1 Introduction

Although chemotherapy is still the most widely used approach in the treatment of tumors, it often suffers from severe side effects, non-targeting capability, and potential drug resistance, which greatly limits its further applications in clinical trials [1,2,3]. In order to solve this problem, combinational therapy is developed via the reasonable integration of other therapeutic methods to increase the efficacy and reduce the side effect of cancer treatment [4,5,6,7]. In this regard, photodynamic therapy (PDT) has been considered to be a promising complementary strategy to chemotherapy because of its negligible drug resistance, minimal invasion, fewer side effects, and less damage to normal tissues [8,9,10,11,12,13]. However, conventional photosensitizers used in PDT are generally large π-conjugated planer small molecules such as porphyrin and hypocrellin derivatives [14,15,16,17]. Such molecules are easy to aggregate in aqueous solution due to π–π stacking and hydrophobic interactions, leading to the quench of the excited state and poor singlet oxygen (1O2) generation quantum yield, which significantly impedes their broader applications. The encapsulation of photosensitizers in supramolecular hosts not only prevents their aggregation and promotes the generation of 1O2, but also offers possibilities for introducing extra functionalities such as targeting ligands, imaging agents, or therapeutic drugs for supramolecular photochemotherapy [18,19,20]. Moreover, the stimuli-responsiveness of supramolecular systems may also provide controllable release of the agents. Therefore, the development of efficient supramolecular hosts for photosensitizers and the exploration of their applications in cancer photochemotherapy are demanding.

Metal–organic cages or metallacages are a type of three-dimensional supramolecular assemblies formed by metal-coordination bonds [21,22,23,24,25,26,27,28,29]. They generally possess well-defined cavities capable of encapsulating guest molecules, enabling their wide applications in chemical sensing and recognition [30,31,32,33,34], adsorption and separation [35,36,37,38,39,40,41], catalysis [42,43,44,45,46], etc. Recently, their application as supramolecular containers for the delivery of photosensitizers for PDT has received much attention due to their convenient preparation, tunable structures, and ease of functionalization [47,48,49,50]. For example, Therrien et al. reported the preparation of prismatic metallacages to deliver and spatial-controlled release porphin for PDT [51]. Mukherjee et al. [52] described a water-soluble barrel-shaped metallacage to encapsulate zinc-phthalocyanine for enchanced PDT. Besides preventing the aggregation of their encapsulated photosensitizers, they can also serve as a type of therapeutic drugs based on the inherent cytotoxicity of metal nodes, increasing the efficacy of cancer treatment via the combination of photo and chemotherapy [53, 54]. However, in most cases, the metallacages only served as containers (Scheme 1); their use as energy donors to promote the 1O2 generation ability of the encapsulated photosensitizers and thus increase the performance of PDT has been rarely addressed. Such study will not only benefit for the rational design and preparation of functional metallacages toward real applications, but also effectively integrate the advantages of both components for improved synergistic therapy, which will promote the development of metallacage-based supramolecular biomaterials.

Scheme 1
scheme 1

Cartoon illustrations of the role of metallacages in photodynamic therapy

Full size image

Herein, we report a type of perylene diimide (PDI)-based emissive metallacages which serves as both hosts and energy donors to promote the 1O2 generation ability of their encapsulated photosensitizers via fluorescence resonance energy transfer (FRET) [55,56,57] and then use the supramolecular system for improved cancer photochemotherapy. The metallacages are prepared via the multicomponent coordination-driven self-assembly of a tetracarboxylic ligand (1), cis-(PEt3)2Pt(OTf)2 (2), and tetrapyridyl perylene diimide (3a or 3b). The distance between the two PDI planes can be finely adjusted via the regulation of PDI ligands, offering Cage 4b as a more efficient host for photosensitizer hypocrellin A (G5) (Ka = (1.51 ± 0.04) × 105 M−1 in CH3CN/H2O). Owing to the good spectral overlap between the emission of the metallacage (Cage 4b) and the absorption of the photosensitizer (G5), efficient FRET takes place from Cage 4b to G5, thus resulting in enhanced 1O2 generation quantum yields. It is worth mentioning that the 1O2 quantum yield (ΦΔ) of complex Cage 4b ⊃ G5 reaches 0.66, which is 1.67 times that of the photosensitizer alone. Complex Cage 4b ⊃ G5 is further assembled with amphiphilic mPEG-DSPE to form nanoparticles (NPs 5) for cancer photodynamic therapy, showing extraordinarily high inhibition rate toward MHCC-97L cells (liver cancer) and NCI-H460 cells (lung cancer). NPs 5 displays the most substantial antitumor effectiveness, demonstrating the highest inhibition of tumor growth in 14 days after the treatment, affirmed by the smallest tumor volume among all the experimental groups. In contrast, the control group treated by 3b, G5, Cage 4b, or Cage 4b ⊃ G5, exhibits significantly lower anticancer activity compared to NPs 5, which is in good agreement with the tumor mass results. This outcome reveals that NPs 5 exhibits the best photodynamic activity for MHCC-97L tumor-bearing nude mice compared with the sole Cage 4b, photosensitizer G5 and other controlled compounds 2 and 3b, consistent with the high 1O2 generation efficiency of complex Cage 4b ⊃ G5. Furthermore, in vivo anticancer studies reveal that NPs 5 exhibits an augmented inhibitory effect on tumor proliferation. All these in vivo anticancer results indicate that the photosensitivity of the photosensitizers can be significantly enhanced through FRET process, leading to desirable performance for improved PDT. This study represents an effective strategy to employ host–guest interactions to enhance the 1O2 generation efficiency via FRET for cancer photochemotherapy, which will promote the biological applications of metallacages.

