Characteristics of Different Nucleic Acid Staining Dyes for DNA

Anal. Chem. 1999, 71, 5470-5480

Articles

Characteristics of Different Nucleic Acid Staining Dyes for DNA Fragment Sizing by Flow Cytometry Xiaomei Yan,† W. Kevin Grace,‡ Thomas M. Yoshida,‡ Robert C. Habbersett,† Nileena Velappan,† James H. Jett,† Richard A. Keller,‡ and Babetta L. Marrone*,†

Life Sciences Division and Chemical Science and Technology Division, MS M888, Los Alamos National Laboratory, Los Alamos, New Mexico 87545

An efficient and reliable double-stranded DNA (dsDNA) staining protocol for DNA fragment sizing by flow cytometry is presented. The protocol employs 0.8 µΜ of PicoGreen to label a wide range of DNA concentrations (0.5 ng/mL to 10 000 ng/mL) without regard to the solution dye/bp ratios and without initial quantification of the DNA analyte concentration. Using a combination of spectrofluorometry and flow cytometry experiments, we found that PicoGreen exhibited better overall performance than all the tested dsDNA binding dyes, such as TOTO1. Fluorometric titration revealed that typical DNA staining protocols designed on the basis of the dye/bp ratio were highly dependent upon the DNA concentration for optimal results. PicoGreen was the least sensitive to the solution dye/bp ratio and was highly fluorescent in the presence of dsDNA. Using this new protocol, accurate histograms of HindIII digested λ DNA were demonstrated for DNA concentrations ranging from 5 to 2000 ng/mL, and for dye/bp ratios from 106:1 to 1:4 at 0.8 µΜ of PicoGreen. The new one-step protocol is broadly applicable to any sensitive, laser-induced fluorescence method for detection of nucleic acids. The sizing of DNA molecules is one of the most widely used analytical methodologies in molecular biology and biochemistry. Measurement of the length distribution of DNA molecules is typically accomplished using labor-intensive, slab-gel electrophoresis.1 However, in recent years, alternative technologies for more rapid DNA fragment sizing have emerged. Capillary electrophoresis has been demonstrated to achieve fast separation with better resolution, higher sensitivity, and the potential for automation.2-5 Atomic force microscopy (AFM)6 and MALDI-TOF mass spec* To whom correspondence should be addressed. Email: [email protected]. Fax: (505)-665-3024. † Life Sciences Division. ‡ Chemical Science and Technology Division. (1) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning, A Laboratory Manual, 2nd ed.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1989. (2) Landers, J. P., Ed. Handbook of Capillary Electrophoresis; CRC Press: Boca Raton, FL, 1994. (3) Righetti, P. G., Ed. Capillary Electrophoresis in Analytical Biotechnology; CRC Press: Boca Raton, FL, 1996.

