Phosphorescent Metal Halide Nanoclusters for Tunable Photodynamic Therapy

Abstract

Photodynamic therapy (PDT) is currently limited by the inability of photosensitizers (PSs) to enter cancer cells and generate sufficient reactive oxygen species. Utilizing phosphorescent triplet states of novel PSs to generate singlet oxygen offers exciting possibilities for PDT. Here, we report phosphorescent octahedral molybdenum (Mo)‐based nanoclusters (NC) with tunable toxicity for PDT of cancer cells without use of rare or toxic elements. Upon irradiation with blue light, these molecules are excited to their singlet state and then undergo intersystem crossing to their triplet state. These NCs display surprising tunability between their cellular cytotoxicity and phototoxicity by modulating the apical halide ligand with a series of short chain fatty acids from trifluoroacetate to heptafluorobutyrate. The NCs are effective in PDT against breast, skin, pancreas, and colon cancer cells as well as their highly metastatic derivatives, demonstrating the robustness of these NCs in treating a wide variety of aggressive cancer cells. Furthermore, these NCs are internalized by cancer cells, remain in the lysosome, and can be modulated by the apical ligand to produce singlet oxygen. Thus, (Mo)‐based nanoclusters are an excellent platform for optimizing PSs. Our results highlight the profound impact of molecular nanocluster chemistry in PDT applications.

Phosphorescent metal halide nanoclusters have been described for tunable photodynamic therapy. Modulating the short chain fatty acid length on the apical halide ligand of phosphorescent octahedral molybdenum‐based nanoclusters facilitates cellular uptake and improves phototoxicity in photodynamic therapy (PDT). Following photoexcitation, the nanoclusters produce singlet oxygen to cause cancer cell death. Image created with BioRender.com.

Introduction

Photodynamic therapy (PDT) has emerged as a promising treatment strategy for many cancers. ^[1] PDT combines a photosensitizer (PS) with light, providing dual selectivity with minimal side effects from either the PS or light alone. This approach for cancer treatment is a favorable alternative to traditional chemotherapy, radiotherapy, and surgical intervention due to its potential for minimal invasiveness, low systemic toxicity, and the absence of initial resistance.[ ² , ³ , ⁴ ] Ideally, a PS remains inactive until activated by a specific wavelength or intensity of photon flux that causes it to react with its environment. Most often, PSs are activated in the presence of oxygen to produce reactive oxygen species (ROS) through two types of oxidative photoreactions. Type I reaction is a photo‐induced electron‐transfer between the PS radical anion (PS⋅⁻) and the triplet ground state oxygen (³O₂) to produce superoxide radical anion (O₂⋅⁻), which can dismutate and form hydrogen peroxide (H₂O₂) and hydroxyl radical (HO⋅⁻). ^[5] Type II reaction is when a PS, typically in an excited triplet state, transfers the absorbed photon energy directly to oxygen to form an excited singlet oxygen (¹O₂). ^[6] Depending on the amount of ROS generated, a cytotoxic effect can cause cellular destruction, as ROS can interact with a wide range of biomolecules and cause damage such as DNA fragmentation and mitochondrial disruption. ^[7] Oxidative damage can trigger cell death through multiple signaling cascades and gene expression changes. A major advantage of PDT is the localized effect caused by the high reactivity and short half‐life of singlet oxygen and the hydroxyl radicals, causing cellular damage only in targeted areas[ ⁸ , ⁹ ] To date, PDT has been FDA approved for treatment of esophageal, esophagus, lung, cutaneous T‐cell lymphoma, and skin cancers, as well as for actinic keratosis, a skin precancer condition.[ ¹⁰ , ¹¹ ] The next generation of PSs are currently being developed for improved PDT.[ ¹⁰ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ ]

An ideal PS has high phototoxicity (toxicity with light activation) and low cytotoxicity (toxicity in the dark) for effective tumor treatment during PDT with minimal side effects. ^[10] We recently demonstrated a platform for controlling the cytotoxicity and phototoxicity of fluorescent organic salts, which generate mitochondrial superoxide, through pairing a photoactive cation with various dipole‐modulating counterions for improved PDT. ^[16] Here, we expand on this work to control the cytotoxicity and phototoxicity of highly emissive phosphorescent octahedral cluster complexes that generate singlet oxygen, opening the possibility for PDT in various cancer types. The hexanuclear molybdenum (Mo)‐based transition metal halide nanocluster (NC) salts have recently emerged as alternative PS candidates for PDT due to their desirable photo‐physical and photo‐chemical properties, lack of toxic elements (such as Pb or Cd), and high degree of chemical stability.[ ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ ] Molybdenum dichloride was first synthesized in 1859 when Blomstrand prepared low‐valence molybdenum halides by passing hydrogen over pentachloride in a hot tube. ^[29] Early on, it was considered to be a trimer; however, the first definite structural characterization was established in 1950 by Vaughan and Pauling in a crystal structure composed of the Mo₆C1₁₄ ²⁻ ion with Mo−Mo octahedron surrounded by eight face‐bridging and six axial halides.[ ³⁰ , ³¹ , ³² , ³³ , ³⁴ ] Later, in the 1980s, studies using electrogenerated chemiluminescence provided insights for application of inorganic systems in light‐emitting devices and electron‐transfer reactions in highly exothermic regimes.[ ³⁵ , ³⁶ , ³⁷ ] More recently, these molecules were used as sensitizers for ¹O₂ generation and oxygen detection in liquids.[ ³⁸ , ³⁹ ] In addition, we demonstrated these highly phosphorescent metal halide nanocluster complexes (A₂Mo₆X₈X^a ₆) in light emitting diodes ^[40] and luminescent solar concentrators ^[41] due to their exceptional electroluminescent and photoluminescent properties.

