Hoechst 33258, DAPI, and Vybrant DyeCycle Violet are common fluorescent dyes used for staining DNA and visualization of chromatin in cell nuclei by fluorescence and confocal microscopy. DNA-bound Hoechst 33258, DAPI, and DyeCycle Violet have their excitation maxima in the UV region of the spectrum and their emission maxima in the 430–470 nm region (exc./em. 355/465; 364/454, 369/437 nm, respectively). These spectral properties make Hoechst 33258, DAPI, and Vybrant DyeCycle Violet convenient nuclear dyes as the remaining range of the visible spectrum can be used for detecting other subcellular targets, using dyes emitting in green, yellow, up to infrared. We demonstrate, however, that the UV-excited dyes are not entirely stable when excited with the UV emitted by mercury arc lamps or 405 nm wavelength light emitted by a laser. Exposure to ultraviolet light of a biological specimen stained with Hoechst 33258, DAPI or Vybrant DyeCycle™ Violet during a typical fluorescence or confocal microscopy observation leads to conversion of these dyes into forms that can be excited by blue light and emit green fluorescence (em. max. approx. 520 nm, with detectable signal intensities in the yellow to orange region of the spectrum). Using mass spectrometry we demonstrate that UV induces protonation of Hoechst 33258 and DAPI. Photoconversion of UV-excited dyes bound to DNA in biological samples may pose problems for image analysis in multicolor fluorescence microscopy. Even relatively small doses of UV result in creating a false positive—a green fluorescence signal derived from the DNA-bound dye, which is expected to fluoresce in the blue region. Following photoconversion, the unexpected emission arising from Hoechst or DAPI may be mistakenly interpreted as a green fluorescent signal expected to arise from another fluorescent probe used in the same experiment. Problems arising from light-induced changes of the spectral properties of DAPI and Hoechst have been reported by the subscribers of the confocal microscopy list server and in recent published work (1).
Materials and Methods
MSU 1.1 human fibroblasts (2) were grown in Dulbecco's Modified Eagles Medium (Sigma-Aldrich, Poland) supplemented with 10% fetal bovine serum, penicillin (50 units/ml), and streptomycin (50 μg/ml), in tissue culture Petri dishes (Techno Plastic Products AG, Switzerland) at 37°C, in a humidified atmosphere of 95% air and 5% CO2. Cells were grown on 20 mm-diameter coverslips (Menzel-Gläser, Germany).
Cells were fixed with 4% formaldehyde (Electron Microscopy Science) using standard formaldehyde fixation protocol and later stored in 1% formaldehyde solution for 1–2 weeks before the experiment.
Cells were stained with DAPI (4′,6-diamidino-2-phenylindole) (3, 4) (Sigma-Aldrich, Poland) at 1 μM, Hoechst 33258 (5–7) (Sigma-Aldrich, Poland) at 2 μg/ml, or Vybrant DyeCycle™ Violet (Invitrogen) at 1 μM. Cells were rinsed three times with PBS, permeabilized with 70% ethanol (30 s) and incubated in a dye solution for 30 minutes at room temperature (RT).
Acquisition of Fluorescence Images
Images were recorded using Leica TCS SP5 confocal microscope (Leica Microsystems, Germany), equipped with a 63 × 1.4 NA oil immersion lens. The 512 × 512 pixel images (field of view 82 × 82 μm or 145 × 145 μm) were recorded. PMT gain was set at 860 V; confocal iris was set at 1 Airy disk. When studying the process of UV-excited dye photoconversion, coverslips with the attached fixed cells were mounted in custom made steel holders, placed in a microscope stage and imaged at RT or 37°C in PBS (fixed cells). For excitation Ar ion gas (458, 476, 488, 496, 514 nm, 100 mW output) and 405 nm diode (3 mW output) lasers were used. For imaging DAPI and Hoechst photoproducts, the intensity of all Ar laser lines was adjusted to 1.25 mW or 0.95 mW. Fluorescent dyes were photoconverted using UV light emitted by a Leica EL6000 mercury metal halide lamp passed through a 360/40 nm emission filter (11 mW, measured with a FieldMaxII Laser Power Meter [Coherent, Santa Clara, CA]). Over 40 experiments were performed to demonstrate the process of DAPI photoconversion and over 60 for Hoechst 33258.
