BODIPY-carbazole multichromophoric systems for green LED-assisted photodynamic therapy of breast cancer: synthesis, optoelectronic properties, DFT and antitumor studies†

Raquel Ledezma, Celín Lozano, Roberto Espinosa, Erick Alfonso, Gleb Turlakov, Geraldina Rodríguez, Rebeca Betancourt, Ronald F. Ziolo, Ivana Moggio* and Eduardo Arias*
Centro de Investigación en Química Aplicada, Boulevard Enrique Reyna No. 140, 25294 Saltillo, Coahuila, Mexico. E-mail: eduardo.arias@ciqa.edu.mx; ivana.moggio@ciqa.edu.mx

First published on 10th July 2025

---

---

Introduction

The literature on BODIPY covers a vast catalogue of derivatives of 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene, whose physicochemical properties can be modulated ad hoc depending on the type and position of functionalization. The boron dipyrrin core can be substituted at the α, β or meso position of the pyrrole ring or at the boron atom.¹ This chemical versatility has enabled the application of BODIPY derivatives in different areas such as lasers,² solar cells,³ OLEDs,⁴ fluorescent labelling for bioimaging or chemical sensing⁵ and photosensitizers (PSs) for PDT.⁶

PDT is a relatively recent antitumor treatment that involves a PS, light and oxygen. The most common mechanism through which the PS acts starts with the absorption of light at a specific wavelength—typically green visible light for BODIPYs. Upon excitation to a singlet excited state, a triplet excited state is accessible through intersystem crossing (ISC), and energy transfer from this state to molecular oxygen present in the tumour cells yields reactive oxygen species (ROS), which induce tumour cell death.⁷ According to this mechanism, the molecular requirements for an efficient PS are (1) solubility in polar solvents for biological assays; (2) no or very low cytotoxicity in the absence of excitation for healthy and tumour cells; (3) high cytotoxicity under excitation but only towards tumour cells; (4) high extinction coefficients at the excitation wavelength; and (5) efficient intersystem crossing. These properties restrict the vast catalogue of organic PSs, such as BODIPYs, which are usually soluble in aromatic or halogenated solvents and exhibit high fluorescence quantum yields, a property that favors de-activation of the first singlet excited state without crossing the triplet state (low ISC quantum yield). In this context, observation of low-energy triplet excited states in BODIPY dyes is not common.⁶ Nonetheless, strategies for the chemical modification of the BODIPY core have been proposed to address this limitation. Intersystem crossing can be promoted, in fact, through the classical approaches such as the heavy atom effect, most commonly by iodine substitution,⁸ or through dimerization⁹ or electron transfer.¹⁰ This last mechanism, known as spin–orbit charge-transfer induced intersystem crossing (SOCT-ISC), can occur in multichromophoric systems where BODIPY acts as the electron acceptor. This electron transfer can proceed either through bonds (TBCT) or through space (TSCT). Among the wide range of chromophores that have been combined with BODIPY, Cz is particularly attractive for PDT due to its strong electron-donating properties, which enable TBCT or TSCT. Both chromophores can be easily functionalized and exhibit good thermal, electrochemical and chemical stability. Although the combination of these two aromatic dyes has been investigated in the context of transistors,¹¹ dye-sensitized solar cells and organic electroluminescent diodes,¹² the current state of research on BODIPY-Cz systems for PDT remains limited. Chart 1 presents representative molecules relevant to the objectives of the present work.

Compounds I and II feature Cz directly functionalized at the meso-position of the BODIPY core,¹³ resulting in a mutual orthogonality that gives no changes in the absorption and emission maxima relative to BODIPY.¹ The fluorescence quantum yields of both compounds in CH₂Cl₂ are very low (0.0022 and 0.0026 for I and II, respectively), which is attributed to energy transfer from Cz to BODIPY upon carbazole excitation at 330 nm. Preliminary antitumor tests against human colon cancer HT29 cell lines showed greater cytotoxic activity for the derivative containing two carbazole units (compound II), but the authors did not provide an explanation for this observation. In ref. 14, SOCT-ISC was investigated by varying the functionalization between Cz and BODIPY. Compound III features a similar functionalization as I, with Cz directly attached to the meso position, while IV contains a phenyl spacer between the two chromophores. Conversely, Cz in V is directly bonded to BODIPY, as in III, but in the substitution occurs at the 2-pyrrol position. Photophysical, theoretical and electrochemical analyses revealed mutual orthogonality in III, while in IV and V, the dihedral angle between donor and acceptor units decreases to ≈50°. Although previous studies have indicated that the orthogonality between the electron donor and acceptor units is a prerequisite for efficient SOCT-ISC via TBCT,¹⁵ this condition is not met in compounds IV and V. Nevertheless, high ISC quantum yields were observed, particularly in polar solvents – an advantageous property that was explored in PDT studies against HeLa cells. In molecule VI, a parallel (face-to-face) conformation between Cz and BODIPY was demonstrated to enable ISC through TSCT mechanism, resulting in a higher intersystem crossing yield compared to the TBCT-based molecule VII.¹⁶

This representative literature survey clearly indicates that achieving ISC is not straightforward to predict as the photophysics of multidye systems can be quite complex. Consequently, the rational design of new photosensitizers for PDT remains a significant research challenge.

Building on our recent interest in BODIPY-multidyes for optoelectronic applications,^17,18 the objective of the present work is to investigate the chemical modulation of the photophysical/electrochemical properties of a series of BODIPY-Cz compounds shown in Chart 2. In this series, one BODIPY core is functionalized with one, two or three carbazole units substituted either at the pyrrole or meso-phenyl positions, with the aim of exploring different strategies to promote ISC and thus develop BODIPY-Cz-based PSs for PDT.

Experimental

Synthesis

Experimental procedures, and chemical and physicochemical characterization of each compound are given in the ESI.† Reactions were carried out at the local ambient pressure of 84.5 kPa.