5470 Analytical Chemistry, Vol. 71, No. 24, December 15, 1999

trometry7 are currently being evaluated for DNA fragment size measurement. Over the past several years, our group8-12 and other groups13-15 have developed an approach for DNA fragment sizing, which uses flow cytometry to detect single molecules of fluorescently labeled nucleic acids. In our method, bis-intercalating dye labeled DNA fragments are passed individually through a tightly focused, low power, continuous-wave laser beam, generating a burst of fluorescence photons. Quantification of total fluorescence emitted by each individual fragment, as it passes through the laser beam, gives the measure of the fragment size. Our rapid (3-10 min for a measurement) flow-cytometric-analysis approach needs less than 1 pg of DNA and is independent of the conformation of the doublestranded DNA (dsDNA) fragment.10 The ability to size individual DNA fragments using flow cytometry has been demonstrated from 212 base pairs (unpublished data) to as many as 351 kilo base pairs (kbp) in length.11 Currently, a major effort in our laboratory is the application of DNA fragment sizing by flow cytometry for the rapid identification of bacteria after whole genomic digestion with a rare-cutting restriction endonuclease.11,12 The success of DNA fragment sizing flow cytometry (DNA FSFC) relies upon precise, stoichiometric staining of the individual DNA molecules with fluorescent dyes. Recent developments in the synthesis of DNA-binding dyes have led to a new family of symmetric dimers of cyanine dyes (two intercalating chromophore (4) Madabhushi, R. S.; Vainer, M.; Dolnik, V.; Enad, S.; Barker, D. L.; Harris, D. W.; Mansfield, E. S. Electrophoresis 1997, 18, 104-111. (5) Kim, Y.; Morris, M. D. Anal. Chem. 1995, 67, 784-786. (6) Fang, Y.; Spisz, T. S.; Wiltshire, T.; D’Costa, N. P.; Bankman, I. N.; Reeves, R. H.; Hoh, J. H. Anal. Chem. 1998, 70, 2123-2129. (7) Ross, P. L.; Davis, P. A.; Belgrader, P. Anal. Chem. 1998, 70, 2067-2073. (8) Goodwin, P. M.; Johnson, M. E.; Martin, J. C.; Ambrose, W. P.; Marrone, B. L.; Jett, J. H.; Keller, R. A. Nucleic Acids Res. 1993, 21, 803-806. (9) Petty, J. T.; Johnson, M. E.; Goodwin, P. M.; Martin, J. C.; Jett, J. H.; Keller, R. A. Anal. Chem. 1995, 67, 1755-1761. (10) Huang, Z.; Petty, J. T.; O’Quinn, B.; Longmire, J. L.; Brown, N. C.; Jett, J. H.;Keller, R. A. Nucleic Acids Res. 1996, 24, 4202-4209. (11) Huang, Z.; Jett, J. H.; Keller, R. A. Cytometry 1999, 35, 169-175. (12) Kim, Y.; Jett, J. H.; Larson, E. J.; Penttila, J. R.; Marrone, B. L.; Keller, R. A. Cytometry, 1999, 36, 324-332. (13) Castro, A.; Fairfield, F. R.; Shera, E. B. Anal. Chem. 1993, 65, 849-852. (14) Schins, J. M.; Agronskaya, A.; de grooth, B. G.; Greve, J. Cytometry 1998, 32, 132-136. (15) Chou, H. P.; Spence, C.; Scherer, A.; Quake, S. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 11-13. 10.1021/ac990780y CCC: $18.00

© 1999 American Chemical Society Published on Web 11/04/1999

moieties bridged by a polyalkylamino linker with a double positive charge) with improved fluorescence properties (Molecular Probes, Inc.). A linker chain length of longer than 10 Å leads to bis-intercalation and, in general, the bis-intercalating derivatives cover four DNA base pairs.16 TOTO-1 and POPO-3 have been used to stain DNA for our flow-cytometric fragment sizing technique due to the compatibility of their excitation spectra with the argon ion laser (488 nm, 514 nm) and the frequency-doubled, diodepumped solid-state Nd/YAG laser (532 nm), respectively. These cyanine dimers have molar extinction coefficients greater than 105 cm-1 M-1 and exhibit 100- to 1000-fold enhancement of their fluorescence quantum yield upon binding to dsDNA.16-18 The dyes are very weakly fluorescent when they are free in solution, which eliminates the need to remove unbound dye from the solution before analysis. Current staining protocols use bis-intercalating dyes to stain the dsDNA according to a prescribed staining dye/bp ratio, typically 1:4 or 1:5.8-15 According to the neighbor-exclusion principle, every other intercalation site along the length of the DNA double helix remains unoccupied.16,19-21 Therefore, at a bound dye to base pair ratio of 1:4, the duplex should be essentially saturated with bis-intercalators. Lower ratios limit the number of dye molecules bound to a single DNA fragment, reducing the burst intensity from an individual fragment. Higher ratios may result in quenching of the fluorescence. However, for the practical application of DNA fragment-sizing flow cytometry for highthroughput sample analysis, we sought a DNA binding dye that could be used in a staining protocol without prior knowledge of the DNA concentration in the sample. We initially screened the nucleic acid stains, such as PicoGreen, TOTO-1, TO-PRO-1, POPO-3, PO-PRO-3, SYTO 16, SYTO 25, ethidium homodimer (EthD-1), and propidium iodide (PI) for their performance in a DNA-staining protocol for DNA FSFC. On the basis of this preliminary screening, the dyes PicoGreen and TOTO-1 were selected for the detailed comparisons presented here on the basis of their relative insensitivity to staining dye/bp ratios, low background fluorescence in solution, and exceptionally high fluorescence enhancement upon binding DNA. EXPERIMENTAL SECTION Chemicals. The dyes used in this study were all purchased from Molecular Probes, Inc. (Eugene, OR). All names of the dyes, except EthD-1 and PI are registered trademarks of Molecular Probes, Inc. KpnI endonuclease restriction enzyme was purchased from New England Biolabs (Beverly, MA) and used under conditions specified by the vendor. Commercially obtained chemicals of ACS purity were used for sample and reagent preparation in ultrapure water (Millipore, Bedford, MA). Sample Preparation. Bacteriophage λ DNA, 48 502 bp in length, and λ DNA/HindIII digest (Promega Co., Madison, WI) were stored at -20 °C. The stock DNA concentration was verified (16) Rye, H. S.; Quesada, M. A.; Peck, K.; Mathies, R. A.; Glazer, A. N. Nucleic Acids Res. 1991, 19, 327-333. (17) Glazer, A. N.; Rye, H. S. Nature (London) 1992, 359, 859-861. (18) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals, 6th edition; Molecular Probes, Inc.: Eugene, OR, 1996. (19) Leman, L. S. J. Mol. Biol. 1961, 3, 18-30. (20) Crothers, D. M. Biopolymers 1968, 6, 575-584. (21) Gao, Q.; Williams, L. D.; Egli, M.; Rabinovich, D.; Chen, SL.; Quigley, G. J.; Rich, A. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 2422-2426.

