Pyridinium- and bromine-substituted distyryl-BODIPY dyes for mitochondria-targeted photodynamic therapy

Abstract

A series of 1-methylpyridinium-substituted brominated distyryl-BODIPY dyes, PyBXI (X = H, M, or Br), was synthesized to achieve cooperative singlet oxygen (¹O₂) production through qualitatively different dual intersystem crossing (ISC) pathways: spin–orbit charge-transfer ISC (SOCT-ISC) and heavy-atom-induced ISC. Upon photoexcitation of the PyBXI dyes, charge-transfer states were preferentially formed through photoinduced electron transfer from the distyryl-BODIPY core to the 1-methylpyridinium moiety, however, followed by nonradiative charge recombination rather than the desired SOCT-ISC. This resulted in negligible fluorescence and ¹O₂ quantum yields in the non-brominated dye PyBHI. The introduction of bromine atoms improved ¹O₂ quantum yield from 0.0034 for the mono-brominated dye PyBMI to 0.0061 for the di-brominated dye PyBBrI, attributable to the heavy atom effect. Nonetheless, the ¹O₂ production efficiency of these dyes remained limited, as photoinduced electron transfer was considered to occur nearly two orders of magnitude faster than singlet-to-triplet ISC. In vitro assays using MCF-7 and HeLa cells demonstrated that PyBBrI induced significant cell death, with IC₅₀ values of ca. 95 and 220 nM, respectively, confirming its potential for use in cancer therapy.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-36213-x.

Introduction

Photodynamic therapy (PDT) is an effective therapeutic modality that has garnered extensive attention for cancer treatment owing to its non-invasive nature, minimal side effects, and low drug resistance^1–4. PDT relies on triplet photosensitizers (PSs), which exhibit the following photosensitization mechanism: Upon the absorption of a specific wavelength of light, singlet (S_n with n ≥ 1) states are formed. Following the S_n (n > 1)→S₁ internal conversion if necessary, intersystem crossing (ISC) occurs from the S₁ state to triplet (T_m with m ≥ 1) states. Following the T_m (m > 1)→T₁ internal conversion if necessary, the transfer of electrons (Type I) or excitation energy (Type II) from the T₁ state generates cytotoxic reactive oxygen species such as singlet oxygen (¹O₂), hydroxyl radicals (^●OH), and superoxide anion radicals (O₂^●−), ultimately inducing cancer cell death. Therefore, the singlet-to-triplet ISC is the key excited-state dynamics that determines the PDT efficiency.

Various molecular scaffolds, such as chlorin,^5,6 hematoporphyrin,^7,8 phthalocyanine,^9,10 and boron dipyrromethene (BODIPY),^11,12 have been used for the design and development of triplet PSs. Among these, BODIPY is especially attractive owing to its high molar extinction coefficients, sufficient fluorescence quantum yields, ease of modular structural modifications, resistance to photobleaching, and stability in various environments^12–14. However, BODIPY inherently suffers from low ISC efficiency, which is a decisive factor limiting its application as a triplet PS. A conventional approach for improving the ISC efficiency of BODIPY-based PSs is to leverage the heavy atom effect. Specifically, the incorporation of heavy halogen atoms such as bromine and iodine into the BODIPY backbone results in the perturbation of spin–orbit coupling (SOC), thereby promoting ISC^15,16. An alternative approach is to exploit spin–orbit charge-transfer ISC (SOCT-ISC),^17,18 which typically occurs in BODIPY-based donor–acceptor dyads^19–21 and dimers^22–24 to conserve the total angular momentum. In these systems, charge-transfer (CT) states are formed via intramolecular photoinduced electron transfer (PET) following photoexcitation. The subsequent charge recombination (CR) process alters the molecular orbital angular momentum, which produces a magnetic torque to induce electron-spin flipping and thus ISC.

Recent studies have focused on combining the aforementioned approaches to develop BODIPY-based donor–acceptor–heavy atom systems, pursuing maximal enhancement of the ISC efficiency by activating qualitatively different multiple ISC channels^25,26. For example, Lin et al. synthesized BODIPY PSs with two iodine atoms at the 2,6-positions and various pyridinyl and pyridinium moieties at the meso position of the BODIPY and evaluated their photodynamic antimicrobial activities²⁶. The Kim group synthesized BODIPY PSs incorporating two halogen atoms (chlorine, bromine, or iodine) at the 2,6-positions and triphenylamine and pyrene moieties at the meso position of the BODIPY and evaluated the ¹O₂ and triplet–triplet annihilation upconversion quantum yields^25,27. Despite these efforts, the practical utilization of donor–acceptor–heavy atom systems remains limited owing to lack of specificity toward cancer cells and nonoptimal spectral range beyond the phototherapeutic window (650–900 nm).

