Expanding the toolbox of bioorthogonal activation of photosensitizers for precise photodynamic therapy through transition metal-mediated deallylation

Abstract

The development of bioorthogonal activation of photosensitizers represents a promising avenue for precise photodynamic therapy (PDT) against cancer. It requires effective decaging strategies that can be applied in vivo, but such highly efficient and biocompatible methodologies remain scarce. We report herein the first use of ruthenium-mediated bioorthogonal deallylation for on-demand activation of boron dipyrromethene (BDP)-based photosensitizers. In this study, we first prepared a series of BDP-based photosensitizers connected to an allyl group with or without a self-immolative spacer via an ester caging unit [Pro-BDP-n (n = 1–4)] and studied their activation upon treatment with several ruthenium complexes [RuLn (n = 1–3)]. Pro-BDP-3, having a methoxy substituent at the self-immolative linker, was found to exhibit the highest release efficiency and fastest decaging kinetics. Using RuL3 as activator, which was modified with a biotin moiety, Pro-BDP-3 could be selectively activated in the endoplasmic reticulum (ER) of biotin receptor-overexpressed cancer cells. The induced ER stress disturbed the intracellular calcium homeostasis and ER-mitochondria crosstalk, resulting in mitochondria dysfunction and eventually cell death via apoptosis. The high treatment efficacy of this approach was also demonstrated in vivo. This work expands the toolbox of bioorthogonal activation of photosensitizers using transition metal-promoted deallylation as an efficient decaging strategy.

Graphical abstract

1. Introduction

Photodynamic therapy (PDT) is a clinically approved treatment modality for a range of cancers and certain noncancerous conditions [1,2]. It utilizes the photochemical reaction between a photosensitizer and oxygen to produce highly cytotoxic reactive oxygen species (ROS) for eradication of cancer cells and tissues. Compared to traditional anticancer modalities, PDT is less invasive and generally induces fewer side effects. Consequently, PDT has emerged as a promising approach for combating various malignancies [3,4]. However, the specificity of the photodynamic action remains a significant challenge that critically influences the treatment outcome. Thus, considerable interest has been directed toward the development of advanced photosensitizing systems capable of targeting the cancer cells and achieving controlled ROS generation at the tumor site, leaving the normal cells unaffected [5]. For chemotherapy, one of the effective strategies to mitigate the side effects on healthy tissues involves the use of prodrugs [6]. Inspired by this strategy, activatable photosensitizers have been developed in which the photosensitizing core is caged with a quencher or a deactivation mechanism that can be relaxed upon interactions with tumor-associated stimuli, leading to restoration of their photosensitizing property [7]. These smart photosensitizers are of great potential for precise PDT, and various advanced designs have been reported in recent years [8].

To facilitate the design and preparation of activatable photosensitizers, bioorthogonal chemistry utilizing highly efficient click reactions offers a powerful tool [9,10]. In fact, click chemistry has revolutionized the research at the interface of chemistry and biology [11]. Apart from the facile bond-formation characteristics, the bond-cleavage properties of certain click reactions have garnered significant interest due to their potential applications in chemical biology and drug activation [12,13]. In the realm of drug delivery, the release of active drug payload via click reaction can proceed with high spatiotemporal selectivity. In contrast to the traditional prodrug activation strategies that involve endogenous stimuli, which can differ among individuals and even among different cells within the same individual, this “click-and-release” strategy enables precise control by using an exogenously administered clickable reagent. Over the past decade, substantial progress has been made in this field with numerous bioorthogonal prodrugs being developed [[14], [15], [16]]. Among the various click methodologies, the inverse-electron-demand Diels-Alder (IEDDA) reaction between tetrazine and trans-cyclooctene has proven to be particularly effective for achieving specific drug activation in cells, living organisms, and animals, owing to its rapid click kinetics and high efficiency in prodrug activation [17]. Encouragingly, one such bioorthogonal prodrug from this category (SQ3370) has successfully entered Phase II clinical trials, which represents the first click chemistry in humans [18]. This body of work underscores the transformative potential of click chemistry in the development of sophisticated drug delivery systems, offering promising avenues for precise and effective treatment of diseases.

For in vivo applications, bioorthogonal bond-cleavage chemistry must exhibit both rapid reaction kinetics and high efficiency. This is critical because the “click-and-release” event must outcompete the clearance of both reagents to ensure sufficient payload release. Despite the success of existing technologies, reactions that fulfill these two criteria, i.e., featuring fast reaction kinetics (k₂ > 100 M⁻¹ s⁻¹) and traceless payload release with high efficiency (>90 %), are exceedingly rare. Even in the case of tetrazine and trans-cyclooctene pair utilized in the clinical trial, significant trade-offs exist between reaction kinetics and activation efficiency [[19], [20], [21]].

Bioorthogonal bond-cleavage chemistry mediated by transition metals has great potential to meet these requirements [[22], [23], [24]]. Distinct from small-molecule-based decaging reactions, the unique and adjustable electrochemical properties of metal ions enable the use of sub-stoichiometric amounts of the catalytic metal ions, allowing for multiple reaction turnovers. Since the report of [Ru(η⁵-Cp∗)(η⁴-cod)Cl] (Cp∗ = pentamethylcyclopentadienyl, cod = 1,5-cyclooctadiene), which can activate rhodamine dye via a deallyloxycarbonyl reaction inside living cells [25], a number of ruthenium derivatives have been developed for effective deprotection of O-allyl carbamate-caged compounds [[22], [23], [24],26,27]. This approach has been applied to prodrug activation, DNA binder activation, and site-specific protein modification [26,[28], [29], [30], [31]]. In comparing Ru-mediated deallylation with other metal-mediated decaging strategies (e.g., Pd and Au), despite direct kinetic comparisons are inherently complex, as reaction rates depend strongly on the specific metal center, ligand environment, and substrate structure, an important advantage of ruthenium-based systems lies in the high tunability of their deallylation kinetics through rational ligand design, providing a versatile platform for optimization toward bioorthogonal applications [32]. This molecular tunability enables fine control over reactivity, solubility, and stability, which is critical for efficient drug activation under physiological conditions. In contrast, palladium- and gold-mediated decaging reactions, while effective, often suffer from limited aqueous solubility and reduced stability in biological environments, frequently necessitating formulation into nanomaterials to achieve biological compatibility [33,34]. Such nanocarrier-based approaches introduce additional layers of complexity, including considerations of size, surface chemistry, long-term biodistribution, and nanomaterial-associated toxicity. Moreover, Pd complexes may raise concerns regarding cytotoxicity and variable reactivity depending on the catalytic system used, while Au-based strategies often exhibit slower reaction kinetics and strong interactions with endogenous thiols, which can attenuate catalytic efficiency in complex biological environments. Therefore, Ru-mediated deallylation represents a compelling balance between catalytic efficiency, chemical robustness, and biological compatibility. Coupled with the growing clinical interest in Ru-based compounds for cancer therapy, these features underscore the promise of Ru-mediated deallylation as a translationally relevant bioorthogonal strategy for prodrug activation.

We report herein the first use of this Ru-mediated bioorthogonal decaging chemistry for activation of photosensitizers toward targeted PDT. The study involved the preparation of a series of boron dipyrromethene (BDP)-based photosensitizers connected to an allyl group with or without a self-immolative spacer via an ester caging unit at the meso position [Pro-BDP-n (n = 1–4)] and ruthenium complexes [RuLn (n = 1–3)] (Fig. 1a), followed by screening of the decaging or activation efficiency. Pro-BDP-3 was found to be particularly potent that underwent deallylation followed by quinone methide elimination readily upon the treatment with RuL3 to give the activated species BDP-COOH with high conversion efficiency (>90 %) (Fig. 1b). In vitro studies found that this pair of bioorthogonal partners exhibited pronounced photocytotoxicity against cancer cells with high biotin receptor levels owing to the preferential cellular uptake of the biotin-modified RuL3, while having minimal impact on the biotin receptor-negative cells. Further mechanistic studies revealed that the photodynamic action occurred mainly at the endoplasmic reticulum (ER) of the cells. The resulting ER stress led to calcium leakage and overload, which disrupted the ER-mitochondrial crosstalk. This disturbance caused mitochondrial damage, triggering cell death via apoptosis (Fig. 1c). The high treatment efficacy of Pro-BDP-3 combined with RuL3 was also demonstrated using a tumor-bearing mouse model. These findings are reported and discussed below in detail.

Fig. 1.

(a) Molecular structures of the allyl-caged Pro-BDP-n (n = 1–4) and the ruthenium complexes RuLn (n = 1–3). (b) Activation of Pro-BDP-3 by RuL3 to give BDP-COOH. (c) Schematic diagram showing the selective uptake of RuL3 for bioorthogonal activation of Pro-BDP-3 and the resulting cell death pathway.

2. Results and discussion

2.1. Design and synthesis of the bioorthogonal components

Owing to the tunable spectroscopic and photophysical properties and the ease of chemical modification, BDP derivatives have been extensively studied as photosensitizers for PDT [35,36]. Recently, the “ester-to-carboxylate” effect has been reported, through which the fluorescence and ROS generation capabilities of these photosensitizers can be activated [37,38]. Therefore, the previously reported BDP-COOH [39] was selected as the parent compound in our design. To mask its photoactivities, an allyl group was introduced via nucleophilic substitution to give Pro-BDP-1, which can be removed readily by the established transition metal-mediated deallylation reaction [[22], [23], [24], [25], [26], [27]]. Considering that the steric environment would significantly influence this decaging process, a self-immolative linker based on 4-hydroxybenzyl alcohol was also inserted to give Pro-BDP-2. To enable further tuning of the decaging efficiency through electronic effect, the electron-donating OCH₃ group and electron-withdrawing NO₂ group were also introduced to the linker to afford Pro-BDP-3 and Pro-BDP-4, respectively. The synthetic route of these activatable photosensitizers is shown in Scheme S1.

To promote the decaging process, ruthenium complexes were selected due to their high biocompatibility, low toxicity, and excellent stability in living systems [[25], [26], [27]]. Based on the seminal work of Meggers et al. [26], treatment of the commercially available RuL1 with the corresponding 2-quinolinecarboxylato ligands, RuL2 and RuL3 were prepared (Scheme S2). To achieve targeted delivery, a biotin moiety was incorporated in RuL3 for targeting the biotin receptor overexpressed in cancer cells. The experimental details for preparation of all these compounds and the characterizing data are given in Supplementary Information.

2.2. Screening of decaging efficiency

The electronic absorption spectra of BDP-COOH and Pro-BDP-n (n = 1–4) were first recorded in phosphate-buffer saline (PBS) at pH 7.4 with 10 % dimethyl sulfoxide (DMSO) (v/v) and 0.1 % Tween 80 (v/v), which were added to enhance the solubility of the compounds and prevent their self-aggregation, at different concentrations (1–10 μM) (Fig. S1–S5). Fig. 2a compiles the spectra recorded at 5 μM, which shows that all the activatable photosensitizers exhibit an intense absorption band at around 700 nm with significant red shift compared to BDP-COOH (at 645 nm). Upon excitation at 610 nm, BDP-COOH displayed intense fluorescence at 682 nm, while the emission of the other four ester analogues was negligible, showing over a 300-fold difference (Fig. 2b), indicating that esterification of BDP-COOH effectively quenched its fluorescence.

Fig. 2.

