All-natural-molecule, bioluminescent photodynamic therapy results in complete tumor regression and prevents metastasis

Abstract

Self-luminescent photodynamic therapy (PDT) has gained attention owing to its potential to enable effective phototherapy without the bottleneck of shallow light penetration into tissues. However, the biosafety concerns and low cytotoxic effect of self-luminescent reagents in vivo have been problems. Here, we demonstrate efficacious bioluminescence (BL)-PDT by using bioluminescence resonance energy transfer (BRET) conjugates of a clinically approved photosensitizer, Chlorin e6, and a luciferase, Renilla reniformis; both derived from biocompatible, natural molecules. With over 80% biophoton utilization efficiency and membrane-fusion liposome-assisted intracellular delivery, these conjugates produce effective, targeted cancer cell killing. Specifically, in an orthotopic mouse model of 4T1 triple-negative breast cancer, BL-PDT showed strong therapeutic effects on large primary tumors and a neoadjuvant outcome in invasive tumors. Furthermore, BL-PDT resulted in complete tumor remission and prevention of metastasis for early-stage tumors. Our results demonstrate the promise of molecularly-activated, clinically viable, depth-unlimited phototherapy.

1. Introduction

Photodynamic therapy (PDT) is a therapeutic treatment modality for many diseases, including cancers [1–2]. PDT employs light and photosensitizers that, upon absorbing photon energy, generate reactive oxygen species (ROS) that can lead to various therapeutic effects, such as killing malignant cells and activating anti-tumor immune responses. However, effective PDT critically relies on a sufficient light dose at target tissues, typically requiring an optical fluence of ~ 50 J cm⁻² at the surface of the tissue. This dose can produce PDT effects at tissue depths up to a few millimeters below the surface [3–7]. This shallow depth has confined PDT to superficial structures and requires extrinsic, fiber-optic light sources to treat endoscope-accessible tissues [8].

To overcome these limited therapeutic depths, a promising strategy for replacing the external light illumination of conventional PDT is by use of self-luminescence based on chemiluminescence (CL) [9–12] or bioluminescence (BL) as an internal light source [13–17]. Although molecular luminescence has a much weaker intensity than conventional light sources, such as lasers and light-emitting diodes, efficient Förster resonance energy transfer (FRET) from the luminescent group to photosensitizers can, in principle, compensate for the difference in intensity and produce substantial PDT effects. Recently, several combinations of self-luminescent donors (CL or BL) and photosensitizer acceptors have been reported, showing ROS-induced cell death in vitro and appreciable tumor growth inhibition in vivo [9–10, 16–17]. However, current self-luminescent PDT is confronted with two major problems: First, the in vivo therapeutic efficiency is modest. This is mainly limited by low luminous efficiency (quantum yields < 0.1 [18]) for the case of chemiluminescence-PDT (CL-PDT) [9–10], which relies on endogenous H₂O₂ at low concentrations, and the low capture and utilization rate (< 40%) of biophotons in the case of bioluminescence-PDT (BL-PDT) [15–17]. Second, the self-luminescent reagents demonstrated so far present cytotoxicity concerns: luminol used in CL-PDT has a high affinity to serum albumin and DNA, raising concerns about biosafety [14, 19]. Quantum dots used in BL-PDT also bring with them the risk of inorganic nano-material toxicity [13–15].

Here, we report highly efficient bioluminescence resonance energy transfer (BRET)-induced BL-PDT enabled by a novel reagent consisting of a clinically used photosensitizer, Chlorin e6 (Ce6), and the Renilla reniformis Luciferase 8 (RLuc8) protein. As both Ce6 and RLuc8 are natural or naturally derived molecules, RLuc8-Ce6 conjugates offer promising potential for clinical translation. Optimized RLuc8-Ce6 conjugates achieve 80% activation efficiency (i.e., the probability of producing one activated Ce6 molecule per BL photon) and are efficiently transported into cancer cells via fusogenic nano-liposomes. We compare the effects of BL-PDT with conventional PDT for 4T1 murine and MDA-MB-231 human triple-negative breast cancer (TNBC) cells in vitro and then in in vivo orthotopic mouse models. Our results show strong therapeutic BL-PDT effects on larger tumors (7~8 mm) as well as neoadjuvant effects at tumor margins. We also demonstrate complete tumor remission and metastasis prevention in the sentinel lymph nodes and lungs.