As aforementioned, one of the limitations of BODIPY PSs for PDT is their lack of specificity toward cancer cells. To address this problem, many studies have focused on developing BODIPY PSs targeting subcellular organelles such as mitochondria²⁸, lysosomes²⁹, and nuclei³⁰. Mitochondria, which play an essential role in metabolism, are attractive targets owing to their unique characteristics. Notably, the mitochondrial membrane potential in cancer cells (ca. −220 mV) is remarkably higher than that in normal cells (ca. −110 to −180 mV), which leads to a ca. 100–1000-fold increase in cellular uptake of lipophilic cations^31–33. To impart a positive charge to PSs, while various cationic groups such as triphenylphosphonium^34,35 and quaternary ammonium^36,37 have been used, pyridinium is particularly appealing. This cation not only endows PSs mitochondrial specificity but also induces intramolecular PET owing to its electron-deficient nature, enabling PSs to hold the potential of undergoing the SOCT-ISC required for ¹O₂ production. Furthermore, the structural properties of pyridinium allow PSs to be effectively localized within mitochondria^38–40.

As part of ongoing efforts to develop donor–acceptor–heavy atom-based BODIPY PSs, this work reports the synthesis of distyryl-BODIPY dyes, PyBXI (X = H, M, or Br), which incorporate an electron donor 1-methylpyridinium at the meso position and heavy bromine atoms at the 2,6-positions of the BODIPY to promote the CT-state-mediated SOCT-ISC and heavy-atom-induced ISC, respectively (Fig. 1a; See Supporting Information S1 and S2 for synthetic details and compound characterization, respectively). Additionally, these dyes exhibit near-infrared (NIR) emission and mitochondrial specificity owing to the two styryl and cationic 1-methylpyridinium moieties, respectively. The photosensitization dynamics of PyBXI dyes, including those associated with CT states and the heavy atom effect, were investigated via steady-state and femtosecond transient absorption (TA) spectroscopy, along with quantum mechanical calculations. The ¹O₂ production capability of the dyes was characterized, focusing on whether the two ISC dynamics can cooperatively produce triplet states, and ultimately, ¹O₂. In vitro assays using two carcinoma cell lines (human breast cancer, MCF-7, and human cervical cancer, HeLa) were performed to evaluate mitochondrial specificity and PDT activity of the dyes.

Fig. 1.

(a) Molecular structure of PyBXI (X = H, M, or Br) dyes. (b) Normalized steady-state absorption (solid lines) and emission (dashed lines) spectra of PyBXI dyes in MeOH. Data recorded in THF and ACN can be found in Supporting Information S3. (c) Solvent-dependent emission spectra of PyBHI. Solvent-dependent (d) Stokes shifts and (e) fluorescence quantum yields of PyBXI dyes.

Results and discussion

Steady-state spectroscopic measurements

Steady-state absorption and emission spectra of PyBXI dyes were recorded in three solvents with different polarities: tetrahydrofuran (THF), acetonitrile (ACN), and methanol (MeOH). Note that the dyes were insoluble in nonpolar solvents, including hexane and cyclohexane. As shown in Fig. 1b and Supporting Information S3 and S4, the absorption spectra of the dyes peaked in the 654–696 nm range, varying slightly with the BODIPY dye and solvent, and extended to the 750–800 nm range. This redshift toward the NIR region relative to the spectral range of simple BODIPY dyes (420–580 nm)⁴¹ suggests that the covalent attachment of the two styryl moieties to the BODIPY backbone effectively extends the π-conjugation length.

The fluorescence emissions from the dyes were relatively inefficient, as described below; however, their spectral profiles remained preserved, as evidenced by Figs. 1b and S17, thereby enabling the estimation of Stokes shifts. As shown in Fig. 1d, the PyBXI dyes exhibited fairly large Stokes shifts (> 1040 cm⁻¹) in polar solvents (ACN and MeOH), which is noteworthy given that BODIPY dyes typically exhibit a Stokes shift of ~ 500 cm⁻¹ owing to their limited conformational flexibility¹⁴. In particular, the observed Stokes shifts were larger than those (~ 660 cm⁻¹) of distyryl-BODIPY dyes, which are identical in structure to the PyBXI dyes except for the tethering of phenyltrimethylammonium instead of 1-methylpyridinium at the meso position of the BODIPY⁴². Furthermore, as shown in Fig. 1c, the non-brominated dye PyBHI exhibited a significant reduction in fluorescence emission as the solvent polarity increased from THF to ACN and MeOH. For quantitative analysis, the fluorescence quantum yield (Ф_F) was measured using Rhodamine 6G in ethanol (Ф_F = 0.95) as a reference and excitation wavelengths corresponding to the absorption peaks. The average Ф_F values determined from three measurements for each condition are shown in Fig. 1e. PyBHI exhibited extremely low Ф_F values across all solvents. Specifically, its Ф_F was 0.05 in THF and further decreased to 0.02 and 0.02 in the more polar ACN and MeOH, respectively. PyBMI and PyBBrI exhibited similar trends. The solvent-polarity-dependent fluorescence capability and unusually large Stokes shift of PyBXI dyes strongly suggest that CT states are formed upon photoexcitation, which will be corroborated by quantum mechanical calculations described below.

