arrays. The control and target probes were ordered from either QIAGEN [Cologne, Germany], IDT [Iowa], or Proligo [Paris, France], because the specification required could not be obtained from one source.
Furthermore, DNA microchips comprising polyacrylamide gel pad matrices with dimensions [width by length by depth] of 300 by 300 by 10 μm and a center-to-center spacing of 500 μm were fabricated as described previously [11]. Oligonucleotide probes [0.5 pmol] and control probes [0.1 pmol] were printed in four replicates onto individual gel pad elements using the Biochip arrayer [Packard, Perkin-Elmer]. Subsequently, the printed microchips were stabilized using pyridine-borane complex and treated with sodium borohydride solution to reduce background fluorescence. The microchips were washed in distilled water for 30 min, dried using a nitrogen stream, and preserved in a dark, dust-free environment before use.
The effects of temperature, salt concentration, and formamide concentration on the emission intensities of different fluorophore-labeled control probes were first investigated. The region immobilized with the control probes on the microchip was covered with a microchamber containing 50 μl of buffer solution at 4°C. The buffer solution contained 20 mM of Tris-HCl [pH 8.0], 5 mM of EDTA, and different concentrations of NaCl [10, 100, and 500 mM] and formamide [0, 10, 20, and 30%]. The assemblage was mounted onto a Peltier thermal stage under a microscopy-based imaging system [8] and heated stepwise from 7.5 to 80°C at an increment of 2.5°C per 3 min. Fluorescent images at the end of each holding interval were captured with an exposure time of approximately 800 ms under either a Cy3 filter [excitation/emission wavelength, 555/610 nm] or a green fluorescence protein [GFP] filter [470/525 nm] [Table 1] through an image acquisition and analysis program, LabArray [//www.eng.nus.edu.sg/civil/Labarray/labarray .htm] [10]. In brief, after capture of the first image at the starting temperature, all probe spots or areas of interest were identified and located with grids. The fluorescence intensities of individual areas of interest at each temperature were then quantified, subtracted from the background, and exported into a Microsoft Excel spreadsheet. Subsequently, a user could obtain the fluorescence intensity curves for individual probes with respect to temperature change after averaging the intensities of all replicates at each temperature. Since individual fluorophores were subjected to light exposure for less than 23 to 25 s, no significant photobleaching was observed with individual fluorophores [data not shown].
Figure 1 shows the effects of temperature and salt concentration on fluorophore stabilities on gel-based microchips. Texas Red, Oregon Green, and FAM [carboxyfluorescein] did not exhibit temperature-dependent characteristics within the range of temperature tested at a salt concentration of 500 mM. In contrast, the remaining five fluorophores were greatly affected by the increase in temperature, with a reduction in fluorescence intensity larger than 56 to 82% of the initial values measured at 7.5°C at a salt concentration of 500 mM. Furthermore, all fluorophores were observed to exhibit different dependencies to salt concentrations [10, 100, and 500 mM]. For Texas Red, rhodamine red, TAMRA [tetramethylrhodamine], and Cy3, the reductions in fluorescent intensities were lower at low salt concentrations than at high salt concentrations. For FAM, the reduction in fluorescent intensity was higher at a 10 mM salt concentration than at 100 mM and 500 mM salt concentrations only when the temperature was greater than 40°C. For fluorescein isothiocyanate [FITC], the reduction in fluorescence intensity was slightly higher at a 10 mM concentration than at high salt concentrations. Oregon Green and Cy2 exhibited irregular emission behavior to low salt concentrations. Changes in formamide concentrations [10, 20, and 30%] did not result in significant changes in the fluorescence signals of the fluorophore dyes [the standard deviation [SD] was less than ±5% of the intensity observed at individual temperatures] [data not shown].
Effects of salt concentration on fluorophore stabilities with respect to temperature on gel-based microchips. The formamide concentration in the buffer was fixed at 0%. [A] Cy3 filter [excitation, 555/50 nm; 610/75 nm]; [B] GFP filter [excitation, 470/40 nm; emission, 525/50 nm].