Studies of Hoechst 33528 and DAPI Immobilized in a Polymer Block
To immobilize the dye, a polyvinyl polymer [described in (8)] doped with DAPI at the concentration of 10 μM, or Hoechst 33258 at the concentration of 20 μg/ml, was made.
Images were analyzed and processed using LAS AF Lite (Leica Microsystems, Germany) and MacBiophotonics ImageJ (http://rsbweb.nih.gov/ij/) software. The displayed images were not manipulated beyond adjusting the γ-function, as noted in figure legends.
For mass spectrometry (MS) measurements the solutions of DAPI at 70 μg/ml and Hoechst 33258 at 100 μg/ml were used. Protonation products were detected in the solutions of the dyes following addition of H2O2 (4% final concentration). Exposure to the UV was performed on a microscope stage of the confocal system, by illuminating the samples with the light emitted from a Leica EL6000 mercury metal halide lamp equipped with a 360/40 nm filter. Measurements were performed with the MicrOTOF-QII mass spectrometer (Bruker, Germany) in a positive ionization mode, using an Appollo Source ESI sprayer. Before measurements the device was calibrated with TuneMix solution. The MS spectra were analyzed using Data Analysis 4.0 software (Bruker, Germany).
DAPI or Hoechst 33258 (10 μM) were dissolved in distilled water or in a 30% hydrogen peroxide solution (Sigma-Aldrich, Poland). Samples were placed in quartz cuvettes and analyzed using Horiba Jobin Yvon Fluoromax-P spectrofluorimeter.
UV-Induced Changes of Spectral Characteristics of Hoechst 33258 and DAPI
We first examined spectral properties of the Hoechst 33258 and DAPI samples used in this work. The free dyes in solutions as well as the dyes bound to DNA were investigated by spectrofluorimetry and spectrally resolved confocal microscopy (Figs. 1A–1F). Their spectral characteristics were in agreement with the published data (3, 7).
When cells stained with Hoechst 33258 were examined using a fluorescence confocal microscope, the expected blue emission excited by 405 nm light was readily seen in nuclei (Fig. 2D) and only a negligible signal of the green fluorescence of Hoechst 33258 was detected (exc./em. 458/480–600 nm) (Fig. 2E). A 60 second exposure of the specimen to the UV emitted by a standard mercury halide lamp resulted in a decrease of the blue fluorescence that would typically be ascribed to photobleaching (see below). Following an exposure to UV, we also detected a concomitant appearance of green fluorescence that was readily excited by blue light in the nuclei of the imaged cells (Fig. 2F). This green emission was only marginally excited by the 405 nm light. The adjacent cells, which had not been exposed to UV previously, exhibited only a negligible green emission. Thus, the green fluorescence was clearly induced by an illumination with the UV or 405 nm light. A similar appearance of the green emitting forms of the dyes was observed when DAPI or Vybrant DyeCycle™ Violet were exposed to UV (Figs. 2G–2I, and data not shown, respectively). The induction of the green fluorescence suggests that new forms of both dyes (photoproducts) were generated. In an attempt to estimate the shape of the excitation spectrum of the putative photoproducts, we exposed the cells stained with Hoechst 33258 or DAPI to UV, generated the green-emitting forms and recorded the fluorescence emissions using several available excitation wavelengths (and equal intensities of the exciting light). Figure 3 demonstrates that the fluorescence of the photoproducts of Hoechst 33258 and DAPI have the highest intensities when excited by the blue light. Figures 3C and 3K demonstrate that, before an illumination of the sample with the UV, the intensity of the green fluorescence excited by 458 nm light was very weak. Clearly, while the 405 laser line was capable of generating photoproducts of Hoechst 33258 and DAPI, it did not excite the green fluorescence of the photoproducts. This indicates that the excitation maxima of Hoechst and DAPI shifted on photoconversion to longer wavelengths, into the region of 450–490 nm. We also measured the emission spectra of Hoechst 33258 and DAPI photoproducts in a confocal microscope (Figs. 1G and 1H). The emission maxima of both photoproducts generated by the UV are in the green region of the spectrum, with the tails of these curves reaching into the red region. Photostability of the photoconverted derivatives of Hoechst 33258 and DAPI was poor, as demonstrated by the photobleaching curves shown in Figures 1I and 1J.