(a) Electronic absorption and (b) fluorescence spectra of BDP-COOH and Pro-BDP-n (n = 1–4) at 5 μM. (c) Release yields of the four activatable photosensitizers (5 μM) with or without the treatment with various ruthenium complexes (0.25 μM) at 37 °C for 12 h as determined by the fluorescence recovery at 682 nm. (d) Time-dependent activation of Pro-BDP-2 and Pro-BDP-3 (5 μM) by RuL2 and RuL3 (0.25 μM) at 37 °C. (e) Comparison of the rate of decay of DPBF (30 μM) as monitored spectroscopically at 416 nm under different conditions (λ_ex > 610 nm). The concentrations of BDP-COOH/Pro-BDP-3 and RuL2/RuL3 were fixed at 5 μM and 0.25 μM, respectively. For the last two groups, Pro-BDP-3 was pre-treated with RuL2 or RuL3 at 37 °C for 12 h. (f) Fluorescence intensities at 682 nm of Pro-BDP-3 (5 μM) upon treatment with various potential interfering species (5 mM) at 37 °C for 1 h or RuL2/RuL3 (0.25 μM) at 37 °C for 12 h. The intensity of BDP-COOH (5 μM) is also included for comparison. The solvent was PBS at pH 7.4 with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v) for all the above measurements.

Leveraging this observation, we proceeded to screen for the optimal combination of activatable photosensitizers and ruthenium complexes by monitoring the recovery of fluorescence in the same medium at 37 °C (Fig. S6). To our delight, the addition of 0.25 μM (0.05 equiv.) RuL2 or RuL3 almost completely restored the fluorescence of Pro-BDP-3 after 12 h, and the release yield was determined to be 94.8 % and 94.1 % for RuL2 and RuL3, respectively (Fig. 2c). Pro-BDP-2 was moderately reactive toward RuL2 and RuL3 with a release yield of 64.2 % and 64.4 %, respectively. Both Pro-BDP-1 and Pro-BDP-4 were less reactive, with all the yields below 15 %. RuL1 was almost unable to activate any of these compounds, underscoring the crucial role of the 2-quinolinecarboxylato ligand in maintaining the deallylation activity of these ruthenium complexes.

To explain these findings, a mechanism is proposed in Fig. S7 based on the one proposed for Ru-promoted cleavage of O-allyl carbamates [26]. It is believed that the ruthenium complex (RuL2 or RuL3) first undergoes solvation (e.g. with H₂O) to displace the η³-allyl group to form a Ru(II) active species. The ruthenium center then coordinates with the allyl group of the BDP in a η² fashion. The subsequent η²-alkene to η³-allyl conversion concomitantly releases BDP-COOH via quinone methide elimination. The resulting η³-allyl intermediate then undergoes nucleophilic attack to reform a η²-alkene complex, which is displaced by the solvent to regenerate the active species. As ruthenium-based intermediates are notoriously transient, no attempts were made to detect these intermediates. Instead, the quinone side product was traced to support the proposed mechanism. To this end, by using the rapid reaction between quinone and glutathione (GSH), 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) was used to quantify GSH and indirectly determine the quinone levels for different conditions [40]. As shown in Fig. S8, the GSH level was significantly reduced when a mixture of GSH (20 μM), Pro-BDP-3 (5 μM), and RuL3 (0.25 μM) in PBS was left for 12 h compared with the levels in the mixtures with BDP-COOH, Pro-BDP-3, and RuL3 alone used as the controls. These results indicated that quinone was generated in the mixture of Pro-BDP-3 and RuL3.

The release of BDP-COOH upon activation of Pro-BDP-3 by RuL3 was further confirmed by high-performance liquid chromatograph (HPLC) analysis. As shown in Fig. S9, a new peak appeared in the chromatogram of Pro-BDP-3 when it was treated with RuL3 for 3 h. With an identical retention time (33.0 min) as that of BDP-COOH, this band was attributed to this activated species. When the reaction time was extended to 12 h, the original peak of Pro-BDP-3 (retention time = 35.7 min) completely disappeared, concomitant with the full emergence of the peak of BDP-COOH. These HPLC results provided direct and conclusive evidence that RuL3 effectively catalyzed the conversion of Pro-BDP-3 to BDP-COOH, thereby robustly corroborating the proposed activation mechanism.

It is worth noting that Pro-BDP-1 could not be deallylated readily, which could be attributed to the steric hindrance caused by the two methyl groups (C1 and C7) that hinders the complexation between the allyl group of Pro-BDP-1 and the ruthenium center. It indicates the importance of the self-immolative linker in the design of these activatable photosensitizers. The reactivity of the other three photosensitizers followed the order Pro-BDP-3 > Pro-BDP-2 > Pro-BDP-4. Surprisingly, the introduction of an electron-withdrawing NO₂ group, which is expected to promote the self-immolation and release of BDP-COOH, reduced the rate of deallylation. The above results may suggest that the rate-determining step is the binding of the Ru(II) active species with the allyl-substituted BDPs instead of the subsequent decaging step.

In addition to the release yield, reaction kinetics were also followed for the deallylation of Pro-BDP-2 and Pro-BDP-3 triggered by RuL2 and RuL3 over a period of 12 h (Fig. S10). As summarized in Fig. 2d, Pro-BDP-3 exhibited a faster decaging rate than Pro-BDP-2 for both ruthenium complexes. This activatable photosensitizer was therefore selected for the subsequent studies. To determine the second-order rate constants (k₂) of the decaging of Pro-BDP-3 using RuL2 and RuL3, the k_obs values at different concentration of the two complexes were measured for the first 6 h, from which the k₂ values were determined to be 58.1 and 49.4 M⁻¹ s⁻¹, respectively (Fig. S11), using the previously reported method [27].

To examine the singlet oxygen generation capability of Pro-BDP-3 under different conditions, 1,3-diphenylisobenzofuran (DPBF) was used as the quencher, of which the electronic absorption at 416 nm decreases upon reaction with singlet oxygen. As shown in Fig. 2e, Pro-BDP-3 alone was unable to degrade DPBF upon light irradiation (λ_ex > 610 nm), suggesting that the singlet oxygen generation ability of Pro-BDP-3 was largely inhibited. In contrast, after the treatment with RuL2 or RuL3 at 37 °C for 12 h, Pro-BDP-3 generated singlet oxygen effectively, as evidenced by the rapid reduction in the absorbance at 416 nm. The rate of decay was comparable with that for BDP-COOH. As expected, both RuL2 and RuL3 could not consume DPBF. These results clearly showed that the ROS generation ability of Pro-BDP-3 could be restored by RuL2 and RuL3.

To examine the selectivity of Pro-BDP-3 toward RuL2 and RuL3, a range of biological agents commonly found in the cellular environment, including Cu²⁺, Mg²⁺, Ca²⁺, Fe³⁺, glutathione (GSH), cysteine (Cys), bovine serum albumin (BSA), H₂O₂, and NaOH were also added to study the effect on fluorescence recovery. As shown in Fig. 2f, none of these agents could significantly restore the fluorescence. In contrast, both RuL2 and RuL3 could largely increase the fluorescence intensity. As expected, the intensity was comparable with that of BDP-COOH, which is the activated form of the photosensitizer. These results showed that the decaging process was highly selective.

Before proceeding to the in vitro studies, the stability of RuL3 under physiological conditions was examined. The complex was first dissolved in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10 % fetal calf serum. After 6, 12, and 24 h, its catalytic activity toward activation of Pro-BDP-3 was assessed by measuring the fluorescence intensity at 682 nm after 12 h, using BDP-COOH for comparison. It was found that under all these conditions, fluorescence was largely restored by >95 %, and the intensity was only marginally reduced when the incubation time increased (Fig. S12). The similar catalytic activity of RuL3 for these conditions indicated that the complex was stable and remained active under physiological conditions for at least 24 h.

Similarly, the stability of Pro-BDP-3 in this serum-containing medium was also studied. It was found that the fluorescence intensity of this compound remained nearly undetectable after incubation in this medium for 6, 12, and 24 h (Fig. S13). This indicated that Pro-BDP-3 remained structurally intact and was not activated by endogenous biomolecules present in the physiological environment. However, when RuL2 or RuL3 was added to Pro-BDP-3 in this medium that had been left for 24 h, followed by further incubation for 12 h, a significant increase in fluorescence was observed, confirming the successful release of the active parent compound BDP-COOH. Based on the fluorescence intensity at 682 nm, the release yield exceeded 90 %, demonstrating that Pro-BDP-3 experienced minimal degradation (less than 10 %) in this serum-containing medium and remained highly responsive to Ru-mediated activation.

2.3. In vitro activation of photoactivities

To examine the cell-selective activation of Pro-BDP-3 by RuL3, the biotin receptor-overexpressed HeLa human cervical cancer cells, 4T1 murine mammary carcinoma cells, and MCF-7 human breast cancer cells were used as the positive cell lines, while NIH 3T3 murine embryonic fibroblast cells, which exhibit low biotin receptor expression, served as a negative control [41]. We first screened the optimal incubation conditions, focusing on the concentration and incubation time of RuL3, by monitoring the recovery of fluorescence in HeLa cells. Given the high cellular uptake of distyryl BDPs [39], HeLa cells were first treated with Pro-BDP-3 (5.0 μM) for 2 h, followed by incubation with RuL3 (1.0, 2.5, or 5.0 μM) for 2, 4, or 6 h. The intracellular fluorescence was then determined by flow cytometry. As shown in Fig. S14a, the intracellular fluorescence intensity increased generally with the concentration and incubation time of RuL3. Based on the combination indices determined [42], the optimal incubation conditions were ascertained to be 2.5 μM (0.5 equiv.) of RuL3 for 4 h (Fig. S14b).

By comparing the intracellular fluorescence intensities under different conditions with that of BDP-COOH, the corresponding intracellular release yields were determined, and the values are summarized in Table S1. Under the optimal incubation conditions as mentioned above, the release yield was determined to be 87.6 %, showing that the bioorthogonal decaging reaction proceeded effectively within the cellular environment.

To investigate whether catalytic amounts of ruthenium complexes could be used for activation of the photosensitizers, HeLa cells were incubated with Pro-BDP-3 (10 μM) for 2 h, followed by incubation with RuL2 or RuL3 (0.01 μM, i.e., 0.1 mol%) for different periods of time. The intracellular fluorescence intensities were determined by flow cytometry, and the release yield was monitored over a period of 48 h by comparing with the intensity of BDP-COOH. As shown in Fig. S15a–c, the fluorescence intensity increased with the incubation time and reached a plateau after 24 h for both complexes. The turnover number (TON) at 48 h was determined to be 124 and 94 for RuL2 and RuL3, respectively, according to the method reported previously [26]. The values were similar to those measured in PBS solution, which were 143 and 136, respectively (Fig. S15d). These results showed that the activation of these photosensitizers could be achieved by using as low as 0.1 mol% of the ruthenium complexes.

Nevertheless, the aforementioned optimized conditions were used to study the activation of Pro-BDP-3 by RuL3 in the four cell lines mentioned above. As shown in Fig. 3a and b, remarkable red fluorescence was observed in the biotin receptor-positive HeLa, 4T1, and MCF-7 cells after being sequentially incubated with Pro-BDP-3 and RuL3, while the intracellular fluorescence of the biotin receptor-negative NIH 3T3 cells was not noticeable. In contrast, by using the non-biotin-conjugated RuL2 instead of RuL3 for incubation, the intracellular fluorescence of NIH 3T3 cells was as strong as that of the other three cell lines, and the intensities were similar to those observed when the cells were incubated with BDP-COOH. Without subsequent incubation with these ruthenium complexes, the fluorescence could hardly be observed for all the four cell lines. These results showed that the biotin moiety of RuL3 could promote the uptake by biotin receptor-positive cancer cells, and the ruthenium complexes play a crucial role in restoring the fluorescence emission of Pro-BDP-3.