In general, metal halide NCs have the formula A₂Mo₆X₈X^a ₆ with an octahedron M₆ (M: Mo, W, etc.) associated with eight tightly bonded halide ions (X: Cl⁻, Br⁻, or I⁻) at inner positions that form the cubic core, the other six halide ligands at apical positions (X^a) can be readily exchanged with a variety of ligand species (monoanionic inorganic or organic) to modify the corresponding photoactive properties[ ⁴² , ⁴³ ] and, two terminal cations (A: K⁺, Cs⁺, or Tetrabutylammonium ‐ TBA⁺) that impact properties such as solubility or non‐radiative decay without altering the NC spectra.[ ⁴⁴ , ⁴⁵ ] The photophysical properties of Mo₆X₈X^a ₆ ²⁻ nanocluster complexes can be effectively tuned by both composition of the tightly bonded halide (X) and substitution of apical ligands (X^a). For inner X positions, as the halide ions varies from Cl⁻ to Br⁻ to I⁻, the change in corresponding electronic transition results in a redshift of the absorption onset (ultraviolet to blue) and blueshift of the emission profile (near infrared to red). ^[46] Ligand exchange at X^a positions exhibits a major impact on radiative/non‐radiative recombination rates, which produces a significant change in the corresponding phosphorescence quantum yield and ¹O₂ production.[ ⁴⁷ , ⁴⁸ ] ¹O₂ production is advantageous for PDT as 1) biological systems have not evolved a specialized antioxidant system to eliminate ¹O₂ production, 2) ¹O₂ is often more reactive than other ROS species such as O₂⋅⁻, 3) the energy transfer to O₂ occurs at a higher rate (k≈1–3×10⁹ M⁻¹ s¹) when compared to the electron transfer (e.g., to produce O₂⋅⁻, estimated as k≤1×10⁷ M⁻¹ s⁻¹), and 4) the energy transfer to O₂ is capable of cleaving most C−C bonds, causing more damage to malignant cells.[ ⁴⁹ , ⁵⁰ ] Therefore, both halide composition engineering and ligand exchange in NCs can be utilized to achieve improved PSs with desirable antitumor effects and ideal characteristics for PDT applications.

Previous work on Mo‐based halide nanoclusters had limited ligand variation and used harsh UV‐light activation in cervical carcinoma derived HELA cells[ ¹⁸ , ²¹ , ²⁵ , ⁵¹ ] or ovarian carcinoma derived A2780 cells.[ ¹⁹ , ²⁴ ] To improve Mo‐based halide nanoclusters as PSs for PDT, it is necessary to systematically investigate ligand effects in a greater range of cancer lines with safe blue light activation for controllable tunability of both cyto‐ and photo‐toxicity. Additionally, PS candidates should also be characterized for cell internalization and specificity for malignant cells.[ ¹⁸ , ²¹ , ²² , ²⁶ ] Here, we systematically studied the impact of three terminating short chain length fatty acid ligand species (CF₂)_n, trifluoroacetate (CF₃COO), pentafluoropropionate (CF₃CF₂COO), and heptafluorobutyrate (CF₃CF₂CF₂COO) on Mo‐based halide NCs (Cs₂Mo₆I₈I^a ₆) as photosensitizers for PDT. We tested 11 cell lines including human non‐transformed cells as well as breast, colon, skin, and pancreatic cancer cells and their highly metastatic derivatives previously selected in immunocompromised mice for higher metastatic capacity.[ ⁵² , ⁵³ ] Parental cancer cell lines include MDA‐MB‐231 (breast cancer), HCC1806 (breast cancer), SW480 (colon cancer), A375 (skin cancer), and BxPC3 (pancreatic cancer), representing some of today's most common and deadly cancer types. ^[54] We tested in the following parental poorly metastatic vs. highly metastatic derivative of each cell line, respectively: MDA‐MB‐231 vs. MDA‐LM2, HCC1806 vs. HCC‐LM2, SW480 vs. SW480‐LvM2, A375 vs. A375‐LM3, and BxPC3 vs. BxPC3‐LvM2.[ ⁵² , ⁵³ , ⁵⁵ ] Based in our previous work ^[40] showing promising optical chemical characterization of the phosphorescent octahedral molybdenum iodide nanoclusters, we aimed to: 1) test the impact of NC ligand length on PS performance during PDT of parental poorly metastatic cancer cells as well as their highly metastatic derivative cells, 2) determine the cellular localization of NCs, and 3) uncover the effect of NC ligand length on the type of ROS generated in cancer cells. We also tested in non‐transformed MCF10 A cells to verify malignant cell specificity of NCs. We found that ligand exchange in NCs allows tunability of cytotoxicity and phototoxicity by dictating the type of ROS produced to enable enhanced PDT in cancer cells. Furthermore, our NCs display effective phototoxic efficacy against highly metastatic cells. This work demonstrates precise control of NC toxicity via ligand exchange and paves the way for enhanced cancer therapy with minimal side effects.