The effect of temperature on the fluorescent intensity of PM probe-target duplexes during melting curve analysis in microchip experiments was subsequently investigated. To carry out hybridization experiments, the probe array region was covered with a microchamber containing 25 μl of hybridization buffer [i.e., 0.5 nmol fluorophore-labeled target oligonucleotide, 0.9 M NaCl, 20 mM Tris-HCl [pH 8.0], and 40% formamide]. After hybridization at 20°C for 16 h, the microchips were briefly washed two to three times with 100 μl of washing buffer [20 mM Tris-HCl [pH 8.0], 5 mM EDTA, and 50 mM NaCl] at 4°C to remove unbound targets. Subsequently, the microchamber was loaded with 50 μl of washing buffer, followed by image acquisition steps and melting curve analysis [15 to 80°C] as described above. At the end of each experiment, the microchip was washed in distilled water for 1 h at 60°C in the dark to remove virtually all remaining bound targets. A gel pad microchip was reused up to 8 to 10 times without significant deterioration in fluorescence intensities. Further, melting curves based on averaged raw fluorescence intensities were generated for individual probe-target duplexes from an Excel file exported from the LabArray. To determine the Td, the fluorescent intensity of a probe-target duplex at the starting temperature was set to 1, and the temperature corresponding to 50% of the normalized intensity was identified as Td.
Figure 2A shows the melting curves of the PM duplex after it was hybridized with Texas Red- and Cy3-labeled oligonucleotide targets separately. The melting curves for those two duplexes were expressed as the percentage of raw fluorescent intensities observed at the starting temperature [i.e., 15°C]. Due to the temperature-associated behavior, both melting curves were distinctly different, with a difference in Td close to 15°C. The melting curve for the Texas Red-labeled target exhibited a typical S-shaped curve for DNA duplex dissociation, and that for the Cy3-labeled target showed a linear decrease. In order to obtain consistent melting curves for both types of fluorophore-labeled targets, a normalization step could be introduced by correcting the raw fluorescent intensities against the measured fluorescent intensities of a control probe attached with the same dye as that of the target [Fig. 2B]. To do so, raw fluorescent intensities of probes obtained from an experiment with a specific fluorescently labeled target were first normalized with the respective control probe labeled with the same fluorophore spots at each corresponding temperature using equation 1:
[1]
where ISC,t is the relative intensity of a test spot with respect to the control spot at temperature t, IS,t is the background-subtracted intensity of a test spot at temperature t, and IC,t is the background-subtracted intensity of a control spot at temperature t. This was followed by another normalization step using equation 2 to set the maximum intensities to 1 [for Td determination].
[2]
where IN,t is the normalized intensity of a test spot at temperature t, and ISC,max is the maximum relative intensity of a test spot in the entire dissociation test. Usually, the maximum intensity occurred at the starting temperature of a melting curve but occasionally could be observed at a higher temperature in this study and a previous report [14]. When this occurred, intensities obtained at individual temperatures were divided by the averaged value of the initial intensity and maximum intensity [14].
Melting curves of PM BC probe-target duplexes. [A] Texas Red- and Cy3-labeled targets without normalization [signal intensity at the initial temperature was set to 1]; [B] Texas Red-, Cy3-, rhodamine red-, and TAMRA-labeled targets normalized against control probes labeled with the same fluorophores. The formamide and salt concentrations in the wash buffer were fixed at 0% and 50 mM, respectively. The key in panel B indicates the fluorophore-labeled target or the type of fluorophore-labeled control probe used for normalization.
After the proposed normalization procedure, nearly identical S-shaped melting curves could be obtained, regardless of the type of fluorophores used [Fig. 2B]. As Texas Red was very stable to temperature changes [Fig. 1], melting curves of Texas Red-labeled target-probe duplexes with and without normalization against the Texas Red-labeled control were almost identical. Thus, the normalization step was not critical for real-time experiments conducted using Texas Red as the fluorescence reporter. During the early development of melting curve analyses on gel-based microchips, Texas Red was the first fluorophore used [9, 15]. This could explain why the temperature dependency issue related to fluorescence emission was not noted and investigated previously. However, Fig. 2A suggests that for experiments conducted using Cy3-labeled targets, melting curves plotted with and without correct normalization against the Cy3-labeled control probe exhibited distinctly different profiles due to the temperature-dependent feature of Cy3. We further tested rhodamine red- and TAMRA-labeled targets and obtained almost identical melting curves following the correct normalization procedure [Fig. 2B].
