Introduction

Photodynamic therapy (PDT) using photosensitizers has garnered attention in the field of cancer treatment because of its high antitumor efficacy, negligible systemic toxicity, spatiotemporal selectivity, and noninvasiveness [1,2,3]. When administered to a body and illuminated with light, photosensitizers generate reactive oxygen species (ROS), mainly singlet oxygen, via intersystem crossing to damage tumor tissues. Although Photofrin® and Laserphyrin® have been commercialized for PDT [4,5,6,7], there remains a risk of skin photosensitivity even in sunlight due to the accumulation of photosensitizers in normal tissues [6, 8]. This is because ROS generation is not cancer-selective; ROS are produced in the presence of light, photosensitizers, or oxygen (3O2). Thus, the side effects of PDT, as well as its therapeutic benefits, depend on the accumulation of photosensitizers, necessitating tumor-selective photosensitizers that preferentially accumulate in tumor tissues over normal tissues.

The unique features of tumors, such as vascular permeability from impaired blood vessels and weak acidity from overexpressed lactic acid, known as the Warburg effect, present opportunities for selective targeting [9,10,11]. Nanoparticles with diameters of 10–200 nm can selectively accumulate in tumor tissues, a phenomenon known as the enhanced permeability and retention (EPR) effect [9, 12]. In addition to passive targeting strategies [13,14,15,16,17,18,19,20], recent developments have focused on activity-controllable photosensitizers triggered by the biochemical hallmarks of tumors, including acidity and overexpressed compounds in cancer cells [21,22,23,24,25,26,27]. These activatable photosensitizers are designed to be in an “OFF state” in normal tissues, where ROS generation by light irradiation is minimized; they are turned to the “ON state” and become active in tumor microenvironments or when exposed to external stimuli such as light, heat, or magnetism. Both passive targeting and active control methods are effective approaches for reducing side effects, such as skin photosensitivity, associated with PDT.

We report the development of chlorophyll‒peptide conjugates as photosensitizers, which combine passive targeting based on the EPR effect with environmentally responsive properties and operate selectively in weakly acidic environments such as tumor tissues (Fig. 1). Chlorophyll derivatives, characterized by a chlorin π-skeleton, exhibit significant absorption in the far-red and near-infrared regions [28] and have been effectively used as photodynamic therapeutic agents [29,30,31]. These derivatives produce ROS when exposed to light in their monomeric state, while their aggregation significantly reduces their phototoxicity through self-quenching. We focused on the aggregation-dependent photophysical properties of chlorophyll derivatives and the unique tumor microenvironments, including vascular permeability and weak acidity, to design chlorophyll‒peptide conjugates that self-aggregate under neutral conditions and disaggregate under acidic conditions. In this strategy, the aggregated conjugates with reduced phototoxicity due to self-quenching circulate in the blood, which is neutral, and selectively accumulate in tumor tissues with defective structures (Fig. 1a). In weakly acidic tumor tissues, the aggregated conjugates disassemble due to electrostatic repulsion caused by the protonation of peptide moieties, resulting in increased phototoxicity (Fig. 1b).

Fig. 1
figure 1

Schematic illustration of the mechanism of tumor-selective PDT. a Tumor-selective accumulation of aggregated chlorophyll‒peptide conjugates on the basis of the EPR effect. b Acid-activatable photosensitizer constructing aggregates of a chlorophyll‒peptide conjugate

Full size image

Our chlorophyll‒peptide conjugates are composed of short peptides and a chlorophyll derivative, which exhibit low immunogenicity, good biodegradability, and high biocompatibility and can be easily synthesized by solid-phase synthesis [32,33,34,35]. These conjugates self-assemble to form nanoparticles, facilitating targeting via the EPR effect. The nanoparticles showed reduced ROS generation and cytotoxicity under neutral conditions upon photoirradiation but exhibited significant phototoxicity under weakly acidic conditions. We anticipate that variations in the physical properties of the conjugates, such as the aggregate size and ROS generation ability, with different peptides will provide valuable insights for designing novel photosensitizers. Acid-activatable conjugates hold promise as cancer-selective photosensitizers capable of reducing phototoxicity in normal tissues.