Fig. 3.

(a) Fluorescence confocal images of HeLa, 4T1, MCF-7, and NIH 3T3 cells after incubation with Pro-BDP-3 (5.0 μM) for 2 h with or without further incubation with RuL2 or RuL3 (2.5 μM) for a further 4 h (red fluorescence; λ_ex = 633 nm, λ_em = 650–900 nm). The cells being incubated with BDP-COOH (5.0 μM) for 2 h were used as the positive control. The cell nuclei were stained with Hoechst (1.0 μM) for 15 min (blue fluorescence; λ_ex = 405 nm, λ_em = 420–500 nm). Scale bar = 20 μm. (b) Corresponding mean red fluorescence intensities quantified by ImageJ. Data are reported as the mean ± standard error of the mean (SEM) for three independent experiments (∗∗∗∗p < 0.0001). (c) Fluorescence confocal images of HeLa, 4T1, MCF-7, and NIH 3T3 cells after the aforementioned treatments and further incubation with H₂DCFDA (10 μM) for 30 min, followed by light irradiation (λ > 610 nm, 25.8 mW/cm²) for 8 min to give a total fluence of 12 J/cm² (green fluorescence; λ_ex = 488 nm, λ_em = 493–550 nm). Scale bar = 20 μm. (d) Corresponding mean green fluorescence intensities of DCF quantified by ImageJ. Data are reported as the mean ± SEM for three independent experiments (∗∗∗∗p < 0.0001). (e) Dark and photo (λ > 610 nm, 25.8 mW/cm², 12 J/cm²) cytotoxicity of BDP-COOH, Pro-BDP-3, RuL2, Pro-BDP-3 + RuL2, RuL3, and Pro-BDP-3 + RuL3 against HeLa, 4T1, MCF-7, and NIH 3T3 cells. The cells were incubated with BDP-COOH, Pro-BDP-3, RuL2, or RuL3 for 2 h. For Pro-BDP-3 + RuL2 and Pro-BDP-3 + RuL3, the cells were first incubated with Pro-BDP-3 for 2 h and then with RuL2 or RuL3 (0.5 equiv.) for a further 4 h. Data are expressed as the mean ± SEM of three independent experiments, each performed in quadruplicate. (f) Photocytotoxicity of these agents at 5.0 μM and the combination treatments at 5.0 μM of Pro-BDP-3 against the four cell lines. The rightmost figure compiles the results for Pro-BDP-3 + RuL3 (∗∗∗∗p < 0.0001). Data are expressed as the mean ± SEM of three independent experiments, each performed in quadruplicate. (g) Live/dead cell viability assay using calcein-AM and PI. The cells were treated as described above, followed by incubation with calcein-AM (1 μM) and PI (2 μM) in binding buffer (2 mL) at 37 °C for 30 min. The live cells were indicated by the green fluorescence of calcein-AM (λ_ex = 488 nm, λ_em = 493–550 nm), while the dead cells were indicated by the red fluorescence of PI (λ_ex = 561 nm, λ_em = 600–800 nm). Scale bar = 50 μm. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

To quantitatively assess the selective internalization of RuL3, inductively coupled plasma mass spectrometry (ICP-MS) was used to measure its cellular uptake across different cell lines, using RuL2 as a negative control. The results showed that these two complexes displayed distinct internalization profiles. While RuL2 was internalized non-specifically by the above four cell lines, regardless of their biotin receptor expression levels, after incubation for 4 h, RuL3 exhibited selective uptake toward the biotin receptor-positive cells (HeLa, 4T1, and MCF-7), with significantly lower levels detected in the biotin receptor-negative NIH-3T3 cells (Fig. S16). These quantitative data provide strong evidence that RuL3 can be selectively internalized by cells overexpressing the biotin receptor, which is a crucial foundation for the targeted activation mechanism proposed in this work.

Similar results were obtained in the activation of intracellular ROS generation. In this study, the non-fluorescent 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA) was used as the probe, which converts to the highly fluorescent 2′,7′-dichlorofluorescein (DCF) upon hydrolysis and oxidation by ROS inside the cells. As shown in Fig. 3c and d, intense green fluorescence of DCF was observed for the four cell lines being incubated with BDP-COOH or Pro-BDP-3 and then with RuL2, followed by light irradiation (λ > 610 nm, 25.8 mW/cm²). When the cells were just incubated with Pro-BDP-3, the fluorescence was negligible. Upon post-incubation with RuL3, bright fluorescence could only be seen for the three biotin-receptor-positive cell lines, but not the negative control. These results showed that the ROS generation of Pro-BDP-3 could also be selectively restored by RuL3 inside the cells.

To reveal the nature of the ROS generated, singlet oxygen sensor green (SOSG) was used for detection, which is a specific probe of singlet oxygen. As shown in Fig. S17, the intracellular fluorescence intensity for the treatment group of Pro-BDP-3 + RuL2 was comparable for all the four cell lines as in the case of BDP-COOH used as a positive control. This indicated that RuL2 effectively activated Pro-BDP-3 to produce singlet oxygen, albeit without cellular selectivity. In contrast, for the treatment of Pro-BDP-3 + RuL3, there was a significant increase in SOSG fluorescence in the biotin receptor-positive cells (HeLa, 4T1, and MCF-7), but not in the receptor-negative NIH-3T3 cells. These results provided more convincing and specific evidence for the selective generation of singlet oxygen following bioorthogonal activation in target cells.

Being encouraged by these results, we proceeded to evaluate the antiproliferative activity of Pro-BDP-3 in combination with RuL2 or RuL3 using the MTT assay [MTT = 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide]. The cytotoxicity of individual components and BDP-COOH was also determined for comparison. As shown in Fig. 3e, all of these agents and the combination treatments Pro-BDP-3 + RuL2 and Pro-BDP-3 + RuL3 showed negligible antiproliferative activity against the four cell lines in the dark at concentrations up to 5.0 μM. Upon light irradiation (λ > 610 nm, 25.8 mW/cm², 12 J/cm²), the parent compound BDP-COOH exhibited high photocytotoxicity against these cells with half maximal inhibitory concentrations (IC₅₀ values) of ca. 2 μM. In contrast, Pro-BDP-3 was essentially noncytotoxic with less than 10 % of these cells being killed at 5.0 μM. Similarly, the cytotoxicity of RuL2 and RuL3 was also negligible. Interestingly, the combination of Pro-BDP-3 and RuL2 successfully restored the photocytotoxicity against all the four cell lines, displaying an inhibitory activity comparable to BDP-COOH. Similar profiles were obtained for the combination of Pro-BDP-3 and RuL3 against HeLa, 4T1, and MCF-7 cells. However, the photocytotoxicity was significantly reduced for NIH 3T3 cells, which could be attributed to the low uptake of RuL3 in these biotin receptor-negative cells, and these results were consistent with those for ROS restoration (Fig. 3c and d). Fig. 3f shows the photocytotoxicity of these agents at 5.0 μM and the combination treatments at 5.0 μM of Pro-BDP-3 against the four cell lines, as well as a comparison of the photocytotoxicity of Pro-BDP-3 + RuL3 against these cells (the rightmost figure), in which the activation of the photodynamic activity of Pro-BDP-3 by RuL2 and RuL3 could clearly be seen, and the cell-selective property of the latter was also evidenced. The IC₅₀ values of BDP-COOH and the combination treatments of Pro-BDP-3 + RuL2 and Pro-BDP-3 + RuL3 for the four cell lines are compiled in Table 1 for comparison. As expected, after activation with RuL2 or RuL3, the IC₅₀ values of Pro-BDP-3 for the three biotin-receptor-positive cell lines were comparable with those of BDP-COOH, which was in fact the activated product. For the biotin receptor-negative NIH 3T3 cells, while the IC₅₀ value of Pro-BDP-3 + RuL2 was similar to that of BDP-COOH, the value could not be determined for Pro-BDP-3 + RuL3.

Table 1.

IC₅₀ values (in μM) of BDP-COOH, Pro-BDP-3 + RuL2, and Pro-BDP-3 + RuL3 against HeLa, 4T1, MCF-7, and NIH 3T3 cells upon light irradiation (λ > 610 nm, 25.8 mW/cm², 12 J/cm²).

	HeLa	4T1	MCF-7	NIH 3T3
BDP-COOH	2.32 ± 0.03	2.45 ± 0.05	2.18 ± 0.03	1.76 ± 0.03
Pro-BDP-3 + RuL2	2.67 ± 0.05	2.83 ± 0.06	1.85 ± 0.05	1.82 ± 0.07
Pro-BDP-3 + RuL3	2.81 ± 0.02	2.75 ± 0.03	2.20 ± 0.02	---[a]

^[a]The value could not be determined up to 5.0 μM.

The above results were further corroborated by a live/dead cell viability assay. As shown in Fig. 3g, without post-treatment with RuL2 or RuL3, Pro-BDP-3 displayed bright green fluorescence of calcein AM in all the four cell lines, showing that the cells remained viable. Upon post-treatment with RuL2, the cells showed strong red fluorescence of propidium iodide (PI), indicating extensive cell death, as in the case of BDP-COOH. With follow-up treatment with RuL3, Pro-BDP-3 could also effectively kill the three biotin-receptor-positive cell lines, giving strong red fluorescence, while the biotin receptor-negative NIH 3T3 cells were not significantly eliminated, leaving substantial live cells as indicated by the green fluorescence. These results were fully consistent with those obtained by MTT assay.

2.4. Studies of cell death mechanism

To reveal the mechanism behind the antiproliferative activity of the activated form of Pro-BDP-3, i.e., BDP-COOH, we initially investigated its subcellular distribution in HeLa cells using the commercially available organelle-selective trackers. It was found that the fluorescence of BDP-COOH generated in situ upon sequential incubation with Pro-BDP-3 and RuL3 showed substantial overlap with that of ER-Tracker Green with a Pearson's coefficient of 0.90 (Fig. 4a and b), showing that the activated form of Pro-BDP-3 localized predominantly in the endoplasmic reticulum (ER). The Pearson's coefficients for the Mito-Tracker and Lyso-Tracker overlapped images were just around 0.6 (Fig. S18). In contrast, the subcellular localization of BDP-COOH used directly for incubation appeared to be non-selective, giving Pearson's coefficients of 0.64, 0.63, and 0.56 for the images of ER-Tracker Green, Mito-Tracker, and Lyso-Tracker, respectively (Fig. 4a and b, and S18). The disparity in distribution might be ascribed to the contrasting hydrophobicity of BDP-COOH and its ester analogue Pro-BDP-3. This hypothesis was supported by the diverse cLogP values calculated by X-LOGP3 as described previously [43], with BDP-COOH exhibiting a value of 6.67 and Pro-BDP-3 exhibiting a value of 9.08.

Fig. 4.

