Light- and copper-activated (photo)cytotoxicity of 8-hydroxyquinoline-based boron photosensitizers with lipid droplet targeting and lipid peroxidation accumulation

Abstract

8-Hydroxyquinoline-based tetracoordinate boron complexes have been observed to exhibit pronounced luminescence and light-activated reactive oxygen species (ROS) generation, while their copper(II) analogs demonstrate significant cytotoxic effects in cancer cells. Nevertheless, both types of complexes are hindered by their inherent hydrophilicity, thereby limiting their efficacy in biological applications. Thus, we developed heavy-metal-free photosensitizers (PSs) based on 8-quinolinolato boron complexes, which exhibit light-activated fluorescence emission and ROS generation upon aggregation. The PSs effectively localize within lipid droplets and exhibit immediate and sustained ROS production upon exposure to light, even under hypoxic conditions, leading to lipid droplet-specific peroxidation, which is in accordance with the intracellular location, leading to ferroptosis-like cell death. Moreover, their fluorescence emission is quenched in the presence of Cu²⁺ ions, and the produced complexes enhance cytotoxicity instead. The photophysical properties of the complexes were comprehensively studied by a combination of experimental measurements, quantum mechanical (QM) and hybrid QM/molecular mechanics (MM) simulations. Thus, this investigation offers insights into new molecular design approaches for multifunctional probes with potential applications in photodynamic therapy and chemotherapy for cancer treatment.

---

Introduction

Photodynamic therapy (PDT) is recognized as a promising approach for the treatment of various diseases, including cancer, due to its minimal invasiveness and controlled cytotoxicity.^1,2 The procedure involves the administration of photosensitizers (PSs), which are designed to be selectively delivered to and accumulated in tumor tissues.^3–5 Upon specific light irradiation, these PSs produce reactive oxygen species (ROS), which induce the death of malignant cells while sparing the surrounding normal cells. Despite its potential, conventional PSs exhibit several limitations: (1) in their aggregated state, which is common within tumors, their fluorescence emission and ROS generation are significantly quenched.^6,7 (2) The uncontrolled ISC process results in negligible fluorescence radiative decay, limiting their applicability in biological settings.^8,9 (3) While the enhancement of the ISC process can be achieved through the introduction of heavy atoms (e.g., metal complexes, bromine, and iodine), this modification can lead to potential dark cytotoxicity.¹⁰ (4) Most PSs operate via a type II photoreaction, which is highly dependent on oxygen concentration.^11–13 This dependence poses a challenge under the hypoxic conditions characteristic of the solid tumor microenvironment.^14,15 (5) PSs often suffer from non-specific localization, reducing the efficacy of the treatment.¹⁶ (6) The limited multifunctionality of PSs restricts their broader application in areas such as synergistic phototherapy, chemosensing, and theranostics.¹⁰ These drawbacks necessitate ongoing research and development to enhance the efficacy and applicability of PSs in clinical settings.^3,10

Furthermore, PDT triggers a range of cellular responses, ultimately leading to cell death through apoptosis and/or necrosis. However, certain cellular adaptations, such as those induced by hypoxia and the upregulation of multidrug resistance proteins, can render cells resistant to apoptosis.^17,18 This highlights the need for the development of PSs that are minimally dependent on oxygen and capable of inducing non-apoptotic cell death pathways to enhance therapeutic efficacy. Ferroptosis, a form of cell death distinct from apoptosis and characterized by iron dependence and specific morphological changes, including mitochondrial abnormalities and the absence of nuclear responses,^19–21 has recently gained attention as a promising strategy for cancer therapy,^22,23 due to its effectiveness against cancer cells that are resistant to apoptosis.²⁴ As reported in ref. 24 and 25, ROS generated by PSs lead to the rapid depletion of glutathione (GSH), which subsequently inhibits glutathione peroxidase 4 (GPX4) activity. This inhibition results in the accumulation of lipid peroxides (LPOs), triggering ferroptosis.^24,25

In a related study, S. Tardito et al. demonstrated that 8-hydroxyquinoline induces a dose-dependent reduction in the viability of human tumor cells.²⁶ However, when copper is co-administered, the ligand's effects are significantly amplified, resulting in substantial cell death across the studied cell lines. Cytotoxic concentrations of 8-hydroxyquinoline-based Cu(II) complexes lead to increased intracellular copper accumulation and extensive vacuolization of the endoplasmic reticulum, preceding a non-apoptotic (paraptotic) form of cell death. Additionally, the detection of heavy metal ions, such as Cu²⁺, has become a critical focus in bioanalysis and environmental monitoring due to their significant roles in both physiological and pathological processes.^24,27–29 Fluorescent probes have emerged as highly effective tools for detecting Cu²⁺ ions, owing to their advantageous properties, including high selectivity, sensitivity, operational simplicity, rapid response, and low cost.^30–33 Notably, to the best of our knowledge, no studies have yet explored the use of boron complexes containing monoanionic bidentate (N^O) ligands as fluorescent probes for the detection of Cu²⁺.

Hence, we synthesized a series of 8-hydroxyquinoline-based boron tetradentate complexes (BQ1–BQ7). The photophysical properties of these complexes were thoroughly examined in various states, supported by computational studies including density functional theory (DFT) calculations and hybrid QM/MM simulations. Notably, BQ7 demonstrated significant potential as a heavy-metal-free PS for type I PDT, exhibiting high fluorescence emission under tumor-like conditions, such as in aggregated states and hypoxic environments. BQ7 was shown to induce lipid droplet-specific peroxidation, leading to ferroptosis-like cell death during PDT. Additionally, BQ7 functioned as a novel Cu²⁺-activated probe with enhanced photo-/dark cytotoxicity due to the formation of 8-hydroxyquinoline-based Cu(II) complexes. This design offers a novel type I PS for phototheranostic applications.

