Fluorophore-mediated Photooxidation of the Guanine Heterocycle

Abstract

Fluorescent dyes are routinely used to visualize DNA or RNA in various experiments, and some dyes also act as photosensitizers capable of catalyzing oxidation reactions. The present studies explored whether the common labeling dyes fluorescein, rhodamine, BODIPY, or cyanine3 (Cy3) can function as photosensitizers to oxidize nucleic acid polymers. Photoirradiation of each dye in the presence of the guanine (G) heterocycle, which is the most sensitive toward oxidation, identified slow rates of nucleobase oxidation in the nucleoside and DNA contexts. For all four fluorophores studied, the only product detected was spiroiminodihydantoin (Sp) suggesting the dyes functioned as Type II photosensitizers and generate singlet oxygen (¹O₂). The nucleoside reactions were then conducted in D₂O solutions, known to increase the lifetime of ¹O_2, which resulted in a ~6-fold increase in the Sp yield, further supporting the classification of these dyes as Type II photosensitizers. Lastly, we inspected the pattern of G reactivity with the dyes upon photoirradiation in the context of a parallel-stranded G-quadruplex. The G nucleotides in the two exterior G-tetrads were found to be oxidation prone, providing the third line of evidence that the dyes are Type II photooxidants. The present work found that the common dyes fluorescein, rhodamine, BODIPY, or Cy3 can drive G oxidation but with a slow rate and low overall yield. This will likely not impact many experiments using dyes to study nucleic acids except for those that have long exposures with high-intensity lights, such as sequencing-by-synthesis experiments using fluorescence as the readout.

Graphical Abstract

Introduction

A typical approach used when studying nucleic acids is covalent tethering of fluorescent dyes for visualization in the experiment. Fluorophores attached to DNA or RNA oligomers have enabled monitoring their migration during gel electrophoresis experiments^[1] and the tracking of these polymers in single-molecule experiments.^[2,3] Additionally, the binding interactions between DNA or RNA with other nucleic acids,^[4] proteins,^[5] or ligands,^[6] as well as in sequencing-by-synthesis experiments can be followed.^[7] There exist many fluorophores tailored for various applications in which fluorescein, rhodamine, BODIPY, and cyanine3 (Cy3) or variants thereof are used with some frequency in nucleic acid visualization (Figure 1). Another feature of numerous dyes is their ability to function as photosensitizers,^[8] including the fluorescein derivative eosin, which has been used to conduct photooxidation in cell-based studies.^[9] In the present work, we interrogated the common fluorophores found in Figure 1 to determine their ability to function as photosensitizers for the oxidation of nucleic acids.

Figure 1.

Structures of the nucleoside and fluorescent dyes studied.

Photooxidants are classified based on the mechanistic pathway by which they oxidize target compounds such as nucleic acids.^[10,11] Type I photosensitizers when photoexcited abstract an electron from nucleobases to yield radical/radical cation intermediates followed by transfer of the abstracted electron to O₂ to form superoxide (O₂^•-). The reaction trajectory of the oxidized nucleobase can include a secondary reaction with O₂^•−.^[12] Alternatively, Type II photosensitizers when photoexcited convert ground-state O₂ to singlet oxygen (¹O₂), an electrophile that reacts via cycloaddition to nucleophilic targets.^[13] As notable examples, the chemical modification of fluorescein yields a Type II photosensitizer utilized in cell-based studies for inducing oxidative stress and labeling the target nucleobase guanine (G) in RNA;^[9] and the iodine-substituted malachite green analog is a fluorogenic photosensitizer employed to study cellular oxidative stress.^[14]

Oxidation of nucleic acids facilitated by exposure to a photosensitizer and light has been studied for nearly 40 years and provides a strong foundation of knowledge on this topic.^{[10,11,15,16]} From these studies, it is consistently found that the guanine (G) heterocycle is the main site of oxidative modification to the nucleobases. This finding results from the fact that the G base is the most electron-rich, priming the heterocycle for attack by oxidizing species.^[17] Detailed studies on G oxidation have led to many proposed pathways leading to end products that differ in structure from the parent compound.^[18–20] In the present work, the G heterocycle in the context of the 2′-deoxynucleoside (dG) or in a synthetic DNA strand was allowed to react with the fluorophores fluorescein, rhodamine, BODIPY, or Cy3 in the presence of visible light. The rates of photosensitized reaction of dG were measured, the reactivity pattern in DNA and the end products of oxidation were determined, and the reactions were probed to address whether the oxidations occurred via Type I or Type II photosensitized oxidations.