2 Experimental and Methods

2.1 Self-Assembly of Metallacages

Compounds 1, 2, and 3a (or 3b) in a 1:4:1 molar ratio were dissolved in acetonitrile (5 mL), and then, the reaction mixture was heated at 50 °C for 8 h. After being cooled to room temperature, the solvent was removed by nitrogen flow. The residue was redissolved in acetonitrile (1 mL) and filtered to give a clear solution, which was precipitated by the addition of diethyl ether to give Cage 4a or Cage 4b as a red powder.

2.2 Growth of Single Crystals

Solutions of Cage 4a or Cage 4b in DMF (350 μL) with varying concentrations (4.0, 6.0, 8.0, 10.0, and 12.0 mg mL−1) were introduced into a 2-mL vial. The vial was then positioned within a 20-mL vial filled with toluene (5 mL). The entire system was left at room temperature for 1 month to get the crystals.

2.3 Preparation of Nanoparticles

Cage 4b ⊃ G5 (3.25 mg) and mPEG-DSPE2000 (8 mg) were dissolved in acetone (1 mL) and pure water (5 mL), respectively, and then stirred for 10 min. The solution of Cage 4b ⊃ G5 was added into the solution of mPEG-DSPE2000 under sonication dropwisely. The mixture was then stirred at room temperature in dark for 36 h to remove acetone. The resulting solution was passed through a 0.2-μm syringe filter to give NPs 5, which was stored at room temperature for further use.

2.4 Singlet Oxygen Quantum Yields Measurements

Singlet oxygen generation experiments were conducted in acetonitrile using a LED lamp with a wavelength of 520 nm as the light source. Prior to the experiment, the absorption of photosensitizers (including 3a, Cage 4a, 3b, Cage 4b, G5, Cage 4a ⊃ G5, and Cage 4b ⊃ G5) at the excitation wavelength of 520 nm was maintained at approximately 0.25. Quantum yields for singlet oxygen generation were determined by monitoring the reduction in the absorption of 1,3-diphenylisobenzofuran (DPBF) at 410 nm due to its photooxidation sensitized by the photosensitizers. Time-dependent absorption spectra of DPBF were recorded every 1.0 min for the homogeneous phase. The quantum yields of singlet oxygen generation (1O2) were calculated using a relative method with optically matched solutions. The calculation involved comparing the decrease in the absorbance of DPBF sensitized by photosensitizers to that of rose bengal (RB) (ΦΔ = 0.54), which served as a reference, based on the equation:

$$\Phi_{\Delta }^{S} = \Phi_{\Delta }^{RB} \frac{{m^{S} F^{RB} }}{{m^{RB} F^{S} }}$$

where superscripts “S” and “RB” denote the photosensitizers and RB, respectively, “ΦΔ” is the quantum yield of singlet oxygen, “m” is the slope of a plot with a difference in the change in the absorbance of DPBF (at 410 nm) with the irradiation time, and “F” is the absorption correction factor, which is given by F = 1–10−OD (OD corresponds to the absorbance of the photosensitizer).

2.5 Cell Culture

The NCI-H460 human lung cancer cell line was acquired from the Shanghai Institute of Cell Biology at the Chinese Academy of Sciences in Shanghai, China. The MHCC-97L (97L) HCC cell line was generously provided by the First Affiliated Hospital of Xi'an Jiaotong University. MHCC-97L cells and NCI-H460 cells were cultured in DMEM and RPMI 1640, respectively, supplemented with 10% fetal bovine serum (FBS) from ExCell Bio, China, and 100X penicillin/streptomycin (100 U mL−1 and 100 μg mL−1) from Solarbio, China, at 37 °C with 5% CO2.

2.6 Confocal Laser Scanning Microscopy Imaging

MHCC-97L and NCI-H460 cells were seeded at a density of 1 × 104 cells on a circular coverslip (diameter 12 mm) in complete DMEM culture medium. Following a 24-h incubation period, the cells were exposed to various compounds at a concentration of 5 μM for 15 min. Subsequently, the cells were rinsed twice with cold PBS. F-actin was stained using 4′,6-diamidino-2-phenylindole (DAPI) and visualized using a Zeiss LSM 710 confocal microscope. The data were analyzed using ImageJ. It was measured at 490 nm using a Bio-Rad 680 microplate reader.