Results and discussion

Synthesis and structural characterization

BQ1–BQ7 were synthesized through a straightforward reaction between BPh₃ and 8-hydroxyquinoline derivatives at ambient temperature using a 1 : 1 molar ratio of BPh₃ to ligand in chloroform solvent (Scheme 1). The structures of BQ1–BQ7 were confirmed by NMR, HR-MS, and IR spectroscopy and elemental analysis (MAD = 0.09–0.23%), and the details are presented in the SI. To further explore the structures of the complexes, three representative complexes including BQ2, BQ6 and BQ7 were studied by single-crystal X-ray diffraction (SC-XRD). The molecular structures of BQ2, BQ6 and BQ7, as shown in Fig. 1, confirm that the boron center in these complexes exhibits a typical tetrahedral coordination with two carbon atoms of the phenyl groups and with the N and O atoms of the quinolinolate ligand. Furthermore, the powder XRD pattern calculated from the SC-XRD structure of BQ7 (Fig. S44c) exhibits strong agreement with the corresponding experimental pattern. In contrast, the discrepancies observed between the calculated and experimental powder XRD patterns for BQ2 and BQ6 (Fig. S44a and S44b, respectively) are appreciable rather than marginal. These divergences indicate the presence of multiple crystalline phases: one consistent with the structure determined by single-crystal analysis and another phase responsible for the significantly more intense diffraction peaks observed in the experimental data.

Scheme 1 Preparation of boron(III) complexes BQ1–BQ7.

Fig. 1 The molecular structures and partial crystal packing of BQ2 (a), BQ6 (b) and BQ7 (c). Relevant geometrical parameters (in Angstroms) for the most important intermolecular interactions in the crystal structures are highlighted and further analyzed in the SI.

Photophysical characteristics

Investigation of the steady-state absorption and emission spectra of BQ1–BQ7 compounds was conducted in different solvents, i.e., from non-polar (e.g., toluene (Tol)) to aprotic polar (e.g., tetrahydrofuran (THF) and acetonitrile (ACN)) ones at 10 µM concentration (Fig. 2a, S1 and S2). Notably, the UV-vis absorption band peak between 386 and 421 nm in these solvents indicates the presence of boron-chelated 8-hydroxyquinoline moieties. For all compounds, the absorption bands are structureless. Additionally, these compounds demonstrate a modest hypsochromic shift in absorption with an unchanged emission peak as solvent polarity increases. Attribution of the hypsochromic shift to a decreased polarizability is less likely, as the hypsochromic shift of the absorption maxima is not paralleled by a similar shift in the emission spectra (vide infra).³⁴ The unsubstituted BQ1 compound exhibited an absorption band maximum ranging from 386 to 400 nm with a prominent emission band maximum at approximately 502 nm (fluorescence quantum yield, FL QY (Φ_F) = 0.41–0.62) in Tol, THF, and ACN. In analogy to its corresponding absorption spectra, the emission bands are structureless. The series of halogen-substituted compounds (BQ2–BQ6) exhibited similar features and a discernible bathochromic shift in their absorption and emission spectra when compared to BQ1 (Table 1). They showed similar Stokes shifts of about 5000 cm⁻¹, which are much larger than the values observed for BODIPY³⁴ or cyanine dyes, indicating significant changes in bond lengths or angles upon excitation. The red shift of the spectra of BQ2–BQ6 compared to BQ1 can be attributed to the electron-withdrawing nature of the halogen substituents, resulting primarily in a decrease in the energy of the lowest unoccupied molecular orbital (LUMO), consequently reducing the highest occupied molecular orbital (HOMO)–LUMO energy gap. The FL QYs and fluorescence decay times of BQ2–BQ6 are decreased as compared to those of BQ1, and this trend is particularly evident in the cases of BQ4 and BQ6. This decrease can mainly be attributed to the enhanced intersystem crossing (ISC) processes induced by the heavy-atom effect which is most marked for the compounds bearing iodine atoms. One should note that the decrease in the FL QY is more marked than that in the FL decay time, suggesting a decrease in the FL rate constant by 40 ± 10%. Of particular interest is BQ7, where the introduction of a methyl group led to a higher emission QY as compared to that of BQ3.

Fig. 2 Normalized absorption (solid lines) and PL emission (dashed lines) spectra of BQ1–BQ7 in (a) toluene (10 µM, λ_ex = 390–410 nm), (b) solid state, and (c) film; (d) fluorescence intensity ratio (I/I₀) of BQ7 in THF/DW (0–99%); (e) absorbance decrease of 1,3-diphenylisobenzofuran (DPBF) in the presence of PSs (e.g., BQ7 and Ru(bpy)₃) in THF/DW (5/95) under light irradiation; (f) the fluorescence intensity of dihydroethidium (DHE) in the presence and absence (as a control group) of PSs in THF/DW (5/95) under light irradiation; (g) fluorescence emission of BQ7 (10 µM) in the presence of metal cations (10.0 eq.) in THF/DW (5/95); (h) UV-vis absorption and fluorescence emission spectra of BQ7 (10 µM) in the presence of Cu²⁺ (0–1.0 eq.) in THF/DW (5/95); (i) fluorescence intensity of BQ7 (10 µM) in the presence of Cu²⁺ (0–1.0 eq.) in THF/DW (5/95) (the Stern–Volmer constant (K_SV)-dependent fluorescence intensity ratio (I/I₀) is provided in the inset).