Experimental

The dG and fluorescein were obtained from commercial vendors and used without further purification (Figure 1). A few dyes were studied as their water-soluble derivatives, including the sulfo-cyanine3 carboxylate dye and BODIPY carboxylate (Lumiprobes; Figure 1), both of which will be referred to throughout the text as Cy3 and BODIPY, respectively. The rhodamine 5 isomer as an alkyne adduct was studied because of its commercial availability (Lumiprobes; Figure 1). The dG stock solution was prepared in 40 mM NaP_i buffer at pH 7.2, and the sample concentration was determined using the extinction coefficient ε_252nm = 13,700 L*mol⁻¹*cm⁻¹ to minimize errors in the time-course studies between samples. The fluorophores were prepared as DMSO concentrated stocks for which the concentrations were determined by dilution into aqueous pH 7.2 buffer and quantified using the following extinction coefficients fluorescein: ε_490nm = 80,000 L*mol⁻¹*cm⁻¹, rhodamine: ε_570nm = 93,000 L*mol⁻¹*cm⁻¹, BODIPY: ε_530nm = 80,000 L*mol⁻¹*cm⁻¹, Cy3: ε_548nm = 162,000 L*mol⁻¹*cm⁻¹.

The reactions were conducted in 200 μL total volume comprised of 1 mM nucleoside and 50 μM fluorescent dye in 40 mM NaP_i aqueous buffer at pH 7.2 with 5% DMSO and at 20 °C with atmospheric O₂ concentrations in an Eppendorf tube. The reactions were initiated by placing the tubes with the lids open 7 cm below a broad-spectrum sun lamp (LED, ~2×10⁻³ W/cm²) and the time-course of the reactions was monitored. For the reactions with fluorescein, BODIPY, and rhodamine, they were monitored every 2 h for up to 8 h post reaction initiation. The Cy3 reactions were monitored up to 41 h. Initially, all reactions at the time points analyzed were lyophilized to dryness and then resuspended in water with 50 μM trichlorophenol (TCP) present that was used as an internal standard for quantification. This compound was selected as an internal standard because at pH 7.2 it is water soluble at this concentration, and it did not coelute with the nucleosides or the photosensitizers in the chromatography described below. The reactions with the internal standard were subjected to reversed-phase HPLC (RP-HPLC) analysis following a literature method to quantify the extent of the oxidation (Figure S1).^[21] After one round of analyzing the reactions by RP-HPLC it was apparent that the dyes fluorescein, rhodamine, and BODIPY failed to fully elute from the column resulting in poor column performance and inconsistent data that made it challenging to obtain reliable values. As an alternative approach, we recognized that dG has a UV absorbance at 260 nm that the dyes do not (Figure S2), and therefore, the nucleoside reactions were monitored by changes in the dG absorption via UV-vis spectroscopy (Figure S3). The quantities of reactant remaining after each time point during the reactions were plotted and fit to a linear curve to determine the reaction rates under the conditions outlined above. The dG oxidation products in the nucleoside model were determined by an HPLC-based assay as previously described.^[21,22]

A test was conducted to support the classification type of the photochemical reactions mediated by the fluorescent dyes. This was achieved by changing the reaction solvent from H₂O to D₂O, which is known to increase the lifetime of ¹O₂ generated by Type II photosensitizers.^[11] These reactions were monitored by the UV-vis method described above.

Finally, oxidation of the G-quadruplex-forming sequence 5′-TTT TTG GGT GGG TGG GTG GGT T with the fluorophores was studied. The oligonucleotide was prepared by solid-phase synthesis using commercially available phosphoramidites that were subsequently cleaved and deprotected following the manufacturer’s protocol. The synthesized oligonucleotide was purified by anion-exchange HPLC, and the concentration of the pure sample was determined as previously reported before its oxidation.^[16] Oxidation in the G-quadruplex (G4) context was achieved by first 5′ labeling the strand with ³²P for visualization purposes as previously described.^[16] The ³²P-labeled strand was annealed by heating for 5 min at 90 °C in 20 mM KP_i (pH 7.4) buffer with 120 mM KCl and slowly cooled to room temperature; this buffer and annealing approach is known to support parallel-strand G4 formation for this sequence.^[23] The oxidations were conducted using the same lamp and set up as outlined above with 10 μM G4 and 50 μM dye concentrations with exposure for 4 h under ambient O₂ conditions. Post reaction, the G4 samples were dialyzed against ddH₂O to remove reaction salts and then lyophilized to dryness. The dry samples were resuspended in freshly prepared aqueous piperidine (200 mM) and heated at 90 °C for 1 h, an approach previously found to induce strand breaks at products of G oxidation from photosensitized oxidations.^[24] The piperidine was removed by lyophilization, after which the samples were loaded into electrophoresis loading dye. The mixture of strands with differing lengths in the dye solution were separated by running them on a denaturing polyacrylamide gel via electrophoresis (PAGE) at 75 W for 3 h. The strands were visualized by storage-phosphor autoradiography, and the corresponding band intensities were quantified using ImageQuant software.