Experimental procedure

General

1H-NMR spectra were recorded in DMSO-d6 (containing 0.05% (w/v) tetramethylsilane (TMS)) via a Bruker Avance III (400 MHz) spectrometer. The internal standard for these measurements was TMS, with a chemical shift (δH) of 0.00 ppm. Laser desorption/ionization-time of flight (LDI-TOF) mass spectra were recorded using a JEOL JMS-S3000plus2 spectrometer. Ultraviolet‒visible (UV‒vis) absorption and fluorescence emission spectra were measured in solution with a 10 mm quartz cell at room temperature via a Shimadzu UV-3600 spectrophotometer and a Shimadzu RF-5300PC spectrophotometer, respectively. Transmission electron microscopy (TEM) was performed on a JEOL JEM-z2500 microscope at an acceleration voltage of 200 kV. Dynamic light scattering (DLS) measurements were conducted via a Malvern Zetasizer Nano ZS at 25 °C in a 10 mm disposable cell. All the reaction reagents, solvents, amino acids, and base resins used for peptide synthesis were obtained from commercial suppliers and utilized as supplied.

Synthesis of Chl-H5

Chl-H5 was synthesized via solid-phase peptide synthesis using CLEAR-Amide Resin (100–200 mesh). The resin (400 mg, 0.42 meq/g) was initially swollen in CH2Cl2 (5 mL) for 12 h and then washed with N,N-dimethylformamide (DMF, 5 mL × 3 times). Fmoc-protected histidine, 1-hydroxy-7-azabenzotriazole, and 1,3-diisopropylcarbodiimide (3 eq. each) were dissolved in DMF (5 mL) and added to a reaction flask. The flask was agitated for 2 h, and the resin was washed with DMF (5 mL × 3 times). For deprotection of the Fmoc group, a 20% (v/v) piperidine/DMF solution (5 mL) was added to the reaction flask and agitated for 20 min. The resulting resin was washed with DMF (5 mL × 3 times) after the deprotection process. After all the amino acids were coupled, pyropheophorbide-a was condensed in the aforementioned fashion. The conjugate was cleaved from the resulting resin by treatment with an aqueous 95% (v/v) trifluoroacetic acid (TFA) solution for 2 h. After the solvents were evaporated in vacuo, cold Et2O was added to the residue to yield dark green precipitates. The solids were dried, dispersed in H2O, and lyophilized to obtain Chl-H5.

1H-NMR (400 MHz, DMSO-d6) δH = 9.78 (1H, s, 10-H), 9.50 (1H, s, 5-H), 8.88 (1H, s, 20-H), 8.89–8.76 (5H, m, 2-H of His), 8.26 (1H, dd, J = 12, 18 Hz, 3-CH), 7.32–7.21 (5H, m, 5-H of His), 6.41 (1H, d, J = 18 Hz, trans-31-CH), 6.24 (1H, d, J = 12 Hz, cis-31-CH), 5.16, 5.08 (each 1H, d, J = 20 Hz, 131-CH2), 4.54 (1H, q, J = 7 Hz, 18-H), 4.30 (1H, d, J = 8 Hz, 17-H), 3.74 (2H, q, J = 7 Hz, 8-CH2), 3.64 (3H, s, 12-CH3), 3.45 (3H, s, 2-CH3), 3.25 (3H, s, 7-CH3), 3.14–2.89 (10H, m, Cα-CH2 of His), 2.43–2.31, 2.12–2.06 (1H + 2H, m, 17-CH2CH2), 1.77 (3H, d, J = 7 Hz, 18-CH3), 1.64 (3H, t, J = 7 Hz, 81-CH3), 0.34, –1.90 (each 1H, s, NH of chlorin macrocycle). [One proton peak in 17-CH2CH2 overlapped with a solvent peak. Imidazole-NH, amide-NH, and peptidyl Cα-H peaks were not detected.] MS (LDI) results were as follows: m/z = 1219.5, 1241.5 (calcd for C63H71N20O7: 1219.6 [M + H]+, C63H70N20O7Na: 1241.6 [M+Na]+).