Table 1 Photophysical properties of BQ1–BQ7 in solvents. Absorption maxima (λ_max), emission maxima (λ_em), molar absorption coefficient (ε), fluorescence QY (Φ_F) and relative singlet oxygen QY (Φ_Δ)

Sol.	λabs (nm)/ε (M^−1 cm^−1) (×10^3)			λems (nm)/ΦF			τF (ns)	ΦΔ
	Tol	THF	ACN	Tol	THF	ACN	Tol	ACN
BQ1	400/2.92	393/2.42	386/2.82	500/0.59	502/0.62	502/0.41	24.3	0.15
BQ2	414/3.30	408/2.68	401/3.29	522/0.17	522/0.15	522/0.13	17.2	0.14
BQ3	412/1.32	400/1.63	395/1.48	522/0.20	522/0.09	522/0.08	14.5	0.14
BQ4	415/2.88	406/2.67	401/3.17	519/0.15	519/0.13	519/0.10	10.0	0.28
BQ5	421/2.94	414/3.34	407/3.51	521/0.03	519/0.02	521/0.02	1.8/4.7	0.43
BQ6	419/3.08	412/2.68	403/3.21	528/0.06	528/0.04	528/0.04	4.4/13.7	0.42
BQ7	405/2.47	397/1.94	389/2.77	516/0.41	514/0.32	514/0.26	23.3	0.16

---

Motivated by the observed reduction in fluorescence emission in BQ4–6, which has been putatively assigned to efficient ISC processes in the latter compounds, the singlet oxygen (¹O₂) QYs of all compounds were measured in ACN employing Ru(bpy)₃²⁺ as a reference (see the results in Table 1). Notably, BQ5 and BQ6, characterized by the presence of iodine atoms, demonstrated the highest efficiencies for ¹O₂ generation (¹O₂ QY, Φ_Δ = 0.43 and Φ_Δ = 0.42, respectively, see Table 1). Incorporation of two bromine atoms into the molecular scaffold (BQ4) leads to a lower ¹O₂ QY (Φ_Δ = 0.28). Intriguingly, boron complexes devoid of heavy atoms (BQ1–3,7) also displayed non-negligible singlet oxygen generation. This suggests that non-radiative decay in these compounds occurs at least partially by ISC.

Further comprehensive photophysical characterization of BQ1–BQ7 was conducted in condensed phases, including Zeonex film (5% doping), the solid-state, and organic glass matrices (in Tol at 77 K) (Fig. S1–S4). The absorption and emission spectra of BQ1–BQ7 in films (Fig. 2c) exhibited a discernible bathochromic shift compared to their respective spectra in the solution phase (see Tables 1 and 2). Conversely, the emission peaks experienced a hypochromic shift in the solid state (Fig. 2c) and in organic glass environments (e.g., Tol, THF, and ACN at 77 K, see Fig. S1 and S2).

Table 2 Photophysical properties of BQ1–BQ7 in film and solid states. Absorption maxima (λ_max), emission maxima (λ_em), and fluorescence QY (Φ_F)

	λabs (nm)	λems (nm)/ΦF			τF ^a (ns)
	Film	Tol 77 K	Film	Solid	Film	Solid
a Average τF, – not recorded.
BQ1	400	484/—	501/0.45	493/0.32	20.8	23.5
BQ2	420	496/—	529/0.40	520/0.19	18.8	12.3
BQ3	421	484/—	530/0.30	491/0.18	20.1	16.5
BQ4	421	449/—	508/0.15	484/0.02	8.0	5.6
BQ5	426	507/—	526/0.03	498/0.01	5.2	1.2
BQ6	426	506/—	529/0.08	503/0.03	4.6	2.6
BQ7	409	498/—	524/0.31	494/0.31	22.6	16.1

---

To gain deeper insights into the fluorescence emission behavior of BQ1–BQ7, excited-state lifetimes were recorded with time-correlated single-photon counting (TCSPC) in solvents, films and solid states (Fig. S1–S4). In Tol at 10 µM concentration, BQ1–BQ4, and BQ7 exhibited single-component emission profiles characterized by prolonged fluorescence lifetimes (with values ranging between 10.0 and 24.3 ns, see Table 1). Conversely, BQ5 and BQ6 exhibited emissions with dual components of monomer and aggregate emission. Notably, the emissions of BQ1–BQ7 in films and solid-state environments exhibited dual components, also suggestive of monomer and aggregate emissions (Tables 2 and S2).

To examine the emission in the aggregated state, the emission characteristics of BQ7 were investigated in THF with distilled water (DW) (0–99%) (Fig. 2d and S7). The emission peak shifted from 528 (at 0% DW) to 520 nm (at 99% DW), accompanied by a substantial enhancement in intensity (∼5 times) at 99% DW. Furthermore, the singlet oxygen (¹O₂) and superoxide radical oxygen (O₂˙⁻) generation of BQ7 were assessed in the aggregated state (at THF/DW (5/95)) utilizing 1,3-diphenylisobenzofuran (DPBF) and dihydroethidium (DHE) as probes (Fig. 2e and f, respectively). Intriguingly, the decrease in DPBF absorption and the increase in DHE intensity observed in the presence of BQ7 surpassed those of Ru(bpy)₃²⁺ at the same absorbance under light irradiation, indicating efficient ISC processes and consequent ROS generation. Thus, BQ7 demonstrates significant potential as an effective PS for fluorescence imaging-guided PDT.