Results and Discussion

Mo‐based nanocluster formulations and cellular uptake

We synthesized Mo‐based NCs with the trifluoroacetate ligand, as it has been shown to significantly increase the phosphorescent quantum yield in iodide‐based NCs and form high rates of singlet‐oxygen. ^[56] Mo‐based NCs were made with apical halides with three different ligands that contain 2, 3 or 4 carbons, triflouroacetate (CF₃COO), pentafluoropropionate (CF₃CF₂COO), and heptafluorobutyrate (CF₃CF₂CF₂COO), to generate: Mo₆I₈(CF₃COO)₆ ²⁻, referred to as C2; Mo₆I₈(CF₃CF₂COO)₆ ²⁻, referred to as C3, and Mo₆I₈(CF₃CF₂CF₂COO)₆ ²⁻, referred to as C4 (Figure 1A). When these NCs are optically excited by blue light in an oxygen‐free condition, radiative recombination takes place and results in phosphorescence with near‐unity photoluminescence quantum yield: 75±5 %, 80±5 %, and 70±5 % for C2, C3 and C4 NCs, respectively. The corresponding absorption and phosphorescent emission spectra are shown in Figure 1B. Alternatively, the Mo‐based NCs also exhibit photosensitization due to the release of singlet oxygen when stimulated in the absorption band between 300–500 nm (UV and blue light) in the presence of oxygen. ^[57] The phosphorescence and singlet oxygen formation are two reversible processes, which shows a strong oxygen dependence as shown in Figure S1. Blue light is commonly used in FDA‐approved PDT for skin cancer and pre‐cancerous lesions[ ⁵⁸ , ⁵⁹ ] as well as colorectal ^[60] and esophageal cancers.[ ⁶¹ , ⁶² ] The broad emission spectrum of Mo‐based NCs in the NIR region (650 ‐ 1200 nm) is in a spectroscopic window with minimal absorbance by biomolecules, reducing background noise. ^[63] These desirable photochemical properties are preferred for PSs used in PDT of cancer cells (Figure 1C). ^[3] Additional chemical structures and mass spectra of the Mo₆I₈I^a ₆ ²⁻ core with the 3 different ligands are shown in Figure S2. The ultraviolet visible spectroscopy (UV‐Vis) absorption profiles of the NCs are shown in Figure S3, which show no peak shift in an aqueous medium or dimethyl sulfoxide (DMSO) for at least 24 h. This indicates high stability of NCs in both environments (Figure S3A).

Figure 1.

Phosphorescent octahedral molybdenum (Mo)‐based nanoclusters structures as photosensitizers for photodynamic therapy. A) Ball‐and‐stick schematics of the nanocluster (NC) core (Mo₆I₈I^a ₆ ²⁻) and three ligand structures (C2: CF₃COO−Ag, C3: CF₃CF₂COO−Ag and C4: CF₃CF₂CF₂COO−Ag) involved in this study. Note: I^a represents six apical iodide of the NC core, which can be exchanged with various ligands (blue shaded area). B) Absorption and emission spectra of the nanoclusters. C) Schematic showing cytotoxicity versus phototoxicity. The tunability between cytotoxicity and phototoxicity is enabled through ligand exchange.

To confirm that the series of Mo‐based NCs are taken up by cancer cells, we treated human breast squamous carcinoma HCC1806 cells with 5 μM of NCs and visualized NC internalization by luminescence microscopy. After 24 h, the phosphorescent signal can be detected next to the nuclei of cells treated by all three NCs, indicating that C2, C3, and C4 are able to penetrate the cell membrane (Figure 2). To established that these NCs are taken by other cancer cell types, we used additional breast (MDA‐MB‐231), pancreatic (BxPC3), colon (SW480), skin (A375) cancer cells (Figure S4). Unlike previous work with nanoparticle formulations,[ ¹⁸ , ⁴⁸ ] C2, C3 and C4 are unencapsulated salt molecules and are internalized by all tested cancer cell lines and remained within cells for at least 24 h. We further tested the NCs in the non‐transformed cell line MCF10 A and found that C4 is not taken up by MCF10 A cells, indicating specificity of C4 for cancer cells (Figure S4).

Figure 2.

Uptake of nanoclusters (NCs) by HCC1806 cells. Phosphorescent images following incubation for 24 h with C2 (CF₃COO), C3 (CF₃CF₂COO) or C4 (CF₃CF₂CF₂COO). Cellular nuclei were stained in blue, using 2′‐[4‐ethoxyphenyl]−5‐[4‐methyl‐1‐piperazinyl]−2,5′‐bi‐1H‐benzimidazole trihydrochloride trihydrate (Hoechst). Images were taken in bright field (BF) and luminescence to validate cellular uptake of nanoclusters. Cells plated without nanocluster exposure were used as control cells (CTRL). Scale bar: 30 μm.