by absorption measurement at 260 nm on a Beckman DU 640 spectrophotometer (Fullerton, CA). The stock concentration of PicoGreen was calculated to be 0.32 mM.22 Mixtures of DNA with dye at various dye/bp ratios were prepared in TE (10 mM TrisHCL, 1 mM EDTA, pH 8.0) buffer under reduced illumination at room temperature. All diluted samples of DNA or dye solutions were also prepared in this buffer. Since most of the dyes are susceptible to photodegradation, the dye working solution and the DNA-dye complexes were either placed in the dark or covered to protect them from light. The order of the addition of the bis-intercalating dye and DNA has been reported to be critical for obtaining reproducible, wellresolved DNA fragment patterns on gels.23 In the present study, the DNA sample and the dye stock were diluted individually to an appropriate concentration first. DNA-dye complex was prepared in the following order: (1) dye was added to the vial; (2) an appropriate amount of TE buffers was added and mixed; and (3) DNA was added. The DNA-dye solutions were incubated for a minimum of 30 min prior to any measurement. The dye to base pair ratio calculation was based on a base pair molecular weight of 660. Fluorescence Spectrometry. Fluorescence measurements were performed on a SPEX Fluorolog-2 spectrofluorometer with double grating and 1.70 nm/mm dispersion. Because of the high extinction coefficients of DNA-bound TOTO-1 (117 000 cm-1 M-1)18 and PicoGreen (70 000 cm-1 M-1 in phosphate buffer),22 concentrations in excess of 1 µM can give rise to inner-filter effects using the typical 7-mm round glass cuvette (200-µL sample volume, Sienco. Inc., Wheat Ridge, CO). Therefore, for the titration studies over a wide range of dye/bp ratios, a capillary tube (0.5mm diameter, 9-µL full volume) along with the capillary cuvette adapter kit (Pharmacia Biotech, San Francisco, CA) was used. The capillary cuvette can reach sensitivity as high as 70% of that obtained with a 7-mm glass cuvette at fluorometer slits width of 1.0 mm. Samples containing TOTO-1 and PicoGreen were excited at 514 and 496 nm, respectively. Unless otherwise specified, the fluorescence data presented in this paper have been corrected for the wavelength-dependent response of the xenon lamp, excitation monochromator, and emission spectrometer. Therefore, the direct comparison of relative fluorescence within one complete data set is possible. DNA Fragment-Sizing Flow Cytometry. The DNA FSFC instrument is similar to that described previously,9,11 except an air-cooled argon ion laser (488 nm, 30 mW, Uniphase, model no. 2014-30SL, San Jose, CA) was used to provide the excitation light. The laser was operated at 10 mW CW and focused by one Planoconvex lens onto the hydrodynamically focused sample stream to yield a calculated spot size of approximately 44 µm diameter (1/e2). Sheath fluid (ultrapure water) was introduced into a 250 × 250 µm2 square-bore flow cell by gravity feed. Sample solution was delivered into the sheath solution through a 40-µm i.d., 240µm o.d. quartz capillary (Polymicro Technologies, Phoenix, AZ) with a ground 9° taper (New Objective, Inc., Cambridge, MA) to ensure smooth laminar flow of the sheath fluid around the end of the capillary. The focused sample stream diameter was ap(22) Singer, V. L.; Jones, L. J.; Yue, S. T. Haugland, R. P. Anal. Biochem. 1997, 249, 228-238. (23) Rye, H. S.; Yue, S.; Wemmer, D. E.; Quesada, M. A.; Haugland, R. P.; Mathies. R.A.; Glazer, A. N. Nucleic Acids Res. 1992, 20, 2803-2812.