Sensing of Cu²⁺ with BQ7

In the literature, metal complexes derived from 8-hydroxyquinoline (Q)-based ligands have garnered considerable attention owing to their broad utility in luminescence materials and biological applications and are typically synthesized via direct reaction of 8-hydroxyquinoline. Here, we explore the feasibility of 8-hydroxyquinoline-based boron complexes transforming into their corresponding metal complexes and their potential use in sensing applications. To this end, BQ1–7 (10 µM) were chosen for reaction with various metal cations, including Co²⁺, Ni²⁺, Fe³⁺, Cd²⁺, Cu²⁺, Al³⁺, Cr³⁺, Zn²⁺, Pb²⁺, and Mn²⁺ (10.0 eq.) in DW containing 5% THF. Intriguingly, upon introduction of Cu²⁺, a substantial quenching of the fluorescence emission of BQ7 was observed, as evidenced both by spectral analysis (Fig. 2g) and under 365 nm light irradiation (Fig. S8). Conversely, negligible perturbation of the emission profile of BQ7 was observed upon addition of other metal cations (Fig. 2g). In addition, no differences in the fluorescence emission of BQ1–BQ6 in the presence of metal cations including Cu²⁺ were observed (Fig. S9), indicating the selective sensing of BQ7. Subsequently, titration experiments were conducted to elucidate the interaction between BQ7 and Cu²⁺, revealing a gradual decrease in emission intensity culminating in complete quenching at 0.5 eq. of Cu²⁺ (Fig. 2h, 2i and S10). The Stern–Volmer constant (K_SV) and the bimolecular quenching rate constant (k_q) were calculated to be 0.34 × 10⁻⁶ M and 41.2 M⁻¹ s⁻¹, respectively. This observation suggests the formation of non-luminescent copper complexes, specifically denoted as (Q7)₂Cu (see Fig. 3a). To corroborate this hypothesis, further titration experiments were undertaken employing ¹H-NMR spectroscopy in acetone-d₆. These investigations revealed a concomitant decrease and increase in proton intensities corresponding to BQ7 and chlorodiphenylborane, respectively. The latter is a byproduct generated in the reaction of BQ7 with CuCl₂ (Fig. S11). Notably, while the signal corresponding to (Q7)₂Cu could not be observed in its doublet form, its presence was further substantiated through mass spectrometric analysis of BQ7 after the addition of Cu²⁺ (at 0.2 eq.) (Fig. S12). Significantly, the observed “on–off” emission of BQ7 upon interaction with Cu²⁺ in the aggregated state underscores its potential utility as a chemosensor and phototheranostic probe, thereby holding promise for diverse biological applications.

Fig. 3 (a) Proposed sensing mechanism of BQ7 with Cu²⁺; (b) frontier molecular orbital images and energies of BQ7 and (Q7)₂Cu at optimized S₀ geometries by using MPW1PW91/6-31+g(d,p) level theory in PCM (with Tol as solvent).

Computational studies

Encouraged by the intriguing photophysical properties of BQ7 in the aggregated state and its sensing capabilities, quantum chemical calculations on the BQ series were performed. The geometries of the ground (S₀) and lowest singlet (S₁) excited states of the BQ series were optimized using density functional theory (DFT)³⁵ and time-dependent DFT (TD-DFT)³⁶ with the Tamm–Dancoff approximation (TDA)³⁷ (see details of the computational protocol in the SI), respectively. The absorption and emission spectra of the BQ derivatives, peaking at 400–420 nm and 500–530 nm, respectively, were attributed to the S₀ → S₁ and S₁ → S₀ electronic transitions, which mainly involve HOMO and LUMO (see Tables S4 and S5). The HOMO and LUMO are predominantly localized on the quinoline ligand and the central boron atom, with minimal contributions from the phenyl ligands (Fig. S13), confirming the local excitation (¹LE) character of S₁. Additionally, the geometry of the (Q7)₂Cu complex was optimized. The ground state of the (Q7)₂Cu complex corresponds to a doublet state, i.e., D₁. Analysis of the frontier molecular orbitals of (Q7)₂Cu reveals that the singly occupied molecular orbital (SOMO) is primarily located on the quinoline ligand, where the LUMO is distributed over the quinoline ligand and the central copper atom (Fig. 3b). The D₁ of (Q7)₂Cu is indeed of ²LMCT character. ²LMCT states are often involved in efficient non-radiative deactivation channels, outcompeting fluorescence or phosphorescence, and thus leading to the quenching of emission.³⁸ Furthermore, Cu(II) complexes experience Jahn–Teller distortions,^39,40 especially in octahedral or distorted tetrahedral geometries. This distortion leads to a non-rigid structure, increasing non-radiative relaxation pathways through structural changes in solvents, further quenching emission.⁴¹ The optimal geometries of the T₁ and T₂ triplet excited states of the BQ series were also optimized by the TDA TD-DFT method.^35,37 The calculated adiabatic energy differences with respect to S₀ are shown in Table 3. The energy of S₁ is significantly higher than that of T₁ and lower than that of T₂, indicating that the most likely ISC pathway is the S₁–T₁ transition. The ISC rate constant (k_ISC) depends mainly on the energy gap and the spin–orbit couplings (SOCs) between the involved singlet and triplet excited states.^42,43 For the BQ1–BQ7 series, S₁ and T₁ are relatively close in energy. The computed SOCs between the S₁ and T₁ states of BQ1–BQ7 are also presented in Table 3. While chlorination does not significantly impact the calculated SOC values (0.07–0.09 cm⁻¹), bromine and iodine substitution significantly increase the computed SOCs, attributable to the heavy-atom effect, such as, the diiodo-substituted BQ5 (0.64 cm⁻¹), dibromo-substituted BQ4 (0.18 cm⁻¹) and mono-iodo-substituted BQ6 (0.46 cm⁻¹). Consequently, BQ5 possesses the highest Φ_Δ value, while BQ2, BQ3, and BQ7 possess nearly identical Φ_Δ values to that of BQ1. These computational results qualitatively agree with the experimental pieces of evidence.