Reversal of Photoconversion of Hoechst 33528 and DAPI
Fluorescence intensity of Hoechst 33258 and its photoproduct follow a conspicuous pattern of behavior following a short illumination with UV. Although the Hoechst blue signal was bleached out readily by the UV, within 1 hour the blue fluorescence recovered to a level of ∼50% of the initial value (Figs. 4A and 4B). At the same time the green signal of the Hoechst 33258 photoproduct, which was induced by the UV illumination, decreased to the level of ∼50% of the initial value (Figs. 4A and 4B). These complementary changes of the intensity of fluorescence of Hoechst 33258 and the photoproduct may indicate that a subpopulation of molecules of the photoproduct reverted to the original form of the dye (i.e., the blue emitting form). Both forms, i.e., the original dye and the photoproduct, reached an equilibrium that lasted for at least 2 hours following the exposure to UV (Figs. 4A and 4B). A similar phenomenon was observed in the case of DAPI (Figs. 4C and 4D).
Photoconversion of Hoechst 33258 and DAPI as a Function of the UV Dose
The concentrations of the UV-induced photoproducts of Hoechst 33258 and DAPI were proportional to the dose of the delivered UV light. A continuous UV illumination of the Hoechst 33258- and DAPI-stained cells resulted in a continuous growth of the intensity of fluorescence of their photoproducts, as demonstrated in Figure 5 (until the whole pool of the dye was photoconverted). It is important to note that this experiment required an exposure of the stained cells not only to the UV (to generate the photoproduct) but to the 458 nm light (in order to record the fluorescence of the photoproducts); therefore, some photobleaching of the photoproducts must have occurred during image recording. Thus, the increasing intensities of fluorescence of the photoproducts, which are shown in Figure 5, represent a minor underestimate of the actual levels of the generated photoproducts.
Reversible Changes of Hoechst 33528 Emission Induced by Hydrogen Peroxide
Several reports indicated that an exposure to exciting light induces photooxidation of fluorescent dyes (9–13). In order to investigate whether an exposure of Hoechst 33258 to hydrogen peroxide will yield a molecule that exhibits fluorescence similar to the photoproducts, we measured the spectral characteristics of Hoechst 33258 in a specimen of fixed cells exposed to H2O2 (30% solution). A green fluorescence signal, similar to the UV-induced change, was detected by spectrally resolved confocal microscopy (Figs. 6A and 6B). In another experiment, we used spectrofluorimetry to measure fluorescence spectra of Hoechst 33258 and DAPI solutions in distilled water, after an exposure to 30% hydrogen peroxide. The intensity of the green fluorescence of the putative oxidized Hoechst 33258 increased gradually during exposure to hydrogen peroxide. Immediately after addition of H2O2 the fluorescence intensity of the blue emission of the original dye increased as well. With time, the intensity of the blue emission decreased again, whereas the green fluorescence continued to increase (Figs. 6C and 6D). In the next experiment, we exposed Hoechst bound to DNA in fixed cells to hydrogen peroxide and subsequently replaced the hydrogen peroxide solution with physiological saline, i.e., a solution that was devoid of any reductants or oxidants (except for the dissolved oxygen). In this environment the fluorescence of the green emitting form decreased and became undetectable (Fig. 6G). Reintroduction of hydrogen peroxide brought about a return of the green fluorescence. Repeating this cycle of exposures to oxidizing and nonoxidizing conditions resulted in appearance and disappearance of the green fluorescence emission signal (Fig. 6G). A similar, but weak conversion of DAPI to the green-emitting form was induced by hydrogen peroxide, as shown in Figures 6E and 6F.
