Nanoparticle delivery of a pH-sensitive prodrug of doxorubicin and a mitochondrial targeting VES-H8R8 synergistically kill multi-drug resistant breast cancer cells

Multi-drug resistance (MDR) remains a major obstacle in cancer treatment while being heavily dependent on mitochondrial activity and drug efflux. We previously demonstrated that cationic lipids, such as the vitamin E succinate modified octahistidine-octaarginine (VES-H8R8) conjugate, target mitochondria, resulting in depolarized mitochondria and inhibited drug efflux in MDR breast cancer cells. We hypothesized that the effective cell uptake, efflux inhibition, and mitochondrial depolarization properties of VES-H8R8 would synergistically enhance the toxicity of a pH-sensitive prodrug of doxorubicin (pDox) when co-encapsulated in nanoparticles (NPs). pDox was successfully synthesized and validated for pH-sensitive release from NPs under lysosome-mimicking, acidic conditions. The synergistic effect of VES-H8R8 and pDox was confirmed against MDR breast cancer cells in vitro. Importantly, synergism was only observed when VES-H8R8 and pDox were co-encapsulated in a single nanoparticulate system. The synergistic mechanism was investigated, confirming superior pDox uptake and retention, Pgp efflux inhibition, mitochondrial depolarization, and enhanced induction of ROS, and apoptosis. This work demonstrates the translational potential of doubly-loaded NPs co-encapsulating pDox with VES-H8R8 to synergistically kill MDR breast cancer cells.

Palmityl-Doxorubicin synthesis. Palmitic acid hydrazide was reacted with the ketone of doxorubicin, to produce the ph-sensitive hydrazone bond containing palmityl-doxorubicin (pDox), as synthesized previously ( Fig. S1A) 22,26 . Briefly, 150 mg of doxorubicin free base (MedKoo Biosciences, Inc, NC, USA) and 82 mg of Palmitic acid hydrazide (1.1 eq) were added to 300 mL of MeOH:DCM solution (1:1) containing 15 uL of trifluoroacetic acid. The reaction proceeded for 24 hours and reaction progress monitored by thin layer chromatography (TLC) using MeOH:DCM (1:3). After the reaction was complete, the solvent was removed by rotary evaporation, and the crude was dissolved in a small volume of MeOH:DCM (1:20). The dissolved crude was purified by silica column chromatography using DCM, and eluting with increasing concentrations of MeOH until the product, pDox, was eluted. Fractions containing pDox were collected, solvent removed by rotary evaporation, and placed in a vacuum dry oven to yield a red oil (yield = 65%). The 1 25), and pDox encapsulation was quantified by fluorescence measurement (ex/em 480/580 nm) against a standard curve. VES-H 8 R 8 encapsulation was quantified using amino acid analysis. Average drug loading was obtained from three separate NP batches, calculated as follows: Figure 1. Schematic representation of the proposed mechanism of synergy of co-encapsulated vitamin E succinate modified octahistidine-octaarginine (VES-H 8 R 8 ) and palmityl-doxorubicin (pDOX). (i) Structure of poly(D,L-lactide-co-2-methyl-2-carboxytrimethylenecarbonate) 12K -g-poly(ethylene glycol) 10K (P(LA-co-TMCC)-g-PEG) used as amphiphilic polymer for NP synthesis. (ii) Structure of vitamin E succinate modified octahistidine-octaarginine (VES-H 8 R 8 ), with vitamin E succinate in red and octahistidine-octaarginine in blue. (iii) Structure of the pH-responsive, palmityl-doxorubicin (pDOX), with doxorubicin in pink, palmityl moiety in green, and the pH-responsive hydrazone bond in black. VES-H 8  NP potency assay. Cells were seeded into 96-well flat-bottomed tissue culture plates at a density of 3,000 cells per well, and allowed to adhere for 24 hours. Treatments were incubated with cells for 24 hours, then replenished with fresh media and incubated for an additional 48 hours. Singly-loaded NPs, DNPs and VNPs, were dose-matched to the respective drug concentration of doubly loaded NPs, VDNPs. Free doxorubicin was used as a control where doxorubicin hydrochloride was