Table 3 Calculated excited state properties of BQ1–BQ7. The singlet and triplet excited states were optimized by the TDA TD-DFT method using MPW1PW91/6-31+g(d,p) in PCM (Tol as solvent). Adiabatic energy difference with respect to the optimized S₀ energies (E_ad), singlet–triplet energy gaps (ΔE_S–T), spin–orbit couplings (SOCs), ISC rate (k_ISC), fluorescence rate (k_fl), and computed fluorescence QYs (Φ_fl)

	Ead (eV)			S1–T1				S1–S0
	S1	T1	T2	ΔES–T (eV)	SOCs (cm^−1)	kISC (s^−1) (FC-HT)	kISC (s^−1) (FC)	kfl (s^−1)	Φfl (cal.)
BQ1	2.62	1.58	3.14	1.04	0.09	6.9 × 10^6	4.8 × 10^5	1.0 × 10^7	0.59
BQ2	2.49	1.48	3.00	1.01	0.09	6.8 × 10^6	4.2 × 10^5	1.1 × 10^7	0.62
BQ3	2.49	1.50	2.30	0.99	0.10	1.8 × 10^6	3.5 × 10^4	3.7 × 10^6	0.67
BQ4	2.48	1.52	2.98	0.96	0.18	1.6 × 10^8	5.0 × 10^5	7.9 × 10^6	0.05
BQ5	2.48	1.54	2.70	0.94	0.64	3.7 × 10^8	9.8 × 10^5	8.9 × 10^6	0.01
BQ6	2.47	1.50	2.70	0.97	0.46	1.1 × 10^8	9.6 × 10^5	4.2 × 10^6	0.02
BQ7	2.52	1.54	3.00	0.98	0.07	5.3 × 10^6	6.0 × 10^4	2.5 × 10^6	0.32

---

In addition, SOCs are not the sole factor influencing ISC rates. To achieve more quantitative agreement, the excited state decay rate constants were calculated. The ISC rate constant for the S₁ → T₁ transition was calculated with the Franck–Condon (FC) approximation but also including Herzberg–Teller (HT) effects⁴⁴ (see the results and a comparison of both approaches in Table 3). The calculated ISC rate constants for BQ1–7 are likely underestimated due to the neglect of HT effects within our theoretical approach. Not surprisingly, BQ5 possesses the largest ISC rate constant (3.7 × 10⁸ s⁻¹), higher than those of BQ4 and BQ6, and significantly greater than those of BQ1–3 and BQ7. Their S₁–T₁ energy gaps are comparable; therefore, the lower ISC rate constants observed for BQ1–3 and BQ7 are primarily attributed to the smaller SOC between the S₁ and T₁ states.

Furthermore, ISC rates are not the only factor impacting fluorescence emission. To obtain a more quantitative agreement of Φ_F, the fluorescence rate constant (k_fl) for the S₁ → S₀ transition was also calculated using the FC/HT approximation. Under the assumption that the nonradiative mechanisms are dominated by the S₁ → T₁ ISC processes, the FL QY can be expressed using the following equation:

Using eqn (1), there is a qualitative/semi-quantitative good agreement between the calculated and experimental FL QYs, with high values for BQ1–3 and BQ7 and the lowest values for BQ4–6 (see Tables 1 and 3). For example, the experimental and computed FL QY values for BQ1, BQ5 and BQ7 are 0.62 and 0.59, 0.03 and 0.01, and 0.41 and 0.32, respectively. This confirms the relevance of heavy-atom and vibronic Herzberg–Teller effects in determining the ISC and fluorescence rate constants of PSs.

Motivated by the enhanced fluorescence emission and ROS generation of BQ7 in its aggregated state, which can be induced by the interactions of BQ7 with its neighboring molecules (e.g., dimers, trimers, and tetramers),⁷ we conducted QM/MM simulations using a multilayer ONIOM model in PCM (water).⁴⁵ In its crystalline form, we observed that BQ7 exhibits π–π stacking interactions between the quinoline and phenyl groups of two molecules (Fig. S15c), which are believed to be the main π–π interaction between them. Consequently, these two molecules (BQ7 dimer) were designated as the QM layer in the QM/MM calculations, while the surrounding molecules were treated as the MM layer (see Fig. 4a). The departing molecular structure for the QM/MM optimizations was extracted from the crystalline data. The geometries of the first singlet (S₁) and triplet (T₁) excited states were optimized with QM/MM (see details in the SI). At the Franck–Condon geometry, the S₁ → S₀ and T₁ → S₀ transitions were predominantly (>96.6%) attributed to the HOMO–LUMO transition, where the HOMO and LUMO are localized on the same molecule (see Fig. 4a and S16), indicating ¹LE and ³LE character for the lowest singlet and triplet excited states of the BQ7 dimer, respectively. Notably, the oscillator strength of S₁ in the BQ7 dimer (f = 0.12) is larger than that of the monomer (f = 0.07), confirming the enhanced emission of BQ7 upon aggregation, which can be induced by interaction of BQ7 with its neighboring molecules (Fig. 4b). In addition, the singlet–triplet adiabatic energy gap between S₁ and T₁ in the BQ7 dimer was calculated to be 0.78 eV, which is lower than that of the monomer state (0.98 eV) (Fig. 4b). The SOC between S₁ and T₁ was computed to be 0.13 cm⁻¹ in the dimer, while those in the monomer are remarkably smaller (0.07 cm⁻¹) (see Fig. 4b). Note that the ISC rate constant (k_ISC) is proportional to the SOC and inversely proportional to the energy gap between the singlet and triplet states. This indicates that the efficient ISC process of BQ7 in the dimer state can be ascribed to an aggregation-induced ISC (AI-ISC) mechanism,⁷ which is beneficial for ROS generation via both type I and type II mechanisms in its aggregated states.