Photoconversion of DAPI and Hoechst 33258 in the Absence of DNA
The UV-induced photoconversion was first observed in Hoechst 33258 or DAPI bound to DNA in fixed cells. However, we also observed that the photoconversion of Hoechst 33258 and DAPI took place in a dye solution placed on a microscope slide and exposed to the exciting UV light (data not shown). In order to establish if photoconversion of Hoechst 33258 and DAPI is indeed independent of binding and the presence of DNA, and whether the water environment is required, we also studied the influence of UV excitation on Hoechst 33258 or DAPI immobilized in a polyvinyl polymer (8), i.e., outside of water or solution, and in the absence of cellular DNA. Under such conditions we also observed generation of a green-emitting form (Fig. 7). Apparently photoconversion of both dyes can occur outside of water solution and in the absence of DNA.
Photoconversion versus Dye Concentration and pH
We investigated whether the concentration of DNA-bound Hoechst 33258 or DAPI has any impact on the process of photoconversion. This experiment is complicated by changes in the cellular staining pattern that accompany the increasing concentrations of the dye. At high Hoechst 33258 or DAPI concentrations (3, 7, and 10 μM) fluorescence signal of Hoechst as well as DAPI bound to DNA diminished, probably due to self-quenching, whereas the signal from RNA became detectable. Nevertheless, the high dye concentrations did not inhibit the photoconversion process (Fig. 8A). The green-emitting forms of Hoechst 33258 and DAPI were detected in both concentrations of the dyes tested.
We have also investigated the influence of pH on photoconversion of Hoechst 33258 and DAPI. First, the acidity itself did have a pronounced impact on Hoechst 33258 fluorescence emission. Acidic environment (pH 2.0 or 5.1) caused a significant increase of the blue emission (ex. 405 nm, em. 430–550 nm) as well as the blue-excited green emission of this dye (ex. 458 nm, em. 480–600 nm). This observation can be explained by the increase of the quantum efficiency of Hoechst fluorescence in a low pH environment, which has been reported (14). Nevertheless, after the UV exposure of Hoechst-stained cells maintained in an acidic environment, no further increase of the green fluorescence was observed, suggesting that photoconversion did not occur, or that the photoproduct was highly unstable. Alkaline pH of 11.5 caused a reduction of the blue and green fluorescence signals of Hoechst 33258. In this environment, the UV-induced photoconversion was still detected (Fig. 8B).
Low pH did not influence the intensity of the blue or green emission of DAPI. Exposure to UV resulted in a clear photoconversion of DAPI. Alkaline pH of 11.5 caused a slight loss of fluorescence of DAPI, but it did not affect the process of its photoconversion (Fig. 8B). We conclude that photoconversion of DAPI can occur in alkaline and acidic environment, whereas Hoechst 33258 can be easily photoconverted only in alkaline environment.
Nature of the Photoconverted Derivatives of Hoechst and DAPI
In order to identify the chemical nature of the products of UV-induced photoconversion of Hoechst 33258 and DAPI, we subjected Hoechst or DAPI solutions to UV illumination (on a microscope slide) or to hydrogen peroxide (4% solution). Subsequently, the samples were analyzed by mass spectrometry.
The peaks originating from a single charge state of both dyes were prominent in the MS spectra of the dyes dissolved in water (Figs. 9A and 9B); this confirmed the presence of the original forms of the dyes in the samples. Very weak signals representing the double charged states of Hoechst and DAPI were also detected. Following exposure to UV or 4% hydrogen peroxide the ratio of the double protonated states of Hoechst and DAPI to the single protonated states of the dyes increased significantly (213.1/425.2 and 139.55/278.1 for Hoechst and DAPI, respectively) (Figs. 9A and 9B, Table 1). This increase indicated that the exposure to UV as well as the incubation with hydrogen peroxide led to protonation of the molecules of both dyes. No other prominent peaks were detected in the MS of both dyes, suggesting that the main products of exposure to UV or hydrogen peroxide were the protonated forms of the dyes.