Synthesis of Chl-(HK)2H

Chl-(HK)2H was obtained as a dark green solid via the same procedure used for Chl-H5 synthesis.

1H-NMR (400 MHz, DMSO-d6) δH = 9.79 (1H, s, 10-H), 9.50 (1H, s, 5-H), 8.90 (1H, s, 20-H), 8.89–8.79 (3H, m, 2-H of His), 8.26 (1H, dd, J = 12, 18 Hz, 3-CH), 7.45–7.20 (3H, m, 5-H of His), 6.42 (1H, d, J = 18 Hz, trans-31-CH), 6.24 (1H, d, J = 12 Hz, cis-31-CH), 5.19, 5.10 (each 1H, d, J = 20 Hz, 131-CH2), 4.57 (1H, q, J = 7 Hz, 18-H), 4.50 (1H, t, J = 7 Hz, 17-H), 4.33–4.12 (5H, m, Cα-H of peptide), 3.74 (2H, q, J = 7 Hz, 8-CH2), 3.65 (3H, s, 12-CH3), 3.44 (3H, s, 2-CH3), 3.25 (3H, s, 7-CH3), 3.14–2.80 (6H, m, Cα-CH2 of His), 2.35–2.31, 2.16–2.09 (1H + 2H, m, 17-CH2CH2), 1.78 (3H, d, J = 7 Hz, 18-CH3), 1.65 (3H, t, J = 7 Hz, 81-CH3), 1.57–1.18 (16H, m, Cα-(CH2)4 of Lys), 0.35, –1.90 (each 1H, s, NH of chlorin macrocycle). [One proton peak in 17-CH2CH2 overlapped with a solvent peak. Imidazole- and amide-NH peaks were not detected.] MS (LDI) results were as follows: m/z = 1202.5, 1223.5 (calcd for C63H80N17O8: 1201.7 [M + H]+, C63H79N17O8Na: 1223.6 [M+Na]+).

Synthesis of Chl-(KH)2K

Chl-(KH)2K was obtained as a dark green solid via the same procedure used for Chl-H5 synthesis.

1H-NMR (400 MHz, DMSO-d6) δH = 9.77 (1H, s, 10-H), 9.49 (1H, s, 5-H), 8.91 (1H, s, 20-H), 8.90–8.81 (2H, m, 2-H of His), 8.26 (1H, dd, J = 12, 18 Hz, 3-CH), 7.46–7.33 (2H, m, 5-H of His), 6.41 (1H, d, J = 18 Hz, trans-31-CH), 6.24 (1H, d, J = 12 Hz, cis-31-CH), 5.16, 5.11 (each 1H, d, J = 20 Hz, 131-CH2), 4.58 (1H, q, J = 7 Hz, 18-H), 4.35 (1H, d, J = 8 Hz, 17-H), 4.24–4.09 (5H, m, Cα-H of peptide), 3.73 (2H, q, J = 7 Hz, 8-CH2), 3.64 (3H, s, 12-CH3), 3.46 (3H, s, 2-CH3), 3.25 (3H, s, 7-CH3), 3.14–2.88 (4H, m, Cα-CH2 of His), 2.36–2.30, 2.20–2.12 (1H + 2H, m, 17-CH2CH2), 1.81 (3H, d, J = 7 Hz, 18-CH3), 1.64 (3H, t, J = 7 Hz, 81-CH3), 1.56–1.22 (24H, m, Cα-(CH2)4 of Lys), 0.35, –1.89 (each 1H, s, NH of chlorin macrocycle). [One proton peak in 17-CH2CH2 overlapped with a solvent peak. Imidazole- and amide-NH peaks were not detected.] MS (LDI) results were as follows: m/z = 1214.6, 1230.5 (calcd for C63H85N17O7Na: 1214.7 [M+Na]+, C63H85N17O7K: 1230.6 [M + K]+).