2. Materials and methods

2.1. Synthesis of Luc-Ce6 conjugate.

The Luc-Ce6 conjugate was synthesized by directly conjugating Renilla luciferase 8 (Rluc8, or simply Luc, G-Biosciences) with chlorin e6 mono 6-amino hexanoic acid amide (Ce6-HA) molecules (Santa Cruz Biotechnology) using the standard N-(3-dimethyl aminopropyl)-N-ethyl-carbodiimide hydrochloride (EDC)/N-hydroxysulfosuccinimide (Sulfo-NHS)-activated condensation reaction. Briefly, 1 mg Ce6-HA (1.4 × 10⁻⁶ mol) powder was thoroughly dispersed in 100 μl dimethyl sulfoxide (DMSO) and then mixed with 400 μl DPBS buffer to make a 0.5 ml, 2 mg/ml solution. Then, 100 μl E.D.C. (2.6 × 10⁻⁵ mol) and 50 μl Sulfo-NHS (4.6 × 10⁻⁶ mol) were added into a Ce6-HA solution and reacted for 20 min at room temperature. After this crosslinking reaction, 447 μl 2-mercaptoethanol (55 mM) was added to the solution to quench the excess EDC, and the sample was incubated for 10 min at room temperature. For conjugation, a new modified Ce6-HA solution was added into a Rluc8-DPBS solution, gently pipetted several times, and then reacted for 3 h in an Eppendorf ThermoMixer-C instrument (300 rpm) at room temperature. The ratio of Ce6-HA and Rluc8 was varied to test different mole conjugation ratios. After conjugation, the mixture solution was filtered using Zeba spin desalting columns (7K MWCO) to eliminate unreacted small molecules and obtain the final Luc-Ce6 DPBS solution.

2.2. Characterization of Luc-Ce6 conjugate.

An electron-multiplication charge-coupled device (EMCCD)-equipped spectrometer (Newton, Andor) was used to measure bioluminescence (BL) spectra from various solutions. To measure BL spectra from RLuc8 and methoxy e-coelenterazine (Me-eCTZ, or simply CTZ), 1 ml, 0.3 μM CTZ was added to a 1 ml, 0.3 μM Luc-Ce6 DPBS solution in a polystyrene cuvette. The BL emission was collected with 0.1 s exposure and 300 frame averaging. The fluorescence spectra of Ce6-HA solutions (from 1.7 to 24.9 μM) and Luc-Ce6 solution (0.38 μM) were measured by using the EMCCD-coupled spectrometer and a xenon lamp filtered by a monochromator. A microplate reader (Epoch 2, Biotek Instruments) in the spectrophotometer mode was used to measure the absorption spectra of Ce6-HA at different concentrations (from 8.3 to 166 μM) and of Luc-Ce6 conjugates (1.39 and 2.78 μM). A dynamic light scattering instrument (Malvern Zetasizer Nano-ZS, 633 nm) was used to measure the hydrodynamic size and Zeta potential.

2.3. Detection of singlet oxygen in solution.

A singlet oxygen sensor green dye (SOSG, Thermo Fisher Scientific), highly selective for ¹O₂, was used to measure the amount of singlet oxygen in the solution. A mixture solution of Luc-Ce6 or Ce6-HA with SOSG (dissolved in methanol) was prepared (10 μM SOSG in the mixture), and CTZ with different molar ratios (20 to 150) was added to BL-PDT sample solutions. The laser-PDT sample solutions were illuminated with a 405 nm laser beam at 1 J cm⁻². The fluorescence spectra of the samples were measured 30 min after treatments.

2.4. Fusogenic nano-liposomes (Fuso-lip) and Luc-Ce6 loading.

The two main components of liposomes, 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) and 1,2-dioleoyl-sn-glycerol-3-phosphoethanolamine (DOPE) were purchased from Avanti Polar Lipids (stored at −80 °C). A 1 mg ml⁻¹ stock solution of DOPE/DOTAP with a weight ratio of 1:1 was prepared. After evaporating the chloroform solvent, DOPE and DOTAP were resuspended in DPBS buffer at 1 mg ml⁻¹ concentration. For Luc-Ce6 loading, a 500 μl, 27.8 μM fresh solution of Luc-Ce6 was added into an 800 μl DOPE/DOTAP solution, stirred vigorously for 10 min (300 rpm), and incubated in an ultrasonic bath (Elmasonic, 37 kHz, 30% power) for 5 min at room temperature. To reduce the size of conjugate-loaded liposomes, a mini-extruder instrument (Avanti Polar Lipids) was used to gently extrude the Luc-Ce6/Fuso-lip solution several times through membrane filters with 100 nm pore sizes for in vitro experiments or 50 nm pore sizes for in vivo experiments. After extrusion, 3,3′-dihexadecyloxacarbo-cyanine perchlorate (DiO) or 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-lindotricarbocyanine iodide (DiR) was added into the liposome solution with a ratio of 1/15 (w/w) for fluorescent staining of the liposomes. The sample was diluted with DPBS buffer to the desired working concentration.

2.5. Cell culture and in vitro cytotoxicity evaluation.

Murine fibroblast L929 cell line cells, murine TNBC 4T1 cells, and human TNBC MDA-MB-231 cells were purchased from ATCC (American Type Culture Collection). Cells were cultured in recommended standard conditions. DMEM basic cell media was supplemented with fetal bovine serum (10%, v/v) and penicillin/ streptomycin (1%, v/v), known as “complete media.” Cells were incubated at 37 °C in a humidified atmosphere of 5% CO₂ and 95% air. For the cytotoxicity studies, cells were seeded on a 96-well plate (~5,000 cells per well) with 100 μl complete media and cultured for 24 h prior to adding different test reagents: CTZ (25, 50, and 100 μM), Fuso-lip (50, 75, 100, and 150 μg ml⁻¹), or Luc-Ce6/Fuso-lip (0.75, 1.25, 1.7, 2.5, 3.3, 5.0, 6.7, and 10 μM). After incubation for 4 h for CTZ and 30 min for Fuso-lip and Luc-Ce6/Fuso-lip, the cells were washed two times by DPBS buffer and cultured in fresh cell media for 24 or 48 h. The cell viability was measured by using a cell counting kit-8 (CCK-8, ApexBio) assay. The absorbance was measured by using a spectrophotometer (Epoch 2, Biotek Instruments).