Fig. 4 (a) HOMO and LUMO images of the BQ7 dimer at optimized S₁ and T₁ geometries, through QM/MM simulation; (b) illustration of the ISC process and fluorescence emission of BQ7 in monomeric and dimeric forms along with SOC and energy gap between S₁ and T₁ states.

Fluorescence sensing of BQ7 in living cells

Encouraged by the intriguing photophysical properties of BQ7 as a potential photosensitizer and fluorescence sensor, we sought to evaluate its feasibility within biological systems. Initially, we assessed the fluorescence of BQ7 within living cells. Time-series imaging was conducted on live HeLa cells in a live cell chamber following the addition of BQ7 at a concentration of 100 μM (Fig. 5a). The fluorescence of BQ7 became detectable approximately 5 minutes post-treatment, reaching saturation at 10 minutes post-treatment. These results indicate that BQ7 successfully diffuses across cellular membranes, with its fluorescence persisting within living cells. Notably, the fluorescence pattern of BQ7 was not uniformly diffused throughout the cell (Fig. 5). Instead, it appeared to accumulate in certain membranous compartments, prompting us to investigate the specific intracellular organelles targeted by BQ7, which could also suggest its potential therapeutic functions.

Fig. 5 (a) Fluorescence images of BQ7 after incubation in living cells (0–15 min); (b) subcellular localization of BQ7 (scale bar: 10 μm). BF: brightfield.

We first examined lysosomes and mitochondria in live HeLa cells. The cells were incubated with 50 μM LysoTracker Deep Red or 100 μM MitoTracker Deep Red, along with 100 μM BQ7. However, confocal laser scanning microscopy (CLSM) imaging revealed a lack of colocalization of BQ7 with either lysosomes or mitochondria (Fig. 5b). Based on the predicted hydrophobicity of BQ7, the vesicular structure, and the high refractivity observed in brightfield images, we hypothesized that BQ7 accumulates in lipid droplets. To test this, we stained live HeLa cells with BODIPY 493/503 dye, which specifically stains neutral lipids and exhibits enhanced signals in lipid droplets. CLSM imaging of these samples demonstrated complete overlap between the signals of BQ7 and BODIPY 493/503. The calculated Pearson colocalization coefficient was 0.5299 for BQ7 and BODIPY 493/503 (n = 4), compared to −0.0136 and 0.0004 for MitoTracker Deep Red (n = 3) and LysoTracker Deep Red (n = 4), respectively.

Therefore, we concluded that BQ7 predominantly colocalizes with lipid droplets inside cells, rather than associating significantly with mitochondria or lysosomes. This specific accumulation in lipid droplets underscores the unique intracellular targeting properties of BQ7, suggesting its potential application in cellular imaging and targeted therapy.

Subsequently, we investigated the feasibility of Cu²⁺-sensing by BQ7 in live cell systems. HeLa cells were initially stained with 100 μM BQ7. Following complete staining, CuCl₂ was introduced to the cells in cell growth medium (DMEM with 10% FBS) at a final concentration of 100 μM. A dramatic decrease in BQ7 fluorescence was observed 5–10 minutes post-treatment, culminating in an almost complete loss of fluorescence (Fig. 6). These findings unequivocally demonstrate the ability of BQ7 to sense Cu²⁺ in live cells. We infer that the fluorescence quenching of BQ7 is attributable to the formation of non-emissive (Q7)₂Cu complexes. Coupled with the colocalization results, we propose that BQ7 can detect Cu²⁺ cations within lipid droplets.

Fig. 6 (a) Fluorescence images of BQ7 before and after treatment with Cu²⁺ in HeLa cells; (b) fluorescence intensity of BQ7 before and after treatment with Cu²⁺ in HeLa cells.

Light-activated photocytotoxicity of BQ7 in living cells

In the previous experiment, we noted the morphological damage in the BQ7 plus irradiation group, along with the delayed increase in ROS. Therefore, it is natural for us to study the cytotoxicity of BQ7 under PDT conditions. HeLa cells were subjected to a dose-dependent cell viability test (BQ7 concentrations of 0–100 μM), with or without light irradiation (Fig. 8). Under normoxic conditions, the HeLa cells showed a dose-dependent decrease in viability measured by CCK8 and spectrophotometry. It was notable that 100 μM BQ7 with light irradiation showed a dramatic decrease 24 hours after irradiation, while the groups without light irradiation maintained their viability even at the highest concentration. Under hypoxic conditions, similar results were observed: 100 μM BQ7 with light irradiation induced cell toxicity 24 hours after irradiation, while the groups without light irradiation remained unaffected. Additionally, we observed an almost complete lack of living cells in the 100 μM BQ7 group at 48 hours after irradiation, while the cells were still viable in the light-unirradiated group at the same time point. Here, we could conclude that BQ7 can induce photo-cytotoxicity under both normoxic and hypoxic conditions, demonstrating its potential for PDT applications.

Next, we investigated the cause of BQ7-mediated cell damage, along with the mechanism of the delayed ROS increase. Live HeLa cells were immediately stained with BODIPY 581/591 (a lipid-specific peroxidation dye) after BQ7 treatment with or without light irradiation. CLSM imaging and quantitative analysis showed that the green to red fluorescence ratio in the lipid droplets increased in the light irradiation group, which indicates the peroxidation of the lipid at these locations (Fig. 9). On the other hand, the global green to red fluorescence was not different among the groups. The delayed ROS elevation observed in Fig. 7 might be a consequence of lipid peroxidation, the detailed mechanism of which will need further investigation.

Fig. 7 Fluorescence images of (a) DCFH-DA and DHE, and (b) delayed measurement of DCFH-DA.

Fig. 8 Cell viability in the presence of BQ7 with and without light irradiation under (a) normoxic and (b) hypoxic conditions at 24 h. (c) Brightfield image of cells at 48 h.