Although the formation of CT states in PyBXI dyes dominated their steady-state spectroscopic properties, the effect of bromine atoms in PyBMI and PyBBrI was also evident. In each of the three solvents, Ф_F of PyBXI gradually decreased as the number of bromine atoms increased (Fig. 1e), attributable to enhanced ISC efficiency owing to the heavy atom effect.

Quantum mechanical calculations

To theoretically corroborate the formation of CT states in PyBXI dyes, density functional theory (DFT) and time-dependent DFT (TD-DFT) calculations were performed using the Gaussian 16 program at the B3LYP-D3/6-31G(d,p) level of theory (see Materials and Methods for calculation details). All calculations were performed in MeOH solvent, with the two triethylene glycol chains on each of the two styryl moieties replaced by hydrogen atoms. The optimized structures, energy diagrams, frontier molecular orbitals, and transition contributions of the PyBXI dyes are shown in Figs. 2 and 3, and S18–S21. Based on the ground-state (S₀) optimization results, both the HOMO and HOMO−1 of all dyes exhibit effective π-delocalization across the distyryl-BODIPY framework. TD-DFT calculations in this Franck–Condon (FC) region clearly indicate that the HOMO−LUMO transition arising from this extended π-delocalization is the primary origin of the intense red absorption observed in these compounds. Optimization at the first excited-state (S₁) energy minimum reveals that all dyes undergo significant structural relaxation. In particular, the dihedral angle of the 1-methylpyridinium moiety, which is nearly orthogonal in the FC region, is markedly reduced. TD-DFT results at this S₁ minimum further demonstrate that this structural change substantially alters the π-electronic structure of the dyes, leading to a distinct redistribution of electron density in the LUMO and LUMO+1 compared to those in the FC region. Whereas the electron densities of the LUMO and LUMO+1 are localized on the distyryl-BODIPY and 1-methylpyridinium, respectively, in the FC region, they become redistributed and accumulated predominantly on the 1-methylpyridinium in both orbitals upon dihedral angle relaxation. Moreover, the S₀→S₁ transition at the S₁ minimum is dominated exclusively by the HOMO−LUMO transition. Taken together, the DFT results in both the FC region and the S₁ minimum suggest that structural relaxation along the S₁ potential energy surface drives a CT process in PyBXI dyes, which is in excellent agreement with the experimental observations.

Fig. 2.

S₀- and S₁-state optimized structures, energy diagrams, and transition orbital contributions of the PyBBrI dye. The frontier molecular orbitals shown here, as well as in Figs. 3 and S18–21, were visualized using an isosurface value of 0.02.

Fig. 3.

Frontier molecular orbital diagrams and transition information for the PyBBrI dye at the (a) S₀- and (b) S₁-state energy minima.

Quantitative measurements of ¹O₂ production

Both spectroscopic experiments and quantum mechanical calculations demonstrate the formation of CT states in PyBXI dyes. In BODIPY-based donor–acceptor dyads and dimers, CT states often lead to SOCT-ISC to conserve the total angular momentum. Specifically, the changes in the molecular orbital angular momentum during CR are compensated via an electron-spin flip. Apart from this CT-state-mediated SOCT-ISC, heavy atoms incorporated in BODIPY dyes perturb the SOC to promote ISC. Accordingly, our PyBXI dyes hold the possibility of exhibiting two qualitatively different ISC pathways, which may cooperatively facilitate triplet-state formation and thus ¹O₂ production.

To assess whether either or both CT-state-mediated and heavy-atom-induced singlet-to-triplet ISC processes occur, ¹O₂ quantum yield (Ф_Δ) was measured using 1,3-diphenylisobenzofuran (DPBF). DPBF reacts with ¹O₂ to form 1,2-dibenzoylbenzene, resulting in a decrease in the DPBF absorption peaked at 410 nm. An air-saturated methanol solution of DPBF (A_{410 nm} ≈ 1.0, with A indicating the absorbance) containing PyBXI (A_{660 nm} ≈ 0.5) was irradiated with a light-emitting diode (LED; 660 nm, 7 mW cm⁻²) for 0–60 min, during which a time series of absorption spectra was collected. As shown in Figs. 4a–c, the decrease in the DPBF absorption band was insignificant for the PyBHI-containing solution and gradually became more pronounced for the PyBMI- and PyBBrI-containing solutions. In agreement with the fact that the reaction of DPBF with ¹O₂ follows first-order kinetics^43,44, the plot of In(A₀/A_t) against the LED irradiation time t (where A₀ and A_t represent the DPBF absorbance at t = 0 and t min, respectively) displayed a linear relationship (Fig. 4d).

Fig. 4.