(a) Study of subcellular localization of BDP-COOH and Pro-BDP-3 + RuL3 in the ER of HeLa cells. The cells were incubated with BDP-COOH or Pro-BDP-3 (5.0 μM) for 2 h, followed by incubation with free medium or RuL3 (2.5 μM), respectively, for 4 h. The cells were then treated with ER-Tracker Green (1.0 μM) for 20 min (BDP: red fluorescence, λ_ex = 633 nm, λ_em = 650–900 nm; ER-Tracker Green: green fluorescence, λ_ex = 488 nm, λ_em = 493–550 nm). Scale bar = 20 μm or 70 μm (for the enlarged images). (b) Corresponding Pearson's values determined by ImageJ (∗∗∗∗p < 0.0001). Data are reported as the mean ± standard deviation (SD) for five independent experiments. (c) Change in fluorescence intensity of ER-Tracker Green in HeLa cells after the photodynamic treatment with BDP-COOH or Pro-BDP-3 + RuL3 over a period of 6 h. The cells were first incubated with these agents as described above. After light irradiation (λ > 610 nm, 25.8 mW/cm², 12 J/cm²), the cells were incubated in culture medium for a further 1, 3, or 6 h, and then stained with ER-Tracker Green (1.0 μM) for 20 min. The cell nuclei were stained with Hoechst (1.0 μM) for a further 15 min (ER-Tracker Green: green fluorescence, λ_ex = 488 nm, λ_em = 493–550 nm; Hoechst: blue fluorescence, λ_ex = 405 nm, λ_em = 420–500 nm). Scale bar = 20 μm. (d) Time-dependent relative mean fluorescence intensity (MFI) of ER-Tracker Green for the above two treatments. (e) Intracellular Ca²⁺ levels in HeLa cells after the photodynamic treatment with BDP-COOH or Pro-BDP-3 + RuL3. After drug incubation and light irradiation as described above, the cells were incubated in culture medium for 6 h and then stained with Fluo-4 AM (1.0 μM) for 30 min (green fluorescence, λ_ex = 488 nm, λ_em = 493–550 nm). The cell nuclei were stained with Hoechst (1.0 μM) for a further 15 min. Scale bar = 20 μm. (f) Analysis of ΔΨ_m on HeLa cells after the photodynamic treatment with BDP-COOH or Pro-BDP-3 + RuL3. After drug incubation and light irradiation as described above, the cells were incubated in culture medium for 6 h and then stained with JC-1 (10 μg mL⁻¹) for 30 min (red fluorescence, λ_ex = 561 nm, λ_em = 570–680 nm; green fluorescence, λ_ex = 488 nm, λ_em = 495–550 nm). (g) Apoptotic percentages of HeLa and NIH 3T3 cells upon various treatments as determined by flow cytometry. The cells were treated with BDP-COOH or Pro-BDP-3 (5.0 μM) for 2 h, followed by incubation with free medium or RuL2/RuL3 (2.5 μM), respectively, for 4 h. After light irradiation (λ > 610 nm, 25.8 mW/cm², 12 J/cm²), the cells were incubated for 6 h and then stained with Annexin V-FITC (1.0 μM, λ_ex = 488 nm, λ_em = 505–545 nm) and PI (1.0 μM, λ_ex = 561 nm, λ_em = 564–606 nm) in binding buffer for 20 min. (h) Percentages of early and late apoptotic HeLa and NIH 3T3 cells after the treatment with BDP-COOH, Pro-BDP-3 + RuL2, and Pro-BDP-3 + RuL3 (∗p < 0.05, ∗∗p < 0.01, ∗∗∗P < 0.001, and ∗∗∗∗P < 0.0001). For (g) and (h), data are reported as the mean ± SEM for three independent experiments. (i) Schematic diagram showing the photodynamic effects induced by Pro-BDP-3 + RuL3 in HeLa cells. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

As reported previously, ER-targeting photosensitizers can disrupt ER and trigger ER stress upon light irradiation [44,45]. Consequently, we sought to investigate whether the ER-localized Pro-BDP-3, after being activated with RuL3, would cause similar effects. To confirm the presence of ER stress, the morphological change in the ER of HeLa cells was monitored using ER-Tracker Green over a period of time. The results showed that after incubation with Pro-BDP-3 and RuL3 followed by light irradiation, the green fluorescence of the ER probe diminished gradually over 6 h, suggesting that the ROS generated by the activated form of Pro-BDP-3 disrupted the ER, leading to the onset of ER stress. In contrast, the green fluorescence was not significantly changed when the cells were incubated with BDP-COOH instead (Fig. 4c and d).

Considering the well-established connection between ER stress and disruption of calcium homeostasis [46], we further examined the ER stress caused by the activated form of Pro-BDP-3 by measuring the intracellular Ca²⁺ levels using Fluo-4 AM, a cell-permeable fluorescent dye for Ca²⁺ detection. Notably, the activated Pro-BDP-3 prominently elevated the intracellular Ca²⁺ levels in HeLa cells (Fig. 4e). In contrast, HeLa cells treated with BDP-COOH directly exhibited only faint green fluorescence of Fluo-4 AM. These results further supported that the combined treatment of Pro-BDP-3 and RuL3 could effectively induce ER stress upon light irradiation, while BDP-COOH could not give this photodynamic effect.

Ca²⁺ ions function as a second messenger and play a crucial role in the complex and extensive information exchange system between ER and mitochondria, known as ER-mitochondria crosstalk. Therefore, ER stress can directly contribute to mitochondrial dysfunction by Ca²⁺ overload [47]. To investigate mitochondrial damage, we utilized 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolo-carbocyanine iodide (JC-1), a probe that is responsive toward the mitochondrial membrane potential (ΔΨ_m). At high ΔΨ_m, JC-1 forms aggregates and emits red fluorescence at 590 nm. Conversely, at low ΔΨ_m it exists as a monomer and emits green fluorescence at 525 nm. A decrease in the ratio of red to green fluorescence intensity indicates mitochondrial depolarization, which is a characteristic of mitochondrial dysfunction. Our findings revealed rapid depolarization of the mitochondrial membrane in HeLa cells with sequential incubation with Pro-BDP-3 and RuL3 at 3 h after the PDT treatment. In contrast, when the cells were treated with BDP-COOH, green fluorescence of JC-1 monomer could only be observed at 6 h after the PDT treatment (Fig. 4f).

As mitochondrial membrane depolarization is closely associated with apoptosis and cell death, we stained HeLa and NIH 3T3 cells with Annexin V – fluorescein isothiocyanate (FITC) and PI to evaluate their apoptotic status following different treatments. As shown in Fig. 4g, the proportion of apoptotic HeLa and NIH 3T3 cells increased significantly upon treatment with BDP-COOH and light irradiation. A similar trend was observed for the cells treated with Pro-BDP-3 + RuL2. Without being activated, Pro-BDP-3 could not give apoptotic cells. Interestingly, due to the selective accumulation of RuL3 in cells with high biotin receptor levels, the photodynamic action initiated by Pro-BDP-3 + RuL3 induced apoptosis exclusively in HeLa cells, but not in NIH 3T3 cells, which was consistent with the observed photocytotoxicity (Fig. 3e, f, and 3g).

Moreover, we also discovered distinct stages of apoptosis in HeLa cells after the treatments of BDP-COOH, Pro-BDP-3 + RuL2, and Pro-BDP-3 + RuL3. Specifically, BDP-COOH primarily induced early apoptosis in HeLa cells, while the cells treated with Pro-BDP-3 + RuL2 or Pro-BDP-3 + RuL3 were predominantly in late apoptosis (Fig. 4h). For NIH 3T3 cells, both BDP-COOH and Pro-BDP-3 + RuL2 primarily induced early apoptosis. For the treatment group of Pro-BDP-3 + RuL3, most of the cells remained viable.

The induction of apoptosis was further confirmed at the protein level. Western blot analysis was performed to examine the expression of some key apoptosis-related markers after various treatments on HeLa cells. It was found that for the treatment groups of Pro-BDP-3 + RuL2 and Pro-BDP-3 + RuL3, there was a notable upregulation of the pro-apoptotic proteins cleaved caspase-3 and Bax with concurrent downregulation of the anti-apoptotic protein Bcl-2 (Fig. S19). In contrast, this result was not observed for the treatment groups of BDP-COOH and Pro-BDP-3 alone. These protein-level findings are consistent with the results obtained from the Annexin V-FITC/PI staining assay, collectively providing robust evidence that the bioorthogonal system effectively triggers apoptotic cell death.

Fig. 4i depicts the overall photodynamic effects of Pro-BDP-3 + RuL3. Owing to the high cLogP value, Pro-BDP-3 preferentially localized in ER. Upon in situ activation by RuL3, Pro-BDP-3 induced ER stress upon irradiation, disrupting the intracellular calcium homeostasis and ER-mitochondria crosstalk. This disruption subsequently led to mitochondria dysfunction, ultimately resulting in apoptosis and cell death.

2.5. In vivo studies

To evaluate the in vivo unmasking of Pro-BDP-3 by RuL3, HeLa xenografted nude mice were first treated with an intravenous injection of RuL3 in PBS with 5 % DMSO (v/v) (100 μL, 10 nmol, 0.5 nmol per gram of mice). After a 2-h interval, the mice were intratumorally injected with Pro-BDP-3 in PBS with 10 % DMSO (v/v) (30 μL, 20 nmol, 1 nmol per gram of mice). This dual injection method was employed to prevent premature activation of Pro-BDP-3 in the blood circulation, as commonly used in bioorthogonal-mediated activation studies [48]. Mice without the injection of RuL3 were used as the control group. Following these treatments, we continuously monitored the whole-body fluorescence images (λ > 700 nm) of the mice for 48 h and quantified the fluorescence intensity per unit area of the tumor at different time points (Fig. 5a). As shown in Fig. 5b and c, for the mice treated with RuL3 and Pro-BDP-3, the fluorescence intensity in the tumor increased over time, reaching its maximum at 12 h post-injection of Pro-BDP-3. At this time point, the intensity was approximately 8-fold higher than that for the control group. At the end of the monitoring period, the mice were sacrificed, and the tumor as well as major organs were harvested for fluorescence imaging (Fig. 5d). As expected, bright fluorescence was observed in the tumor of the positive treatment group, and the intensity was 5-fold higher than that of the control group (Fig. 5e). No significant fluorescence signal was detected in major organs, including heart, liver, spleen, lung, and kidney. These results showed that the biotin-conjugated RuL3 could accumulate in the biotin receptor-overexpressed HeLa tumor and activate Pro-BDP-3 therein through bioorthogonal deallylation to restore the fluorescence emission.

Fig. 5.

(a) Timeline for monitoring the activation of in vivo fluorescence emission of Pro-BDP-3 by RuL3. (b) In vivo fluorescence images of HeLa tumor-bearing nude mice after intratumoral injection of Pro-BDP-3 (1 nmol per gram of mice) with or without pre-intravenous injection of RuL3 (0.5 nmol per gram of mice) over a period of 48 h. (c) Change in fluorescence intensity per unit area of the tumor for the above two treatment groups along with time. (d) Ex vivo fluorescence images of tumors and main organs at 48 h post-injection for the above two treatment groups. (e) Fluorescence intensities of the tumors harvested from the above two groups of mice at 48 h post-injection. (f) Timeline for studying the activation of in vivo PDT effect of Pro-BDP-3 by RuL3. (g) Change in tumor size of the mice after the different treatments. (h) Relative tumor growth curves of the mice after the different treatments. (i) Tumor growth inhibition rates, defined as 1 – (mean volume of the treated tumors)/(mean volume of the control tumors), for different treatments. (j) Tumor burden, defined as the tumor weight/body weight on day 12, for different treatments. (k) Variation of the body weight of the mice after different treatments. For (c), (e), (h)–(k), data are reported as the mean ± SD (n = 4 for imaging study and 5 for PDT study).