2.7 Intracellular Reactive Oxygen Species Production Study

Cells were seeded in a glass culture dish for 24 h, then continued to be cultured in fresh medium containing compounds (2 μM) for 8 h. Subsequently, the cells were exposed to 1 mL of fresh serum-free medium containing 10 μM DCFH-DA at 37 °C for 20 min. After thorough washing, the cells were subjected to white light treatment (50 mW cm−2) for 1 min, and flow cytometry analysis was conducted to examine intracellular reactive oxygen species production.

2.8 MTT Assay

Cells were seeded in 96-well plates at a density of 5000 cells per well in 180 μL of complete medium and incubated in a 5% CO2 atmosphere at 37 °C for 24 h. Subsequently, the culture medium was replaced with 20 μL of freshly prepared culture medium containing compounds at various concentrations. The cells were further incubated for 48 h, after which the medium was substituted with fresh culture medium, and MTT solution (5 mg mL−1) was added. The cells were incubated for an additional 4 h to enable viable cells to reduce the yellow tetrazolium salt (MTT) into dark blue formazan crystals. Finally, 100 μL of lysis buffer was added to the wells and incubated for another 4 h at 37 °C. The absorbance was measured at 490 nm using a Bio-Rad 680 microplate reader.

2.9 Cellular Staining

Cells were initially seeded in glass dishes for 24 h and subsequently subjected to four different treatments: (1) PBS solution; (2) PBS solution incubated with white light (50 mW cm−2) for 1 min; (3) media containing NPs 5 (2 μM) for 8 h without exposure to light; and (4) media containing NPs 5 (2 μM) for 8 h with illumination using white light (50 mW cm−2) for 1 min. Following the diverse treatments, cells were further incubated at 37 °C for 2 h and then stained with propidium iodide (PI) (60 μg mL−1) and fluorescein diacetate (FDA) (100 μg mL−1) in PBS for 10 min. Subsequently, the cells were meticulously washed and imaged by confocal laser scanning microscopy (CLSM), with excitation wavelengths of 488 nm for FDA and 543 nm for PI. The emission filters used were 500–550 nm for FDA and 550–650 nm for PI.

2.10 In Vivo Tumor Therapy

All experimental protocols were approved by the Ethics Committee of the Xi'an Jiaotong University Health Science Center, Xi’an, China. The mice were randomly divided into seven groups (n = 5/group). To set up the tumor xenograft model, a total of 5 × 106 MHCC-97L cells were subcutaneously injected into the right armpit of BALB/c female nude mice (4 weeks old) and permitted the tumor to reach a size over 100 mm3 in volume. Thirty-five tumor-bearing female nude mice were divided into seven groups (PBS group, cis-(PEt3)2Pt(OTf)2 group, 3b group, G5 group, Cage 4b group, Cage 4b ⊃ G5 group, and NPs 5 group). These compounds were injected into the tumor at a final cis-(PEt3)2Pt(OTf)2 concentration of 2 mg kg−1, then treated with laser irradiation (520 nm, 50 mW cm−2 for 30 s). After 6-h post injection, the mice were exposed to Caliper IVIS Lumina II System to investigate the in vivo tracking of metallacages. Tumor size (V = W2 × L/2 mm3) was measured, and the body weight was recorded every 2 days for 14 days. At day 14, the tumors were collected and fixed in 4% paraformaldehyde overnight, embedded in paraffin, and sectioned at a thickness of 5 μm. The sections were stained with hematoxylin and eosin (H&E) or immunohistochemistry (IHC) of Ki 67.

2.11 Statistical Analysis

All results are representative of data generated in each independent experiment. All numerical values were expressed as the mean ± SEM. For multiple comparisons, statistical analysis was performed using one-way ANOVA followed by a Bonferroni posttest. Data analysis was performed using SPSS 18.0 software and considered statistically significant at p < 0.05.