Ligand chain length modulates toxicity of Mo‐based nanoclusters

To understand the impact of ligand chain length of Mo‐based NCs on cellular toxicity, we conducted cell viability experiments. We first performed light dosimetry studies by treating cancer cells with blue light only (460 nm) for varying amounts of time to determine non‐toxic light doses for PDT studies. We found that treatment with 460 nm for 5 minutes did not decrease cell viability in HCC1806, HCC‐LM2, A375 and A375‐LM3 cells. For sensitive cell lines MDA‐MB‐231, MDA‐LM2, BxPC3, BxPC3‐LvM2, SW40, and SW480‐LvM2 cells, for which 5 minutes of light irradiation statistically changed cell viability in comparison with the control (no light irradiation), the chosen treatment was 460 nm for 2 minutes (Figure S5). We also tested treatment with UV light (395 nm); in contrast to previous work, UV irradiation alone for only 1 minute decreased cell viability by≥20 % in all cell lines tested (Figures S6 and S7).[ ¹⁹ , ²⁴ ] We first focused on the parental HCC1806 breast cancer cell line and the highly metastatic derivative HCC1806 LM2 cell line due to their high blue light resistance in comparison with the other cell lines tested (Figure S5).[ ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ ] Cellular viability was determined by comparing the cell viability between day 1 and day 4 following PDT treatment with varying concentrations of C2, C3 and C4. Similar results were observed for both poorly and highly metastatic cell lines, showing promise for treatment of aggressive cancers (Figures 3 and S8). C2 was highly cytotoxic with increasing C2 concentrations independent of blue light irradiation with little difference in the dark and light irradiated IC50 (1.3 μM and 0.65 μM for HCC1806, 0.58 μM and 0.33 μM for HCC1806 LM2 cells in dark and light irradiated IC50s, respectively) (Figure S8). C3 had high phototoxicity with low cytotoxicity at low concentrations (<5 μM); however, C3 became highly cytotoxic at a higher concentration of 15 μM with a dark IC50 3.5 times higher than light irradiated IC50 (2.34 μM and 0.70 μM for HCC1806, 2.14 μM and 0.61 μM for HCC1806 LM2 cells in dark and light irradiated IC50s, respectively; Figure S8). C4 displayed desirable characteristics as a PS with no cytotoxicity even for high concentrations up to 15 μM, with dark IC50 of 26.94 and 58.99 μM for HCC1806 and HCC1806 LM2, respectively, and potent phototoxicity with more than tenfold decrease in light irradiated IC50 of 0.24 μM and 0.20 μM for HCC1806 and HCC1806 LM2, respectively (Figure S8).

Figure 3.

Dose response to treatment is dictated by the ligand length of nanoclusters. Toxicity of C2 (CF₃COO), C3 (CF₃CF₂COO) and C4 (CF₃CF₂CF₂COO) was determined in breast cancer cell lines. HCC1806 cancer cells were incubated with various concentrations of NCs for 24 h then irradiated with or without 460 nm blue light for 5 minutes daily for 4 days. On day 4, cell viability was determined by trypan blue assay. Data are displayed as means ± S.E.M., *Statistical comparison between dark and 460 nm exposure groups. n=3. (*** p‐value≤0.001).

We further performed PDT on ten additional cell lines to verify efficacy in other cancer types. The cells were treated with NCs and 2 minutes of 460 nm blue light for MDA‐MB‐231, MDA‐LM2, BxPC3, BxPC3‐LvM2, SW40, SW480‐LvM2 cells, or 5 minutes of 460 nm blue light for HCC1806, HCC‐LM2, A375 and A375‐LM3 cells, according to the blue light tolerance of each cell line (Figure S5). C2 displayed the highest cytotoxicity, as cell viability decreased even without exposure to blue light in all cancer cell lines, including the highly metastatic lines. C3 displayed relatively high cytotoxicity in breast, pancreas, and colon cancer cells (Figures 4A–D); however, it had low cytotoxicity combined with high phototoxicity in both skin cancer cell lines A375 and A375 LM3 (Figure 4E). C4 had excellent phototoxicity with no significant cytotoxicity in many cell lines except in colon cancer cell lines (SW480 and SW480 LvM2), which were resistant to C4 phototoxicity (Figure 4D).

Figure 4.

Ligand length determines the toxicity of nanoclusters and specificity in cell lines. Relative viabilities of ten cancer cells lines after treatment with nanoclusters with or without 460 nm blue light irradiation. Cell viability was determined on day 4 by trypan blue staining and cell counting. Parental (above) and highly metastatic (below) cell lines (A) HCC1806, (B) MDA‐MD‐231, (C) BxPC3, (D) SW480, (E) A375 were incubated with 5 μM of indicated nanoclusters for 24 h and exposed to the dark or 460 nm blue light for 2–5 minutes, depending on the tolerance of each cell line (Figure S5). Data are displayed as means ± S.E.M., *Statistical comparison between the treatment and control groups (CTRL dark), and between dark and 460 nm exposure groups. n=3. (* p‐value≤0.05, **p‐value≤0.01, ***p‐value≤0.001).

The toxicity of 460 nm blue light and Mo‐based NCs was also tested on the non‐tumorigenic cell line MCF10 A. C2 and C3 at 5 μM were highly phototoxic following 5 minutes of blue light activation (Figure S10A). In contrast to cancer cells, C4 had no significant effect on these non‐tumorigenic cells (Figure S10B). Taken together, these data demonstrate that the cyto‐ and photo‐toxicities of Mo‐based NCs can be precisely tuned via ligand chain length and their potential for treating highly metastatic cancers without impacting non‐tumorigenic cells.