Fig. 9 Fluorescence images of (a) peroxidation of lipid droplets measured by BODIPY 581/591 and (b) quantitation of relative lipid droplet peroxidation. (c) Fluorescence image of BQ7 under the treatment of PDT showing nuclei and mitochondria at 0 and 180 min.

We also observed delayed mitochondrial fission and swelling, which is evidence of mitochondrial damage, 180 minutes after PDT treatment with BQ7. At the same time, until 180 minutes, we could not observe any evidence of apoptotic cell death, including apoptotic bodies, nuclear fragmentation or DNA condensation. The morphological changes resemble those of ferroptosis.²⁰ These results suggest that the PDT treatment with BQ7 leads to lipid droplet-specific peroxidation, which is in accordance with the intracellular location of BQ7, leading to ferroptosis-like cell death.

Cu²⁺-mediated (photo)cytotoxicity of BQ7 in living cells

In relation to the ability of BQ7 to react with Cu²⁺ and form (Q7)₂Cu complexes in live cells. Cu complexes of quinoline showed cytotoxicity toward cancer cells.²⁶ We performed a CCK8 viability assay after treatment with 100 μM BQ7 and 100 μM CuCl₂ with or without light irradiation. The HeLa cells were treated with BQ7 in DMEM for 30 min. After a DPBS wash, CuCl₂ in DMEM was applied for 10 min. After another DPBS wash, the samples were irradiated. We found a significant Cu²⁺ dose-dependent decrease in cell viability in most of the groups tested with a greater impact in the light-irradiated groups (Fig. 10). Among the 100 μM CuCl₂-treated groups, the 100 μM BQ7 group with light irradiation exhibited the highest cytotoxicity. Therefore, we can conclude that the presence of Cu²⁺ can enhance the photo- and dark-cytotoxicity of BQ7 cancer cells, highlighting its potential for Cu²⁺-mediated PDT and CMT applications.

Fig. 10 Cell viability in the presence of BQ7 (0–100 µM) and Cu²⁺ (0–100 µM) with (a) and without (b) light irradiation.

Conclusion

In summary, a series of 8-hydroxyquinoline-based boron PSs (BQ1–BQ7) was synthesized, exhibiting intense fluorescence emission in solvents, films, solid states, and organic glass matrices. The ¹O₂ QYs of the BQ series were enhanced by the presence of heavy atoms, such as iodine and bromine. However, the heavy-metal-free BQ7 PS demonstrated efficient fluorescence emission and ROS generation through both type I and type II mechanisms in the aggregated state. QM/MM simulations revealed an increase in the oscillator strength of the S₁ state and the ISC process between the S₁ and T₁ states in the dimer form of BQ7 as compared to its monomer form. Furthermore, BQ7 showed the ability to detect Cu²⁺ ions via an on–off fluorescence emission in aggregates and within lipid droplets. PDT treatment with BQ7 under both normoxic and hypoxic conditions resulted in lipid droplet-specific peroxidation, consistent with its intracellular localization, leading to ferroptosis-like cell death. Additionally, the treatment with Cu²⁺ enhanced the cytotoxicity of BQ7, suggesting a novel molecular design strategy for heavy-metal-free type I PSs aimed at ferroptosis-mediated PDT and chemotherapy.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: experimental and characterization data include NMR spectra, crystallographic data, photophysical measurements, biological experiments, and computational investigations. See DOI: https://doi.org/10.1039/d5dt01016k.

Acknowledgements

This work was sponsored by the NAFOSTED – FWO program with project numbers FWO.104.2020.03 and G0E5321N in Vietnam and the Flemish region, respectively. L. V. M. thanks the Hercules Foundation for supporting the purchase of the diffractometer through project AKUL/09/0035. The authors thank Prof. So Yeong Lee, Dr Flip de Jong, and Dr Jonathan Vandenwijngaerden for their valuable support in doing experiments. The authors thank Prof. Tan Le Hoang Doan for performing elemental analysis using the Euro Vector Elemental Analyser EA3100 Series at the Advanced Materials Technology Institute, Vietnam National University Ho Chi Minh City (VNUHCM-AMTI).