2.6. Intracellular delivery of Luc-Ce6 conjugate.

Cells were plated in glass-bottomed well plates with the same density. After culturing for 24 h, the cell media was removed, a Luc-Ce6/Fuso-lip DPBS solution was added, and the cells were incubated for 30 min. After removing the Luc-Ce6/Fuso-lip solution, the cells were washed three times and cultured in fresh cell media for further experiments or fixed by 4% paraformaldehyde (PFA) for imaging. As a control group, Ce6-HA solution was added to the culture plate and incubated for 2 h before fixing with 4% PFA. The cell nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). A lyso-tracker probe (Invitrogen, Red DND-99) was used to stain intracellular lysosomes. 1 mM probe stock solution was diluted to 1 μM in DPBS buffer and added to the cells in glass-bottom plates to reach 50 nM in the media. After 1 h incubation, the cells were washed 3 times with DPBS buffer, fixed with 4% PFA, and examined with fluorescence confocal microscopy. A mitochondria tracker Red CMXRos dye (Invitrogen) and 2’,7’-dichlorodihydrofluorescein diacetate (DCFDA, Cell Biolabs) was used for mitochondrial and intracellular ROS staining, respectively. The Mito-Tracker probe was dissolved in DPBS buffer to prepare a 50 nM solution. DCFDA was dissolved in DMSO to form a stock solution (of 500 μM) and diluted into the Mito-Tracker DPBS buffer to obtain a 5 μM working solution. After incubation with Luc-Ce6/Fuso-lip, 4T1 cells in a glass-bottom plate were washed twice with DPBS and added to the Mito-Tracker/DCFDA-DPBS mixture solution. After further incubation for 30 min at 37 °C, the cells were washed three times, fixed, and stained with DAPI for imaging.

2.7. Measurement of intracellular ROS.

4T1 Cells were plated on a glass-bottomed plate. After cultured for 24 h, Luc-Ce6/Fuso-lip or Ce6 DPBS solution was added and incubated for 30 min or 2 h, respectively. After washing twice with DPBS, a 50 nM DCFDA/DPBS labeling solution was added into the cell culture and incubated for 30 min at 37 °C. For the BL-PDT group, CTZ (100 μM) was added, and the cells were incubated for 4 h. For the laser-PDT group, a 405 nm laser was illuminated with a fluence of 1 J cm⁻². After completing the treatment, the cells were washed three times and fixed by 4% PFA for imaging.

2.8. Assessment of therapeutic effects in vitro.

4T1 (5,000 cells per well) and MDA-MB-231 cells (10,000 cells per well) were seeded in 96-well plates and incubated for 24 h. After washing with DPBS, the cells received Luc-Ce6/Fuso-lip at different concentrations and incubated for 1 h at 37 °C. Next, the cells were washed with DPBS, received 100 μM CTZ, and incubated for 4 h. The cells were then washed and rested for 24 h at 37 °C. The viability of the cells was measured by using the CCK-8 assay. Control cells underwent the same washing and incubation process without adding Luc-Ce6 and CTZ reagents. For laser-PDT, cells were incubated with a Ce6-HA DPBS solution for 2 h and then irradiated with a 405 nm laser for 1 J cm⁻². A Live/Dead^™ cell imaging kit (Invitrogen) was used for imaging analysis of cell viability. 4T1 cells were incubated for 24 h, rinsed with DPBS, and stained with the live/dead kit for 15 min at room temperature. Finally, the cells were washed with DPBS and fixed by 4% PFA for imaging. For apoptosis assay, 4T1 cells were plated in 24-well plates at a density of 5 × 10⁴ cells per well and cultured for 24 h. After incubation with Luc-Ce6/Fuso-lip for 30 min or Ce6 for 2 h, 4T1 cells were treated with 100 μM CTZ for BL-PDT and 405 nm laser (1 J cm⁻²) for laser-PDT. After incubation for 24 h, the cells were harvested, washed twice with cold DPBS, and then resuspended in DPBS buffer. The cell solution was then mixed and incubated with Annexin-V-FITC and PI (Abcam) working solution for 20 min at room temperature. Flow cytometry (BD FACSAria Cell Sorting System) was used to measure the apoptosis signals from the fluorescent probes.

2.9. Tissue penetration efficiency study in vitro.

To measure the penetration depth of the 405 nm laser, chicken breast tissue slices with a thickness of 50 μm each were prepared through standard freezing tissues and cryosection. Each breast tissue slice was fully spread and put on a glass slide. The glass slides were stacked on top of each other to simulate thicker tissues (50 μm for one slice to 300 μm for 6 stacks). 4T1 cells were prepared in a plate and incubated with Ce6-HA for 2 h. Then, 405 nm laser light with a fluence 1 J cm⁻² was illuminated on top of the first tissue slice. The optical attenuation through tissue stacks was measured using an optical power meter (Thorlabs, PM400). After the laser-PDT, the cells were incubated for 24 h, and their viability was measured using the CCK-8 assay.