Subcellular localization and mechanism of toxicity

To investigate subcellular localization of Mo‐based NCs, we investigated the NC localization within specific organelles in the cell. HCC1806 breast cancer cells were incubated with C2, C3, or C4 for 24 h and stained with three different organelle‐specific fluorophores, LysoTracker, ER‐Tracker or RHO123, to determine NC localization in lysosomes, endoplasmic reticulum, or mitochondria, respectively. As shown in Figure 5A, most of the NCs accumulate in lysosomes (Pearson correlation coefficient: C2=0.74, C3=0.61 and C4=0.60). Lysosomes are membrane‐enclosed compartments filled with hydrolytic enzymes that accumulate material from the extracellular milieu, internalized by the cell, as well as from the intracellular material that have been taken up for degradation. ^[72] Lysosomes behave as key organelles to maintain cell homeostasis, and even low levels of ¹O₂ production can trigger cell death.[ ⁷³ , ⁷⁴ ] Further, the mitochondrial photodamage triggered by lysosomal ROS damage can increase cell death signaling pathways.[ ⁷⁵ , ⁷⁶ ] Importantly, photodamage toward lysosomes causes lysosomal membrane permeabilization and leads to proton and hydrolase leakage, resulting in cell death.[ ⁷⁷ , ⁷⁸ ] We observe moderate accumulation of NCs in the endoplasmic reticulum (Pearson correlation coefficient: C2=0.51, C3=0.48 and C4=0.45), likely due to internalization pathways leading to endosomal escape to deliver NC‐filled endosomes to the endoplasmic reticulum (Figure 5B).[ ⁷⁹ , ⁸⁰ ] No accumulation of NCs is observed in the mitochondria (Figure 5C).

Figure 5.

The nanoclusters accumulate mainly in cell lysosomes. Colocalization assay was performed in HCC1806 cells pre‐incubated with nanocluster at 5 μM (24 h) and stained with (A) Lysotracker Green, (B) ER‐Tracker or (C) RHO123 according to the manufacture instructions. Cellular nuclei were stained in blue, using 2′‐[4‐ethoxyphenyl]−5‐[4‐methyl‐1‐piperazinyl]−2,5′‐bi‐1H‐benzimidazole trihydrochloride trihydrate (Hoechst). Cells plated without nanocluster exposure were considered as control cells (CTRL). Scale bar: 30 μm.

Next, we performed fluorescence studies to determine specific ROS produced by the NCs using ROS sensitive probes. Chloromethyl‐2′, 7′‐dichlorodihydrofuorescein diacetate (CM‐H₂DCFDA) was used to analyze general cytoplasmic ROS levels, MitoSOX was used to measure mitochondrial superoxide, and Singlet Oxygen Sensor Green (SOSG) was used as a singlet oxygen probe. Using general cytoplasmic ROS probe CM‐H₂DCFDA, we observed for C2 almost 2‐fold higher fluorescence than the positive control H₂O₂. C3 produces as high levels of cytoplasmic ROS as the positive control when treated with blue light; however, C4 general cytoplasmic ROS production is similar to the negative control (CTRL) with or without light activation (Figure 6A and 6B). Mitochondrial superoxide was not detected in any of the NCs (Figure 6A and 6 C), consistent with the absence of NCs in the mitochondria (Figure 5C). Singlet oxygen production with C4 was approximatively 3‐fold higher when compared to the negative control, with similar production of ¹O₂ as the positive control rose bengal (RB) when cells were exposed to light (Figure 6A and 6D). Similar levels of ¹O₂ were observed for the negative control, C2 and C3 with and without blue light irradiation.

Figure 6.

Nanoclusters kill cancer cells through ROS production. (A) Representative luminescent images of ROS production assay in HCC1806 cancer cells treated with 5 μM of C2 (CF₃COO), C3 (CF₃CF₂COO) and C4 (CF₃CF₂CF₂COO) (24 h) just after PDT treatment (460 nm/2 minutes). The cells were stained with CM‐H₂DCFDA, MitoSOX and SOSG, according to manufacturer instructions, to detect General cytoplasmic ROS, Mitochondrial superoxide and Singlet Oxygen production respectively. Note: Positive controls (Positive CTRL) used H₂O₂ at 500 μM for cytoplasmic ROS production, Antimycin (Anti) at 100 μM for mitochondrial superoxide and Rose Bengal (RB) at 80 μM for singlet oxygen production. Cells plated without nanocluster were consider as a control (CTRL). The quantitative analysis of (B) General Cytoplasmic ROS, (C) Mitochondrial superoxide, and (D) Singlet Oxygen production was determined from the fluorescent images. Data are displayed as means±S.E.M., *Statistical comparison between dark and 460 nm exposure groups. n=3. (**p‐value≤0.01, ***p‐value≤0.001). Scale bar=30 μm (100x).

It has been established that the phosphorescence of Mo‐based NCs is quenched by the ground triplet state of O₂, thus generating singlet oxygen. ^[81] Consistently, our work shows that Mo‐based NCs produce singlet oxygen, but with differing amounts modulated by the apical halide ligand of the octahedral molybdenum nanocluster. C2 produces the lowest levels of singlet oxygen and the highest levels of general cytoplasmic ROS, which includes species such as superoxide, hydroxyl radical, peroxides, hydroperoxides, and singlet oxygen in the cytoplasm. C3 produces an intermediate level of singlet oxygen and general cytoplasmic ROS. Interestingly, C2 and C3 are the NCs with the lower and intermediate phototoxic effect, respectively, in different cancer cell types. Conversely, C4 produces the highest level of singlet oxygen and lowest levels of general cytoplasmic ROS. ROS modulation by ligand chain length is consistent with previous work that showed changes in ROS production when photosensitizers are conjugated with different fatty acids, which impact photosensitizer release into the cytoplasm, degree of lipid peroxidation, and phototoxicity upon irradiation.[ ⁸² , ⁸³ , ⁸⁴ ] Together, these data indicate the tunability of Mo‐based NCs, which can be modulated for ¹O₂ production for more efficient PDT.

