Targeting of a Photosensitizer to the Mitochondrion Enhances the Potency of Photodynamic Therapy.

Photodynamic therapy (PDT) involves use of a photosensitizer, whose activation with light leads to the production of singlet oxygen (SOS), generation of reactive oxygen species (ROS), and initiation of associated cell toxicity. Because a cell's mitochondria constitute sites where oxygen levels are high, ROS can be readily produced, and apoptosis is commonly initiated. Therefore, an ideal PDT agent might be a potent photosensitizer that could naturally accumulate in mitochondria. Although a number of mitochondria-targeting moieties, including triphenylphosphine, guanidinium, and bisguanidium, have been identified, a quantitative comparison of their efficacies in targeting mitochondria has not been performed. In this study, we have prepared triphenylphosphine, guanidinium, and bisguanidium derivatives of the FDA-approved PDT agent verteporfin (Visudyne, benzoporphyrin derivative-monoacid ring A: BPD-MA) and compared their abilities to induce the intracellular perturbations common to potent PDT agents. Cellular parameters examined included subcellular localization of the verteporfin, real-time monitoring of SOS production, quantitation of reactive oxygen species (ROS) generation, analysis of mitochondria and chromatin integrity, characterization of cytoskeletal disruption and evaluation of cytochrome C release as a measure of apoptosis. An analysis of these parameters demonstrates that the triphenylphosphine derivative (0323) has better mitochondria-targeting efficacy, SOS production, and mitochondria membrane toxicity than either unmodified verteporfin or its guanidinium derivatives. Consistent with this potency, 0323 also induced the most prominent mitochondria swelling, actin depolymerization, pyknosis, and cytochrome C release. We conclude that triphenylphosphine has a better mitochondria-targeting moiety than guanidinium or bis-guanidinium and those PDT photosensitizers with improved cytotoxicities can be prepared by conjugating a mitochondria-targeting moiety to the desired photosensitizer.

Introduction

Photodynamic therapy (PDT) combines light energy, oxygen, and a light-absorbing molecule called a photosensitizer to induce an oxidative stress in the illuminated (usually neoplastic) cells.[1] Each of these components is nontoxic by itself, but, when combined, can induce a chain of reactions leading to the production of reactive oxygen species (ROS) causing cell death. The photosensitizer (PS) is a molecule that can absorb a photon that excites an outer shell electron to a singlet state (S1), which can then either return to its ground state (i.e., by fluorescence or heat emission) or undergo intersystem crossing to a triplet state,[2] which can then react with oxygen to generate free radicals (type I reaction) or singlet oxygen species (type II reaction). These reactive oxygen species then undergo a series of downstream reactions to cause oxidative damage.[3]

One of the challenges in photodynamic therapy (PDT) is to develop novel photosensitizers that can improve the therapeutic efficacy without imposing undesirable requirements for prolonged/enhanced exposure to activating light. The major parameter used to compare the potencies of different photosensitizers is singlet oxygen quantum yield (ϕΔ), which is the efficiency by which a photosensitizer uses light energy to convert molecular oxygen to singlet oxygen.[4] Photosensitizers that contain the highest ϕΔ are desirable because singlet oxygen is a powerful oxidant that can directly induce cellular toxicity. However, it has been demonstrated that a direct correlation between singlet oxygen production and cell death does not exist, but rather the toxicity of PDT depends on the subcellular localization of the PDT agent.[5] Because singlet oxygen is highly reactive and readily quenched (its lifetime is of the order of 40 ns and its diffusion radius is ∼20 nm[6]), its site of production will almost certainly be its site of oxidative damage (i.e., a mammalian cell’s diameter is of the order of ∼10–30 μm and mitochondria organelles are ∼900 nm wide).[7] The solution, if achievable, would be to target a PDT agent specifically to an organelle where SOS/ROS could be efficiently produced and the toxic effects of SOS/ROS would be sensitively experienced. Based on the reports that mitochondria contain high concentrations of oxygen[9] and that even low levels of singlet oxygen produced in the mitochondria are more toxic than large amounts produced in the nucleus,[8] it seemed logical to try to develop a photosensitizer that would target the mitochondria. Additional observations that the mitochondria are decisive regulators of apoptosis and that these organelles produce most of the cell’s energy only added motivation to this objective.[10,11] Indeed, most of the clinically approved photosensitizers including Foscan,[12] Photofrin,[13] and Visudyne[14] already partially localize to the mitochondria.