The PDT efficacy of RuL3 + Pro-BDP-3 was then investigated using 4T1 tumor-bearing nude mice according to the timeline shown in Fig. 5f. Once the tumor size reached approximately 80–100 mm³, the mice were randomly divided into six groups (n = 5) and received different treatments: (1) PBS with light irradiation, (2) RuL3 + Pro-BDP-3, (3) Pro-BDP-3 with light irradiation, (4) RuL3 only, (5) RuL3 + Pro-BDP-3 with light irradiation, and (6) BDP-COOH with light irradiation. The mice were first intravenously injected with either PBS or RuL3 in PBS with 5 % DMSO (100 μL, 10 nmol, 0.5 nmol per gram of mice) (except group 3 and 6). After 2 h, Pro-BDP-3 in PBS with 10 % DMSO (v/v) (30 μL, 20 nmol, 1 nmol per gram of mice) was administrated via intratumoral injection for group 2, 3, and 5. For the positive control group 6, the same dose of BDP-COOH was also injected intratumorally. For the light treatment groups, the tumors were irradiated with a diode laser at 675 nm operated at 0.1 W for 10 min (total fluence = 60 J/cm²) at 12 h after the injection of Pro-BDP-3 or BDP-COOH, when the localization of Pro-BDP-3 at the tumor reached almost the maximum (Fig. 5c). As depicted in Fig. 5g–j and S20, the combination treatment of RuL3 and Pro-BDP-3 with laser irradiation significantly inhibited the tumor growth as in the case of BDP-COOH with light irradiation, showing that Pro-BDP-3 was fully decaged and activated at the tumor site. In the absence of laser irradiation, the inhibition effect was not significant compared to the PBS-treated control group. Moreover, Pro-BDP-3 alone did not exert a significant PDT effect, indicating its dependence on activation by RuL3. In addition, the body weight of the mice was monitored after all these treatments. As shown in Fig. 5k, no significant changes were observed over a period of 12 days, showing that the different treatments did not cause noticeable toxic side effects to the mice. To provide further evidence, hematoxylin and eosin (H&E) staining of the major organs, including heart, liver, spleen, lungs, and kidneys was performed after all these treatments. As shown in Fig. S21, no noticeable histopathological abnormalities were observed for all the organ sections, indicating a favorable safety profile for all these treatments. In contrast, for the treatment group of RuL3 + Pro-BDP-3 with light irradiation, the tumor section showed extensive cell death as in the case of the positive control BDP-COOH with light irradiation, while this observation was not detected in the tumor sections for the other treatments. The overall results firmly established the feasibility of RuL3-mediated bioorthogonal deallylation of Pro-BDP-3 for in vivo targeted PDT.

3. Conclusions

Compared to traditional activating strategies of photosensitizers, bioorthogonal approach offers several advantages, particularly the high spatiotemporal selectivity that enables precise PDT. The recent initiation of clinical trials of SQ3370 represents a major milestone in the development of bioorthogonal prodrugs. In this study, we have developed a simple yet efficient bioorthogonal strategy for site-specific activation of far-red-absorbing photosensitizers. This approach involves the use of allyl-caged distyryl BDP-based photosensitizers that can be activated through the highly efficient ruthenium-mediated deallylation. Pro-BDP-3, in particular, was the most promising candidate in the series being studied, exhibiting the highest release efficiency and fastest decaging kinetics. By using the biotin-modified ruthenium complex RuL3 as an activator, which could be selectively taken up by biotin receptor-positive cancer cells and tumors, Pro-BDP-3 could be specifically activated to restore the fluorescence emission and ROS generation, leading to efficient PDT, both in vitro and in vivo. Detailed in vitro mechanistic studies revealed that Pro-BDP-3 localized in the ER of HeLa cells, inducing ER stress upon photosensitization. It resulted in disruption of the intracellular calcium homeostasis and ER-mitochondria crosstalk, leading to mitochondria dysfunction and cell death through apoptosis.

While the established bioorthogonal activation of photosensitizers using IEDDA reactions, for example between tetrazine and trans-cyclooctene moieties, has been shown to be effective, both in vitro and in vivo, this transition metal-mediated deallylation strategy provides additional advantages. As mentioned above, owing to the catalytic nature of the activators, only a sub-stoichiometric amount of metal catalysts is needed for activation. This is particularly advantageous for activating substrates that require high intracellular concentrations to exert their effects, as using stoichiometric bioorthogonal pairs at such high doses may lead to practical challenges and potential cytotoxicity from the activators. In addition, the deallylation kinetics can be tuned by changing the coordination environment, providing a facile approach to optimize the activation process. Preparation of allyl derivatives and metal complexes is also relatively straightforward compared with the chemical modification of conventional bioorthogonal components. Overall speaking, our results showed that RuL3-mediated bioorthogonal deallylation is highly specific and efficient that can expand the toolbox of bioorthogonal activation of photosensitizers for precise PDT.

4. Materials and methods

4.1. General

All the solvents and reagents were of high-performance liquid chromatography (HPLC) and reagent grade, respectively, and used as received. All the reactions were monitored by thin-layer chromatography (TLC) performed on pre-coated 0.20 mm silica gel 60 UV254 plates and visualized under UV light. Chromatographic purification was carried out on silica gel (Macherey-Nagel, 230–400 mesh) with the indicated eluent. ¹H and ¹³C{¹H} NMR spectra were recorded with a Bruker AVANCE III HD-300, HD-400, or HD-500 spectrometer in deuterated solvents. Spectra were referenced internally using the residual solvent [¹H: δ = 7.26 (for CDCl₃) or 1.94 (for CD₃CN)] or solvent [¹³C: δ = 77.2 (for CDCl₃) or 118.2 (for CD₃CN)] resonances relative to SiMe₄. The values of the chemical shifts are expressed in δ values (ppm) and the coupling constants (J) in Hz. Electrospray ionization (ESI) mass spectra were recorded on a Finnigan MAT 95 XL spectrometer. Electronic absorption and steady-state fluorescence spectra were taken on a Cary 5G UV–Vis–NIR spectrophotometer and a Hitachi F-7000 spectrofluorometer, respectively.

4.2. Synthesis of 1-allyloxy-4-bromomethyl-2-nitrobenzene (4)

To a stirred solution of 4-allyloxy-3-nitrobenzyl alcohol [49] (0.21 g, 1.0 mmol) in CH₂Cl₂ (15 mL) at 0 °C, PBr₃ (0.27 g, 1.0 mmol) was added. After the complete consumption of 4-allyloxy-3-nitrobenzyl alcohol as indicated by TLC, the reaction was quenched with water (10 mL), and the mixture was extracted with ethyl acetate (20 mL). The organic layer was dried over anhydrous Na₂SO₄, filtered, and evaporated under reduced pressure. The crude product 4 was directly used for next step without further purification.

4.3. General procedure for synthesis of Pro-BDP-n (n = 1–4)

A mixture of BDP-COOH [48] (104 mg, 0.1 mmol), alkyl bromides 1–4 (0.11 mmol), and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU, 17 μL, 0.11 mmol) in CH₂Cl₂ (10 mL) was stirred at room temperature for 6 h. After the complete consumption of BDP-COOH as indicated by TLC, the reaction was quenched with 20 % aqueous NH₄Cl (10 mL), and the mixture was extracted with CH₂Cl₂ (20 mL). The organic layer was dried over anhydrous Na₂SO₄, filtered, and evaporated under reduced pressure. The crude product was purified by silica gel column chromatography using CH₂Cl₂/MeOH (30:1, v/v) as the eluent to give Pro-BDP-n (n = 1–4).

Pro-BDP-1: green solid (60 mg, 55 %), ¹H NMR (300 MHz, CDCl₃): δ 8.19 (d, J = 16.5 Hz, 2H, C=CH), 7.60 (d, J = 9.0 Hz, 4H, Ar-H), 7.54 (d, J = 16.5 Hz, 2H, C=CH), 6.96 (d, J = 9.0 Hz, 4H, Ar-H), 5.96–6.09 (m, 1H, OCH₂CH), 5.46–5.53 (m, 1H, C=CH), 5.40 (dd, J = 10.5, 1.2 Hz, 1H, C=CH), 4.89 (d, J = 6.0 Hz, 2H, OCH₂), 4.19 (t, J = 5.1 Hz, 4H, OCH₂), 3.89 (t, J = 4.8 Hz, 4H, OCH₂), 3.75–3.78 (m, 4H, OCH₂), 3.70–3.72 (m, 4H, OCH₂), 3.65–3.68 (m, 4H, OCH₂), 3.54–3.58 (m, 4H, OCH₂), 3.39 (s, 6H, OCH₃), 2.21 (s, 6H, CH₃). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 165.1, 160.3, 151.7, 143.5, 140.1, 130.2, 130.0, 129.5, 125.1, 121.1, 116.5, 115.0, 82.6, 71.9, 70.9, 70.7, 70.6, 69.7, 67.6, 59.1, 15.8. HRMS (ESI): m/z calcd for C₄₅H₅₃BF₂I₂N₂O₁₀ [M]^- 1084.1864, found 1804.1858.

Pro-BDP-2: green solid (54 mg, 45 %), ¹H NMR (300 MHz, CDCl₃): δ 8.17 (d, J = 16.5 Hz, 2H, C=CH), 7.59 (d, J = 8.4 Hz, 4H, Ar-H), 7.52 (d, J = 16.5 Hz, 2H, C=CH), 7.38 (d, J = 8.7 Hz, 2H, Ar-H), 6.93–6.97 (m, 6H, Ar-H), 6.00–6.12 (m, 1H, OCH₂CH), 5.39–5.46 (m, 1H, C=CH), 5.33–5.36 (m, 3H, C=CH and Ar-CH₂), 4.56–4.58 (m, 2H, OCH₂), 4.19 (t, J = 4.8 Hz, 4H, OCH₂), 3.89 (t, J = 4.8 Hz, 4H, OCH₂), 3.75–3.78 (m, 4H, OCH₂), 3.65–3.71 (m, 8H, OCH₂), 3.54–3.57 (m, 4H, OCH₂), 3.38 (s, 6H, OCH₃), 2.10 (s, 6H, CH₃). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 165.2, 160.2, 159.3, 151.6, 143.6, 140.0, 132.8, 131.2, 130.2, 129.5, 129.4, 125.6, 125.3, 117.9, 116.5, 115.0, 82.5, 71.9, 70.9, 70.6, 70.5, 69.6, 68.8, 67.5, 59.1, 15.6. HRMS (ESI): m/z calcd for C₅₂H₅₉BF₂I₂N₂NaO₁₁ [M+Na]⁺ 1213.2170, found 1213.2170.