Time series of absorption spectra of air-saturated MeOH solutions of DPBF containing (a) PyBHI, (b) PyBMI, and (c) PyBBrI under LED light irradiation (660 nm, 7 mW cm⁻², 0–60 min). (d) PyBXI-dependent temporal changes in the DPBF absorbance at 410 nm, plotted according to first-order kinetics (dots) and linear fits of data (lines).

The Ф_Δ values of the PyBXI dyes were determined relative to methylene blue in MeOH (Ф_Δ ​​= 0.5; see Materials and Methods and Supporting Information S6). The average Ф_Δ values ​determined from three independent measurements were 0.0009 ± 0.0005, 0.0034 ± 0.0006, and 0.0061 ± 0.0016 for PyBHI, PyBMI, and PyBBrI, respectively. Strikingly, the Ф_Δ values for the PyBXI dyes were nearly an order of magnitude smaller than those of their respective analogues, which are identical in structure to the PyBXI dyes except for the tethering of phenyltrimethylammonium instead of 1-methylpyridinium at the meso position of the BODIPY, thereby lacking CT characteristics. These results were contrary to our initial expectation that both CT-state-mediated SOCT-ISC and heavy-atom-induced ISC would act cooperatively to maximize ¹O₂ production in PyBXI dyes. The extremely low Ф_Δ of the non-brominated dye PyBHI suggests that the CT state formed by PET from the distyryl-BODIPY to the 1-methylpyridinium primarily dissipates nonradiatively rather than inducing SOCT-ISC. This result is inconsistent with a previous study by Harriman et al.⁴⁵, in which a BODIPY derivative bearing an N-methylpyridinium group at the meso position and two ethyl groups at the 2,6-positions (BOD-CAT) was reported to exhibit rapid ISC mediated by a CT-state. This discrepancy may arise from structural differences between the two systems. A previous study by Filatov et al. on BODIPY–anthracene dyads⁴⁶ demonstrated that alkyl substitution on the BODIPY core significantly modulates the HOMO and LUMO energy levels and, consequently, their energetic alignment relative to those of anthracene, critically affecting CT efficiency and triplet-state generation. In this context, the fact that PyBHI, compared to BOD-CAT, lacks alkyl substituents at the 2,6-positions but is distyrylated at the 3,5-positions of the BODIPY core may lead to an energy level alignment between the HOMOs and LUMOs of the distyryl-BODIPY and 1-methylpyridinium that is unfavorable for SOCT-ISC, thereby promoting CR. Despite being low overall, the Ф_Δ values of the PyBXI dyes increased with the increase in the number of bromine atoms, demonstrating the effectiveness of heavy-atom-induced ISC.

Additional experiments using 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) confirmed that the production of reactive oxygen species, including H₂O₂, by PyBXI dyes was insignificant (Supporting Information S7).

Femtosecond TA spectroscopy

To decipher the observed photosensitizing capability of PyBXI dyes, their excited-state dynamics were investigated via femtosecond TA measurements. TA spectroscopy measures changes in absorbance over time after photoexcitation. Because this photoexcitation transiently alters the populations of ground and excited states, the absorption and emission of excited species and reduced ground-state populations result in three characteristic TA spectral signals: positive excited-state absorption (ESA), negative stimulated emission (SE), and negative ground-state bleaching (GSB)⁴⁷. The temporal evolution and decay of these signals carry essential information regarding the excited-state dynamics.

The TA spectra of PyBXI dyes are shown in Figs. 5a–c. The negative signals above 650 nm exhibited a redshift from PyBHI to PyBMI and PyBBrI, consistent with their steady-state absorption and emission bands and attributable to GSB and SE. Apart from these redshifts, all PyBXI dyes exhibited similar TA spectral evolution and decay, where a pronounced double-exponential feature with time constants of approximately 200 fs and 250 ps were observed (Figs. 5d–f and S24). Considering the spectroscopic and theoretical evidence provided above and the similarity in the decay timescale with the CT-state lifetimes of BODIPY-based donor–acceptor dyads and dimers^48,49, it is concluded that CT is the primary process governing the excited-state dynamics of PyBXI dyes.

Fig. 5.

Time series of femtosecond TA spectra of (a) PyBHI, (b) PyBMI, and (c) PyBBrI in MeOH. Decay profiles of the GSB signal at selected wavelengths for (d) PyBHI, (e) PyBMI, and (f) PyBBrI.

However, in contrast to PyBHI, PyBMI and PyBBrI exhibited additional long-lived signals with a time constant longer than 10 ns. Further nanosecond TA measurements clearly revealed that these long-lived signals persisted for more than 1 µs (Fig. S25). The long-lived component in these brominated dyes is attributable to triplet-state lifetimes, consistent with previous reports^50,51. These residual signals in the TA decays were more prominent in PyBBrI than in PyBMI, which is expected as the increased number of heavy bromine atoms enhances SOC and thus facilitates ISC. The lack of such a long-lived residual component in PyBHI indicates no triplet-state formation, from which we conclude that the CT states formed by PET from the distyryl-BODIPY to 1-methylpyridinium in PyBXI dyes mainly dissipate nonradiatively rather than being followed by SOCT-ISC. Given this absence of CT-state-mediated SOCT-ISC, the triplet-state formation in PyBMI and PyBBrI is most likely driven by heavy-atom-induced ISC.