Based on current and previous work, the ligand chain length of the NCs likely tune cellular toxicity by facilitating the cellular internalization of these molecules.[ ⁷² , ⁸⁵ ] The NCs accumulate mostly in the lysosomes, with C2 having the highest affinity for the lysosome with a Pearson correlation coefficient (PCC) of 0.74, followed by C3 with a PCC of 0.61, and finally C4 with the lowest PCC of 0.60 (Figure 5A). This data indicates that NCs are taken up and accumulate mainly in the lysosome, the final organelles in endocytic pathways. Previous works have established that fatty acids are taken up by endocytosis (Figure 7).[ ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ ] During endocytosis, the NCs are surrounded by an endocytic vesicle, which fuse with early endosomes that mature and become late endosomes. Simultaneously, the Golgi apparatus produces transport vesicles carrying acid hydrolases that fuse with the late endosome, which mature into lysosomes.[ ⁹⁰ , ⁹¹ ] Once in the lysosomes, the NCs can stimulate lysosomal membrane permeabilization (LMP), which disrupts lysosomes and leaks lysosomal content to induce cell death. ^[92] The high lysosomal affinity of C2 causes interaction in the lysosomal membrane and induces LMP, resulting in its high dark cytotoxicity without light activation (Figure 3). C3 has lower lysosomal affinity, which slightly decreases lysosomal disruption to cause less cell death by LMP, resulting in decreased dark cytotoxicity. C4 has the lowest lysosomal membrane affinity and therefore the lowest dark cytotoxicity. Once irradiated with light, all three NCs generate ROS (Figure 6), which is a known promoter of LMP.[ ⁸² , ⁹³ ] Thus, light activation activates LMP, inducing phototoxicity for all three NCs.

Figure 7.

Overview of nanocluster uptake, localization and toxicity. The short chain fatty acids ligand length facilitates NC endocytosis, by which these molecules are taken up from outside the cell in endocytic vesicles. The endocytic vesicles fuse with transport vesicles carrying hydrolases from the Golgi apparatus, which mature into lysosomes. Depending on the stimulus, the lysosomal membrane can be disrupted causing lysosomal membrane permeabilization (LMP). LMP is responsible for releasing cathepsins and other hydrolases from the lysosomal lumen to the cytosol, which culminates in cell death. C2 has highest affinity for the lysosome and activates LMP even without light activation, causing its high dark cytotoxicity. C4 has low cytotoxicity but high phototoxicity, indicating that LMP is triggered only after light irradiation. Blue light irradiation triggers ROS production to stimulate LMP, release hydrolases and ¹O₂ in the cytosol. Image created with BioRender.com.

Collectively, these results show that the modulation of Mo‐based NCs with short‐chain fatty acid ligands impact cellular internalization, ROS production, and PDT efficacy.[ ⁹⁴ , ⁹⁵ ] Our data indicate C4 to be a great PS candidate for PDT, especially in HCC1806 breast cancer cells and its highly metastatic derivative cells, as it presents phototoxicity even in low concentrations, accumulates in the lysosome and effectively generates ¹O₂ only after light irradiation without presenting cytotoxicity even at high doses. C3 also shows promise with desirable phototoxicity at low doses and should be further investigated, especially for both primary and metastatic colon cancer.

Conclusion

In summary, we have demonstrated the ability to tune the cyto‐ and photo‐toxicities of metal halide nanocluster molecules in a wide range of cancer cell lines as phosphorescent photosensitizers. Phototoxicity can be enhanced concomitantly with suppressed cytotoxicity by varying the length of apical fatty acid ligand chains of the octahedral molybdenum nanocluster. These modified nanoclusters are internalized by 10 cancer cell lines from breast, pancreatic, colon, and skin tumors. In addition, the NC displayed excellent phototoxicity in cancer cells, including highly metastatic cancer cell lines. Moreover, C4 was not internalized by the non‐tumorigenic cell line MCF10A, demonstrating specificity for cancer cells without accumulation in non‐tumorigenic cells. The tunability in ROS production was also dependent on the length of the fluorinated fatty acid ligand chain. We found that C3 and C4 are promising candidates as photosensitizing agents for PDT. At low concentrations, C3 presented high phototoxicity with light activation and no cytotoxicity in the dark in skin or colon cancer cell lines. C4 presented high phototoxicity and no cytotoxicity in most of the cell types studied here, except in colon cancer cells. Furthermore, in non‐tumorigenic cells, C4 presented desirable low cytotoxicity. We have shown that the intercellular photosensitization of Mo‐based nanoclusters is modulated by the terminal ligands and length of the fluorinated fatty acid ligand chain. Thus, our platform of tuning Mo‐based NCs is a promising alternative to improve the next generation PSs for therapeutic agents for PDT. While blue light‐activated molecules are widely used in PDT to treat superficial cancers and other diseases, the low penetration depth of blue light is a limitation for treating deeply imbedded tumors. Future work will improve blue light delivery in deep tumors and explore shifting activation wavelengths of NCs.