Theoretical calculations

All of the theoretical calculations were performed using Orca, version 5.0.3,¹⁹ and Crest (Conformer–Rotamer Ensemble Sampling Tool),²⁰ which was used to generate all of the possible conformers for each studied BODIPY-Cz compound. First of all, each conformer was optimized using the BP86²¹ functional and the def2-SVP basis set. Then, the minimum energy structure was treated with the CAM-B3LYP functional²² using triple-zeta def2-TZVP,²³ in combination with the corresponding auxiliary def2/J²⁴ and def2-TZVP/C²⁵ basis sets, in chloroform, and using the Gaussian charge scheme with a VdW-type cavity.²⁶ Dispersion that includes D3BJ correction with Becke–Johnson damping was considered.²⁷ Scalar relativistic zeroth-order regular approximation (ZORA)²⁸ was applied for compound RL-5. For light atoms, H, C, O, B, F and N, Hamiltonian-specific recontractions of all-electron nonrelativistic Karlsruhe basis sets were used, whereas for I, a specially designed segmented all-electron relativistically contracted (SARC) basis set²⁹ was employed for both geometry optimization and TDDFT calculations, in combination with the corresponding auxiliary SARC/J³⁰ and def2-TZVP/C basis sets.

Analytical frequencies were computed for all optimized structures, and the absence of imaginary modes confirmed that true minima were obtained in all cases. The vertical excitation energies, transitional dipole moments, as well as triplet excited states were obtained using DLPNO-STEOM-CCSD on the DFT-optimized geometries, with one to six roots requested depending on the chemical nature of the studied compounds, where the vertical energy corresponds to HOMO–LUMO transition. Vibronic effects were computed at the Franck–Condon level using the ‘ESD’ module of ORCA, based on the STEOM excitation energy, transition dipole moment and DFT-calculated frequencies. A Gaussian line shape was used to simulate the absorption spectra, with a linewidth of 500 cm⁻¹. The excited state geometries as well as their respective Hessians were obtained via the ESD module using the vertical gradient approach, which assumes that the excited state (ES) Hessian is equal to that of the ground state (GS). This approach allows extrapolation of the ES geometry from the TD-DFT ES gradient and the GS Hessian. Excited-state analysis³¹ as well as electrostatic potential evaluation³² were performed using Multiwfn 3.8.³³ The EPM and the distribution of frontier molecular orbitals (FMOs) were plotted for the optimized geometries using the VMD package.³⁴

Equipment and methods

¹H (400 MHz) and ¹³C (100 MHz) NMR spectra were obtained at room temperature using a Bruker Advance III 400 MHz spectrometer, with CDCl₃ as solvent and the residual non-deuterated chloroform signal as the internal reference. MALDI-TOF mass spectra were acquired using a Bruker Autoflex Speed TOF/TOF operated in reflection mode. Positive ions were detected using DCTB as matrix and sodium trifluoroacetate as the cationized agent. The electrochemical properties of all of the molecules were investigated by cyclic voltammetry (CV) and square wave voltammetry (SWV) using a BASi electrochemical cell C-3 coupled to a potentiostat/galvanostat Squidstat Plus from Admiral Instruments. The system consisted of a 3-electrode cell: glassy carbon as the working electrode, platinum wire as the counter electrode and Ag/AgCl in 3M NaCl as the reference electrode (viability of −35 ± 20 mV vs. the calomel electrode). The measurements were performed at ambient temperature (T ∼25 °C) in previously distilled dichloromethane, bubbled with Ar for 5 min, using Bu₄NPF₆ (0.1 M) as the supporting electrolyte, which had been twice recrystallized in methanol. The sample concentration ranged from 0.5 to 1.0 mM. Cyclic voltammetry was performed with scan rates of 50, 100 and 200 mV s⁻¹ within the electrochemical window of −1.75 V to 1.85 V vs. the reference electrode. For the electrochemically reversible electron transfer processes, the standard redox potential (E₀) was estimated using the formal potential, defined as the average of the anodic and cathodic peak potentials obtained from the cyclic voltammetry measurements. The anodic and cathodic peaks were determined from the second derivative of the oxidative and reductive scans, respectively. The square wave voltammetry (SWV) measurements were carried out with a potential step of 4 mV, a pulsed amplitude of 25 mV and a frequency of 15 Hz. The main oxidation and reduction potentials in SWV were determined from the maxima of the oxidative and reductive scans. All redox potentials were calibrated against Ag/AgCl in 3M NaCl using Ferrocene/Ferrocenium (Fc/Fc⁺) as an internal reference at 100 mV s⁻¹ in CV (E⁰ = 0.41 V) and under the same conditions in SWV (E^1/2 = 0.43 V).

Photophysical properties were analysed in acclimatized rooms (17 ± 1 °C) using spectroscopic-grade chloroform. UV-Vis spectra were recorded on an Agilent Cary 6 UV/Vis spectrophotometer. The molar extinction coefficient (ε) was calculated from the slope of the absorbance vs. molar concentration plot using at least four different solutions. Fluorescence spectra were recorded using a Horiba PTI QuantaMaster QM-8450-22-C spectrofluorometer equipped with an integrating sphere for the determination of the absolute quantum yields. Spectra were obtained with background correction and with consistent slits and detector bias between the sample and the solvent. Slit widths were adjusted to ensure that the uncorrected spectra remained within the linear detection range (maximum 10⁶ counts). The excitation wavelength was set 10 nm below the main absorption maximum, and the corresponding absorbance was kept below 0.1. For each compound, at least four solutions were analysed, and the quantum yield was averaged and reported as ϕ. Stokes’ shift (Δν) values were calculated from the absorption and fluorescence maxima in wavenumbers. Average fluorescence lifetimes (τ) were determined using time-correlated single photon counting (TCSPC) on the same instrument, with a nanoLED laser at a wavelength close to the excitation used for the emission spectra. Data fitting was performed using the FelixGX software provided with the instrument. Radiative (k_rad) and non-radiative (k_nr) rate constants were calculated using the following equations: k_rad = ϕ/τ and k_nr = (1 − ϕ)/τ. A 0.01% suspension of Ludox AS40 (Aldrich) in ultrapure water was used for the prompt signal. Calibration of the equipment was performed using a POPOP [4-bis(4-methyl-5-phenyl-2-oxazolyl)benzene] solution in methanol (optical density 0.1, lifetime 0.93 ns).³⁵ Singlet oxygen generation was determined in DMSO by adapting the method reported in ref. 36. Briefly, solutions of 1,3-diphenylisobenzofuran (DPBF, 50 μM) and of the BODIPY RL-4 and RL-5 (2.5 μM) were prepared in aerated DMSO (20 min) and stored in the dark until use. A volume of 1 mL of the PS solution was added to 10 mL of the DPBF solution, and the mixture was immediately transferred into a cuvette placed in the UV-vis spectrophotometer. The solution was irradiated for 10 minutes using a green LED (532 nm, LED LE RTDUW S2WP, Osram) positioned at the open top of the cuvette. A voltage of 3 V was applied using a programmable power supply (GW PPT-3615 GPIB). The absorbance of DPBF at 418 nm was monitored at fixed intervals using the scanning kinetic mode of the instrument. The natural logarithm of the absorbance ratio, ln(A/A₀), where A₀ is the initial absorbance, was plotted against the irradiation time for the sample and Rose Bengal (RB) as the standard. The relative singlet-oxygen generation rate (ϕ_Δ) of each photosensitizer was determined by the ratio of the corresponding slopes of the fitted curves. A negative control was also analysed by adding 1 mL of DMSO to the DPBF stock solution.