Methods to improve mitochondria targeting have included conjugation of such mitochondria-targeting moieties as guanidinium, bisguanidinium, or triphenylphosphine groups to the main compound to introduce a delocalized positive charge and increase the compound’s lipophilicity, which collectively are thought to increase a conjugate’s affinity for mitochondria. A previous study by Sibrian-Vazquez et al. showed an increased localization of porphyrin compounds to the mitochondria upon synthesizing their guanidinium and bisguanidinium derivatives.[15] Although these guanidinium and bisguanidinium were shown to localize to the mitochondria, the modified compounds were never compared to the parent compound, making it unclear whether these molecules actually increased mitochondria localization. Triphenylphosphine group is another extensively studied mitochondria-targeting moiety that has been used for the delivery of chemotherapeutics,[16−19] antioxidants,[20] imaging probes,[21] photosensitizers,[22−24] and other molecules.[25] However, to our knowledge, there has not been a comparison of these moieties in their mitochondria-targeting efficiencies.

Herein, we present a multimodal comparison of the PDT agent, verteporfin, with three of its mitochondria-targeted derivatives in a human head and neck cancer cell line. The compound verteporfin was chosen because it was already known to partially concentrate in mitochondria, which if increased upon conjugation to guanidinium, bisguanidinium, or triphenylphosphine would provide a robust test of the value of mitochondrial targeting for maximization of PDT potency. Moreover, verteporfin is an FDA-approved drug for the treatment of wet age-related macular degeneration[26,27] and has a longer absorption wavelength than the other mitochondria-targeting photosensitizers mentioned above, which is desirable to target deeper tissue.

Results and Discussion

The molecules synthesized for this study included 0317, which contains monoguanidinium group in the carboxyl terminus; 0320, which contains a cyclic bis-guanidinium group that bridges the carboxyl termini; and 0323, which contains a triphenylphosphine group on the carboxyl terminus (Figure ). As shown in Figure , the monoacid ring A of verteporfin was coupled to guanidine and (2-aminoethyl)triphenylphosphonium bromide 2-(1H-7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyl uronium hexafluorophosphate methanaminium (HATU) as a coupling agent to obtain the monoguanidinium and triphenylphosphonium derivative of verteporfin named 0317 and 0323, respectively. To achieve bis- and tris-guanidinium derivative of verteporfin for increased mitochondria-targeting, we tried the conventional methyl ester hydrolysis conditions to obtain di- and tri- free carboxylic acid derivative of verteporfin. However, our attempt with strong and mild basic conditions for methyl ester hydrolysis were unsuccessful, and we observed complex products in liquid chromatography–mass spectrometry (LCMS). Finally, we optimized using Me3SnOH, which had been established by Nicolaou and co-workers[28] to hydrolyze selectively the aliphatic methyl esters without damaging the olephenic methyl ester in the E ring of verteporfin. To obtain a tris-guanidinium derivative of verteporfin, the tri-carboxylic acid derivative was further conjugated with guanidine using HATU coupling condition. The reaction was monitored by LCMS, and the major product shows the molecular weight of 755.2, which is not that of our expected tris-guanidinium derivative 0319 (MW: 816.9). The LCMS analysis indicates that the isolated product with the molecular weight of 755.2 was the cyclic bis-guanidinium derivative 0320.

Figure 1

Synthesis of the modified verteporfin to target mitochondria. Reagents and conditions: (i) HATU, diisopropylethylamine (DIPEA), dimethyl sulfoxide (DMSO), room temperature (r.t.) 30 min then guanidine (2 equiv) stir overnight, (ii) HATU, DIPEA, DMSO, r.t. 25 min then (2-aminoethyl)triphenylphosphonium bromide (2 equiv) stir overnight, (iii) Me3SnOH (10 equiv), 1,2-DCE, 80 °C, 1 h, (iv) HATU, DIPEA, DMSO, r.t. 30 min then guanidine (6 equiv) stir overnight.