Plausible photophysical pathways for PyBXI dyes are illustrated in Fig. 6. PyBHI preferentially undergoes PET to form a CT state rather than emitting fluorescence, followed by nonradiative CR rather than the desired SOCT-ISC. Consequently, both the fluorescence and ¹O₂ production efficiencies of this dye are low, as evidenced by the extremely small Ф_F and Ф_Δ values. In the case of the brominated dyes PyBMI and PyBBrI, after the formation of the S₁ state upon photoexcitation, a pair of nonradiative PET-CR processes, similar to PyBHI, and heavy-atom-induced ISC concurrently occur. However, the former dynamics is expected to dominate the latter dynamics, considering that time-resolved spectroscopic studies of various BODIPY derivatives have verified that the rate of PET^52–54 is typically two orders of magnitude higher than that of ISC^55,56. Nonetheless, the heavy-atom-induced ISC is still influential. In other words, the increase in the number of heavy bromine atoms enhances the ¹O₂ production efficiency (Ф_Δ) from 0.0034 for PyBMI to 0.0061 for PyBBrI.

Fig. 6.

Jablonski diagram illustrating plausible photophysical pathways in (a) PyBHI and (b) PyBMI and PyBBrI. The energy levels are approximately depicted based on TD-DFT calculations (see Materials and Methods for calculation details) and spectroscopic data. As the energy levels of the CT states exhibit a pronounced dependence on solvent polarity, as demonstrated by the Stokes shift data in Fig. 1d, they are represented as a band.

In vitro cell imaging and viability assays

Although the PyBXI dyes do not exhibit cooperative ¹O₂ production, we evaluated their potential for in vitro PDT using MCF-7 and HeLa cells. In the dyes, the 1-methylpyridinium moiety has two functions: It acts as an electron acceptor to induce PET for CT-state formation, as described above, and its cationic characteristic allows the dyes to target cancer cell mitochondria whose membrane potential (ca. −220 mV) is more negative compared with that of normal cell mitochondria (ca. −108 to −180 mV). Considering the latter aspect, the mitochondrial targeting ability of PyBXI dyes was assessed via cell imaging. MCF-7 and HeLa cells were treated with 1 µM of PyBXI dissolved in DMSO for 24 h, followed by staining with the mitochondrial dye MitoTracker Green (MTG) and nuclear dye 4′,6-diamidino-2-phenylindole (DAPI). Subsequently, the cells were subjected to confocal laser scanning microscopy (CLSM; see Materials and Methods for experimental details). Figures 7a and b show CLSM images of the two cell lines, obtained via the selective excitation of PyBXI, MTG, and DAPI. Irrespective of the cell line and BODIPY dye, the PyBXI dyes (red) colocalized with the MTG (green) in the cytoplasm around the DAPI (blue)-stained nuclei. Moreover, the fluorescence intensity profiles of PyBXI and MTG along an intersection of one of the cells shown in Figs. 7a and b overlapped each other (Supporting Information S9). The excellent colocalization of the two dyes demonstrates the ability of PyBXI dyes to target cancer cell mitochondria, attributable to the cationic 1-methylpyridinium moiety of the dyes.

Fig. 7.

CLSM images of (a) MCF-7 and (b) HeLa cells treated with 1 µM of the PyBXI dyes (red), MTG (green), and DAPI (blue). Merged images are shown at the bottom in (a) and (b). Scale bar = 10 μm. Cell viabilities of (c) MCF-7 and (d) HeLa cells after treatment with the PyBXI dyes of different concentrations (0–640 nM). The treated cells were incubated either in the dark or under LED light irradiation (680 nm, 40 mW, 30 min) prior to viability evaluation. Data denote mean values, and error bars denote standard deviation (n = 3).