Biological assays

Results and discussion

Synthesis

Theoretical study

Since the stabilization of excited triplet states of BODIPY-carbazole compounds is crucial for PDT, DFT calculations were carried out to identify the best candidate. A meso phenyl BODIPY substituted with methyl groups at the 1,3,5,7 positions was selected as the core because phenyl in the meso position is orthogonal to the BODIPY plane due to steric hindrances, as shown in Fig. 1 and Fig. S3 (ESI†). This orthogonality is retained when Cz is substituted at the 2- or 2,6-position(s). However, once the BODIPY is substituted with Cz at both 2,6-positions, the dihedral angle between Cz and BODIPY ranges from 52° to 54° (Fig. 1). The absence of sterically hindered substituents at the 2,4-positions of Cz allows for free rotation of the Cz moiety.

Fig. 2 shows the frontier molecular orbitals of the studied molecules. For RL-3, the HOMO orbitals are localized on Cz because of its electron-donating character, while the LUMO orbitals are centred on the BODIPY core, reflecting its electron-accepting character. For the rest of the molecules, the HOMO orbitals are distributed over both Cz and BODIPY, indicating extended conjugation. However, the LUMO orbitals are mainly located in the BODIPY core. This graphical representation confirms that both aryls are electronically interacting, as previously observed in similar Cz-BODIPY dyads linked via ethynylene bonds.⁴⁰ Although those molecules exhibit more extensive conjugation because of the coplanarity induced by the ethynylene linker, their frontier molecular orbital distribution follows the same trend observed in the present study. It is worth mentioning that in the present series, carbazole is bonded to the BODIPY at its meta position. It would be interesting to synthesize and investigate homologues with carbazoles attached at different positions on the BODIPY core, as variations in the coplanarity are expected to affect the photophysical properties. For instance, Liao et al. investigated carbazole derivatives bonded at 3 (para) or 2 (meta) positions of the BODIPY cores substituted with cyanoacetic acid at 2 and 6 positions for organic solar cell applications. The authors found that the 2-linked molecules exhibited slightly more coplanar backbones compared to their 3-linked analogues, which resulted in more efficient intramolecular charge transfer and thus higher current densities in the devices.⁴¹ Hou et al.¹⁴ reported a dihedral angle of 90° for compound IV and 55° for compound V. These compounds feature carbazole substitution at its 3 position with BODIPY. The impact of this difference can be seen when comparing meso-BODIPY-substituted compound IV to RL-3. Therefore, the final geometry results from a combination of steric hindrance imposed by the different aryl substituents and the specific linking position. We noticed that the HOMO and LUMO energy levels and orbital distributions of RL-7 are nearly identical to those of RL-6, suggesting that the third Cz moiety at the meso phenyl position is electronically isolated and does not contribute to the conjugation. This implies that from RL-4, RL-6 to RL-7, the increase in HOMO energy correlates with the number of carbazole units participating in conjugation. For RL-5, the HOMO energy lies between that of RL-3 and the rest of the series, likely due to the presence of an iodine atom, which also significantly lowers the LUMO energy. Theoretical excitation properties were determined using a hybrid computational approach that combined STEOM-DLPNO-CCSD vertical excitation energies with a standard TDDFT method to simulate a vibronic-like absorption spectra at ground state (GS) geometry.

Table 1 presents the excited state (ES) properties of the studied series, where the vertical energy E_0–0 and corresponding λ_abs show good agreement with the experimental spectroscopic results (vide infra), as well as consistency in trends across the series. The simulated spectra (Fig. S4, ESI†) display excitonic features with vibronic replicas, where the peak intensity and wavelength of the excitonic band depend on the number of Cz units and their positions with respect to the BODIPY core.

---

Geometry relaxation of RL-5, studied using TDDFT, revealed that the dihedral angle between the donor and acceptor changes from 54° in the GS to 41° in the ES, resulting in a more planar conformation that facilitates charge transfer, as further confirmed by electrochemical analysis (vide infra).

Hole–electron analysis was also conducted to corroborate the formation of a charge transfer state in RL5. Table 2 presents data of the first three excitations of RL-5, where the D index quantifies the distance between the hole and electron, and the t index indicates the degree of separation. For the first excitation, i.e. S₀ → S₁, the D index is 0.509 Å, while for the other excitations, the D values are significantly higher.

---

The nature of the excitation can be monitored by the increasing values of the D index. Specifically, for S₀ → S₂ excitation, the excitation is classified as a hybrid local charge transfer (HLCT),⁴² while for S₀ → S₃, it corresponds to a highly charge transfer state.

This behavior can be better visualized through the hole–electron distribution map (Fig. 3a). In the S₀ → S₁ excitation, the barycenters of the positive and negative moieties are fully localized on the BODIPY core, while in the S₀ → S₂ and S₀ → S₃ excitations, this distribution becomes increasingly delocalized. The t index is also useful for distinguishing these excitations: the S₀ → S₁ excitation exhibits a negative t value, indicating a low probability of charge separation. In contrast, the positive and increasing t values for the S₀ → S₂ and S₀ → S₃ excitations are consistent with a hybrid local (S₀ → S₂) and strongly charge transfer (S₀ → S₃) excited state.