To compare these compounds, we quantified and compared their mitochondria localization, intracellular singlet oxygen production, and mitochondria membrane potential (Δψm) change. Furthermore, we studied the morphological changes of the cytoskeleton and mitochondria and reactive oxygen species production. Because the goal of PDT is to kill pathological cells, cell death was analyzed by observing nuclear pyknosis and cytochrome C release from the mitochondria. Results from this study should facilitate the discovery and optimization of novel PDT therapeutics by determining which substituent will promote the most photosensitizer accumulation in the mitochondria. Moreover, the methods employed here should help guide development of newer photosensitizers by establishing methods to compare their efficacy.

Cellular Localization of Modified Compounds

The subcellular fate of the modified compounds was determined by fluorescent microscopy. Pearson’s correlation coefficient was used to determine the localization of the photosensitizers (Figure i–l) in the mitochondria, which was labeled with Mitotracker green (Figure e–h). As shown in Figure , when the Mitotracker and 0323 fluorescence images were merged, a very strong colocalization signal was observed (Figure p). The calculated Pearson’s correlation coefficients (±SEM) were 0.722 (± 0.044) for BPD-MA, 0.678 (± 0.029) for 0317, 0.631 (± 0.029) for 0320, and 0.860 (± 0.012) for 0323. Using a one-way ANOVA with Tukey post-hoc test, the modified compound 0323 showed a significantly higher localization to the mitochondria than BPD-MA (p-value = 0.033), whereas the other modified compounds showed no difference compared to BPD-MA.

Figure 2

Localization of BPD-MA and modified compounds in the mitochondria. (a–d) Bright field images of human KB cells. Cells stained with Mitotracker green (e–h) and photosensitizers (i–l). (m–p) Merged image of green and red channels showing colocalization (yellow) of photosensitizers and Mitotracker green. (q) Modified compound 0323 showed a significantly higher colocalization compared to BPD-MA using Pearson’s correlation coefficient and one-way ANOVA with Tukey’s post-hoc test (p = 0.0326). n = 5 images for each group. *p < 0.05. It should be noted that the photosensitizer fluorescence images do not correlate with the amount of compound intracellularly. These images were modified after capture to help the readers optimally visualize the intracellular localization of each compound. These adjustments have no effect on the calculation of colocalization because Pearson’s Correlation Coefficient is independent of pixel intensity/brightness.

The results demonstrate that verteporfin (BPD-MA) localizes to the mitochondria, which agrees with earlier studies reporting its localization to the mitochondria of pancreatic cancer cells,[32] endothelial cells, and prostate cancer cells.[33] There was improved mitochondria targeting of BPD-MA by using the triphenylphosphine (TPP) moiety, which also agrees with the literature showing TPP as an effective mitochondria-targeting molecule.[24] Although guanidinium and biguanidinium have also been used to target the mitochondria, our results demonstrate that they reduced mitochondria localization of BPD-MA. The study by Sibrian-Vasquez et al., which showed an increased mitochondria localization, did not quantify the localization and compare the synthesized compounds to the unmodified porphyrins.[15] Because porphyrins naturally localize to the mitochondria, without comparison to an untargeted control porphyrin, it is difficult to draw any conclusion regarding the ability of an added substituent to improve mitochondria localization. Moreover, because BPD-MA, which is a porphyrin derivative, has already been shown to significantly localize in the mitochondria, further improvements may be difficult.