dissolved in dimethylsulfoxide (DMSO) at a concentration of 1 mg/mL. Free VES-H8R8 was also used as a control dissolved in DMSO at a concentration of 1 mg/mL. Free doxorubicin was used as a control where doxorubicin hydrochloride was dissolved in dimethylsulfoxide (DMSO) at a concentration of 1 mg/mL. Free VES-H 8 R 8 was also used as a control dissolved in DMSO at a concentration of 1 mg/mL. After the addition of NP formulations, cells were allowed to grow for 24 h at 37 °C in a 5% CO 2 and 95% air humidified incubator. The Presto Blue (Life Technology) assay was used as a proxy for cell relative viability, calculated as follows:  www.nature.com/scientificreports www.nature.com/scientificreports/ USA). Both DNP-and VDNP-treated cells exhibited fluorescence in the FL-2 channel due to the fluorescence of pDox. As such, VNPs were used to explore the effects of mitochondrial membrane polarization. VNPs was incubated at its IC 50 (14 µM) as well as the IC 50 of the concentration of VES-H 8 R 8 in VDNPs (4 µM). 20,000 cells were seeded into 48-well plates and allowed to adhere for 24 h at 37 °C in a 5% CO2 and 95% air humidified incubator. The cells were treated the next day for 5 hours in full media. Carbonyl cyanide m-chlorophenyl hydrazone (CCCP) is a rapid mitochondrial depolarizer and was used as a positive control (50 µM). Following treatment, the cells were washed with PBS thrice, and then incubated at 37 °C with 10 µM of JC-1 in full media for 30 min. The cells were then washed thrice in PBS, harvested, and placed on ice before measuring fluorescence in a flow cytometer within an hour. Cell debris and doublets were gated out using FSH-A vs FSH-H, and at least 10,000 events were collected. A gate was set according to DMSO and CCCP treated cells in the FL-2 channel (585 nm / 40 nm) to measure the proportion of JC-1 aggregate fluorescence versus JC-1 monomer fluorescence in the FL-1 channel (533 nm / 30 mn). Averages were obtained from three biological repeats. Changes in the mitochondria membrane potential (ΔΨm) were expressed using the following equation: Apoptotic cells have exposed phosphatidylserine on the external cell membrane that binds to Annexin V-Cy5, while necrotic cells exhibit membrane damage that allows Live or Dead Fixable Dead Cell Staining dye to bind to DNA. Briefly, 20,000 EMT6/AR-1 cells were seeded into 48-well plates and allowed to adhere for 24 h at 37 °C in a 5% CO2 and 95% air humidified incubator. To analyze apoptosis, the EMT6/AR-1 cells were treated the next day for 2, 5, or 24 hours at the IC 50 of VDNPs, where DNPs, VNPs, and DNPs+VNPs were dose-matched to the respective concentration of pDox or VES-H 8 R 8 in VDNPs. Untreated control cells were used to measure background fluorescence in the FL-4 channel. Identical setup for the apoptosis assay was used to analyze necrotic cells, but only for a 24 hour incubation. Following treatment, floating cells were collected, adhered cells were harvested, washed with PBS, and then incubated with Annexin V-Cy5 (1 uL/1 mL) or with Live or Dead Fixable Dead Cell Staining Kit for 20 minutes at 25 °C as per manufacturer's protocol. Cells were then placed on ice before measuring fluorescence in a flow cytometer within an hour. Cell debris and doublets were gated out using FSH-A vs FSH-H, and at least 10,000 events were collected. Gates were set in the FL-4 channel (ex = 640 nm, em = 675 nm / 25 nm) for Annexin-V-Cy5 or Live or Dead Fixable Dead Cell Staining relative to unstained controls. Proportion of apoptotic or necrotic cells were averaged from 3 biological repeats.
Statistical analysis. All statistical analyses were performed using Graph Pad Prism version 6.00 for Windows (Graph Pad Software, San Diego, California, www.graphpad.com). Differences among groups were assessed by one-way ANOVA with Bonferroni post hoc correction. Graphs are annotated where p-values are represented as *p ≤ 0.05, **p ≤ 0.01, or ***p ≤ 0.001. All data are presented as mean ± standard deviation.