The temperature dependency effect of fluorophores on mismatch discrimination during melting curve analysis on microchips was further examined [Table 2]. Among the four fluorophores tested, the observed Tds for the PM duplexes higher than after correct normalization were 3.98 to 5.0°C [SD = 0.37 to 0.61°C] one-MM duplexes and 10.3 to 12.3°C [SD = 0.24 to 1.03°C] higher than the two-MM duplex. These results suggested that correct normalization was also necessary to achieve MM discrimination.
Tds of PM duplexes and ΔTds between MM duplexes and the PM duplex under different fluorophore labeling conditionsa
| Texas Red | 54.20 | 0.50 | −3.98 | 0.37 | −4.24 | 0.53 | −10.34 | 1.03 |
| Cy3 | 55.08 | 0.73 | −4.09 | 0.45 | −3.89 | 0.43 | −10.77 | 0.84 |
| TAMARA | 55.83 | 0.65 | −4.38 | 0.43 | −4.37 | 0.54 | −10.40 | 0.69 |
| Rhodamine red | 55.62 | 0.66 | −5.02 | 0.61 | −4.82 | 0.51 | −12.27 | 0.24 |
It was reported by Kotova and coworkers [5] that tetramethylrhodamine-isothiocyanate- and FITC-labeled targets exhibited similar melting curves but that the Texas Red-labeled targets showed melting curves that shifted toward higher temperatures for both the PM and MM duplexes. They suspected that Texas Red-labeled targets could have higher affinities to immobilized probes than the tetramethylrhodamine-isothiocyanate- and FITC-labeled oligonucleotides. In contrast, our results suggested that this difference in melting curves was likely due to the lack of a proper normalization step. This was because the relevant correction factors [equation 1] were not introduced to counter the effects of the differences in emitted fluorescence intensities among various fluorophores with respect to temperature changes.
Furthermore, the emitted fluorescence intensities-versus-temperature graphs were useful in facilitating the selection of fluorophore labels in future microchip experiments with dual-color labeling. The selection of fluorophore labels with similar emission characteristics with respect to changes in external experimental conditions, such as temperature, salt concentration, and formamide concentration, is preferable in order to minimize any dye-associated bias during the comparison of results from hybridization experiments employing dual-color labeling. When two fluorophore labels that possess different emission characteristics with respect to external working conditions are used, correction factors for individual labels can be obtained based on respective calibration curves similar to those in Fig. 1 and can be introduced to eliminate any dye-associated bias. It is also possible to use fluorophores with linear temperature dependency as temperature sensors in microfluidic devices where temperature monitoring by physical instruments is difficult [12].
In summary, we have demonstrated the changes in emission characteristics of various commonly used fluorescent labels with respect to important hybridization parameters [e.g., temperature and salt concentration]. Although the exact reasons for the difference in fluorescence emission characteristics among those fluorophores were not clear, the proposed normalization could facilitate a fair comparison among different dissociation curves obtained during melting curve analysis on DNA microchips.