Analytical Chemistry, Vol. 71, No. 24, December 15, 1999

5471

Table 1. Comparison of Selected Nucleic Acid Dye Properties

a

dye (efficiently excitable laser lines, nm)a

fluorescence enhancementb

relative tolerance of dye/bp ratio

relative stability of dilution test

PicoGreen (496, 488) TOTO-1 (514, 496, 488) TO-PRO-1 (514, 496, 488) POPO-3 (532, 514) PO-PRO-3 (532, 514) EthD-1 (514,532) PI (532, 514) SYTO 16 (488, 496) SYTO 25 (514, 532)

+++ +++ ++ ++ + + + ++ ++

+++ ++ +++ + + ++ ++ ++ +++

+++ +++ ++ +++ ++ +++ ++ ++ ++

Wavelengths are listed according to excitation efficiency. b Above 1000 (+++), above 100 (++), above 10 (+).

proximately 16 µm. The laser probe volume was calculated to be about 9 pL, defined by the overlap of the laser beam and the sample stream. N2 gas and a precision pressure regulator were employed to control the sample-introduction rate. The transit time of the sample through the detection probe volume was controlled by the sheath flow (∼55 µL/min) to produce a linear velocity of ∼15 mm/s at the center of the cuvette, corresponding to a 3-ms transit time through the laser beam. The sample volumetric flow rate was ∼5 nL/min, corresponding to an analysis rate of ∼85 DNA fragments/s. The fluorescence was collected perpendicular to both the laser excitation and the flow axes with a 40 × 0.85 NA microscope objective (Nikon Inc., Melville, NY) and spatially filtered by a 0.8 (horizontal) × 3.0 (vertical) mm slit. The slit defines an approximate detection area of 20 µm × 75 µm in the sample stream. The collected light was then spectrally filtered by a 530DF30 interference filter (Omega Optical, Brattleboro, VT). The transmitted fluorescence was focused by an 8-mm focal length lens onto a single-photon-counting, avalanche photodiode detector (SPCM-AQ-121, EG&G Canada, Vaudreuil, Canada). The photoelectron pulses were counted on a multichannel scaler (MCS, Oxford Instruments, Oak Ridge, TN). The MCS summed the number of pulses in 50-µs-wide bins. Two to four hundred scans of data (8192 bins per scan) were collected and transferred to a personal computer and analyzed by a program written in IDL (Interactive Data Language, Research Systems, Inc., Boulder, CO). For the data analysis, a burst was counted when a series of bins exceeded a count threshold set above the average background over a specified time period. A typical background rate was 3 photoelectrons (PE) per MCS bin, and the typical threshold was chosen to be 6 photoelectrons per bin. The area of each burst identified was integrated and histogrammed to give a burst size distribution. Histograms were fit to a sum of Gaussians. Burstsize means (centroids of histogram peaks), determined by the fit, were plotted versus the fragment lengths and fit by linear regression. Microspheres (Fluoresbrite YG, 0.952 ( 0.016 µm diameter, cat no. 18860, Polysciences Inc., Warrington, PA) were used to align the optics. RESULTS AND DISCUSSION Table 1 compares the selected properties of nine commonly used nucleic acid stains in terms of fluorescence enhancement, relative dye/bp ratio tolerance, and relative dilution stability. Clearly, PicoGreen exhibits the best overall performance, followed by TOTO-1, TO-PRO-1, and SYTO 25. For the two most often used 5472 Analytical Chemistry, Vol. 71, No. 24, December 15, 1999

Figure 1. Absorption and corrected fluorescence emission spectra of TOTO-1 (a) and PicoGreen (b) in the absence (broken line) or presence (solid line) of λ DNA at a dye/bp ratio of 1:5. The concentration of DNA is 1.32 µg/mL (2 × 10-6 M bp) for both absorption and fluorescence measurements. All spectra were normalized to facilitate comparison.

intercalating dyes in our lab for DNA FSFC, TOTO-1 is relatively sensitive to the staining dye/bp ratios compared with PicoGreen. For POPO-3, even more limited dye/bp ratios compared with TOTO-1 were observed. The relatively intense fluorescence background of unbound POPO-3 and PO-PRO-3 prohibited a titration study at higher dye concentrations. The DNA-dye complexes formed from all those tested dyes are stable upon dilution in the time range of hours, and dimeric dyes have higher binding affinity than their monomeric counterparts. The results presented in this paper focus on the comparison between TOTO-1 and PicoGreen regarding absorption and emission spectra; dye/bp ratio effects; DNA concentration effects; stability of the DNA binding complex; and potential for one-step staining of DNA fragments for accurate sizing by flow cytometry. Measurement of Absorption and Fluorescence Emission Spectra. The absorption and corrected fluorescence emission spectra for TOTO-1 and PicoGreen in the absence or presence of