Pro-BDP-3: green solid (74 mg, 60 %), ¹H NMR (400 MHz, CDCl₃): δ 8.17 (d, J = 16.8 Hz, 2H, C=CH), 7.59 (d, J = 8.8 Hz, 4H, Ar-H), 7.52 (d, J = 16.8 Hz, 2H, C=CH), 6.94–7.01 (m, 6H, Ar-H), 6.89 (d, J = 8.0 Hz, 1H, Ar-H), 6.04–6.14 (m, 1H, OCH₂CH), 5.42 (dd, J = 17.2, 1.6 Hz, 1H, C=CH), 5.37 (s, 2H, Ar-CH₂), 5.32 (dd, J = 10.8, 1.6 Hz, 1H, C=CH), 4.65 (d, J = 9.6 Hz, 2H, OCH₂), 4.19 (t, J = 4.8 Hz, 4H, OCH₂), 3.90 (s, 3H, OCH₃), 3.89 (t, J = 4.8 Hz, 4H, OCH₂), 3.75–3.77 (m, 4H, OCH₂), 3.69–3.71 (m, 4H, OCH₂), 3.65–3.68 (m, 4H, OCH₂), 3.55–3.57 (m, 4H, OCH₂), 3.38 (s, 6H, OCH₃), 2.09 (s, 6H, CH₃). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 165.2, 160.3, 151.7, 149.6, 148.7, 143.6, 140.0, 132.9, 130.2, 129.5, 126.1, 125.2, 122.4, 118.3, 116.5, 115.0, 113.2, 112.9, 71.9, 70.9, 70.7, 70.6, 69.7, 67.6, 65.9, 59.1, 56.1, 15.3. HRMS (ESI): m/z calcd for C₅₃H₆₁BF₂I₂N₂NaO₁₂ [M+Na]⁺ 1243.2276, found 1243.2277.

Pro-BDP-4: green solid (65 mg, 52 %), ¹H NMR (400 MHz, CDCl₃): δ 8.19 (d, J = 16.8 Hz, 2H, C=CH), 7.97 (d, J = 2.4 Hz, 1H, Ar-H), 7.62 (dd, J = 8.8, 2.4 Hz, 1H, Ar-H), 7.59 (d, J = 8.8 Hz, 4H, Ar-H), 7.53 (d, J = 16.8 Hz, 2H, C=CH), 7.12 (d, J = 8.8 Hz, 1H, Ar-H), 6.97 (d, J = 8.8 Hz, 4H, Ar-H), 5.99–6.08 (m, 1H, OCH₂CH), 5.48–5.53 (m, 1H, C=CH), 5.38 (s, 2H, Ar-CH₂), 5.34–5.36 (m, 1H, C=CH), 4.72–4.74 (m, 2H, OCH₂), 4.19 (t, J = 4.8 Hz, 4H, OCH₂), 3.89 (t, J = 4.8 Hz, 4H, OCH₂), 3.75–3.77 (m, 4H, OCH₂), 3.69–3.71 (m, 4H, OCH₂), 3.65–3.68 (m, 4H, OCH₂), 3.55–3.57 (m, 4H, OCH₂), 3.39 (s, 6H, OCH₃), 2.10 (s, 6H, CH₃). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 165.2, 160.3, 152.4, 151.9, 143.3, 140.3, 139.8, 134.9, 131.3, 130.2, 129.5, 126.8, 125.8, 123.8, 123.7, 118.8, 116.5, 115.3, 115.0, 71.9, 70.9, 70.7, 70.6, 70.2, 69.7, 67.6, 59.1, 15.7. HRMS (ESI): m/z calcd for C₅₂H₅₈BF₂I₂N₃NaO₁₃ [M+Na]⁺ 1258.2021, found 1258.2021.

4.4. Synthesis of Boc-protected quinoline derivative 6

A mixture of compound 5 [50] (50 mg, 0.22 mmol), tert-butyl N-(3-bromopropyl)carbamate (57 mg, 0.24 mmol), and Cs₂CO₃ (85 mg, 0.26 mmol) in acetone (15 mL) was stirred under reflux for 6 h. After cooling to room temperature, the solvent was evaporated in vacuo, and the residue was dissolved in CH₂Cl₂ (20 mL). After being washed with saturated brine (20 mL), the organic layer was dried over anhydrous Na₂SO₄, filtered, and evaporated under reduced pressure. The crude product was purified by silica gel column chromatography using hexane/ethyl acetate (10:1, v/v) as the eluent to give 6 (76 mg, 90 %) as a white solid. ¹H NMR (400 MHz, CDCl₃): δ 8.24 (d, J = 7.6 Hz, 1H, Ar-H), 8.23 (d, J = 7.6 Hz, 1H, Ar-H), 7.76 (td, J = 7.6, 1.2 Hz, 1H, Ar-H), 7.60 (td, J = 7.6, 0.9 Hz, 1H, Ar-H), 7.56 (s, 1H, Ar-H), 6.08–6.18 (m, 1H, OCH₂CH), 5.45–5.50 (m, 1H, C=CH), 5.32–5.36 (m, 1H, C=CH), 4.99 (d, J = 6.0 Hz, 2H, OCH₂), 4.79 (br s, 1H, NH), 4.35 (t, J = 6.0 Hz, 2H, OCH₂), 3.43–3.47 (m, 2H, NCH₂), 2.15–2.21 (m, 2H, OCH₂CH₂), 1.44 (s, 9H, tBu). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 165.4, 162.4, 156.0, 149.1, 148.5, 131.9, 130.5, 130.4, 127.7, 122.2, 121.7, 119.4, 100.8, 79.5, 67.0, 37.9, 29.4, 28.4. HRMS (ESI): m/z calcd for C₂₁H₂₇N₂O₅ [M+H]⁺ 387.1914, found 387.1908.

4.5. Synthesis of biotin-modified quinoline derivative 7

Compound 6 (100 mg, 0.26 mmol) was dissolved in CH₂Cl₂ (10 mL) containing 20 % trifluoroacetic acid (TFA). The mixture was stirred at room temperature for 1 h. Subsequently, the solvent was removed under reduced pressure. The residue was dissolved in CH₂Cl₂ (15 mL) followed by addition of biotin N-hydroxysuccinimide ester (NHS) (106 mg, 0.31 mmol) and N,N-diisopropylethylamine (DIPEA) (43 mg, 0.33 mmol). The mixture was stirred at room temperature overnight. The resulting mixture was diluted with CH₂Cl₂ (50 mL) and then washed with saturated brine (50 mL). The organic layer was dried over anhydrous Na₂SO₄, filtered, and evaporated under reduced pressure. The crude product was purified by silica gel column chromatography using CH₂Cl₂/MeOH (50:1, v/v) as the eluent to give 7 (73 mg, 70 %) as a faint yellow solid. ¹H NMR (400 MHz, CDCl₃): δ 8.39 (d, J = 8.4 Hz, 1H, Ar-H), 8.32 (d, J = 8.4 Hz, 1H, Ar-H), 7.95 (t, J = 7.6 Hz, 1H, Ar-H), 7.76 (t, J = 7.6 Hz, 1H, Ar-H), 7.66 (s, 1H, Ar-H), 6.81 (br s, 2H, NH), 6.03–6.13 (m, 1H, OCH₂CH), 5.49 (dd, J = 16.8, 1.2 Hz, 1H, C=CH), 5.38 (dd, J = 10.0, 1.2 Hz, 1H, C=CH), 4.97 (d, J = 6.4 Hz, 2H, OCH₂), 4.48–4.52 (m, 2H, OCH₂), 4.45 (dd, J = 8.0, 4.8 Hz, 1H, NHCH), 4.23 (dd, J = 8.0, 4.8 Hz, 1H, NHCH), 3.65–3.70 (m, 2H, NCH₂), 3.45 (dd, J = 13.2, 6.8 Hz, 1H, SCH), 3.05–3.10 (m, 1H, SCH), 2.80 (dd, J = 13.2, 4.8 Hz, 1H, SCH), 2.49 (d, J = 12.8 Hz, 1H, NH), 2.14–2.27 (m, 4H, CH₂), 1.50–1.74 (m, 4H, CH₂), 1.33–1.39 (m, 2H, CH₂). ¹³C{¹H} NMR (126 MHz, CDCl₃): δ 173.6, 165.5, 163.8, 162.4, 149.1, 148.4, 131.8, 130.6, 130.2, 127.7, 122.1, 121.7, 119.5, 100.8, 67.0, 66.6, 61.7, 60.1, 55.5, 40.6, 36.5, 35.8, 28.9, 28.0, 27.9, 25.6. HRMS (ESI): m/z calcd for C₂₆H₃₃N₄O₅S [M+H]⁺ 513.2166, found 513.2168.

4.6. Synthesis of RuL2

[Ru(η⁵-Cp)(CH₃CN)₃]PF₆ (RuL1) (34 mg, 78 μmol) was dissolved in CH₂Cl₂ (3 mL) in a 50 mL three-necked flask. Compound 6 (30 mg, 78 μmol) dissolved in CH₂Cl₂ (3 mL) was added into the flask. The reaction was carried out under nitrogen atmosphere for 30 min. The solvent was then removed, and the resulting residue was washed with CH₂Cl₂ and then with diethyl ether containing CH₂Cl₂ to yield a yellow solid (8.5 mg, 16 %). ¹H NMR (500 MHz, CD₃CN): δ 8.46 (d, J = 8.5 Hz, 1H, Ar-H), 8.05 (t, J = 8.5 Hz, 1H, Ar-H), 7.86 (t, J = 8.5 Hz, 1H, Ar-H), 7.77 (d, J = 8.5 Hz, 1H, Ar-H), 7.55 (s, 1H, Ar-H), 6.14 (s, 5H, Cp), 4.58–4.61 (m, 2H, allyl), 4.44–4.47 (m, 2H, OCH₂), 4.39–4.41 (m, 1H, allyl), 4.31–4.32 (m, 1H, allyl), 4.10–4.12 (m, 1H, allyl), 3.30–3.34 (m, 2H, NCH₂), 2.09–2.12 (m, 2H, OCH₂CH₂), 1.36 (s, 9H, tBu). ¹³C{¹H} NMR (126 MHz, CD₃CN): δ 172.9, 167.7, 154.3, 149.0, 134.3, 129.6, 129.2, 125.0, 124.6, 105.2, 103.9, 96.9, 70.9, 69.6, 69.5, 65.8, 65.4, 37.8, 29.9, 28.5. HRMS (ESI) m/z calcd for C₂₃H₂₆N₂NaO₅Ru [M-allyl + Na]⁺ 535.0783, found 535.0778.

4.7. Synthesis of RuL3

This complex was prepared in situ by treating [Ru(η⁵-Cp)(CH₃CN)₃]PF₆ (RuL1) (34 mg, 78 μmol) with 7 (40 mg, 78 μmol) in CH₂Cl₂ (6 mL) for 30 min and used directly for the subsequent studies as in the cases of similar biotinylated ruthenium complexes [[51], [52], [53]]. Upon evaporation, a yellow solid was obtained (15.8 mg, 25 %). HRMS (ESI) m/z calcd for C₃₁H₃₇N₄O₅RuS [M]⁺ 679.1531, found 679.1523.

4.8. Study in fluorescence recovery in solution

Pro-BDP-n (n = 1–4) were dissolved in PBS at pH = 7.4 with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v), respectively, to give solutions at 5 μM. RuLn (n = 1–3) in DMSO were then added respectively to give a final concentration of 0.25 μM. The mixtures were stirred at 37 °C over a period of 12 h. The fluorescence spectra and intensity at 682 nm were record at each time point upon excitation at 610 nm.

