Single photon DNA photocleavage at 830 nm by quinoline dicarbocyanine dyes

Abstract

We have synthesized symmetrical carbocyanine dyes in which two 4-quinolinium rings are joined by a pentamethine bridge that is meso-substituted with H or Cl. Irradiation of the halogenated dye at 830 nm produces hydroxyl radicals that generate DNA direct strand breaks. This represents the first reported example of DNA photocleavage upon single photon excitation of a chromophore at wavelengths above 800 nm.

Graphical Abstract

Photodynamic therapy (PDT) is an emerging cancer treatment aimed at minimizing the side effects associated with traditional chemotherapeutic agents. In PDT, excitation of a photosensitizer (PS) with low energy light triggers the production of highly localized reactive oxygen species (ROS) in diseased tissues with minimal involvement of surrounding cells.^1–3 In the most common PDT mechanism, singlet oxygen (¹O₂) is generated by Type 2 energy transfer between the triplet excited state (³PS*) of the PS and ground state triplet oxygen (³O₂).⁴ The triplet state can also react with ³O₂ by Type 1 electron transfer to yield superoxide anion radicals (O₂•). Spontaneous dismutation of O₂•⁻ generates H₂O₂, which gives rise to hydroxyl radicals (•OH) via Fenton chemistry.^{4, 5} With respective diffusion distances of 50–100 nm⁶ and 0.8–6.0 nm,⁷ the short-lived and highly reactive ¹O₂ and •OH formed upon dye excitation cause extensive oxidative damage to DNA and other cellular macromolecules in their vicinity.⁸

The clinical PDT agents porfimer sodium (Photofrin®), talaporfin (Aptocine™), and verteporfin (Visudyne®) directly sensitize genomic DNA cleavage when irradiated in tissue culture and/or in circulating cells.² While their absorption bands are compatible with visible light sources that emit at wavelengths ≤ 689 nm, alternative photosensitizers with near-infrared maxima extending from ~700 nm to 900 nm are desired. This is due to enhanced penetration of incident irradiation afforded by minimal light absorption in this range by molecules in the body. Thus, the light depth attained at 835 nm through biological tissue is approximately twice that at ~630 nm, the wavelength used to activate porfimer sodium.¹

Although infrared light passes through tissues more efficiently, the result of red-shifting the λ_max of a chromophore is to reduce triplet state energy.⁹ This places limits on near-IR ROS production.⁵ For example, Type 2 singlet oxygen is generated only when a PS has a triplet state energy equivalent to or higher than the excitation energy of ¹O₂ (95 kJmol⁻¹, ~1270 nm).⁴ When taking into consideration the minimal energy gap (≤ 63 kJmol⁻¹) between the first excited ¹PS* and ³PS* states of PDT agents, this translates into an ~810 nm upper absorption limit for ¹O₂ production.¹⁰ In order for a photosensitizer to form Type 1 superoxide, the oxidation potential of the PS triplet state should be higher than the oxidation potential of ground state triplet oxygen (E° (³O₂/O₂•⁻) = 0.16 V at pH 7.0),¹¹ but lowering triplet state energy inopportunely decreases excited state oxidation potentials.¹² For the above reasons, there are relatively few examples of DNA photocleaving agents that are effective ROS generators in the near-infrared range.² Until the present report, the longest wavelengths to trigger DNA strand scission upon direct, single photon chromophore activation were all under 800 nm. In a seminal study, Chakravarty et al. cleaved plasmid DNA in high yield by employing an anthracenyl-bis(pyridyl)Fe(III) catecholate complex to photosensitize hydroxyl radical production under single photon 785 nm illumination.¹³

Near-infrared cyanine dyes are currently being developed as PDT agents^{14, 15} and are used in clinical settings as fluorescent probes in the diagnosis and imaging of cancer.¹⁶ DNA interactions are facilitated by the cyanines’ two flanking heteroaromatic nitrogen rings, which share a positive charge that is delocalized through a central polymethine bridge.¹⁷ DNA cleavage by hydroxyl radicals¹⁸ and singlet oxygen^18–20 has been reported, with excitation wavelengths in the visible range^{19, 20} up to 700 nm.¹⁸ Certain cyanine dyes, particularly those with 2-quinolinium rings, avidly interact with the DNA minor groove as monomers, dimers, and higher order aggregates.¹⁷ In the design and syntheses of cyanine dyes 4 and 5 (Scheme 1; ESI Scheme S1; Figs. S1 to S5), a 4-quinolinium motif in combination with a highly conjugated central pentamethine bridge was selected to maximize near-infrared light absorption (> 700 nm). Cyanines possessing extended polymethine chains can lose color in aqueous solutions due to spontaneous dye auto-oxidation (no hν).²¹ Substituting the H atom at the meso position of the pentamethine bridge of 4-quinolinium dye 4 with electron withdrawing Cl (5) was intended to reduce the dye oxidation process²² and, through a heavy atom effect, accelerate intersystem crossing rates from the singlet to the triplet excited PS state.²³

Scheme 1.