2.10. In vivo tumor model.

According to the NIH guidelines, all animal studies and related protocols (2015N000205) have been reviewed and approved by the MGH Institutional Animal Care and Use Committee (IACUC). BALB/c female mice (5~6 weeks old) were purchased from Jackson Laboratories. A metastatic TNBC orthotopic model was produced by carefully injecting ~ 10⁶ 4T1 tumor cells into the mammary fat pad of a mouse. The tumor volume was measured from Week 1 following inoculation by external manual examination. The tumor volume was calculated as (length times widtĥ2) 2⁻¹. Each week post-inoculation, tumor tissues, sentinel Lymph nodes (SLNs), and lungs were harvested to analyze the extent of tumor invasion and metastasis. Tissue sections were stained with Hematoxylin & Eosin (H&E) and antibodies against Ki-67 and epidermal growth factor receptor (EGFR).

2.11. In vivo biodistribution after intratumoral delivery.

Mice with orthotopic 4T1 TNBC tumors were randomly divided into two groups (n=3 each). Fuso-lip stained with DiR dyes was used. A Luc-Ce6/Fuso-lip PBS solution was injected into the tumor at several locations to cover the entire volume of the tumor at a 2.5 mg kg⁻¹ dose (based on the Luc-Ce6 amount). Mice were sacrificed at 6 h after injection. The entire tumor was carefully collected and quickly frozen in OCT gel (Tissue-Tek^®) for cryosection. To obtain the spatial distribution of Luc-Ce6 in the tumor, four tumor slices (8 μm thick each) were cut with an inter-slice distance of 50 μm. The sham-treated control group received the same multipoint intratumoral injection of PBS. The tumor slices were imaged using a fluorescence microscope (Keyence, BZ-X700).

2.12. In vivo antitumor efficacy.

For initial assessment, 4T1 tumor-bearing mice were randomly divided into 4 groups (n=3 each), each receiving PBS, Luc-Ce6/Fuso-lip only, BL-PDT, and laser-PDT. On Day 14, after tumor inoculation, Luc-Ce6/Fuso-lip PBS solution was directly injected into the tumor at multiple locations at 2.5 mg kg⁻¹ dose. 50 μl PBS was injected into the PBS group. For the BL-PDT group, 50 μl CTZ (100 μM) was intravenously injected into the tail vein at 6 h after the injection of Luc-Ce6/Fuso-lip, and additional CTZ with the same amount was administered every 4 h for a total 20 h duration. For the laser-PDT group, the skin directly on top of the tumor was irradiated with a 405 nm laser with a dose of 90 J cm⁻². 3 days after treatment, the mice were euthanized, and the whole tumors were collected and fixed in 4% PFA for 2 days. Then the tumor tissues were transferred for staining for H&E, Ki-67, EGFR, and TUNEL following the standard protocol. The tissue slices were scanned by using a Nanozoomer Slide Scanner (Hamamatsu).

For testing early-stage tumor treatments, mice with orthotopic 4T1 tumors were randomly divided into 4 groups (n = 5 each). On day 7, after tumor inoculation, Luc-Ce6/Fuso-lip PBS solution was injected into the tumor in 2.5 mg/kg dose. Starting from 6 h after the injection, CTZ (100 μM) was intravenously injected into the BL-PDT group 6 times with an interval of 4 h. In addition, a single 405 nm laser irradiation (90 J cm⁻² at the skin) was given to the laser-PDT group. Following the treatments, the tumor size, body weight, and general health conditions of the mice were measured every other day. 17 days after the completion of the treatment, the mice were euthanized. Sentinel lymph nodes (SLNs) and lungs were collected and fixed with 4% PFA for 1 day and embedded into paraffin for sectioning. Tissue sections of the SLNs and lungs were stained with H&E and antibodies against Ki-67 and EGFR. Additionally, various other organs, such as the heart, liver, kidney, and spleen, were also harvested, fixed by 4% PFA, and examined by H&E histology.

2.13. BL-PDT for neoadjuvant therapy.

2 weeks post tumor inoculation, mice with orthotopic 4T1 tumors were randomly divided into PBS control, Laser-PDT, and BL-PDT groups (n=3 each). For the laser-PDT and BL-PDT groups, Luc-Ce6/Fuso-lip-PBS solution (1.0 mg kg⁻¹) was carefully injected into the tumor boundaries on all sides using a fine needle. For distribution measurement, 6 h after the injection, the tumor, and its neighboring normal tissues were harvested and quickly frozen in OCT gel for cryo-sectioning. For testing neoadjuvant therapy, 50 μl CTZ (100 μM) was intravenously injected every 4 h, a total of 6 times for the BL-PDT group. In addition, 90 J cm⁻² of 405 nm laser was given to the laser-PDT group. 5 days after treatment, tissues, including the tumor boundary, were harvested and fixed in 4% PFA for staining for H&E, Ki-67, EGFR, and TUNEL.