Experimental Section

Nanocluster synthesis: Cs₂Mo₆I₈I^a ₆: MoI₂ powder (2 A Biotech) was uniformly mixed with CsI powder (Sigma‐Aldrich) with a stoichiometric ratio of 3 : 1. The mixture was then transferred into a quartz ampule (12 cm long, 1.5 cm diameter), and the ampule was sealed under vacuum. The ampule was heated at the reaction temperature of 750 °C for 72 h to form Cs₂Mo₆I₈I^a ₆. After cooling down to room temperature, the Cs₂Mo₆I₈I^a ₆ powder in the ampule was dissolved in acetone (wine‐colored solution) and the undissolved impurity (unreacted black powder) was removed. The acetone was removed by rotary evaporation to obtain red Cs₂Mo₆I₈I^a ₆ powder. Detailed characterization has been previous described. ^[41] NC ligand exchange: Cs₂Mo₆I₈I^a ₆ was weighed and dissolved in acetone in a flask, and silver trifluoroacetate (CF₃COO−Ag), silver pentafluoropropionate (CF₃CF₂COO−Ag) or silver heptafluorobutyrate (CF₃CF₂CF₂COO−Ag) (Sigma‐Aldrich) was added into the Cs₂Mo₆I₈I^a ₆ solution with a stoichiometric ratio of 6 : 1, respectively. The reaction was kept in the dark in a nitrogen atmosphere for 48 h. After the ligand exchange reaction, the precipitated AgI was filtered out and the solution (cider‐colored) was dried by rotary evaporation to obtain orange Cs₂Mo₆I₈(CF₃COO)₆, Cs₂Mo₆I₈(CF₃CF₂COO)₆ or Cs₂Mo6I₈(CF₃CF₂CF₂COO)₆ powder. The powder was purified by silica column chromatography (20 % ethanol / 80 % acetone, gradually increasing to 100 % ethanol) to yield the pure nanocluster product. Column chromatography was performed using Silicycle 60 Å, 35–75 μm silica gel. The final purification step boosts the NC phosphorescence quantum yield by ∼10 % by eliminating the non‐radiative impurities from the reactions. All the NC products were confirmed by high resolution mass spectrometry (Xevo G2‐QTOF – Waters) (Figure S1).

Cell lines: Human cells lines HCC1806, MDA‐MB‐231, BxPC3, SW480 and A375 and their highly metastatic derivative cell lines HCC‐LM2, MDA‐LM2, BxPC‐LvM2, SW‐LvM2, and A375‐LM3 were received from professor Dr. Hani Goodarzi (University of California, San Francisco). All cells were array based miRNA profiling and in vivo selected as a poorly metastatic cancer lines, and their respective highly metastatic sub‐lines. ^[55] All cells were cultured in a 37 °C, 5 % CO₂ humidified incubator without light exposure. The cell lines MDA‐MB‐231, MDA‐LM2, A375‐par, and A375‐LM3 were cultured in DMEM (Gibco Cat. No 10–013‐CV) base medium supplemented with 4.5 g l⁻¹ glucose, 10 % Fetal Bovine Serum (FBS), 4 mM l‐glutamine, 1 mM sodium pyruvate, penicillin (100 Uml⁻¹), streptomycin (100 μgml⁻¹) (Sigma‐Aldrich F0392) and amphotericin B (100 μgml⁻¹) (Sigma‐Aldrich A2942). The cell lines HCC1806, HCC‐LM2, BxPC3, and BxPC‐LvM2 were cultured in RPMI 1640 (Gibco Cat. No 11875‐093) base medium supplemented with 10 % FBS, 2 mM l‐glutamine, penicillin (100 Uml⁻¹), streptomycin (100 μgml⁻¹) and amphotericin B (1 μgml⁻¹). The SW480 and SW‐LvM2 cell lines were cultured in McCoy's 5A modified medium supplemented with 10 % FBS, penicillin (100 Uml⁻¹), streptomycin (100 μgml⁻¹) and amphotericin B (1 μgml⁻¹). Mouse mammary cancer cells 6DT1 were cultured in DMEM base medium supplemented with 10 % FBS, 2 mM l‐glutamine, penicillin (100 Uml⁻¹), and streptomycin (100 μgml⁻¹)

Viability studies: Cells were seeded in 6‐well tissue culture plates and allowed to adhere overnight. The following day, media was aspirated and replaced with media containing each nanocluster dissolved in DMSO at indicated concentrations. Each well was irradiated with a 460 nm lamp (18650 Blue light LDE flashlight 460 nm–Amazon online store) for 2 minutes to verify ROS production or 5 minutes to test cell viability, based on cell tolerance of blue light irradiation (Figure S4). Control cells were left in a dark incubator without irradiation. Immediately after irradiation, the media was replenished with fresh media containing nanoclusters and allowed to incubate for another 24 h. The same procedure was repeated at 48 and 72 h. Viable cell number was determined at 24 h (day 1) and 96 h (day 4) using 4 % trypan blue and a Nexcelom Cellometer Auto T4 cell counter. All assays were done using three biological replicates. The fold change in cell proliferation over days of treatment was calculated using the following equation:

$${\displaystyle FoldChange={log}_2\frac{Day4viablecellcount}{Day1viablecellcount}}$$