Results and Discussion
NP characterization and pDox release. Blank NPs, singly loaded NPs, VES-H 8 R 8 -NPs (VNPs) and pDox-NPs (DNPs), and doubly loaded NPs, VES-H 8 R 8 -pDox-NPs (VDNPs) were prepared by nanoprecipitation in water. The average hydrodynamic sizes of the NPs were ~60 nm, indicating that the size of the NPs is independent of drug encapsulation. (Table 1) Most of the NPs also exhibited polydispersity indices below 0.20, indicative of narrow size distribution 29 . The zeta potential of blank NPs was −12.9 ± 1.6 mV, while the zeta potential of singly loaded NPs were −4.6 ± 0.7 mV and 11.8 ± 1.8 mV for DNPs and VNPs, respectively. VDNPs, exhibited a zeta potential of 13.6 ± 1.7 mV. While all the values are in the range of neutral zeta potential (−20 to +20 mV), the trends from negative to positive charge is indicative of successful encapsulation of cationic VES-H 8 R 8 and pDox 30,31 . The more neutral zeta potential of DNPs is attributed to the protonated amine of pDox neutralizing the free acids of P(LA-co-TMCC)-g-PEG, while the more positive zeta potential of VDNPs is attributed to positive charges found within both pDox and VES-H 8  www.nature.com/scientificreports www.nature.com/scientificreports/ of 15.1 ± 0.9%. Interestingly, while pDox alone precipitates, encapsulated pDox is soluble (Fig. S2A), indicating successful drug encapsulation. VDNPs achieved similar drugs loadings to each singly-loaded NPs, with pDox loading of 10.9 ± 1.3% and VES-H 8 R 8 loading of 11.7 ± 0.7%, while the total drug loading amounts to 22.6%. The concentration of each drug in the encapsulated NPs is shown in Fig. S2B.
The NP stability was monitored by hydrodynamic size and polydispersity indices (PDI) measurements over time ( Fig. 2A,B). Interestingly, while NPs with high drug loadings typically exhibit limited stability under physiological conditions, all formulations maintained size and PDI over at least 96 h at 37 °C 33 . These results show that these P(LA-co-TMCC)-g-PEG NPs are stable in vitro and should be efficacious in vivo as they have been previously shown to be stable in serum, with a half-life of 71 ± 12 h 29 , and to enhance tumour accumulation of chemotherapeutics vs. that with conventional excipient delivery 34 . The pH-dependent release of doxorubicin (Dox) from the pDox prodrug was validated by mass spectrometry, where all of the pDox was hydrolyzed within 24 h at pH 5.0 (Fig. S3). Dox release was studied at both physiological and acidic pH, using a dialysis test as previously reported 35,36 . Free Dox control showed the expected rapid release, with >85% Dox released within 4 h at both pH 5.0 and 7.4 (Fig. 2C). Both DNPs and VDNPs showed limited Dox release (<20%) over 48 h at pH 7.4. However, the same formulations released up to 81% of Dox under acidic conditions (pH 5.0). Together, these results confirm the pH-sensitive release of pDox from the NPs. The similar results obtained for DNPs and VDNPs also indicate that VES-H 8 R 8 co-encapsulation do not influence pDox hydrolysis and subsequent Dox release. More importantly, the release study at pH 5.0 is representative of late lysosomal acidic conditions, suggesting suitable endolysosomal escape of Dox following cell uptake 36 .  anti-cancer activity of pDox and VES-H 8 R 8 were assessed in the parental breast cancer cell line, EMT6/P, and the MDR breast cancer cell line, EMT6/AR-1, using the Presto Blue metabolic assay. The ability of treated cancer cells to metabolize Presto Blue was a proxy for anti-cancer activity (Tables 2, S1). pDox alone could not be tested at concentrations greater than 20 µM because of limited solubility. Importantly, the pH-responsive release of doxorubicin from pDox was demonstrated at pH 5.0 (Fig. S3). This supports the pH-responsive release of doxorubicin when pDox encapsulated NPs get trafficked through the endo-lysosomal pathway. While the half maximal inhibitory concentration (IC 50 ) of Dox in EMT6/P is 0.16 ± 0.02 µM, it increases up to 82.19 ± 7.36 µM in EMT6/ AR-1, indicative of Dox efflux and MDR status in agreement with the literature 37,38 . The IC 50 of VES-H 8 R 8 in EMT6/AR-1 and EMT6/P was 12.02 ± 1.94 µM and 11.12 ± 2.56 µM, respectively. The IC 50 of VES-H 8 R 8 on the parental and that on MDR breast cancer cell lines were not significantly different, suggesting that the cell uptake and the mechanism of cancer cell death of