Figure 2. Fluorometric titration of three different concentrations (given in bp and ng/mL) of λ phage DNA with TOTO-1 (a, c, and e) and PicoGreen (b, d, and f) versus dye/bp ratio. Triangles represent the measured total fluorescence at each dye/bp ratio. Open circles represent the free-dye fluorescence in the absence of DNA at the corresponding dye amount. Measurements were done using capillary cuvettes.

dsDNA are shown in Figure 1. The spectra were obtained by incubating 1.32 µg/mL of λ DNA with each dye at a 1:5 dye/bp ratio. The fluorescence emission spectra in Figure 1 demonstrate the highly enhanced fluorescence when these dyes bind to DNA (about 4700-fold for PicoGreen and 3400-fold for TOTO-1) and their negligible, intrinsic, free-dye-fluorescence background. For both TOTO-1 and PicoGreen, red shifts and hyperchromaticities reflect changes in the absorption spectra of bound dye, either resulting from the formation of a different dye conformation or a change in the micro-environment (hydrophobic effect) of dye molecules. The magnitudes of these changes were different for the two dyes. TOTO-1 produced a more pronounced shift in maximum absorption wavelength upon binding with DNA (from 480 to 514 nm) and a larger degree in hyperchromaticity around 514 nm. For PicoGreen, there was a small red shift of the peak absorption (from 498 nm for the free dye to 500 nm for the bound dye). The complexes of PicoGreen or TOTO-1 with dsDNA are spectrally compatible with commercially available air-cooled argonion lasers (488 nm). The nonradiative decay of photoexcited cyanine dyes is believed to be controlled by the rate of rotation or torsion about the central methine bridge which links the heterocycles at either

end.24,25 When TOTO-1 binds to dsDNA, the benzothiazole ring and the quinoline ring are believed to be clamped by the nucleic acid bases, preventing this rotation and resulting in a decrease in radiationless decay and an increase in the fluorescence quantum yield. The structure of PicoGreen is proprietary, and the mode of binding to dsDNA has not been fully characterized. Optimal Dye/bp Ratio Study. Since the initial DNA concentration can vary widely for the nucleic acid samples prepared for DNA FSFC, a fluorometric titration study over a wide range of DNA concentrations was performed. Figure 2 shows the fluorometric titration, using a capillary cuvette, of phage λ DNA with TOTO-1 and PicoGreen at three different DNA concentrations (2 × 10-6 M bp, 2 × 10-7 M bp, and 2 × 10-8 M bp) over a staining dye/bp ratio of 1:50 to 100:1. The background levels shown in parts e and f of Figure 2 were due to the capillary-scattering effect. The fluorescence response versus dye/bp ratio was highly dependent on the DNA concentration. The higher the DNA concentration, the narrower the range of useful dye/bp ratios with (24) Netzel, T. L.; Nafisi, K.; Zhao, M.; Lenhard, J. R.; Johnson, I. J. Phys. Chem. 1995, 99, 17936-17947. (25) Spielmann, H. P.; Wemmer, D. E.; Jacobsen, J. P. Biochemistry 1995, 34, 8542-8553.