2.14. Statistical analysis.

The data are presented as means ± standard deviation, as noted in each case. The number of samples, n, is provided. A two-way analysis of variance (ANOVA) was used. P < 0.05 was considered to indicate a statistically significant difference. P < 0.05, P < 0.01, and P < 0.001 are indicated with single, double, and triple asterisks, respectively.

3. Results

3.1. Synthesis and BRET efficiency of Luc-Ce6 conjugate

Scheme 1a illustrates the working principle of the Luc-Ce6 conjugate. Briefly, Ce6-HA molecules (710 Da) were conjugated to RLuc8 proteins (~37 kDa) using EDC/Sulfo-NHS chemistry (Figure S1). The conjugation was formed between the surface-active amino residues of Rluc8 protein (at least 39 from a total of 311 amino acid residues [20–21]) and the acid amide of the Ce6-HA, resulting in Ce6 conjugated on the surface of RLuc8. When a substrate methoxy e-coelenterazine (Me-eCTZ, or briefly CTZ hereinafter) binds with RLuc8 (briefly Luc8), the BL reaction produces energy that is desired to be transferred to one of the Ce6 molecules via BRET. In addition, the excited Ce6 may react with an oxygen molecule nearby and convert it to singlet oxygen (¹O₂). This probability of ROS generation from activated Ce6 is 65–70% in typical in vitro and in vivo conditions [22–23].

Scheme 1.

BL-PDT on 4T1 triple-negative breast cancer tumors (TNBC) in vivo. (a) Schematic of the structure of Luc-Ce6 conjugate and its bioluminescence resonant energy transfer (BRET) principle. Upon administration of substrate Me-eCTZ, the generated bioluminescence resonant energy from the Luc8 core is transferred to the conjugated Ce6 molecule and subsequently activated Ce6 to produce therapeutic reactive oxygen species. (b) Schematic illustration of intracellular delivery of Luc-Ce6 by fusogenic nanoliposomes (Fuso-lip). (c) BL-PDT by locally injected Luc-Ce6 and systematically injected CTZ results in complete tumor remission and prevention of metastasis of TNBC tumors.

A mixture of Luc8 and CTZ in solution emits BL photons with a center at 415 nm, which overlaps well with the absorption peak of Ce6 at 406 nm (Figure 1a). When Luc8 (0.3 μM) and Ce6-HA (7.5 μM) were simply mixed in solution with CTZ, fluorescence from Ce6 at 660 nm appeared (Fig. 1a) due to the absorption of BL photons by Ce6 and BRET from Luc8/CTZ to Ce6 molecules in proximity (< 5 nm). We fabricated Luc-Ce6 conjugates with different ratios of Ce6 to Luc8. As the ratio increased from 15:1 to 42:1, the BL intensity at 440 nm decreased. The fluorescence intensity at 660 nm reached the maximum at a ratio of 25:1 (Figures 1b, S2b, and S3b). A simple kinetic model considering a BRET efficiency and hindrance of Ce6 to the interaction of Luc explained the experimental result reasonably well (Figure 1c). We determined 25:1 to be the optimal ratio and used this value in all subsequent experiments.

Figure 1. Physicochemical characterization of the Luc-Ce6 conjugate.

(a) Normalized BL spectrum of Luc (blue), absorption spectra of Ce6 (green), and BL spectra of the mixture of Luc and Ce6 (the ratio is 1:25) (purple). (b) BL spectra of Luc-Ce6 conjugates with different mole conjugation ratios of Ce6 molecules to Luc protein. (c) The dependence of spectral energy calculated from BL spectrums for the donor and acceptor on the number of Ce6 molecules per Luc8. (d) BL spectra of Luc8 protein (0.3 μM), the mixture of Luc (0.3 μM) and Ce6 (7.5 μM), and Luc-Ce6 conjugate (0.3 μM for Luc). A: donor emission (400–600 nm) area, B: acceptor emission (600–800 nm) area. (e) Changes in the BL and SOSG fluorescence intensity as the proportion of substrate CTZ varied at two different concentrations of Luc-Ce6. (f) SOSG fluorescence intensity as a function of Ce6 concentration in solution after laser-PDT.

Figure 1d shows the representative emission spectrum of Luc-Ce6 conjugates (0.3 μM), integrated over 1 min (BL of Luc lasts about 1 min [15]), with CTZ (~1 μM). The BRET ratio, defined by the ratio of total acceptor emission (B, 600–800 nm) to the total donor emission (A, 400–600 nm), was 0.71. The ratio of activated Ce6 molecules (transferred photons) to the total number of produced BL photons was 80% (Table S1). This is a remarkably high efficiency of excitation. For far-field illumination, the absorption cross-section of Ce6 is ~ 1 × 10⁻¹⁵ cm² at 405 nm, which means 10¹⁵ photons in 1 cm² are needed to excite a single Ce6 molecule. In conventional PDT using far-field illumination, most photons are not absorbed by photosensitizers and are wasted. By contrast, 80% of BL photon is used to excite Ce6 (Table S1).