The biocompatibility and photocytotoxicity of PyBXI dyes were assessed via cell viability assays. MCF-7 and HeLa cells were incubated in 5% CO₂ for 24 h at 37 °C and then treated with PyBXI dyes of different concentrations (0–640 nM) for an additional 24 h (see Materials and Methods for experimental details). As shown by the gray bars in Figs. 7c and d, irrespective of the cell line and BODIPY dye, the control group treated under the dark condition exhibited no significant reduction in cell viability (> 90%) across the tested BODIPY doses. This suggests that the synthesized PyBXI dyes are highly biocompatible. As shown by the green bars in Figs. 7c and d, under LED light irradiation (680 nm, 40 mW, 30 min), the PDT effect of PyBHI toward both MCF-7 and HeLa cells was insignificant, given the minimal difference in cell viability compared with that of the control group. The viability of the PyBMI-treated MCF-7 and HeLa cells reduced to ca. 70 and 80%, respectively, only at the highest BODIPY dose of 640 nM. In contrast, the most significant PDT effect was observed in the case of the di-brominated dye PyBBrI. The cell viability began to effectively reduce from the BODIPY dose of 80–160 nM and reached 28 and < 5 nM at the highest dose (640 nM) for MCF-7 and HeLa cells, respectively. Accordingly, the half-maximal inhibitory concentration (IC₅₀) values ​​of PyBBrI for MCF-7 and HeLa cells were ca. 95 and 220 nM, respectively. PyBBrI exhibited significant PDT efficacy despite its notably low Ф_Δ, as determined from the DPBF experiment. This apparent discrepancy may be interpreted by several factors. One possibility is dye aggregation under biological conditions, a phenomenon previously reported for BODIPY derivatives to enhance ISC^57,58. To examine this possibility, we performed steady-state absorption and emission measurements on PyBXI in DMSO/water mixtures with varying water fractions. As the water content increased from 0 to 90%, PyBXI exhibited clear spectral signatures of H-type aggregation: slight blue shifts in both absorption and emission, a pronounced increase in the 0–1/0–0 vibronic peak ratio in absorption, and substantial fluorescence quenching (Fig. S27). To further evaluate the ROS generation in the aggregated state, a 10:90 DMSO/water mixture containing 9,10-anthracenediyl-bis(methylene)dimalonic acid (ABDA) and PyBXI was irradiated with an LED (660 nm, 7 mW cm⁻²). Although the irradiation intensity was considerably lower than that used in the PDT assays and ABDA is generally less sensitive to ¹O₂ than DPBF, a perceptible decrease in ABDA absorption in the < 400 nm region was observed over 1 h of irradiation, confirming ¹O₂ generation by the H-type aggregates (Fig. S28). These results support the likelihood that PyBBrI formed H-type aggregates in the intracellular environment, which could have modulated its excited-state pathways—likely by reducing the singlet-triplet energy gap and suppressing radiative decay—thereby facilitating ISC. Another possibility involves complex intracellular interactions. The optimal mitochondrial localization of PyBBrI may have contributed to its high PDT efficacy^59–61. Moreover, interactions between PyBBrI and subcellular biomolecules may have restricted the free rotation of the two long and flexible triethylene glycol chains attached to each of the two styryl moieties. This conformational constraint may have suppressed undesirable nonradiative decay pathways⁶², thereby facilitating ISC dynamics. Furthermore, cellular environmental factors, such as pH, ionic strength, and electrostatic interactions, may have modulated the electronic structures of the distyryl-BODIPY (donor) and 1-methylpyridinium (acceptor)^63–66 in a manner that favors SOCT-ISC over CR following the CT-state formation. Collectively, these factors may account for the appreciable in vitro PDT activity of PyBBrI.

Conclusions

We synthesized 1-methylpyridinium-substituted brominated distyryl-BODIPY dyes, PyBXI (X = H, M, or Br). The spectral window of these dyes fell within the NIR region owing to the extension of π-conjugation achieved via the distyrylation of the BODIPY backbone. We hypothesized that while heavy bromine atoms undoubtedly facilitate ISC by perturbing SOC, CT states formed by PET from distyryl-BODIPY to 1-methylpyridinium may be followed by SOCT-ISC, cooperatively generating triplet states and thus cytotoxic ¹O₂. Indeed, CT states were formed in PyBXI dyes, as evidenced by spectroscopic experiments and quantum mechanical calculations. Specifically, the evidence included a Stokes shifts of > 1000 cm⁻¹, larger than those of conventional BODIPYs; solvent-polarity-dependent fluorescence reduction; and S₀→S_{1 or 2} transitions with negligible oscillator strength, accompanied by electron density movement from distyryl-BODIPY to 1-methylpyridinium. Nonetheless, the CT states mainly dissipated nonradiatively rather than mediating SOCT-ISC, causing the non-brominated dye PyBHI to exhibit negligible Ф_F and Ф_Δ values. Apart from this absence of CT-state-mediated SOCT-ISC, heavy-atom-induced ISC occurred in the brominated dyes: Ф_Δ increased from the mono-brominated dye PyBMI to di-brominated dye PyBBrI. Notably, the ¹O₂ production efficiency of these dyes was approximately an order of magnitude smaller than that of their respective analogues identical in structure to the PyBXI dyes except for the incorporation of phenyltrimethylammonium instead of 1-methylpyridinium, resulting in a lack of CT characteristics. These results indicate that a pair of nonradiative PET-CR processes is dominant over the heavy-atom-induced ISC, as PET is known to proceed faster than ISC by nearly two orders of magnitude. The proposed excited-state dynamics of PyBXI dyes were directly confirmed by femtosecond TA spectroscopy. The TA signals of PyBMI and PyBBrI decayed bi-exponentially with time constants of ca. 250 ps and > 10 ns, corresponding to CT-state and triplet-state lifetimes, respectively. In contrast, PyBHI exhibited only the ca. 250 ps decay. In vitro CLSM using two carcinoma cell lines revealed that all PyBXI dyes exhibited mitochondrial targeting ability, attributable to the cationic 1-methylpyridinium moiety. In cell viability assays, all PyBXI dyes showed no dark cytotoxicity, and only the di-brominated dye PyBBrI induced significant cell death, with IC₅₀ values of ca. 95 and 220 nM for MCF-7 and HeLa cells, respectively. These results highlight the potential of PyBBrI for use in cancer therapy.