5473

the maximum fluorescence observed at a lower dye/bp ratio. A titration study on KpnI digested λ DNA showed exactly the same pattern. As shown in Figure 2a, titration of λ DNA at a relatively high concentration (2 × 10-6 M bp, 1320 ng/mL) resulted in a total loss of TOTO-1 fluorescence when the dye/bp ratio approached 1:1. The transition from the maximum signal to total quenching was very abrupt. For a DNA concentration 10-fold less (Figure 2c), the fluorescence intensity was maximum at a dye/bp ratio of 1:1. The fluorescence remained high at 2:1 and then declined very quickly to the baseline at 5:1. For the DNA concentration of another 10-fold less (Figure 2e), the acceptable range of dye/bp ratio extended to 10:1, with the optimal dye/bp ratio observed at 2:1 to 5:1. Compared with TOTO-1, PicoGreen was far less sensitive to the dye/bp ratio. A useful signal was achieved to a dye/bp ratio as high as 10:1 at a DNA concentration of 2 × 10-6 M bp. No testing was made above a dye concentration of 20 µM. At a DNA concentration of 2 × 10-7 M bp, the PicoGreen fluorescence reached a plateau after 1:1 and did not noticeably decrease until it reached a dye/bp ratio of 50:1 (Figure 2d). For the relatively low DNA concentration (2 × 10-8 M bp), the trend followed that shown in Figure 2d, but with a more gradual transition at very high dye/bp ratios. It is evident that although the commonly cited 1:4 or 1:5 dye/bp ratios in the literature are suitable for higher DNA concentrations, such as 2 × 10-6 M bp, these ratios are not optimal for lower DNA concentrations. A trend similar to that shown in Figure 2 has been reported before. Ray et al.26 observed that for both TOTO- and YOYOdsDNA complexes, the dye/bp ratio approaching binding-site saturation was dependent on the DNA concentration. Rye et al.23 also reported that a lower dye/bp ratio allowed them to stain higher DNA concentrations with TOTO-1, YOYO-1, and the Ethidium dyes for analysis by gel electrophoresis. Binding Between DNA Base Pairs and Dye. The standard fluorescent dyes such as ethidium bromide and acridine orange are known to bind to DNA by intercalation between base pairs of dsDNA and by electrostatic interaction with the phosphate groups on the DNA backbone.27-29 NMR studies of the interactions of TOTO-1 with a double-stranded 8-mer indicate that it is bisintercalated with a local distortion of the DNA structure in the intercalating region.25 The same researchers also reported that TOTO-1 exhibits strong sequence selectivity for the site CTAG.30,31 From molecular mechanical simulations on bis-intercalation of TOTO-1 into dsDNA, Rao and Kollman concluded that vibrational entropy considerations could play an important role in excluding structures where intercalators occupy locations between adjacent base pairs.32 TOTO-1 has been referred to as a bis-intercalator,17,23 but there is also evidence indicating that electrostatic interaction and groove association are also involved in TOTO-1 binding to (26) Rye, H. S.; Dabora, J. M.; Quesada, M. A.; Mathies, R. A. Glazer, A. N. Anal. Biochem. 1993, 208, 144-150. (27) Waring, M. J. J. Mol. Biol. 1965, 13, 269-282. (28) Lepecq, J. B.; Paoletti, C. J. Mol. Biol. 1967, 27, 87-106. (29) Krugh, T. R.; Reinhardt, D. G. J. Mol. Biol. 1975, 97, 133-162. (30) Jacobsen, J. P.; Pedersen, J. B.; Hansen, L. F.; Wemmer, D. E. Nucleic Acids Res. 1995, 23, 753-760. (31) Hansen, L. F.; Jensen, L. K.; Jacobsen, J. P. Nucleic Acids Res. 1996, 24, 859-867. (32) Rao, S. N.; Kollman, P. A. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 57355739.

5474


DNA. The structure of the hole-burned and the fluorescence linenarrowed spectra indicate the contribution of two binding modes of TO-PRO-3 and TOTO-3 with DNA: externally bound to the DNA chain and intercalation or base-stacked configurations.33 Netzel et al.24 reported that the dimeric DNA stains show either bi- or triexponential fluorescence decay kinetics reflecting multiple modes of dye/dsDNA association. YOYO-1 was found to exhibit at least two distinct binding modes: at low dye/bp ratios, the binding mode appears to consist primarily of intercalation, while at high dye/bp ratios a second mode involving the external binding begins to contribute, and the binding constant for this mode was smaller than the binding constant for intercalation.34,35 The results shown in Figure 2 indicate there is a continuous shift of the optimal dye/bp ratio to higher values and the useful dye/bp ratio range increases with a decrease of the DNA concentration. This interesting trend led us to simulate the DNA site binding efficiency on the basis of the measured dissociation constant of TOTO-1 (1.1 × 10-9 M) in TE.36 Assuming there is only bis-intercalation binding and one dye molecule (A) covering four base pairs (named one DNA binding site, B), binding data can be analyzed in terms of a single, reversible equilibrium between A and B and the DNA-dye complex (C)

A+BSC

(1)

This equilibrium can be described by dissociation constant

KD ) [A]f[B]f/[C]

(2)

where [A]f is free dye concentration, [B]f is free DNA binding site concentration, and [C] is complex concentration. Assuming [C] ) C, eq 2 can be expressed as

KD )