It is usually accepted that one of the mechanisms that explains the triplet generation involves the orthogonality between the donor and acceptor moieties, known as the SOCT-ISC.¹⁵ However, this is not always the case.¹⁴ As shown in Fig. 1, none of the optimized geometries exhibit complete orthogonality between the BODIPY and Cz units, although the structures are not fully coplanar either.

In order to investigate the possibility of other orthogonal geometries that could be energetically close to the minima and thus accessible, a detailed conformational study was carried out on RL-5 and RL-3 as representative molecules. Fig. 3b shows the energy surface scan versus the dihedral angle θ between the BODIPY plane and the meso substituent. For RL-5, two energy minima were found at ∼50° and 130°, both of which correspond to conformations unfavourable for triplet generation. In contrast, the minimum potential energy for RL-3 is observed at 80–90°; however, triplet generation remains insufficient because Cz is not directly bonded to the BODIPY core, but rather through the meso-phenyl group.

According to ref. 43, the spin density on atoms linking the electron donor and the acceptor aryl groups in dyads is another important parameter for estimating the likelihood of SOCT-ISC. In our series, the spin density of the 1,3,5,7-tetramethyl BODIPY (RL-1) is higher at the meso position and lower at the 2,7-positions, which predicts a correspondingly low ISC yield. SOC and electron excitation analyses were carried out to better understand the intersystem crossing and the presence of the local or intramolecular charge transfer states. All SOC values were calculated at S₀-optimized geometry for RL-4 and RL-5 (Table S1, ESI†). From that table and Fig. 4, it can be observed that the energy gaps between S₁ and T₁ are relatively large for both RL-4 and RL-5, while the gaps are smaller for the S₁–T_2,3 transitions. This behaviour is further confirmed by the higher SOC values observed.

The larger SOC values for RL-5 clearly indicate a more probable and efficient triplet formation, attributed to the iodine atom incorporation. It is well known that the formation of ³CT state requires a very small singlet–triplet energy gap,⁴⁴ as occurs in the radical-pair intersystem crossing (RP-ISC) mechanism. However, the energy gap of the studied RL-4 and RL-5 dyads is rather large. Thus, based on all of these analyses, it can be argued that ISC is only possible via SOC between ¹LE and ³LE states. For this reason, RL-5 is considered the most interesting molecule for evaluating singlet oxygen generation in the context of PDT.

Electrochemical properties

Fig. 5 and Fig. S5 (ESI†) show the voltammograms of some representative BODIPYs from the series, while Table 3 summarizes all the determined electrochemical parameters. Successive cycles (Fig. S7, ESI†) did not show any change in the voltammograms, indicating that diffusion-controlled transport is the dominant process. The voltammograms of BODIPY RL-1 and Cz RL-2 were included as references. RL-1 displays one-electron reduction and one-electron oxidation reversible waves with a large peak separation of ΔE = 2.50 V, determined from the respective formal potentials, whereas RL-2 shows a typical.

---

One-electron reversible oxidation wave (E° 1.29 V) and a second one-electron reversible oxidation peak due to the formation of 3,3′-bicarbazole (E° = +0.82 V). As N is substituted with an alkyl chain, the 9,9′-bicarbazyl formation was not detected. No reduction waves were observed due to the electron-donating character of Cz.

From the voltammograms of the series (Fig. 5a), we observed that: (i) Each molecule exhibits the characteristic redox features of both BODIPY and Cz. The number of electrons participating in each molecule depends of the number and position of the Cz donor units. (ii) When Cz is substituted on the phenyl in meso position (RL-3), the voltammogram almost matches the sum of the individual components. This suggests a lack of electronic interaction between the two chromophores, as has been found for other systems,¹⁷ and is consistent with theoretical and spectroscopic results (vide infra). However, a slight increase in the width of the waves is observed and attributed to the possible dimerization of the molecule (Fig. S6, ESI†). This may occur because of the tendency of Cz radicals to couple with another Cz radical, mainly at the 3,6- and, less frequently, at the 1,8-positions of the Cz. (iii) When Cz is substituted at the 2-position (RL-4), two broad and diffuse oxidation bands are observed, whereas only one is observed in the reduction process. Electron transfer from Cz to BODIPY appears to be a fast process because distinct bands are only resolved at high scan rates (∼200 mV s⁻¹, Fig. S7, ESI†). In the cathodic region, the band appearing at −1.32 V is attributed to reversible one-electron reduction of the BODIPY. However, in the case of the addition of an iodine at the 2-position, RL-5, the bands display a fully reversible reduction potential of the BODIPY at E₀ = −1.13 V, independently of the scan rate because of the electron-donating effect of the iodine. In the anodic region, the first diffuse band, attributed to the quasi-reversible oxidation of the BODIPY, is shifted to E₀ = 1.21 V, while the second band remains at the same potential with a broadening similar to that observed for RL-4. To understand the widening of these bands, the second derivative (d²J/dV²) of the cyclic voltammograms was applied (Fig. 5c); moreover, SWV was also carried out (Fig. S8, ESI†). These studies corroborated the presence of two peaks in the anodic region, owing to the participation of one electron from Cz and one electron from BODIPY. In the cathodic region, only one reversible reduction electron wave associated with the BODIPY was identified. (iv) The difference in the anodic region between two Cz-substituted (RL-6) and three Cz-substituted (RL-7) BODIPY core, compared to RL-4, lies in Cz-associated band that exhibits higher current intensity. Additionally, this band becomes broader and more anodically shifted because of the greater number of electrons participating, compared to the BODIPY-associated band, which remains at ∼1.0 V. In contrast, the reduction band tends to disappear for RL-7, suggesting that the negative charge is stabilized when BODIPY core is substituted with three Cz units. (v) The ΔE value, which is related to the band gap, decreases with the number of Cz units in the molecule because of the extension of conjugation throughout the structure. (vi) Overall, the electrostatic potential maps (EPM) shown in Fig. 5b and Fig. S9 (ESI†) indicate that the anion is mainly localized on the BODIPY and the carbazole units, while the cation is primarily centred on the BODIPY. This explains why two oxidation bands and one reduction band are observed in the CV curves. (vii) Simulated oxidation–reduction potentials are similar to those of experimental values determined at the half-wave redox electrochemical potentials.

Photophysical properties

Table 4 summarizes the photophysical properties of the BODIPY-Cz compounds in chloroform, while Fig. 6a and b show the corresponding absorption/fluorescence spectra.