Results and Discussion

The dG nucleoside photosensitized oxidations with fluorescein, rhodamine, BODIPY, or Cy3 dyes were set up with 1 mM nucleoside and 50 μM dye in 40 mM NaP_i buffer (pH 7.2) at 20 °C with ambient O₂ present. The reaction mixtures were exposed to light from a broad-spectrum sun lamp to initiate the photosensitized oxidations. First, a control study was conducted by exposing dG without a dye present to find that no reaction occurred when monitoring the reaction by RP-HPLC for up to 24 h. Next, time-course profiles for the reactions with dye present were monitored by RP-HPLC. This approach worked for monitoring the Cy3 photosensitized reactions (Figure S1); however, the dyes fluorescein, rhodamine, and BODIPY failed to elute from the column, even with high MeCN, resulting in this analytical approach yielding poor data that could not be used to faithfully follow the reactions. Thus, we took advantage of the dG nucleoside having a strong absorption band at 260 nm, absent from the dyes, that enabled the monitoring of the reaction as a function of the change in the UV-vis absorbance (Figures S2 and S3). The 260-nm absorbance decreased as the dG was consumed because, as described below, the product of the oxidation does not absorb light at this wavelength. In the end, the nucleoside reactions were monitored by UV-vis to follow the time-course of loss in the 260-nm absorption value.

The reactions conducted to study dG photooxidation with fluorescein, rhodamine, or BODIPY followed the loss of dG every 2 h for up to 8 h post-reaction initiation (Figure 2A). The rate of dG loss with the photosensitizers fluorescein was 2.4 nmol/h, rhodamine was 12.4 nmol/h, and BODIPY was 2.4 nmol/h (Figure 2B). When Cy3 was the photosensitizer, the reaction progressed very slowly, and as a consequence, was monitored out to 41 h post-reaction initiation (Figure 2A). The subsequent rate of dG loss was determined to be 0.4 nmol/h with Cy3 as the photooxidant (Figure 2B). First, it must be noted that UV-vis also allowed monitoring the integrity of the dyes as the reactions progressed by their visible absorption bands. Most prominent for BODIPY and rhodamine was the loss in the visible absorbance for these dyes that indicated concurrent photobleaching throughout the reactions (Figure S3). Consequently, the rates of dG loss reported are convoluted with the rates of photobleaching of the dyes and do not reflect accurate values for oxidation. Nonetheless, the rates of dG loss detected when exposed to common laboratory dyes and light demonstrate oxidation of the nucleobase can occur. The rates of dG loss followed the order rhodamine > fluorescein > BODIPY >> Cy3.

Figure 2.

Analysis of dG consumption upon exposure to visible light in the presence of four fluorescent dyes. (A) Time-dependent loss of dG in the photochemical reactions with the dyes fluorescein, rhodamine, BODIPY, or Cy3. (B) The observed rates of dG loss determined from the plots in panel A. *The values for rhodamine and BODIPY oxidations are derived from reactions in which photobleaching of the dyes was observed and do not reflect the true reaction rates. The dyes fluorescein and Cy3 did not appear to photobleach during the experiments but this was not rigorously tested.

Photooxidation of dG proceeds along two main pathways (Scheme 1).^{[10,11,15,16]} In one pathway with Type I photosensitizers, electron abstraction from the G heterocycle yields an acidic radical cation and neutral radical pair that react at the C5 position of the base to furnish imidazolone (Iz); Iz slowly reacts with water to yield oxazolone (Z) as a relatively stable end product. In the second pathway, G radical cation can add H₂O at the C8 position followed by completion of the oxidation and tautomerization to yield 8-oxo-7,8-dihydroguanine (OG). The OG formed is many orders of magnitude more sensitive to oxidation than the parent compound G^[25] and undergoes a further two-electron oxidation to yield spiroiminodihydantoin (Sp) in nucleoside reactions at pH > 7, single-stranded DNA, and G4 DNA contexts. Alternatively, oxidation of OG can yield 5-guanidinohydantoin (Gh) in nucleoside reactions at pH < 6 and reactions in double-stranded DNA contexts.^[19] Singlet oxygen can react with dG by a [4+2] cycloaddition across the imidazole ring to yield OG in the presence of reductants (thiols or ascorbate, absent in the present work) and Sp/Gh in reactions without reductant present.^[26–28] The point of this mechanistic discussion is that the products of dG oxidation provide clues as to the nature of the photosensitizer (i.e., Type I or II); the presence of Iz/Z serves as a bellwether of a Type I classification, and their absence indicates that a Type II reaction may be occurring.