3 Results and Discussion

3.1 Preparation and Characterization of Metallacages

Metallacage (Cage 4a or Cage 4b) was prepared by the self-assembly of tetracarboxylic ligand (1), cis-(PEt3)2Pt(OTf)2 (2) and PDI ligand (3a or 3b) in a 1:4:1 molar ratio (Fig. 1a, see Supporting Information of synthetic details). Their structures were confirmed by the combination of 31P{1H}, 1H NMR and electrospray ionization time-of-flight mass spectrometry (ESI-TOF–MS). The 31P{1H} NMR spectra exhibited two doublet peaks at 5.19 and − 0.47 ppm for Cage 4a, and 5.36 and − 0.12 ppm for Cage 4b (Fig. 1b, c). These doublet peaks, along with concurrent 195Pt satellites, indicated that each platinum atom coordinated with one pyridyl nitrogen of the PDI face and one carboxylic oxygen of the tetracarboxylic pillar. In the 1H NMR spectra, the α-pyridyl protons Ha and the β-pyridyl protons Hb of Cage 4a split into two sets of signals with noticeable chemical shifts (Fig. 1d, e), corresponding to protons inside and outside the cavity. For Cage 4b, only one set of protons (Hf, Hg, Hh, Hi and Hj) was observed, and all the protons exhibited obvious downfield chemical shifts (Fig. 1f, g), suggesting the formation of coordination bonds. The coordination stoichiometries of Cage 4a and Cage 4b were determined by ESI-TOF–MS (Fig. 1h, i). Isotopically well-resolved peaks with charge states ranging from 4 + to 8 + were observed for the two metallacages due to the loss of counterions (OTf−), confirming the composition of the metallacages. For example, peaks at m/z = 742.9245, 1040.5464, 1278.2423, 1635.2900, 818.8931, 957.4337, 1141.8347, and 1400.1527 were found, corresponding to [Cage 4a − 8OTf]8+, [Cage 4a − 6OTf]6+, [Cage 4a − 5OTf]5+, [Cage 4a − 4OTf]4+, [Cage 4b − 8OTf]8+, [Cage 4b − 7OTf]7+, [Cage 4b − 6OTf]6+, and [Cage 4b − 5OTf]5+, respectively. These data were entirely consistent with their calculated values. These findings align with previous reports [58,59,60,61], affirming the successful construction of the relevant discrete multicomponent metallacages.

Fig. 1
figure 1

a Self-assembly of Cage 4a and Cage 4b. b, c Partial 31P{1H} NMR (150 MHz, CD3CN, 295 K) spectra of Cage 4a and Cage 4b. d–g Partial 1H NMR spectra (600 MHz, CD3CN, 295 K) of 3a, Cage 4a, 3b and Cage 4b. h, i ESI-TOF–MS spectra of Cage 4a and Cage 4b. j Crystal structures of Cage 4a. k Simulated structures of Cage 4b. Hydrogen atoms, triethylphosphine units, counterions, and solvent molecules were omitted for clarity

Full size image

To further elucidate the coordination structures of the metallacages, single crystals of Cage 4a suitable for X-ray diffraction analysis were successfully obtained through vapor diffusion of dioxane into DMF over 1 month. Cage 4a is composed of two tetrapyridyl PDI units and two tetracarboxyl panels connected by eight platinum(II) atoms, forming a barrel-shaped structure. The dimension of Cage 4a is 1.60 × 1.08 × 1.02 nm3, based on the distance between platinum atoms. The distance between the two parallel PDI faces is 1.25 nm, which is determined by the length of the tetracarboxylic ligands. As all the attempts to get the single crystals of Cage 4b failed, molecular simulation was conducted to unveil its coordination structure. The dimension of Cage 4b is 2.48 × 1.35 × 0.86 nm3. Because the distance between the neighboring pyridyl groups in 3b is much longer than that in 3a, tetracarboxylic ligand 1 is connected to the two neighboring pyridyl groups using its long side in Cage 4b, which has been also observed in our previous study [62]. The distance between the two PDI faces in Cage 4b is only 0.98 nm, much shorter than that in Cage 4a, which is better for the π–π stacking interactions between the guests and the metallacage hosts. Therefore, it is anticipated that both metallacages are capable of complexing planar guest molecules, while metallacage Cage 4b should exhibit stronger binding abilities with guests than Cage 4a.

3.2 Host–Guest Complexation Study

Considering that Cage 4a and Cage 4b possess electron-deficient PDI faces and available cavities, their complexation with polycyclic aromatic hydrocarbons (PAHs) including pyrene (G1), triphenylene (G2), perylene (G3), and coronene (G4) was first studied. The 1H NMR spectra (Figs. 2 and S10) revealed that all complexes underwent rapid exchange on the NMR time scale, which was because the open windows of the metallacages would allow guest molecules to readily enter and exit. The protons of all the guests experienced significant upfield chemical shifts, indicative of the effective host–guest complexation by Cage 4a and Cage 4b. Job’s plots (Figs. S11 − S18), based on UV/vis spectroscopic absorbance data, affirmed that the complexes of Cage 4a and Cage 4b with G1, G2, G3, and G4 in solution maintained a 1:1 stoichiometry. Fluorescence titration experiments (Figs. S19 − S26) were conducted to determine the association constants (Ka) of these complexes in solution. It was observed that Ka of Cage 4b was almost an order of magnitude larger than those of Cage 4a for all the four guests in acetonitrile (Table 1). This enhanced affinity can be attributed to the shorter distance between the two PDI faces of Cage 4b, which facilitates stronger π–π stacking interactions between the electron-deficient PDI and the electron-rich PAHs.