Synthesis of Chl-K5

Chl-K5 was obtained as a dark green solid via the same procedure used for Chl-H5 synthesis.

1H-NMR (400 MHz, DMSO-d6) δH = 9.79 (1H, s, 10-H), 9.50 (1H, s, 5-H), 8.91 (1H, s, 20-H), 8.27 (1H, dd, J = 12, 18 Hz, 3-CH), 6.42 (1H, d, J = 18 Hz, trans-31-CH), 6.24 (1H, d, J = 12 Hz, cis-31-CH), 5.22, 5.12 (each 1H, d, J = 20 Hz, 131-CH2), 4.59 (1H, q, J = 7 Hz, 18-H), 4.35 (1H, d, J = 8 Hz, 17-H), 4.25–4.14 (5H, m, Cα-H of Lys), 3.74 (2H, q, J = 7 Hz, 8-CH2), 3.65 (3H, s, 12-CH3), 3.46 (3H, s, 2-CH3), 3.26 (3H, s, 7-CH3), 2.35–2.31, 2.22–2.11 (1H + 2H, m, 17-CH2CH2), 1.81 (3H, d, J = 7 Hz, 18-CH3), 1.65 (3H, t, J = 7 Hz, 81-CH3), 1.72–1.14 (40H, m, Cα-(CH2)4 of Lys), 0.37, –1.88 (each 1H, s, NH of chlorin macrocycle). [One proton peak in 17-CH2CH2 overlapped with a solvent peak. Imidazole- and amide-NH peaks were not detected.] MS (LDI) results were as follows: m/z = 1174.7, 1196.7 (calcd for C63H96N15O7: 1174.8 [M + H]+, C63H95N15O7Na: 1196.7 [M+Na]+).

Preparation of self-assembled conjugates

The conjugate was dissolved in a small amount of ethanol (50 μL). A 20 nmol aliquot of the solution was transferred into test tubes to 20 nmol, and then the solvent was dried. After 2.0 mL of phosphate buffer solution (PBS) at pH 5.8 or 7.4 was added to the dried conjugates, the sample was ultrasonicated for 5 min to prepare self-aggregates dispersed in the PBS medium.

Evaluation of ROS generation

A total of 100 μg of singlet oxygen sensor green (SOSG) [36] was dissolved in 33 μL of MeOH and diluted with 16.467 mL of pure water to prepare an SOSG stock solution (10 μM). A total of 1.0 mL of the SOSG stock solution was added to 1.0 mL of the assembled conjugate dispersion in 1% (v/v) EtOH–PBS at pH 7.4 or 5.8 to obtain a sample for ROS evaluation. The sample was photoirradiated for 5 min (λ = 420 nm) with the light source of a SHIMADZU RF-5300PC fluorescence spectrophotometer, and then, ROS generation was evaluated by fluorescence emission spectroscopy with excitation at 410 nm.

Cytotoxicity test

The cytotoxicity of Chl-(HK)2H was evaluated via the propidium iodide (PI) exclusion test [37]. HeLa cells (1.0 × 105) were plated on 35 mm glass-bottom dishes and cultured for 24 h. The cells were treated with PI (1 µg/mL) at 37 °C. A preprepared dispersion of Chl-(HK)2H aggregates in PBS at pH 5.8 or 7.4 (10 μM) was added to the cell medium and adjusted to a concentration of 0.1 or 1 μM. For the dark cytotoxicity assay, the medium was changed to Dulbecco’s modified Eagle medium (pH 5.8 or 7.4), and the cells were incubated with Chl-(HK)2H at 0.1 or 1 μM. After 0–60 min, the cells were observed via fluorescence microscopy using an EVOS M5000 imaging system (Thermo Fisher Scientific) equipped with a LUCPLFLN20X objective lens (Olympus) and an EVOS Texas Red 2.0 light cube (Thermo Fisher Scientific). For the PDT study, HeLa cells treated with Chl-(HK)2H at 0.1 or 1 μM were photoirradiated (λ = 420 nm) with a 300 W xenon lamp (MAX-303, Asahi Spectra) equipped with an HMX0420 bandpass filter (Asahi Spectra) for 10–30 min. Cell viability was estimated by determining the percentage of non-PI-stained (live) cells relative to the total number of cells counted (n > 600 cells for each condition).