4.9. GSH consumption study

GSH (20 μM) was mixed with BDP-COOH (5 μM), Pro-BDP-3 (5 μM), RuL3 (0.25 μM), or Pro-BDP-3 (5 μM) + RuL3 (0.25 μM) in PBS with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v). The mixtures were left under ambient conditions for 12 h. After that, a DTNB (Ellman's Reagent) kit was employed to measure the GSH content in the mixtures by monitoring the absorbance at 412 nm using UV–Vis spectroscopy.

4.10. HPLC analysis of activation of Pro-BDP-3 by RuL3

Pro-BDP-3 was dissolved in PBS with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v) to give a 5 μM solution. It was then treated with RuL3 (0.25 μM) at 37 °C for 3 h and 12 h. The mixtures were analyzed by HPLC, using Pro-BDP-3 and BDP-COOH (10 μM) for comparison. The HPLC experiments were performed on a Waters HPLC system equipped with a Waters 1525 binary pump, a Waters 2998 photodiode array detector, and a Waters 2475 Multi λ fluorescence detector. Reverse-phase HPLC separation was performed on an XBridge® Peptide BEH C18 column (130 Å, 3.5 μm, 4.6 mm × 250 mm) at a flow rate of 1 mL min⁻¹. The solvents used for HPLC analysis were of HPLC grade. The conditions were set as follows: 10 % A (0.1 % TFA in acetonitrile) + 90 % B (0.1 % TFA in deionized water) in the first 5 min, and then changed to 100 % A in 15 min, maintained under this condition for 20 min, changed to 10 % A + 90 % B in 5 min, and finally maintained under this condition for a further 5 min.

4.11. Study in singlet oxygen generation in solution

The ROS generation ability of Pro-BDP-3 (5 μM) with or without prior treatment with RuL2 or RuL3 (0.25 μM) at 37 °C for 12 h, as well as that of the reference compounds BDP-COOH, RuL2, and RuL3 was studied in PBS at pH = 7.4 with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v), using DPBF as the probe. A solution of the sample (5 μM for the BDPs or 0.25 μM for the Ru complexes) and DPBF (30 μM) was irradiated with red light from a 300 W halogen lamp after passing through a water tank for cooling and a color filter with a cut-on wavelength at 610 nm (Newport). The absorption maximum of DPBF at 416 nm was monitored along with the irradiation time.

4.12. Selectivity study in solution

Pro-BDP-3 was dissolved in PBS at pH = 7.4 with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v) to give a 5 μM solution. It was then treated with RuL2 or RuL3 (0.25 μM) at 37 °C for 12 h. The fluorescence intensity at 682 nm of these mixtures was recorded upon excitation at 610 nm. For comparison, a range of biological agents commonly found in the cellular environment, including Cu²⁺, Mg²⁺, Ca²⁺, Fe³⁺, GSH, Cys, BSA, H₂O₂, and NaOH (5 mM) were also added respectively to the solution of Pro-BDP-3 (5 μM). The mixtures were stirred at 37 °C for 1 h. After that, the fluorescence intensity at 682 nm was recorded upon excitation at 610 nm. BDP-COOH (5 μM) in the same medium was used as the positive control.

4.13. Stability of RuL3 in physiological conditions

A solution of RuL3 (1 μM) in DMEM with 10 % fetal calf serum was kept at 37 °C for 6, 12, or 24 h. After that, the pretreated RuL3 (0.25 μM) was mixed with Pro-BDP-3 (5 μM) in PBS with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v) at 37 °C for 12 h. The fluorescence intensity at 682 nm was then recorded upon excitation at 610 nm, using BDP-COOH (5 μM) as the control.

4.14. Stability of Pro-BDP-3 in physiological conditions

A solution of Pro-BDP-3 (5 μM) in DMEM with 10 % fetal calf serum was kept at 37 °C for 6, 12, or 24 h. For the mixture being left for 24 h, RuL2 or RuL3 (0.25 μM) was added, and the resulting mixtures were kept at 37 °C for a further 12 h. The fluorescence intensities at 682 nm of all these mixtures were recorded upon excitation at 610 nm. BDP-COOH (5 μM) was used as the control.

4.15. Determination of turnover number in solution

Pro-BDP-3 was dissolved in PBS with 10 % DMSO (v/v) and 0.1 % Tween 80 (v/v) to give a 10 μM solution. It was then treated with RuL2 or RuL3 (0.01 μM) at 37 °C over a period of 48 h. The fluorescence intensity at 682 nm of these mixtures was recorded at different time points upon excitation at 610 nm. By comparing with the fluorescence intensity of BDP-COOH (10 μM) under the same conditions, the release yields were determined. The TONs of RuL2 and RuL3 during this period of time were also calculated by the equation: [Pro-BDP-3] x release yield (%)/[RuL2/RuL3] [26].

4.16. Cell lines and culture conditions

The HeLa human cervical cancer cells (ATCC, CCL-2), 4T1 murine mammary carcinoma cells (ATCC, CRL-2539), MCF-7 human breast cancer cells (ATCC, HTB-22), and NIH 3T3 murine embryonic fibroblast cells were maintained in Dulbecco's modified Eagle's medium (DMEM, ThermoFisher, cat. no. 11965092) supplemented with fetal calf serum (10 %) and penicillin-streptomycin (100 unit/mL and 100 μg/mL, respectively). The cells were grown at 37 °C in a humidified 5 % CO₂ atmosphere.

4.17. Preparation of stock solutions

Appropriate amounts of RuL2, RuL3, BDP-COOH, and Pro-BDP-3 were dissolved in DMSO to form stock solutions with a concentration of 10 mM. These solutions were then diluted with PBS or the culture medium with 0.1 % Tween 80 (v/v) to the indicated concentrations for the in vitro studies. The concentrations of DMSO in all cases were below 0.1 % (v/v).

4.18. Optimization of intracellular activation conditions

Approximately 2 × 10⁵ HeLa cells in DMEM (2 mL) were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere for 12 h. After discarding the medium, the cells were incubated with Pro-BDP-3 in DMEM (5.0 μM, 2 mL) for 2 h. After rinsing the cells with PBS (1 mL) three times, the cells were incubated with RuL3 under different conditions. The cells were then harvested and washed with PBS for three times. The fluorescence intensity of BDP was determined by flow cytometry (CytoFLEX S Beckman) with 10⁴ cells counted in each sample. The dye was excited at 638 nm and its fluorescence was monitored at 700–725 nm, and the results were analyzed using the FlowJo. Subsequently, the zero interaction potency (ZIP) synergy scores (for concentration and incubation time) were calculated by SynergyFinder [38].

4.19. Determination of turnover number in cells

Approximately 2 × 10⁵ HeLa cells in DMEM were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (10 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells were treated with RuL2 or RuL3 (0.01 μM, 2 mL) in the medium for 2, 12, 24, and 48 h. Afterward, the culture medium was discarded, and the cells were rinsed with PBS for three times. They were then detached using trypsin and pelleted by centrifugation at 7000 rpm for 5 min. After collecting the cells, the intracellular fluorescence intensity was determined by flow cytometry (λ_ex = 638 nm, λ_em = 700–725 nm). The release yield of Pro-BDP-3 was determined by comparing the fluorescence intensity with that of BDP-COOH. The TONs of RuL2 and RuL3 were calculated by the equation: [Pro-BDP-3] x release yield (%)/[RuL2/RuL3] [26].

4.20. Determination of intracellular ruthenium contents

Approximately 2 × 10⁵ HeLa, 4T1, MCF-7, and NIH 3T3 cells in DMEM were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with RuL2 or RuL3 (10 μM, 2 mL) for 4 h. The culture medium was then discarded, and the cells were rinsed with PBS for three times. They were then detached using trypsin and pelleted by centrifugation at 7000 rpm for 5 min. The cells collected were sequentially digested with 100 μL of 70 % nitric acid for 2 h, 50 μL of 30 % hydrogen peroxide for 1.5 h, and 50 μL of concentrated hydrochloric acid for 1.5 h at 35 °C to obtain a homogeneous solution. The mixtures were then analyzed using ICP-MS (ICP-OES AVIO 220 MAX Spectrometer). A control experiment was performed with the untreated cells.

4.21. Intracellular activation of Pro-BDP-3

Approximately 2 × 10⁵ HeLa, 4T1, MCF-7, and NIH 3T3 cells in DMEM were seeded on 35 mm confocal dishes (MatTek Corporation, no. P35G-0-14-C) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for a further 4 h. The cells were then stained with Hoechst 33342 (ThermoFisher, cat. no. H1399, 1.0 μM, 2 mL) for 15 min. After being rinsed with PBS for three times, the cells were examined with a Zeiss laser scanning microscope (Zeiss LSM880). The BDP unit was excited at 633 nm and its fluorescence was monitored at 650–900 nm. The Hoechst 33342 was excited at 405 nm and its fluorescence was monitored at 420–500 nm. The images were digitized and analyzed using the Zen software. The mean fluorescence intensity was determined using ImageJ.

4.22. Intracellular ROS generation

Approximately 2 × 10⁵ HeLa, 4T1, MCF-7, and NIH 3T3 cells in DMEM were seeded on 35 mm confocal dishes and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for a further 4 h. The medium was removed, and then the cells were further incubated with H₂DCFDA (Sigma-Aldrich, cat. no. HY-D0940, 10 μM, 2 mL) for 30 min. The cells were irradiated with a 300 W halogen lamp after passing through a water tank for cooling and a color glass filter (Newport) cut-on at 610 nm. The fluence rate (λ > 610 nm) was 25.8 mW/cm². Illumination of 8 min led to a total fluence of 12 J/cm². Finally, the cells were examined with a Zeiss laser scanning microscope. The oxidized product DCF was excited at 488 nm and its fluorescence was monitored at 493–550 nm. The images were digitized and analyzed using the Zen software. The mean fluorescence intensity was determined using ImageJ.

4.23. Intracellular singlet oxygen generation

Approximately 2 × 10⁵ HeLa, 4T1, MCF-7, and NIH 3T3 cells in DMEM were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated with BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for an additional 4 h. The medium was removed, and then the cells were further incubated with SOSG (Thermo Fisher, cat. no. S36002, 10 μM, 2 mL) for 30 min. The cells were irradiated with a 300 W halogen lamp after passing through a water tank for cooling and a color glass filter (Newport) cut-on at 610 nm. The fluence rate (λ > 610 nm) was 25.8 mW/cm². Illumination of 8 min led to a total fluence of 12 J/cm². Finally, the cells were collected and examined with flow cytometry (λ_ex = 488 nm, λ_em = 493–550 nm).

4.24. Study of cytotoxicity

Approximately 1 × 10⁴ HeLa, 4T1, MCF-7, and NIH 3T3 cells per well in DMEM were inoculated in 96-well plates and incubated for 24 h. The culture medium was removed, and then the cells were incubated with different concentrations of BDP-COOH, Pro-BDP-3, RuL2, or RuL3 in DMEM (0, 1, 2, 3, 4, and 5 μM, 100 μL) for 2 h, followed by incubation with fresh medium for a further 4 h. Alternatively, the cells were incubated with different concentrations of Pro-BDP-3 in DMEM (0, 1, 2, 3, 4, and 5 μM, 100 μL) for 2 h. After being rinsed with PBS three times, the cells were incubated with RuL2 or RuL3 in DMEM (at half of the concentration of Pro-BDP-3, 100 μL) for a further 4 h. For the light treatment groups, the cells were rinsed with PBS for three times, followed by light irradiation with the aforementioned light source for 8 min. After incubation in culture medium for 6 h, an MTT solution in PBS (3 mg/mL, 50 μL) was added to each well, followed by incubation for a further 4 h. After removing the MTT solution, DMSO (150 μL) was added to each well. The plate was agitated on a BioTek microplate reader at ambient temperature for 10 s before measuring the absorbance at 490 nm for each well. The average absorbance of the blank wells, which did not contain the cells, was subtracted from the readings of the other wells. The cell viability was determined by the following equation: % viability = [Σ(A_i/A_control) × 100]/n, where A_i is the absorbance of the ith datum (i = 1, 2, …, n), A_control is the average absorbance of the control wells in which no sample was treated, and n (=4) is the number of data points.