Scheme 1.

Pathways to photooxidation products of the G heterocycle.

Following an established HPLC-based protocol from our laboratory the dG oxidation products were determined.^[21] In the four different fluorophore-dye mediated oxidations, the products Iz or Z were not observed, and the only detected product was Sp (Figure S4). These reactions were conducted without reductant present at neutral pH; thus, the observation of only Sp supports the conclusion that the dyes are Type II photosensitizers (Scheme 1). Additionally, Sp has a stereocenter in the base leading to its formation as a pair of diastereomers with known absolute configurations based on their elution from the HPLC column used.^[22] The R vs. S diastereomer ratio was nearly 1:1, as previously found for ¹O₂-mediated oxidation of the G heterocycle (Figure S5).^[16,26] The loss of aromaticity in the Sp product causes the lesion to no longer absorb light at 260 nm (Figure S2), enabling the UV-vis method for monitoring the time-course of reactions.

To gain further evidence that the fluorophore dyes fluorescein, rhodamine, BODIPY, and Cy3 were Type II photosensitizers under the conditions of the present study, the reactions were repeated and analyzed when the solvent was changed from H₂O to D₂O. The lifetime of ¹O₂ in D₂O significantly increases leading to greater reactivities for Type II photosensitizer in the deuterated solvent.^[11] For each of the cases, when the solvent was D₂O the rate of dG loss measured increased significantly (Figure 3). For the fluorescein sensitized oxidation, the rate of dG loss in D₂O increased by ~5-fold to 35 nmol/h relative to the H₂O reaction. The rhodamine sensitized oxidation gave a ~4-fold increase to 50 nmol/h in D₂O with respect to the H₂O reactions. The BODIPY sensitized oxidation of dG resulted in a ~6-fold increase in the rate to 15 nmol/h in D₂O relative to H₂O. Lastly, the Cy3 sensitized oxidation in D₂O proceeded with a ~6-fold increase in the reaction rate to 2 nmol/h when compared to H₂O. All reactions provided a rate enhancement when in D₂O relative to H₂O solvent providing additional support for these dyes operating as Type II photosensitizers in these reactions.

Figure 3.

Rates of dG consumption increase in buffered D₂O compared to H₂O in support of the dyes as Type II photosensitizers.

Intrigued by our previous observation that the reactivity profile of the G bases in parallel-folded G4 contexts changed with established Type I vs. Type II photooxidants,^[16] we conducted G4 oxidations with the fluorophore dyes as photosensitizers to further probe their photosensitizer classification. The sequence selected for study was previously found to adopt only a parallel-stranded G4 fold,^[23] in which Type I photosensitizers will oxidize the 5′-most G nucleotides in each of the G runs, and Type II photosensitizers will mediate damage at the exterior G tetrads, seen as the 5′ and 3′ most G nucleotides in each GGG run (Figure 4A).^[16] The Sp product of G oxidation was previously established to be labile to hot piperidine cleavage resulting in a strand break that can be visualized and quantified by gel electrophoresis when the 5` end of the DNA is labeled with ³²P.^[24] Figure 4B provides an example of PAGE analysis for oxidation of the G4 fold with rhodamine after a 4 h irradiation using a broad-spectrum sun lamp; PAGE images for the other dyes are provided in Figure S6. The oxidation lane is compared to a Maxam-Gilbert G Lane to identify the position of the G nucleotides oxidized in the sequence. The PAGE analysis distinctly shows G oxidation, as observed by strand breaks, at the 5′ and 3′ G nucleotides of each GGG track. Quantification of the band intensities further supports the visual interpretation of the PAGE (Figures 4C and S7). Identical cleavage patterns were observed for oxidation of the G4 with the fluorophores fluorescein and Cy3 when exposed to light from a broad spectrum sunlamp for up to 4 h, with each having slightly different yields (Figure 4C). While the nucleoside studies support BODIPY as a Type II photooxidant, G4 oxidation with BODIPY gave poor reaction yields that could not be confidently evaluated to make a determination on the G reaction pattern. However, these G4 oxidations add final support for the fluorescein, rhodamine, and Cy3 dyes functioning as Type II photosensitizers.