Chemical structures of the 4-quinolinium pentamethine carbocyanine dyes under study.

In UV-visible absorption spectra recorded over time, 4-quinolinium carbocyanines 4 and 5 appeared to be stable in DMSO, with absorption maxima of 823 nm (4, X = H) and 795 nm (5, X = Cl) (ESI Fig. S6; red lines in Fig. 1). In aqueous buffer however, the dyes exhibited markedly different signatures. The peak heights of the 823 nm and 795 nm absorption bands seen in DMSO were less prominent and new blue-shifted absorption maxima at 644 nm (4) and 547 nm (5) appeared (black lines in Fig. 1).²⁴ In contrast to DMSO, the major bands of both cyanines lost intensity in the aqueous medium, although substitution of the meso hydrogen for electron withdrawing Cl considerably slowed down absorption loss over time (Fig. S7 A,B). Interestingly, the addition of calf thymus (CT) DNA to the aqueous buffer stabilized the dyes, especially in the case of 5 (X = Cl), which, in contrast to 4, was extremely stable when the DNA was present (Figs. 1 and S7C,D). A second effect of DNA addition was to red-shift the long wavelength cyanine dye absorption maxima, e.g., from 801 nm to 826.5 nm (4, X = H); 777 nm to 805 nm (5, X = Cl).

Fig. 1.

UV-visible spectra of 10 μM of dyes 4 and 5: in DMSO (red line, t = 0 min); 10 mM sodium phosphate pH 7.0 buffer, without DNA (black line, t = 0 min) and with 150 μM bp CT DNA (blue line, t = 0 to purple line, t = 25 min).

The DNA-induced absorption changes shown in Fig. 1 indicated that it should be possible to employ the 4-quinolinium dyes to sensitize photocleavage at near-infrared wavelengths ≥ 800 nm. Towards this end, cyanines 4 and 5 were equilibrated with pUC19 plasmid and irradiated with 808 nm and 830 nm LED lasers (pH = 7.0). Duplicate sets of reactions were maintained at 10 °C (Fig. 2) and at 22 °C (Fig. S8) during irradiation. Additional samples were kept in the dark at 10 °C, 22 °C, and 37 °C (Fig. S9). DNA products resolved on agarose gels showed that 808 nm and 830 nm near-infrared light caused the dyes to convert uncut supercoiled plasmid to nicked DNA through the formation of direct strand breaks (Figs. 2 and S8). Dye 4, which rapidly degrades in aqueous buffer (Figs. 1 and S7 A,C), generated lower levels of cleavage. Considerably more strand breakage was sensitized by the stable halogenated dye (5). Increasing the reaction temperature from 10 °C to 22 °C had no effect on cleavage yields (Fig. S8). Moreover, minimal levels of strand breakage were observed in all of the dark control reactions containing dye, even after the temperature was increased from 10 °C to 37 °C (Fig. S9). Taken together, the above results appear to rule against a thermal process and instead suggest that the DNA direct strand breaks produced by exposing cyanine dyes 4 and 5 to 808 nm and 830 nm LED lasers are photochemical in nature. In our next experiment, plasmid samples containing 5 μM up to 50 μM of 4-quinolinium cyanine 5 (X = Cl) were irradiated at 830 nm for 30 min. DNA cleavage was observed at the lowest dye concentrations tested (5 μM) and gradually increased until approaching a plateau at ~ 30 μM of dye (Fig. S10). When 20 μM reactions were irradiated as a function of time, DNA cleavage occurred up until the 120 min experimental endpoint (830 nm hν; Fig S11). To the best of our knowledge, the plasmid experiments reported in this paper constitute the first reported examples of DNA strand breakage upon single photon dye excitation at wavelengths above 800 nm.