References

1. A. P. Castano, P. Mroz and M. R. Hamblin, Nat. Rev. Cancer, 2006, 6, 535–545.
2. J. F. Lovell, T. W. B. Liu, J. Chen and G. Zheng, Chem. Rev., 2010, 110, 2839–2857.
3. X. Li, J. F. Lovell, J. Yoon and X. Chen, Nat. Rev. Clin. Oncol., 2020, 17, 657–674.
4. S. S. Lucky, K. C. Soo and Y. Zhang, Chem. Rev., 2015, 115, 1990–2042.
5. Z. Zhou, K. Ni, H. Deng and X. Chen, Adv. Drug Delivery Rev., 2020, 158, 73–90.
6. Y. Hong, J. W. Y. Lam and B. Z. Tang, Chem. Soc. Rev., 2011, 40, 5361–5388.
7. J. Zhang, S. Mukamel and J. Jiang, J. Phys. Chem. B, 2020, 124, 2238–2244.
8. G. Feng, G.-Q. Zhang and D. Ding, Chem. Soc. Rev., 2020, 49, 8179–8234.
9. T. C. Pham, T. T. H. Hoang, D. N. Tran, G. Kim, T. V. Nguyen, T. V. Pham, S. Nandanwar, D. L. Tran, M. Park and S. Lee, ACS Appl. Mater. Interfaces, 2023, 15, 47969–47977.
10. T. C. Pham, V.-N. Nguyen, Y. Choi, S. Lee and J. Yoon, Chem. Rev., 2021, 121, 13454–13619.
11. W. Fan, P. Huang and X. Chen, Chem. Soc. Rev., 2016, 45, 6488–6519.
12. B. M. Luby, C. D. Walsh and G. Zheng, Angew. Chem., Int. Ed., 2019, 58, 2558–2569.
13. Z. Zhou, J. Song, L. Nie and X. Chen, Chem. Soc. Rev., 2016, 45, 6597–6626.
14. J. M. Brown and W. R. Wilson, Nat. Rev. Cancer, 2004, 4, 437–447.
15. A. Sharma, J. F. Arambula, S. Koo, R. Kumar, H. Singh, J. L. Sessler and J. S. Kim, Chem. Soc. Rev., 2019, 48, 771–813.
16. R. Wang, X. Li and J. Yoon, ACS Appl. Mater. Interfaces, 2021, 13, 19543–19571.
17. L. Flamant, A. Notte, N. Ninane, M. Raes and C. Michiels, Mol. Cancer, 2010, 9, 191.
18. J. Liu, C. Chen, T. Wei, O. Gayet, C. Loncle, L. Borge, N. Dusetti, X. Ma, D. Marson, E. Laurini, S. Pricl, Z. Gu, J. Iovanna, L. Peng and X.-J. Liang, Exploration, 2021, 1, 21–34.
19. J. P. F. Angeli, R. Shah, D. A. Pratt and M. Conrad, Trends Pharmacol. Sci., 2017, 38, 489–498.
20. J. Ju, Y.-n. Song and K. Wang, Aging Dis., 2021, 12, 261–276.
21. M. Liu, B. Liu, Q. Liu, K. Du, Z. Wang and N. He, Coord. Chem. Rev., 2019, 382, 160–180.
22. J. Han, H. Kang, X. Li, N. Kwon, H. Li, S. Park and J. Yoon, ACS Appl. Nano Mater., 2021, 4, 5954–5962.
23. Y. Wen, H. Chen, L. Zhang, M. Wu, F. Zhang, D. Yang, J. Shen and J. Chen, Free Radicals Biol. Med., 2021, 173, 41–51.
24. J. Yang, X. Jin, Z. Cheng, H. Zhou, L. Gao, D. Jiang, X. Jie, Y. Ma and W. Chen, ACS Sustainable Chem. Eng., 2021, 9, 13206–13214.
25. W. S. Yang and B. R. Stockwell, Trends Cell Biol., 2016, 26, 165–176.
26. S. Tardito, A. Barilli, I. Bassanetti, M. Tegoni, O. Bussolati, R. Franchi-Gazzola, C. Mucchino and L. Marchiò, J. Med. Chem., 2012, 55, 10448–10459.
27. Y. Li, H. Zhou, S. Yin, H. Jiang, N. Niu, H. Huang, S. A. Shahzad and C. Yu, Sens. Actuators, B, 2016, 235, 33–38.
28. L. Wang, Z.-L. Wei, Z.-Z. Chen, C. Liu, W.-K. Dong and Y.-J. Ding, Microchem. J., 2020, 155, 104801.
29. G. Zhang, Y. Li, J. Xu, C. Zhang, S. Shuang, C. Dong and M. M. Choi, Sens. Actuators, B, 2013, 183, 583–588.
30. S. Wu, X. Ma, Y. Wang, J. Zhou, X. Li and X. Wang, Spectrochim. Acta, Part A, 2021, 249, 119330.
31. L. Zhang, J. Sun, S. Liu, X. Cui, W. Li and J. Fang, Inorg. Chem. Commun., 2013, 35, 311–314.
32. Z. Han, D. Nan, H. Yang, Q. Sun, S. Pan, H. Liu and X. Hu, Sens. Actuators, B, 2019, 298, 126842.
33. M. Li, L.-C. Feng, S.-S. Feng and W.-K. Dong, J. Mol. Struct., 2022, 1261, 132926.
34. W. Qin, M. Baruah, M. Van der Auweraer, F. C. De Schryver and N. Boens, J. Phys. Chem. A, 2005, 109, 7371–7384.
35. W. Kohn, A. D. Becke and R. G. Parr, J. Phys. Chem., 1996, 100, 12974–12980.
36. C. Adamo and D. Jacquemin, Chem. Soc. Rev., 2013, 42, 845–856.
37. S. Hirata and M. Head-Gordon, Chem. Phys. Lett., 1999, 314, 291–299.
38. A. M. May and J. L. Dempsey, Chem. Sci., 2024, 15, 6661–6678.
39. M. V. Veidis, G. H. Schreiber, T. E. Gough and G. J. Palenik, J. Am. Chem. Soc., 1969, 91, 1859–1860.
40. M. A. Halcrow, Chem. Soc. Rev., 2013, 42, 1784–1795.
41. J. Wang, J. Ren, Q. Tang, X. Wang, Y. Wang, Y. Wang, Z. Du, W. Wang, L. Huang, L. A. Belfiore and J. Tang, Materials, 2022, 15, 1719.
42. C. M. Marian, Wiley Interdiscip. Rev.:Comput. Mol. Sci., 2012, 2, 187–203.
43. T. J. Penfold, E. Gindensperger, C. Daniel and C. M. Marian, Chem. Rev., 2018, 118, 6975–7025.
44. A. Baiardi, J. Bloino and V. Barone, J. Chem. Phys., 2014, 141, 114108.
45. L. W. Chung, W. M. C. Sameera, R. Ramozzi, A. J. Page, M. Hatanaka, G. P. Petrova, T. V. Harris, X. Li, Z. Ke, F. Liu, H.-B. Li, L. Ding and K. Morokuma, Chem. Rev., 2015, 115, 5678–5796.

---

Footnote

† Contributed equally.

---