Singlet Oxygen Species Production

The effect of BPD-MA and its derivatives on intracellular singlet oxygen species generation was investigated. Previous studies have used singlet oxygen quantum yield to compare different photosensitizers. However, these values do not accurately predict the intracellular singlet oxygen production, which depends on the localization of the photosensitizer and its proximity to oxygen. Therefore, singlet oxygen sensor green (SOSG), which reacts with singlet oxygen to produce a highly fluorescent endoperoxide molecule, was used as a marker of intracellular singlet oxygen production.[34] Due to its low cellular penetrance as a consequence of extracellular protein binding, the method described by Choudhury et al. was used.[35] Higher intensity of SOSG fluorescence, which corresponds to an increase in the singlet oxygen species production, was observed in photosensitizer-treated cells (Figure f–i) compared to the nontreated control group (Figure a) following irradiation. To further determine the singlet oxygen production, fluorescence intensity was quantified[36] using ImageJ’s mean gray value (n = 75 cells) (Figure j). There was significantly higher singlet oxygen production with 0317 (24 142 au ± 451.5) treatment compared to BPD-MA (21 072 au ± 423.1) treatment after light irradiation (p-value = 0.001) and with 0323 (35339 au ± 666) treated cells compared to verteporfin (p-value < 0.0001) and 0317 (p-value < 0.0001) treatment. Treatment with 0320 (18 674 au ± 405.5) did not show any difference from nontreated cells (17 456 au ± 442.4) (p-value = 0.3944).

Figure 3

Intracellular singlet oxygen production after irradiation. (a) KB cells with no drug. (b–e) Fluorescence of photosensitizers and (f–i) corresponding intracellular singlet oxygen production. (j) Singlet oxygen quantification after light treatment shows statistically significant increase in BPD-MA-treated (p-value < 0.0001), 0317-treated (p-value < 0.0001), and 0323-treated (p-value < 0.0001) cells compared with no drug; 0317-treated (p-value = 0.001) and 0323-treated (p-value < 0.0001) compared to BPD-MA-treated cells; and 0323-treated (p-value < 0.0001) compared with 0317-treated cells. Statistics were done using one-way ANOVA with Tukey post-hoc analysis. n = 75 cells for all groups. # used for comparison with no drug, * for comparison with BPD-MA and $ for comparison with 0317. ***p < 0.001, ****p < 0.0001. It should be noted, contrary to the previous figure, the postcapture parameters (i.e., LUTs) for the photosensitizer fluorescence images were constant in this figure to provide the viewer with a visual measure of the relative fluorescence of each molecule.

Figure 4

Changes in mitochondria membrane potential. JC-1 signal in control cells (a, b) and cells treated with BPD-MA (c, d), 0317 (e, f), 0320 (g, h), and 0323 (i, j) directly after (a–i) and 12 min (b–j) after irradiation. (k) Green-to-red channel ratio during longitudinal study for 12 min upon irradiation. Higher ratio corresponds to reduction in membrane potential. 0323-treated cells show significant reduction in mitochondria membrane potential at 6 min (p = 0.002), 8 min (p < 0.0001), 10 min (p < 0.0001), and 12 min (p < 0.0001). Statistical analysis was done using 2-way ANOVA (n = 20 cells for time = −1 for all groups; n = 30 for nontreated group thereafter and n = 40 for photosensitizer groups). **p < 0.01, ****p < 0.0001.

We observe the highest production of singlet oxygen species after irradiation of 0323-treated cells. From the previous colocalization experiments, it was shown that 0323 had the greatest accumulation in the mitochondria compared to the other photosensitizers. Because the mitochondria is an organelle with the highest intracellular oxygen concentration,[9] and because oxygen is required for PDT,[37] it was not surprising that the photosensitizer with the highest mitochondria localization also had the highest ROS production. It should also be noted that the excited state energy following near-infrared (NIR) illumination can be dissipated by multiple mechanisms, and that compounds that fluoresce efficiently are usually poor PDT agents and vice versa. Thus, compounds that efficiently convert excited state energy into emitted fluorescence must necessarily have less excited state energy available for the production of reactive oxygen species (i.e., via intersystem crossing, formation of singlet oxygen, and further reaction to generate other ROS). The fact that 0323 has the lowest fluorescence and highest ROS, whereas BPD-MA has the highest fluorescence and the lowest ROS is consistent with this explanation.