Fig. 2.

Agarose gels showing cyanine dye-sensitized photocleavage of pUC19 plasmid DNA irradiated with 808 nm (A) and 830 nm (B) LED lamps (2.8 W/cm², 30 min hν at 10 °C). Reactions contained 10 mM sodium phosphate buffer pH 7.0 and 38 μM bp DNA in the absence and presence of 20 μM of dye. Yields and standard deviation were obtained over 3 trials. Abbreviations: L = linear; N = nicked; S = supercoiled.

The superior DNA photo-cleaving abilities of 5 led us to investigate its DNA binding mode(s). In circular dichroism (CD) spectra of calf thymus (CT) DNA, the achiral cyanine generated an intense, bisignate induced CD (ICD) band indicative of exciton coupling (Fig. 3, black line). The band passes through zero ~ at the 547 nm absorption maximum of the DNA-bound form of the chlorinated cyanine (Fig. 3, green line) and has an overall appearance consistent with the formation of a right-handed helical H aggregate in which cofacial cyanine dimers assemble in an end-to-end fashion in the DNA minor groove.^{25, 26} UV-visible absorption titrations were then conducted in which small volumes of calf thymus (CT) DNA titrant were sequentially added to a solution containing a fixed amount of 5 (Fig. S12). A known effect of raising DNA concentration is to disrupt cyanine aggregation in favor of monomeric dye.^{25, 27} Adding DNA to 5 increased the intensity of the dye’s 805 nm absorption maximum while decreasing the height of its 547 band. A similar absorption change was observed when 5 was moved from buffered aqueous solvent expected to promote dye aggregation (black line in Fig. 1B, no DNA) to DMSO, a polar organic solvent that stabilizes dye monomers (red line in Fig. 1B, no DNA).²⁴ Taken together, the CD and UV-visible data suggest that the 547 nm absorption peak (Figs. 1B and 3) may arise from a DNA-bound H-aggregate, while absorption at 805 nm represents DNA-bound monomer. For 5, the absence of an ICD signal corresponding to 805 nm absorption sheds light on a possible mode of DNA interaction of the putative monomeric dye form (Figs 3 and S13). While polymethine cyanines that engage in intercalation or groove binding typically generate ICDs, induced CD signals are frequently absent in the case of chromophores that bind to DNA externally.^27–30

Fig. 3.

Double y-axis plots superimposing the circular dichroism (CD) and UV-visible absorption (Abs) spectra of dye 5 (22 °C). Samples contained 10 mM sodium phosphate buffer pH 7.0, 10 μM of dye and/or 120 μM bp (CD) to 150 μM bp (Abs) of CT DNA.

The fluorescence spectra of dye 5 (X = Cl) were recorded next (Fig. S14). Excitation wavelengths of 550 nm and 800 nm were selected to target the putative H-aggregated and monomeric dye forms, respectively. Without DNA, 550 nm excitation resulted in little if any emission (Fig. S14A). When DNA was present however, the dye fluoresced markedly. In contrast, no emission was observed under the 800 nm irradiation (Fig S14B). Upon DNA intercalation, the conformational mobility responsible for rapid non-radiative decay of cyanine dyes from singlet excited state is restricted, leading to large fluorescence enhancements.^{17, 27} The same principle has been applied to cyanine dyes that become emissive upon interacting with the DNA minor groove.¹⁷ Taken together with the ICD spectra, the fluorescence data suggest that the putative dye monomers responsible for 805 nm cyanine absorption may be interacting with DNA in a conformationally flexible fashion distinct from classical minor groove binding and intercalation. While exciting the non-fluorescent dye monomer of 5 with 808 and 830 nm LED lasers sensitized plasmid DNA cleavage in good yields (Fig. 2), less strand breakage occurred under a 532 nm laser, possibly due to a depletion of excited state populations caused by the fluorescence of the corresponding putative DNA-bound H aggregate (Fig. S15).