Results and discussion

Design and synthesis of chlorophyll‒peptide conjugates

We designed chlorophyll‒peptide conjugates that change their aggregation ability with varying pH, consisting of a pH-responsive peptide with histidine and/or lysine and a chlorophyll-a derivative, pyropheophorbide-a (Scheme 1, lower, center). Lysine is a hydrophilic amino acid that is constantly ionized under acidic and neutral conditions. The introduction of hydrophilic groups was required to improve the dispersibility of the aggregated conjugates in aqueous solution, so lysine residues with hydrophilic side chains were selected. Histidine is hydrophobic under neutral conditions but hydrophilic under acidic conditions due to the protonation of the imidazolyl side chain. Histidine residues were chosen because this pH-responsive change in hydrophilicity is important for providing chlorophyll with the disassembling property induced under weakly acidic conditions. In this study, we synthesized four chlorophyll‒peptide conjugates with different numbers of lysine and histidine residues, namely, Chl-H5, Chl-(HK)2H, Chl-(KH)2K, and Chl-K5 (Scheme 1, upper, right), to achieve both water dispersibility and pH responsiveness (acid-activating properties) of their aggregates.

Scheme 1
scheme 1

Synthesis of chlorophyll‒peptide conjugates

Full size image

Naturally occurring chlorophyll-a was extracted from Spirulina, and pyropheophorbide-a was prepared via its demetallation, demethoxycarbonylation at the 132-position, and hydrolysis of the 172-esterifying group [28, 38]. Although pheophorbide-a has often been used as a photosensitizer, we chose the chemically stable pyropheophorbide-a without the methoxycarbonyl group at the 132-position in the highly reactive β-keto-ester moiety. Peptides were synthesized on a resin via solid-phase synthesis, and the free (deprotected) N-terminus of the peptides was condensed with pyropheophorbide-a, which possesses a carboxy group. Chlorophyll‒peptide conjugates were obtained by cleavage from the resin via 95% (v/v) TFA–H2O (Scheme 1). The conjugates were characterized using 1H-NMR spectroscopy and LDI-TOF spectrometry (the experimental details are described in the Experimental Section, and their spectra are shown in Figs. S1–S4 in the Supplemental Information).

Self-aggregation of chlorophyll‒peptide conjugates

The morphology and size of the aggregates of the chlorophyll‒peptide conjugates were evaluated via TEM. The conjugate was dissolved in a small amount of ethanol, dried to form a thin film, and then rehydrated with PBS at pH 7.4, followed by sonication to prepare self-aggregates. The dispersion was cast onto a carbon-coated copper grid and incubated for 5 min, and excess PBS was removed to prepare TEM samples of the aggregated conjugates. Nanoparticles were observed in the aggregates of Chl-H5, Chl-(HK)2H, and Chl-(KH)2K prepared in PBS at pH 7.4, with average particle diameters of 329 ± 335 (33–2800 nm), 73 ± 50, and 251 ± 202 nm determined by TEM measurements, respectively (Fig. 2). The larger standard deviation value is due to the variation in the particle size. In particular, the aggregates of Chl-H5 and Chl-(KH)2K were widely distributed, ranging from large aggregates larger than 1 μm to small aggregates with diameters of several dozen nm. For the Chl-K5 aggregates, no nanoparticles could be identified; instead, ameba-like aggregates were observed. We also attempted to determine the size dispersion by DLS but were unable to measure it successfully because the wavelength of the laser used overlapped with the Qy absorption band of the conjugates.