We used a singlet oxygen sensor green (SOSG) dye to detect ¹O₂ generated by Luc-Ce6 and CTZ in solution [24]. The SOSG fluorescence intensity increased nearly linearly with increasing CTZ (Figure 1e) before it saturated when the ratio of CTZ to Luc8 exceeded 100. A similar trend was observed with conventional laser-PDT using Ce6-HA as the photosensitizer and 405 nm laser irradiation (Figure 1f and Figure S5). Regarding singlet oxygen generation, 280 μM CTZ on 2.8 μM Luc-Ce6 is equivalent to laser fluence of 1 J cm⁻² on 10 μM Ce6.

3.2. Intracellular delivery of Luc-Ce6 conjugates

Intracellular delivery of protein constructs has been challenging [25–26]. We found fusogenic nano-liposomes (Fuso-lip) to have a high affinity to the cell membrane (Fig. S6 and Fig. S7). Fuse-lip, upon fusing into the cellular membrane, can directly inject cargo into the cytoplasm (Scheme 1b) [27]. We loaded Luc-Ce6 conjugates in Fuso-lip using a charge differential drive followed by membrane extrusion through 50 nm pores. The final Luc-Ce6/Fuso-lip complex had a zeta potential of −17.8 mV (Figure S8). We tested intracellular delivery with 4T1 murine breast cancer cells. We found fuso-lip to fuse immediately with the cell membrane and deliver Luc-Ce6 conjugates into the cytosol (Figures 2a and S9). The cells incubated with 5 μM Luc-Ce6 (containing 75 μM Ce6) in Fuso-lip exhibited even stronger Ce6 fluorescence than cells incubated with 25 μM Ce6-HA for 120 min incubation time. Flow cytometry analysis supported the effectiveness of Luc-Ce6 delivery by Fuso-lip, which was comparable to the intracellular uptake of small Ce6-HA molecules (Figure 2b). The intracellular distribution of Luc-Ce6 was not correlated with the distribution of lysosomes (Figures 2c, S10, and S11). This supports that the uptake mechanism of Luc-Ce6/Fuso-lip escapes from typical endocytosis and does not involve the endosomal-lysosomal system [28]. Ce6 molecules accumulate in the mitochondria membrane after entering cells partly because of their hydrophobic aromatic rings [29]. This association with mitochondria is important for the cytotoxicity of PDT [30]. We observed an overlap of mitochondria with Ce6 and Luc-Ce6 (Figures 2d, S12, and S13), as well as the fluorescence from DCFDA, an intracellular ROS sensor, in both laser-PDT (Fig. S14) and BL-PDT (Figure S15). Given the similar intracellular distributions, we expect the cytotoxic mechanism to be similar between BL-PDT and laser-PDT using Ce6.

Figure 2. Intracellular delivery of Luc-Ce6 conjugates.

(a) Fluorescence images of 4T1 cells after incubation with 5 μM Luc-Ce6 (red)/Fuso-lip (green) for 30 min, or Ce6-HA molecules (25 μM) for 2 h. (b) Flow cytometry data of cells incubated with 5 μM Luc-Ce6/Fuso-lip and Ce6-HA with different concentrations. (c) Confocal fluorescence images of 4T1 cells after incubation with Luc-Ce6 (red)/Fuso-lip over time. Purple: lysosomes, blue: nucleus. (d) Confocal fluorescence images of Luc-Ce6/Fuso-lip-treated and Ce6-treated 4T1 cells. Purple: mitochondria, green: intracellular ROS. (e) The hydrodynamic sizes of the Luc protein, Luc-Ce6 conjugate, Fuso-lip, and Luc-Ce6/Fuso-lip. The final Fuso-lip with Luc-Ce6 is about 100 nm. (f) Fluorescence microscopy images of the whole tumor cryosection from the Luc-Ce6/Fuso-lip-treated mice. Fuso-lip: yellow, Luc-Ce6: red, BF: bright field.

We further assessed the delivery and distribution of tumors after intratumoral injection of Luc-Ce6/Fuso-lip. The hydrodynamic size of the Luc-Ce6/Fuso-lip complex was about 100 nm (Figure 2e). Luc-Ce6/Fuso-lip (2.5 mg kg⁻¹) was intratumorally injected into the TNBC orthotopic tumor (~180 mm³) using a fine needle at several sites. In addition, we harvested tumors 6 h after intratumoral injection of Luc-Ce6/Fuso-lip and found the reagent distributed extensively in the tumor (Figures 2f, S16, and S17).

3.3. In vitro BL-PDT on 4T1 and MDA-MB-231 cells

We evaluated the intrinsic cytotoxicity of CTZ, Fuso-lip, and Luc-Ce6/Fuso-lip reagents with 4T1 (murine) and MDA-MB-231 (human) TNBC cell lines as well as L929 (mouse fibroblast) cells in vitro using a Cell Counting Kit-8 (CCK-8) assay (Figure S18). No significant cytotoxicity was observed from all three cell types for 48 h, even with high concentrations of Luc-Ce6 (10 μM) and CTZ (100 μM).