Fig. 2
figure 2

Cartoon illustrations of the complexation. Partial 1H NMR spectra (600 MHz, CD3CN, 295 K) of a Cage 4a ⊃ G3, b G3, c Cage 4b ⊃ G3, d Cage 4a ⊃ G5, e G5, f Cage 4b ⊃ G5. [Host] = [Guest] = 2 mM

Full size image
Table 1 Association constants between Cage 4a, Cage 4b, and different guests
Full size table

The above results encouraged us to explore the potential of the metallacages for encapsulating planar photosensitizer Hypocrellin A (G5). The complexation between Cage 4a/Cage 4b and G5 was initially investigated using 1H NMR spectroscopy (Fig. 2). Significant upfield shifts were observed for protons H13 of G5 owing to the shielding effect of the metallacages, indicating good host−guest complexation. Job’s plots (Figs. S27 and S28) and ESI mass spectra (Fig. S29) both supported the 1:1 binding stoichiometry of Cage 4a ⊃ G5 and Cage 4b ⊃ G5. For example, three peaks at m/z = 1232.9257, 1509.0952, and 1923.8585 were observed, which was attributed to [Cage 4b ⊃ G5 − 6OTf]6+, [Cage 4b ⊃ G5 − 5OTf]5+, and [Cage 4b ⊃ G5 − 4OTf]4+, respectively. The Ka of Cage 4a ⊃ G5 and Cage 4b ⊃ G5 in acetonitrile (Fig. S30 and S31) was determined to be (8.91 ± 0.21) × 103 and (1.83 ± 0.03) × 104 M−1, respectively. Notably, such values increased into (5.40 ± 0.21) × 104 and (1.51 ± 0.04) × 105 M−1, respectively, in water/acetonitrile (1:1), due to the solvophobic effect of G5 in aqueous environments [63, 64], which is highly advantageous for biological applications.

Since the absorption of G5 exhibited a substantial overlap with the emission of Cage 4a and Cage 4b (Figs. 3b and S32a), FRET may occur from the metallacages to G5 [65, 66]. As the gradual addition of G5 to the solution of Cage 4a or Cage 4b, the emission of the metallacages decreased while that of the G5 increased (Figs. 3c and S32b), providing compelling evidence for the efficient energy transfer between G5 and Cage 4b. The formation of host–guest complexes shortened the distance between the metallacage donors and G5, offering good energy transfer efficiency (ΦET). ΦET of Cage 4a ⊃ G5 and Cage 4b ⊃ G5 were calculated to be 36.99% and 53.37%, respectively. The higher ΦET of Cage 4b ⊃ G5 compared with that of Cage 4a ⊃ G5 is attributed to the higher binding affinity of Cage 4b ⊃ G5, because higher binding affinity would give higher complexation ratio for the complexes and thus increase the energy transfer efficiency. Such FRET process is expected to boost the light-harvesting performance and enhance the generation of 1O2.

Fig. 3
figure 3

a Cartoon illustration of the cancer photochemotherapy of NPs 5 self-assembled from Cage 4b ⊃ G5 and mPEG-DSPE2000. b Normalized absorption spectrum of G5 and emission spectrum of Cage 4b. c Fluorescence spectra of Cage 4b and G5 in a mixture of CH3CN and water (v/v = 1/1) with different concentrations of G5 (λex = 365 nm, c = 10 µM for Cage 4b). d Plots of the absorption decays of DPBF at 410 nm versus the irradiation time in the presence of 3a, Cage 4a, 3b, Cage 4b, G5, Cage 4a ⊃ G5, or Cage 4b ⊃ G5 (λex = 520 nm). e Photographs of NPs 5, Tyndall effect and NPs 5 in CH3CN and water (v/v = 1/1) after illumination. f SEM image of NPs 5. g Particle size distributions of NPs 5 measured by DLS (c = 1.0 mM)

Full size image

To validate the hypothesis, electron spin resonance (ESR) spectroscopy was utilized to investigate the generation of 1O2 through sensitization of Cage 4a ⊃ G5 and Cage 4b ⊃ G5 using 2,2,6,6-tetramethylpiperidine (TEMP) as a 1O2 sensor, which was capable of generating stable tetramethylpiperidine oxide radical (TEMPO) upon trapping 1O2. All complexes exhibited a 1:1:1 triplet signal upon light irradiation (Fig. S33), which was consistent with the 1O2 signal generated by the complexes. Compared to Cage 4b ⊃ G5, the signal from Cage 4a ⊃ G5 appeared relatively weak. In addition, the 1O2 generation quantum yields of 3a, Cage 4a, 3b, Cage 4b, G5, Cage 4a ⊃ G5, and Cage 4b ⊃ G5 were measured using a reactive 1O2 scavenger, 1,3-diphenylisobenzofuran (DPBF) [67]. Time-dependent UV/vis absorption spectra of a mixture solution of DPBF and different species upon irradiation at 520 nm were recorded (Fig. S34). A gradual decrease in the characteristic absorption band centered at 410 nm for DPBF was observed with increasing exposure time, indicating the accumulation of 1O2. The absorption bands corresponding to Cage 4a, Cage 4b, Cage 4a ⊃ G5, and Cage 4b ⊃ G5 remained almost unchanged, suggesting their good photostability. The 1O2 quantum yields (ΦΔ) of all the compounds were calculated using Rose Bengal (RB) with a known efficiency (ΦΔRB = 0.54) as the reference. The ΦΔ values were determined to be 12%, 9%, 14%, 11%, 45%, 53%, and 66% for 3a, Cage 4a, 3b, Cage 4b, G5, Cage 4a ⊃ G5, and Cage 4b ⊃ G5, respectively (Fig. 3d). Based on these results, both Cage 4a/Cage 4b and G5 exhibited the capacity for 1O2 generation. Additionally, the FRET between metallacages and G5 further enhanced the efficiency of 1O2 generation for complexes Cage 4a ⊃ G5 and Cage 4b ⊃ G5. Notably, Cage 4b ⊃ G5 demonstrated a higher ΦΔ compared with Cage 4a ⊃ G5, attributed to its better host−guest complexation.