Mechanisms contributing to DNA photocleavage were evaluated by using chemical agents to modulate direct strand break formation. Irradiating reactions in an argon-purged glove box reduced strand scission by 5 up to 75%, strongly implicated the involvement of ground state triplet oxygen (Table S1, Fig S16). A role for Type 1 hydroxyl radicals was then suggested by the •OH scavenger sodium benzoate, which decreased photocleavage by ~40%. Although DNA strand scission was inhibited by the popular Type 2 singlet oxygen scavenger sodium azide, cleavage was supressed by ~12% when H₂O was replaced with D₂O, a solvent that produces a 10-fold enhancement in the lifetime of singlet oxygen (Table S1, Figs. S17 and S18).³¹ This rules against ¹O₂ as a contributing ROS and instead points to the residual ability of sodium azide to react with Type 1 •OH.³²

To confirm ROS involvement, we used the fluorescent probes 3’-(4-hydroxyphenyl)fluorescein (HPF) to detect hydroxyl radicals and Singlet Oxygen Sensor Green® (SOSG) to trap ¹O_2..³³ In preliminary controls, hydroxyl radicals generated by the Fenton reagent (ammonium iron(II) sulphate + H₂O₂) and singlet oxygen photosensitized by methylene blue were detected by HPF and SOSG, respectively (Fig S19). Upon irradiating the cyanine dye at 830 nm, there was a substantial fluorescence increase when HPF was present in the reaction, vs. a small decrease in the case of SOSG (Fig 4). Finally, adding the •OH scavenger sodium benzoate significantly reduced HPF signal intensity. Considered together with the data in Table S1, the fluorescence experiments point to hydroxyl radicals as the primary ROS responsible for the near-infrared DNA photocleavage sensitized by chlorinated cyanine dye 5.

Fig. 4.

Fluorescence spectra recorded in 10 mM sodium phosphate buffer pH 7.0 without and with 20 μM of 5 and: (A) 3 μM HPF (λ_ex = 490 nm) ± 100 mM sodium benzoate (SB); (B) 0.75 μM SOSG (λ_ex = 480 nm). Reactions were kept in the dark or irradiated (hν) at 830 nm for 30 min (22 ºC).

To evaluate the cellular internalization and intracellular localization of the synthesized dye, ES2 ovarian carcinoma cells³⁴ were analyzed with fluorescence microscopy³⁵ following incubation with dye 5 (10 μg/mL) for 24 h. The obtained images revealed that the intrinsic NIR fluorescence signal of dye 5 was predominantly localized in the perinuclear region and partially in nuclei of cancer cells (Fig 5A). To assess the efficiency of the internalized molecules to generate intracellular ROS upon light activation and cause a phototherapeutic effect, ES2 cancer cells were treated with dye 5 for 24 h and then exposed to a 808 nm laser diode (0.3 W/cm²) for 10 min. Intracellular ROS levels were measured with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA),³⁶ a fluorescent probe that detects hydroxyl radicals and other oxidants. The viability of the treated cells was evaluated with the dye Calcein AM³⁷. The obtained results revealed that total intracellular ROS levels in the treated cells increased by ~2 times (Fig S20) while the cell viability decreased by ~53% (Fig 5B) when compared to controls.

Fig. 5.

(A) Representative superimposed fluorescence microscopy images reveal intracellular localization of dye 5 (red) in ES2 cancer cells after incubation for 24 h followed by staining nuclei with Hoechst 33342 (blue). (B) ES2 cancer cell viability for: Cells- no treatment; Light-cells exposed to a 808 nm laser (0.3 W/cm²) for 10 min; (5) – cells incubated with dye 5 (10 μg/mL = 20 μM) for 24 h under dark conditions; (5) + Light - cells incubated with dye 5 (10 μg/mL) for 24 h and exposed to a 808 nm laser (0.3 W/cm²) for 10 min. *p < 0.05 when compared with non-treated cells.

In summary, we have synthesized two symmetrical carbocyanine dyes in which dual 4-quinolinium rings are linked to a pentamethine bridge meso-substituted with either H (4) or Cl (5). The electron withdrawing chlorine atom substantially stabilizes dye 5 in aqueous DNA solutions (Fig. S7 B,D). When bound to DNA in a conformationally flexible, monomeric fashion, single photon 830 nm excitation of the dye generates •OH radicals that produce direct strand breaks in plasmid DNA (pH 7.0, no piperidine or base). This is significant not only because 830 nm light deeply penetrates biological tissue, but also because of the intrinsic limitations on ROS production imposed by the low excited state energies of many near-IR chromophores. We have additionally shown that dye 5 is taken up by ES2 cancer cells and generates phototoxic ROS when illuminated with near-infrared light. Our findings suggest that 4-quinolinium-based carbocyanine dyes may have the potential to serve as hydroxyl radical sensitizing agents in phototherapeutic applications.