4.25. Live/dead cell co-staining

The live/death cell co-staining was performed using a commercially available kit (Elabscience, cat. no. E-CK-A354) containing calcein-AM and PI. Approximately 2 × 10⁵ HeLa, 4T1, MCF-7, and NIH 3T3 cells in DMEM were seeded on 35 mm confocal dishes and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for a further 4 h. After being rinsed with PBS for three times, the cells were irradiated with the aforementioned light source for 8 min. After the PDT treatment, the cells were incubated for a further 6 h. The cells were then incubated with a solution of calcein-AM and PI in binding buffer (1 μM, 2 mL) for 30 min. Finally, the cells were examined with a Zeiss laser scanning microscope. Calcein-AM was excited at 488 nm and its fluorescence was monitored at 493–550 nm, while the excitation of PI was achieved at 561 nm with the emission at 600–800 nm. The images were digitized and analyzed using the Zen software.

4.26. Subcellular localization study

Approximately 2 × 10⁵ HeLa cells in DMEM were seeded on a 35 mm confocal dish and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated with BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) in the medium for 2 h. After being rinsed with PBS for three times, the cells were incubated in a fresh medium or RuL3 (2.5 μM, 2 mL) in the medium, respectively, for 4 h. After repeated rinsing with PBS, the cells were stained with LysoTracker Green DND-26 (ThermoFisher, cat. no. L7526, 1.0 μM, 2 mL), MitoTracker Green FM (ThermoFisher, cat. no. M7514, 1.0 μM, 2 mL) or ER-Tracker Green (ThermoFisher, cat. no. E34251, 1.0 μM, 2 mL) in Hank's Balanced Salt Solution (HBSS) for 30 min. The cells were then rinsed with PBS twice and refed with HBSS. The cells were then examined with a Zeiss laser scanning microscope. The BDP unit was excited at 633 nm and its fluorescence was monitored at 650–900 nm. The trackers were excited at 488 nm and their fluorescence was monitored at 493–550 nm. The images were digitized and analyzed using the Zen software. The Pearson's correlation coefficients were quantified by ImageJ.

4.27. ER stress detection

Approximately 2 × 10⁵ HeLa cells in DMEM were seeded on a 35 mm confocal dish and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated with BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) in the medium for 2 h. After being rinsed with PBS for three times, the cells were incubated in a fresh medium or RuL3 (2.5 μM, 2 mL) in the medium, respectively, for 4 h. After being rinsed with PBS for three times, the cells were irradiated with the aforementioned light source for 8 min. The cells were then incubated for 1, 3, or 6 h at 37 °C in a humidified 5 % CO₂ atmosphere. The cells were then rinsed with PBS for three times and then stained with ER-Tracker Green (1.0 μM, 2 mL) in HBSS for 30 min, followed by staining with Hoechst 33342 (ThermoFisher, cat. no. H1399, 1.0 μM, 2 mL) for a further 15 min. After being rinsed with PBS for three times, the cells were examined with a Zeiss laser scanning microscope (Zeiss LSM880). ER-Tracker Green was excited at 488 nm and its fluorescence was monitored at 493–550 nm. Hoechst 33342 was excited at 405 nm and its fluorescence was monitored at 420–500 nm. The images were digitized and analyzed using the Zen software. The mean fluorescence intensity was determined using ImageJ.

4.28. Intracellular Ca²⁺ detection

HeLa cells were treated as described above. After the photodynamic treatment, the cells were incubated for 6 h at 37 °C in a humidified 5 % CO₂ atmosphere. After being rinsed with PBS for three times, the cells were stained with Fluo-4 AM (Beyotime, cat. no. S1060, 1.0 μM) in PBS for 30 min, followed by staining with Hoechst 33342 (ThermoFisher, cat. no. H1399, 1 μM, 2 mL) for a further 15 min. After being rinsed with PBS for three times, the cells were examined with a Zeiss laser scanning microscope (Zeiss LSM880). Fluo-4 AM was excited at 488 nm and its fluorescence was monitored at 493–550 nm. Hoechst 33342 was excited at 405 nm and its fluorescence was monitored at 420–500 nm. The images were digitized and analyzed using the Zen software.

4.29. Investigation of mitochondria membrane potential

HeLa cells were treated as described above. After the photodynamic treatment, the cells were incubated for 3 or 6 h at 37 °C in a humidified 5 % CO₂ atmosphere. After being rinsed with PBS for three times, the cells were treated with JC-1 (10 μg/mL, ThermoFisher, no. T3168) in a serum-free medium and then incubated for a further 30 min in the dark. The medium was removed, and then the cells were rinsed with PBS for three times. The fluorescence of the monomeric form (green: λ_ex = 488 nm, λ_em = 495–550 nm) and the J-aggregate form (red: λ_ex = 561 nm, λ_em = 570–680 nm) of JC-1 was examined with a Zeiss laser scanning microscope (Zeiss LSM880). The images were digitized and analyzed using the Zen software.

4.30. Analysis of the cell death pathway

Approximately 2 × 10⁵ HeLa and NIH 3T3 cells in DMEM were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for a further 4 h. After being rinsed with PBS for three times, the cells were irradiated with the aforementioned light source for 8 min. The cells were then incubated for 6 h at 37 °C in a humidified 5 % CO₂ atmosphere. The cells were then harvested and rinsed with PBS for three times. After staining with Annexin V-FITC (1.0 μM) and PI (1.0 μM) in binding buffer (ThermoFisher, cat. no. V13242, 1 mL) for 20 min, the cells were examined by flow cytometry with 10⁴ cells counted in each sample under the FITC (λ_ex = 488 nm, λ_em = 505–545 nm) and PI (λ_ex = 561 nm, λ_em = 564–606 nm) channels for annexin V-FITC and PI, respectively. The results were digitized and analyzed using FlowJo.

4.31. Western blot analysis

Approximately 2 × 10⁵ HeLa cells in DMEM were seeded on 6-well plates (NEST 703001) and incubated overnight at 37 °C in a humidified 5 % CO₂ atmosphere. The medium was removed, and then the cells were incubated in the medium with or without BDP-COOH or Pro-BDP-3 (5.0 μM, 2 mL) for 2 h. After being rinsed with PBS for three times, the cells incubated with Pro-BDP-3 were further treated with RuL2 or RuL3 (2.5 μM, 2 mL) in the medium for 4 h. For the cells incubated with BDP-COOH or Pro-BDP-3 alone, they were incubated in fresh medium for a further 4 h. After being rinsed with PBS for three times, the cells were irradiated with the aforementioned light source for 8 min. The cells were then incubated for 6 h at 37 °C in a humidified 5 % CO₂ atmosphere. Then the cells were collected and subjected to standard western blot. The extracted proteins were separated using SDS-PAGE and then transferred to polyvinylidene difluoride membranes. To prevent interference from nonspecific binding, the membranes were blocked with 5 % nonfat milk solution for 1 h at room temperature and incubated overnight at 4 °C with the indicated primary antibodies against cleaved caspase-3 (CST, no. 9661), Bax (Abcam, no. 2774), and Bcl-2 (Abcam, no. 2876). After incubation with the goat-antirabbit IgG-HRP secondary antibody (Thermo Fisher scientific, Oregon, USA) at room temperature for 1 h, the immunoreactive bands were detected using a chemiluminescence system (Bio-Rad, CA, USA). For actin (CST, no. 14968), the procedure was consistent with cleaved caspase-3 evaluation.

4.32. In vivo fluorescence imaging

Female Balb/c nude mice (20–25 g) were obtained from the Laboratory Animal Services Centre of the City University of Hong Kong. All animal experiments had been approved by the Animal Experimentation Ethics Committee of the City University of Hong Kong (Ref. A-0453). The mice were kept under pathogen-free conditions with free access to food and water. HeLa cells in PBS (1 × 10⁷ cells in 100 μL) were inoculated subcutaneously on the back of the mice. When the size of the tumor reached 80–100 mm³ [Volume = (Length × Width²)/2], the mice were administered with RuL3 in PBS with 5 % DMSO (v/v) (100 μL, 10 nmol) intravenously, followed by the intratumoral injection of Pro-BDP-3 in PBS with 10 % DMSO (v/v) (30 μL, 20 nmol) after 2 h. The mice without the injection of RuL3 were treated as the control. The fluorescence images of the mice were captured before and after the injection at different time points for 48 h with an IVIS animal imaging system (excitation wavelength at 680 nm, emission wavelength at 700–800 nm). The mice were sacrificed after 48 h and the tumors and major organs were harvested. The ex vivo biodistribution of the activated Pro-BDP-3 was examined based on the fluorescence intensity recorded with the IVIS animal imaging system. The images were digitized and analyzed using the IVIS imaging system software (Living Imaging 3.2). Four mice were used for each experiment.

4.33. In vivo PDT treatment

4T1 tumor-bearing nude mice as prepared above were randomly divided into six groups (n = 5): (1) intravenous injection with PBS with laser irradiation, (2) intravenous injection with RuL3 followed by intratumoral injection with Pro-BDP-3 without laser irradiation, (3) intratumoral injection with Pro-BDP-3 with laser irradiation, (4) intravenous injection with RuL3 without laser irradiation, (5) intravenous injection with RuL3 followed by intratumoral injection with Pro-BDP-3 with laser irradiation, and (6) BDP-COOH with light irradiation. For treatment group 5, the mice were administered with RuL3 in PBS with 5 % DMSO (v/v) (100 μL, 10 nmol) intravenously, followed by the intratumoral injection of Pro-BDP-3 in PBS with 10 % DMSO (v/v) (30 μL, 20 nmol) after 2 h. At 12 h post-injection, the tumor was irradiated with a diode laser (Biolitec Ceralas) at 675 nm operated at 0.1 W. Illumination on a spot size of 1.0 cm² for 10 min led to a total fluence of 60 J/cm². The tumor size of the nude mice was monitored periodically for the next 12 days. After 12 days, the mice were sacrificed, and the internal organs and the tumor were harvested. They were then fixed with 4 % paraformaldehyde and stained with H&E for histological analysis.

4.34. Statistical analysis

Data shown on figures are presented as the mean ± SEM or SD. The data were analyzed using the Student's t-test with p values < 0.05 considered as significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Statistical calculations were performed using a Microsoft Excel spreadsheet (Microsoft Corporation, Redmond, WA, USA).

CRediT authorship contribution statement

Feijie Xu: Writing – original draft, Validation, Methodology, Investigation, Formal analysis. Qianqian Wu: Investigation, Methodology. Lin He: Methodology, Investigation. Lin Yang: Methodology, Investigation. Pui-Chi Lo: Writing – review & editing, Writing – original draft, Supervision, Project administration, Funding acquisition, Formal analysis.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is/are the supplementary data to this article:

Data availability

Data will be made available on request.