Changes in Mitochondria Membrane Potential

Singlet oxygen is highly reactive and damaging to membranes.[39] Thus, we hypothesized that the increased production of singlet oxygen in mitochondria will cause damage to its membranes. To test this hypothesis, we measured the mitochondria membrane potential (Δψm) with the JC-1 assay. JC-1 is a cationic dye that accumulates in the mitochondria depending on the Δψm.[40] During low Δψm, low concentrations of JC-1 localize in the mitochondria, forming mostly monomers that emit green fluorescence. However, with higher Δψm, there is an increased concentration of JC-1 in the mitochondria, leading to the formation of JC-1 aggregates that emit red fluorescence. Quantitative analysis of Δψm was performed by ratiometric analysis[41] of green monomer (green channel) to red aggregate (red channel) (Figure k). The ratio of monomer to aggregate fluorescence was used as a surrogate for analyzing the Δψm change. Before light irradiation, no difference between control and photosensitizer groups was observed, indicating no dark toxicity. As expected, the ratio of green monomer to red complex gradually increased during longitudinal observation for 12 min in BPD-MA-treated cells, indicating the decline of Δψm.[42] The modified compound 0323, which previously showed the highest localization in the mitochondria and greatest singlet oxygen production, also exhibited the largest Δψm toxicity, which was significantly higher from 6 min onward (p = 0.002 at 6 min, p < 0.0001 at 8 min, p < 0.0001 at 10 min, and p < 0.0001 at 12 min) compared to BPD-MA (Figure k).

As expected, the mitochondria membrane potential was decreased after 690 nm irradiation in photosensitizer-treated KB cells and the magnitude of depolarization directly correlated with the extent of mitochondria localization. Thus, the compounds BPD-MA and 0323 had the greatest mitochondrial localization and the highest Δψm toxicity. As noted above, singlet oxygen species are short lived and have a diffusion radius of only 20 nm, requiring that the location in which they are produced also be the site where they cause the most damage. Consistent with this, the superior accumulation of BPD-MA and 0323 in the mitochondria agrees with the greater observed damage to Δψm. Furthermore, our longitudinal study revealed a decline in membrane potential before mitochondria swelling, suggesting this phenomenon may be an event preceding early processes associated with apoptosis. Because the compound 0320 did not show a significant increase in SOS or change in Δψm in KB cells, this compound was not further studied.

Qualitative Changes: Actin, Mitochondria, and ROS Production

The next step in evaluating BPD-MA and its modified derivatives was to study the morphological changes in the cytoskeleton and mitochondria together with the production of reactive oxygen species (ROS) following irradiation. Nontreated cells displayed fibrous structures of actin filaments (Figure a), whereas BPD-MA-treated (Figure b), 0317-treated (Figure c) and 0323-treated (Figure d) cells showed fragmentation of the actin filaments. Importantly, 0317-treated cells showed less fragmentation and more polymerized actin than BPD-MA- and 0323-treated cells. Granular swelling of the mitochondria, which is a sign of cell damage, was also more prominent in BPD-MA-treated (Figure f) and 0323-treated (Figure h) cells compared to control (Figure e) and 0317-treated (Figure g) cells, which showed a normal tubular mitochondria structure. There was also an increase in ROS in BPD-MA-treated (Figure j), 0317-treated (Figure k), and 0323-treated (Figure l) cells compared to control cells (Figure i). Curiously, the greatest increase in reactive oxygen species was found to be in the nucleus, suggesting a significant effect of PDT on the reactive oxygen species production in the nucleus[43] and/or mitochondria nucleus retrograde signaling.[44] This prediction is supported by our finding of early signs of cell injury and pyknosis (next section)[45] upon initiation of PDT. Interestingly, 0323-treated cells showed apoptotic bodies indicated by the arrow in Figure l, which will be investigated further in a section below.

Figure 5

Qualitative changes observed in cytoskeleton, mitochondria morphology, and ROS following irradiation. (a–d) Actin cytoskeleton in nontreated cells and photosensitizer-treated cells. (e–h) Mitochondria morphology studies. (i–l) Changes in ROS production following light irradiation. (l) Apoptotic body indicated by the arrow.

We believe that the above chain of events begins with 0323 and BPD-MA localization in the mitochondria and their subsequent generation of reactive species, which then lead to mitochondrial membrane depolarization and loss of its integrity. Considering the integral role the mitochondrial membrane plays in the regulation of electrolyte balance,[46] its deterioration can be expected to bring about a charge imbalance that will cause mitochondria swelling. The mitochondria membrane potential is crucial for ATP production through oxidative phosphorylation.[46] Actin assembly and disassembly, which are essential for the structural integrity of the cell,[47] molecular signaling within the cell,[48] and mitochondria motility,[49,50] are dependent on ATP for their regulation. Therefore, damaging the Δψm will lead to its depolymerization, as shown in this study.