Fig. 2
figure 2

TEM images of self-aggregates of the chlorophyll‒peptide conjugates prepared in PBS at pH 7.4 on a carbon‒coated copper grid

Full size image

Chl-H5, bearing a peptide consisting only of histidine, which is hydrophobic under neutral conditions, exhibited high aggregation in aqueous solution and formed large amorphous particles. In contrast, Chl-K5, which contains only highly hydrophilic lysine residues, did not form nanoparticles because of its lower aggregation propensity. These results suggest that the ratio of histidine to lysine residues significantly affects the size and morphology of the aggregates. Since particles with diameters of 10–200 nm are known to accumulate in tumor tissues [9, 12], Chl-(HK)2H aggregates, with an average particle size of 73 nm, are expected to act as agents capable of cancer-selective accumulation.

pH-dependent photophysical properties

To investigate the acid responsiveness of the aggregated conjugates, their pH-dependent photophysical properties were examined. All the monomeric conjugates in ethanol presented two sharp and strong absorption peaks at 410 and 666 nm, referred to as the Soret and Qy bands [28, 38], respectively (Fig. 3a, black lines). The UV–vis absorption spectra of the aggregates in PBS at pH 7.4 were broadened and slightly redshifted, with Chl-H5 displaying the most broadened and shifted spectrum (Fig. 3a, blue lines). These spectral changes are often observed when chlorophyll and its derivatives form disordered aggregates, implying that the nanoparticles of the conjugate, as shown in Fig. 2, are micelle-like aggregates built by the solvophobic effect. A comparison of the absorption spectra of the aggregates prepared under weakly acidic and neutral conditions revealed an increase in the absorption peak intensity with decreasing pH for Chl-H5 (Fig. 3a, blue to red lines). The aggregation of Chl-(HK)2H and Chl-(KH)2H, which contain three histidine residues, also slightly increased the Soret absorption bands in acidic PBS. Owing to the protonation of the imidazolyl group in the histidine side chain under acidic conditions, the conjugates produce electrostatic repulsion between peptide moieties in addition to increasing the hydrophilicity. Thus, the slight increase in absorption intensity was attributed to the broadened absorption spectrum becoming closer to the monomer spectrum because of the partial disaggregation of the micellar aggregates caused by the electrostatic repulsion and reduction in the hydrophobicity of the conjugates. Chl-K5, lacking any histidyl residues, revealed no change in the absorption spectrum with pH.

Fig. 3
figure 3

a UV‒vis absorption spectra of chlorophyll‒peptide conjugates in EtOH (black) and PBS at pH 5.8 (red) and 7.4 (blue). All the concentrations of the conjugates were 10 µM. b Fluorescence emission spectra of chlorophyll‒peptide conjugates in PBS at pH 7.4 (blue) and 5.8 (red) upon excitation at 410 nm. All the concentrations of the conjugates were 1 µM

Full size image

Although the determination of the pKa of the conjugates by pH titration experiments did not succeed unfortunately, the disassembly of the aggregated conjugates in response to increasing acidity in an aqueous solution was monitored via fluorescence emission spectra. Samples for fluorescence measurements were prepared by diluting the dispersion used for UV–vis absorption spectroscopy ten-fold with PBS at pH 5.8 or 7.4. All the fluorescence emission spectra of the aggregated conjugates revealed an increase in the monomer-derived fluorescence emission at 668 nm with decreasing pH (Fig. 3b). Almost no increase in the absorption intensity with decreasing pH was observed for Chl-K5, but changes in the fluorescence emission spectrum were detected, implying that the amino group of the lysine side chain contributed to acid-induced disaggregation. TEM images of the conjugates prepared in different pH PBS solutions also indicated acid-induced disaggregation. Compared with the TEM images of the aggregates prepared under neutral conditions, the images of the aggregates prepared in PBS at pH 5.8 revealed transformation into amoeba-like aggregates or an increase in the aggregate size (Fig. S5). These transformations associated with acidification are caused by partial disaggregation due to electrostatic repulsion induced by protonation of the peptide moiety on the conjugates.