(A - C)(B - C) C

(3)

where A is total dye concentration and B is total DNA binding site concentration (total DNA base pair concentration divided by 4). The DNA site binding efficiency is defined as C/B, and the free dye concentration can be obtained from A - C. According to eq 3, we can simulate the DNA site binding efficiency (Figure 3a) and the free dye concentration (Figure 3b) at any given DNA concentration and staining dye/bp ratio for TOTO-1. The general trend shown in Figure 3a reveals that at very low DNA concentrations, saturation of the DNA binding sites requires higher staining dye/bp ratios. As the DNA concentration is increased, lower dye: bp ratios (such as 1:5 or 1:2) are able to reach maximum DNA binding efficiency. The trend displayed in Figure 3a agrees well with the continuous shift of the optimal dye/bp ratio to lower values with the increase of the DNA concentrations shown in Figure 2 (a, c, e) for TOTO-1. For example, for a DNA concentration of 13.2 ng/mL, the calculated DNA site binding efficiencies (33) Milanovich, N.; Suh, M.; Jankowiak, R.; Small, G. J.; Hayes, J. M. J. Phys. Chem. 1996, 100, 9181-9186. (34) Larsson, A.; Carlsson, C.; Jonsson, M.; Albinsson, N. J. Am. Chem. Soc. 1994, 116, 8459-8465. (35) Carlsson, C.; Larsson, A.; Jonsson, M.; Albinsson, B.; Norden, B. J. Phys. Chem. 1994, 98, 10313-10321. (36) Yan, X. et. al, unpublished data, paper in preparation.

Figure 3. Simulated DNA binding efficiency (a) and free-dye concentration (b) versus TOTO-1 staining dye/bp ratio at five different DNA concentrations. A dissociation constant KD of 1.1 nM was used.

at staining dye/bp ratios of 2:1 and 5:1 are 97.0% and 98.9%, respectively. Experimental data in Figure 2e shows fluorescence reaching a maximum at both 2:1 and 5:1 dye/bp ratios. For a higher DNA concentration of 1320 ng/mL, the calculated DNA site binding efficiencies at staining dye/bp ratios of 1:5 and 1:2 are 79.2 and 99.8%, respectively, which agree with the experimental data in Figure 2a. If quenching could be eliminated experimentally, the titration curve displayed in Figure 2a should reach a plateau beyond a dye/bp ratio of 1:2, as was simulated in Figure 3a. In considering our experimental results with TOTO-1 and PicoGreen, most likely two types of binding modes coexist. The intense fluorescence corresponds to bis-intercalation with a very high binding affinity. The second mode probably involves external binding resulting in quenched fluorescence. Since the secondary binding sites usually have a lower binding affinity (higher dissociation constant) than that associated with the primary (intercalation) sites, their population requires a higher free-dye concentration. According to the simulation given in Figure 3b, the free-dye concentrations corresponding to the totally quenched signals at 1:1 in Figure 2a and 5:1 in Figure 2c are 1500 and 950 nM, respectively. These concentrations may be considerably higher than the dissociation constant of second-mode binding, resulting in fully quenched fluorescence. The high apparent tolerance of PicoGreen to the higher free dye concentrations suggests that the secondary dissociation constant of PicoGreen

is larger than that of TOTO-1. Though the molecular structure of PicoGreen is proprietary, PicoGreen is apparently a substituted unsymmetrical cyanine monomer and belongs to a new class of nucleic acid stains in which specific substituents have been introduced to particular positions on the unsymmetrical cyanine core molecule.37 Deduced from Table 1, it seems that more tolerance to the higher dye/bp ratios is a general trend for monointercalating dyes. Therefore, we speculate that the sensitivity of the nucleic acid labeling dye to the staining dye/bp ratio may be primarily dependent on the dissociation constant of its secondary binding mode. The higher the second KD, the greater the useful range of dye/bp ratios. Effect of Dye/bp Ratio on DNA Fragment Sizing by Flow Cytometry. Although 1:4 or 1:5 dye/bp ratios are typically used as the DNA staining protocol for DNA fragment sizing by flow cytometry and other methods,38,39 a detailed study of the effect of varying the staining dye/bp ratio on the fluorescence intensity from individual fragments has not been reported before. Figure 4 shows the size distribution histograms of TOTO-1 (a) and PicoGreen (b) labeled KpnI digested λ DNA (3 × 10-8 nM bp, 20 (37) Singer, V. L.; Jin, X.; Jones, L.; Yue, S.; Haugland, R. P. Biotechnology International, Volume I.; Universal Medical Press: San Francisco, CA, 1997; 267-276. (38) Gurrieri, S.; Wells, K. S.; Johnson, I. D.; Bustamante, C. Anal. Biochem. 1997, 249, 44-53. (39) Lyon, W. A.; Nie, S. Anal. Chem. 1997, 69, 3400-3405.