Analysis of Cell Death

Viability assays were then performed using propidium iodide on photosensitizer-treated KB cells irradiated with 690 nm light. Four different treatment groups were examined, namely a control group, which consisted of KB cells treated with no photosensitizer (Figure a) and treated groups preincubated with BPD-MA (Figure b), 0317 (Figure c) or 0323 (Figure d). As shown in Figure , after irradiation, propidium iodide staining was only observed in photosensitizer-treated cells, with 0323 showing the greatest cytotoxicity (93% PI positive), 0317 demonstrating an intermediate toxicity (54% positive), and BPD-MA displaying the lowest cytotoxicity (25% positive). Pyknosis, which involves an irreversible condensation of chromatin, was also observed.

Figure 6

Modified compounds reduce cell viability following irradiation. KB cells were incubated for 1 h with either no photosensitizer or 50 nM BPD-MA, 0317, or 0323 and exposed for 5 s to the NIR laser light. After 1 h, the cells were incubated with Hoechst dye (nuclear stain) and propidium iodide (vitality stain) before evaluation by fluorescence microscopy. (a–d) Merged images of Hoechst (blue) and propidium iodide (magenta) are shown, where dead cells appear magenta.

Apoptosis Through Cytochrome C Release from the Mitochondria

Considering the roles of mitochondria damage[51] and actin network disassembly[52] in mediating apoptosis, we evaluated the most effective photosensitizer to this point (0323) and the parent compound (BPD-MA) for the release of cytochrome C as a measure of one of the hallmarks of apoptosis. Two major mechanisms of cell death consist of necrosis and apoptosis. Apoptosis is desirable in PDT because it minimizes inflammation to adjacent normal tissue.[53] One of the first events that occurs in apoptosis is the release of cytochrome C from the inner membrane of the mitochondria to the cytosol.[54] Whereas release of cytochrome C from mitochondria was prominent in BPD-MA-treated (Figure b) and 0323-treated (Figure c) cells, little if any cytochrome C release was observed from the mitochondria in control cells (Figure a). This is consistent with previous studies, which have shown that BPD-MA induces apoptosis.[55,56] Moreover, apoptotic bodies, which are hallmarks of cell apoptosis,[57] can be observed in the photosensitizer-treated groups (see arrows). Another function of cytochrome C upon release from the mitochondria is to induce chromatin condensation following its translocation to the nucleus.[58] Interestingly, cytochrome C is localized more in the nucleus of 0323-treated cells than in BPD-MA-treated cells. This study shows PDT with the modified compound 0323 induces apoptosis similar to BPD-MA.

Figure 7

Induction of apoptosis following PDT. (a) Cytochrome C localization in KB cells with no photosensitizer. Cytochrome C release in BPD-MA-treated (b) and 0323-treated (c) cells. Arrows indicate apoptotic bodies.

Conclusions

In conclusion, we have developed a mitochondria-targeted photosensitizer by adding a triphenylphosphine (TPP+) group to carboxyl terminus of the parent PDT agent, verteporfin (0323). This mitochondria-targeting moiety was more efficacious than monoguanidium (0317) or bis-guanidinium (0320) in targeting the mitochondria. Improving the mitochondria localization of verteporfin resulted in increased singlet oxygen species (SOS), depolarization of mitochondria membrane potential, and cell death. Moreover, elevated actin depolymerization, increased reactive oxygen species, enhanced mitochondria swelling, pyknosis, and apoptosis were also observed. Based on these observations, we recommend tethering a TPP moiety to the desired PDT agent when developing a photosensitizer for cytotoxic applications.