Using the DCFDA intracellular assay, we detected considerable ROS generation in 4T1 cells after BL-PDT (10 μM Luc-Ce6, 100 μM CTZ) (Figure 3a). The ROS level was comparable to that generated by conventional PDT (1 J cm⁻²) using 4-times higher Ce6 amount (Figure 3a). The DCFDA signal increased with an increasing amount of Luc-Ce6 for a fixed amount of CTZ (Figures 3b, S19, and S20). BL-PDT resulted in near-complete cell killing confirmed by live/dead fluorescence dyes and morphological changes (Figures 3c and S21–23). Cell killing by laser-PDT occurs only in regions illuminated with sufficient laser energy (Figures 3c and S21). When we placed a layer of chicken breast tissues on top of cell samples in the optical path, simulating deep tissue therapy, no phototoxicity was observed, even with a modest tissue thickness of 300 μm (Figure S24). Using the CCK-8 assay, we compared the dose-dependent cytotoxicity (Figs. 3d and 3e). The half-maximal inhibitory concentration (IC₅₀) values of Luc-Ce6 in BL-PDT were 5.56 μM for 4T1 cells and 1.51 μM for MDA-MB-231 cells. By comparison, IC₅₀ values of Ce6 in laser-PDT were 20.8 μM for 4T1 and 5.39 μM for MDA-MB-231 cells, respectively. Moreover, we used an Annexin V-FITC/PI apoptosis detection kit and flow cytometry to confirm the apoptotic death of 4T1 cells 24 hours after BL-PDT and laser-PDT (Figure S26). In general, BL-PDT showed a considerable therapeutic effect in vitro.

Figure 3. In vitro assessment of Luc-Ce6/Fuso-lip-based BL-PDT.

(a) Confocal microscopy images of intracellular ROS-sensing CDFDA (green) in 4T1 cells after BL-PDT (Luc-Ce6 10 μM and CTZ 100 μM) and laser-PDT (Ce6 and 405 nm laser, 1 J cm⁻²). (b) Measured DCFDA intensity for BL-PDT and laser-PDT (1 J cm⁻²) for various concentrations of Luc-Ce6 and Ce6. (c) Fluorescence images of 4T1 cells stained with Calcein AM (live cells, green) and BOBO-3 Iodide (dead cells, red). Scale bars, 50 μm. (d-e) Relative viability of 4T1 (d) and MDA-MB-231 cells (e) after BL-PDT and laser-PDT for different concentrations of Luc-Ce6 and Ce6. The horizontal grey band indicates the half-maximal inhibitory concentration (IC₅₀) levels.

3.4. In vivo BL-PDT for TNBC tumors at the advanced stage

For preclinical testing, we first established an orthotopic model of metastatic TNBC using ~10⁶ 4T1 cells in the mammary fat pad of a BALB/c mouse. The implanted tumor grew rapidly, reaching 4–5 mm in size in 1 week and 9–12 mm in 3 weeks (Figure S27). Tumor invasion into surrounding tissues was observed on Day 14 (Figure S27c). Consistently, no apparent metastasis was found in the sentinel lymph nodes (SLN) and lungs on Day 7. However, on Day 14, small metastases were detected in the SLNs and lungs (Fig. S28 and Figure S29). On Day 21, multiple metastatic foci were evident throughout the lungs. For this result, we estimated that early metastasis occurs between 1 and 2 weeks.

We test the effect of BL PDT on relatively large-size tumors (7–8 mm). On Day 14, we injected Luc-Ce6/Fuso-lip into primary tumors, waited for 6 h for the Luc-Ce6 to diffuse into the tumor cells, and intravenously injected the first dose of CTZ (100 μM, 50 μl). Then, 5 additional injections of CTZ with the same dose were given in the subsequent 20 h (Figure 4a). The tumors were harvested 3 days after the treatment. Their frozen tissue sections were examined using a terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay to detect cell apoptosis. The BL-PDT treated groups showed strong TUNEL signals over larger areas and central locations (Figure 4b). PBS-treated and Luc-Ce6 sham-treated tumors had no obvious TUNEL signals (Figures 4c and S30). Laser-PDT treated groups, which had received the same intratumoral injection of Luc-Ce6/Fuso-lip and treated with a 405-nm laser (90 J cm⁻²), showed TUNEL signals only in superficial regions along the tumor boundary within < 0.5 mm depth from the surface (Figure 4e). This result showed the advantage of BL-PDT over laser-PDT in terms of therapeutic depth.

Figure 4.

In vivo assessment of BL-PDT in a 4T1 TNBC orthotopic model. (a) Schematic of the treatment schedule for 4T1 tumors at the advanced stage. (b-d) H&E and immunofluorescence (IF) images of tumor sections after BL-PDT treatment (b), laser-PDT (c, 405 nm, 90 J cm⁻²), and PBS treatment (d). Green: TUNEL, blue: DAPI for nuclei. (e) Principle of neoadjuvant BL-PDT. (f) TUNEL, H&E, and IF-stained tumor sections, including surrounding normal tissues, 5 days after BL-PDT and sham (PBS) treatment.