---

In general, all compounds exhibit a main absorption peak (Fig. 6a), attributed to the S₀ → S₁ electronic transition at around 502-540 nm, similar to the BODIPY precursor RL-1 and in agreement with the literature.¹ A redshift of this peak, from 502 nm of RL-1 to 517 nm for RL-4 and to ∼540 nm for RL-6 and RL-7, reveals that carbazoles substituted at 2- and 2, 6-positions electronically interact with the BODIPY. This interaction extends the conjugation throughout the structure and consequently reduces the optical band gap E_g slightly from 2.4 eV to 2.1 eV. The introduction of an iodine atom at the 6-position of the BODIPY in RL-5 imparts a bathochromic shift of 18 nm compared to its analogue RL-4 because of the positive inductive effect of the halogen attaining the same absorption maximum as RL-6. When Cz is substituted at the meso-phenyl RL-3, no significant change is observed in the peak position of 504 nm, indicating that electronic interaction is weak or null, as confirmed by theoretical calculus, voltammetry and in the literature.¹⁴ The extinction coefficient ε is on the order of 10⁴ M⁻¹ cm⁻¹ for the entire series as typical for BODIPYs. Interestingly, RL-3 exhibits the highest values. Similar trends have been reported in the literature. For instance, Hou et al.¹⁴ reported an ε value of 8.7 × 10⁴ M⁻¹ cm⁻¹ in DCM for the meso functionalized compound IV, compared to 5.9 × 10⁴ M⁻¹ cm⁻¹ for compound V. This behavior can be explained by the negligible electronic interaction between BODIPY and carbazole in RL-3, where Cz functions merely as an electron-donating substituent (rather than as a conjugated moiety), leading to a hyperchromic effect on the BODIPY electronic transition. Additional peaks observed in the UV region can be ascribed to electronic transitions of Cz, as in the corresponding reference spectra, and/or to S₀ → S_n excitations of BODIPY. The half-height bandwidth (HHBW_abs) of the main absorption maxima increases correspondingly with the conjugation extension. RL-3 has a similar value of 20 nm than RL-1, which means that BODIPY is not electronically interacting with Cz in RL-3. However, in RL-4, where Cz is substituted at the 2-position, the HHBW_abs increases to 28 nm, and further to ∼50 nm for RL-6 and RL-7. This trend reflects the greater number of rotamers arising from the free rotation around the single bonds of the BODIPY-Cz, although the charge transfer character of the electronic transition cannot be ruled out, as evidenced by theoretical studies and supported by the literature.¹⁴

In general, the features observed in the absorption properties are retained in the fluorescence spectra (Fig. 6b). The meso-substituted BODIPY RL-3 presents a sharp (HHBW_emis of 25 nm) and structured emission band with features that are mirror-like to the S₀ → S₁ excitation, with a maximum at 516 nm, typical of BODIPYs and similar to the reference compound RL-1.

Consistently, the Stokes shift value (Δν ∼ 460 cm⁻¹) falls within the range of molecules that do not undergo significant geometrical changes upon excitation. Indeed, RL-3 exhibits photophysical properties very similar to those of RL-1, but markedly different from RL-2, including a fluorescence quantum yield ϕ ∼ 0.6, lifetimes values τ = 2.5 ns, Fig. S10 (ESI†), and radiative k_rad (0.2 ns⁻¹) and non-radiative k_nr (0.1 ns⁻¹) constants corroborating that phenyl-Cz is isolated from the BODIPY. Conversely, for the BODIPYs bearing Cz units at the 2- and 2,6-positions, the emission peak becomes broader and unresolved, with a redshifted maximum. This shift increases with the number of carbazoles substituents, i.e. RL-4 ≪ RL-6 = RL-7. As also observed in the absorption spectra, the iodine atom in RL-5 imparts a redshift of 11 nm in the emission relative to RL-4. The Stokes shift values for these four compounds are also significantly larger compared to the reference RL-1 and to the meso phenyl-Cz RL-3. The average value Δν of ∼1700 cm⁻¹ (1450 cm⁻¹ for RL-5) indicates a change of the geometry from aromatic to quinoid, from the ground to the excited state value. As expected, the fluorescence quantum yield is similar between RL-3 and the reference RL-1 (ϕ ∼0.6), as well as for RL-6 and RL-7. The value obtained for RL-1 is in agreement with those reported for this molecule by other authors.^14,45 For that series, the deviation from the unity was ascribed to losses due to the free rotation of the meso-phenyl around the single bond to the BODIPY core, albeit to a lesser extent than in less substituted homologues.² This explanation is also applicable to the present series, as the theoretical studies suggest free rotations of the carbazole unit(s). Indeed, internal conversion losses are likely not significant, as RL-6 and RL-7 exhibit much larger Stokes shift than RL-3, but comparable ϕ values. Fluorescence resonance energy transfer between (photon) donor and (photon) acceptor moieties may occur in the studied molecules, considering that both BODIPY and carbazole units could act as (photon) donors or acceptors. Under carbazole excitation, energy transfer from carbazole to BODIPY is possible, as previously observed by Sengul et al.¹³ However, for the present study, where PDT is expected under green irradiation, the excitation was performed at the BODIPY absorption maximum. In this case, only BODIPY absorbs and can act as a photon donor, but its emission is at 514 nm, where carbazole does not absorb. However, some degree of overlap between the absorption and emission of the same molecule, i.e. self-absorption, is possible. Electron transfer from the donor Cz to the electron-withdrawing BODIPY occurs, as stated by theoretical calculations and electrochemistry. Intersystem crossing is also another possible de-activation pathway, eventually through spin–orbit charge transfer, as has been demonstrated in the literature for molecules III–V.¹⁴ ISC is most likely to be important for RL-5 because of the heavy atom effect derived from the iodine, as confirmed by the theoretical calculations. In order to support this hypothesis, singlet oxygen generation was investigated for RL-4 and RL-5. For this study, DMSO was used as the solvent, but no relevant changes were found in the photophysical properties between chloroform and DMSO (Fig. S11 and Table S2, ESI†).