The fluorescence emission intensity of the conjugates in PBS tended to increase with decreasing number of histidine residues, regardless of pH. However, the incremental ratios of the luminescence of Chl-H5, Chl-(HK)2H, Chl-(KH)2K, and Chl-K5 with increasing acidity (FL intensity at 668 nm at pH 5.8/FL intensity at 668 nm at pH 7.4) were 2.1, 2.8, 3.9, and 2.3, respectively, and were not proportional to the number of histidine residues.

ROS generation

The amount of ROS, particularly singlet oxygen, generated by light irradiation of the aggregates of chlorophyll‒peptide conjugates under acidic or neutral conditions was estimated by fluorescence emission spectral measurements with the SOSG probe. Since SOSG transforms into a fluorophore with emission maxima at approximately 530 nm upon reacting with singlet oxygen, the amount of singlet oxygen produced by light irradiation of the aggregated conjugates can be evaluated by an increase in the fluorescence emission intensity [39]. The dispersion of aggregated conjugates in PBS at pH 5.8 or 7.4 containing SOSG was prepared, and the SOSG samples were either irradiated with light (λ = 420 nm) or left in the dark for 5 min. ROS generation was assessed by comparing the fluorescence emission spectra of photoirradiated and non-photoirradiated SOSG samples. All the fluorescence emission spectra of the SOSG samples at pH 5.8 showed increased intensity at 528 nm upon photoirradiation (Fig. 4, red dotted to solid lines), indicating that singlet oxygen is produced via photoexcitation of the conjugates and subsequent energy transfer to oxygen. In contrast, the increase in emission at 528 nm upon light irradiation of the SOSG samples at pH 7.4 was reduced (Fig. 4, blue dotted to solid lines). Specifically, the SOSG samples containing the conjugates Chl-H5, Chl-(HK)2H, and Chl-(KH)2K, which have histidine residues, showed hardly any spectral changes under light irradiation. These findings suggest that these conjugates can significantly reduce the amount of ROS generation induced by light irradiation under neutral conditions, which is promising for reducing phototoxicity in normal tissues.

Fig. 4
figure 4

Fluorescence emission spectra of the SOSG probe with chlorophyll‒peptide conjugates in PBS at pH 5.8. (red) and 7.4 (blue) upon excitation at 410 nm. The dotted and solid lines show the spectra of the solution after standing for 5 min in the dark and after photoirradiation (λ = 420 nm) for 5 min, respectively. The concentrations of SOSG and the conjugates were 5 and 1 µM, respectively

Full size image

Physical properties of conjugates with different peptides

The sizes of the aggregates and the photophysical properties of the chlorophyll‒peptide conjugates, which are important for tumor-selective photosensitizers, are shown in Table 1. The fluorescence emission intensity and photoinduced ROS generation of the conjugates decreased with increasing ratio of histidine residue, regardless of pH. This is attributed to the influence of the hydrophobic histidine side chains on self-aggregation and the accompanying quenching. Chl-H5, which possesses a peptide composed only of histidine, could be expected to reduce phototoxicity under neutral conditions. However, its acid-induced activation of photophysical properties was less likely to occur because the conjugate formed large aggregates due to strong hydrophobic effects. Conversely, hydrophilic Chl-K5 exhibited high fluorescence emission and ROS generation even in neutral PBS and was not inactivated. Therefore, the balance between hydrophilic and hydrophobic functional groups, as well as pH responsivity, is key to developing acid-activatable photosensitizers. Indeed, Chl-(HK)2H and Chl-(KH)2K, which contained a mixture of the two amino acids, presented greater fluorescence intensity and increased ROS generation with decreasing pH (7.4 to 5.8) than did Chl-H5 and Chl-K5. Notably, Chl-(HK)2H displayed the greatest change in singlet oxygen generation with pH (Table 1, right column) and self-assembled to form nanoparticles capable of tumor-selective accumulation based on the EPR effect (10 < 73 < 200 nm, second left column). Thus, this aggregate is expected to be a photosensitizer capable of both passive and active targeting.