Experimental Section

Chemicals and Materials

2-(1H-7-Azabenzotriazol-1-yl)-1,1,3,3-tetramethyl uronium hexafluorophosphate methanaminium (HATU) was obtained from Genscript Inc. (Piscataway, New Jersey). Guanidine hydrochloride, 1,2-dichloro ethane (1,2-DCE), diisopropylethylamine (DIPEA), dimethyl sulfoxide (DMSO), and all other reagents were purchased from Sigma-Aldrich. Trimethyltin hydroxide (Me3SnOH) was purchased from Alfa Aesar. For the synthesis of 0323, (2-aminoethyl)triphenylphosphonium bromide was synthesized according to reported procedures.[29,30]

Synthesis of Mono/Bicationic Derivatives of Verteporfin

For synthesis of mitochondria targeted monocationic derivatives of verteporfin (1 equiv) dissolved in anhydrous DMSO containing diisopropylethylamine (5 equiv) and HATU (1 equiv) were stirred for 30 min under argon atmosphere. To obtain 0317 and 0323, 3-fold molar excess of guanidine hydrochloride salt and 3-fold molar excess of (2-aminoethyl)triphenylphosphonium bromide, respectively, were then added and stirred overnight at room temperature, as outlined in Figure . The crude products were purified by preparative reverse phase high-performance liquid chromatography (HPLC) using a mobile phase of A = 20 mM ammonium acetate buffer, pH 7; B = acetonitrile; gradient 0–50% B in 30 min, 13 mL min–1, λ = 280 nm. Pure fractions were analyzed by LCMS and (see Figure S2 in the Supporting Information) pooled and lyophilized to furnish 0317 and 0323.

For synthesis of the polycationic derivatives of verteporfin, we used Me3SnOH reagent for mild methyl ester hydrolysis of verteporfin. Verteporfin (1 equiv) dissolved in anhydrous 1,2-DCE containing Me3SnOH (10 equiv) was stirred for 1 h at 80 °C. The reaction was monitored by LCMS and the crude product was purified by preparative reverse-phase HPLC using a mobile phase of A = 20 mM ammonium acetate buffer, pH 7; B = acetonitrile; gradient 0–50% B in 30 min, 13 mL min–1, λ = 280 nm. The methyl ester hydrolyzed verteporfin (1 equiv) dissolved in anhydrous DMSO containing diisopropylethylamine (5 equiv) and HATU (1 equiv) was stirred for 30 min under argon atmosphere. A 10-fold molar excess of guanidine hydrochloride salt was then added and stirred overnight at room temperature, as outlined in Figure . The crude product was purified by preparative reverse-phase HPLC using a mobile phase of A = 20 mM ammonium acetate buffer, pH 7; B = acetonitrile; gradient 0–50% B in 30 min, 13 mL min–1, λ = 280 nm. Pure fractions were analyzed by LCMS and (see Figure S3 in the Supporting Information) pooled and lyophilized to furnish 0320. The measured extinction coefficients of the aforementioned PDT agents were found to be 3.5 × 104 M–1 cm–1 (BPD-MA), 2.4 × 104 M–1 cm–1 (0317), and 2.2 × 104 M–1 cm–1 (0323).

Cell Culture, Plating

KB cells were cultured with 5 mL of folic acid (−) RPMI media (Gibco) with 5% fetal bovine serum (FBS) (Atlanta biologicals) and 1% penicillin/streptomycin (GIbco) in 25 mL cell culture flasks (Corning, Sigma-Aldrich). Splitting was done 2–3 times per week or when cell culture flasks reached about 90% confluence. Cells were plated on 4-well chambers (Nunc Lab-Tek Chamber Slide System) in 500 μL media and allowed to grow for 2–3 days or when plates reached 90% confluence. Due to the nature of photosensitizers, all drug and dye incubations were done in the dark.

Colocalization Experiment

Following cell plating, the cells were co-incubated with 50 nM drug and 15 nM Mitotracker green (obtained from Life Technologies) for 1 h in RPMI media. The cells were then washed two times with phosphate-buffered saline (PBS) and kept in CO2-independent media (Life technologies) for imaging. Because the excitation of the drug may cause morphological changes in the mitochondria within seconds, Mitotracker green was captured first at 488/520 nm and photosensitizers were captured at 690 nm.