Table 1. Mass spectrometry analysis of UV- or H2O2-generated products of Hoechst and DAPI.
A. Mass or mass-to-charge values of the original blue-emitting dyes and their H2O2-generated products of protonation
MONOISOTOPIC MASS (Da)
B. Proportional contents of double-protonated forms
In 4% H2O2
AFTER UV ILLUMINATION
Hoechst 33258 and DAPI bleach under the exciting light and their blue emission may seem lost permanently. We demonstrate that bleaching of these dyes has an interesting aspect—a change of spectral properties rather than just a loss of the blue emission. During exposure to UV both dyes (and Vybrant DyeCycle Violet, which we investigated in some experiments) undergo photoconversion into different but still fluorescent forms. The UV-induced chemical change is manifested by a shift of the excitation and the emission bands toward the longer wavelength. Generation of the photoproducts is proportional to the dose of the exciting light. It does not require the presence of DNA or water, and occurs at various dye concentrations, in acidic, neutral, and alkaline pH (except for the acidic environment in the case of Hoechst 33258). Spectrofluorimetry and mass spectrometry indicate that the chemical change induced by UV or hydrogen peroxide is protonation. This hypothesis appears to agree with the earlier description of changes of quantum efficiency of fluorescence on protonation of Hoechst 33258 (15). The preliminary mass spectrometry data require further refinement as it is not known if the photoproducts are stable during ionization procedure. As a consequence it is not known if the peaks in the mass spectra can be used for accurate quantitative assessment of concentrations of the photoproducts. The products of photoconversion are readily bleached by the exciting blue light. It remains to be established if this loss of fluorescence of the photoproducts constitutes further protonation or a different chemical reaction.
Following UV-induced photoconversion, further reactions between the dye and the photoproduct appear to take place. After photoconversion, both Hoechst 33258 and DAPI appear to reach some equilibrium with their respective photoproducts. It is possible that these reactions occur spontaneously in the dark, but we cannot rule out a possibility that small doses of UV and/or blue light are required for this process to occur (see the conditions of the experiment shown in Fig. 4).
It is conceivable that, following generation of the photoproducts, FRET between the original dye and the photoproduct occurs under the UV excitation. We observed an increase of the blue emission of Hoechst and DAPI during bleaching of the photoproduct. Such a phenomenon would be expected to occur in the case of acceptor photobleaching. Further studies are needed to establish if FRET occurs in this system.
Photoconversion of fluorescent proteins and reversible redox reactions of low molecular weight dyes attracted considerable attention recently because of potential applications of these phenomena in superresolution microscopies (16–18). Because photoconversion of Hoechst 33258 and DAPI is apparently reversible, both dyes may become useful in superresolution microscopy approaches that rely on photoswitching of fluorescent probes. Another interesting aspect of photoconversion of Hoechst and DAPI is their potential use for detection of oxidative stress exerted on DNA. Preliminary experiments conducted in our laboratory demonstrated that such an application is feasible.
A less welcome aspect of the photoconversion of Hoechst 33258 and DAPI is the issue of generation of unexpected green fluorescence signals in microscopy specimens labeled with the UV-excited and some other green, yellow or red emitting dyes. Even a small dose of UV may cause an increase in the green fluorescence signal derived from the investigated DNA dyes. This emission can be mistaken for a green fluorescence emitted by a different, blue-excited probe like fluorescein, Alexa 488 or GFP.
The authors would like to thank Mrs Małgorzata Jemioła-Rzemińska, M.Sc., for help with spectrofluorimetry measurements and Mr Artur Piróg, M.Sc., for assistance during mass spectrometry measurements. Travel fellowship awarded by ISAC and ‘Doctus’ scholarship awarded by Małopolska Centre of Entrepreneurship to DZ-B are gratefully acknowledged.