Table 1 Average diameter, fluorescence emission, and ROS generation data of the chlorophyll‒peptide conjugates
Full size table

Cytotoxicity of conjugates

The dark toxicity and phototoxicity of Chl-(HK)2H were investigated through cytotoxicity tests in HeLa cells [37]. Samples for the PI assay were prepared by adding PI and Chl-(HK)2H aggregates to HeLa cells in media adjusted to pH 5.8 and 7.4, resulting in final conjugate concentrations of 0.1 and 1 µM, respectively. The samples were incubated under light-shielded and illuminated conditions (λ = 420 nm) for 10, 20, and 30 min, and cell viability was determined by counting the number of living and dead cells via fluorescence microscopy (Figs. S6 and S7). The cell viability of the samples incubated in the dark, irrespective of pH or concentration, was greater than 98%, which was almost identical to that of the control samples without the conjugate (Fig. 5). Even with a 10 µM concentration of the conjugate incubated in the dark, the cell viability remained higher than 90%, indicating that Chl-(HK)2H did not exhibit dark toxicity (Fig. S8). When the samples with 0.1 and 1 µM Chl-(HK)2H were photoirradiated at pH 5.8, the cell viability gradually decreased, eventually reaching 50% and 7%, respectively, after 30 min of incubation. Since no cell death was observed after light irradiation of the control sample, these results demonstrated that the ROS generated by light irradiation of Chl-(HK)2H induced cell death (Fig. 5).

Fig. 5
figure 5

Cell viabilities of HeLa cells incubated with Chl-(HK)2H for 0, 10, 20, and 30 min at pH 7.4 (blue) and 5.8 (red). The two graphs on the right and left show the cell viabilities after incubation in the dark and with photoirradiation (λ = 420 nm), respectively: mean ± SD, n = 3

Full size image

A phototoxicity study of the PI assay samples at pH 7.4 was conducted similarly. After 30 min of light irradiation of the 1 µM sample, the cell viability decreased to 4%, which was almost identical to that at pH 5.8, revealing high cytotoxicity even under neutral conditions with high concentrations of Chl-(HK)2H. However, the cell viability of the 0.1 µM sample at pH 7.4 decreased less, which was significantly different from the cytotoxicity results in an acidic medium. Therefore, low concentrations of Chl-(HK)2H function as photosensitizers capable of reducing cytotoxicity under neutral conditions, similar to normal tissue, and inducing cell death under weakly acidic conditions, such as in tumor tissue.

Conclusion

We designed and synthesized chlorophyll‒peptide conjugates Chl-H5, Chl-(HK)2H, Chl-(KH)2K, and Chl-K5 containing histidine and/or lysine residues to develop photosensitizers capable of both cancer-selective accumulation through passive targeting and acid-activatable phototoxicity. The ratio of hydrophilic lysine to hydrophobic histidine significantly affected the size of the self-aggregates of the conjugates and their pH-dependent photophysical properties. In particular, Chl-(HK)2H self-assembled into nanoparticles with a diameter of approximately 70 nm, which are expected to achieve tumor-selective accumulation based on the EPR effect and a reduction in ROS generation via self-aggregation under neutral conditions. In contrast, the aggregates of the conjugates partially disaggregated, and the photosensitizing activity increased with decreasing pH. Cytotoxicity testing using HeLa cells revealed that Chl-(HK)2H exhibited negligible dark toxicity regardless of the aqueous pH and sample concentration. Cell death was observed when a low concentration of the conjugate was exposed to light irradiation under weakly acidic conditions, confirming its function as a photosensitizer. We propose that the conjugate is useful as a photosensitizer capable of both passive targeting and acid activation based on the characteristics of tumor tissue. We expect that this study will contribute to the development of cancer-selective photosensitizers.