Materials and methods

Materials and instrumentations

All reagents were obtained from commercial sources and used without additional purification unless specified. 2,4-dimethylpyrrole, 4-pyridiniumcarboxaldehyde, trifluoroacetic acid and boron trifluoride diethyl etherate (BF₃·Et₂O) were purchased from Sigma Aldrich (USA). 2,3-dichloro-5,6-dicyano-1,4-benzoquinone was bought from Alfa Aesar (USA). N-Bromosuccinimide was obtained from TCI (Japan). Triethylamine (TEA), iodomethane, sodium hydrogen carbonate (NaHCO₃), and magnesium sulfate (MgSO₄) were procured from Daejung Chemical (South Korea). Solvents were of analytical grade and were distilled prior to use. All compounds were characterized by ¹H- and ¹³C-NMR spectroscopy on a Bruker AM 250 spectrometer (USA) and high-resolution electrospray ionization mass spectrometry (HR-ESI-MS) on an SYNAPT G2-Si high-definition mass spectrometer (United Kingdom).

Spectroscopic characteristics

Steady-state absorption and emission spectra of PyBXI dissolved in solvents with different polarity were measured at room temperature using a spectrophotometer (UV-1650PC; Shimadzu) and a fluorometer (F-7000; Hitachi), respectively. The fluorescence quantum yields (Φ_F) were determined by a comparative method using Rhodamine 6G (Φ_F = 0.95 in ethanol) as a reference according to the following equation:

1

where A, E, and n refer to the optical density at the excitation wavelength, the integrated fluorescence intensity, and the refractive index of the solvent used, respectively. The superscripts S and R represent sample and reference, respectively.

Singlet oxygen quantum yield measurements

The ¹O₂ quantum yields (Φ_Δ) were determined by a comparative method using 1,3-diphenylisobenzofuran (DPBF) as a ¹O₂ quencher and methylene blue (MB, Φ_Δ = 0.5 in methanol) as a reference. The DPBF (A_{410 nm} ≈ 1.0) was prepared in air-saturated methanol solutions containing PyBXI (A_{660 nm} ≈ 0.5) and irradiated with red LED light (λ_max = 660 nm) with the intensity of 7 mW cm⁻². During the irradiation, the attenuation of the absorption band at 410 nm as a result of the photodegradation of DPBF was collected periodically between 0 and 60 min. The first-order kinetics was assumed and the Φ_Δ were determined according to expression as follows:

2

where m means the slope of the time-dependent photodegradation curve and F is the correction factor for the difference in absorbance between sample and reference and is given as F = 1–10^−A. Since the Ф_Δ were too small, they were measured three times and presented as an average value with standard deviation.

Theoretical calculations

Geometry optimizations and electronic structure analyses were performed using DFT and TD-DFT. All calculations were carried out under methanol solvent conditions using the B3LYP functional of the Gaussian 16 program package and the 6-31G(d,p) basis sets with a dispersion correction (D3). The conductor-like polarizable continuum model (CPCM) was employed to account for the solvent effect of methanol in all calculations. To reduce computational cost, the peripheral triethylene glycol (TEG) chains of the PyBXI dyes were replaced with hydrogen atoms.

Transient absorption spectroscopy

Femtosecond transient absorption (TA) measurements with a time resolution of 200 fs were performed using a custom-built apparatus. A Ti:Sapphire regenerative amplifier system (800 nm, 350 µJ, 10 kHz, 35 fs, Spitfire Pro, Spectra Physics) was used as a fundamental laser source, which, first, were divided into two parts by a 5:5 (R:T) beam splitter. A half portion passed through optical parametric amplifier (OPA) (TOPAS, Spectra Physics) for a generation of pump pulse in the range 290–2600 nm. The other portion was used for the generation of probe continuum pulses by using a sapphire window. The time delay between pump and probe beams was controlled by a linear motor stage in the beamline for a probe generation stage. Spectra of the dispersed WLC probe are monitored by a high-speed spectrometer (Ultrafast Systems). Due to the limit of camera frame rate, we obtained TA spectra with a rate of 2 kHz. In each scan, we averaged at least 1500 times of TA spectra to secure an acceptable signal-to-noise ratio. To prevent polarization-dependent signals, the polarization angle of the pump pulse was set at the magic angle (54.7°) to the horizontally polarized probe pulse. With the optical Kerr signal measurements by n-hexane, cross-correlation FWHM in pump-probe experiments was estimated to be about 200 fs. After every experiment, steady-state absorption spectra were carefully checked, and we confirmed that there was no degradation of the sample.