Singlet Oxygen Experiment

The cells were co-incubated with 50 nM drug and 10 μM singlet oxygen sensor green (SOSG, Life Technologies) for 1 h in FBS (−) MEM. They were washed two times with PBS and kept in CO2-independent media for imaging. The cells were then irradiated with 690 nm light followed by SOSG capture at 488/520 nm.

Mitochondria Membrane Potential Experiment

The cells were incubated with 50 nM BPD-MA, 0317, 0320, or 0323 in RPMI for 1 h followed by two washes with PBS. They were then incubated in 7.665 μM JC-1 (a probe of mitochondrial membrane potential; Life Technologies) in RPMI for 10 min followed by two washes with PBS and kept in CO2-independent media. JC-1 was captured before irradiation at 590 nm (J-aggregates) and 529 nm (green monomers). The cells were then irradiated with 690 nm light. JC-1 was immediately captured at 590 and 529 nm at 2 min intervals for 12 min.

Cytoskeleton and Mitochondria Structure Experiment

The cells were incubated with 50 nM BPD-MA, 0317, or 0323 in RPMI for 1 h followed by two washes with PBS and kept in CO2-independent media. They were then irradiated using 690 nm light. Cells were then fixed with 4% PFA for 15 min, permeabilized with 1% Triton X-100 for 30 min and blocked with goat kit[31] for 60 min. The cells were then stained with F-actin green labeling kit (Life Technologies) for 30 min, 1 μM Mitotracker deep-red FM (Life Technologies) for 30 min, and 1 μg mL–1 Hoechst for 30 min. They were then washed two times with PBS and images were captured at 520 nm for actin, 461 nm for Hoechst, and 665 nm for mitochondria stain.

ROS Experiment

The cells were incubated with 50 nM BPD-MA, 0317, or 0323 in RPMI for 1 h followed by two washes with PBS. They were then incubated with 5 μM ROS in RPMI for 30 min followed by two washes with PBS and kept in CO2-independent media. The cells were irradiated with 690 nm light and ROS were captured at 488/520 nm.

Cell Death Experiment

The KB cells were incubated with 50 nM of BPD-MA, 0317, or 0323 diluted in RPMI media for 1 h. They were washed two times with PBS (Gibco), followed by irradiation for 5 s by the NIR laser light. After 1 h, the cells were incubated with 5 μg mL–1Hoechst dye (nuclear stain) and 1 μg mL–1propidium iodide (vitality stain) for 30 min. The cells were then washed two times with PBS, kept in CO2-independent media, and then imaged at 350/461 and 535/617 nm for Hoechst and propidium iodide dyes, respectively.

Apoptosis Experiment

The cells were incubated with 50 nM BPD-MA or 0323 in RPMI for 1 h followed by two washes with PBS and kept in CO2-independent media. They were then irradiated using 690 nm light and then incubated at 37 °C and 5% CO2 for 2 days. The cells were then fixed using 4% PFA for 15 min, permeablized using 1% Triton X-100 for 30 min, and blocked with a goat kit to prevent nonspecific binding for 30 min. They were then stained with 5 μg mL–1 mouse IgG anti-cytochrome C (BioLegend) for 20 h and washed three times with PBS. The cells were then counterstained with Alexa Fluor 594 Goat antimouse antibody (594) for 90 min and washed three times with PBS. Cytochrome C was then captured using 617 nm emission wavelength.

Microscopy and Imaging Analysis

Images were taken using Nikon Ti-S epifluorescence microscope equipped with incubating chamber set at 37 °C and an oil immersion 60× objective, and the images were obtained with NIS-Elements software. Colocalization analysis was done using the JACoP plugin tool in Fiji ImageJ to calculate the Pearson’s correlation coefficient. Singlet oxygen production analysis was done using ImageJ to obtain SOSG intensities. Fluorescence intensity map was obtained using ImageJ. Mitochondria membrane potential analysis was also done using ImageJ to calculate green-to-red channel intensity ratio. For ImageJ analysis, the region of interest feature was selected to precisely quantify the mean gray value intensity of each cell.

Statistics

Statistical analysis was performed using Prism 7. One-way and two-way ANOVA were used to compare the population mean of different treatment groups at 0.05 significant level. Tukey post-hoc test was further applied to determine the statistical significance of mean difference among the groups.