Neoadjuvant chemotherapy and radiation therapy are often performed before surgery to create conditions for better surgical resection for TNBC in the clinic [31–32]. We tested the neoadjuvant potential of BL-PDT for invasive tumors at the advanced stage (Figure 4a). We injected Luc-Ce6/Fuso-lip at the root of the tumor around its boundary on Day 14 (Figure 4e) and activated them in the tumor boundary by intravascularly administered CTZ (Figure S31). 5 days after the BL-PDT, the tumors were collected for histological analysis. Most tumor cells in the border appeared round, atrophic, and apoptotic (Figures 4f and S32). This morphological change visualized the margin more vividly than untreated tumors in control animals (Figure 4d), in which the tumor invasion continued progressing. A laser-PDT-treated group also showed blurred tumor boundaries due to limited penetration depth of light (Figure S33). Our result shows that neoadjuvant BL-PDT can act on deep tissue, shrink the tumor size, and delineate the tumor margin.

3.5. In vivo BL-PDT results in tumor regression and prevents metastasis

Having confirmed the cell-killing effect in vivo, we tested BL-PDT on tumors in their early stage of development. Figure 5a depicts the treatment schedule. BL-PDT was applied on tumors 7 days after 4T1 cell inoculation, a time point when no metastasis is expected to have occurred. For BL-PDT-treated mice (n=5), we measured the shrinkage of tumor volume within 2 days after treatment. Complete tumor regression was achieved by Day 18 without relapse (Figure 5c). The body weights of the treated mice were normal (Figure 5d). In stark contrast, laser-PDT-treated mice showed nearly unaffected exponential growth curves (Figures 5b and 5c). Various organs were harvested on Day 25. Histology of tissue sections from the heart, liver, kidney, and spleen showed no sign of tissue damage (Figure S34) nor metastasis in SLNs and lungs (Figures 5f–g and S35). On Day 25, all the control, sham, and laser-PDT-treated mice showed substantial metastasis to the SLN and lung (Figures 5e–g). Our result showed that BL-PDT could eradicate early-stage tumors and prevent metastasis in mice.

Figure 5.

The effects of BL-PDT on early-stage 4T1 tumor. (a) Treatment schedule. (b) Growth curves of tumors in different groups: PBS injection only, Luc-Ce6 injection only, Luc-Ce6 and laser activation (laserPDT, 405 nm, 90 J cm⁻²), and Luc-Ce6 and substrate CTZ activation (BL-PDT). Error bars are the standard deviation of 5 animals per group. ***, P < 0.001; n.s.: not significant. (c) Photos of tumors harvested from different groups at different times. (d) Body weight. (e) Illustration of BL-PDT inducing apoptosis of the primary tumor and preventing metastasis to the lung and sentinel lymph node. (f-g) H&E histology and immunofluorescence (IF)-stained sentinel lymph nodes (f) and lungs (g) harvested from tumor-bearing mice at 18 days: healthy mice, PBS injection only (sham), laser-PDT, and BL-PDT. Blue: DAPI, yellow: Ki-67 (nucleus), red: EGFR (cell membrane). While metastasis is apparent in the lymph nodes and lungs in the PBS and laser-PDT groups, no evident metastasis was found in the control and BL-PDT groups.

4. Discussion

The optimized Luc-Ce6 conjugates with high efficiency of utilizing self-luminescent biophotons and membrane-fusion liposomal delivery enabled us to demonstrate highly effective BL-PDT for in vivo treatments of TNBC. The molecularly activatable Luc-Ce6 conjugates enable depth-independent PDT, a significant advantage over conventional light-activatable photosensitizers. In addition, the all-natural-material reagents have no toxicity concerns. Therefore, they are more suitable for broad in vivo applications or clinical transformation than chemical luminal-based CL-PDT, typical inorganic nanoparticle-based BL-PDT, and Cerenkov radiation-induced PDT. Additionally, the inherent highly therapeutic performance of Luc-Ce6 conjugate, owing to the adequate use of bio-photons, enables a more substantial therapeutic effect in vivo, compared with recently synthetic protein-based BL-PDT. Furthermore, the triggerable action of CTZ substrate allows BL-PDT to be performed at a desired time when the reagents have been delivered to target regions for the best therapeutic outcome. It also helps to achieve low background cytotoxicity, an advantage over non-activable drugs like conventional chemotherapy agents.

The ROS-mediated mechanism makes PDT a unique modality different from chemotherapies, radiation therapies, and immunotherapies [33–34]. Our results demonstrated the potential of BL-PDT as a standalone, complementary, or synergistic treatment option for cancer treatment used like a typical pharmaceutic agent. In this study, we have investigated intratumoral delivery by local injection of Luc-Ce6/Fuso-lip. Although this method could be viable for treating small tumors or neo-adjuvant treatment of larger tumors, as demonstrated, it will be worthwhile to develop an excellent systemic delivery method of BRET conjugates for broader applications such as treating metastases because the therapeutic effect of BL-PDT dramatically relies on the amount of Luc-Ce6 aggregation [35–36]. Exploring different all-natural luciferases, luciferins, and photosensitizers for different target tissues and organelles will also be interesting. In addition, BL-PDT can apply to treat non-cancers, such as deep skin lesions. Finally, the BRET strategy may be extended to other phototherapies, such as antibacterial blue light therapies that use endogenous photosensitizers [37–38]. Our promising results encourage further developments toward clinical BRET-based phototherapies.

Data availability

The data supporting this paper’s figures are available upon reasonable request from the authors.