Femtosecond TA measurements with a time resolution of 50 fs were performed using a Ti:Sapphire regenerative amplifier (Integra C, Quantronix). The fundamental output was divided into two beam paths. One portion was directed into a custom-built non-collinear optical parametric amplifier (NOPA) to generate broadband pump pulses ranging 550–750 nm. The NOPA output was compressed to sub-20 fs duration using a chirped mirror pair (GDD ≈ −40 fs² per bounce, 470–810 nm, Layertec) in combination with a fused silica wedge pair (Newport). Broadband probe pulses covering 450–1100 nm were produced by focusing a fraction of the fundamental into a 3 mm YAG crystal (EKSMA) to generate a white-light supercontinuum. The pump and probe beams were spatially overlapped on the sample in a non-collinear geometry, and the transmitted probe was detected by a CMOS-based spectrometer (Ocean FX, Ocean Optics). The pump beam was modulated using an optical chopper (MC2000, Thorlabs). All experiments were conducted in a 2 mm path length quartz cuvette (Stana), with the sample concentration adjusted to yield an optical density of 0.1–0.3 at 650 nm. The pump pulse energy was set to 80 nJ, which provided an optimal signal-to-noise ratio in the TA spectra.

Nanosecond TA measurements were conducted using an automated TA spectrometer (EOS, Ultrafast Systems) with photoexcitation at 420 nm provided by a diode-pumped Q-switched Nd:YAG laser and OPO (NT242, EKSPLA).

Cells and cell cultures

MCF-7 (human breast adenocarcinoma) and HeLa (human cervix adenocarcinoma) were supplied by the Korean Cell Line Bank. They were maintained in RPMI 1640 medium (Gibco, Carlsbad, CA, USA) with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin) (Welgene Inc., South Korea) at 37 °C in a humidified 5% CO₂ incubator.

Cell proliferation assay

MCF-7 and HeLa cells (2 × 10³ cells/well) were seeded in 96-well plates and incubated at 37 °C in 5% CO₂ for 24 h. Then the cells were treated with BODIPY dyes at different concentrations for another 24 h. The cell proliferation was measured according to the manufacturer’s instructions for CellTiter 96^® AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA). Then the absorbance was determined at 690 nm using an ELISA plate reader (Thermo Fisher Scientific, Inc., Waltham, MA, USA).

Confocal laser scanning microscopy

MCF-7 and HeLa cells were treated with BODIPY dyes (1 µM in DMSO) for 24 h. Then, they were further incubated with mitochondria-staining MitoTracker Green (MTG) (Invitrogen) for 45 min. Cells were then fixed with 4% paraformaldehyde for 10 min and permeabilized for 10 min with 0.1% Triton X-100 followed by counterstaining with 4′,6-diamidino-2-phenylindole (DAPI) for 1 h at room temperature. Finally, images of the cells were taken using confocal microscopy (LSM-700, Carl Zeiss, Germany).

Photodynamic therapy activity

MCF-7 and HeLa cells (2 × 10³ cells/well) were maintained similarly as described above but with further incubation of 2 h at 37 °C in 5% CO₂ under dark conditions. Then, the media were changed with phenol-red free RPMI 1640 followed by irradiation with a red LED (680 nm, 40 mW) for 30 min. The LED device was custom-built. The irradiation power was measured by placing the well plate vertically 30 cm above the LED source, and the PDT experiments were conducted under identical conditions. The cells were further incubated for 24 h and the cell proliferation (% of the control) was measured using the same method described above.

Statistical analysis

All data are expressed as the means ± standard deviations and were compared by one-way analysis of variance and Tukey’s test, using Prism GraphPad 6 software (San Diego, CA, USA). Group means were considered as significantly different at p < 0.05.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J.Y. and H.-J.K. directed the project. I.W.B. and D.K.M. conducted synthesis and characterization. C.K. conducted spectroscopic experiments, quantum mechanical calculations, and data analysis. J.O., T.H.H., and J.M.L. conducted femtosecond TA spectroscopy and data analysis. J.L. conducted in vitro assays and data analysis. J.J., S.B., and M.S. analyzed data. C.K., I.W.B., J.J., J.O., and J.Y. wrote the manuscript and all authors commented on the manuscript.

Funding

This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korea government (MSIT) (Nos. RS-2024-00351180, RS-2025-25418541, 2021R1A6A1A03039503, and RS-2024-00343229), Korea Basic Science Institute (National Research Facilities and Equipment Center) grant (2022R1A6C101B794), and Global - Learning & Academic research institution for Master’s·PhD students, and Postdocs (LAMP) Program of the National Research Foundation of Korea (NRF) grant funded by the Ministry of Education (No. RS-2023-00301914).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.