Fig. 7 shows the plot of the absorbance decrease ln(A/A₀) at 418 nm for DPBF in the presence of RL-4, RL-5 or Rose Bengal (RB) as the standard under green light irradiation. DPBF alone is stable under the LED irradiation, as can be observed in the control experiment over 10 minutes. In the same time regime, the absorbance of DPBF aerated solution decreases in the presence of the photosensitizer because of the photo-oxidation of DPBF promoted by the singlet oxygen generation. The slope of the fitted curve relative to RB gives a singlet oxygen generation yield (ϕ_Δ) of 20% for RL-4 and 60% for RL-5, which is in agreement with the much lower fluorescence quantum yield of RL-5. The ϕ_Δ value of RL-5 is quite high, considering that with just one iodine atom, it is close to the value obtained in ref. 46 for 2,6-diiodinated meso-phenyl BODIPY, which suggests its application as PS in PDT.

Biological properties

Reactive oxygen species (ROS), a byproduct of aerobic respiration, can be toxic to DNA, proteins and lipids, and may even be lethal. However, ROS can also be essential, as many responses to ROS are important for normal physiology and the development of several diseases, including cancer. ROS affect cancer cells and stromal components of the tumour to regulate cancer progression and survival.⁴⁷

Based on the promising singlet oxygen generation of RL-5, it was analysed as a photosensitizer for photodynamic therapy by fixing the excitation wavelength for activating the antitumor property at 532 nm, close to its absorption maximum. Before evaluation, the effect of irradiation alone on the cells was assessed as an internal control. As shown in Fig. S12 and S13 (ESI†), LED treatment did not affect the viability of either healthy fibroblasts (1132SK) or breast cancer cells (MDA-MB-231), in agreement with other reports on keratinocytes NHL and murine fibroblasts NIH-3T3.^48,49

In vitro cytotoxicity was evaluated on four cell lines: breast cancer MDA-MB-231 (Fig. 8a), lung cancer A549 (Fig. 8b), healthy human fibroblasts 1132SK (Fig. 8c), and murine fibroblasts NIH-3T3 (Fig. 8d). In all cases, experiments were conducted under green LED irradiation, either in presence of RL-5 (blue squares) at concentrations ranging from 0 to 20 μM³⁶ or without RL-5 (black circles). For both human tumour cell lines, a toxic effect on cell viability was observed following irradiation with RL-5 at low concentrations (Fig. 8a and b). In particular, cell viability at just 0.5 μM of RL-5 was 85.3%, and the antitumor effect increased with the concentration, reducing cell viability to 44.87% at 10 μM, and to 11.78% at 20 μM. Care must be taken to not exceed 20 μM of RL-5 because a significant toxic effect was observed in the cells even without irradiation, with cell viability decreasing to 66.38%. In contrast, no relevant effect was observed between 0.5 and 10 μM. Treatment of lung cancer cells proved to be less sensitive to RL-5-irradiation, with 57.3% of cell viability at a 20 μM dose, and RL-5 was poorly toxic without irradiation, with cell viability of ∼78%.

We found that healthy human 1132SK fibroblasts were not affected by the treatment, with or without irradiation, in concentrations between 0.5 and 1 μM of RL-5. In contrast, at a dose of 10 μM under irradiation, the cell viability was 85.97%, and it dropped to 49.14% at 20 μM. This concentration is quite high for standard medical treatment, thus opening the opportunity to use RL-5 in phototherapy at low doses. In the case of murine fibroblasts NIH-3T3, which are frequently used in the literature as healthy control cell lines, a strong cytotoxic effect of RL-5 was observed at even a low dose (1 μM), with a cell viability of 65.23%.

In the literature, BODIPYs with various functionalization have been tested on distinct tumour cell lines and under different excitation conditions and treatment protocols, making strict comparisons difficult because of variations in one or more parameters. In ref. 8, the authors synthesized a series of 2,6-diiodinated BODIPYs that exhibited exceptionally high cytotoxicity, with IC₅₀ values in the range of 1–2 nM under green LED excitation conditions, similar to our series. However, those BODIPYs were not functionalized with carbazole, and the tumour cells used were different, being human ovarian carcinoma cell line SKOV3. In the work reported in ref. 12, as mentioned in the introduction, compound II, a BODIPY-Cz analogue of RL-5, showed an IC₅₀ of 8.3 ng mL⁻¹, but the test was performed on human colon cancer HT29 cell lines and without light excitation. Overall, our findings demonstrate that BODIPY-Cz RL-5 exhibits a differential cytotoxic effect between healthy and tumour cell lines, with high cytotoxic effect at low doses (0.5–10 μM), after 24 h of treatment and only 2 h of irradiation. We suggest that an even greater cytotoxic effect could be achieved with prolonged exposure times beyond 24 h and/or by modifying the irradiation parameters; however, this is work still in progress.

Conclusions

In the present work, we report the synthesis of a series of BODIPYs bearing one, two or three carbazole units at the 2- or 2,6- or 2,6, 8-(meso)phenyl positions to investigate the effect of the chemical modulation on the ability to promote intersystem crossing for potential applications in photodynamic therapy. Electron transfer from Cz to the BODIPY core, which can facilitate ISC via the spin–orbit charge-transfer induced intersystem crossing (SOCT-ISC) mechanism, occurs only when Cz is substituted at the 2- or 2,6-positions. In contrast, when Cz is attached at the meso-phenyl position, both Cz and BODIPY behave as two electronically isolated chromophores because of their mutual orthogonality. Electrostatic potential maps and cyclic voltammetry measurement of the series revealed that the molecules exhibit donor–acceptor (push–pull) character, where the anion is primarily localized on both BODIPY and Cz, while the cation is centred only on the BODIPY core. Nevertheless, theoretical calculations showed that the dihedral angle between the two chromophores is 50° and 130°, deviating from the orthogonality typically required for efficient SOCT-ISC. Moreover, the SOC values and the electron excitation analysis suggest that intersystem crossing is feasible only for RL-5, where SOC is promoted by the heavy iodine atom between ¹LE and ³LE. Singlet oxygen generation was confirmed for RL-5, with a quantum yield of 0.6, remarkably high, considering that the molecule contains only one iodine atom. Based on these findings, RL-5 was selected for PDT studies on breast cancer (MDA-MB-231), lung cancer (A549), and healthy human fibroblasts (1132SK) and murine fibroblasts (NIH-3T3) as control, following 24 h of treatment and 2 h of green LED light exposure. RL-5 exhibited a differential cytotoxic effect between healthy and tumour cell lines, showing high cytotoxicity at low doses (0.5 to 10 μM) after 24 h of treatment and just 2 h of irradiation. Notably, the MDA-MB-231 breast cancer cells were the most sensitive to the reactive oxygen species generated by RL-5. Even at a low dose (0.5 μM), cell viability was 85.3%. The antitumor effect increased with concentration, resulting in cell viabilities of 44.87% at 10 μM and 11.78% at 20 μM. Although strict comparisons with other BODIPY-based PDT studies are limited because of differences in chemical functionalization, tumour cells and treatment conditions, these assays demonstrate the potential of RL-5 to be used at low doses for antitumor therapy.