5475

Figure 4. Histograms of the fluorescence burst sizes of a KpnI digest of λ DNA stained with TOTO-1 (a) and PicoGreen (b) at several different dye/bp ratios indicated in each panel. A DNA concentration of 20 ng/mL (30 nM bp) was used for all dye/bp ratios presented. The bin width was 10 photoelectrons (PE) per bin. The actual fragment sizes are shown in (a), the second panel from the top.

ng/mL) at several different dye/bp ratios. Each set of experiments was conducted from low to high dye/bp ratios. Figure 4 shows that good histograms were obtained at all the tested dye/bp ratios of PicoGreen, whereas for TOTO-1, the fluorescence was totally quenched at a dye/bp ratio of 50:1. For the TOTO-1 sample stained at a 1:5 dye/bp ratio, a poor histogram was obtained. However, if the sample is stained at a higher DNA concentration (e.g., 300 ng/mL) with the same 1:5 dye/bp ratio and then diluted prior to running on the instrument, good fragment-sizing data was obtained, which agrees with previous reports.8-12 From the standpoint of sensitivity, at the present instrument setup, PicoGreen had a weaker signal than TOTO-1 at the lower dye/bp ratios. The first peak (1503 bp) in each of the 1:5 and 1:1 PicoGreen stained samples was not resolved from the background. As the dye/bp ratio was increased, the burst sizes from individual fragments increased, revealing smaller fragments that were easily distinguished from the background. These results suggest that the dye/bp ratio is more critical for DNA fragment sizing by flow cytometry, in which single DNA molecules are measured, than for capillary electrophoresis in which a bulk segment of identical-length fragments are measured. The dye/ bp ratio is especially critical for measurement of small ( 1:1. Bear in mind that the basis for the flowcytometric approach to DNA fragment sizing depends on the direct relationship between the amount of dye molecules bound to individual DNA fragments and the DNA fragment length. CONCLUSION We have developed an easy DNA staining protocol that allows us to eliminate prior quantification of the DNA concentration in each sample and to stain a wide range of DNA concentrations using 0.8 µM of PicoGreen without encountering nonstoichiometric dye-binding effects. We have determined that PicoGreen is an excellent fluorophore for the staining of dsDNA, allocating

Figure 8. Correlation of the centroids of the burst sizes obtained from the fitted Gaussian curve (gray line) with known DNA fragment lengths. 50 ng/mL of λ/HindIII digest was stained with 0.8 µM of PicoGreen.


5479

accurate size measurement by DNA FSFC. The staining is not influenced by the dye/bp ratio over a broad range of DNA concentrations, though the PicoGreen staining is somewhat less sensitive than the TOTO-1 staining for the present DNA FSFC instrument. Simulations based on the bis-intercalation dissociation constant of TOTO-1 support the interpretation of our results. Most likely there are two types of binding modes that coexist. The intense fluorescence corresponds to the bis-intercalation mode, with a very high binding affinity. The secondary quenching mode may involve external binding with lower binding affinity. These two modes may also have different on- or off-rates, which determine the time domain of the occurrence from each mode. Preliminary experimentation shows that addition of salt reversed the quenching effect. More detailed studies are needed to investigate the quenching mechanism and related kinetics. A onestep staining protocol using 0.8 µM PicoGreen is currently used in our laboratory for the high-throughput size analysis of DNA fragments from restriction enzyme digests of bacteria. The onestep PicoGreen staining protocol also works well for size analysis of Polymerase Chain Reaction (PCR) and Amplified Fragment

5480


Length Polymorphism (AFLP) products using DNA FSFC and can be performed without purification of the DNA from other reaction constituents. Moreover, our results using PicoGreen for DNA FSFC should be applicable to the development of improved DNA staining protocols for other fluorescence-based, DNA fragment-sizing methods. ACKNOWLEDGMENT This work was supported by the Chemical and Biological Nonproliferation Program (NN-20) of the Department of Energy, by the NIH-funded National Flow Cytometry Resource (PR-01315), and by the FBI Hazardous Materials Response Unit, for which we are most grateful. The statements and conclusions herein are those of the authors and do not necessarily represent the views of the FBI. We thank Dr. John P. Nolan for many helpful discussions. Received for review July 15, 1999. Accepted September 30, 1999. AC990780Y

Characteristics of Different Nucleic Acid Staining Dyes for DNA

Recommend Documents