VES-H 8 R 8 are independent of the Pgp efflux. We also investigated the anti-cancer activity of the combination of free Dox and VES-H 8 R 8 in both breast cancer cell lines. The molar ratio of Dox to VES-H 8 R 8 was based on the drug loading of pDox and VES-H 8 R 8 in NPs (Table 1). The combination of free Dox and VES-H 8 R 8 in EMT6/P resulted in an IC 50 with a total drug concentration of 0.30 ± 0.03 µM, similar to that of Dox alone (0.16 ± 0.02 µM), indicating that VES-H 8 R 8 did not increase the anti-cancer activity of Dox in the drug-sensitive cancer cell line. On the contrary, the combination of free Dox and VES-H 8 R 8 in the MDR EMT6/AR-1 cells resulted in an IC 50 of 32.64 ± 1.70 µM requiring concentrations of 3.9-fold less Dox than Dox alone and 1.54-fold less VES-H 8 R 8 than VES-H 8 R 8 alone.
Next, we investigated the benefits of co-encapsulating VES-H 8 R 8 and pDox in NPs. VNPs exhibited an IC 50 in the low µM range in both EMT6/AR-1 and EMT6/P cells, similar to the anti-cancer activity of free VES-H 8 R 8 ( Table 2 and Fig. S1). Here, the total drug concentration is the sum of the concentrations of pDox and Ves-H 8 R 8 , as the polymers used in the NPs were previously shown to be non-toxic to cancer cells 39 . On the Dox-sensitive EMT6/P cells, DNPs, the mixture of DNPs and VNPs, and VDNPs exhibited IC 50  . The requirement of co-encapsulation is fundamental for in vivo translation as co-encapsulated drug nanoformulations administered in vivo can lead to superior tumor reduction and maintenance of drug ratios at the tumor site relative to the co-administered free drugs 19 .
To assess the synergistic advantage of co-encapsulating VES-H 8 R 8 and pDox in NPs, a combination index (CI) was calculated based on the median effect analysis 27 . A median effect plot was produced by converting the dose response curves in Fig. S2C,D into log [(fa) −1 − 1] −1 vs log (D), where fa is the fraction of non-viable cells and D is the drug concentration (Fig. S3). R 2 > 0.90 indicates statistical validity of the analysis and conforms to mass action law 28 . Furthermore, nonparallel lines in the median effect plots infers non-exclusivity in the mechanism of action of VES-H 8 R 8 and pDox 28 . The median effect plot was further analyzed to obtain a CI. CI > 1 reflects drugs antagonism, CI = 1 reflects drug additive effect, and CI < 1 shows drug synergism 27 . The CI of the mixture of DNPs and VNPs was 0.99 ± 0.07, indicating that the mixture of DNPs and VNPs was additive in anti-cancer activity. Importantly, the CI of VDNPs was 0.61 ± 0.04, suggesting synergism of the co-encapsulated drugs against the MDR EMT6/AR-1 cells. In previous studies, a similar CI was achieved with NP formulations combining cytarabine and daunorubicin at a molar ratio of 5:1, which demonstrated improved survival of patients 20,42 . The synergism of VDNPs may be attributed to the pH-dependent release of Dox from NPs, which would poison the topoisomerase II enzyme, coupled with the toxicity of VES-H 8 R 8 by targeting of the mitochondria with subsequent depolarization 14,43,44 . Furthermore, the dose reduction index indicates the fold decrease of a drug required VDNPs enhances uptake, and retention of pDox. To elucidate the synergistic mechanism of VDNPs, we investigated the time-dependent uptake of free Dox, DNPs and VDNPs in MDR EMT6/AR-1 cells over 2, 5, and 24 h, using confocal microscopy (Fig. 3A). In EMT6/AR-1 cells, the free Dox control showed minimal Dox fluorescence at all time points, confirming active drug efflux in MDR cancer cells. Cells treated with DNP and VDNP exhibited both punctate organelles and fluorescence in the cytosol and nucleus at 24 h after treatment, indicating that the NPs were in the endo-lysosomal compartments and the doxorubicin was in the cytosol and the nucleus, which is consistent with previous reports 45 . Both DNPs and VDNPs showed time-dependent Dox accumulation up to 24 h post treatment. More importantly, VDNPs qualitatively exhibited enhanced uptake relative to DNPs and free Dox at all time points. DNP-and VDNP-treated cells exhibit punctate organelles, indicative of the endo/lysosomal entrapment, as well as diffuse cytoplasmic fluorescence, indicative of endosomal escape 46 . For both DNPs and VDNPs, there were indications of Dox release from the NPs and colocalization with cell nuclei, the intracellular target of Dox 44 .