Author contributions

Raquel Ledezma: investigation; Celín Lozano: investigation; Roberto Espinosa: investigation and formal analysis; Erick Alfonso: investigation; Eduardo Arias: conceptualization, formal analysis, writing and project administration; Ivana Moggio: conceptualization, formal analysis, writing and project administration; Gleb Turlakov: investigation and writing; Geraldina Rodríguez: investigation; Rebeca Betancourt: writing review; Ronald F. Ziolo: project administration and writing review.

Conflicts of interest

There are no conflicts to declare.

Data availability

Some of the data supporting this article have been included as part of the ESI,† and other data are available upon request.

Acknowledgements

We wish to acknowledge the U.S. Air Force Office of Scientific Research for the grant FA9550-23-1-0271, to IPICyT for the use of the Thubat Kaal II cluster of the National Supercomputing Center for theoretical calculations and to Maricela Garcia and Teresa Rodríguez for their technical help. R. E. thanks the CIQA's Cellular growth analytical laboratory.

References

1. A. Loudet and K. Burgess, Chem. Rev., 2007, 107, 4891–4932,  DOI:10.1021/cr078381n .
2. I. Esnal, I. Valois-Escamilla, C. F. A. Gómez-Durán, A. Urías-Benavides, M. L. Betancourt-Mendiola, I. López-Arbeloa, J. Bañuelos, I. García-Moreno, A. Costela and E. Peña-Cabrera, ChemPhysChem, 2013, 14, 4134,  DOI:10.1002/cphc.201300818 ; A. Prakash, J. C. Janardhanan, V. K. Praveen, P. Radhakrishnan and A. Mujeeb, J. Luminescence, 2022, 252, 119343,  DOI:10.1016/j.jlumin.2022.119343 .
3. B. M. Squeo, L. Ganzer, T. Virgili and M. Pasini, Molecules, 2021, 26, 153,  DOI:10.3390/molecules26010153 .
4. D. Ma, G. Zhao, H. Chen, R. Zhou, G. Zhang, W. Tian, W. Jiang and Y. Sun, Dyes Pigm., 2022, 203, 110377,  DOI:10.1016/j.dyepig.2022.110377 ; X. Song, D. Zhang, Y. Zhang, Y. Lu and L. Duan, Adv. Opt. Mater., 2020, 8, 2000483,  DOI:10.1002/adom.202000483 .
5. P. Kaur and K. Singh, J. Mater. Chem. C, 2019, 7, 11361–11405,  10.1039/C9TC03719E .
6. A. Kamkaew, S. Hui Lim, H. Boon Lee, L. Voon Kiew, L. Yong Chung and K. Burgess, Chem. Soc. Rev., 2013, 42, 77–88,  10.1039/c2cs35216h .
7. D. E. J. G. J. Dolmans, D. Fukumura and R. K. Jain, Nat. Rev. Cancer, 2003, 3, 380–387,  DOI:10.1038/nrc1071 .
8. E. Caruso, M. Gariboldi, A. Sangion, P. Gramatica and S. Banfi, J. Photochem. Photobiol., B, 2017, 167, 269–281,  DOI:10.1016/j.jphotobiol.2017.01.012 .
9. Y. Cakmak, S. Kolemen, S. Duman, Y. Dede, Y. Dolen, B. Kilic, Z. Kostereli, L. Tatar Yildirim, A. Lale Dogan, D. Guc and E. U. Akkaya, Angew. Chem., Int. Ed., 2011, 50, 11937–11941,  DOI:10.1002/anie.201105736 .
10. M. A. Filatov, S. Karuthedath, P. M. Polestshuk, H. Savoie, K. J. Flanagan, C. Sy, E. Sitte, M. Telitchko, F. Laquai, R. W. Boyle and M. O. Senge, J. Am. Chem. Soc., 2017, 139, 6282–6285,  DOI:10.1021/jacs.7b00551 .
11. S. Wanwong, P. Khomein and S. Thayumanavan, Chem. Cent. J., 2018, 12, 60,  DOI:10.1186/s13065-018-0430-5 .
12. I. Gupta and P. E. Kesavan, Front. Chem., 2019, 7, 841,  DOI:10.3389/fchem.2019.00841 .
13. I. F. Sengul, E. Okutan, H. Kandemir, E. Astarcı and B. Çoşut, Dyes Pigm., 2015, 123, 32–38,  DOI:10.1016/j.dyepig.2015.07.025 .
14. Y. Hou, I. Kurganskii, A. Elmali, H. Zhang, Y. Gao, L. Lv, J. Zhao, A. Karatay, L. Luo and M. Fedin, J. Chem. Phys., 2020, 152, 114701,  DOI:10.1063/1.5145052 .
15. Z. E. X. Dance, S. M. Mickley, T. M. Wilson, A. B. Ricks, A. M. Scott, M. A. Ratner and M. R. Wasielewski, J. Phys. Chem. A, 2008, 112, 4194–4201,  DOI:10.1021/jp800561g .
16. H. Liang, M. Lu, Z. Mahmood, Z. Li, Z. Chen, G. Chen, M.-D. Li, Y. Huo and S. Ji, Angew. Chem., Int. Ed., 2023, 62, e202312600,  DOI:10.1002/anie.202312600 .
17. J. R. Flores, G. Castruita-De León, G. Turlakov, E. Arias, I. Moggio, S. M. Montemayor, R. Torres, R. Ledezma, R. F. Ziolo and J. González-Torres, Chem. – Eur. J., 2021, 27, 2493–2505,  DOI:10.1002/chem.202004481 .
18. J. L. Vázquez, I. Velazco-Cabral, M. Flores-Álamo, G. Turlakov, G. Rodríguez, I. Moggio, E. Arias, E. Peña-Cabrera and M. A. Vázquez, Chem. – Eur. J., 2022, 28, e202202446,  DOI:10.1002/chem.202202446 .