When the treatment killed all cells in the plate, we considered it as number 1, as a minimal possible number, so the treatment was calculated using the following equation: Fold Change=log₂ 1/(Day 1 viable cell count). Relative cell viability (Figure S8) was calculated in comparison with the control (absence of the NC without irradiation), considered as 100 %. The half maximal inhibitory concentration (IC50) was calculated using GraphPad software Prism 8.0 and the following equation:

$${\displaystyle Y=\frac{100}{(1+{10}^{\left((logIC50-X\right){}^{*}HillSlope\left)\right)})}}$$

Luminescence imaging studies: For nanocluster uptake studies, cells were seeded in 3 cm tissue culture plates at a density of 50.000 cells per well in culture cell medium. The nanoclusters were added to the cells plate and incubated for 24 h at 37 °C with 5 % CO₂ until the day of imaging. For live cell imaging, the media was aspirated, and the cells were washed with phosphate buffered saline (PBS, Sigma Aldrich) three times before being imaged. Images were obtained using a Leica Dmi8 microscope with a customized set of lenses (λ_ex=350/50x nm, λ_em=650 lp nm) manufactured by Chroma Technology Corp attached to an empty filter cube, size P (1 mm) manufactured by Leica‐microsystems, with a PE4000 LED light source, DFC9000GT camera, and LAS X imaging software.

Colocalization analysis: The cells were grown on 0.5 mm coverslips placed in 3 cm tissue culture plates containing media. Following cell plating, the cells were incubated with 5 μM C2 (CF₃COO), C3 (CF₃CF₂COO) and C4 (CF₃CF₂CF₂COO) and allowed to incubate for 24 h. The cells were further each independently incubated with 3,6‐diamino‐9‐(2‐(methoxycarbonyl) phenyl chloride (Rhodamine123), LysoTracker Green DND‐26 (Invitrogen) or endoplasmic reticulum (ER) ‐Tracker Green (BODIPY FL Glibenclamide, Invitrogen), according to the manufacturer's instructions. After incubation, the cells were washed three times with PBS and the images were taken with a 63x oil‐immersion objective lens. The organelle trackers were excited and detected as recommended by the supplier. The nanoclusters were detected under a customized mounted set of lenses (λ_ex=350/50x nm, λem=650 lp nm) manufactured by Chroma Technology Corp attached to an empty filter cube, size P (1 mm) compatible with the DMi8 microscope manufactured by Leica‐microsystems.

Detection of Reactive Oxygen Species (ROS): The cells were grown on 0.5 mm coverslips placed in 3 cm tissue culture plates containing media, following cell plating, the cells were incubated with 5 μM C2 (CF₃COO), C3 (CF₃CF₂COO) and C4 (CF₃CF₂CF₂COO) and allowed to incubate for 24 h. After, the cells were washed two times with PBS and kept in media without nanoclusters for light irradiation. Immediately after irradiation, the cells were stained with chloromethyl‐2′, 7′‐dichlorodihydrofuorescein diacetate (CM‐H₂DCFDA, Invitrogen), MitoSOX (Invitrogen) or Singlet Oxygen Sensor Green (SOSG, Invitrogen) according to the manufacturer's instructions. Hydrogen peroxide (H₂O₂), Antimycin A and Rose Bengal were used as positive controls for CM‐H₂DCFDA, MitoSOX, and SOSG, respectively. In all experiments, the cells were stained with 1 μM 2′‐[4‐ethoxyphenyl]‐5‐[4‐methyl‐1‐piperazinyl]−2,5′‐bi‐1H‐benzimidazole trihydrochloride trihydrate (Hoechst 33342, Invitrogen) for 5 minutes and washed two times with PBS for imaging.

Optical Characterization: Transmittance (T(λ)) of NC solutions were measured by using a double‐beam Lambda 800 UV/VIS spectrometer in the transmission mode. A reference cuvette sample containing only the solvent was placed on the reference beam side, which allows the reflectance to be subtracted. The absorption spectrum (A(λ)) can be readily converted by following A(λ)=1‐ T(λ). The photoluminescence spectra were measured with a PTI QuantaMaster 40 spectrofluorometer with excitation at 400 nm. The photoluminescence quantum yields of NC solution were measured by using a Hamamatsu Quantaurus fluorometer.

Statistical analysis: All data are expressed as means±S.E.M. Each experiment was repeated at least three times independently. Statistical analyses were performed using Prism 8.0 software (GraphPad Software Inc., San Diego, CA, USA). Data were tested using one‐way ANOVA analysis performed with an ad hoc Tukey test, or a two‐way ANOVA analysis was performed with an ad hoc Bonferroni multiple comparison test, where appropriate. Statistical significance was defined as * p‐value<0.05, ** p‐value<0.01, *** p‐value<0.001, and **** p‐value<0.0001. For image processing and colocalization quantitation, the images were imported into the Fiji version of ImageJ (http://fiji.sc). For colocalization analysis, the Fiji plugin, designated as Coloc 2, was used to calculate the Pearson coefficient. ^[79]

Funding

This work was supported by the National Science Foundation under CAREER grant no. CBET 1845006 to S.Y.L., and the National Science Foundation under grant no. CBET 1702591 to R.R.L. S.Y.L. and H.C.D.M. were also supported by the National Cancer Institute of the National Institutes of Health under Award Number R01CA270136 and the National Institute of Environmental Health Sciences of the National Institutes of Health under Award Number R01ES030695.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Medeiros H. C. D., Yang C., Herrera C. K., Broadwater D., Ensink E., Bates M., Lunt R. R., Lunt S. Y., Chem. Eur. J. 2023, 29, e202202881.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.