Based on the outstanding 1O2 generation capability of Cage 4b ⊃ G5, its application for cancer photodynamic therapy was further explored. 1,2-Distearoyl-phosphatidylethanolamine (DSPE)/polyethylene glycol (PEG) conjugate (mPEG-DSPE2000) [68], a commonly used polymer, was assembled with complex Cage 4b ⊃ G5 to form nanoparticles NPs 5, which were further used for further biological experiments (Fig. 3a). The photo image (Fig. 3e) of NPs 5 presented obvious “Tyndall effect” in solution, which indicated the successful preparation of dispersed colloidal nanoparticles. The size and morphology of NPs 5 were examined using scanning electron microscopy (SEM) and dynamic light scattering (DLS), respectively. The SEM image (Fig. 3f) revealed that NPs 5 exhibited micellar structures with diameters of 140–150 nm. DLS analysis (Fig. 3g) indicated an average hydrodynamic diameter of 142 nm, which agreed well with the SEM results. It is believed that nanoparticles within this size range can exhibit an enhanced permeability and retention effect [69], potentially augmenting their uptake and retention within tumors and increasing their anticancer activity. The size distribution of nanoparticles appeared broad, suggesting that these amphiphilic structures may readily aggregate in aqueous solution. No notable alterations in size (Fig. 3g) or absorption (Fig. S35) were observed for these nanoparticles even after 7 days, demonstrating the exceptional stability of complex-loaded NPs 5, attributed to the protective role of the amphiphilic mPEG-DSPE, which securely houses the complex within the hydrophobic interior.

3.3 Cell Imaging and Anticancer Study

The UV/vis absorption and emission spectra of 3a, Cage 4a, 3b, Cage 4b, G5, Cage 4a ⊃ G5, and Cage 4b ⊃ G5 were recorded to assess their optical characteristics (Fig. S36). Ligand 3a exhibited three absorption bands centered at 466, 496, and 534 nm, which was consistent with the typical absorption of PDI derivatives [70, 71]. The corresponding molar absorption coefficients (ε) were determined to be 1.50 × 104, 3.21 × 104, and 4.08 × 104 M−1 cm−1, respectively. Likewise, ligand 3b, Cage 4a, and Cage 4b, complexes Cage 4a ⊃ G5 and Cage 4b ⊃ G5 all displayed similar absorption bands in the visible region. Furthermore, two distinct emission peaks were observed for 3a at wavelengths of 515 and 551 nm. In contrast, 3b exhibited much weaker emission with peaks centered at 572 and 621 nm, which is probably due to the increased molecular motions in 3b compared with 3a. Cage 4a and Cage 4b exhibited two emission bands centered at 516 and 560 nm, and 556 and 597 nm, respectively. The fluorescence quantum yields (ΦF, Figs. S37–S41) of 3a, 3b, Cage 4a, Cage 4b, and G5 were determined to be 93.91%, 0.96%, 34.06%, 4.93%, and 3.68%, respectively. As for Cage 4a ⊃ G5 and Cage 4b ⊃ G5, the complexation leads to bathochromic shift and increased emission via FRET, with the ΦF values of 42.09% and 27.19%, respectively (Figs. S42 and S43). Therefore, Cage 4b, Cage 4b ⊃ G5, and 5 were further used as contrast agents for bioimaging. Liver cancer MHCC-97L cells and lung cancer NCI-H460 cells were stained with DAPI and treated with above-mentioned compounds simultaneously. After 6 h, the intensity of the fluorescence reached the maximum value which did not increase as time went by, so the images were taken at 6-h post incubation (Figs. 4a and S44) using confocal laser scanning microscopy (CLSM). Based on the merged figures, bright red fluorescence originating from these compounds was observed within the cells. The cellular uptake of complex Cage 4b ⊃ G5 and NPs 5 was assessed through flow cytometry (FCM) analysis. In both MHCC-97L and NCI-H460 cells (Figs. 4b and S45 − S47), stronger emission was observed for NPs 5, suggesting that the cellular uptake of NPs 5 was better than complex Cage 4b ⊃ G5 after incubation. Collectively, these findings confirm the suitability of Cage 4b, Cage 4b ⊃ G5, and NPs 5 as contrast agents for cell imaging.