19. F. Neese, F. Wennmohs, U. Becker and C. Riplinger, J. Chem. Phys., 2020, 152, 224108,  DOI:10.1063/5.0004608 .
20. P. Pracht, F. Bohle and S. Grimme, Phys. Chem. Chem. Phys., 2020, 22, 7169–7192,  10.1039/C9CP06869D .
21. A. D. Becke, Phys. Rev. A: At., Mol., Opt. Phys., 1988, 38, 3098,  DOI:10.1103/PhysRevA.38.3098 .
22. T. Yanai, D. P. Tew and N. C. Handy, Chem. Phys. Lett., 2004, 393, 51–57,  DOI:10.1016/j.cplett.2004.06.011 .
23. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297–3305,  10.1039/B508541A .
24. F. Weigend, Phys. Chem. Chem. Phys., 2006, 8, 1057–1065,  10.1039/B515623H .
25. A. Hellweg, C. Hattig, S. Hofener and W. Klopper, Theor. Chem. Acc., 2007, 117, 587–597,  DOI:10.1007/s00214-007-0250-5 .
26. M. Garcia-Ratés and F. Neese, J. Comput. Chem., 2020, 41, 922–939,  DOI:10.1002/jcc.26139 .
27. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104 CrossRef CAS PubMed ; S. Grimme, S. Ehrlich and L. Goerigk, J. Comput. Chem., 2011, 32, 1456–1465,  DOI:10.1002/jcc.21759 .
28. E. van Lenthe, E. J. Baerends and J. G. Snijders, J. Chem. Phys., 1993, 99, 4597–4610,  DOI:10.1063/1.466059 .
29. J. D. Rolfes, F. Neese and D. A. Pantazis, J. Comput. Chem., 2020, 41, 1842–1849,  DOI:10.1002/jcc.26355 .
30. D. A. Pantazis and F. Neese, J. Chem. Theory Comput., 2009, 5, 2229–2238,  DOI:10.1021/ct900090f .
31. Z. Liu, T. Lu and Q. Chen, Carbon, 2020, 165, 461–467,  DOI:10.1016/j.carbon.2020.05.023 .
32. J. Zhang and T. Lu, Phys. Chem. Chem. Phys., 2021, 23, 20323–20328,  10.1039/D1CP02805G .
33. T. Lu and F. Chen, J. Comput. Chem., 2012, 33, 580–592,  DOI:10.1002/jcc.22885 .
34. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38,  DOI:10.1016/0263-7855(96)00018-5 .
35. S. A. El-Daly, S. A. El-Azim, F. M. Elmekawey, B. Y. Elbaradei, S. A. Shama and A. M. Asiri, Int. J. Photoenergy, 2012, 2012, 458126,  DOI:10.1155/2012/458126 .
36. S. Banfi, G. Nasini, S. Zaza and E. Caruso, Tetrahedron, 2013, 69, 4845–4856,  DOI:10.1016/j.tet.2013.04.064 .
37. M. L. Agazzi, M. B. Ballatore, A. M. Durantini, E. N. Durantini and A. C. Tomé, J. Photochem. Rev., 2019, 40, 21–48,  DOI:10.1016/j.jphotochemrev.2019.04.001 .
38. M. Feoktistova, P. Geserick and M. Leverkus, Cold Spring Harb. Protoc., 2016, 2016 DOI:10.1101/pdb.prot087379 .
39. P. Kumar, A. Nagarajan and P. D. Uchil, Cold Spring Harb. Protoc., 2018, 2018(6) DOI:10.1101/pdb.prot095489 .
40. Y. Li, Y. Wei and X. Zhou, J. Photochem. Photobiol. A: Chemistry, 2020, 400, 112713,  DOI:10.1016/j.jphotochem.2020.112646 .
41. J. Liao, H. Zhao, Y. Xu, W. Zhou, F. Peng, Y. Wanga and Y. Fang, RSC Adv., 2017, 7, 33975–33985,  10.1039/c7ra04402j .
42. X. Nie, Z. Mahmood, Z. D. Liu, M. Li, D. Hu, W. Chen, L. Xing, S. Su, Y. Huo and S. Ji, Energy Environ. Mater., 2024, 7, e12597,  DOI:10.1002/eem2.12597 .
43. Z. Wang, M. Ivanov, Y. Gao, L. Bussotti, P. Foggi, H. Zhang, N. Russo, B. Dick, J. Zhao, M. Di Donato, G. Mazzone, L. Luo and M. Fedin, Chem. – Eur. J., 2020, 26, 1091–1102,  DOI:10.1002/chem.201904306 .
44. Y. Hou, X. Zhang, K. Chen, D. Liu, Z. Wang, Q. Liu, J. Zhao and A. Barbon, J. Mater. Chem. C, 2019, 7, 12048–12074,  10.1039/C9TC04285G .
45. X.-F. Zhang and N. Feng, Chem. – Asian J., 2017, 12, 2447–2456,  DOI:10.1002/asia.201700794 .
46. M. J. Ortiz, A. R. Agarrabeitia, G. Duran-Sampedro, J. Banuelos Prieto, T. A. Lopez, W. A. Massad, H. A. Montejano, N. A. Garcia and I. Lopez Arbeloa, Tetrahedron, 2012, 68, 1153–1162,  DOI:10.1016/J.TET.2011.11.070 .
47. E. C. Cheung and K. H. Vousden, Nat. Rev. Cancer, 2022, 22, 280–297,  DOI:10.1038/s41568-021-00435-0 .
48. R. Lubart, H. Friedmann, M. Sinykov and N. Grossman, Laser Therapy, 1995, 7, 101–105,  DOI:10.1117/12.199225 .
49. R. Lubart, Y. Wollman, H. Friedmann, S. Rochkind and I. Laulicht, J. Photochem. Photobiol., B, 1992, 12, 305–310,  DOI:10.1016/1011-1344(92)85032-P .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5nj01483b

This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2025