We further quantified the difference in Dox fluorescence between free Dox, DNPs and VDNPs using flow cytometry (Fig. 3B). Here, we incubated treatment groups for up to 24 h to measure time-dependent uptake. To measure retention, we incubated treatment groups for 24 h, then replenished with fresh media and incubated the cells for an additional 24 h to measure the amount of Dox retained after 48 h. At all time points, VDNPs delivered significantly more Dox in the MDR EMT6/AR-1 cells relative to DNPs and free Dox controls (*** <0.001). Specifically, after a 24 h incubation, DNPs treated cells exhibited a mean fluorescence intensity (M.F.I.)  14,48 . Therefore, we hypothesized that encapsulated VES-H 8 R 8 inhibits the Pgp efflux of Dox, leading to increased Dox accumulation in the cytosol. To test our hypothesis, we incubated DNPs with free vitamin E succinate and evaluated Dox retention. After 24 h of incubation, followed by 24 h in fresh medium, the treated EMT6/AR-1 cells exhibited a 2.0 fold increase in Dox retention relative to DNPs alone at 48 h (Data not shown). The enhanced uptake and retention of VDNPs clearly demonstrate the benefits of co-encapsulating pDox and VES-H 8 R 8 . Improved uptake and retention along with Pgp efflux inhibition help to explain the synergism observed in the anti-cancer activity of VDNPs.

Doubly loaded NPs induce mitochondrial depolarization and ROS.
To deepen the understanding of the synergistic mechanism of VDNPs, we confirmed the effect of VES-H 8 R 8 on mitochondrial membrane potential when encapsulated in NPs. To monitor mitochondrial membrane potential, JC-1 was used as a mitochondria-specific polarization probe. To avoid Dox fluorescence interference with the JC-1 assay, we used VES-H 8 R 8 loaded NPs. Mitochondrial membrane polarization data are shown normalized to no treatment group (Fig. 4A). As expected, blank NPs had no effect on mitochondrial membrane polarization, and the positive control group, carbonyl cyanide m-chlorophenyl hydrazone (CCCP), depolarized the mitochondrial membrane. At the IC 50 concentration (14 µM), VNPs significantly depolarized the mitochondrial membrane relative to blank NPs and no treatment control, validating that VES-H 8 R 8 is capable of mitochondrial depolarization when encapsulated in NPs (*** <0.001). More importantly, VNPs significantly depolarized the mitochondria at 4 µM, the equivalent dose of VES-H 8 R 8 at the IC 50 of VDNPs, relative to the no treatment group, indicating that depolarization is involved in the synergistic anti-cancer activity of VDNPs (* <0.05). These results are consistent with the mitochondrial depolarization induced by VES-H 8 R 8 14 . Dox has also been demonstrated to indirectly depolarize cancer cell mitochondria through Dox-induced oxidative stress [49][50][51] . It is possible that the co-delivery of VES-H 8 R 8 and pDox in VDNPs requires lower doses of each drug to affect mitochondria. VES-H 8 R 8 may directly act upon the mitochondria while Dox induces intracellular events, such as reactive oxygen species (ROS) production, that indirectly affect mitochondria. The combined effects of encapsulated VES-H 8 R 8 and pDox in VDNPs on mitochondrial depolarization further aid in understanding the synergistic cytotoxicity observed.