Fig. 4
figure 4

a CLSM images of MHCC-97L cells after the incubation with DAPI and G5, Cage 4b, Cage 4b ⊃ G5, or NPs 5. b Fluorescence intensity of cells incubated with 4b ⊃ G5 or NPs 5 at different times. c Cell inhibition of MHCC-97L cells incubated with 3b, G5, Cage 4b, Cage 4b ⊃ G5, or NPs 5 without/with light irradiation. d CLSM images of MHCC-97L cells after the incubation with PBS and NPs 5 without/with light irradiation (excitation wavelength: FDA 488 nm, PI 543 nm; emission filter: FDA 500–550 nm, PI 550–650 nm)

Full size image

Considering the 1O2 generation capacity and good stability of G5 and PDI derivatives, 3b, G5, Cage 4b, Cage 4b ⊃ G5, and NPs 5 were employed to study the intracellular 1O2 production. Flow cytometry (FCM) experiments (Figs. S48 and S49) utilizing 2,7-dichlorodi-hydrofluorescein diacetate (DCFH-DA) as a reactive 1O2 scavenger [72] revealed that only cells treated with light exhibited good fluorescence, indicating that all the compounds were capable of producing 1O2 under light irradiation. Significantly, the fluorescence intensity of Cage 4b ⊃ G5 and NPs 5 was much stronger than that of other compounds, suggesting that these two compounds possessed the highest 1O2 generation capacity among all tested compounds in cells, consistent with their 1O2 generation quantum yields in solution (Fig. 3d). Moreover, due to the higher cellular uptake of NPs 5 compared to Cage 4b ⊃ G5, NPs 5 exhibited better intracellular 1O2 production.

The anticancer efficacy for PDT involving 2, 3b, G5, Cage 4b, Cage 4b ⊃ G5, and NPs 5 against two human cancer cell lines (MHCC-97L and NCI-H460 cells) was assessed using 3-(4’,5’-dimethylthiazol-2’-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Figs. 4c and S50). In order to ensure the comparability of test results, the absolute concentration of all the compounds was set to be 1.25 μM. Compared with compounds without light treatment, all tested substances exhibited significantly enhanced anticancer activities upon light irradiation. Taking NPs 5 as an example, in the absence of irradiation, a small amount of apoptotic and necrotic cells was detected, which may be related to the chemotherapeutic effect of the platinum(II) contained in the metallacage. However, when treated with laser irradiation (50 mW cm−2, white light, 1 min), the inhibition rate increased from 29 to 92% for MHCC-97L cells. For NCI-H460 cells, NPs 5 achieved 92% inhibition rate under irradiation. Notably, NPs 5 showed the highest cytotoxicity among these compounds. This is likely attributed to the occurrence of FRET from Cage 4b to G5, leading to an increase in the efficiency of 1O2 generation in Cage 4b ⊃ G5. The photodynamic therapeutic effect of NPs 5 was further demonstrated by staining live cells and apoptotic cells with fluorescein diacetate (FDA) and propidium iodide (PI), respectively. As depicted in Fig. 4d, the control groups (PBS, PBS with light irradiation, and NPs 5 without light) exhibited green fluorescent live cells. Only the experimental group (NPs 5 with laser irradiation) predominantly showed red fluorescence, indicating apoptotic cells, providing further evidence of effective photodynamic therapy.

In vivo experiments were further conducted to evaluate the theranostic activities of 2, 3b, G5, Cage 4b, Cage 4b ⊃ G5, and NPs 5. In this study, MHCC-97L tumor-bearing nude mice with subcutaneous xenograft tumor models were employed. Initially, in vivo fluorescence imaging of mice was carried out after tail vein injection of these compounds to evaluate their uptake and biodistribution capability (Figs. S51 and S52). A noticeable fluorescence signal was concentrated at the tumor site in mice after the treatment with 3b, G5, Cage 4b, Cage 4b ⊃ G5, or NPs 5. In contrast, negligible fluorescence signal was detected in other organs such as the liver and kidneys, indicating a significant accumulation of compounds in the tumor area 6-h post injection. Notably, the fluorescence signal of NPs 5 was notably higher than that of other compounds, suggesting that these nanoparticles exhibited long-term fluorescence and possessed excellent tumor accumulation and retention capabilities. The strong fluorescence exhibited by NPs 5 facilitated imaging techniques for monitoring the pharmacokinetics in mice with tumors. Following injection, NPs 5 displayed sustained systemic distribution, and the fluorescence of NPs 5 localized in tumor tissues progressively increased over time (Fig. 5a). Fluorescence imaging conducted at 6-h post injection revealed heightened fluorescence at the tumor site compared to normal organs. This signal remained for at least 12-h post injection, indicating obvious biodistribution of NPs 5 within the sample. Conversely, mice administered with Cage 4b ⊃ G5 displayed a more rapid decline in fluorescence signal throughout the body. This can be attributed to the prolonged circulation times of PEGylated NPs 5, which markedly enhanced their permeability and retention within tumors.