1. Bodrossy, L., and A. Sessitsch. 2004. Oligonucleotide microarrays in microbial diagnostics. Curr. Opin. Microbiol. 7:245-254. [PubMed] [Google Scholar]
2. El Fantroussi, S., H. Urakawa, A. E. Bernhard, J. J. Kelly, P. A. Noble, H. Smidt, G. M. Yershov, and D. A. Stahl. 2003. Direct profiling of environmental microbial populations by thermal dissociation analysis of native rRNAs hybridized to oligonucleotide microarrays. Appl. Environ. Microbiol. 69:2377-2382. [PMC free article] [PubMed] [Google Scholar]
3. Fotin, A. V., A. L. Drobyshev, D. Y. Proudnikov, A. N. Perov, and A. D. Mirzabekov. 1998. Parallel thermodynamic analysis of duplexes on oligodeoxyribonucleotide microchips. Nucleic Acids Res. 26:1515-1521. [PMC free article] [PubMed] [Google Scholar]
4. Jeon, S., J. Turner, and S. Granick. 2003. Noncontact temperature measurement in microliter-sized volumes using fluorescent-labeled DNA oligomers. J. Am. Chem. Soc. 125:9908-9909. [PubMed] [Google Scholar]
5. Kotova, E. Y., E. Y. Kreindlin, V. E. Barsky, and A. D. Mirzabekov. 2000. Effects of various fluorochromes and competition between labeled oligonucleotides on their hybridization to oligonucleotides immobilized on biological microchips. Mol. Biol. 34:207-214. [Google Scholar]
6. Kotova, E. Y., E. Y. Kreindlin, V. E. Barsky, and A. D. Mirzabekov. 2000. Optical properties of fluorochromes promising for use in biological microchips. Mol. Biol. 34:266-271. [PubMed] [Google Scholar]
7. Li, E. S.-Y., and W.-T. Liu. 2004. DNA microarray technology in microbial ecology studies—principle, application and current limitations. Microbes Environ. 18:175-187. [Google Scholar]
8. Li, E. S.-Y., J. K. K. Ng, J.-H. Wu, and W.-T. Liu. 2004. Evaluating single-base-pair discriminating capability of planar oligonucleotide microchips using a non-equilibrium dissociation approach. Environ. Microbiol. 6:1197-1202. [PubMed] [Google Scholar]
9. Liu, W.-T., A. D. Mirzabekov, and D. A. Stahl. 2001. Optimization of an oligonucleotide microchip for microbial identification studies: a non-equilibrium dissociation approach. Environ. Microbiol. 3:619-629. [PubMed] [Google Scholar]
10. Ng, J. K., and W.-T. Liu. 2005. LabArray: real-time imaging and analytical tool for microarrays. Bioinformatics 21:689-690. [PubMed] [Google Scholar]
11. Proudnikov, D., E. Timofeev, and A. Mirzabekov. 1998. Immobilization of DNA in polyacrylamide gel for the manufacture of DNA and DNA-oligonucleotide microchips. Anal. Biochem. 259:34-41. [PubMed] [Google Scholar]
12. Ross, D., M. Gaitan, and L. E. Locascio. 2001. Temperature measurement in microfluidic systems using a temperature-dependent fluorescent dye. Anal. Chem. 73:4117-4123. [PubMed] [Google Scholar]
13. Urakawa, H., S. El Fantroussi, H. Smidt, J. C. Smoot, E. H. Tribou, J. J. Kelly, P. A. Noble, and D. A. Stahl. 2003. Optimization of single-base-pair mismatch discrimination in oligonucleotide microarrays. Appl. Environ. Microbiol. 69:2848-2856. [PMC free article] [PubMed] [Google Scholar]
14. Urakawa, H., P. A. Noble, S. El Fantroussi, J. J. Kelly, and D. A. Stahl. 2002. Single-base-pair discrimination of terminal mismatches by using oligonucleotide microarrays and neural network analyses. Appl. Environ. Microbiol. 68:235-244. [PMC free article] [PubMed] [Google Scholar]
15. Yershov, G., V. Barsky, A. Belgovskiy, E. Kirillov, E. Kreindlin, I. Ivanov, S. Parinov, D. Guschin, A. Drobishev, S. Dubiley, and A. Mirzabekov. 1996. DNA analysis and diagnostics on oligonucleotide microchips. Proc. Natl. Acad. Sci. USA 93:4913-4918. [PMC free article] [PubMed] [Google Scholar]
16. Zhou, J. H. 2003. Microarrays for bacterial detection and microbial community analysis. Curr. Opin. Microbiol. 6:288-294. [PubMed] [Google Scholar]