As both Dox and VES-H 8 R 8 have been demonstrated to induce ROS, we investigated the effect of VDNPs on ROS levels, at the IC 50 of VDNPs. As controls, the singly-loaded NPs and the mixture of singly loaded NPs were dose-matched to the respective drug in VDNPs. VDNPs significantly induced ROS at all time points (Fig. 4B). Specifically, VDNPs increased ROS levels to 224.9 ± 29.6%, 356.0 ± 23.5%, and 450.4 ± 40.8% relative to the no treatment group, upon incubation at 2, 5, and 24 h, respectively. Both DNPs and VNPs controls did not significantly induce an increase in ROS levels compared to no treatment. VDNPs significantly induced ROS levels compared to the mixture of DNPs and VNPs at 6 and 24 h of incubation, further highlighting the requirement of co-encapsulating VES-H 8 R 8 and pDox to maximize synergism. (*** <0.001) The increase in ROS upon VDNPs treatment is consistent with the oxidative stress typically observed with both Dox and VES-H 8 R 8 treatment 14,51 . It is clear that mitochondrial membrane depolarization and induction of ROS is involved in the synergistic cytotoxicity of VDNPs.

Doubly loaded NPs induce mainly apoptosis. Mitochondrial membrane depolarization and induction
of ROS typically result in apoptosis 52 . Furthermore, VES-H 8 R 8 was previously shown to induce apoptosis in the MDR EMT6/AR-1 cells 14 . To avoid fluorescent bleed over from Dox, apoptotic and necrotic cells were stained separately with Annexin V-Cy5 and 7-AAD. VDNPs incubated at its IC 50 significantly increased the proportion of apoptotic cells at 2, 5, and 24 h of incubation relative to singly-loaded NPs, DNPs and VNPs, and their mixture (** <0.01) (Fig. 5A). Specifically, VDNPs increased apoptotic levels to 21.7 ± 3.3%, 26.0 ± 6.3%, and 29.6 ± 5.6% upon incubation at 2, 5, and 24 h, respectively. Importantly, co-treating cells simultaneously with DNPs and VNPs did not significantly induce apoptosis relative to no treatment control group and both singly loaded NPs, further supporting the required co-encapsulation to trigger apoptosis. Thus, for annexin V-Cy5 binding, VDNP treatment likely resulted in EMT6/AR-1 cells exposing phosphatidylserine on the cell surface membrane. The induction of apoptosis by VDNPs agrees with the induced apoptosis observed with the treatment of EMT6/ AR-1 cells with VES-H 8 R 8 , as previously shown 14 . As necrotic cells with compromised membranes could also stain for annexin-V, as a false positive for apoptosis, we evaluated the induction of necrosis upon VDNPs treatment. VDNPs significantly increased the proportion of necrotic cells relative to no treatment control (** <0.01) (Fig. 5B). However, the increase in necrotic cell number was not as large as the increase in the proportion of apoptotic cells, suggesting that VDNPs treatment causes minimal membrane damage, and therefore mainly induces VDNPs induced apoptosis in a time-dependent manner whereas singly-loaded NPs, palmityl-doxoribicin NPs (DNPs) and vitamin E succinate modified octahistidine-octaarginine NPs (VNPs), or their mixture (VNPs + DNPs) did not induce apoptosis at 2, 5, and 24 h of incubation. (B) VDNPs increased the proportion of cells that were necrotic after a 24 h incubation, relative to singly-loaded NPs, DNPs and VNPs, or their mixture (VNPs + DNPs). VDNPs were incubated at the IC 50 , where the other NP formulations were dose-matched to the respective drug concentration for either palmityl-doxorubicin (pDox) or vitamin E succinate modified octahistidine-octaarginine (VES-H 8 R 8 ). Data are presented as a mean (n = 3) ± SD, and statistical analysis was performed using one-way ANOVA and Tukey's multiple comparison test (** <0.01, *** <0.001).