Fig. 5
figure 5

a In vivo fluorescence imaging of the tumor-bearing mice after tail vein injection of Cage 4b ⊃ G5 and NPs 5. b Digital photos of the final tumor tissues harvested from the mice treated with different formulations at day 14. c Tumor volume changes, d tumor mass (∗p < 0.1, ∗  ∗p < 0.01, ∗  ∗  ∗p < 0.001), e body weight changes, and f spleen mass after injection of various formulas with 520 nm (50 mW cm−2) irradiation (35 tumor-bearing female nude mice were divided into seven groups, n = 5/group). g H&E and Ki 67 staining of postoperative sections collected from different groups treated tumor tissues after 14 d therapies (The scale bar is 100 μm)

Full size image

The in vivo antitumor efficacy of the photochemotherapy was further assessed. The mice were randomly divided into seven groups (n = 5/group) when the tumor volume reached ∼100 mm3 and were subcutaneously injected with PBS buffer, 2, 3b, G5, Cage 4b, Cage 4b ⊃ G5, or NPs 5, all at a dose of 2 mg kg−1 in platinum weight, then treated with laser irradiation (50 mW cm−2 for 30 s). The average tumor size of the different groups was observed every 2 days for 2 weeks to evaluate the therapeutic effect. All the test groups, including 3b, G5, Cage 4b, Cage 4b ⊃ G5, and NPs 5, exhibited better antitumor activities compared to the control group administered with PBS and 2, as evidenced by the decreased size of the tumor after treatment (Fig. 5b). As depicted in Fig. 5c, weak therapeutic effects were observed for the mice that only underwent single chemotherapy (2) or single PDT (3b and G5) compared to the PBS group. An exceptional antitumor outcome was achieved for the photochemotherapy group (Cage 4b ⊃ G5 and NPs 5). Notably, the mice treated with NPs 5 exhibited the highest tumor growth inhibition at 14-day post treatment, confirmed by the smallest tumor volume among all the test groups, which aligned with the tumor mass outcome (Fig. 5d). Moreover, the body weights of the mice remained very similar (Fig. 5e). There was no significant reduction in body weight, and no distinct signs of toxic effects such as changes in urination or nervous behavior were observed for all the tested compounds. The spleen mass (Fig. 5f) remained nearly unchanged, indicating that these compounds can be applied as therapeutic agents for cancer treatment.

Hematoxylin and eosin (H&E) staining assay was also conducted to assess the proliferation and apoptosis of cells in the tumor tissue (Fig. 5g). All treatment groups exhibited varying degrees of necrosis compared to the PBS group, indicating that all tested compounds possessed certain antitumor activities. Remarkably, NPs 5 with laser irradiation resulted in noticeable shrinkage and alterations in the tumor cells and the highest level of tumor apoptosis and necrosis, signifying a pronounced inhibitory effect on tumor proliferation. Furthermore, Ki67-positive immunohistochemical staining was utilized, where areas of cell proliferation were marked by brown spots in the captured images in Fig. 5g. Notably, the group treated with NPs 5 under laser irradiation showed the most substantial decrease in the count of Ki67-positive tumor cells compared with other treatment groups. In essence, these results unequivocally confirmed the amplified synergistic therapeutic performance achieved through the combination of chemotherapy and laser irradiation-activated PDT. These results suggested that the host–guest complexation between the metallacages and photosensitizers, which has been demonstrated to increase the 1O2 generation via FRET, played a pivotal role to increase the efficacy of cancer photochemotherapy. The routine blood test analysis and blood biochemical assay were conducted to further assess the potential long-term toxicity of NPs 5 in vivo. All markers remained within normal ranges (Figs. S53 and S54), indicating no significant toxicity or inflammatory response. These findings emphasize the efficacy of combining chemotherapy and PDT in tumor treatment, thereby improving the survival quality of mice and extending their lifespan.

4 Conclusions

In summary, two emissive perylene diimide metallacages with different cavity sizes were prepared and further utilized for complexing hypocrellin-type photosensitizer through host–guest interactions. The aggregation of hypocrellin A in an aqueous solution was suppressed after complexation with the metallacage. Noticeably, the efficient host–guest complexation shortened the distance between the metallacages and hypocrellin A, offering effective FRET from the metallacages to hypocrellin A and increasing the 1O2 generation quantum yields. Thus, the host–guest complexes were further assembled into supramolecular nanoparticles, demonstrating superior photodynamic activities for cancer therapy compared with sole metallacages and photosensitizer, as indicated by both in vitro and in vivo studies. This study offers an efficient strategy to address the photosensitivity limitation of conventional photosensitizers through the host–guest complexation-based FRET, which will promote the development of metallacage-based delivery system for cancer therapy.