Figure 4.

Photooxidation of a parallel-stranded G4 occurs at the solvent-exposed G nucleotides reflecting ¹O₂ as the active oxidant formed by the fluorophore photosensitized reactions. (A) The anticipated patterns of G oxidation in a parallel-strand G4 for Type I and II photosensitizers. (B) Example PAGE analysis for a Maxam-Gilbert G Lane, no reaction, and 4 h irradiation in the presence of rhodamine. (C) Quantification of the band intensities of the oxidations. The original PAGE for the sections shown in panel A and data for the other reactions are provided in Figure S7.

The present work set out to address whether the common fluorescent dyes fluorescein, rhodamine, BODIPY, or Cy3 used to label nucleic acids for visualization could serve as photosensitizers to oxidize the polymer. Decades of prior work have consistently demonstrated G nucleotides are the sites of photooxidation by Type I and II photosensitizers.^[15,18] Inspection of G oxidation in the nucleoside and DNA contexts found the dyes can drive the oxidation of the heterocycle to yield the spirocycle Sp via a Type II photochemical oxidation reaction. This product is highly distorting to double-stranded DNA causing a decrease in thermal stability and is nearly a complete roadblock for polymerase bypass.^[29–31] Unexpected oxidation of G nucleotides by the dye used to visualize the nucleic acids could impact the results of the experiment whether it is a single-molecule study, binding assay, gel-based enzyme assay, or sequencing-by-synthesis experiment. However, the rate of oxidation is very slow under the present reaction conditions using a LED sun lamp, which could be enhanced with high-powered lasers for photoexcitation of the fluorophores. Likely fluorophore-mediated photooxidation of G in DNA or RNA using these standard dyes will not be a major issue unless the experiments run for long times with continuous light exposure such as multicycle sequencing-by-synthesis experiments. For comparison, G oxidations conducted with the well-studied Type I photooxidant riboflavin or with Rose Bengal, a high-yielding photosensitizer for ¹O₂ formation, lead to G oxidation rates that are more than two orders of magnitude higher (unpublished results).

The present observation that these four dyes can function as Type II photosensitizers to convert ³O₂ to ¹O₂ is consistent with prior literature reports. For example, BODIPY is a known Type II photosensitizer.^[32] The fluorescein derivative eosin is known to be a Type II photosensitizer in cellular studies.^[33] Further, rhodamine dyes are known as singlet oxygen generators,^[34] and the present work further supports this finding. Lastly, cyanine dyes can generate ¹O₂,^[35] consistent with the present report. This work provides a side-by-side comparison of the dyes under identical reaction conditions. A typical issue with fluorescent dyes is that they photobleach, which is also consistent with the present findings. The photobleaching observed could result from ¹O₂ or low-level Type I photosensitization by the dyes to yield O₂^•- not detected in the present studies.

Conclusions

For the studies described herein, we were curious whether common laboratory fluorophores that are attached to DNA or RNA for visualization in various experiments could function as photosensitizers to drive the oxidation of the G nucleotide. To test this, G in the contexts of a nucleoside or G4-folded DNA was exposed to light in the presence of fluorescent dyes that included fluorescein, rhodamine, BODIPY, or Cy3 to inspect for oxidation (Figure 1). All four fluorescent dyes were found to drive the photooxidation of the G heterocycle. When dG was studied, the measured reaction rates were found to be slow (Figure 2). In all cases the product of the reaction was found to be the Sp spirocycle that suggested, based on prior work,^[26] that the oxidation proceeded via a Type II mechanism with ¹O₂ as the active oxidant (Scheme 1). When reactions were conducted in D₂O the rates increased further supporting ¹O₂ as the oxidant (Figure 3). Lastly, when a parallel-stranded G4 was allowed to react with the dyes in the presence of light, the pattern of oxidation was on each G in an exterior G-tetrad, providing the third line of evidence that ¹O₂ was the active oxidant (Figure 4). The results support the conclusion that the cohort of fluorophores studied can function as photosensitizers to drive G oxidation; however, the reaction rates are extremely slow with the low-intensity light source studied (~2×10⁻³ W/cm²). Oxidation of G as an artifact when using fluorescent dyes as visualization probes is likely not an issue for most experiments except those that expose the nucleic acid to dyes and light for extended periods such as sequencing-by-synthesis using fluorescence as a readout.